Add files using upload-large-folder tool

LTA_openwebtext_dualt/logs/elfopt_8gpu/lta_owt_len1024_elfopt_muon_ema_ddit768x12_8gpu_5epoch_20260513_023024.log

@@ -0,0 +1,617 @@
+[rank6]:[W513 02:30:29.571793088 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 6]  using GPU 6 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+[rank2]:[W513 02:30:29.573074597 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 2]  using GPU 2 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+[rank0]:[W513 02:30:29.621398720 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 0]  using GPU 0 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+t-20260513102957-v877z-worker-0:10222:10222 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10222:10222 [0] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10222:10222 [0] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10222:10222 [0] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10222:10222 [0] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10228:10228 [6] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10228:10228 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10224:10224 [2] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10224:10224 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10228:10228 [6] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10228:10228 [6] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10224:10224 [2] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10224:10224 [2] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10228:10228 [6] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10224:10224 [2] NCCL INFO Comm config Blocking set to 1
+[rank1]:[W513 02:30:29.693479602 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 1]  using GPU 1 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+[rank3]:[W513 02:30:29.694131805 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 3]  using GPU 3 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+[rank7]:[W513 02:30:29.696815174 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 7]  using GPU 7 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+t-20260513102957-v877z-worker-0:10223:10223 [1] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10223:10223 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10223:10223 [1] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10223:10223 [1] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10225:10225 [3] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10225:10225 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10225:10225 [3] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10225:10225 [3] NCCL INFO NCCL version 2.25.1+cuda12.8
+[rank5]:[W513 02:30:29.704393470 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 5]  using GPU 5 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+t-20260513102957-v877z-worker-0:10223:10223 [1] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10225:10225 [3] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10229:10229 [7] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10229:10229 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10229:10229 [7] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10229:10229 [7] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10229:10229 [7] NCCL INFO Comm config Blocking set to 1
+[rank4]:[W513 02:30:29.710692738 ProcessGroupNCCL.cpp:4571] [PG ID 0 PG GUID 0 Rank 4]  using GPU 4 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. Specify device_ids in barrier() to force use of a particular device, or call init_process_group() with a device_id.
+t-20260513102957-v877z-worker-0:10227:10227 [5] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10227:10227 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10227:10227 [5] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10227:10227 [5] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10227:10227 [5] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10226:10226 [4] NCCL INFO cudaDriverVersion 12080
+t-20260513102957-v877z-worker-0:10226:10226 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10226:10226 [4] NCCL INFO Bootstrap: Using eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10226:10226 [4] NCCL INFO NCCL version 2.25.1+cuda12.8
+t-20260513102957-v877z-worker-0:10226:10226 [4] NCCL INFO Comm config Blocking set to 1
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NET/Plugin: Loaded net plugin NCCL RDMA Plugin v9 (v9)
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NET/Plugin: Loaded collnet plugin SHARP (v9)
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Plugin Path : /opt/hpcx/nccl_rdma_sharp_plugin/lib/libnccl-net.so
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO P2P plugin v9 IBext_v9
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth1
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NCCL_IB_PCI_RELAXED_ORDERING set by environment to 1.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NET/IB : Using [0]mlx5_1:1/RoCE [1]mlx5_4:1/RoCE [2]mlx5_5:1/RoCE [3]mlx5_6:1/RoCE [4]mlx5_7:1/RoCE [5]mlx5_8:1/RoCE [6]mlx5_9:1/RoCE [7]mlx5_10:1/RoCE [RO]; OOB eth1:10.82.112.1<0>
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Using network IBext_v9
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO ncclCommInitRankConfig comm 0x98ac490 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 69020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO ncclCommInitRankConfig comm 0xbd43ac0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 65040 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO ncclCommInitRankConfig comm 0xa99de10 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId 73020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO ncclCommInitRankConfig comm 0x98a1870 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 67020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO ncclCommInitRankConfig comm 0x9b15700 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 6b020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO ncclCommInitRankConfig comm 0xaa777e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId 75020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO ncclCommInitRankConfig comm 0xa87b0d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId 71020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO ncclCommInitRankConfig comm 0xa1cc500 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 6f020 commId 0xc3d8f44253f33569 - Init START
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO RAS client listening socket at ::1<28028>
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Bootstrap timings total 0.008778 (create 0.000020, send 0.000074, recv 0.008247, ring 0.000142, delay 0.000000)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Bootstrap timings total 0.011209 (create 0.000023, send 0.000069, recv 0.000102, ring 0.010580, delay 0.000001)
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Bootstrap timings total 0.000624 (create 0.000023, send 0.000070, recv 0.000111, ring 0.000105, delay 0.000000)
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Bootstrap timings total 0.002691 (create 0.000020, send 0.000076, recv 0.000032, ring 0.000105, delay 0.000000)
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Bootstrap timings total 0.022499 (create 0.000026, send 0.000074, recv 0.017801, ring 0.002214, delay 0.000001)
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Bootstrap timings total 0.078898 (create 0.000026, send 0.000068, recv 0.070159, ring 0.008328, delay 0.000000)
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Bootstrap timings total 0.004769 (create 0.000021, send 0.000079, recv 0.000076, ring 0.004295, delay 0.000000)
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Bootstrap timings total 0.072414 (create 0.000030, send 0.000069, recv 0.061280, ring 0.004294, delay 0.000001)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO MNNVL busId 0x67020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO MNNVL busId 0x6b020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO MNNVL busId 0x75020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO MNNVL busId 0x6f020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO MNNVL busId 0x65040 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO MNNVL busId 0x69020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO MNNVL busId 0x71020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO MNNVL busId 0x73020 fabric UUID 0.0 cliqueId 0x0 state 3 healthMask 0x0
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /var/run/nvidia-topologyd/virtualTopology.xml
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Setting affinity for GPU 3 to 03ffffff,ffffffff,ffffffff
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Setting affinity for GPU 2 to 03ffffff,ffffffff,ffffffff
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Setting affinity for GPU 1 to 03ffffff,ffffffff,ffffffff
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Setting affinity for GPU 4 to 0fffff,ffffffff,ffffffff,fc000000,00000000,00000000
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Setting affinity for GPU 0 to 03ffffff,ffffffff,ffffffff
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Setting affinity for GPU 7 to 0fffff,ffffffff,ffffffff,fc000000,00000000,00000000
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Setting affinity for GPU 5 to 0fffff,ffffffff,ffffffff,fc000000,00000000,00000000
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Setting affinity for GPU 6 to 0fffff,ffffffff,ffffffff,fc000000,00000000,00000000
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO NVLS multicast support is available on dev 5
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO NVLS multicast support is available on dev 7
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO NVLS multicast support is available on dev 2
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO NVLS multicast support is available on dev 4
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO NVLS multicast support is available on dev 3
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO NVLS multicast support is available on dev 6
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO NVLS multicast support is available on dev 1
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO NVLS multicast support is available on dev 0
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO comm 0xaa777e0 rank 7 nRanks 8 nNodes 1 localRanks 8 localRank 7 MNNVL 0
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO comm 0xa99de10 rank 6 nRanks 8 nNodes 1 localRanks 8 localRank 6 MNNVL 0
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO comm 0x98ac490 rank 2 nRanks 8 nNodes 1 localRanks 8 localRank 2 MNNVL 0
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO comm 0x9b15700 rank 3 nRanks 8 nNodes 1 localRanks 8 localRank 3 MNNVL 0
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO comm 0x98a1870 rank 1 nRanks 8 nNodes 1 localRanks 8 localRank 1 MNNVL 0
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO comm 0xa87b0d0 rank 5 nRanks 8 nNodes 1 localRanks 8 localRank 5 MNNVL 0
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO comm 0xbd43ac0 rank 0 nRanks 8 nNodes 1 localRanks 8 localRank 0 MNNVL 0
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO comm 0xa1cc500 rank 4 nRanks 8 nNodes 1 localRanks 8 localRank 4 MNNVL 0
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 00/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 01/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 02/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 03/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 04/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 05/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 06/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 07/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 08/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 09/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 10/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 11/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 12/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 13/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 14/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 15/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 16/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 17/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 18/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 19/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 20/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 21/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 22/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Channel 23/24 : 0 1 2 3 4 5 6 7
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO P2P Chunksize set to 524288
+t-20260513102957-v877z-worker-0:10225:10374 [3] NCCL INFO [Proxy Service] Device 3 CPU core 48
+t-20260513102957-v877z-worker-0:10225:10375 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 50
+t-20260513102957-v877z-worker-0:10223:10376 [1] NCCL INFO [Proxy Service] Device 1 CPU core 2
+t-20260513102957-v877z-worker-0:10223:10377 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 4
+t-20260513102957-v877z-worker-0:10229:10378 [7] NCCL INFO [Proxy Service] Device 7 CPU core 146
+t-20260513102957-v877z-worker-0:10229:10379 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 150
+t-20260513102957-v877z-worker-0:10224:10380 [2] NCCL INFO [Proxy Service] Device 2 CPU core 48
+t-20260513102957-v877z-worker-0:10224:10381 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 50
+t-20260513102957-v877z-worker-0:10227:10382 [5] NCCL INFO [Proxy Service] Device 5 CPU core 106
+t-20260513102957-v877z-worker-0:10227:10383 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 108
+t-20260513102957-v877z-worker-0:10228:10384 [6] NCCL INFO [Proxy Service] Device 6 CPU core 94
+t-20260513102957-v877z-worker-0:10228:10385 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 98
+t-20260513102957-v877z-worker-0:10226:10386 [4] NCCL INFO [Proxy Service] Device 4 CPU core 94
+t-20260513102957-v877z-worker-0:10226:10387 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 96
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Check P2P Type intraNodeP2pSupport 1 directMode 0
+t-20260513102957-v877z-worker-0:10222:10388 [0] NCCL INFO [Proxy Service] Device 0 CPU core 52
+t-20260513102957-v877z-worker-0:10222:10389 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 48
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO 24 coll channels, 24 collnet channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO CC Off, workFifoBytes 1048576
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO ncclCommInitRankConfig comm 0xa1cc500 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 6f020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO ncclCommInitRankConfig comm 0xa99de10 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId 73020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO ncclCommInitRankConfig comm 0xaa777e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId 75020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO ncclCommInitRankConfig comm 0xa87b0d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId 71020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol.
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO ncclCommInitRankConfig comm 0xbd43ac0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 65040 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10226:10301 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 4 nranks 8 total 2.15 (kernels 0.20, alloc 1.02, bootstrap 0.00, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.35, rest 0.03)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead.
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO ncclCommInitRankConfig comm 0x98ac490 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 69020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10228:10295 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 6 nranks 8 total 2.21 (kernels 0.22, alloc 1.03, bootstrap 0.02, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.35, rest 0.03)
+t-20260513102957-v877z-worker-0:10227:10300 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 5 nranks 8 total 2.16 (kernels 0.20, alloc 1.02, bootstrap 0.00, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.36, rest 0.03)
+t-20260513102957-v877z-worker-0:10229:10299 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 7 nranks 8 total 2.17 (kernels 0.21, alloc 1.02, bootstrap 0.00, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.36, rest 0.03)
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO ncclCommInitRankConfig comm 0x9b15700 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 6b020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO ncclCommInitRankConfig comm 0x98a1870 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 67020 commId 0xc3d8f44253f33569 - Init COMPLETE
+t-20260513102957-v877z-worker-0:10222:10294 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 8 total 2.21 (kernels 0.21, alloc 0.99, bootstrap 0.07, allgathers 0.00, topo 0.54, graphs 0.01, connections 0.35, rest 0.03)
+t-20260513102957-v877z-worker-0:10224:10296 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 2 nranks 8 total 2.21 (kernels 0.21, alloc 0.98, bootstrap 0.08, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.36, rest 0.03)
+t-20260513102957-v877z-worker-0:10225:10298 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 3 nranks 8 total 2.17 (kernels 0.21, alloc 1.02, bootstrap 0.01, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.36, rest 0.02)
+t-20260513102957-v877z-worker-0:10223:10297 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 8 total 2.17 (kernels 0.21, alloc 1.02, bootstrap 0.01, allgathers 0.01, topo 0.54, graphs 0.01, connections 0.36, rest 0.03)
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/CUMEM
+t-20260513102957-v877z-worker-0:10228:10394 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10222:10393 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10229:10391 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10227:10396 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10225:10395 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10223:10390 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10224:10397 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+t-20260513102957-v877z-worker-0:10226:10392 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1
+{
+  "device": "cuda:0",
+  "rank": 0,
+  "world_size": 8,
+  "samples": "owt_cached_chunks:8734897",
+  "vocab_size": 50257,
+  "tokenizer_vocab_size": 50257,
+  "save_dir": "runs/lta_owt_len1024_elfopt_muon_ema_ddit768x12_8gpu_5epoch_20260513_023024",
+  "batch_size": 8,
+  "grad_accum": 8,
+  "effective_batch_size": 512,
+  "global_batch_size": 512,
+  "lr_schedule": "constant_warmup",
+  "optimizer": "muon",
+  "warmup_steps": 8531,
+  "min_lr": 0.0,
+  "weight_decay": 0.0,
+  "adamw_param_groups": "nanogpt",
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_eps": 1e-08,
+  "muon_momentum": 0.95,
+  "muon_ns_steps": 5,
+  "muon_update_scale": 1.0,
+  "ema_decay": 0.9999,
+  "ema_start_step": 0,
+  "model_type": "ddit",
+  "dual_t": true,
+  "corrupt_t_mode": "independent",
+  "corrupt_min_t": null,
+  "corrupt_max_t": null,
+  "prefix_block_prob": 0.0,
+  "prefix_block_len": 128,
+  "dirichlet_endpoint_mode": "categorical_dual_t",
+  "dirichlet_semantic_t_mode": "same",
+  "dirichlet_semantic_t_value": 0.0,
+  "categorical_wrong_from_full_vocab": true,
+  "categorical_wrong_from_batch_valid_tokens": false,
+  "mask_mixture_original_prob": 0.0,
+  "mask_mixture_lowk_prob": 0.0,
+  "mask_mixture_lowcorrupt_prob": 0.0,
+  "mask_mixture_block_prob": 0.0,
+  "mask_mixture_all_prob": 0.0,
+  "mask_mixture_lowk_clean_tokens": "1,2,4,8,16,32,64",
+  "mask_mixture_lowcorrupt_tokens": "1,2,4,8,16,32,64",
+  "mask_mixture_block_tokens": "64,128",
+  "simplex_bridge_sampler": "dirichlet",
+  "logistic_normal_sigma_min": 0.18,
+  "logistic_normal_sigma_max": 2.2,
+  "logistic_normal_tau_min": 0.65,
+  "logistic_normal_tau_max": 1.15,
+  "torch_compile": false,
+  "compile_mode": "max-autotune",
+  "state_format": "prob",
+  "target_loss": "hard_ce",
+  "meanflow_weight": 0.0,
+  "bridge_noise_init": "logistic_normal",
+  "noise_sigma": -1.0,
+  "wrap": true,
+  "wrap_mode": "stream",
+  "wrap_record_buffer_size": 200,
+  "owt_cached_chunks": true,
+  "owt_chunk_cache_dir": "/e2e-data/evad-tech-vla/wanghan58/data/small_benchmarks/langflow_2604_11748/openwebtext_lta_cached_chunks/gpt2_len1024_train_minus_100k",
+  "owt_chunk_cache_rebuild": false,
+  "owt_chunk_cache_write_batch": 4096,
+  "owt_exact_repeat_per_chunk": 0,
+  "online_chunk_shuffle": false,
+  "online_chunk_shuffle_buffer": 10000,
+  "openwebtext_split": "all",
+  "detokenizer": "auto",
+  "resolved_detokenizer": null,
+  "num_workers": 0,
+  "latest_every": 1000,
+  "resume_path": ""
+}
+t-20260513102957-v877z-worker-0:10222:10486 [0] NCCL INFO NVLS comm 0xbd43ac0 headRank 0 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10226:10487 [4] NCCL INFO NVLS comm 0xa1cc500 headRank 4 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10223:10488 [1] NCCL INFO NVLS comm 0x98a1870 headRank 1 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10224:10489 [2] NCCL INFO NVLS comm 0x98ac490 headRank 2 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10229:10490 [7] NCCL INFO NVLS comm 0xaa777e0 headRank 7 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10228:10491 [6] NCCL INFO NVLS comm 0xa99de10 headRank 6 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10227:10492 [5] NCCL INFO NVLS comm 0xa87b0d0 headRank 5 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+t-20260513102957-v877z-worker-0:10225:10493 [3] NCCL INFO NVLS comm 0x9b15700 headRank 3 nHeads 8 buffSize 1048576 nvlsPerRankSize 33554432 nvlsTotalSize 268435456
+step=50 micro_steps=400 elapsed=48.6s lr=1.195639e-05 loss_all=10.8125 acc_all=0.5523 loss_corrupt=10.8125 acc_corrupt=0.3757 corrupt_frac=0.5552 loss=10.8125 loss_recon=10.8125 loss_meanflow=0.0000 mean_model_t=0.4944 mean_corrupt_t=0.5036 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4934 init_acc_corrupt=0.4725 init_gold_top10=0.5009 init_gold_top100=0.5309
+step=100 micro_steps=800 elapsed=47.4s lr=2.367835e-05 loss_all=10.8125 acc_all=0.5656 loss_corrupt=10.8125 acc_corrupt=0.3881 corrupt_frac=0.5519 loss=10.8125 loss_recon=10.8125 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.5064 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4916 init_acc_corrupt=0.4749 init_gold_top10=0.5030 init_gold_top100=0.5319
+step=150 micro_steps=1200 elapsed=47.3s lr=3.540030e-05 loss_all=10.7976 acc_all=0.5622 loss_corrupt=10.8031 acc_corrupt=0.3899 corrupt_frac=0.5621 loss=10.8031 loss_recon=10.8031 loss_meanflow=0.0000 mean_model_t=0.5121 mean_corrupt_t=0.5055 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4925 init_acc_corrupt=0.4729 init_gold_top10=0.5021 init_gold_top100=0.5300
+step=200 micro_steps=1600 elapsed=47.3s lr=4.712226e-05 loss_all=10.7770 acc_all=0.5614 loss_corrupt=10.7891 acc_corrupt=0.3900 corrupt_frac=0.5531 loss=10.7891 loss_recon=10.7891 loss_meanflow=0.0000 mean_model_t=0.5059 mean_corrupt_t=0.5118 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4887 init_acc_corrupt=0.4772 init_gold_top10=0.5058 init_gold_top100=0.5343
+step=250 micro_steps=2000 elapsed=47.7s lr=5.884422e-05 loss_all=10.7465 acc_all=0.5551 loss_corrupt=10.7697 acc_corrupt=0.3777 corrupt_frac=0.5506 loss=10.7697 loss_recon=10.7697 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.5016 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5008 init_acc_corrupt=0.4645 init_gold_top10=0.4934 init_gold_top100=0.5233
+step=300 micro_steps=2400 elapsed=47.4s lr=7.056617e-05 loss_all=10.7068 acc_all=0.5429 loss_corrupt=10.7407 acc_corrupt=0.3755 corrupt_frac=0.5615 loss=10.7407 loss_recon=10.7407 loss_meanflow=0.0000 mean_model_t=0.4975 mean_corrupt_t=0.5060 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4925 init_acc_corrupt=0.4735 init_gold_top10=0.5019 init_gold_top100=0.5309
+step=350 micro_steps=2800 elapsed=47.2s lr=8.228813e-05 loss_all=10.6568 acc_all=0.5378 loss_corrupt=10.7063 acc_corrupt=0.3645 corrupt_frac=0.5504 loss=10.7063 loss_recon=10.7063 loss_meanflow=0.0000 mean_model_t=0.4953 mean_corrupt_t=0.5088 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4963 init_acc_corrupt=0.4697 init_gold_top10=0.4980 init_gold_top100=0.5278
+step=400 micro_steps=3200 elapsed=47.6s lr=9.401008e-05 loss_all=10.5988 acc_all=0.5316 loss_corrupt=10.6653 acc_corrupt=0.3571 corrupt_frac=0.5454 loss=10.6653 loss_recon=10.6653 loss_meanflow=0.0000 mean_model_t=0.4950 mean_corrupt_t=0.5016 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5001 init_acc_corrupt=0.4643 init_gold_top10=0.4942 init_gold_top100=0.5236
+step=450 micro_steps=3600 elapsed=47.2s lr=1.057320e-04 loss_all=10.5302 acc_all=0.5254 loss_corrupt=10.6143 acc_corrupt=0.3554 corrupt_frac=0.5518 loss=10.6143 loss_recon=10.6143 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.4974 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4970 init_acc_corrupt=0.4686 init_gold_top10=0.4972 init_gold_top100=0.5272
+step=500 micro_steps=4000 elapsed=47.2s lr=1.174540e-04 loss_all=10.4504 acc_all=0.5179 loss_corrupt=10.5520 acc_corrupt=0.3546 corrupt_frac=0.5567 loss=10.5520 loss_recon=10.5520 loss_meanflow=0.0000 mean_model_t=0.5000 mean_corrupt_t=0.5107 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4885 init_acc_corrupt=0.4787 init_gold_top10=0.5065 init_gold_top100=0.5332
+step=550 micro_steps=4400 elapsed=47.2s lr=1.291759e-04 loss_all=10.3596 acc_all=0.5119 loss_corrupt=10.4883 acc_corrupt=0.3445 corrupt_frac=0.5446 loss=10.4883 loss_recon=10.4883 loss_meanflow=0.0000 mean_model_t=0.5052 mean_corrupt_t=0.5019 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4987 init_acc_corrupt=0.4672 init_gold_top10=0.4957 init_gold_top100=0.5252
+step=600 micro_steps=4800 elapsed=47.4s lr=1.408979e-04 loss_all=10.2615 acc_all=0.5024 loss_corrupt=10.4195 acc_corrupt=0.3340 corrupt_frac=0.5407 loss=10.4195 loss_recon=10.4195 loss_meanflow=0.0000 mean_model_t=0.5048 mean_corrupt_t=0.4963 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5042 init_acc_corrupt=0.4609 init_gold_top10=0.4897 init_gold_top100=0.5213
+step=650 micro_steps=5200 elapsed=47.2s lr=1.526199e-04 loss_all=10.1555 acc_all=0.4913 loss_corrupt=10.3425 acc_corrupt=0.3260 corrupt_frac=0.5479 loss=10.3425 loss_recon=10.3425 loss_meanflow=0.0000 mean_model_t=0.4990 mean_corrupt_t=0.4946 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5096 init_acc_corrupt=0.4549 init_gold_top10=0.4844 init_gold_top100=0.5153
+step=700 micro_steps=5600 elapsed=47.2s lr=1.643418e-04 loss_all=10.0314 acc_all=0.4856 loss_corrupt=10.2464 acc_corrupt=0.3260 corrupt_frac=0.5511 loss=10.2464 loss_recon=10.2464 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4957 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5051 init_acc_corrupt=0.4599 init_gold_top10=0.4890 init_gold_top100=0.5194
+step=750 micro_steps=6000 elapsed=47.3s lr=1.760638e-04 loss_all=9.8863 acc_all=0.4836 loss_corrupt=10.1351 acc_corrupt=0.3265 corrupt_frac=0.5437 loss=10.1351 loss_recon=10.1351 loss_meanflow=0.0000 mean_model_t=0.5055 mean_corrupt_t=0.5025 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4950 init_acc_corrupt=0.4706 init_gold_top10=0.4993 init_gold_top100=0.5293
+step=800 micro_steps=6400 elapsed=47.4s lr=1.877857e-04 loss_all=9.7404 acc_all=0.4758 loss_corrupt=10.0222 acc_corrupt=0.3227 corrupt_frac=0.5505 loss=10.0222 loss_recon=10.0222 loss_meanflow=0.0000 mean_model_t=0.5058 mean_corrupt_t=0.5005 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4982 init_acc_corrupt=0.4659 init_gold_top10=0.4959 init_gold_top100=0.5264
+step=850 micro_steps=6800 elapsed=47.2s lr=1.995077e-04 loss_all=9.5817 acc_all=0.4695 loss_corrupt=9.8946 acc_corrupt=0.3218 corrupt_frac=0.5568 loss=9.8946 loss_recon=9.8946 loss_meanflow=0.0000 mean_model_t=0.5045 mean_corrupt_t=0.5047 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4955 init_acc_corrupt=0.4712 init_gold_top10=0.4988 init_gold_top100=0.5291
+step=900 micro_steps=7200 elapsed=47.2s lr=2.112296e-04 loss_all=9.4086 acc_all=0.4659 loss_corrupt=9.7618 acc_corrupt=0.3203 corrupt_frac=0.5578 loss=9.7618 loss_recon=9.7618 loss_meanflow=0.0000 mean_model_t=0.4962 mean_corrupt_t=0.5050 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4990 init_acc_corrupt=0.4659 init_gold_top10=0.4954 init_gold_top100=0.5254
+step=950 micro_steps=7600 elapsed=47.2s lr=2.229516e-04 loss_all=9.2066 acc_all=0.4715 loss_corrupt=9.6143 acc_corrupt=0.3233 corrupt_frac=0.5466 loss=9.6143 loss_recon=9.6143 loss_meanflow=0.0000 mean_model_t=0.5056 mean_corrupt_t=0.5013 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4970 init_acc_corrupt=0.4677 init_gold_top10=0.4974 init_gold_top100=0.5269
+step=1000 micro_steps=8000 elapsed=48.0s lr=2.346735e-04 loss_all=9.0073 acc_all=0.4722 loss_corrupt=9.4652 acc_corrupt=0.3230 corrupt_frac=0.5506 loss=9.4652 loss_recon=9.4652 loss_meanflow=0.0000 mean_model_t=0.4978 mean_corrupt_t=0.4901 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5091 init_acc_corrupt=0.4545 init_gold_top10=0.4849 init_gold_top100=0.5158
+step=1050 micro_steps=8400 elapsed=49.4s lr=2.463955e-04 loss_all=8.7539 acc_all=0.4867 loss_corrupt=9.2592 acc_corrupt=0.3387 corrupt_frac=0.5455 loss=9.2592 loss_recon=9.2592 loss_meanflow=0.0000 mean_model_t=0.5068 mean_corrupt_t=0.4987 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4964 init_acc_corrupt=0.4689 init_gold_top10=0.4977 init_gold_top100=0.5280
+step=1100 micro_steps=8800 elapsed=47.3s lr=2.581175e-04 loss_all=8.5291 acc_all=0.4858 loss_corrupt=9.1069 acc_corrupt=0.3332 corrupt_frac=0.5487 loss=9.1069 loss_recon=9.1069 loss_meanflow=0.0000 mean_model_t=0.5076 mean_corrupt_t=0.4923 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5131 init_acc_corrupt=0.4507 init_gold_top10=0.4807 init_gold_top100=0.5131
+step=1150 micro_steps=9200 elapsed=47.3s lr=2.698394e-04 loss_all=8.2256 acc_all=0.5009 loss_corrupt=8.8483 acc_corrupt=0.3518 corrupt_frac=0.5492 loss=8.8483 loss_recon=8.8483 loss_meanflow=0.0000 mean_model_t=0.4880 mean_corrupt_t=0.5105 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4939 init_acc_corrupt=0.4728 init_gold_top10=0.5006 init_gold_top100=0.5300
+step=1200 micro_steps=9600 elapsed=47.2s lr=2.815614e-04 loss_all=7.9819 acc_all=0.4942 loss_corrupt=8.6724 acc_corrupt=0.3445 corrupt_frac=0.5594 loss=8.6724 loss_recon=8.6724 loss_meanflow=0.0000 mean_model_t=0.5078 mean_corrupt_t=0.4905 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5065 init_acc_corrupt=0.4569 init_gold_top10=0.4872 init_gold_top100=0.5201
+step=1250 micro_steps=10000 elapsed=47.1s lr=2.932833e-04 loss_all=7.6269 acc_all=0.5052 loss_corrupt=8.4087 acc_corrupt=0.3514 corrupt_frac=0.5458 loss=8.4087 loss_recon=8.4087 loss_meanflow=0.0000 mean_model_t=0.4963 mean_corrupt_t=0.4999 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5008 init_acc_corrupt=0.4647 init_gold_top10=0.4933 init_gold_top100=0.5238
+step=1300 micro_steps=10400 elapsed=47.2s lr=3.050053e-04 loss_all=7.3064 acc_all=0.5058 loss_corrupt=8.1541 acc_corrupt=0.3535 corrupt_frac=0.5543 loss=8.1541 loss_recon=8.1541 loss_meanflow=0.0000 mean_model_t=0.4980 mean_corrupt_t=0.4979 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5009 init_acc_corrupt=0.4644 init_gold_top10=0.4933 init_gold_top100=0.5236
+step=1350 micro_steps=10800 elapsed=47.3s lr=3.167272e-04 loss_all=6.9403 acc_all=0.5105 loss_corrupt=7.8690 acc_corrupt=0.3577 corrupt_frac=0.5550 loss=7.8690 loss_recon=7.8690 loss_meanflow=0.0000 mean_model_t=0.4935 mean_corrupt_t=0.5024 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4980 init_acc_corrupt=0.4682 init_gold_top10=0.4961 init_gold_top100=0.5263
+step=1400 micro_steps=11200 elapsed=47.3s lr=3.284492e-04 loss_all=6.6098 acc_all=0.5081 loss_corrupt=7.6173 acc_corrupt=0.3549 corrupt_frac=0.5601 loss=7.6173 loss_recon=7.6173 loss_meanflow=0.0000 mean_model_t=0.4999 mean_corrupt_t=0.4967 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5063 init_acc_corrupt=0.4582 init_gold_top10=0.4878 init_gold_top100=0.5183
+step=1450 micro_steps=11600 elapsed=47.5s lr=3.401711e-04 loss_all=6.2064 acc_all=0.5188 loss_corrupt=7.2967 acc_corrupt=0.3645 corrupt_frac=0.5511 loss=7.2967 loss_recon=7.2967 loss_meanflow=0.0000 mean_model_t=0.4940 mean_corrupt_t=0.5044 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4966 init_acc_corrupt=0.4682 init_gold_top10=0.4978 init_gold_top100=0.5272
+step=1500 micro_steps=12000 elapsed=47.1s lr=3.518931e-04 loss_all=5.8854 acc_all=0.5207 loss_corrupt=7.0334 acc_corrupt=0.3677 corrupt_frac=0.5555 loss=7.0334 loss_recon=7.0334 loss_meanflow=0.0000 mean_model_t=0.4951 mean_corrupt_t=0.5001 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4963 init_acc_corrupt=0.4692 init_gold_top10=0.4981 init_gold_top100=0.5273
+step=1550 micro_steps=12400 elapsed=47.5s lr=3.636151e-04 loss_all=5.6304 acc_all=0.5225 loss_corrupt=6.8543 acc_corrupt=0.3653 corrupt_frac=0.5503 loss=6.8543 loss_recon=6.8543 loss_meanflow=0.0000 mean_model_t=0.4989 mean_corrupt_t=0.5031 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4987 init_acc_corrupt=0.4659 init_gold_top10=0.4958 init_gold_top100=0.5247
+step=1600 micro_steps=12800 elapsed=47.3s lr=3.753370e-04 loss_all=5.4321 acc_all=0.5248 loss_corrupt=6.7023 acc_corrupt=0.3652 corrupt_frac=0.5462 loss=6.7023 loss_recon=6.7023 loss_meanflow=0.0000 mean_model_t=0.5041 mean_corrupt_t=0.4979 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.5059 init_acc_corrupt=0.4579 init_gold_top10=0.4881 init_gold_top100=0.5198
+step=1650 micro_steps=13200 elapsed=47.2s lr=3.870590e-04 loss_all=5.2358 acc_all=0.5320 loss_corrupt=6.4720 acc_corrupt=0.3799 corrupt_frac=0.5497 loss=6.4720 loss_recon=6.4720 loss_meanflow=0.0000 mean_model_t=0.5084 mean_corrupt_t=0.5027 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4927 init_acc_corrupt=0.4734 init_gold_top10=0.5017 init_gold_top100=0.5309
+step=1700 micro_steps=13600 elapsed=47.2s lr=3.987809e-04 loss_all=5.0956 acc_all=0.5353 loss_corrupt=6.3531 acc_corrupt=0.3809 corrupt_frac=0.5492 loss=6.3531 loss_recon=6.3531 loss_meanflow=0.0000 mean_model_t=0.4959 mean_corrupt_t=0.5046 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4967 init_acc_corrupt=0.4688 init_gold_top10=0.4975 init_gold_top100=0.5289
+step=1750 micro_steps=14000 elapsed=47.1s lr=4.105029e-04 loss_all=4.9791 acc_all=0.5365 loss_corrupt=6.2240 acc_corrupt=0.3836 corrupt_frac=0.5514 loss=6.2240 loss_recon=6.2240 loss_meanflow=0.0000 mean_model_t=0.4896 mean_corrupt_t=0.5021 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4968 init_acc_corrupt=0.4682 init_gold_top10=0.4975 init_gold_top100=0.5266
+step=1800 micro_steps=14400 elapsed=47.2s lr=4.222248e-04 loss_all=4.8119 acc_all=0.5441 loss_corrupt=6.0439 acc_corrupt=0.3934 corrupt_frac=0.5495 loss=6.0439 loss_recon=6.0439 loss_meanflow=0.0000 mean_model_t=0.5011 mean_corrupt_t=0.5048 mean_loss_t_weight=1.0000 prior_center_loss_beta=0.0000 wrong_frac=0.4894 init_acc_corrupt=0.4760 init_gold_top10=0.5053 init_gold_top100=0.5336

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/infer_step_0010000_state_fromstate_t1p45.log

@@ -0,0 +1,8 @@
+[watch-infer] 2026-05-20_23:01:09 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0010000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt step=10000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt", "step": 10000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 8.05236863767061, "nll_per_token": 2.085966288854589, "tokens": 7891, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 7.63913892725258, "nll_per_token": 2.0332848909696546, "tokens": 7911, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 1.8837622158001541, "unique_tokens": 274, "token_count": 8192, "distinct_1": 0.033447265625, "distinct_2": 0.09547244094488189, "top_token_mass": 0.3001708984375}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0010000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-20_23:01:31 done step_0010000

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/infer_step_0020000_state_fromstate_t1p45.log

@@ -0,0 +1,8 @@
+[watch-infer] 2026-05-21_01:03:31 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0020000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt step=20000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt", "step": 20000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 43.29514508349064, "nll_per_token": 3.768040505939089, "tokens": 7277, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 44.71103365440866, "nll_per_token": 3.8002203090662934, "tokens": 7167, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.680576719579072, "unique_tokens": 989, "token_count": 8192, "distinct_1": 0.1207275390625, "distinct_2": 0.436884842519685, "top_token_mass": 0.0694580078125}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0020000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_01:04:00 done step_0020000

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/infer_step_0030000_state_fromstate_t1p45.log

@@ -0,0 +1,8 @@
+[watch-infer] 2026-05-21_03:42:01 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0030000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt step=30000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt", "step": 30000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 32.519971378109084, "nll_per_token": 3.4818544045472755, "tokens": 7020, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 34.09647656081684, "nll_per_token": 3.529194052288641, "tokens": 6748, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.1030114352732854, "unique_tokens": 881, "token_count": 8192, "distinct_1": 0.1075439453125, "distinct_2": 0.3661417322834646, "top_token_mass": 0.100341796875}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0030000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_03:42:30 done step_0030000

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/infer_step_0040000_state_fromstate_t1p45.log

@@ -0,0 +1,8 @@
+[watch-infer] 2026-05-21_06:20:32 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0040000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt step=40000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt", "step": 40000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 35.01775070788609, "nll_per_token": 3.555855096008144, "tokens": 7383, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 35.25691068369559, "nll_per_token": 3.5626615578078398, "tokens": 7376, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.4672587399947643, "unique_tokens": 1033, "token_count": 8192, "distinct_1": 0.1260986328125, "distinct_2": 0.42913385826771655, "top_token_mass": 0.06982421875}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0040000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_06:21:00 done step_0040000

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/infer_step_0050000_state_fromstate_t1p45.log

@@ -0,0 +1,8 @@
+[watch-infer] 2026-05-21_08:58:02 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0050000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt step=50000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt", "step": 50000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 39.13239939115062, "nll_per_token": 3.666950752797587, "tokens": 5574, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 38.076086050208254, "nll_per_token": 3.6395864223457512, "tokens": 5543, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 2.8907815168852227, "unique_tokens": 899, "token_count": 8192, "distinct_1": 0.1097412109375, "distinct_2": 0.3246801181102362, "top_token_mass": 0.1265869140625}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0050000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_08:58:34 done step_0050000

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/processed_every10k_state_fromstate_t1p45.txt

@@ -0,0 +1,5 @@
+runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt
+runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt
+runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt
+runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt
+runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt

LTA_openwebtext_dualt/logs/lm1b_v8192_len128_infer_watch/watch_every10k_state_t1p45.nohup.log

@@ -0,0 +1,46 @@
+[watch-infer] run_dir=runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739
+[watch-infer] out_base=docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45
+[watch-infer] processed_file=logs/lm1b_v8192_len128_infer_watch/processed_every10k_state_fromstate_t1p45.txt
+[watch-infer] decode=dual_line_resample anchor=state final_from=state final_sample=argmax temp=1.45 steps=128 cmax=1024 n=64
+[watch-infer] 2026-05-20_22:59:09 no step_*.pt yet
+[watch-infer] 2026-05-20_23:00:09 no step_*.pt yet
+[watch-infer] 2026-05-20_23:01:09 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0010000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt step=10000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0010000.pt", "step": 10000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 8.05236863767061, "nll_per_token": 2.085966288854589, "tokens": 7891, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 7.63913892725258, "nll_per_token": 2.0332848909696546, "tokens": 7911, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 1.8837622158001541, "unique_tokens": 274, "token_count": 8192, "distinct_1": 0.033447265625, "distinct_2": 0.09547244094488189, "top_token_mass": 0.3001708984375}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0010000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-20_23:01:31 done step_0010000
+[watch-infer] 2026-05-21_01:03:31 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0020000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt step=20000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0020000.pt", "step": 20000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 43.29514508349064, "nll_per_token": 3.768040505939089, "tokens": 7277, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 44.71103365440866, "nll_per_token": 3.8002203090662934, "tokens": 7167, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.680576719579072, "unique_tokens": 989, "token_count": 8192, "distinct_1": 0.1207275390625, "distinct_2": 0.436884842519685, "top_token_mass": 0.0694580078125}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0020000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_01:04:00 done step_0020000
+[watch-infer] 2026-05-21_03:42:01 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0030000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt step=30000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0030000.pt", "step": 30000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 32.519971378109084, "nll_per_token": 3.4818544045472755, "tokens": 7020, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 34.09647656081684, "nll_per_token": 3.529194052288641, "tokens": 6748, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.1030114352732854, "unique_tokens": 881, "token_count": 8192, "distinct_1": 0.1075439453125, "distinct_2": 0.3661417322834646, "top_token_mass": 0.100341796875}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0030000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_03:42:30 done step_0030000
+[watch-infer] 2026-05-21_06:20:32 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0040000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt step=40000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0040000.pt", "step": 40000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 35.01775070788609, "nll_per_token": 3.555855096008144, "tokens": 7383, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 35.25691068369559, "nll_per_token": 3.5626615578078398, "tokens": 7376, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 3.4672587399947643, "unique_tokens": 1033, "token_count": 8192, "distinct_1": 0.1260986328125, "distinct_2": 0.42913385826771655, "top_token_mass": 0.06982421875}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0040000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_06:21:00 done step_0040000
+[watch-infer] 2026-05-21_08:58:02 infer runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt -> docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0050000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[ckpt] runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt step=50000
+[decode-base] n=64 max_len=128 steps=128 model_t=flow
+[decode-time] schedule=linear s=[0.0,0.25] gumbel=(2.2,0.8) force_final=True t0=0.000000 t_mid=0.500000 t_end=1.000000 dt_mean=0.007812 dt_max=0.007812
+[decode] temp=1.45 final=state rule=dual_line_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform decode_freq_penalty=0/0/0-1^1 final_sample=argmax/1/k64/p0.95 freq_penalty=0/0/0 start_t=0 start_init=noise time_path=0.0000->1.0000 generated 64/64
+[summary] {"type": "summary", "checkpoint": "runs/lta_lm1b_compact_gpt2bpe_v8192_len128_repro_mask1_uniformt_fp32_ddit768x12_gbs512_4gpu_1m_20260520_221739/step_0050000.pt", "step": 50000, "decode": {"steps": 128, "model_t_mode": "flow", "decode_time_schedule": "linear", "decode_s_min_frac": 0.0, "decode_s_max_frac": 0.25, "decode_force_final_t": true, "decode_time_grid": [0.0, 0.0078125, 0.015625, 0.0234375, 0.03125, 0.0390625, 0.046875, 0.0546875, 0.0625, 0.0703125, 0.078125, 0.0859375, 0.09375, 0.1015625, 0.109375, 0.1171875, 0.125, 0.1328125, 0.140625, 0.1484375, 0.15625, 0.1640625, 0.171875, 0.1796875, 0.1875, 0.1953125, 0.203125, 0.2109375, 0.21875, 0.2265625, 0.234375, 0.2421875, 0.25, 0.2578125, 0.265625, 0.2734375, 0.28125, 0.2890625, 0.296875, 0.3046875, 0.3125, 0.3203125, 0.328125, 0.3359375, 0.34375, 0.3515625, 0.359375, 0.3671875, 0.375, 0.3828125, 0.390625, 0.3984375, 0.40625, 0.4140625, 0.421875, 0.4296875, 0.4375, 0.4453125, 0.453125, 0.4609375, 0.46875, 0.4765625, 0.484375, 0.4921875, 0.5, 0.5078125, 0.515625, 0.5234375, 0.53125, 0.5390625, 0.546875, 0.5546875, 0.5625, 0.5703125, 0.578125, 0.5859375, 0.59375, 0.6015625, 0.609375, 0.6171875, 0.625, 0.6328125, 0.640625, 0.6484375, 0.65625, 0.6640625, 0.671875, 0.6796875, 0.6875, 0.6953125, 0.703125, 0.7109375, 0.71875, 0.7265625, 0.734375, 0.7421875, 0.75, 0.7578125, 0.765625, 0.7734375, 0.78125, 0.7890625, 0.796875, 0.8046875, 0.8125, 0.8203125, 0.828125, 0.8359375, 0.84375, 0.8515625, 0.859375, 0.8671875, 0.875, 0.8828125, 0.890625, 0.8984375, 0.90625, 0.9140625, 0.921875, 0.9296875, 0.9375, 0.9453125, 0.953125, 0.9609375, 0.96875, 0.9765625, 0.984375, 0.9921875, 1.0], "decode_rule": "dual_line_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "decode_freq_penalty_alpha": 0.0, "decode_freq_penalty_beta": 0.0, "decode_freq_penalty_floor": 0.0, "decode_freq_penalty_start": 0.0, "decode_freq_penalty_end": 1.0, "decode_freq_penalty_power": 1.0, "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 1024.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "final_sample_mode": "argmax", "final_sample_temp": 1.0, "final_top_k": 64, "final_top_p": 0.95, "final_freq_penalty_alpha": 0.0, "final_freq_penalty_beta": 0.0, "final_freq_penalty_floor": 0.0, "lock_bos": false, "n_samples": 64, "seed": 20260520}, "raw_genppl": {"ppl": 39.13239939115062, "nll_per_token": 3.666950752797587, "tokens": 5574, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 38.076086050208254, "nll_per_token": 3.6395864223457512, "tokens": 5543, "kept_samples": 64, "total_samples": 64, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 2.8907815168852227, "unique_tokens": 899, "token_count": 8192, "distinct_1": 0.1097412109375, "distinct_2": 0.3246801181102362, "top_token_mass": 0.1265869140625}}
+[done] docs/lta_samples/metrics_20260520/lm1b_v8192_len128_every10k_dualline_state_fromstate_t1p45/step_0050000/decode_dual_line_resample_state_fromstate_t1p45_c1024_steps128.jsonl
+[watch-infer] 2026-05-21_08:58:34 done step_0050000

LTA_openwebtext_dualt/logs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/maskfloor_gamma2.dirichlet_resample.eval.log

@@ -0,0 +1,8 @@
+[ckpt] runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_maskfloor_gamma2/latest.pt step=1000
+[decode-base] n=16 max_len=1024 steps=64 model_t=flow
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 4/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 8/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 12/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 16/16
+[summary] {"type": "summary", "checkpoint": "runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_maskfloor_gamma2/latest.pt", "step": 1000, "decode": {"steps": 64, "model_t_mode": "flow", "decode_rule": "dirichlet_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 64.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "lock_bos": false, "n_samples": 16, "seed": 20260503}, "raw_genppl": {"ppl": 205.89546320407374, "nll_per_token": 5.327368579813691, "tokens": 11862, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 186.53793153722665, "nll_per_token": 5.228634604662277, "tokens": 11770, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 2.3015467149505735, "unique_tokens": 2801, "token_count": 16384, "distinct_1": 0.17095947265625, "distinct_2": 0.34836265884652984, "top_token_mass": 0.55841064453125}}
+[done] docs/lta_samples/metrics_20260513/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/maskfloor_gamma2_dirichlet_resample.jsonl

LTA_openwebtext_dualt/logs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/old_drop_lowt_ce.flowmap.eval.log

@@ -0,0 +1,8 @@
+[ckpt] runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_old_drop_lowt_ce/latest.pt step=1000
+[decode-base] n=16 max_len=1024 steps=64 model_t=flow
+[decode] temp=1.45 final=state rule=flowmap support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 4/16
+[decode] temp=1.45 final=state rule=flowmap support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 8/16
+[decode] temp=1.45 final=state rule=flowmap support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 12/16
+[decode] temp=1.45 final=state rule=flowmap support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 16/16
+[summary] {"type": "summary", "checkpoint": "runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_old_drop_lowt_ce/latest.pt", "step": 1000, "decode": {"steps": 64, "model_t_mode": "flow", "decode_rule": "flowmap", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 64.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "lock_bos": false, "n_samples": 16, "seed": 20260503}, "raw_genppl": {"ppl": 18.545462901559578, "nll_per_token": 2.9202251716200354, "tokens": 406, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 12.73431918253559, "nll_per_token": 2.5443006466596554, "tokens": 390, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 0.04249307349759278, "unique_tokens": 18, "token_count": 16384, "distinct_1": 0.0010986328125, "distinct_2": 0.002993646138807429, "top_token_mass": 0.9935302734375}}
+[done] docs/lta_samples/metrics_20260513/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/old_drop_lowt_ce_flowmap.jsonl

LTA_openwebtext_dualt/logs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/old_linear_nomaskfloor.dirichlet_resample.eval.log

@@ -0,0 +1,8 @@
+[ckpt] runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_old_linear_nomaskfloor/latest.pt step=1000
+[decode-base] n=16 max_len=1024 steps=64 model_t=flow
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 4/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 8/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 12/16
+[decode] temp=1.45 final=state rule=dirichlet_resample support=1 semantic=1 anchor=state cfg=0/1@0:uniform start_t=0 start_init=noise generated 16/16
+[summary] {"type": "summary", "checkpoint": "runs/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234_old_linear_nomaskfloor/latest.pt", "step": 1000, "decode": {"steps": 64, "model_t_mode": "flow", "decode_rule": "dirichlet_resample", "support_power": 1.0, "semantic_power": 1.0, "anchor_mode": "state", "cfg_scale": 0.0, "cfg_power": 1.0, "cfg_start": 0.0, "cfg_prior": "uniform", "start_t": 0.0, "start_init": "noise", "noise_init": "dirichlet", "noise_sigma": -1.0, "dirichlet_concentration": 1.0, "concentration_min": 1.0, "concentration_max": 64.0, "target_prob": 1.0, "endpoint_temp": 1.45, "final_from": "state", "lock_bos": false, "n_samples": 16, "seed": 20260503}, "raw_genppl": {"ppl": 29.662822677002694, "nll_per_token": 3.389894499983462, "tokens": 15117, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "stripped_genppl": {"ppl": 26.313876829298174, "nll_per_token": 3.2700964361277274, "tokens": 14711, "kept_samples": 16, "total_samples": 16, "empty_rate": 0.0, "skipped_samples": 0}, "diversity": {"sample_entropy": 2.398402485948885, "unique_tokens": 704, "token_count": 16384, "distinct_1": 0.04296875, "distinct_2": 0.12243401759530792, "top_token_mass": 0.28253173828125}}
+[done] docs/lta_samples/metrics_20260513/lowt_hard_vs_easy_c1024_fullvocab_ctx1024_384x6_20260513_202234/old_linear_nomaskfloor_dirichlet_resample.jsonl

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/__init__.py

@@ -0,0 +1,33 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+import logging
+
+__version__ = '0.3.8'
+
+
+class DistlibException(Exception):
+    pass
+
+
+try:
+    from logging import NullHandler
+except ImportError:  # pragma: no cover
+
+    class NullHandler(logging.Handler):
+
+        def handle(self, record):
+            pass
+
+        def emit(self, record):
+            pass
+
+        def createLock(self):
+            self.lock = None
+
+
+logger = logging.getLogger(__name__)
+logger.addHandler(NullHandler())

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/compat.py

@@ -0,0 +1,1138 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2013-2017 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+from __future__ import absolute_import
+
+import os
+import re
+import shutil
+import sys
+
+try:
+    import ssl
+except ImportError:  # pragma: no cover
+    ssl = None
+
+if sys.version_info[0] < 3:  # pragma: no cover
+    from StringIO import StringIO
+    string_types = basestring,
+    text_type = unicode
+    from types import FileType as file_type
+    import __builtin__ as builtins
+    import ConfigParser as configparser
+    from urlparse import urlparse, urlunparse, urljoin, urlsplit, urlunsplit
+    from urllib import (urlretrieve, quote as _quote, unquote, url2pathname,
+                        pathname2url, ContentTooShortError, splittype)
+
+    def quote(s):
+        if isinstance(s, unicode):
+            s = s.encode('utf-8')
+        return _quote(s)
+
+    import urllib2
+    from urllib2 import (Request, urlopen, URLError, HTTPError,
+                         HTTPBasicAuthHandler, HTTPPasswordMgr, HTTPHandler,
+                         HTTPRedirectHandler, build_opener)
+    if ssl:
+        from urllib2 import HTTPSHandler
+    import httplib
+    import xmlrpclib
+    import Queue as queue
+    from HTMLParser import HTMLParser
+    import htmlentitydefs
+    raw_input = raw_input
+    from itertools import ifilter as filter
+    from itertools import ifilterfalse as filterfalse
+
+    # Leaving this around for now, in case it needs resurrecting in some way
+    # _userprog = None
+    # def splituser(host):
+    # """splituser('user[:passwd]@host[:port]') --> 'user[:passwd]', 'host[:port]'."""
+    # global _userprog
+    # if _userprog is None:
+    # import re
+    # _userprog = re.compile('^(.*)@(.*)$')
+
+    # match = _userprog.match(host)
+    # if match: return match.group(1, 2)
+    # return None, host
+
+else:  # pragma: no cover
+    from io import StringIO
+    string_types = str,
+    text_type = str
+    from io import TextIOWrapper as file_type
+    import builtins
+    import configparser
+    from urllib.parse import (urlparse, urlunparse, urljoin, quote, unquote,
+                              urlsplit, urlunsplit, splittype)
+    from urllib.request import (urlopen, urlretrieve, Request, url2pathname,
+                                pathname2url, HTTPBasicAuthHandler,
+                                HTTPPasswordMgr, HTTPHandler,
+                                HTTPRedirectHandler, build_opener)
+    if ssl:
+        from urllib.request import HTTPSHandler
+    from urllib.error import HTTPError, URLError, ContentTooShortError
+    import http.client as httplib
+    import urllib.request as urllib2
+    import xmlrpc.client as xmlrpclib
+    import queue
+    from html.parser import HTMLParser
+    import html.entities as htmlentitydefs
+    raw_input = input
+    from itertools import filterfalse
+    filter = filter
+
+try:
+    from ssl import match_hostname, CertificateError
+except ImportError:  # pragma: no cover
+
+    class CertificateError(ValueError):
+        pass
+
+    def _dnsname_match(dn, hostname, max_wildcards=1):
+        """Matching according to RFC 6125, section 6.4.3
+
+        http://tools.ietf.org/html/rfc6125#section-6.4.3
+        """
+        pats = []
+        if not dn:
+            return False
+
+        parts = dn.split('.')
+        leftmost, remainder = parts[0], parts[1:]
+
+        wildcards = leftmost.count('*')
+        if wildcards > max_wildcards:
+            # Issue #17980: avoid denials of service by refusing more
+            # than one wildcard per fragment.  A survey of established
+            # policy among SSL implementations showed it to be a
+            # reasonable choice.
+            raise CertificateError(
+                "too many wildcards in certificate DNS name: " + repr(dn))
+
+        # speed up common case w/o wildcards
+        if not wildcards:
+            return dn.lower() == hostname.lower()
+
+        # RFC 6125, section 6.4.3, subitem 1.
+        # The client SHOULD NOT attempt to match a presented identifier in which
+        # the wildcard character comprises a label other than the left-most label.
+        if leftmost == '*':
+            # When '*' is a fragment by itself, it matches a non-empty dotless
+            # fragment.
+            pats.append('[^.]+')
+        elif leftmost.startswith('xn--') or hostname.startswith('xn--'):
+            # RFC 6125, section 6.4.3, subitem 3.
+            # The client SHOULD NOT attempt to match a presented identifier
+            # where the wildcard character is embedded within an A-label or
+            # U-label of an internationalized domain name.
+            pats.append(re.escape(leftmost))
+        else:
+            # Otherwise, '*' matches any dotless string, e.g. www*
+            pats.append(re.escape(leftmost).replace(r'\*', '[^.]*'))
+
+        # add the remaining fragments, ignore any wildcards
+        for frag in remainder:
+            pats.append(re.escape(frag))
+
+        pat = re.compile(r'\A' + r'\.'.join(pats) + r'\Z', re.IGNORECASE)
+        return pat.match(hostname)
+
+    def match_hostname(cert, hostname):
+        """Verify that *cert* (in decoded format as returned by
+        SSLSocket.getpeercert()) matches the *hostname*.  RFC 2818 and RFC 6125
+        rules are followed, but IP addresses are not accepted for *hostname*.
+
+        CertificateError is raised on failure. On success, the function
+        returns nothing.
+        """
+        if not cert:
+            raise ValueError("empty or no certificate, match_hostname needs a "
+                             "SSL socket or SSL context with either "
+                             "CERT_OPTIONAL or CERT_REQUIRED")
+        dnsnames = []
+        san = cert.get('subjectAltName', ())
+        for key, value in san:
+            if key == 'DNS':
+                if _dnsname_match(value, hostname):
+                    return
+                dnsnames.append(value)
+        if not dnsnames:
+            # The subject is only checked when there is no dNSName entry
+            # in subjectAltName
+            for sub in cert.get('subject', ()):
+                for key, value in sub:
+                    # XXX according to RFC 2818, the most specific Common Name
+                    # must be used.
+                    if key == 'commonName':
+                        if _dnsname_match(value, hostname):
+                            return
+                        dnsnames.append(value)
+        if len(dnsnames) > 1:
+            raise CertificateError("hostname %r "
+                                   "doesn't match either of %s" %
+                                   (hostname, ', '.join(map(repr, dnsnames))))
+        elif len(dnsnames) == 1:
+            raise CertificateError("hostname %r "
+                                   "doesn't match %r" %
+                                   (hostname, dnsnames[0]))
+        else:
+            raise CertificateError("no appropriate commonName or "
+                                   "subjectAltName fields were found")
+
+
+try:
+    from types import SimpleNamespace as Container
+except ImportError:  # pragma: no cover
+
+    class Container(object):
+        """
+        A generic container for when multiple values need to be returned
+        """
+
+        def __init__(self, **kwargs):
+            self.__dict__.update(kwargs)
+
+
+try:
+    from shutil import which
+except ImportError:  # pragma: no cover
+    # Implementation from Python 3.3
+    def which(cmd, mode=os.F_OK | os.X_OK, path=None):
+        """Given a command, mode, and a PATH string, return the path which
+        conforms to the given mode on the PATH, or None if there is no such
+        file.
+
+        `mode` defaults to os.F_OK | os.X_OK. `path` defaults to the result
+        of os.environ.get("PATH"), or can be overridden with a custom search
+        path.
+
+        """
+
+        # Check that a given file can be accessed with the correct mode.
+        # Additionally check that `file` is not a directory, as on Windows
+        # directories pass the os.access check.
+        def _access_check(fn, mode):
+            return (os.path.exists(fn) and os.access(fn, mode)
+                    and not os.path.isdir(fn))
+
+        # If we're given a path with a directory part, look it up directly rather
+        # than referring to PATH directories. This includes checking relative to the
+        # current directory, e.g. ./script
+        if os.path.dirname(cmd):
+            if _access_check(cmd, mode):
+                return cmd
+            return None
+
+        if path is None:
+            path = os.environ.get("PATH", os.defpath)
+        if not path:
+            return None
+        path = path.split(os.pathsep)
+
+        if sys.platform == "win32":
+            # The current directory takes precedence on Windows.
+            if os.curdir not in path:
+                path.insert(0, os.curdir)
+
+            # PATHEXT is necessary to check on Windows.
+            pathext = os.environ.get("PATHEXT", "").split(os.pathsep)
+            # See if the given file matches any of the expected path extensions.
+            # This will allow us to short circuit when given "python.exe".
+            # If it does match, only test that one, otherwise we have to try
+            # others.
+            if any(cmd.lower().endswith(ext.lower()) for ext in pathext):
+                files = [cmd]
+            else:
+                files = [cmd + ext for ext in pathext]
+        else:
+            # On other platforms you don't have things like PATHEXT to tell you
+            # what file suffixes are executable, so just pass on cmd as-is.
+            files = [cmd]
+
+        seen = set()
+        for dir in path:
+            normdir = os.path.normcase(dir)
+            if normdir not in seen:
+                seen.add(normdir)
+                for thefile in files:
+                    name = os.path.join(dir, thefile)
+                    if _access_check(name, mode):
+                        return name
+        return None
+
+
+# ZipFile is a context manager in 2.7, but not in 2.6
+
+from zipfile import ZipFile as BaseZipFile
+
+if hasattr(BaseZipFile, '__enter__'):  # pragma: no cover
+    ZipFile = BaseZipFile
+else:  # pragma: no cover
+    from zipfile import ZipExtFile as BaseZipExtFile
+
+    class ZipExtFile(BaseZipExtFile):
+
+        def __init__(self, base):
+            self.__dict__.update(base.__dict__)
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *exc_info):
+            self.close()
+            # return None, so if an exception occurred, it will propagate
+
+    class ZipFile(BaseZipFile):
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *exc_info):
+            self.close()
+            # return None, so if an exception occurred, it will propagate
+
+        def open(self, *args, **kwargs):
+            base = BaseZipFile.open(self, *args, **kwargs)
+            return ZipExtFile(base)
+
+
+try:
+    from platform import python_implementation
+except ImportError:  # pragma: no cover
+
+    def python_implementation():
+        """Return a string identifying the Python implementation."""
+        if 'PyPy' in sys.version:
+            return 'PyPy'
+        if os.name == 'java':
+            return 'Jython'
+        if sys.version.startswith('IronPython'):
+            return 'IronPython'
+        return 'CPython'
+
+
+import sysconfig
+
+try:
+    callable = callable
+except NameError:  # pragma: no cover
+    from collections.abc import Callable
+
+    def callable(obj):
+        return isinstance(obj, Callable)
+
+
+try:
+    fsencode = os.fsencode
+    fsdecode = os.fsdecode
+except AttributeError:  # pragma: no cover
+    # Issue #99: on some systems (e.g. containerised),
+    # sys.getfilesystemencoding() returns None, and we need a real value,
+    # so fall back to utf-8. From the CPython 2.7 docs relating to Unix and
+    # sys.getfilesystemencoding(): the return value is "the user’s preference
+    # according to the result of nl_langinfo(CODESET), or None if the
+    # nl_langinfo(CODESET) failed."
+    _fsencoding = sys.getfilesystemencoding() or 'utf-8'
+    if _fsencoding == 'mbcs':
+        _fserrors = 'strict'
+    else:
+        _fserrors = 'surrogateescape'
+
+    def fsencode(filename):
+        if isinstance(filename, bytes):
+            return filename
+        elif isinstance(filename, text_type):
+            return filename.encode(_fsencoding, _fserrors)
+        else:
+            raise TypeError("expect bytes or str, not %s" %
+                            type(filename).__name__)
+
+    def fsdecode(filename):
+        if isinstance(filename, text_type):
+            return filename
+        elif isinstance(filename, bytes):
+            return filename.decode(_fsencoding, _fserrors)
+        else:
+            raise TypeError("expect bytes or str, not %s" %
+                            type(filename).__name__)
+
+
+try:
+    from tokenize import detect_encoding
+except ImportError:  # pragma: no cover
+    from codecs import BOM_UTF8, lookup
+
+    cookie_re = re.compile(r"coding[:=]\s*([-\w.]+)")
+
+    def _get_normal_name(orig_enc):
+        """Imitates get_normal_name in tokenizer.c."""
+        # Only care about the first 12 characters.
+        enc = orig_enc[:12].lower().replace("_", "-")
+        if enc == "utf-8" or enc.startswith("utf-8-"):
+            return "utf-8"
+        if enc in ("latin-1", "iso-8859-1", "iso-latin-1") or \
+           enc.startswith(("latin-1-", "iso-8859-1-", "iso-latin-1-")):
+            return "iso-8859-1"
+        return orig_enc
+
+    def detect_encoding(readline):
+        """
+        The detect_encoding() function is used to detect the encoding that should
+        be used to decode a Python source file.  It requires one argument, readline,
+        in the same way as the tokenize() generator.
+
+        It will call readline a maximum of twice, and return the encoding used
+        (as a string) and a list of any lines (left as bytes) it has read in.
+
+        It detects the encoding from the presence of a utf-8 bom or an encoding
+        cookie as specified in pep-0263.  If both a bom and a cookie are present,
+        but disagree, a SyntaxError will be raised.  If the encoding cookie is an
+        invalid charset, raise a SyntaxError.  Note that if a utf-8 bom is found,
+        'utf-8-sig' is returned.
+
+        If no encoding is specified, then the default of 'utf-8' will be returned.
+        """
+        try:
+            filename = readline.__self__.name
+        except AttributeError:
+            filename = None
+        bom_found = False
+        encoding = None
+        default = 'utf-8'
+
+        def read_or_stop():
+            try:
+                return readline()
+            except StopIteration:
+                return b''
+
+        def find_cookie(line):
+            try:
+                # Decode as UTF-8. Either the line is an encoding declaration,
+                # in which case it should be pure ASCII, or it must be UTF-8
+                # per default encoding.
+                line_string = line.decode('utf-8')
+            except UnicodeDecodeError:
+                msg = "invalid or missing encoding declaration"
+                if filename is not None:
+                    msg = '{} for {!r}'.format(msg, filename)
+                raise SyntaxError(msg)
+
+            matches = cookie_re.findall(line_string)
+            if not matches:
+                return None
+            encoding = _get_normal_name(matches[0])
+            try:
+                codec = lookup(encoding)
+            except LookupError:
+                # This behaviour mimics the Python interpreter
+                if filename is None:
+                    msg = "unknown encoding: " + encoding
+                else:
+                    msg = "unknown encoding for {!r}: {}".format(
+                        filename, encoding)
+                raise SyntaxError(msg)
+
+            if bom_found:
+                if codec.name != 'utf-8':
+                    # This behaviour mimics the Python interpreter
+                    if filename is None:
+                        msg = 'encoding problem: utf-8'
+                    else:
+                        msg = 'encoding problem for {!r}: utf-8'.format(
+                            filename)
+                    raise SyntaxError(msg)
+                encoding += '-sig'
+            return encoding
+
+        first = read_or_stop()
+        if first.startswith(BOM_UTF8):
+            bom_found = True
+            first = first[3:]
+            default = 'utf-8-sig'
+        if not first:
+            return default, []
+
+        encoding = find_cookie(first)
+        if encoding:
+            return encoding, [first]
+
+        second = read_or_stop()
+        if not second:
+            return default, [first]
+
+        encoding = find_cookie(second)
+        if encoding:
+            return encoding, [first, second]
+
+        return default, [first, second]
+
+
+# For converting & <-> &amp; etc.
+try:
+    from html import escape
+except ImportError:
+    from cgi import escape
+if sys.version_info[:2] < (3, 4):
+    unescape = HTMLParser().unescape
+else:
+    from html import unescape
+
+try:
+    from collections import ChainMap
+except ImportError:  # pragma: no cover
+    from collections import MutableMapping
+
+    try:
+        from reprlib import recursive_repr as _recursive_repr
+    except ImportError:
+
+        def _recursive_repr(fillvalue='...'):
+            '''
+            Decorator to make a repr function return fillvalue for a recursive
+            call
+            '''
+
+            def decorating_function(user_function):
+                repr_running = set()
+
+                def wrapper(self):
+                    key = id(self), get_ident()
+                    if key in repr_running:
+                        return fillvalue
+                    repr_running.add(key)
+                    try:
+                        result = user_function(self)
+                    finally:
+                        repr_running.discard(key)
+                    return result
+
+                # Can't use functools.wraps() here because of bootstrap issues
+                wrapper.__module__ = getattr(user_function, '__module__')
+                wrapper.__doc__ = getattr(user_function, '__doc__')
+                wrapper.__name__ = getattr(user_function, '__name__')
+                wrapper.__annotations__ = getattr(user_function,
+                                                  '__annotations__', {})
+                return wrapper
+
+            return decorating_function
+
+    class ChainMap(MutableMapping):
+        '''
+        A ChainMap groups multiple dicts (or other mappings) together
+        to create a single, updateable view.
+
+        The underlying mappings are stored in a list.  That list is public and can
+        accessed or updated using the *maps* attribute.  There is no other state.
+
+        Lookups search the underlying mappings successively until a key is found.
+        In contrast, writes, updates, and deletions only operate on the first
+        mapping.
+        '''
+
+        def __init__(self, *maps):
+            '''Initialize a ChainMap by setting *maps* to the given mappings.
+            If no mappings are provided, a single empty dictionary is used.
+
+            '''
+            self.maps = list(maps) or [{}]  # always at least one map
+
+        def __missing__(self, key):
+            raise KeyError(key)
+
+        def __getitem__(self, key):
+            for mapping in self.maps:
+                try:
+                    return mapping[
+                        key]  # can't use 'key in mapping' with defaultdict
+                except KeyError:
+                    pass
+            return self.__missing__(
+                key)  # support subclasses that define __missing__
+
+        def get(self, key, default=None):
+            return self[key] if key in self else default
+
+        def __len__(self):
+            return len(set().union(
+                *self.maps))  # reuses stored hash values if possible
+
+        def __iter__(self):
+            return iter(set().union(*self.maps))
+
+        def __contains__(self, key):
+            return any(key in m for m in self.maps)
+
+        def __bool__(self):
+            return any(self.maps)
+
+        @_recursive_repr()
+        def __repr__(self):
+            return '{0.__class__.__name__}({1})'.format(
+                self, ', '.join(map(repr, self.maps)))
+
+        @classmethod
+        def fromkeys(cls, iterable, *args):
+            'Create a ChainMap with a single dict created from the iterable.'
+            return cls(dict.fromkeys(iterable, *args))
+
+        def copy(self):
+            'New ChainMap or subclass with a new copy of maps[0] and refs to maps[1:]'
+            return self.__class__(self.maps[0].copy(), *self.maps[1:])
+
+        __copy__ = copy
+
+        def new_child(self):  # like Django's Context.push()
+            'New ChainMap with a new dict followed by all previous maps.'
+            return self.__class__({}, *self.maps)
+
+        @property
+        def parents(self):  # like Django's Context.pop()
+            'New ChainMap from maps[1:].'
+            return self.__class__(*self.maps[1:])
+
+        def __setitem__(self, key, value):
+            self.maps[0][key] = value
+
+        def __delitem__(self, key):
+            try:
+                del self.maps[0][key]
+            except KeyError:
+                raise KeyError(
+                    'Key not found in the first mapping: {!r}'.format(key))
+
+        def popitem(self):
+            'Remove and return an item pair from maps[0]. Raise KeyError is maps[0] is empty.'
+            try:
+                return self.maps[0].popitem()
+            except KeyError:
+                raise KeyError('No keys found in the first mapping.')
+
+        def pop(self, key, *args):
+            'Remove *key* from maps[0] and return its value. Raise KeyError if *key* not in maps[0].'
+            try:
+                return self.maps[0].pop(key, *args)
+            except KeyError:
+                raise KeyError(
+                    'Key not found in the first mapping: {!r}'.format(key))
+
+        def clear(self):
+            'Clear maps[0], leaving maps[1:] intact.'
+            self.maps[0].clear()
+
+
+try:
+    from importlib.util import cache_from_source  # Python >= 3.4
+except ImportError:  # pragma: no cover
+
+    def cache_from_source(path, debug_override=None):
+        assert path.endswith('.py')
+        if debug_override is None:
+            debug_override = __debug__
+        if debug_override:
+            suffix = 'c'
+        else:
+            suffix = 'o'
+        return path + suffix
+
+
+try:
+    from collections import OrderedDict
+except ImportError:  # pragma: no cover
+    # {{{ http://code.activestate.com/recipes/576693/ (r9)
+    # Backport of OrderedDict() class that runs on Python 2.4, 2.5, 2.6, 2.7 and pypy.
+    # Passes Python2.7's test suite and incorporates all the latest updates.
+    try:
+        from thread import get_ident as _get_ident
+    except ImportError:
+        from dummy_thread import get_ident as _get_ident
+
+    try:
+        from _abcoll import KeysView, ValuesView, ItemsView
+    except ImportError:
+        pass
+
+    class OrderedDict(dict):
+        'Dictionary that remembers insertion order'
+
+        # An inherited dict maps keys to values.
+        # The inherited dict provides __getitem__, __len__, __contains__, and get.
+        # The remaining methods are order-aware.
+        # Big-O running times for all methods are the same as for regular dictionaries.
+
+        # The internal self.__map dictionary maps keys to links in a doubly linked list.
+        # The circular doubly linked list starts and ends with a sentinel element.
+        # The sentinel element never gets deleted (this simplifies the algorithm).
+        # Each link is stored as a list of length three:  [PREV, NEXT, KEY].
+
+        def __init__(self, *args, **kwds):
+            '''Initialize an ordered dictionary.  Signature is the same as for
+            regular dictionaries, but keyword arguments are not recommended
+            because their insertion order is arbitrary.
+
+            '''
+            if len(args) > 1:
+                raise TypeError('expected at most 1 arguments, got %d' %
+                                len(args))
+            try:
+                self.__root
+            except AttributeError:
+                self.__root = root = []  # sentinel node
+                root[:] = [root, root, None]
+                self.__map = {}
+            self.__update(*args, **kwds)
+
+        def __setitem__(self, key, value, dict_setitem=dict.__setitem__):
+            'od.__setitem__(i, y) <==> od[i]=y'
+            # Setting a new item creates a new link which goes at the end of the linked
+            # list, and the inherited dictionary is updated with the new key/value pair.
+            if key not in self:
+                root = self.__root
+                last = root[0]
+                last[1] = root[0] = self.__map[key] = [last, root, key]
+            dict_setitem(self, key, value)
+
+        def __delitem__(self, key, dict_delitem=dict.__delitem__):
+            'od.__delitem__(y) <==> del od[y]'
+            # Deleting an existing item uses self.__map to find the link which is
+            # then removed by updating the links in the predecessor and successor nodes.
+            dict_delitem(self, key)
+            link_prev, link_next, key = self.__map.pop(key)
+            link_prev[1] = link_next
+            link_next[0] = link_prev
+
+        def __iter__(self):
+            'od.__iter__() <==> iter(od)'
+            root = self.__root
+            curr = root[1]
+            while curr is not root:
+                yield curr[2]
+                curr = curr[1]
+
+        def __reversed__(self):
+            'od.__reversed__() <==> reversed(od)'
+            root = self.__root
+            curr = root[0]
+            while curr is not root:
+                yield curr[2]
+                curr = curr[0]
+
+        def clear(self):
+            'od.clear() -> None.  Remove all items from od.'
+            try:
+                for node in self.__map.itervalues():
+                    del node[:]
+                root = self.__root
+                root[:] = [root, root, None]
+                self.__map.clear()
+            except AttributeError:
+                pass
+            dict.clear(self)
+
+        def popitem(self, last=True):
+            '''od.popitem() -> (k, v), return and remove a (key, value) pair.
+            Pairs are returned in LIFO order if last is true or FIFO order if false.
+
+            '''
+            if not self:
+                raise KeyError('dictionary is empty')
+            root = self.__root
+            if last:
+                link = root[0]
+                link_prev = link[0]
+                link_prev[1] = root
+                root[0] = link_prev
+            else:
+                link = root[1]
+                link_next = link[1]
+                root[1] = link_next
+                link_next[0] = root
+            key = link[2]
+            del self.__map[key]
+            value = dict.pop(self, key)
+            return key, value
+
+        # -- the following methods do not depend on the internal structure --
+
+        def keys(self):
+            'od.keys() -> list of keys in od'
+            return list(self)
+
+        def values(self):
+            'od.values() -> list of values in od'
+            return [self[key] for key in self]
+
+        def items(self):
+            'od.items() -> list of (key, value) pairs in od'
+            return [(key, self[key]) for key in self]
+
+        def iterkeys(self):
+            'od.iterkeys() -> an iterator over the keys in od'
+            return iter(self)
+
+        def itervalues(self):
+            'od.itervalues -> an iterator over the values in od'
+            for k in self:
+                yield self[k]
+
+        def iteritems(self):
+            'od.iteritems -> an iterator over the (key, value) items in od'
+            for k in self:
+                yield (k, self[k])
+
+        def update(*args, **kwds):
+            '''od.update(E, **F) -> None.  Update od from dict/iterable E and F.
+
+            If E is a dict instance, does:           for k in E: od[k] = E[k]
+            If E has a .keys() method, does:         for k in E.keys(): od[k] = E[k]
+            Or if E is an iterable of items, does:   for k, v in E: od[k] = v
+            In either case, this is followed by:     for k, v in F.items(): od[k] = v
+
+            '''
+            if len(args) > 2:
+                raise TypeError('update() takes at most 2 positional '
+                                'arguments (%d given)' % (len(args), ))
+            elif not args:
+                raise TypeError('update() takes at least 1 argument (0 given)')
+            self = args[0]
+            # Make progressively weaker assumptions about "other"
+            other = ()
+            if len(args) == 2:
+                other = args[1]
+            if isinstance(other, dict):
+                for key in other:
+                    self[key] = other[key]
+            elif hasattr(other, 'keys'):
+                for key in other.keys():
+                    self[key] = other[key]
+            else:
+                for key, value in other:
+                    self[key] = value
+            for key, value in kwds.items():
+                self[key] = value
+
+        __update = update  # let subclasses override update without breaking __init__
+
+        __marker = object()
+
+        def pop(self, key, default=__marker):
+            '''od.pop(k[,d]) -> v, remove specified key and return the corresponding value.
+            If key is not found, d is returned if given, otherwise KeyError is raised.
+
+            '''
+            if key in self:
+                result = self[key]
+                del self[key]
+                return result
+            if default is self.__marker:
+                raise KeyError(key)
+            return default
+
+        def setdefault(self, key, default=None):
+            'od.setdefault(k[,d]) -> od.get(k,d), also set od[k]=d if k not in od'
+            if key in self:
+                return self[key]
+            self[key] = default
+            return default
+
+        def __repr__(self, _repr_running=None):
+            'od.__repr__() <==> repr(od)'
+            if not _repr_running:
+                _repr_running = {}
+            call_key = id(self), _get_ident()
+            if call_key in _repr_running:
+                return '...'
+            _repr_running[call_key] = 1
+            try:
+                if not self:
+                    return '%s()' % (self.__class__.__name__, )
+                return '%s(%r)' % (self.__class__.__name__, self.items())
+            finally:
+                del _repr_running[call_key]
+
+        def __reduce__(self):
+            'Return state information for pickling'
+            items = [[k, self[k]] for k in self]
+            inst_dict = vars(self).copy()
+            for k in vars(OrderedDict()):
+                inst_dict.pop(k, None)
+            if inst_dict:
+                return (self.__class__, (items, ), inst_dict)
+            return self.__class__, (items, )
+
+        def copy(self):
+            'od.copy() -> a shallow copy of od'
+            return self.__class__(self)
+
+        @classmethod
+        def fromkeys(cls, iterable, value=None):
+            '''OD.fromkeys(S[, v]) -> New ordered dictionary with keys from S
+            and values equal to v (which defaults to None).
+
+            '''
+            d = cls()
+            for key in iterable:
+                d[key] = value
+            return d
+
+        def __eq__(self, other):
+            '''od.__eq__(y) <==> od==y.  Comparison to another OD is order-sensitive
+            while comparison to a regular mapping is order-insensitive.
+
+            '''
+            if isinstance(other, OrderedDict):
+                return len(self) == len(
+                    other) and self.items() == other.items()
+            return dict.__eq__(self, other)
+
+        def __ne__(self, other):
+            return not self == other
+
+        # -- the following methods are only used in Python 2.7 --
+
+        def viewkeys(self):
+            "od.viewkeys() -> a set-like object providing a view on od's keys"
+            return KeysView(self)
+
+        def viewvalues(self):
+            "od.viewvalues() -> an object providing a view on od's values"
+            return ValuesView(self)
+
+        def viewitems(self):
+            "od.viewitems() -> a set-like object providing a view on od's items"
+            return ItemsView(self)
+
+
+try:
+    from logging.config import BaseConfigurator, valid_ident
+except ImportError:  # pragma: no cover
+    IDENTIFIER = re.compile('^[a-z_][a-z0-9_]*$', re.I)
+
+    def valid_ident(s):
+        m = IDENTIFIER.match(s)
+        if not m:
+            raise ValueError('Not a valid Python identifier: %r' % s)
+        return True
+
+    # The ConvertingXXX classes are wrappers around standard Python containers,
+    # and they serve to convert any suitable values in the container. The
+    # conversion converts base dicts, lists and tuples to their wrapped
+    # equivalents, whereas strings which match a conversion format are converted
+    # appropriately.
+    #
+    # Each wrapper should have a configurator attribute holding the actual
+    # configurator to use for conversion.
+
+    class ConvertingDict(dict):
+        """A converting dictionary wrapper."""
+
+        def __getitem__(self, key):
+            value = dict.__getitem__(self, key)
+            result = self.configurator.convert(value)
+            # If the converted value is different, save for next time
+            if value is not result:
+                self[key] = result
+                if type(result) in (ConvertingDict, ConvertingList,
+                                    ConvertingTuple):
+                    result.parent = self
+                    result.key = key
+            return result
+
+        def get(self, key, default=None):
+            value = dict.get(self, key, default)
+            result = self.configurator.convert(value)
+            # If the converted value is different, save for next time
+            if value is not result:
+                self[key] = result
+                if type(result) in (ConvertingDict, ConvertingList,
+                                    ConvertingTuple):
+                    result.parent = self
+                    result.key = key
+            return result
+
+    def pop(self, key, default=None):
+        value = dict.pop(self, key, default)
+        result = self.configurator.convert(value)
+        if value is not result:
+            if type(result) in (ConvertingDict, ConvertingList,
+                                ConvertingTuple):
+                result.parent = self
+                result.key = key
+        return result
+
+    class ConvertingList(list):
+        """A converting list wrapper."""
+
+        def __getitem__(self, key):
+            value = list.__getitem__(self, key)
+            result = self.configurator.convert(value)
+            # If the converted value is different, save for next time
+            if value is not result:
+                self[key] = result
+                if type(result) in (ConvertingDict, ConvertingList,
+                                    ConvertingTuple):
+                    result.parent = self
+                    result.key = key
+            return result
+
+        def pop(self, idx=-1):
+            value = list.pop(self, idx)
+            result = self.configurator.convert(value)
+            if value is not result:
+                if type(result) in (ConvertingDict, ConvertingList,
+                                    ConvertingTuple):
+                    result.parent = self
+            return result
+
+    class ConvertingTuple(tuple):
+        """A converting tuple wrapper."""
+
+        def __getitem__(self, key):
+            value = tuple.__getitem__(self, key)
+            result = self.configurator.convert(value)
+            if value is not result:
+                if type(result) in (ConvertingDict, ConvertingList,
+                                    ConvertingTuple):
+                    result.parent = self
+                    result.key = key
+            return result
+
+    class BaseConfigurator(object):
+        """
+        The configurator base class which defines some useful defaults.
+        """
+
+        CONVERT_PATTERN = re.compile(r'^(?P<prefix>[a-z]+)://(?P<suffix>.*)$')
+
+        WORD_PATTERN = re.compile(r'^\s*(\w+)\s*')
+        DOT_PATTERN = re.compile(r'^\.\s*(\w+)\s*')
+        INDEX_PATTERN = re.compile(r'^\[\s*(\w+)\s*\]\s*')
+        DIGIT_PATTERN = re.compile(r'^\d+$')
+
+        value_converters = {
+            'ext': 'ext_convert',
+            'cfg': 'cfg_convert',
+        }
+
+        # We might want to use a different one, e.g. importlib
+        importer = staticmethod(__import__)
+
+        def __init__(self, config):
+            self.config = ConvertingDict(config)
+            self.config.configurator = self
+
+        def resolve(self, s):
+            """
+            Resolve strings to objects using standard import and attribute
+            syntax.
+            """
+            name = s.split('.')
+            used = name.pop(0)
+            try:
+                found = self.importer(used)
+                for frag in name:
+                    used += '.' + frag
+                    try:
+                        found = getattr(found, frag)
+                    except AttributeError:
+                        self.importer(used)
+                        found = getattr(found, frag)
+                return found
+            except ImportError:
+                e, tb = sys.exc_info()[1:]
+                v = ValueError('Cannot resolve %r: %s' % (s, e))
+                v.__cause__, v.__traceback__ = e, tb
+                raise v
+
+        def ext_convert(self, value):
+            """Default converter for the ext:// protocol."""
+            return self.resolve(value)
+
+        def cfg_convert(self, value):
+            """Default converter for the cfg:// protocol."""
+            rest = value
+            m = self.WORD_PATTERN.match(rest)
+            if m is None:
+                raise ValueError("Unable to convert %r" % value)
+            else:
+                rest = rest[m.end():]
+                d = self.config[m.groups()[0]]
+                while rest:
+                    m = self.DOT_PATTERN.match(rest)
+                    if m:
+                        d = d[m.groups()[0]]
+                    else:
+                        m = self.INDEX_PATTERN.match(rest)
+                        if m:
+                            idx = m.groups()[0]
+                            if not self.DIGIT_PATTERN.match(idx):
+                                d = d[idx]
+                            else:
+                                try:
+                                    n = int(
+                                        idx
+                                    )  # try as number first (most likely)
+                                    d = d[n]
+                                except TypeError:
+                                    d = d[idx]
+                    if m:
+                        rest = rest[m.end():]
+                    else:
+                        raise ValueError('Unable to convert '
+                                         '%r at %r' % (value, rest))
+            # rest should be empty
+            return d
+
+        def convert(self, value):
+            """
+            Convert values to an appropriate type. dicts, lists and tuples are
+            replaced by their converting alternatives. Strings are checked to
+            see if they have a conversion format and are converted if they do.
+            """
+            if not isinstance(value, ConvertingDict) and isinstance(
+                    value, dict):
+                value = ConvertingDict(value)
+                value.configurator = self
+            elif not isinstance(value, ConvertingList) and isinstance(
+                    value, list):
+                value = ConvertingList(value)
+                value.configurator = self
+            elif not isinstance(value, ConvertingTuple) and isinstance(value, tuple):
+                value = ConvertingTuple(value)
+                value.configurator = self
+            elif isinstance(value, string_types):
+                m = self.CONVERT_PATTERN.match(value)
+                if m:
+                    d = m.groupdict()
+                    prefix = d['prefix']
+                    converter = self.value_converters.get(prefix, None)
+                    if converter:
+                        suffix = d['suffix']
+                        converter = getattr(self, converter)
+                        value = converter(suffix)
+            return value
+
+        def configure_custom(self, config):
+            """Configure an object with a user-supplied factory."""
+            c = config.pop('()')
+            if not callable(c):
+                c = self.resolve(c)
+            props = config.pop('.', None)
+            # Check for valid identifiers
+            kwargs = dict([(k, config[k]) for k in config if valid_ident(k)])
+            result = c(**kwargs)
+            if props:
+                for name, value in props.items():
+                    setattr(result, name, value)
+            return result
+
+        def as_tuple(self, value):
+            """Utility function which converts lists to tuples."""
+            if isinstance(value, list):
+                value = tuple(value)
+            return value

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/database.py

@@ -0,0 +1,1359 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 The Python Software Foundation.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+"""PEP 376 implementation."""
+
+from __future__ import unicode_literals
+
+import base64
+import codecs
+import contextlib
+import hashlib
+import logging
+import os
+import posixpath
+import sys
+import zipimport
+
+from . import DistlibException, resources
+from .compat import StringIO
+from .version import get_scheme, UnsupportedVersionError
+from .metadata import (Metadata, METADATA_FILENAME, WHEEL_METADATA_FILENAME,
+                       LEGACY_METADATA_FILENAME)
+from .util import (parse_requirement, cached_property, parse_name_and_version,
+                   read_exports, write_exports, CSVReader, CSVWriter)
+
+__all__ = [
+    'Distribution', 'BaseInstalledDistribution', 'InstalledDistribution',
+    'EggInfoDistribution', 'DistributionPath'
+]
+
+logger = logging.getLogger(__name__)
+
+EXPORTS_FILENAME = 'pydist-exports.json'
+COMMANDS_FILENAME = 'pydist-commands.json'
+
+DIST_FILES = ('INSTALLER', METADATA_FILENAME, 'RECORD', 'REQUESTED',
+              'RESOURCES', EXPORTS_FILENAME, 'SHARED')
+
+DISTINFO_EXT = '.dist-info'
+
+
+class _Cache(object):
+    """
+    A simple cache mapping names and .dist-info paths to distributions
+    """
+
+    def __init__(self):
+        """
+        Initialise an instance. There is normally one for each DistributionPath.
+        """
+        self.name = {}
+        self.path = {}
+        self.generated = False
+
+    def clear(self):
+        """
+        Clear the cache, setting it to its initial state.
+        """
+        self.name.clear()
+        self.path.clear()
+        self.generated = False
+
+    def add(self, dist):
+        """
+        Add a distribution to the cache.
+        :param dist: The distribution to add.
+        """
+        if dist.path not in self.path:
+            self.path[dist.path] = dist
+            self.name.setdefault(dist.key, []).append(dist)
+
+
+class DistributionPath(object):
+    """
+    Represents a set of distributions installed on a path (typically sys.path).
+    """
+
+    def __init__(self, path=None, include_egg=False):
+        """
+        Create an instance from a path, optionally including legacy (distutils/
+        setuptools/distribute) distributions.
+        :param path: The path to use, as a list of directories. If not specified,
+                     sys.path is used.
+        :param include_egg: If True, this instance will look for and return legacy
+                            distributions as well as those based on PEP 376.
+        """
+        if path is None:
+            path = sys.path
+        self.path = path
+        self._include_dist = True
+        self._include_egg = include_egg
+
+        self._cache = _Cache()
+        self._cache_egg = _Cache()
+        self._cache_enabled = True
+        self._scheme = get_scheme('default')
+
+    def _get_cache_enabled(self):
+        return self._cache_enabled
+
+    def _set_cache_enabled(self, value):
+        self._cache_enabled = value
+
+    cache_enabled = property(_get_cache_enabled, _set_cache_enabled)
+
+    def clear_cache(self):
+        """
+        Clears the internal cache.
+        """
+        self._cache.clear()
+        self._cache_egg.clear()
+
+    def _yield_distributions(self):
+        """
+        Yield .dist-info and/or .egg(-info) distributions.
+        """
+        # We need to check if we've seen some resources already, because on
+        # some Linux systems (e.g. some Debian/Ubuntu variants) there are
+        # symlinks which alias other files in the environment.
+        seen = set()
+        for path in self.path:
+            finder = resources.finder_for_path(path)
+            if finder is None:
+                continue
+            r = finder.find('')
+            if not r or not r.is_container:
+                continue
+            rset = sorted(r.resources)
+            for entry in rset:
+                r = finder.find(entry)
+                if not r or r.path in seen:
+                    continue
+                try:
+                    if self._include_dist and entry.endswith(DISTINFO_EXT):
+                        possible_filenames = [
+                            METADATA_FILENAME, WHEEL_METADATA_FILENAME,
+                            LEGACY_METADATA_FILENAME
+                        ]
+                        for metadata_filename in possible_filenames:
+                            metadata_path = posixpath.join(
+                                entry, metadata_filename)
+                            pydist = finder.find(metadata_path)
+                            if pydist:
+                                break
+                        else:
+                            continue
+
+                        with contextlib.closing(pydist.as_stream()) as stream:
+                            metadata = Metadata(fileobj=stream,
+                                                scheme='legacy')
+                        logger.debug('Found %s', r.path)
+                        seen.add(r.path)
+                        yield new_dist_class(r.path,
+                                             metadata=metadata,
+                                             env=self)
+                    elif self._include_egg and entry.endswith(
+                            ('.egg-info', '.egg')):
+                        logger.debug('Found %s', r.path)
+                        seen.add(r.path)
+                        yield old_dist_class(r.path, self)
+                except Exception as e:
+                    msg = 'Unable to read distribution at %s, perhaps due to bad metadata: %s'
+                    logger.warning(msg, r.path, e)
+                    import warnings
+                    warnings.warn(msg % (r.path, e), stacklevel=2)
+
+    def _generate_cache(self):
+        """
+        Scan the path for distributions and populate the cache with
+        those that are found.
+        """
+        gen_dist = not self._cache.generated
+        gen_egg = self._include_egg and not self._cache_egg.generated
+        if gen_dist or gen_egg:
+            for dist in self._yield_distributions():
+                if isinstance(dist, InstalledDistribution):
+                    self._cache.add(dist)
+                else:
+                    self._cache_egg.add(dist)
+
+            if gen_dist:
+                self._cache.generated = True
+            if gen_egg:
+                self._cache_egg.generated = True
+
+    @classmethod
+    def distinfo_dirname(cls, name, version):
+        """
+        The *name* and *version* parameters are converted into their
+        filename-escaped form, i.e. any ``'-'`` characters are replaced
+        with ``'_'`` other than the one in ``'dist-info'`` and the one
+        separating the name from the version number.
+
+        :parameter name: is converted to a standard distribution name by replacing
+                         any runs of non- alphanumeric characters with a single
+                         ``'-'``.
+        :type name: string
+        :parameter version: is converted to a standard version string. Spaces
+                            become dots, and all other non-alphanumeric characters
+                            (except dots) become dashes, with runs of multiple
+                            dashes condensed to a single dash.
+        :type version: string
+        :returns: directory name
+        :rtype: string"""
+        name = name.replace('-', '_')
+        return '-'.join([name, version]) + DISTINFO_EXT
+
+    def get_distributions(self):
+        """
+        Provides an iterator that looks for distributions and returns
+        :class:`InstalledDistribution` or
+        :class:`EggInfoDistribution` instances for each one of them.
+
+        :rtype: iterator of :class:`InstalledDistribution` and
+                :class:`EggInfoDistribution` instances
+        """
+        if not self._cache_enabled:
+            for dist in self._yield_distributions():
+                yield dist
+        else:
+            self._generate_cache()
+
+            for dist in self._cache.path.values():
+                yield dist
+
+            if self._include_egg:
+                for dist in self._cache_egg.path.values():
+                    yield dist
+
+    def get_distribution(self, name):
+        """
+        Looks for a named distribution on the path.
+
+        This function only returns the first result found, as no more than one
+        value is expected. If nothing is found, ``None`` is returned.
+
+        :rtype: :class:`InstalledDistribution`, :class:`EggInfoDistribution`
+                or ``None``
+        """
+        result = None
+        name = name.lower()
+        if not self._cache_enabled:
+            for dist in self._yield_distributions():
+                if dist.key == name:
+                    result = dist
+                    break
+        else:
+            self._generate_cache()
+
+            if name in self._cache.name:
+                result = self._cache.name[name][0]
+            elif self._include_egg and name in self._cache_egg.name:
+                result = self._cache_egg.name[name][0]
+        return result
+
+    def provides_distribution(self, name, version=None):
+        """
+        Iterates over all distributions to find which distributions provide *name*.
+        If a *version* is provided, it will be used to filter the results.
+
+        This function only returns the first result found, since no more than
+        one values are expected. If the directory is not found, returns ``None``.
+
+        :parameter version: a version specifier that indicates the version
+                            required, conforming to the format in ``PEP-345``
+
+        :type name: string
+        :type version: string
+        """
+        matcher = None
+        if version is not None:
+            try:
+                matcher = self._scheme.matcher('%s (%s)' % (name, version))
+            except ValueError:
+                raise DistlibException('invalid name or version: %r, %r' %
+                                       (name, version))
+
+        for dist in self.get_distributions():
+            # We hit a problem on Travis where enum34 was installed and doesn't
+            # have a provides attribute ...
+            if not hasattr(dist, 'provides'):
+                logger.debug('No "provides": %s', dist)
+            else:
+                provided = dist.provides
+
+                for p in provided:
+                    p_name, p_ver = parse_name_and_version(p)
+                    if matcher is None:
+                        if p_name == name:
+                            yield dist
+                            break
+                    else:
+                        if p_name == name and matcher.match(p_ver):
+                            yield dist
+                            break
+
+    def get_file_path(self, name, relative_path):
+        """
+        Return the path to a resource file.
+        """
+        dist = self.get_distribution(name)
+        if dist is None:
+            raise LookupError('no distribution named %r found' % name)
+        return dist.get_resource_path(relative_path)
+
+    def get_exported_entries(self, category, name=None):
+        """
+        Return all of the exported entries in a particular category.
+
+        :param category: The category to search for entries.
+        :param name: If specified, only entries with that name are returned.
+        """
+        for dist in self.get_distributions():
+            r = dist.exports
+            if category in r:
+                d = r[category]
+                if name is not None:
+                    if name in d:
+                        yield d[name]
+                else:
+                    for v in d.values():
+                        yield v
+
+
+class Distribution(object):
+    """
+    A base class for distributions, whether installed or from indexes.
+    Either way, it must have some metadata, so that's all that's needed
+    for construction.
+    """
+
+    build_time_dependency = False
+    """
+    Set to True if it's known to be only a build-time dependency (i.e.
+    not needed after installation).
+    """
+
+    requested = False
+    """A boolean that indicates whether the ``REQUESTED`` metadata file is
+    present (in other words, whether the package was installed by user
+    request or it was installed as a dependency)."""
+
+    def __init__(self, metadata):
+        """
+        Initialise an instance.
+        :param metadata: The instance of :class:`Metadata` describing this
+        distribution.
+        """
+        self.metadata = metadata
+        self.name = metadata.name
+        self.key = self.name.lower()  # for case-insensitive comparisons
+        self.version = metadata.version
+        self.locator = None
+        self.digest = None
+        self.extras = None  # additional features requested
+        self.context = None  # environment marker overrides
+        self.download_urls = set()
+        self.digests = {}
+
+    @property
+    def source_url(self):
+        """
+        The source archive download URL for this distribution.
+        """
+        return self.metadata.source_url
+
+    download_url = source_url  # Backward compatibility
+
+    @property
+    def name_and_version(self):
+        """
+        A utility property which displays the name and version in parentheses.
+        """
+        return '%s (%s)' % (self.name, self.version)
+
+    @property
+    def provides(self):
+        """
+        A set of distribution names and versions provided by this distribution.
+        :return: A set of "name (version)" strings.
+        """
+        plist = self.metadata.provides
+        s = '%s (%s)' % (self.name, self.version)
+        if s not in plist:
+            plist.append(s)
+        return plist
+
+    def _get_requirements(self, req_attr):
+        md = self.metadata
+        reqts = getattr(md, req_attr)
+        logger.debug('%s: got requirements %r from metadata: %r', self.name,
+                     req_attr, reqts)
+        return set(
+            md.get_requirements(reqts, extras=self.extras, env=self.context))
+
+    @property
+    def run_requires(self):
+        return self._get_requirements('run_requires')
+
+    @property
+    def meta_requires(self):
+        return self._get_requirements('meta_requires')
+
+    @property
+    def build_requires(self):
+        return self._get_requirements('build_requires')
+
+    @property
+    def test_requires(self):
+        return self._get_requirements('test_requires')
+
+    @property
+    def dev_requires(self):
+        return self._get_requirements('dev_requires')
+
+    def matches_requirement(self, req):
+        """
+        Say if this instance matches (fulfills) a requirement.
+        :param req: The requirement to match.
+        :rtype req: str
+        :return: True if it matches, else False.
+        """
+        # Requirement may contain extras - parse to lose those
+        # from what's passed to the matcher
+        r = parse_requirement(req)
+        scheme = get_scheme(self.metadata.scheme)
+        try:
+            matcher = scheme.matcher(r.requirement)
+        except UnsupportedVersionError:
+            # XXX compat-mode if cannot read the version
+            logger.warning('could not read version %r - using name only', req)
+            name = req.split()[0]
+            matcher = scheme.matcher(name)
+
+        name = matcher.key  # case-insensitive
+
+        result = False
+        for p in self.provides:
+            p_name, p_ver = parse_name_and_version(p)
+            if p_name != name:
+                continue
+            try:
+                result = matcher.match(p_ver)
+                break
+            except UnsupportedVersionError:
+                pass
+        return result
+
+    def __repr__(self):
+        """
+        Return a textual representation of this instance,
+        """
+        if self.source_url:
+            suffix = ' [%s]' % self.source_url
+        else:
+            suffix = ''
+        return '<Distribution %s (%s)%s>' % (self.name, self.version, suffix)
+
+    def __eq__(self, other):
+        """
+        See if this distribution is the same as another.
+        :param other: The distribution to compare with. To be equal to one
+                      another. distributions must have the same type, name,
+                      version and source_url.
+        :return: True if it is the same, else False.
+        """
+        if type(other) is not type(self):
+            result = False
+        else:
+            result = (self.name == other.name and self.version == other.version
+                      and self.source_url == other.source_url)
+        return result
+
+    def __hash__(self):
+        """
+        Compute hash in a way which matches the equality test.
+        """
+        return hash(self.name) + hash(self.version) + hash(self.source_url)
+
+
+class BaseInstalledDistribution(Distribution):
+    """
+    This is the base class for installed distributions (whether PEP 376 or
+    legacy).
+    """
+
+    hasher = None
+
+    def __init__(self, metadata, path, env=None):
+        """
+        Initialise an instance.
+        :param metadata: An instance of :class:`Metadata` which describes the
+                         distribution. This will normally have been initialised
+                         from a metadata file in the ``path``.
+        :param path:     The path of the ``.dist-info`` or ``.egg-info``
+                         directory for the distribution.
+        :param env:      This is normally the :class:`DistributionPath`
+                         instance where this distribution was found.
+        """
+        super(BaseInstalledDistribution, self).__init__(metadata)
+        self.path = path
+        self.dist_path = env
+
+    def get_hash(self, data, hasher=None):
+        """
+        Get the hash of some data, using a particular hash algorithm, if
+        specified.
+
+        :param data: The data to be hashed.
+        :type data: bytes
+        :param hasher: The name of a hash implementation, supported by hashlib,
+                       or ``None``. Examples of valid values are ``'sha1'``,
+                       ``'sha224'``, ``'sha384'``, '``sha256'``, ``'md5'`` and
+                       ``'sha512'``. If no hasher is specified, the ``hasher``
+                       attribute of the :class:`InstalledDistribution` instance
+                       is used. If the hasher is determined to be ``None``, MD5
+                       is used as the hashing algorithm.
+        :returns: The hash of the data. If a hasher was explicitly specified,
+                  the returned hash will be prefixed with the specified hasher
+                  followed by '='.
+        :rtype: str
+        """
+        if hasher is None:
+            hasher = self.hasher
+        if hasher is None:
+            hasher = hashlib.md5
+            prefix = ''
+        else:
+            hasher = getattr(hashlib, hasher)
+            prefix = '%s=' % self.hasher
+        digest = hasher(data).digest()
+        digest = base64.urlsafe_b64encode(digest).rstrip(b'=').decode('ascii')
+        return '%s%s' % (prefix, digest)
+
+
+class InstalledDistribution(BaseInstalledDistribution):
+    """
+    Created with the *path* of the ``.dist-info`` directory provided to the
+    constructor. It reads the metadata contained in ``pydist.json`` when it is
+    instantiated., or uses a passed in Metadata instance (useful for when
+    dry-run mode is being used).
+    """
+
+    hasher = 'sha256'
+
+    def __init__(self, path, metadata=None, env=None):
+        self.modules = []
+        self.finder = finder = resources.finder_for_path(path)
+        if finder is None:
+            raise ValueError('finder unavailable for %s' % path)
+        if env and env._cache_enabled and path in env._cache.path:
+            metadata = env._cache.path[path].metadata
+        elif metadata is None:
+            r = finder.find(METADATA_FILENAME)
+            # Temporary - for Wheel 0.23 support
+            if r is None:
+                r = finder.find(WHEEL_METADATA_FILENAME)
+            # Temporary - for legacy support
+            if r is None:
+                r = finder.find(LEGACY_METADATA_FILENAME)
+            if r is None:
+                raise ValueError('no %s found in %s' %
+                                 (METADATA_FILENAME, path))
+            with contextlib.closing(r.as_stream()) as stream:
+                metadata = Metadata(fileobj=stream, scheme='legacy')
+
+        super(InstalledDistribution, self).__init__(metadata, path, env)
+
+        if env and env._cache_enabled:
+            env._cache.add(self)
+
+        r = finder.find('REQUESTED')
+        self.requested = r is not None
+        p = os.path.join(path, 'top_level.txt')
+        if os.path.exists(p):
+            with open(p, 'rb') as f:
+                data = f.read().decode('utf-8')
+            self.modules = data.splitlines()
+
+    def __repr__(self):
+        return '<InstalledDistribution %r %s at %r>' % (
+            self.name, self.version, self.path)
+
+    def __str__(self):
+        return "%s %s" % (self.name, self.version)
+
+    def _get_records(self):
+        """
+        Get the list of installed files for the distribution
+        :return: A list of tuples of path, hash and size. Note that hash and
+                 size might be ``None`` for some entries. The path is exactly
+                 as stored in the file (which is as in PEP 376).
+        """
+        results = []
+        r = self.get_distinfo_resource('RECORD')
+        with contextlib.closing(r.as_stream()) as stream:
+            with CSVReader(stream=stream) as record_reader:
+                # Base location is parent dir of .dist-info dir
+                # base_location = os.path.dirname(self.path)
+                # base_location = os.path.abspath(base_location)
+                for row in record_reader:
+                    missing = [None for i in range(len(row), 3)]
+                    path, checksum, size = row + missing
+                    # if not os.path.isabs(path):
+                    #     path = path.replace('/', os.sep)
+                    #     path = os.path.join(base_location, path)
+                    results.append((path, checksum, size))
+        return results
+
+    @cached_property
+    def exports(self):
+        """
+        Return the information exported by this distribution.
+        :return: A dictionary of exports, mapping an export category to a dict
+                 of :class:`ExportEntry` instances describing the individual
+                 export entries, and keyed by name.
+        """
+        result = {}
+        r = self.get_distinfo_resource(EXPORTS_FILENAME)
+        if r:
+            result = self.read_exports()
+        return result
+
+    def read_exports(self):
+        """
+        Read exports data from a file in .ini format.
+
+        :return: A dictionary of exports, mapping an export category to a list
+                 of :class:`ExportEntry` instances describing the individual
+                 export entries.
+        """
+        result = {}
+        r = self.get_distinfo_resource(EXPORTS_FILENAME)
+        if r:
+            with contextlib.closing(r.as_stream()) as stream:
+                result = read_exports(stream)
+        return result
+
+    def write_exports(self, exports):
+        """
+        Write a dictionary of exports to a file in .ini format.
+        :param exports: A dictionary of exports, mapping an export category to
+                        a list of :class:`ExportEntry` instances describing the
+                        individual export entries.
+        """
+        rf = self.get_distinfo_file(EXPORTS_FILENAME)
+        with open(rf, 'w') as f:
+            write_exports(exports, f)
+
+    def get_resource_path(self, relative_path):
+        """
+        NOTE: This API may change in the future.
+
+        Return the absolute path to a resource file with the given relative
+        path.
+
+        :param relative_path: The path, relative to .dist-info, of the resource
+                              of interest.
+        :return: The absolute path where the resource is to be found.
+        """
+        r = self.get_distinfo_resource('RESOURCES')
+        with contextlib.closing(r.as_stream()) as stream:
+            with CSVReader(stream=stream) as resources_reader:
+                for relative, destination in resources_reader:
+                    if relative == relative_path:
+                        return destination
+        raise KeyError('no resource file with relative path %r '
+                       'is installed' % relative_path)
+
+    def list_installed_files(self):
+        """
+        Iterates over the ``RECORD`` entries and returns a tuple
+        ``(path, hash, size)`` for each line.
+
+        :returns: iterator of (path, hash, size)
+        """
+        for result in self._get_records():
+            yield result
+
+    def write_installed_files(self, paths, prefix, dry_run=False):
+        """
+        Writes the ``RECORD`` file, using the ``paths`` iterable passed in. Any
+        existing ``RECORD`` file is silently overwritten.
+
+        prefix is used to determine when to write absolute paths.
+        """
+        prefix = os.path.join(prefix, '')
+        base = os.path.dirname(self.path)
+        base_under_prefix = base.startswith(prefix)
+        base = os.path.join(base, '')
+        record_path = self.get_distinfo_file('RECORD')
+        logger.info('creating %s', record_path)
+        if dry_run:
+            return None
+        with CSVWriter(record_path) as writer:
+            for path in paths:
+                if os.path.isdir(path) or path.endswith(('.pyc', '.pyo')):
+                    # do not put size and hash, as in PEP-376
+                    hash_value = size = ''
+                else:
+                    size = '%d' % os.path.getsize(path)
+                    with open(path, 'rb') as fp:
+                        hash_value = self.get_hash(fp.read())
+                if path.startswith(base) or (base_under_prefix
+                                             and path.startswith(prefix)):
+                    path = os.path.relpath(path, base)
+                writer.writerow((path, hash_value, size))
+
+            # add the RECORD file itself
+            if record_path.startswith(base):
+                record_path = os.path.relpath(record_path, base)
+            writer.writerow((record_path, '', ''))
+        return record_path
+
+    def check_installed_files(self):
+        """
+        Checks that the hashes and sizes of the files in ``RECORD`` are
+        matched by the files themselves. Returns a (possibly empty) list of
+        mismatches. Each entry in the mismatch list will be a tuple consisting
+        of the path, 'exists', 'size' or 'hash' according to what didn't match
+        (existence is checked first, then size, then hash), the expected
+        value and the actual value.
+        """
+        mismatches = []
+        base = os.path.dirname(self.path)
+        record_path = self.get_distinfo_file('RECORD')
+        for path, hash_value, size in self.list_installed_files():
+            if not os.path.isabs(path):
+                path = os.path.join(base, path)
+            if path == record_path:
+                continue
+            if not os.path.exists(path):
+                mismatches.append((path, 'exists', True, False))
+            elif os.path.isfile(path):
+                actual_size = str(os.path.getsize(path))
+                if size and actual_size != size:
+                    mismatches.append((path, 'size', size, actual_size))
+                elif hash_value:
+                    if '=' in hash_value:
+                        hasher = hash_value.split('=', 1)[0]
+                    else:
+                        hasher = None
+
+                    with open(path, 'rb') as f:
+                        actual_hash = self.get_hash(f.read(), hasher)
+                        if actual_hash != hash_value:
+                            mismatches.append(
+                                (path, 'hash', hash_value, actual_hash))
+        return mismatches
+
+    @cached_property
+    def shared_locations(self):
+        """
+        A dictionary of shared locations whose keys are in the set 'prefix',
+        'purelib', 'platlib', 'scripts', 'headers', 'data' and 'namespace'.
+        The corresponding value is the absolute path of that category for
+        this distribution, and takes into account any paths selected by the
+        user at installation time (e.g. via command-line arguments). In the
+        case of the 'namespace' key, this would be a list of absolute paths
+        for the roots of namespace packages in this distribution.
+
+        The first time this property is accessed, the relevant information is
+        read from the SHARED file in the .dist-info directory.
+        """
+        result = {}
+        shared_path = os.path.join(self.path, 'SHARED')
+        if os.path.isfile(shared_path):
+            with codecs.open(shared_path, 'r', encoding='utf-8') as f:
+                lines = f.read().splitlines()
+            for line in lines:
+                key, value = line.split('=', 1)
+                if key == 'namespace':
+                    result.setdefault(key, []).append(value)
+                else:
+                    result[key] = value
+        return result
+
+    def write_shared_locations(self, paths, dry_run=False):
+        """
+        Write shared location information to the SHARED file in .dist-info.
+        :param paths: A dictionary as described in the documentation for
+        :meth:`shared_locations`.
+        :param dry_run: If True, the action is logged but no file is actually
+                        written.
+        :return: The path of the file written to.
+        """
+        shared_path = os.path.join(self.path, 'SHARED')
+        logger.info('creating %s', shared_path)
+        if dry_run:
+            return None
+        lines = []
+        for key in ('prefix', 'lib', 'headers', 'scripts', 'data'):
+            path = paths[key]
+            if os.path.isdir(paths[key]):
+                lines.append('%s=%s' % (key, path))
+        for ns in paths.get('namespace', ()):
+            lines.append('namespace=%s' % ns)
+
+        with codecs.open(shared_path, 'w', encoding='utf-8') as f:
+            f.write('\n'.join(lines))
+        return shared_path
+
+    def get_distinfo_resource(self, path):
+        if path not in DIST_FILES:
+            raise DistlibException('invalid path for a dist-info file: '
+                                   '%r at %r' % (path, self.path))
+        finder = resources.finder_for_path(self.path)
+        if finder is None:
+            raise DistlibException('Unable to get a finder for %s' % self.path)
+        return finder.find(path)
+
+    def get_distinfo_file(self, path):
+        """
+        Returns a path located under the ``.dist-info`` directory. Returns a
+        string representing the path.
+
+        :parameter path: a ``'/'``-separated path relative to the
+                         ``.dist-info`` directory or an absolute path;
+                         If *path* is an absolute path and doesn't start
+                         with the ``.dist-info`` directory path,
+                         a :class:`DistlibException` is raised
+        :type path: str
+        :rtype: str
+        """
+        # Check if it is an absolute path  # XXX use relpath, add tests
+        if path.find(os.sep) >= 0:
+            # it's an absolute path?
+            distinfo_dirname, path = path.split(os.sep)[-2:]
+            if distinfo_dirname != self.path.split(os.sep)[-1]:
+                raise DistlibException(
+                    'dist-info file %r does not belong to the %r %s '
+                    'distribution' % (path, self.name, self.version))
+
+        # The file must be relative
+        if path not in DIST_FILES:
+            raise DistlibException('invalid path for a dist-info file: '
+                                   '%r at %r' % (path, self.path))
+
+        return os.path.join(self.path, path)
+
+    def list_distinfo_files(self):
+        """
+        Iterates over the ``RECORD`` entries and returns paths for each line if
+        the path is pointing to a file located in the ``.dist-info`` directory
+        or one of its subdirectories.
+
+        :returns: iterator of paths
+        """
+        base = os.path.dirname(self.path)
+        for path, checksum, size in self._get_records():
+            # XXX add separator or use real relpath algo
+            if not os.path.isabs(path):
+                path = os.path.join(base, path)
+            if path.startswith(self.path):
+                yield path
+
+    def __eq__(self, other):
+        return (isinstance(other, InstalledDistribution)
+                and self.path == other.path)
+
+    # See http://docs.python.org/reference/datamodel#object.__hash__
+    __hash__ = object.__hash__
+
+
+class EggInfoDistribution(BaseInstalledDistribution):
+    """Created with the *path* of the ``.egg-info`` directory or file provided
+    to the constructor. It reads the metadata contained in the file itself, or
+    if the given path happens to be a directory, the metadata is read from the
+    file ``PKG-INFO`` under that directory."""
+
+    requested = True  # as we have no way of knowing, assume it was
+    shared_locations = {}
+
+    def __init__(self, path, env=None):
+
+        def set_name_and_version(s, n, v):
+            s.name = n
+            s.key = n.lower()  # for case-insensitive comparisons
+            s.version = v
+
+        self.path = path
+        self.dist_path = env
+        if env and env._cache_enabled and path in env._cache_egg.path:
+            metadata = env._cache_egg.path[path].metadata
+            set_name_and_version(self, metadata.name, metadata.version)
+        else:
+            metadata = self._get_metadata(path)
+
+            # Need to be set before caching
+            set_name_and_version(self, metadata.name, metadata.version)
+
+            if env and env._cache_enabled:
+                env._cache_egg.add(self)
+        super(EggInfoDistribution, self).__init__(metadata, path, env)
+
+    def _get_metadata(self, path):
+        requires = None
+
+        def parse_requires_data(data):
+            """Create a list of dependencies from a requires.txt file.
+
+            *data*: the contents of a setuptools-produced requires.txt file.
+            """
+            reqs = []
+            lines = data.splitlines()
+            for line in lines:
+                line = line.strip()
+                # sectioned files have bare newlines (separating sections)
+                if not line:  # pragma: no cover
+                    continue
+                if line.startswith('['):  # pragma: no cover
+                    logger.warning(
+                        'Unexpected line: quitting requirement scan: %r', line)
+                    break
+                r = parse_requirement(line)
+                if not r:  # pragma: no cover
+                    logger.warning('Not recognised as a requirement: %r', line)
+                    continue
+                if r.extras:  # pragma: no cover
+                    logger.warning('extra requirements in requires.txt are '
+                                   'not supported')
+                if not r.constraints:
+                    reqs.append(r.name)
+                else:
+                    cons = ', '.join('%s%s' % c for c in r.constraints)
+                    reqs.append('%s (%s)' % (r.name, cons))
+            return reqs
+
+        def parse_requires_path(req_path):
+            """Create a list of dependencies from a requires.txt file.
+
+            *req_path*: the path to a setuptools-produced requires.txt file.
+            """
+
+            reqs = []
+            try:
+                with codecs.open(req_path, 'r', 'utf-8') as fp:
+                    reqs = parse_requires_data(fp.read())
+            except IOError:
+                pass
+            return reqs
+
+        tl_path = tl_data = None
+        if path.endswith('.egg'):
+            if os.path.isdir(path):
+                p = os.path.join(path, 'EGG-INFO')
+                meta_path = os.path.join(p, 'PKG-INFO')
+                metadata = Metadata(path=meta_path, scheme='legacy')
+                req_path = os.path.join(p, 'requires.txt')
+                tl_path = os.path.join(p, 'top_level.txt')
+                requires = parse_requires_path(req_path)
+            else:
+                # FIXME handle the case where zipfile is not available
+                zipf = zipimport.zipimporter(path)
+                fileobj = StringIO(
+                    zipf.get_data('EGG-INFO/PKG-INFO').decode('utf8'))
+                metadata = Metadata(fileobj=fileobj, scheme='legacy')
+                try:
+                    data = zipf.get_data('EGG-INFO/requires.txt')
+                    tl_data = zipf.get_data('EGG-INFO/top_level.txt').decode(
+                        'utf-8')
+                    requires = parse_requires_data(data.decode('utf-8'))
+                except IOError:
+                    requires = None
+        elif path.endswith('.egg-info'):
+            if os.path.isdir(path):
+                req_path = os.path.join(path, 'requires.txt')
+                requires = parse_requires_path(req_path)
+                path = os.path.join(path, 'PKG-INFO')
+                tl_path = os.path.join(path, 'top_level.txt')
+            metadata = Metadata(path=path, scheme='legacy')
+        else:
+            raise DistlibException('path must end with .egg-info or .egg, '
+                                   'got %r' % path)
+
+        if requires:
+            metadata.add_requirements(requires)
+        # look for top-level modules in top_level.txt, if present
+        if tl_data is None:
+            if tl_path is not None and os.path.exists(tl_path):
+                with open(tl_path, 'rb') as f:
+                    tl_data = f.read().decode('utf-8')
+        if not tl_data:
+            tl_data = []
+        else:
+            tl_data = tl_data.splitlines()
+        self.modules = tl_data
+        return metadata
+
+    def __repr__(self):
+        return '<EggInfoDistribution %r %s at %r>' % (self.name, self.version,
+                                                      self.path)
+
+    def __str__(self):
+        return "%s %s" % (self.name, self.version)
+
+    def check_installed_files(self):
+        """
+        Checks that the hashes and sizes of the files in ``RECORD`` are
+        matched by the files themselves. Returns a (possibly empty) list of
+        mismatches. Each entry in the mismatch list will be a tuple consisting
+        of the path, 'exists', 'size' or 'hash' according to what didn't match
+        (existence is checked first, then size, then hash), the expected
+        value and the actual value.
+        """
+        mismatches = []
+        record_path = os.path.join(self.path, 'installed-files.txt')
+        if os.path.exists(record_path):
+            for path, _, _ in self.list_installed_files():
+                if path == record_path:
+                    continue
+                if not os.path.exists(path):
+                    mismatches.append((path, 'exists', True, False))
+        return mismatches
+
+    def list_installed_files(self):
+        """
+        Iterates over the ``installed-files.txt`` entries and returns a tuple
+        ``(path, hash, size)`` for each line.
+
+        :returns: a list of (path, hash, size)
+        """
+
+        def _md5(path):
+            f = open(path, 'rb')
+            try:
+                content = f.read()
+            finally:
+                f.close()
+            return hashlib.md5(content).hexdigest()
+
+        def _size(path):
+            return os.stat(path).st_size
+
+        record_path = os.path.join(self.path, 'installed-files.txt')
+        result = []
+        if os.path.exists(record_path):
+            with codecs.open(record_path, 'r', encoding='utf-8') as f:
+                for line in f:
+                    line = line.strip()
+                    p = os.path.normpath(os.path.join(self.path, line))
+                    # "./" is present as a marker between installed files
+                    # and installation metadata files
+                    if not os.path.exists(p):
+                        logger.warning('Non-existent file: %s', p)
+                        if p.endswith(('.pyc', '.pyo')):
+                            continue
+                        # otherwise fall through and fail
+                    if not os.path.isdir(p):
+                        result.append((p, _md5(p), _size(p)))
+            result.append((record_path, None, None))
+        return result
+
+    def list_distinfo_files(self, absolute=False):
+        """
+        Iterates over the ``installed-files.txt`` entries and returns paths for
+        each line if the path is pointing to a file located in the
+        ``.egg-info`` directory or one of its subdirectories.
+
+        :parameter absolute: If *absolute* is ``True``, each returned path is
+                          transformed into a local absolute path. Otherwise the
+                          raw value from ``installed-files.txt`` is returned.
+        :type absolute: boolean
+        :returns: iterator of paths
+        """
+        record_path = os.path.join(self.path, 'installed-files.txt')
+        if os.path.exists(record_path):
+            skip = True
+            with codecs.open(record_path, 'r', encoding='utf-8') as f:
+                for line in f:
+                    line = line.strip()
+                    if line == './':
+                        skip = False
+                        continue
+                    if not skip:
+                        p = os.path.normpath(os.path.join(self.path, line))
+                        if p.startswith(self.path):
+                            if absolute:
+                                yield p
+                            else:
+                                yield line
+
+    def __eq__(self, other):
+        return (isinstance(other, EggInfoDistribution)
+                and self.path == other.path)
+
+    # See http://docs.python.org/reference/datamodel#object.__hash__
+    __hash__ = object.__hash__
+
+
+new_dist_class = InstalledDistribution
+old_dist_class = EggInfoDistribution
+
+
+class DependencyGraph(object):
+    """
+    Represents a dependency graph between distributions.
+
+    The dependency relationships are stored in an ``adjacency_list`` that maps
+    distributions to a list of ``(other, label)`` tuples where  ``other``
+    is a distribution and the edge is labeled with ``label`` (i.e. the version
+    specifier, if such was provided). Also, for more efficient traversal, for
+    every distribution ``x``, a list of predecessors is kept in
+    ``reverse_list[x]``. An edge from distribution ``a`` to
+    distribution ``b`` means that ``a`` depends on ``b``. If any missing
+    dependencies are found, they are stored in ``missing``, which is a
+    dictionary that maps distributions to a list of requirements that were not
+    provided by any other distributions.
+    """
+
+    def __init__(self):
+        self.adjacency_list = {}
+        self.reverse_list = {}
+        self.missing = {}
+
+    def add_distribution(self, distribution):
+        """Add the *distribution* to the graph.
+
+        :type distribution: :class:`distutils2.database.InstalledDistribution`
+                            or :class:`distutils2.database.EggInfoDistribution`
+        """
+        self.adjacency_list[distribution] = []
+        self.reverse_list[distribution] = []
+        # self.missing[distribution] = []
+
+    def add_edge(self, x, y, label=None):
+        """Add an edge from distribution *x* to distribution *y* with the given
+        *label*.
+
+        :type x: :class:`distutils2.database.InstalledDistribution` or
+                 :class:`distutils2.database.EggInfoDistribution`
+        :type y: :class:`distutils2.database.InstalledDistribution` or
+                 :class:`distutils2.database.EggInfoDistribution`
+        :type label: ``str`` or ``None``
+        """
+        self.adjacency_list[x].append((y, label))
+        # multiple edges are allowed, so be careful
+        if x not in self.reverse_list[y]:
+            self.reverse_list[y].append(x)
+
+    def add_missing(self, distribution, requirement):
+        """
+        Add a missing *requirement* for the given *distribution*.
+
+        :type distribution: :class:`distutils2.database.InstalledDistribution`
+                            or :class:`distutils2.database.EggInfoDistribution`
+        :type requirement: ``str``
+        """
+        logger.debug('%s missing %r', distribution, requirement)
+        self.missing.setdefault(distribution, []).append(requirement)
+
+    def _repr_dist(self, dist):
+        return '%s %s' % (dist.name, dist.version)
+
+    def repr_node(self, dist, level=1):
+        """Prints only a subgraph"""
+        output = [self._repr_dist(dist)]
+        for other, label in self.adjacency_list[dist]:
+            dist = self._repr_dist(other)
+            if label is not None:
+                dist = '%s [%s]' % (dist, label)
+            output.append('    ' * level + str(dist))
+            suboutput = self.repr_node(other, level + 1)
+            subs = suboutput.split('\n')
+            output.extend(subs[1:])
+        return '\n'.join(output)
+
+    def to_dot(self, f, skip_disconnected=True):
+        """Writes a DOT output for the graph to the provided file *f*.
+
+        If *skip_disconnected* is set to ``True``, then all distributions
+        that are not dependent on any other distribution are skipped.
+
+        :type f: has to support ``file``-like operations
+        :type skip_disconnected: ``bool``
+        """
+        disconnected = []
+
+        f.write("digraph dependencies {\n")
+        for dist, adjs in self.adjacency_list.items():
+            if len(adjs) == 0 and not skip_disconnected:
+                disconnected.append(dist)
+            for other, label in adjs:
+                if label is not None:
+                    f.write('"%s" -> "%s" [label="%s"]\n' %
+                            (dist.name, other.name, label))
+                else:
+                    f.write('"%s" -> "%s"\n' % (dist.name, other.name))
+        if not skip_disconnected and len(disconnected) > 0:
+            f.write('subgraph disconnected {\n')
+            f.write('label = "Disconnected"\n')
+            f.write('bgcolor = red\n')
+
+            for dist in disconnected:
+                f.write('"%s"' % dist.name)
+                f.write('\n')
+            f.write('}\n')
+        f.write('}\n')
+
+    def topological_sort(self):
+        """
+        Perform a topological sort of the graph.
+        :return: A tuple, the first element of which is a topologically sorted
+                 list of distributions, and the second element of which is a
+                 list of distributions that cannot be sorted because they have
+                 circular dependencies and so form a cycle.
+        """
+        result = []
+        # Make a shallow copy of the adjacency list
+        alist = {}
+        for k, v in self.adjacency_list.items():
+            alist[k] = v[:]
+        while True:
+            # See what we can remove in this run
+            to_remove = []
+            for k, v in list(alist.items())[:]:
+                if not v:
+                    to_remove.append(k)
+                    del alist[k]
+            if not to_remove:
+                # What's left in alist (if anything) is a cycle.
+                break
+            # Remove from the adjacency list of others
+            for k, v in alist.items():
+                alist[k] = [(d, r) for d, r in v if d not in to_remove]
+            logger.debug('Moving to result: %s',
+                         ['%s (%s)' % (d.name, d.version) for d in to_remove])
+            result.extend(to_remove)
+        return result, list(alist.keys())
+
+    def __repr__(self):
+        """Representation of the graph"""
+        output = []
+        for dist, adjs in self.adjacency_list.items():
+            output.append(self.repr_node(dist))
+        return '\n'.join(output)
+
+
+def make_graph(dists, scheme='default'):
+    """Makes a dependency graph from the given distributions.
+
+    :parameter dists: a list of distributions
+    :type dists: list of :class:`distutils2.database.InstalledDistribution` and
+                 :class:`distutils2.database.EggInfoDistribution` instances
+    :rtype: a :class:`DependencyGraph` instance
+    """
+    scheme = get_scheme(scheme)
+    graph = DependencyGraph()
+    provided = {}  # maps names to lists of (version, dist) tuples
+
+    # first, build the graph and find out what's provided
+    for dist in dists:
+        graph.add_distribution(dist)
+
+        for p in dist.provides:
+            name, version = parse_name_and_version(p)
+            logger.debug('Add to provided: %s, %s, %s', name, version, dist)
+            provided.setdefault(name, []).append((version, dist))
+
+    # now make the edges
+    for dist in dists:
+        requires = (dist.run_requires | dist.meta_requires
+                    | dist.build_requires | dist.dev_requires)
+        for req in requires:
+            try:
+                matcher = scheme.matcher(req)
+            except UnsupportedVersionError:
+                # XXX compat-mode if cannot read the version
+                logger.warning('could not read version %r - using name only',
+                               req)
+                name = req.split()[0]
+                matcher = scheme.matcher(name)
+
+            name = matcher.key  # case-insensitive
+
+            matched = False
+            if name in provided:
+                for version, provider in provided[name]:
+                    try:
+                        match = matcher.match(version)
+                    except UnsupportedVersionError:
+                        match = False
+
+                    if match:
+                        graph.add_edge(dist, provider, req)
+                        matched = True
+                        break
+            if not matched:
+                graph.add_missing(dist, req)
+    return graph
+
+
+def get_dependent_dists(dists, dist):
+    """Recursively generate a list of distributions from *dists* that are
+    dependent on *dist*.
+
+    :param dists: a list of distributions
+    :param dist: a distribution, member of *dists* for which we are interested
+    """
+    if dist not in dists:
+        raise DistlibException('given distribution %r is not a member '
+                               'of the list' % dist.name)
+    graph = make_graph(dists)
+
+    dep = [dist]  # dependent distributions
+    todo = graph.reverse_list[dist]  # list of nodes we should inspect
+
+    while todo:
+        d = todo.pop()
+        dep.append(d)
+        for succ in graph.reverse_list[d]:
+            if succ not in dep:
+                todo.append(succ)
+
+    dep.pop(0)  # remove dist from dep, was there to prevent infinite loops
+    return dep
+
+
+def get_required_dists(dists, dist):
+    """Recursively generate a list of distributions from *dists* that are
+    required by *dist*.
+
+    :param dists: a list of distributions
+    :param dist: a distribution, member of *dists* for which we are interested
+                 in finding the dependencies.
+    """
+    if dist not in dists:
+        raise DistlibException('given distribution %r is not a member '
+                               'of the list' % dist.name)
+    graph = make_graph(dists)
+
+    req = set()  # required distributions
+    todo = graph.adjacency_list[dist]  # list of nodes we should inspect
+    seen = set(t[0] for t in todo)  # already added to todo
+
+    while todo:
+        d = todo.pop()[0]
+        req.add(d)
+        pred_list = graph.adjacency_list[d]
+        for pred in pred_list:
+            d = pred[0]
+            if d not in req and d not in seen:
+                seen.add(d)
+                todo.append(pred)
+    return req
+
+
+def make_dist(name, version, **kwargs):
+    """
+    A convenience method for making a dist given just a name and version.
+    """
+    summary = kwargs.pop('summary', 'Placeholder for summary')
+    md = Metadata(**kwargs)
+    md.name = name
+    md.version = version
+    md.summary = summary or 'Placeholder for summary'
+    return Distribution(md)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/index.py

@@ -0,0 +1,508 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2013-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+import hashlib
+import logging
+import os
+import shutil
+import subprocess
+import tempfile
+try:
+    from threading import Thread
+except ImportError:  # pragma: no cover
+    from dummy_threading import Thread
+
+from . import DistlibException
+from .compat import (HTTPBasicAuthHandler, Request, HTTPPasswordMgr,
+                     urlparse, build_opener, string_types)
+from .util import zip_dir, ServerProxy
+
+logger = logging.getLogger(__name__)
+
+DEFAULT_INDEX = 'https://pypi.org/pypi'
+DEFAULT_REALM = 'pypi'
+
+
+class PackageIndex(object):
+    """
+    This class represents a package index compatible with PyPI, the Python
+    Package Index.
+    """
+
+    boundary = b'----------ThIs_Is_tHe_distlib_index_bouNdaRY_$'
+
+    def __init__(self, url=None):
+        """
+        Initialise an instance.
+
+        :param url: The URL of the index. If not specified, the URL for PyPI is
+                    used.
+        """
+        self.url = url or DEFAULT_INDEX
+        self.read_configuration()
+        scheme, netloc, path, params, query, frag = urlparse(self.url)
+        if params or query or frag or scheme not in ('http', 'https'):
+            raise DistlibException('invalid repository: %s' % self.url)
+        self.password_handler = None
+        self.ssl_verifier = None
+        self.gpg = None
+        self.gpg_home = None
+        with open(os.devnull, 'w') as sink:
+            # Use gpg by default rather than gpg2, as gpg2 insists on
+            # prompting for passwords
+            for s in ('gpg', 'gpg2'):
+                try:
+                    rc = subprocess.check_call([s, '--version'], stdout=sink,
+                                               stderr=sink)
+                    if rc == 0:
+                        self.gpg = s
+                        break
+                except OSError:
+                    pass
+
+    def _get_pypirc_command(self):
+        """
+        Get the distutils command for interacting with PyPI configurations.
+        :return: the command.
+        """
+        from .util import _get_pypirc_command as cmd
+        return cmd()
+
+    def read_configuration(self):
+        """
+        Read the PyPI access configuration as supported by distutils. This populates
+        ``username``, ``password``, ``realm`` and ``url`` attributes from the
+        configuration.
+        """
+        from .util import _load_pypirc
+        cfg = _load_pypirc(self)
+        self.username = cfg.get('username')
+        self.password = cfg.get('password')
+        self.realm = cfg.get('realm', 'pypi')
+        self.url = cfg.get('repository', self.url)
+
+    def save_configuration(self):
+        """
+        Save the PyPI access configuration. You must have set ``username`` and
+        ``password`` attributes before calling this method.
+        """
+        self.check_credentials()
+        from .util import _store_pypirc
+        _store_pypirc(self)
+
+    def check_credentials(self):
+        """
+        Check that ``username`` and ``password`` have been set, and raise an
+        exception if not.
+        """
+        if self.username is None or self.password is None:
+            raise DistlibException('username and password must be set')
+        pm = HTTPPasswordMgr()
+        _, netloc, _, _, _, _ = urlparse(self.url)
+        pm.add_password(self.realm, netloc, self.username, self.password)
+        self.password_handler = HTTPBasicAuthHandler(pm)
+
+    def register(self, metadata):  # pragma: no cover
+        """
+        Register a distribution on PyPI, using the provided metadata.
+
+        :param metadata: A :class:`Metadata` instance defining at least a name
+                         and version number for the distribution to be
+                         registered.
+        :return: The HTTP response received from PyPI upon submission of the
+                request.
+        """
+        self.check_credentials()
+        metadata.validate()
+        d = metadata.todict()
+        d[':action'] = 'verify'
+        request = self.encode_request(d.items(), [])
+        self.send_request(request)
+        d[':action'] = 'submit'
+        request = self.encode_request(d.items(), [])
+        return self.send_request(request)
+
+    def _reader(self, name, stream, outbuf):
+        """
+        Thread runner for reading lines of from a subprocess into a buffer.
+
+        :param name: The logical name of the stream (used for logging only).
+        :param stream: The stream to read from. This will typically a pipe
+                       connected to the output stream of a subprocess.
+        :param outbuf: The list to append the read lines to.
+        """
+        while True:
+            s = stream.readline()
+            if not s:
+                break
+            s = s.decode('utf-8').rstrip()
+            outbuf.append(s)
+            logger.debug('%s: %s' % (name, s))
+        stream.close()
+
+    def get_sign_command(self, filename, signer, sign_password, keystore=None):  # pragma: no cover
+        """
+        Return a suitable command for signing a file.
+
+        :param filename: The pathname to the file to be signed.
+        :param signer: The identifier of the signer of the file.
+        :param sign_password: The passphrase for the signer's
+                              private key used for signing.
+        :param keystore: The path to a directory which contains the keys
+                         used in verification. If not specified, the
+                         instance's ``gpg_home`` attribute is used instead.
+        :return: The signing command as a list suitable to be
+                 passed to :class:`subprocess.Popen`.
+        """
+        cmd = [self.gpg, '--status-fd', '2', '--no-tty']
+        if keystore is None:
+            keystore = self.gpg_home
+        if keystore:
+            cmd.extend(['--homedir', keystore])
+        if sign_password is not None:
+            cmd.extend(['--batch', '--passphrase-fd', '0'])
+        td = tempfile.mkdtemp()
+        sf = os.path.join(td, os.path.basename(filename) + '.asc')
+        cmd.extend(['--detach-sign', '--armor', '--local-user',
+                    signer, '--output', sf, filename])
+        logger.debug('invoking: %s', ' '.join(cmd))
+        return cmd, sf
+
+    def run_command(self, cmd, input_data=None):
+        """
+        Run a command in a child process , passing it any input data specified.
+
+        :param cmd: The command to run.
+        :param input_data: If specified, this must be a byte string containing
+                           data to be sent to the child process.
+        :return: A tuple consisting of the subprocess' exit code, a list of
+                 lines read from the subprocess' ``stdout``, and a list of
+                 lines read from the subprocess' ``stderr``.
+        """
+        kwargs = {
+            'stdout': subprocess.PIPE,
+            'stderr': subprocess.PIPE,
+        }
+        if input_data is not None:
+            kwargs['stdin'] = subprocess.PIPE
+        stdout = []
+        stderr = []
+        p = subprocess.Popen(cmd, **kwargs)
+        # We don't use communicate() here because we may need to
+        # get clever with interacting with the command
+        t1 = Thread(target=self._reader, args=('stdout', p.stdout, stdout))
+        t1.start()
+        t2 = Thread(target=self._reader, args=('stderr', p.stderr, stderr))
+        t2.start()
+        if input_data is not None:
+            p.stdin.write(input_data)
+            p.stdin.close()
+
+        p.wait()
+        t1.join()
+        t2.join()
+        return p.returncode, stdout, stderr
+
+    def sign_file(self, filename, signer, sign_password, keystore=None):  # pragma: no cover
+        """
+        Sign a file.
+
+        :param filename: The pathname to the file to be signed.
+        :param signer: The identifier of the signer of the file.
+        :param sign_password: The passphrase for the signer's
+                              private key used for signing.
+        :param keystore: The path to a directory which contains the keys
+                         used in signing. If not specified, the instance's
+                         ``gpg_home`` attribute is used instead.
+        :return: The absolute pathname of the file where the signature is
+                 stored.
+        """
+        cmd, sig_file = self.get_sign_command(filename, signer, sign_password,
+                                              keystore)
+        rc, stdout, stderr = self.run_command(cmd,
+                                              sign_password.encode('utf-8'))
+        if rc != 0:
+            raise DistlibException('sign command failed with error '
+                                   'code %s' % rc)
+        return sig_file
+
+    def upload_file(self, metadata, filename, signer=None, sign_password=None,
+                    filetype='sdist', pyversion='source', keystore=None):
+        """
+        Upload a release file to the index.
+
+        :param metadata: A :class:`Metadata` instance defining at least a name
+                         and version number for the file to be uploaded.
+        :param filename: The pathname of the file to be uploaded.
+        :param signer: The identifier of the signer of the file.
+        :param sign_password: The passphrase for the signer's
+                              private key used for signing.
+        :param filetype: The type of the file being uploaded. This is the
+                        distutils command which produced that file, e.g.
+                        ``sdist`` or ``bdist_wheel``.
+        :param pyversion: The version of Python which the release relates
+                          to. For code compatible with any Python, this would
+                          be ``source``, otherwise it would be e.g. ``3.2``.
+        :param keystore: The path to a directory which contains the keys
+                         used in signing. If not specified, the instance's
+                         ``gpg_home`` attribute is used instead.
+        :return: The HTTP response received from PyPI upon submission of the
+                request.
+        """
+        self.check_credentials()
+        if not os.path.exists(filename):
+            raise DistlibException('not found: %s' % filename)
+        metadata.validate()
+        d = metadata.todict()
+        sig_file = None
+        if signer:
+            if not self.gpg:
+                logger.warning('no signing program available - not signed')
+            else:
+                sig_file = self.sign_file(filename, signer, sign_password,
+                                          keystore)
+        with open(filename, 'rb') as f:
+            file_data = f.read()
+        md5_digest = hashlib.md5(file_data).hexdigest()
+        sha256_digest = hashlib.sha256(file_data).hexdigest()
+        d.update({
+            ':action': 'file_upload',
+            'protocol_version': '1',
+            'filetype': filetype,
+            'pyversion': pyversion,
+            'md5_digest': md5_digest,
+            'sha256_digest': sha256_digest,
+        })
+        files = [('content', os.path.basename(filename), file_data)]
+        if sig_file:
+            with open(sig_file, 'rb') as f:
+                sig_data = f.read()
+            files.append(('gpg_signature', os.path.basename(sig_file),
+                         sig_data))
+            shutil.rmtree(os.path.dirname(sig_file))
+        request = self.encode_request(d.items(), files)
+        return self.send_request(request)
+
+    def upload_documentation(self, metadata, doc_dir):  # pragma: no cover
+        """
+        Upload documentation to the index.
+
+        :param metadata: A :class:`Metadata` instance defining at least a name
+                         and version number for the documentation to be
+                         uploaded.
+        :param doc_dir: The pathname of the directory which contains the
+                        documentation. This should be the directory that
+                        contains the ``index.html`` for the documentation.
+        :return: The HTTP response received from PyPI upon submission of the
+                request.
+        """
+        self.check_credentials()
+        if not os.path.isdir(doc_dir):
+            raise DistlibException('not a directory: %r' % doc_dir)
+        fn = os.path.join(doc_dir, 'index.html')
+        if not os.path.exists(fn):
+            raise DistlibException('not found: %r' % fn)
+        metadata.validate()
+        name, version = metadata.name, metadata.version
+        zip_data = zip_dir(doc_dir).getvalue()
+        fields = [(':action', 'doc_upload'),
+                  ('name', name), ('version', version)]
+        files = [('content', name, zip_data)]
+        request = self.encode_request(fields, files)
+        return self.send_request(request)
+
+    def get_verify_command(self, signature_filename, data_filename,
+                           keystore=None):
+        """
+        Return a suitable command for verifying a file.
+
+        :param signature_filename: The pathname to the file containing the
+                                   signature.
+        :param data_filename: The pathname to the file containing the
+                              signed data.
+        :param keystore: The path to a directory which contains the keys
+                         used in verification. If not specified, the
+                         instance's ``gpg_home`` attribute is used instead.
+        :return: The verifying command as a list suitable to be
+                 passed to :class:`subprocess.Popen`.
+        """
+        cmd = [self.gpg, '--status-fd', '2', '--no-tty']
+        if keystore is None:
+            keystore = self.gpg_home
+        if keystore:
+            cmd.extend(['--homedir', keystore])
+        cmd.extend(['--verify', signature_filename, data_filename])
+        logger.debug('invoking: %s', ' '.join(cmd))
+        return cmd
+
+    def verify_signature(self, signature_filename, data_filename,
+                         keystore=None):
+        """
+        Verify a signature for a file.
+
+        :param signature_filename: The pathname to the file containing the
+                                   signature.
+        :param data_filename: The pathname to the file containing the
+                              signed data.
+        :param keystore: The path to a directory which contains the keys
+                         used in verification. If not specified, the
+                         instance's ``gpg_home`` attribute is used instead.
+        :return: True if the signature was verified, else False.
+        """
+        if not self.gpg:
+            raise DistlibException('verification unavailable because gpg '
+                                   'unavailable')
+        cmd = self.get_verify_command(signature_filename, data_filename,
+                                      keystore)
+        rc, stdout, stderr = self.run_command(cmd)
+        if rc not in (0, 1):
+            raise DistlibException('verify command failed with error code %s' % rc)
+        return rc == 0
+
+    def download_file(self, url, destfile, digest=None, reporthook=None):
+        """
+        This is a convenience method for downloading a file from an URL.
+        Normally, this will be a file from the index, though currently
+        no check is made for this (i.e. a file can be downloaded from
+        anywhere).
+
+        The method is just like the :func:`urlretrieve` function in the
+        standard library, except that it allows digest computation to be
+        done during download and checking that the downloaded data
+        matched any expected value.
+
+        :param url: The URL of the file to be downloaded (assumed to be
+                    available via an HTTP GET request).
+        :param destfile: The pathname where the downloaded file is to be
+                         saved.
+        :param digest: If specified, this must be a (hasher, value)
+                       tuple, where hasher is the algorithm used (e.g.
+                       ``'md5'``) and ``value`` is the expected value.
+        :param reporthook: The same as for :func:`urlretrieve` in the
+                           standard library.
+        """
+        if digest is None:
+            digester = None
+            logger.debug('No digest specified')
+        else:
+            if isinstance(digest, (list, tuple)):
+                hasher, digest = digest
+            else:
+                hasher = 'md5'
+            digester = getattr(hashlib, hasher)()
+            logger.debug('Digest specified: %s' % digest)
+        # The following code is equivalent to urlretrieve.
+        # We need to do it this way so that we can compute the
+        # digest of the file as we go.
+        with open(destfile, 'wb') as dfp:
+            # addinfourl is not a context manager on 2.x
+            # so we have to use try/finally
+            sfp = self.send_request(Request(url))
+            try:
+                headers = sfp.info()
+                blocksize = 8192
+                size = -1
+                read = 0
+                blocknum = 0
+                if "content-length" in headers:
+                    size = int(headers["Content-Length"])
+                if reporthook:
+                    reporthook(blocknum, blocksize, size)
+                while True:
+                    block = sfp.read(blocksize)
+                    if not block:
+                        break
+                    read += len(block)
+                    dfp.write(block)
+                    if digester:
+                        digester.update(block)
+                    blocknum += 1
+                    if reporthook:
+                        reporthook(blocknum, blocksize, size)
+            finally:
+                sfp.close()
+
+        # check that we got the whole file, if we can
+        if size >= 0 and read < size:
+            raise DistlibException(
+                'retrieval incomplete: got only %d out of %d bytes'
+                % (read, size))
+        # if we have a digest, it must match.
+        if digester:
+            actual = digester.hexdigest()
+            if digest != actual:
+                raise DistlibException('%s digest mismatch for %s: expected '
+                                       '%s, got %s' % (hasher, destfile,
+                                                       digest, actual))
+            logger.debug('Digest verified: %s', digest)
+
+    def send_request(self, req):
+        """
+        Send a standard library :class:`Request` to PyPI and return its
+        response.
+
+        :param req: The request to send.
+        :return: The HTTP response from PyPI (a standard library HTTPResponse).
+        """
+        handlers = []
+        if self.password_handler:
+            handlers.append(self.password_handler)
+        if self.ssl_verifier:
+            handlers.append(self.ssl_verifier)
+        opener = build_opener(*handlers)
+        return opener.open(req)
+
+    def encode_request(self, fields, files):
+        """
+        Encode fields and files for posting to an HTTP server.
+
+        :param fields: The fields to send as a list of (fieldname, value)
+                       tuples.
+        :param files: The files to send as a list of (fieldname, filename,
+                      file_bytes) tuple.
+        """
+        # Adapted from packaging, which in turn was adapted from
+        # http://code.activestate.com/recipes/146306
+
+        parts = []
+        boundary = self.boundary
+        for k, values in fields:
+            if not isinstance(values, (list, tuple)):
+                values = [values]
+
+            for v in values:
+                parts.extend((
+                    b'--' + boundary,
+                    ('Content-Disposition: form-data; name="%s"' %
+                     k).encode('utf-8'),
+                    b'',
+                    v.encode('utf-8')))
+        for key, filename, value in files:
+            parts.extend((
+                b'--' + boundary,
+                ('Content-Disposition: form-data; name="%s"; filename="%s"' %
+                 (key, filename)).encode('utf-8'),
+                b'',
+                value))
+
+        parts.extend((b'--' + boundary + b'--', b''))
+
+        body = b'\r\n'.join(parts)
+        ct = b'multipart/form-data; boundary=' + boundary
+        headers = {
+            'Content-type': ct,
+            'Content-length': str(len(body))
+        }
+        return Request(self.url, body, headers)
+
+    def search(self, terms, operator=None):  # pragma: no cover
+        if isinstance(terms, string_types):
+            terms = {'name': terms}
+        rpc_proxy = ServerProxy(self.url, timeout=3.0)
+        try:
+            return rpc_proxy.search(terms, operator or 'and')
+        finally:
+            rpc_proxy('close')()

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/locators.py

@@ -0,0 +1,1303 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+
+import gzip
+from io import BytesIO
+import json
+import logging
+import os
+import posixpath
+import re
+try:
+    import threading
+except ImportError:  # pragma: no cover
+    import dummy_threading as threading
+import zlib
+
+from . import DistlibException
+from .compat import (urljoin, urlparse, urlunparse, url2pathname, pathname2url,
+                     queue, quote, unescape, build_opener,
+                     HTTPRedirectHandler as BaseRedirectHandler, text_type,
+                     Request, HTTPError, URLError)
+from .database import Distribution, DistributionPath, make_dist
+from .metadata import Metadata, MetadataInvalidError
+from .util import (cached_property, ensure_slash, split_filename, get_project_data,
+                   parse_requirement, parse_name_and_version, ServerProxy,
+                   normalize_name)
+from .version import get_scheme, UnsupportedVersionError
+from .wheel import Wheel, is_compatible
+
+logger = logging.getLogger(__name__)
+
+HASHER_HASH = re.compile(r'^(\w+)=([a-f0-9]+)')
+CHARSET = re.compile(r';\s*charset\s*=\s*(.*)\s*$', re.I)
+HTML_CONTENT_TYPE = re.compile('text/html|application/x(ht)?ml')
+DEFAULT_INDEX = 'https://pypi.org/pypi'
+
+
+def get_all_distribution_names(url=None):
+    """
+    Return all distribution names known by an index.
+    :param url: The URL of the index.
+    :return: A list of all known distribution names.
+    """
+    if url is None:
+        url = DEFAULT_INDEX
+    client = ServerProxy(url, timeout=3.0)
+    try:
+        return client.list_packages()
+    finally:
+        client('close')()
+
+
+class RedirectHandler(BaseRedirectHandler):
+    """
+    A class to work around a bug in some Python 3.2.x releases.
+    """
+    # There's a bug in the base version for some 3.2.x
+    # (e.g. 3.2.2 on Ubuntu Oneiric). If a Location header
+    # returns e.g. /abc, it bails because it says the scheme ''
+    # is bogus, when actually it should use the request's
+    # URL for the scheme. See Python issue #13696.
+    def http_error_302(self, req, fp, code, msg, headers):
+        # Some servers (incorrectly) return multiple Location headers
+        # (so probably same goes for URI).  Use first header.
+        newurl = None
+        for key in ('location', 'uri'):
+            if key in headers:
+                newurl = headers[key]
+                break
+        if newurl is None:  # pragma: no cover
+            return
+        urlparts = urlparse(newurl)
+        if urlparts.scheme == '':
+            newurl = urljoin(req.get_full_url(), newurl)
+            if hasattr(headers, 'replace_header'):
+                headers.replace_header(key, newurl)
+            else:
+                headers[key] = newurl
+        return BaseRedirectHandler.http_error_302(self, req, fp, code, msg,
+                                                  headers)
+
+    http_error_301 = http_error_303 = http_error_307 = http_error_302
+
+
+class Locator(object):
+    """
+    A base class for locators - things that locate distributions.
+    """
+    source_extensions = ('.tar.gz', '.tar.bz2', '.tar', '.zip', '.tgz', '.tbz')
+    binary_extensions = ('.egg', '.exe', '.whl')
+    excluded_extensions = ('.pdf',)
+
+    # A list of tags indicating which wheels you want to match. The default
+    # value of None matches against the tags compatible with the running
+    # Python. If you want to match other values, set wheel_tags on a locator
+    # instance to a list of tuples (pyver, abi, arch) which you want to match.
+    wheel_tags = None
+
+    downloadable_extensions = source_extensions + ('.whl',)
+
+    def __init__(self, scheme='default'):
+        """
+        Initialise an instance.
+        :param scheme: Because locators look for most recent versions, they
+                       need to know the version scheme to use. This specifies
+                       the current PEP-recommended scheme - use ``'legacy'``
+                       if you need to support existing distributions on PyPI.
+        """
+        self._cache = {}
+        self.scheme = scheme
+        # Because of bugs in some of the handlers on some of the platforms,
+        # we use our own opener rather than just using urlopen.
+        self.opener = build_opener(RedirectHandler())
+        # If get_project() is called from locate(), the matcher instance
+        # is set from the requirement passed to locate(). See issue #18 for
+        # why this can be useful to know.
+        self.matcher = None
+        self.errors = queue.Queue()
+
+    def get_errors(self):
+        """
+        Return any errors which have occurred.
+        """
+        result = []
+        while not self.errors.empty():  # pragma: no cover
+            try:
+                e = self.errors.get(False)
+                result.append(e)
+            except self.errors.Empty:
+                continue
+            self.errors.task_done()
+        return result
+
+    def clear_errors(self):
+        """
+        Clear any errors which may have been logged.
+        """
+        # Just get the errors and throw them away
+        self.get_errors()
+
+    def clear_cache(self):
+        self._cache.clear()
+
+    def _get_scheme(self):
+        return self._scheme
+
+    def _set_scheme(self, value):
+        self._scheme = value
+
+    scheme = property(_get_scheme, _set_scheme)
+
+    def _get_project(self, name):
+        """
+        For a given project, get a dictionary mapping available versions to Distribution
+        instances.
+
+        This should be implemented in subclasses.
+
+        If called from a locate() request, self.matcher will be set to a
+        matcher for the requirement to satisfy, otherwise it will be None.
+        """
+        raise NotImplementedError('Please implement in the subclass')
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        raise NotImplementedError('Please implement in the subclass')
+
+    def get_project(self, name):
+        """
+        For a given project, get a dictionary mapping available versions to Distribution
+        instances.
+
+        This calls _get_project to do all the work, and just implements a caching layer on top.
+        """
+        if self._cache is None:  # pragma: no cover
+            result = self._get_project(name)
+        elif name in self._cache:
+            result = self._cache[name]
+        else:
+            self.clear_errors()
+            result = self._get_project(name)
+            self._cache[name] = result
+        return result
+
+    def score_url(self, url):
+        """
+        Give an url a score which can be used to choose preferred URLs
+        for a given project release.
+        """
+        t = urlparse(url)
+        basename = posixpath.basename(t.path)
+        compatible = True
+        is_wheel = basename.endswith('.whl')
+        is_downloadable = basename.endswith(self.downloadable_extensions)
+        if is_wheel:
+            compatible = is_compatible(Wheel(basename), self.wheel_tags)
+        return (t.scheme == 'https', 'pypi.org' in t.netloc,
+                is_downloadable, is_wheel, compatible, basename)
+
+    def prefer_url(self, url1, url2):
+        """
+        Choose one of two URLs where both are candidates for distribution
+        archives for the same version of a distribution (for example,
+        .tar.gz vs. zip).
+
+        The current implementation favours https:// URLs over http://, archives
+        from PyPI over those from other locations, wheel compatibility (if a
+        wheel) and then the archive name.
+        """
+        result = url2
+        if url1:
+            s1 = self.score_url(url1)
+            s2 = self.score_url(url2)
+            if s1 > s2:
+                result = url1
+            if result != url2:
+                logger.debug('Not replacing %r with %r', url1, url2)
+            else:
+                logger.debug('Replacing %r with %r', url1, url2)
+        return result
+
+    def split_filename(self, filename, project_name):
+        """
+        Attempt to split a filename in project name, version and Python version.
+        """
+        return split_filename(filename, project_name)
+
+    def convert_url_to_download_info(self, url, project_name):
+        """
+        See if a URL is a candidate for a download URL for a project (the URL
+        has typically been scraped from an HTML page).
+
+        If it is, a dictionary is returned with keys "name", "version",
+        "filename" and "url"; otherwise, None is returned.
+        """
+        def same_project(name1, name2):
+            return normalize_name(name1) == normalize_name(name2)
+
+        result = None
+        scheme, netloc, path, params, query, frag = urlparse(url)
+        if frag.lower().startswith('egg='):  # pragma: no cover
+            logger.debug('%s: version hint in fragment: %r',
+                         project_name, frag)
+        m = HASHER_HASH.match(frag)
+        if m:
+            algo, digest = m.groups()
+        else:
+            algo, digest = None, None
+        origpath = path
+        if path and path[-1] == '/':  # pragma: no cover
+            path = path[:-1]
+        if path.endswith('.whl'):
+            try:
+                wheel = Wheel(path)
+                if not is_compatible(wheel, self.wheel_tags):
+                    logger.debug('Wheel not compatible: %s', path)
+                else:
+                    if project_name is None:
+                        include = True
+                    else:
+                        include = same_project(wheel.name, project_name)
+                    if include:
+                        result = {
+                            'name': wheel.name,
+                            'version': wheel.version,
+                            'filename': wheel.filename,
+                            'url': urlunparse((scheme, netloc, origpath,
+                                               params, query, '')),
+                            'python-version': ', '.join(
+                                ['.'.join(list(v[2:])) for v in wheel.pyver]),
+                        }
+            except Exception:  # pragma: no cover
+                logger.warning('invalid path for wheel: %s', path)
+        elif not path.endswith(self.downloadable_extensions):  # pragma: no cover
+            logger.debug('Not downloadable: %s', path)
+        else:  # downloadable extension
+            path = filename = posixpath.basename(path)
+            for ext in self.downloadable_extensions:
+                if path.endswith(ext):
+                    path = path[:-len(ext)]
+                    t = self.split_filename(path, project_name)
+                    if not t:  # pragma: no cover
+                        logger.debug('No match for project/version: %s', path)
+                    else:
+                        name, version, pyver = t
+                        if not project_name or same_project(project_name, name):
+                            result = {
+                                'name': name,
+                                'version': version,
+                                'filename': filename,
+                                'url': urlunparse((scheme, netloc, origpath,
+                                                   params, query, '')),
+                            }
+                            if pyver:  # pragma: no cover
+                                result['python-version'] = pyver
+                    break
+        if result and algo:
+            result['%s_digest' % algo] = digest
+        return result
+
+    def _get_digest(self, info):
+        """
+        Get a digest from a dictionary by looking at a "digests" dictionary
+        or keys of the form 'algo_digest'.
+
+        Returns a 2-tuple (algo, digest) if found, else None. Currently
+        looks only for SHA256, then MD5.
+        """
+        result = None
+        if 'digests' in info:
+            digests = info['digests']
+            for algo in ('sha256', 'md5'):
+                if algo in digests:
+                    result = (algo, digests[algo])
+                    break
+        if not result:
+            for algo in ('sha256', 'md5'):
+                key = '%s_digest' % algo
+                if key in info:
+                    result = (algo, info[key])
+                    break
+        return result
+
+    def _update_version_data(self, result, info):
+        """
+        Update a result dictionary (the final result from _get_project) with a
+        dictionary for a specific version, which typically holds information
+        gleaned from a filename or URL for an archive for the distribution.
+        """
+        name = info.pop('name')
+        version = info.pop('version')
+        if version in result:
+            dist = result[version]
+            md = dist.metadata
+        else:
+            dist = make_dist(name, version, scheme=self.scheme)
+            md = dist.metadata
+        dist.digest = digest = self._get_digest(info)
+        url = info['url']
+        result['digests'][url] = digest
+        if md.source_url != info['url']:
+            md.source_url = self.prefer_url(md.source_url, url)
+            result['urls'].setdefault(version, set()).add(url)
+        dist.locator = self
+        result[version] = dist
+
+    def locate(self, requirement, prereleases=False):
+        """
+        Find the most recent distribution which matches the given
+        requirement.
+
+        :param requirement: A requirement of the form 'foo (1.0)' or perhaps
+                            'foo (>= 1.0, < 2.0, != 1.3)'
+        :param prereleases: If ``True``, allow pre-release versions
+                            to be located. Otherwise, pre-release versions
+                            are not returned.
+        :return: A :class:`Distribution` instance, or ``None`` if no such
+                 distribution could be located.
+        """
+        result = None
+        r = parse_requirement(requirement)
+        if r is None:  # pragma: no cover
+            raise DistlibException('Not a valid requirement: %r' % requirement)
+        scheme = get_scheme(self.scheme)
+        self.matcher = matcher = scheme.matcher(r.requirement)
+        logger.debug('matcher: %s (%s)', matcher, type(matcher).__name__)
+        versions = self.get_project(r.name)
+        if len(versions) > 2:   # urls and digests keys are present
+            # sometimes, versions are invalid
+            slist = []
+            vcls = matcher.version_class
+            for k in versions:
+                if k in ('urls', 'digests'):
+                    continue
+                try:
+                    if not matcher.match(k):
+                        pass  # logger.debug('%s did not match %r', matcher, k)
+                    else:
+                        if prereleases or not vcls(k).is_prerelease:
+                            slist.append(k)
+                except Exception:  # pragma: no cover
+                    logger.warning('error matching %s with %r', matcher, k)
+                    pass  # slist.append(k)
+            if len(slist) > 1:
+                slist = sorted(slist, key=scheme.key)
+            if slist:
+                logger.debug('sorted list: %s', slist)
+                version = slist[-1]
+                result = versions[version]
+        if result:
+            if r.extras:
+                result.extras = r.extras
+            result.download_urls = versions.get('urls', {}).get(version, set())
+            d = {}
+            sd = versions.get('digests', {})
+            for url in result.download_urls:
+                if url in sd:  # pragma: no cover
+                    d[url] = sd[url]
+            result.digests = d
+        self.matcher = None
+        return result
+
+
+class PyPIRPCLocator(Locator):
+    """
+    This locator uses XML-RPC to locate distributions. It therefore
+    cannot be used with simple mirrors (that only mirror file content).
+    """
+    def __init__(self, url, **kwargs):
+        """
+        Initialise an instance.
+
+        :param url: The URL to use for XML-RPC.
+        :param kwargs: Passed to the superclass constructor.
+        """
+        super(PyPIRPCLocator, self).__init__(**kwargs)
+        self.base_url = url
+        self.client = ServerProxy(url, timeout=3.0)
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        return set(self.client.list_packages())
+
+    def _get_project(self, name):
+        result = {'urls': {}, 'digests': {}}
+        versions = self.client.package_releases(name, True)
+        for v in versions:
+            urls = self.client.release_urls(name, v)
+            data = self.client.release_data(name, v)
+            metadata = Metadata(scheme=self.scheme)
+            metadata.name = data['name']
+            metadata.version = data['version']
+            metadata.license = data.get('license')
+            metadata.keywords = data.get('keywords', [])
+            metadata.summary = data.get('summary')
+            dist = Distribution(metadata)
+            if urls:
+                info = urls[0]
+                metadata.source_url = info['url']
+                dist.digest = self._get_digest(info)
+                dist.locator = self
+                result[v] = dist
+                for info in urls:
+                    url = info['url']
+                    digest = self._get_digest(info)
+                    result['urls'].setdefault(v, set()).add(url)
+                    result['digests'][url] = digest
+        return result
+
+
+class PyPIJSONLocator(Locator):
+    """
+    This locator uses PyPI's JSON interface. It's very limited in functionality
+    and probably not worth using.
+    """
+    def __init__(self, url, **kwargs):
+        super(PyPIJSONLocator, self).__init__(**kwargs)
+        self.base_url = ensure_slash(url)
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        raise NotImplementedError('Not available from this locator')
+
+    def _get_project(self, name):
+        result = {'urls': {}, 'digests': {}}
+        url = urljoin(self.base_url, '%s/json' % quote(name))
+        try:
+            resp = self.opener.open(url)
+            data = resp.read().decode()  # for now
+            d = json.loads(data)
+            md = Metadata(scheme=self.scheme)
+            data = d['info']
+            md.name = data['name']
+            md.version = data['version']
+            md.license = data.get('license')
+            md.keywords = data.get('keywords', [])
+            md.summary = data.get('summary')
+            dist = Distribution(md)
+            dist.locator = self
+            # urls = d['urls']
+            result[md.version] = dist
+            for info in d['urls']:
+                url = info['url']
+                dist.download_urls.add(url)
+                dist.digests[url] = self._get_digest(info)
+                result['urls'].setdefault(md.version, set()).add(url)
+                result['digests'][url] = self._get_digest(info)
+            # Now get other releases
+            for version, infos in d['releases'].items():
+                if version == md.version:
+                    continue    # already done
+                omd = Metadata(scheme=self.scheme)
+                omd.name = md.name
+                omd.version = version
+                odist = Distribution(omd)
+                odist.locator = self
+                result[version] = odist
+                for info in infos:
+                    url = info['url']
+                    odist.download_urls.add(url)
+                    odist.digests[url] = self._get_digest(info)
+                    result['urls'].setdefault(version, set()).add(url)
+                    result['digests'][url] = self._get_digest(info)
+#            for info in urls:
+#                md.source_url = info['url']
+#                dist.digest = self._get_digest(info)
+#                dist.locator = self
+#                for info in urls:
+#                    url = info['url']
+#                    result['urls'].setdefault(md.version, set()).add(url)
+#                    result['digests'][url] = self._get_digest(info)
+        except Exception as e:
+            self.errors.put(text_type(e))
+            logger.exception('JSON fetch failed: %s', e)
+        return result
+
+
+class Page(object):
+    """
+    This class represents a scraped HTML page.
+    """
+    # The following slightly hairy-looking regex just looks for the contents of
+    # an anchor link, which has an attribute "href" either immediately preceded
+    # or immediately followed by a "rel" attribute. The attribute values can be
+    # declared with double quotes, single quotes or no quotes - which leads to
+    # the length of the expression.
+    _href = re.compile("""
+(rel\\s*=\\s*(?:"(?P<rel1>[^"]*)"|'(?P<rel2>[^']*)'|(?P<rel3>[^>\\s\n]*))\\s+)?
+href\\s*=\\s*(?:"(?P<url1>[^"]*)"|'(?P<url2>[^']*)'|(?P<url3>[^>\\s\n]*))
+(\\s+rel\\s*=\\s*(?:"(?P<rel4>[^"]*)"|'(?P<rel5>[^']*)'|(?P<rel6>[^>\\s\n]*)))?
+""", re.I | re.S | re.X)
+    _base = re.compile(r"""<base\s+href\s*=\s*['"]?([^'">]+)""", re.I | re.S)
+
+    def __init__(self, data, url):
+        """
+        Initialise an instance with the Unicode page contents and the URL they
+        came from.
+        """
+        self.data = data
+        self.base_url = self.url = url
+        m = self._base.search(self.data)
+        if m:
+            self.base_url = m.group(1)
+
+    _clean_re = re.compile(r'[^a-z0-9$&+,/:;=?@.#%_\\|-]', re.I)
+
+    @cached_property
+    def links(self):
+        """
+        Return the URLs of all the links on a page together with information
+        about their "rel" attribute, for determining which ones to treat as
+        downloads and which ones to queue for further scraping.
+        """
+        def clean(url):
+            "Tidy up an URL."
+            scheme, netloc, path, params, query, frag = urlparse(url)
+            return urlunparse((scheme, netloc, quote(path),
+                               params, query, frag))
+
+        result = set()
+        for match in self._href.finditer(self.data):
+            d = match.groupdict('')
+            rel = (d['rel1'] or d['rel2'] or d['rel3'] or
+                   d['rel4'] or d['rel5'] or d['rel6'])
+            url = d['url1'] or d['url2'] or d['url3']
+            url = urljoin(self.base_url, url)
+            url = unescape(url)
+            url = self._clean_re.sub(lambda m: '%%%2x' % ord(m.group(0)), url)
+            result.add((url, rel))
+        # We sort the result, hoping to bring the most recent versions
+        # to the front
+        result = sorted(result, key=lambda t: t[0], reverse=True)
+        return result
+
+
+class SimpleScrapingLocator(Locator):
+    """
+    A locator which scrapes HTML pages to locate downloads for a distribution.
+    This runs multiple threads to do the I/O; performance is at least as good
+    as pip's PackageFinder, which works in an analogous fashion.
+    """
+
+    # These are used to deal with various Content-Encoding schemes.
+    decoders = {
+        'deflate': zlib.decompress,
+        'gzip': lambda b: gzip.GzipFile(fileobj=BytesIO(b)).read(),
+        'none': lambda b: b,
+    }
+
+    def __init__(self, url, timeout=None, num_workers=10, **kwargs):
+        """
+        Initialise an instance.
+        :param url: The root URL to use for scraping.
+        :param timeout: The timeout, in seconds, to be applied to requests.
+                        This defaults to ``None`` (no timeout specified).
+        :param num_workers: The number of worker threads you want to do I/O,
+                            This defaults to 10.
+        :param kwargs: Passed to the superclass.
+        """
+        super(SimpleScrapingLocator, self).__init__(**kwargs)
+        self.base_url = ensure_slash(url)
+        self.timeout = timeout
+        self._page_cache = {}
+        self._seen = set()
+        self._to_fetch = queue.Queue()
+        self._bad_hosts = set()
+        self.skip_externals = False
+        self.num_workers = num_workers
+        self._lock = threading.RLock()
+        # See issue #45: we need to be resilient when the locator is used
+        # in a thread, e.g. with concurrent.futures. We can't use self._lock
+        # as it is for coordinating our internal threads - the ones created
+        # in _prepare_threads.
+        self._gplock = threading.RLock()
+        self.platform_check = False  # See issue #112
+
+    def _prepare_threads(self):
+        """
+        Threads are created only when get_project is called, and terminate
+        before it returns. They are there primarily to parallelise I/O (i.e.
+        fetching web pages).
+        """
+        self._threads = []
+        for i in range(self.num_workers):
+            t = threading.Thread(target=self._fetch)
+            t.daemon = True
+            t.start()
+            self._threads.append(t)
+
+    def _wait_threads(self):
+        """
+        Tell all the threads to terminate (by sending a sentinel value) and
+        wait for them to do so.
+        """
+        # Note that you need two loops, since you can't say which
+        # thread will get each sentinel
+        for t in self._threads:
+            self._to_fetch.put(None)    # sentinel
+        for t in self._threads:
+            t.join()
+        self._threads = []
+
+    def _get_project(self, name):
+        result = {'urls': {}, 'digests': {}}
+        with self._gplock:
+            self.result = result
+            self.project_name = name
+            url = urljoin(self.base_url, '%s/' % quote(name))
+            self._seen.clear()
+            self._page_cache.clear()
+            self._prepare_threads()
+            try:
+                logger.debug('Queueing %s', url)
+                self._to_fetch.put(url)
+                self._to_fetch.join()
+            finally:
+                self._wait_threads()
+            del self.result
+        return result
+
+    platform_dependent = re.compile(r'\b(linux_(i\d86|x86_64|arm\w+)|'
+                                    r'win(32|_amd64)|macosx_?\d+)\b', re.I)
+
+    def _is_platform_dependent(self, url):
+        """
+        Does an URL refer to a platform-specific download?
+        """
+        return self.platform_dependent.search(url)
+
+    def _process_download(self, url):
+        """
+        See if an URL is a suitable download for a project.
+
+        If it is, register information in the result dictionary (for
+        _get_project) about the specific version it's for.
+
+        Note that the return value isn't actually used other than as a boolean
+        value.
+        """
+        if self.platform_check and self._is_platform_dependent(url):
+            info = None
+        else:
+            info = self.convert_url_to_download_info(url, self.project_name)
+        logger.debug('process_download: %s -> %s', url, info)
+        if info:
+            with self._lock:    # needed because self.result is shared
+                self._update_version_data(self.result, info)
+        return info
+
+    def _should_queue(self, link, referrer, rel):
+        """
+        Determine whether a link URL from a referring page and with a
+        particular "rel" attribute should be queued for scraping.
+        """
+        scheme, netloc, path, _, _, _ = urlparse(link)
+        if path.endswith(self.source_extensions + self.binary_extensions +
+                         self.excluded_extensions):
+            result = False
+        elif self.skip_externals and not link.startswith(self.base_url):
+            result = False
+        elif not referrer.startswith(self.base_url):
+            result = False
+        elif rel not in ('homepage', 'download'):
+            result = False
+        elif scheme not in ('http', 'https', 'ftp'):
+            result = False
+        elif self._is_platform_dependent(link):
+            result = False
+        else:
+            host = netloc.split(':', 1)[0]
+            if host.lower() == 'localhost':
+                result = False
+            else:
+                result = True
+        logger.debug('should_queue: %s (%s) from %s -> %s', link, rel,
+                     referrer, result)
+        return result
+
+    def _fetch(self):
+        """
+        Get a URL to fetch from the work queue, get the HTML page, examine its
+        links for download candidates and candidates for further scraping.
+
+        This is a handy method to run in a thread.
+        """
+        while True:
+            url = self._to_fetch.get()
+            try:
+                if url:
+                    page = self.get_page(url)
+                    if page is None:    # e.g. after an error
+                        continue
+                    for link, rel in page.links:
+                        if link not in self._seen:
+                            try:
+                                self._seen.add(link)
+                                if (not self._process_download(link) and
+                                        self._should_queue(link, url, rel)):
+                                    logger.debug('Queueing %s from %s', link, url)
+                                    self._to_fetch.put(link)
+                            except MetadataInvalidError:  # e.g. invalid versions
+                                pass
+            except Exception as e:  # pragma: no cover
+                self.errors.put(text_type(e))
+            finally:
+                # always do this, to avoid hangs :-)
+                self._to_fetch.task_done()
+            if not url:
+                # logger.debug('Sentinel seen, quitting.')
+                break
+
+    def get_page(self, url):
+        """
+        Get the HTML for an URL, possibly from an in-memory cache.
+
+        XXX TODO Note: this cache is never actually cleared. It's assumed that
+        the data won't get stale over the lifetime of a locator instance (not
+        necessarily true for the default_locator).
+        """
+        # http://peak.telecommunity.com/DevCenter/EasyInstall#package-index-api
+        scheme, netloc, path, _, _, _ = urlparse(url)
+        if scheme == 'file' and os.path.isdir(url2pathname(path)):
+            url = urljoin(ensure_slash(url), 'index.html')
+
+        if url in self._page_cache:
+            result = self._page_cache[url]
+            logger.debug('Returning %s from cache: %s', url, result)
+        else:
+            host = netloc.split(':', 1)[0]
+            result = None
+            if host in self._bad_hosts:
+                logger.debug('Skipping %s due to bad host %s', url, host)
+            else:
+                req = Request(url, headers={'Accept-encoding': 'identity'})
+                try:
+                    logger.debug('Fetching %s', url)
+                    resp = self.opener.open(req, timeout=self.timeout)
+                    logger.debug('Fetched %s', url)
+                    headers = resp.info()
+                    content_type = headers.get('Content-Type', '')
+                    if HTML_CONTENT_TYPE.match(content_type):
+                        final_url = resp.geturl()
+                        data = resp.read()
+                        encoding = headers.get('Content-Encoding')
+                        if encoding:
+                            decoder = self.decoders[encoding]   # fail if not found
+                            data = decoder(data)
+                        encoding = 'utf-8'
+                        m = CHARSET.search(content_type)
+                        if m:
+                            encoding = m.group(1)
+                        try:
+                            data = data.decode(encoding)
+                        except UnicodeError:  # pragma: no cover
+                            data = data.decode('latin-1')    # fallback
+                        result = Page(data, final_url)
+                        self._page_cache[final_url] = result
+                except HTTPError as e:
+                    if e.code != 404:
+                        logger.exception('Fetch failed: %s: %s', url, e)
+                except URLError as e:  # pragma: no cover
+                    logger.exception('Fetch failed: %s: %s', url, e)
+                    with self._lock:
+                        self._bad_hosts.add(host)
+                except Exception as e:  # pragma: no cover
+                    logger.exception('Fetch failed: %s: %s', url, e)
+                finally:
+                    self._page_cache[url] = result   # even if None (failure)
+        return result
+
+    _distname_re = re.compile('<a href=[^>]*>([^<]+)<')
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        result = set()
+        page = self.get_page(self.base_url)
+        if not page:
+            raise DistlibException('Unable to get %s' % self.base_url)
+        for match in self._distname_re.finditer(page.data):
+            result.add(match.group(1))
+        return result
+
+
+class DirectoryLocator(Locator):
+    """
+    This class locates distributions in a directory tree.
+    """
+
+    def __init__(self, path, **kwargs):
+        """
+        Initialise an instance.
+        :param path: The root of the directory tree to search.
+        :param kwargs: Passed to the superclass constructor,
+                       except for:
+                       * recursive - if True (the default), subdirectories are
+                         recursed into. If False, only the top-level directory
+                         is searched,
+        """
+        self.recursive = kwargs.pop('recursive', True)
+        super(DirectoryLocator, self).__init__(**kwargs)
+        path = os.path.abspath(path)
+        if not os.path.isdir(path):  # pragma: no cover
+            raise DistlibException('Not a directory: %r' % path)
+        self.base_dir = path
+
+    def should_include(self, filename, parent):
+        """
+        Should a filename be considered as a candidate for a distribution
+        archive? As well as the filename, the directory which contains it
+        is provided, though not used by the current implementation.
+        """
+        return filename.endswith(self.downloadable_extensions)
+
+    def _get_project(self, name):
+        result = {'urls': {}, 'digests': {}}
+        for root, dirs, files in os.walk(self.base_dir):
+            for fn in files:
+                if self.should_include(fn, root):
+                    fn = os.path.join(root, fn)
+                    url = urlunparse(('file', '',
+                                      pathname2url(os.path.abspath(fn)),
+                                      '', '', ''))
+                    info = self.convert_url_to_download_info(url, name)
+                    if info:
+                        self._update_version_data(result, info)
+            if not self.recursive:
+                break
+        return result
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        result = set()
+        for root, dirs, files in os.walk(self.base_dir):
+            for fn in files:
+                if self.should_include(fn, root):
+                    fn = os.path.join(root, fn)
+                    url = urlunparse(('file', '',
+                                      pathname2url(os.path.abspath(fn)),
+                                      '', '', ''))
+                    info = self.convert_url_to_download_info(url, None)
+                    if info:
+                        result.add(info['name'])
+            if not self.recursive:
+                break
+        return result
+
+
+class JSONLocator(Locator):
+    """
+    This locator uses special extended metadata (not available on PyPI) and is
+    the basis of performant dependency resolution in distlib. Other locators
+    require archive downloads before dependencies can be determined! As you
+    might imagine, that can be slow.
+    """
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        raise NotImplementedError('Not available from this locator')
+
+    def _get_project(self, name):
+        result = {'urls': {}, 'digests': {}}
+        data = get_project_data(name)
+        if data:
+            for info in data.get('files', []):
+                if info['ptype'] != 'sdist' or info['pyversion'] != 'source':
+                    continue
+                # We don't store summary in project metadata as it makes
+                # the data bigger for no benefit during dependency
+                # resolution
+                dist = make_dist(data['name'], info['version'],
+                                 summary=data.get('summary',
+                                                  'Placeholder for summary'),
+                                 scheme=self.scheme)
+                md = dist.metadata
+                md.source_url = info['url']
+                # TODO SHA256 digest
+                if 'digest' in info and info['digest']:
+                    dist.digest = ('md5', info['digest'])
+                md.dependencies = info.get('requirements', {})
+                dist.exports = info.get('exports', {})
+                result[dist.version] = dist
+                result['urls'].setdefault(dist.version, set()).add(info['url'])
+        return result
+
+
+class DistPathLocator(Locator):
+    """
+    This locator finds installed distributions in a path. It can be useful for
+    adding to an :class:`AggregatingLocator`.
+    """
+    def __init__(self, distpath, **kwargs):
+        """
+        Initialise an instance.
+
+        :param distpath: A :class:`DistributionPath` instance to search.
+        """
+        super(DistPathLocator, self).__init__(**kwargs)
+        assert isinstance(distpath, DistributionPath)
+        self.distpath = distpath
+
+    def _get_project(self, name):
+        dist = self.distpath.get_distribution(name)
+        if dist is None:
+            result = {'urls': {}, 'digests': {}}
+        else:
+            result = {
+                dist.version: dist,
+                'urls': {dist.version: set([dist.source_url])},
+                'digests': {dist.version: set([None])}
+            }
+        return result
+
+
+class AggregatingLocator(Locator):
+    """
+    This class allows you to chain and/or merge a list of locators.
+    """
+    def __init__(self, *locators, **kwargs):
+        """
+        Initialise an instance.
+
+        :param locators: The list of locators to search.
+        :param kwargs: Passed to the superclass constructor,
+                       except for:
+                       * merge - if False (the default), the first successful
+                         search from any of the locators is returned. If True,
+                         the results from all locators are merged (this can be
+                         slow).
+        """
+        self.merge = kwargs.pop('merge', False)
+        self.locators = locators
+        super(AggregatingLocator, self).__init__(**kwargs)
+
+    def clear_cache(self):
+        super(AggregatingLocator, self).clear_cache()
+        for locator in self.locators:
+            locator.clear_cache()
+
+    def _set_scheme(self, value):
+        self._scheme = value
+        for locator in self.locators:
+            locator.scheme = value
+
+    scheme = property(Locator.scheme.fget, _set_scheme)
+
+    def _get_project(self, name):
+        result = {}
+        for locator in self.locators:
+            d = locator.get_project(name)
+            if d:
+                if self.merge:
+                    files = result.get('urls', {})
+                    digests = result.get('digests', {})
+                    # next line could overwrite result['urls'], result['digests']
+                    result.update(d)
+                    df = result.get('urls')
+                    if files and df:
+                        for k, v in files.items():
+                            if k in df:
+                                df[k] |= v
+                            else:
+                                df[k] = v
+                    dd = result.get('digests')
+                    if digests and dd:
+                        dd.update(digests)
+                else:
+                    # See issue #18. If any dists are found and we're looking
+                    # for specific constraints, we only return something if
+                    # a match is found. For example, if a DirectoryLocator
+                    # returns just foo (1.0) while we're looking for
+                    # foo (>= 2.0), we'll pretend there was nothing there so
+                    # that subsequent locators can be queried. Otherwise we
+                    # would just return foo (1.0) which would then lead to a
+                    # failure to find foo (>= 2.0), because other locators
+                    # weren't searched. Note that this only matters when
+                    # merge=False.
+                    if self.matcher is None:
+                        found = True
+                    else:
+                        found = False
+                        for k in d:
+                            if self.matcher.match(k):
+                                found = True
+                                break
+                    if found:
+                        result = d
+                        break
+        return result
+
+    def get_distribution_names(self):
+        """
+        Return all the distribution names known to this locator.
+        """
+        result = set()
+        for locator in self.locators:
+            try:
+                result |= locator.get_distribution_names()
+            except NotImplementedError:
+                pass
+        return result
+
+
+# We use a legacy scheme simply because most of the dists on PyPI use legacy
+# versions which don't conform to PEP 440.
+default_locator = AggregatingLocator(
+                    # JSONLocator(), # don't use as PEP 426 is withdrawn
+                    SimpleScrapingLocator('https://pypi.org/simple/',
+                                          timeout=3.0),
+                    scheme='legacy')
+
+locate = default_locator.locate
+
+
+class DependencyFinder(object):
+    """
+    Locate dependencies for distributions.
+    """
+
+    def __init__(self, locator=None):
+        """
+        Initialise an instance, using the specified locator
+        to locate distributions.
+        """
+        self.locator = locator or default_locator
+        self.scheme = get_scheme(self.locator.scheme)
+
+    def add_distribution(self, dist):
+        """
+        Add a distribution to the finder. This will update internal information
+        about who provides what.
+        :param dist: The distribution to add.
+        """
+        logger.debug('adding distribution %s', dist)
+        name = dist.key
+        self.dists_by_name[name] = dist
+        self.dists[(name, dist.version)] = dist
+        for p in dist.provides:
+            name, version = parse_name_and_version(p)
+            logger.debug('Add to provided: %s, %s, %s', name, version, dist)
+            self.provided.setdefault(name, set()).add((version, dist))
+
+    def remove_distribution(self, dist):
+        """
+        Remove a distribution from the finder. This will update internal
+        information about who provides what.
+        :param dist: The distribution to remove.
+        """
+        logger.debug('removing distribution %s', dist)
+        name = dist.key
+        del self.dists_by_name[name]
+        del self.dists[(name, dist.version)]
+        for p in dist.provides:
+            name, version = parse_name_and_version(p)
+            logger.debug('Remove from provided: %s, %s, %s', name, version, dist)
+            s = self.provided[name]
+            s.remove((version, dist))
+            if not s:
+                del self.provided[name]
+
+    def get_matcher(self, reqt):
+        """
+        Get a version matcher for a requirement.
+        :param reqt: The requirement
+        :type reqt: str
+        :return: A version matcher (an instance of
+                 :class:`distlib.version.Matcher`).
+        """
+        try:
+            matcher = self.scheme.matcher(reqt)
+        except UnsupportedVersionError:  # pragma: no cover
+            # XXX compat-mode if cannot read the version
+            name = reqt.split()[0]
+            matcher = self.scheme.matcher(name)
+        return matcher
+
+    def find_providers(self, reqt):
+        """
+        Find the distributions which can fulfill a requirement.
+
+        :param reqt: The requirement.
+         :type reqt: str
+        :return: A set of distribution which can fulfill the requirement.
+        """
+        matcher = self.get_matcher(reqt)
+        name = matcher.key   # case-insensitive
+        result = set()
+        provided = self.provided
+        if name in provided:
+            for version, provider in provided[name]:
+                try:
+                    match = matcher.match(version)
+                except UnsupportedVersionError:
+                    match = False
+
+                if match:
+                    result.add(provider)
+                    break
+        return result
+
+    def try_to_replace(self, provider, other, problems):
+        """
+        Attempt to replace one provider with another. This is typically used
+        when resolving dependencies from multiple sources, e.g. A requires
+        (B >= 1.0) while C requires (B >= 1.1).
+
+        For successful replacement, ``provider`` must meet all the requirements
+        which ``other`` fulfills.
+
+        :param provider: The provider we are trying to replace with.
+        :param other: The provider we're trying to replace.
+        :param problems: If False is returned, this will contain what
+                         problems prevented replacement. This is currently
+                         a tuple of the literal string 'cantreplace',
+                         ``provider``, ``other``  and the set of requirements
+                         that ``provider`` couldn't fulfill.
+        :return: True if we can replace ``other`` with ``provider``, else
+                 False.
+        """
+        rlist = self.reqts[other]
+        unmatched = set()
+        for s in rlist:
+            matcher = self.get_matcher(s)
+            if not matcher.match(provider.version):
+                unmatched.add(s)
+        if unmatched:
+            # can't replace other with provider
+            problems.add(('cantreplace', provider, other,
+                          frozenset(unmatched)))
+            result = False
+        else:
+            # can replace other with provider
+            self.remove_distribution(other)
+            del self.reqts[other]
+            for s in rlist:
+                self.reqts.setdefault(provider, set()).add(s)
+            self.add_distribution(provider)
+            result = True
+        return result
+
+    def find(self, requirement, meta_extras=None, prereleases=False):
+        """
+        Find a distribution and all distributions it depends on.
+
+        :param requirement: The requirement specifying the distribution to
+                            find, or a Distribution instance.
+        :param meta_extras: A list of meta extras such as :test:, :build: and
+                            so on.
+        :param prereleases: If ``True``, allow pre-release versions to be
+                            returned - otherwise, don't return prereleases
+                            unless they're all that's available.
+
+        Return a set of :class:`Distribution` instances and a set of
+        problems.
+
+        The distributions returned should be such that they have the
+        :attr:`required` attribute set to ``True`` if they were
+        from the ``requirement`` passed to ``find()``, and they have the
+        :attr:`build_time_dependency` attribute set to ``True`` unless they
+        are post-installation dependencies of the ``requirement``.
+
+        The problems should be a tuple consisting of the string
+        ``'unsatisfied'`` and the requirement which couldn't be satisfied
+        by any distribution known to the locator.
+        """
+
+        self.provided = {}
+        self.dists = {}
+        self.dists_by_name = {}
+        self.reqts = {}
+
+        meta_extras = set(meta_extras or [])
+        if ':*:' in meta_extras:
+            meta_extras.remove(':*:')
+            # :meta: and :run: are implicitly included
+            meta_extras |= set([':test:', ':build:', ':dev:'])
+
+        if isinstance(requirement, Distribution):
+            dist = odist = requirement
+            logger.debug('passed %s as requirement', odist)
+        else:
+            dist = odist = self.locator.locate(requirement,
+                                               prereleases=prereleases)
+            if dist is None:
+                raise DistlibException('Unable to locate %r' % requirement)
+            logger.debug('located %s', odist)
+        dist.requested = True
+        problems = set()
+        todo = set([dist])
+        install_dists = set([odist])
+        while todo:
+            dist = todo.pop()
+            name = dist.key     # case-insensitive
+            if name not in self.dists_by_name:
+                self.add_distribution(dist)
+            else:
+                # import pdb; pdb.set_trace()
+                other = self.dists_by_name[name]
+                if other != dist:
+                    self.try_to_replace(dist, other, problems)
+
+            ireqts = dist.run_requires | dist.meta_requires
+            sreqts = dist.build_requires
+            ereqts = set()
+            if meta_extras and dist in install_dists:
+                for key in ('test', 'build', 'dev'):
+                    e = ':%s:' % key
+                    if e in meta_extras:
+                        ereqts |= getattr(dist, '%s_requires' % key)
+            all_reqts = ireqts | sreqts | ereqts
+            for r in all_reqts:
+                providers = self.find_providers(r)
+                if not providers:
+                    logger.debug('No providers found for %r', r)
+                    provider = self.locator.locate(r, prereleases=prereleases)
+                    # If no provider is found and we didn't consider
+                    # prereleases, consider them now.
+                    if provider is None and not prereleases:
+                        provider = self.locator.locate(r, prereleases=True)
+                    if provider is None:
+                        logger.debug('Cannot satisfy %r', r)
+                        problems.add(('unsatisfied', r))
+                    else:
+                        n, v = provider.key, provider.version
+                        if (n, v) not in self.dists:
+                            todo.add(provider)
+                        providers.add(provider)
+                        if r in ireqts and dist in install_dists:
+                            install_dists.add(provider)
+                            logger.debug('Adding %s to install_dists',
+                                         provider.name_and_version)
+                for p in providers:
+                    name = p.key
+                    if name not in self.dists_by_name:
+                        self.reqts.setdefault(p, set()).add(r)
+                    else:
+                        other = self.dists_by_name[name]
+                        if other != p:
+                            # see if other can be replaced by p
+                            self.try_to_replace(p, other, problems)
+
+        dists = set(self.dists.values())
+        for dist in dists:
+            dist.build_time_dependency = dist not in install_dists
+            if dist.build_time_dependency:
+                logger.debug('%s is a build-time dependency only.',
+                             dist.name_and_version)
+        logger.debug('find done for %s', odist)
+        return dists, problems

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/manifest.py

@@ -0,0 +1,384 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 Python Software Foundation.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+"""
+Class representing the list of files in a distribution.
+
+Equivalent to distutils.filelist, but fixes some problems.
+"""
+import fnmatch
+import logging
+import os
+import re
+import sys
+
+from . import DistlibException
+from .compat import fsdecode
+from .util import convert_path
+
+
+__all__ = ['Manifest']
+
+logger = logging.getLogger(__name__)
+
+# a \ followed by some spaces + EOL
+_COLLAPSE_PATTERN = re.compile('\\\\w*\n', re.M)
+_COMMENTED_LINE = re.compile('#.*?(?=\n)|\n(?=$)', re.M | re.S)
+
+#
+# Due to the different results returned by fnmatch.translate, we need
+# to do slightly different processing for Python 2.7 and 3.2 ... this needed
+# to be brought in for Python 3.6 onwards.
+#
+_PYTHON_VERSION = sys.version_info[:2]
+
+
+class Manifest(object):
+    """
+    A list of files built by exploring the filesystem and filtered by applying various
+    patterns to what we find there.
+    """
+
+    def __init__(self, base=None):
+        """
+        Initialise an instance.
+
+        :param base: The base directory to explore under.
+        """
+        self.base = os.path.abspath(os.path.normpath(base or os.getcwd()))
+        self.prefix = self.base + os.sep
+        self.allfiles = None
+        self.files = set()
+
+    #
+    # Public API
+    #
+
+    def findall(self):
+        """Find all files under the base and set ``allfiles`` to the absolute
+        pathnames of files found.
+        """
+        from stat import S_ISREG, S_ISDIR, S_ISLNK
+
+        self.allfiles = allfiles = []
+        root = self.base
+        stack = [root]
+        pop = stack.pop
+        push = stack.append
+
+        while stack:
+            root = pop()
+            names = os.listdir(root)
+
+            for name in names:
+                fullname = os.path.join(root, name)
+
+                # Avoid excess stat calls -- just one will do, thank you!
+                stat = os.stat(fullname)
+                mode = stat.st_mode
+                if S_ISREG(mode):
+                    allfiles.append(fsdecode(fullname))
+                elif S_ISDIR(mode) and not S_ISLNK(mode):
+                    push(fullname)
+
+    def add(self, item):
+        """
+        Add a file to the manifest.
+
+        :param item: The pathname to add. This can be relative to the base.
+        """
+        if not item.startswith(self.prefix):
+            item = os.path.join(self.base, item)
+        self.files.add(os.path.normpath(item))
+
+    def add_many(self, items):
+        """
+        Add a list of files to the manifest.
+
+        :param items: The pathnames to add. These can be relative to the base.
+        """
+        for item in items:
+            self.add(item)
+
+    def sorted(self, wantdirs=False):
+        """
+        Return sorted files in directory order
+        """
+
+        def add_dir(dirs, d):
+            dirs.add(d)
+            logger.debug('add_dir added %s', d)
+            if d != self.base:
+                parent, _ = os.path.split(d)
+                assert parent not in ('', '/')
+                add_dir(dirs, parent)
+
+        result = set(self.files)    # make a copy!
+        if wantdirs:
+            dirs = set()
+            for f in result:
+                add_dir(dirs, os.path.dirname(f))
+            result |= dirs
+        return [os.path.join(*path_tuple) for path_tuple in
+                sorted(os.path.split(path) for path in result)]
+
+    def clear(self):
+        """Clear all collected files."""
+        self.files = set()
+        self.allfiles = []
+
+    def process_directive(self, directive):
+        """
+        Process a directive which either adds some files from ``allfiles`` to
+        ``files``, or removes some files from ``files``.
+
+        :param directive: The directive to process. This should be in a format
+                     compatible with distutils ``MANIFEST.in`` files:
+
+                     http://docs.python.org/distutils/sourcedist.html#commands
+        """
+        # Parse the line: split it up, make sure the right number of words
+        # is there, and return the relevant words.  'action' is always
+        # defined: it's the first word of the line.  Which of the other
+        # three are defined depends on the action; it'll be either
+        # patterns, (dir and patterns), or (dirpattern).
+        action, patterns, thedir, dirpattern = self._parse_directive(directive)
+
+        # OK, now we know that the action is valid and we have the
+        # right number of words on the line for that action -- so we
+        # can proceed with minimal error-checking.
+        if action == 'include':
+            for pattern in patterns:
+                if not self._include_pattern(pattern, anchor=True):
+                    logger.warning('no files found matching %r', pattern)
+
+        elif action == 'exclude':
+            for pattern in patterns:
+                self._exclude_pattern(pattern, anchor=True)
+
+        elif action == 'global-include':
+            for pattern in patterns:
+                if not self._include_pattern(pattern, anchor=False):
+                    logger.warning('no files found matching %r '
+                                   'anywhere in distribution', pattern)
+
+        elif action == 'global-exclude':
+            for pattern in patterns:
+                self._exclude_pattern(pattern, anchor=False)
+
+        elif action == 'recursive-include':
+            for pattern in patterns:
+                if not self._include_pattern(pattern, prefix=thedir):
+                    logger.warning('no files found matching %r '
+                                   'under directory %r', pattern, thedir)
+
+        elif action == 'recursive-exclude':
+            for pattern in patterns:
+                self._exclude_pattern(pattern, prefix=thedir)
+
+        elif action == 'graft':
+            if not self._include_pattern(None, prefix=dirpattern):
+                logger.warning('no directories found matching %r',
+                               dirpattern)
+
+        elif action == 'prune':
+            if not self._exclude_pattern(None, prefix=dirpattern):
+                logger.warning('no previously-included directories found '
+                               'matching %r', dirpattern)
+        else:   # pragma: no cover
+            # This should never happen, as it should be caught in
+            # _parse_template_line
+            raise DistlibException(
+                'invalid action %r' % action)
+
+    #
+    # Private API
+    #
+
+    def _parse_directive(self, directive):
+        """
+        Validate a directive.
+        :param directive: The directive to validate.
+        :return: A tuple of action, patterns, thedir, dir_patterns
+        """
+        words = directive.split()
+        if len(words) == 1 and words[0] not in ('include', 'exclude',
+                                                'global-include',
+                                                'global-exclude',
+                                                'recursive-include',
+                                                'recursive-exclude',
+                                                'graft', 'prune'):
+            # no action given, let's use the default 'include'
+            words.insert(0, 'include')
+
+        action = words[0]
+        patterns = thedir = dir_pattern = None
+
+        if action in ('include', 'exclude',
+                      'global-include', 'global-exclude'):
+            if len(words) < 2:
+                raise DistlibException(
+                    '%r expects <pattern1> <pattern2> ...' % action)
+
+            patterns = [convert_path(word) for word in words[1:]]
+
+        elif action in ('recursive-include', 'recursive-exclude'):
+            if len(words) < 3:
+                raise DistlibException(
+                    '%r expects <dir> <pattern1> <pattern2> ...' % action)
+
+            thedir = convert_path(words[1])
+            patterns = [convert_path(word) for word in words[2:]]
+
+        elif action in ('graft', 'prune'):
+            if len(words) != 2:
+                raise DistlibException(
+                    '%r expects a single <dir_pattern>' % action)
+
+            dir_pattern = convert_path(words[1])
+
+        else:
+            raise DistlibException('unknown action %r' % action)
+
+        return action, patterns, thedir, dir_pattern
+
+    def _include_pattern(self, pattern, anchor=True, prefix=None,
+                         is_regex=False):
+        """Select strings (presumably filenames) from 'self.files' that
+        match 'pattern', a Unix-style wildcard (glob) pattern.
+
+        Patterns are not quite the same as implemented by the 'fnmatch'
+        module: '*' and '?'  match non-special characters, where "special"
+        is platform-dependent: slash on Unix; colon, slash, and backslash on
+        DOS/Windows; and colon on Mac OS.
+
+        If 'anchor' is true (the default), then the pattern match is more
+        stringent: "*.py" will match "foo.py" but not "foo/bar.py".  If
+        'anchor' is false, both of these will match.
+
+        If 'prefix' is supplied, then only filenames starting with 'prefix'
+        (itself a pattern) and ending with 'pattern', with anything in between
+        them, will match.  'anchor' is ignored in this case.
+
+        If 'is_regex' is true, 'anchor' and 'prefix' are ignored, and
+        'pattern' is assumed to be either a string containing a regex or a
+        regex object -- no translation is done, the regex is just compiled
+        and used as-is.
+
+        Selected strings will be added to self.files.
+
+        Return True if files are found.
+        """
+        # XXX docstring lying about what the special chars are?
+        found = False
+        pattern_re = self._translate_pattern(pattern, anchor, prefix, is_regex)
+
+        # delayed loading of allfiles list
+        if self.allfiles is None:
+            self.findall()
+
+        for name in self.allfiles:
+            if pattern_re.search(name):
+                self.files.add(name)
+                found = True
+        return found
+
+    def _exclude_pattern(self, pattern, anchor=True, prefix=None,
+                         is_regex=False):
+        """Remove strings (presumably filenames) from 'files' that match
+        'pattern'.
+
+        Other parameters are the same as for 'include_pattern()', above.
+        The list 'self.files' is modified in place. Return True if files are
+        found.
+
+        This API is public to allow e.g. exclusion of SCM subdirs, e.g. when
+        packaging source distributions
+        """
+        found = False
+        pattern_re = self._translate_pattern(pattern, anchor, prefix, is_regex)
+        for f in list(self.files):
+            if pattern_re.search(f):
+                self.files.remove(f)
+                found = True
+        return found
+
+    def _translate_pattern(self, pattern, anchor=True, prefix=None,
+                           is_regex=False):
+        """Translate a shell-like wildcard pattern to a compiled regular
+        expression.
+
+        Return the compiled regex.  If 'is_regex' true,
+        then 'pattern' is directly compiled to a regex (if it's a string)
+        or just returned as-is (assumes it's a regex object).
+        """
+        if is_regex:
+            if isinstance(pattern, str):
+                return re.compile(pattern)
+            else:
+                return pattern
+
+        if _PYTHON_VERSION > (3, 2):
+            # ditch start and end characters
+            start, _, end = self._glob_to_re('_').partition('_')
+
+        if pattern:
+            pattern_re = self._glob_to_re(pattern)
+            if _PYTHON_VERSION > (3, 2):
+                assert pattern_re.startswith(start) and pattern_re.endswith(end)
+        else:
+            pattern_re = ''
+
+        base = re.escape(os.path.join(self.base, ''))
+        if prefix is not None:
+            # ditch end of pattern character
+            if _PYTHON_VERSION <= (3, 2):
+                empty_pattern = self._glob_to_re('')
+                prefix_re = self._glob_to_re(prefix)[:-len(empty_pattern)]
+            else:
+                prefix_re = self._glob_to_re(prefix)
+                assert prefix_re.startswith(start) and prefix_re.endswith(end)
+                prefix_re = prefix_re[len(start): len(prefix_re) - len(end)]
+            sep = os.sep
+            if os.sep == '\\':
+                sep = r'\\'
+            if _PYTHON_VERSION <= (3, 2):
+                pattern_re = '^' + base + sep.join((prefix_re,
+                                                    '.*' + pattern_re))
+            else:
+                pattern_re = pattern_re[len(start): len(pattern_re) - len(end)]
+                pattern_re = r'%s%s%s%s.*%s%s' % (start, base, prefix_re, sep,
+                                                  pattern_re, end)
+        else:  # no prefix -- respect anchor flag
+            if anchor:
+                if _PYTHON_VERSION <= (3, 2):
+                    pattern_re = '^' + base + pattern_re
+                else:
+                    pattern_re = r'%s%s%s' % (start, base, pattern_re[len(start):])
+
+        return re.compile(pattern_re)
+
+    def _glob_to_re(self, pattern):
+        """Translate a shell-like glob pattern to a regular expression.
+
+        Return a string containing the regex.  Differs from
+        'fnmatch.translate()' in that '*' does not match "special characters"
+        (which are platform-specific).
+        """
+        pattern_re = fnmatch.translate(pattern)
+
+        # '?' and '*' in the glob pattern become '.' and '.*' in the RE, which
+        # IMHO is wrong -- '?' and '*' aren't supposed to match slash in Unix,
+        # and by extension they shouldn't match such "special characters" under
+        # any OS.  So change all non-escaped dots in the RE to match any
+        # character except the special characters (currently: just os.sep).
+        sep = os.sep
+        if os.sep == '\\':
+            # we're using a regex to manipulate a regex, so we need
+            # to escape the backslash twice
+            sep = r'\\\\'
+        escaped = r'\1[^%s]' % sep
+        pattern_re = re.sub(r'((?<!\\)(\\\\)*)\.', escaped, pattern_re)
+        return pattern_re

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/markers.py

@@ -0,0 +1,167 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+"""
+Parser for the environment markers micro-language defined in PEP 508.
+"""
+
+# Note: In PEP 345, the micro-language was Python compatible, so the ast
+# module could be used to parse it. However, PEP 508 introduced operators such
+# as ~= and === which aren't in Python, necessitating a different approach.
+
+import os
+import re
+import sys
+import platform
+
+from .compat import string_types
+from .util import in_venv, parse_marker
+from .version import LegacyVersion as LV
+
+__all__ = ['interpret']
+
+_VERSION_PATTERN = re.compile(
+    r'((\d+(\.\d+)*\w*)|\'(\d+(\.\d+)*\w*)\'|\"(\d+(\.\d+)*\w*)\")')
+_VERSION_MARKERS = {'python_version', 'python_full_version'}
+
+
+def _is_version_marker(s):
+    return isinstance(s, string_types) and s in _VERSION_MARKERS
+
+
+def _is_literal(o):
+    if not isinstance(o, string_types) or not o:
+        return False
+    return o[0] in '\'"'
+
+
+def _get_versions(s):
+    return {LV(m.groups()[0]) for m in _VERSION_PATTERN.finditer(s)}
+
+
+class Evaluator(object):
+    """
+    This class is used to evaluate marker expressions.
+    """
+
+    operations = {
+        '==': lambda x, y: x == y,
+        '===': lambda x, y: x == y,
+        '~=': lambda x, y: x == y or x > y,
+        '!=': lambda x, y: x != y,
+        '<': lambda x, y: x < y,
+        '<=': lambda x, y: x == y or x < y,
+        '>': lambda x, y: x > y,
+        '>=': lambda x, y: x == y or x > y,
+        'and': lambda x, y: x and y,
+        'or': lambda x, y: x or y,
+        'in': lambda x, y: x in y,
+        'not in': lambda x, y: x not in y,
+    }
+
+    def evaluate(self, expr, context):
+        """
+        Evaluate a marker expression returned by the :func:`parse_requirement`
+        function in the specified context.
+        """
+        if isinstance(expr, string_types):
+            if expr[0] in '\'"':
+                result = expr[1:-1]
+            else:
+                if expr not in context:
+                    raise SyntaxError('unknown variable: %s' % expr)
+                result = context[expr]
+        else:
+            assert isinstance(expr, dict)
+            op = expr['op']
+            if op not in self.operations:
+                raise NotImplementedError('op not implemented: %s' % op)
+            elhs = expr['lhs']
+            erhs = expr['rhs']
+            if _is_literal(expr['lhs']) and _is_literal(expr['rhs']):
+                raise SyntaxError('invalid comparison: %s %s %s' %
+                                  (elhs, op, erhs))
+
+            lhs = self.evaluate(elhs, context)
+            rhs = self.evaluate(erhs, context)
+            if ((_is_version_marker(elhs) or _is_version_marker(erhs))
+                    and op in ('<', '<=', '>', '>=', '===', '==', '!=', '~=')):
+                lhs = LV(lhs)
+                rhs = LV(rhs)
+            elif _is_version_marker(elhs) and op in ('in', 'not in'):
+                lhs = LV(lhs)
+                rhs = _get_versions(rhs)
+            result = self.operations[op](lhs, rhs)
+        return result
+
+
+_DIGITS = re.compile(r'\d+\.\d+')
+
+
+def default_context():
+
+    def format_full_version(info):
+        version = '%s.%s.%s' % (info.major, info.minor, info.micro)
+        kind = info.releaselevel
+        if kind != 'final':
+            version += kind[0] + str(info.serial)
+        return version
+
+    if hasattr(sys, 'implementation'):
+        implementation_version = format_full_version(
+            sys.implementation.version)
+        implementation_name = sys.implementation.name
+    else:
+        implementation_version = '0'
+        implementation_name = ''
+
+    ppv = platform.python_version()
+    m = _DIGITS.match(ppv)
+    pv = m.group(0)
+    result = {
+        'implementation_name': implementation_name,
+        'implementation_version': implementation_version,
+        'os_name': os.name,
+        'platform_machine': platform.machine(),
+        'platform_python_implementation': platform.python_implementation(),
+        'platform_release': platform.release(),
+        'platform_system': platform.system(),
+        'platform_version': platform.version(),
+        'platform_in_venv': str(in_venv()),
+        'python_full_version': ppv,
+        'python_version': pv,
+        'sys_platform': sys.platform,
+    }
+    return result
+
+
+DEFAULT_CONTEXT = default_context()
+del default_context
+
+evaluator = Evaluator()
+
+
+def interpret(marker, execution_context=None):
+    """
+    Interpret a marker and return a result depending on environment.
+
+    :param marker: The marker to interpret.
+    :type marker: str
+    :param execution_context: The context used for name lookup.
+    :type execution_context: mapping
+    """
+    try:
+        expr, rest = parse_marker(marker)
+    except Exception as e:
+        raise SyntaxError('Unable to interpret marker syntax: %s: %s' %
+                          (marker, e))
+    if rest and rest[0] != '#':
+        raise SyntaxError('unexpected trailing data in marker: %s: %s' %
+                          (marker, rest))
+    context = dict(DEFAULT_CONTEXT)
+    if execution_context:
+        context.update(execution_context)
+    return evaluator.evaluate(expr, context)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/metadata.py

@@ -0,0 +1,1068 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012 The Python Software Foundation.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+"""Implementation of the Metadata for Python packages PEPs.
+
+Supports all metadata formats (1.0, 1.1, 1.2, 1.3/2.1 and 2.2).
+"""
+from __future__ import unicode_literals
+
+import codecs
+from email import message_from_file
+import json
+import logging
+import re
+
+
+from . import DistlibException, __version__
+from .compat import StringIO, string_types, text_type
+from .markers import interpret
+from .util import extract_by_key, get_extras
+from .version import get_scheme, PEP440_VERSION_RE
+
+logger = logging.getLogger(__name__)
+
+
+class MetadataMissingError(DistlibException):
+    """A required metadata is missing"""
+
+
+class MetadataConflictError(DistlibException):
+    """Attempt to read or write metadata fields that are conflictual."""
+
+
+class MetadataUnrecognizedVersionError(DistlibException):
+    """Unknown metadata version number."""
+
+
+class MetadataInvalidError(DistlibException):
+    """A metadata value is invalid"""
+
+# public API of this module
+__all__ = ['Metadata', 'PKG_INFO_ENCODING', 'PKG_INFO_PREFERRED_VERSION']
+
+# Encoding used for the PKG-INFO files
+PKG_INFO_ENCODING = 'utf-8'
+
+# preferred version. Hopefully will be changed
+# to 1.2 once PEP 345 is supported everywhere
+PKG_INFO_PREFERRED_VERSION = '1.1'
+
+_LINE_PREFIX_1_2 = re.compile('\n       \\|')
+_LINE_PREFIX_PRE_1_2 = re.compile('\n        ')
+_241_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
+               'Summary', 'Description',
+               'Keywords', 'Home-page', 'Author', 'Author-email',
+               'License')
+
+_314_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
+               'Supported-Platform', 'Summary', 'Description',
+               'Keywords', 'Home-page', 'Author', 'Author-email',
+               'License', 'Classifier', 'Download-URL', 'Obsoletes',
+               'Provides', 'Requires')
+
+_314_MARKERS = ('Obsoletes', 'Provides', 'Requires', 'Classifier',
+                'Download-URL')
+
+_345_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
+               'Supported-Platform', 'Summary', 'Description',
+               'Keywords', 'Home-page', 'Author', 'Author-email',
+               'Maintainer', 'Maintainer-email', 'License',
+               'Classifier', 'Download-URL', 'Obsoletes-Dist',
+               'Project-URL', 'Provides-Dist', 'Requires-Dist',
+               'Requires-Python', 'Requires-External')
+
+_345_MARKERS = ('Provides-Dist', 'Requires-Dist', 'Requires-Python',
+                'Obsoletes-Dist', 'Requires-External', 'Maintainer',
+                'Maintainer-email', 'Project-URL')
+
+_426_FIELDS = ('Metadata-Version', 'Name', 'Version', 'Platform',
+               'Supported-Platform', 'Summary', 'Description',
+               'Keywords', 'Home-page', 'Author', 'Author-email',
+               'Maintainer', 'Maintainer-email', 'License',
+               'Classifier', 'Download-URL', 'Obsoletes-Dist',
+               'Project-URL', 'Provides-Dist', 'Requires-Dist',
+               'Requires-Python', 'Requires-External', 'Private-Version',
+               'Obsoleted-By', 'Setup-Requires-Dist', 'Extension',
+               'Provides-Extra')
+
+_426_MARKERS = ('Private-Version', 'Provides-Extra', 'Obsoleted-By',
+                'Setup-Requires-Dist', 'Extension')
+
+# See issue #106: Sometimes 'Requires' and 'Provides' occur wrongly in
+# the metadata. Include them in the tuple literal below to allow them
+# (for now).
+# Ditto for Obsoletes - see issue #140.
+_566_FIELDS = _426_FIELDS + ('Description-Content-Type',
+                             'Requires', 'Provides', 'Obsoletes')
+
+_566_MARKERS = ('Description-Content-Type',)
+
+_643_MARKERS = ('Dynamic', 'License-File')
+
+_643_FIELDS = _566_FIELDS + _643_MARKERS
+
+_ALL_FIELDS = set()
+_ALL_FIELDS.update(_241_FIELDS)
+_ALL_FIELDS.update(_314_FIELDS)
+_ALL_FIELDS.update(_345_FIELDS)
+_ALL_FIELDS.update(_426_FIELDS)
+_ALL_FIELDS.update(_566_FIELDS)
+_ALL_FIELDS.update(_643_FIELDS)
+
+EXTRA_RE = re.compile(r'''extra\s*==\s*("([^"]+)"|'([^']+)')''')
+
+
+def _version2fieldlist(version):
+    if version == '1.0':
+        return _241_FIELDS
+    elif version == '1.1':
+        return _314_FIELDS
+    elif version == '1.2':
+        return _345_FIELDS
+    elif version in ('1.3', '2.1'):
+        # avoid adding field names if already there
+        return _345_FIELDS + tuple(f for f in _566_FIELDS if f not in _345_FIELDS)
+    elif version == '2.0':
+        raise ValueError('Metadata 2.0 is withdrawn and not supported')
+        # return _426_FIELDS
+    elif version == '2.2':
+        return _643_FIELDS
+    raise MetadataUnrecognizedVersionError(version)
+
+
+def _best_version(fields):
+    """Detect the best version depending on the fields used."""
+    def _has_marker(keys, markers):
+        return any(marker in keys for marker in markers)
+
+    keys = [key for key, value in fields.items() if value not in ([], 'UNKNOWN', None)]
+    possible_versions = ['1.0', '1.1', '1.2', '1.3', '2.1', '2.2']  # 2.0 removed
+
+    # first let's try to see if a field is not part of one of the version
+    for key in keys:
+        if key not in _241_FIELDS and '1.0' in possible_versions:
+            possible_versions.remove('1.0')
+            logger.debug('Removed 1.0 due to %s', key)
+        if key not in _314_FIELDS and '1.1' in possible_versions:
+            possible_versions.remove('1.1')
+            logger.debug('Removed 1.1 due to %s', key)
+        if key not in _345_FIELDS and '1.2' in possible_versions:
+            possible_versions.remove('1.2')
+            logger.debug('Removed 1.2 due to %s', key)
+        if key not in _566_FIELDS and '1.3' in possible_versions:
+            possible_versions.remove('1.3')
+            logger.debug('Removed 1.3 due to %s', key)
+        if key not in _566_FIELDS and '2.1' in possible_versions:
+            if key != 'Description':  # In 2.1, description allowed after headers
+                possible_versions.remove('2.1')
+                logger.debug('Removed 2.1 due to %s', key)
+        if key not in _643_FIELDS and '2.2' in possible_versions:
+            possible_versions.remove('2.2')
+            logger.debug('Removed 2.2 due to %s', key)
+        # if key not in _426_FIELDS and '2.0' in possible_versions:
+            # possible_versions.remove('2.0')
+            # logger.debug('Removed 2.0 due to %s', key)
+
+    # possible_version contains qualified versions
+    if len(possible_versions) == 1:
+        return possible_versions[0]   # found !
+    elif len(possible_versions) == 0:
+        logger.debug('Out of options - unknown metadata set: %s', fields)
+        raise MetadataConflictError('Unknown metadata set')
+
+    # let's see if one unique marker is found
+    is_1_1 = '1.1' in possible_versions and _has_marker(keys, _314_MARKERS)
+    is_1_2 = '1.2' in possible_versions and _has_marker(keys, _345_MARKERS)
+    is_2_1 = '2.1' in possible_versions and _has_marker(keys, _566_MARKERS)
+    # is_2_0 = '2.0' in possible_versions and _has_marker(keys, _426_MARKERS)
+    is_2_2 = '2.2' in possible_versions and _has_marker(keys, _643_MARKERS)
+    if int(is_1_1) + int(is_1_2) + int(is_2_1) + int(is_2_2) > 1:
+        raise MetadataConflictError('You used incompatible 1.1/1.2/2.1/2.2 fields')
+
+    # we have the choice, 1.0, or 1.2, 2.1 or 2.2
+    #   - 1.0 has a broken Summary field but works with all tools
+    #   - 1.1 is to avoid
+    #   - 1.2 fixes Summary but has little adoption
+    #   - 2.1 adds more features
+    #   - 2.2 is the latest
+    if not is_1_1 and not is_1_2 and not is_2_1 and not is_2_2:
+        # we couldn't find any specific marker
+        if PKG_INFO_PREFERRED_VERSION in possible_versions:
+            return PKG_INFO_PREFERRED_VERSION
+    if is_1_1:
+        return '1.1'
+    if is_1_2:
+        return '1.2'
+    if is_2_1:
+        return '2.1'
+    # if is_2_2:
+        # return '2.2'
+
+    return '2.2'
+
+# This follows the rules about transforming keys as described in
+# https://www.python.org/dev/peps/pep-0566/#id17
+_ATTR2FIELD = {
+    name.lower().replace("-", "_"): name for name in _ALL_FIELDS
+}
+_FIELD2ATTR = {field: attr for attr, field in _ATTR2FIELD.items()}
+
+_PREDICATE_FIELDS = ('Requires-Dist', 'Obsoletes-Dist', 'Provides-Dist')
+_VERSIONS_FIELDS = ('Requires-Python',)
+_VERSION_FIELDS = ('Version',)
+_LISTFIELDS = ('Platform', 'Classifier', 'Obsoletes',
+               'Requires', 'Provides', 'Obsoletes-Dist',
+               'Provides-Dist', 'Requires-Dist', 'Requires-External',
+               'Project-URL', 'Supported-Platform', 'Setup-Requires-Dist',
+               'Provides-Extra', 'Extension', 'License-File')
+_LISTTUPLEFIELDS = ('Project-URL',)
+
+_ELEMENTSFIELD = ('Keywords',)
+
+_UNICODEFIELDS = ('Author', 'Maintainer', 'Summary', 'Description')
+
+_MISSING = object()
+
+_FILESAFE = re.compile('[^A-Za-z0-9.]+')
+
+
+def _get_name_and_version(name, version, for_filename=False):
+    """Return the distribution name with version.
+
+    If for_filename is true, return a filename-escaped form."""
+    if for_filename:
+        # For both name and version any runs of non-alphanumeric or '.'
+        # characters are replaced with a single '-'.  Additionally any
+        # spaces in the version string become '.'
+        name = _FILESAFE.sub('-', name)
+        version = _FILESAFE.sub('-', version.replace(' ', '.'))
+    return '%s-%s' % (name, version)
+
+
+class LegacyMetadata(object):
+    """The legacy metadata of a release.
+
+    Supports versions 1.0, 1.1, 1.2, 2.0 and 1.3/2.1 (auto-detected). You can
+    instantiate the class with one of these arguments (or none):
+    - *path*, the path to a metadata file
+    - *fileobj* give a file-like object with metadata as content
+    - *mapping* is a dict-like object
+    - *scheme* is a version scheme name
+    """
+    # TODO document the mapping API and UNKNOWN default key
+
+    def __init__(self, path=None, fileobj=None, mapping=None,
+                 scheme='default'):
+        if [path, fileobj, mapping].count(None) < 2:
+            raise TypeError('path, fileobj and mapping are exclusive')
+        self._fields = {}
+        self.requires_files = []
+        self._dependencies = None
+        self.scheme = scheme
+        if path is not None:
+            self.read(path)
+        elif fileobj is not None:
+            self.read_file(fileobj)
+        elif mapping is not None:
+            self.update(mapping)
+            self.set_metadata_version()
+
+    def set_metadata_version(self):
+        self._fields['Metadata-Version'] = _best_version(self._fields)
+
+    def _write_field(self, fileobj, name, value):
+        fileobj.write('%s: %s\n' % (name, value))
+
+    def __getitem__(self, name):
+        return self.get(name)
+
+    def __setitem__(self, name, value):
+        return self.set(name, value)
+
+    def __delitem__(self, name):
+        field_name = self._convert_name(name)
+        try:
+            del self._fields[field_name]
+        except KeyError:
+            raise KeyError(name)
+
+    def __contains__(self, name):
+        return (name in self._fields or
+                self._convert_name(name) in self._fields)
+
+    def _convert_name(self, name):
+        if name in _ALL_FIELDS:
+            return name
+        name = name.replace('-', '_').lower()
+        return _ATTR2FIELD.get(name, name)
+
+    def _default_value(self, name):
+        if name in _LISTFIELDS or name in _ELEMENTSFIELD:
+            return []
+        return 'UNKNOWN'
+
+    def _remove_line_prefix(self, value):
+        if self.metadata_version in ('1.0', '1.1'):
+            return _LINE_PREFIX_PRE_1_2.sub('\n', value)
+        else:
+            return _LINE_PREFIX_1_2.sub('\n', value)
+
+    def __getattr__(self, name):
+        if name in _ATTR2FIELD:
+            return self[name]
+        raise AttributeError(name)
+
+    #
+    # Public API
+    #
+
+#    dependencies = property(_get_dependencies, _set_dependencies)
+
+    def get_fullname(self, filesafe=False):
+        """Return the distribution name with version.
+
+        If filesafe is true, return a filename-escaped form."""
+        return _get_name_and_version(self['Name'], self['Version'], filesafe)
+
+    def is_field(self, name):
+        """return True if name is a valid metadata key"""
+        name = self._convert_name(name)
+        return name in _ALL_FIELDS
+
+    def is_multi_field(self, name):
+        name = self._convert_name(name)
+        return name in _LISTFIELDS
+
+    def read(self, filepath):
+        """Read the metadata values from a file path."""
+        fp = codecs.open(filepath, 'r', encoding='utf-8')
+        try:
+            self.read_file(fp)
+        finally:
+            fp.close()
+
+    def read_file(self, fileob):
+        """Read the metadata values from a file object."""
+        msg = message_from_file(fileob)
+        self._fields['Metadata-Version'] = msg['metadata-version']
+
+        # When reading, get all the fields we can
+        for field in _ALL_FIELDS:
+            if field not in msg:
+                continue
+            if field in _LISTFIELDS:
+                # we can have multiple lines
+                values = msg.get_all(field)
+                if field in _LISTTUPLEFIELDS and values is not None:
+                    values = [tuple(value.split(',')) for value in values]
+                self.set(field, values)
+            else:
+                # single line
+                value = msg[field]
+                if value is not None and value != 'UNKNOWN':
+                    self.set(field, value)
+
+        # PEP 566 specifies that the body be used for the description, if
+        # available
+        body = msg.get_payload()
+        self["Description"] = body if body else self["Description"]
+        # logger.debug('Attempting to set metadata for %s', self)
+        # self.set_metadata_version()
+
+    def write(self, filepath, skip_unknown=False):
+        """Write the metadata fields to filepath."""
+        fp = codecs.open(filepath, 'w', encoding='utf-8')
+        try:
+            self.write_file(fp, skip_unknown)
+        finally:
+            fp.close()
+
+    def write_file(self, fileobject, skip_unknown=False):
+        """Write the PKG-INFO format data to a file object."""
+        self.set_metadata_version()
+
+        for field in _version2fieldlist(self['Metadata-Version']):
+            values = self.get(field)
+            if skip_unknown and values in ('UNKNOWN', [], ['UNKNOWN']):
+                continue
+            if field in _ELEMENTSFIELD:
+                self._write_field(fileobject, field, ','.join(values))
+                continue
+            if field not in _LISTFIELDS:
+                if field == 'Description':
+                    if self.metadata_version in ('1.0', '1.1'):
+                        values = values.replace('\n', '\n        ')
+                    else:
+                        values = values.replace('\n', '\n       |')
+                values = [values]
+
+            if field in _LISTTUPLEFIELDS:
+                values = [','.join(value) for value in values]
+
+            for value in values:
+                self._write_field(fileobject, field, value)
+
+    def update(self, other=None, **kwargs):
+        """Set metadata values from the given iterable `other` and kwargs.
+
+        Behavior is like `dict.update`: If `other` has a ``keys`` method,
+        they are looped over and ``self[key]`` is assigned ``other[key]``.
+        Else, ``other`` is an iterable of ``(key, value)`` iterables.
+
+        Keys that don't match a metadata field or that have an empty value are
+        dropped.
+        """
+        def _set(key, value):
+            if key in _ATTR2FIELD and value:
+                self.set(self._convert_name(key), value)
+
+        if not other:
+            # other is None or empty container
+            pass
+        elif hasattr(other, 'keys'):
+            for k in other.keys():
+                _set(k, other[k])
+        else:
+            for k, v in other:
+                _set(k, v)
+
+        if kwargs:
+            for k, v in kwargs.items():
+                _set(k, v)
+
+    def set(self, name, value):
+        """Control then set a metadata field."""
+        name = self._convert_name(name)
+
+        if ((name in _ELEMENTSFIELD or name == 'Platform') and
+            not isinstance(value, (list, tuple))):
+            if isinstance(value, string_types):
+                value = [v.strip() for v in value.split(',')]
+            else:
+                value = []
+        elif (name in _LISTFIELDS and
+              not isinstance(value, (list, tuple))):
+            if isinstance(value, string_types):
+                value = [value]
+            else:
+                value = []
+
+        if logger.isEnabledFor(logging.WARNING):
+            project_name = self['Name']
+
+            scheme = get_scheme(self.scheme)
+            if name in _PREDICATE_FIELDS and value is not None:
+                for v in value:
+                    # check that the values are valid
+                    if not scheme.is_valid_matcher(v.split(';')[0]):
+                        logger.warning(
+                            "'%s': '%s' is not valid (field '%s')",
+                            project_name, v, name)
+            # FIXME this rejects UNKNOWN, is that right?
+            elif name in _VERSIONS_FIELDS and value is not None:
+                if not scheme.is_valid_constraint_list(value):
+                    logger.warning("'%s': '%s' is not a valid version (field '%s')",
+                                   project_name, value, name)
+            elif name in _VERSION_FIELDS and value is not None:
+                if not scheme.is_valid_version(value):
+                    logger.warning("'%s': '%s' is not a valid version (field '%s')",
+                                   project_name, value, name)
+
+        if name in _UNICODEFIELDS:
+            if name == 'Description':
+                value = self._remove_line_prefix(value)
+
+        self._fields[name] = value
+
+    def get(self, name, default=_MISSING):
+        """Get a metadata field."""
+        name = self._convert_name(name)
+        if name not in self._fields:
+            if default is _MISSING:
+                default = self._default_value(name)
+            return default
+        if name in _UNICODEFIELDS:
+            value = self._fields[name]
+            return value
+        elif name in _LISTFIELDS:
+            value = self._fields[name]
+            if value is None:
+                return []
+            res = []
+            for val in value:
+                if name not in _LISTTUPLEFIELDS:
+                    res.append(val)
+                else:
+                    # That's for Project-URL
+                    res.append((val[0], val[1]))
+            return res
+
+        elif name in _ELEMENTSFIELD:
+            value = self._fields[name]
+            if isinstance(value, string_types):
+                return value.split(',')
+        return self._fields[name]
+
+    def check(self, strict=False):
+        """Check if the metadata is compliant. If strict is True then raise if
+        no Name or Version are provided"""
+        self.set_metadata_version()
+
+        # XXX should check the versions (if the file was loaded)
+        missing, warnings = [], []
+
+        for attr in ('Name', 'Version'):  # required by PEP 345
+            if attr not in self:
+                missing.append(attr)
+
+        if strict and missing != []:
+            msg = 'missing required metadata: %s' % ', '.join(missing)
+            raise MetadataMissingError(msg)
+
+        for attr in ('Home-page', 'Author'):
+            if attr not in self:
+                missing.append(attr)
+
+        # checking metadata 1.2 (XXX needs to check 1.1, 1.0)
+        if self['Metadata-Version'] != '1.2':
+            return missing, warnings
+
+        scheme = get_scheme(self.scheme)
+
+        def are_valid_constraints(value):
+            for v in value:
+                if not scheme.is_valid_matcher(v.split(';')[0]):
+                    return False
+            return True
+
+        for fields, controller in ((_PREDICATE_FIELDS, are_valid_constraints),
+                                   (_VERSIONS_FIELDS,
+                                    scheme.is_valid_constraint_list),
+                                   (_VERSION_FIELDS,
+                                    scheme.is_valid_version)):
+            for field in fields:
+                value = self.get(field, None)
+                if value is not None and not controller(value):
+                    warnings.append("Wrong value for '%s': %s" % (field, value))
+
+        return missing, warnings
+
+    def todict(self, skip_missing=False):
+        """Return fields as a dict.
+
+        Field names will be converted to use the underscore-lowercase style
+        instead of hyphen-mixed case (i.e. home_page instead of Home-page).
+        This is as per https://www.python.org/dev/peps/pep-0566/#id17.
+        """
+        self.set_metadata_version()
+
+        fields = _version2fieldlist(self['Metadata-Version'])
+
+        data = {}
+
+        for field_name in fields:
+            if not skip_missing or field_name in self._fields:
+                key = _FIELD2ATTR[field_name]
+                if key != 'project_url':
+                    data[key] = self[field_name]
+                else:
+                    data[key] = [','.join(u) for u in self[field_name]]
+
+        return data
+
+    def add_requirements(self, requirements):
+        if self['Metadata-Version'] == '1.1':
+            # we can't have 1.1 metadata *and* Setuptools requires
+            for field in ('Obsoletes', 'Requires', 'Provides'):
+                if field in self:
+                    del self[field]
+        self['Requires-Dist'] += requirements
+
+    # Mapping API
+    # TODO could add iter* variants
+
+    def keys(self):
+        return list(_version2fieldlist(self['Metadata-Version']))
+
+    def __iter__(self):
+        for key in self.keys():
+            yield key
+
+    def values(self):
+        return [self[key] for key in self.keys()]
+
+    def items(self):
+        return [(key, self[key]) for key in self.keys()]
+
+    def __repr__(self):
+        return '<%s %s %s>' % (self.__class__.__name__, self.name,
+                               self.version)
+
+
+METADATA_FILENAME = 'pydist.json'
+WHEEL_METADATA_FILENAME = 'metadata.json'
+LEGACY_METADATA_FILENAME = 'METADATA'
+
+
+class Metadata(object):
+    """
+    The metadata of a release. This implementation uses 2.1
+    metadata where possible. If not possible, it wraps a LegacyMetadata
+    instance which handles the key-value metadata format.
+    """
+
+    METADATA_VERSION_MATCHER = re.compile(r'^\d+(\.\d+)*$')
+
+    NAME_MATCHER = re.compile('^[0-9A-Z]([0-9A-Z_.-]*[0-9A-Z])?$', re.I)
+
+    FIELDNAME_MATCHER = re.compile('^[A-Z]([0-9A-Z-]*[0-9A-Z])?$', re.I)
+
+    VERSION_MATCHER = PEP440_VERSION_RE
+
+    SUMMARY_MATCHER = re.compile('.{1,2047}')
+
+    METADATA_VERSION = '2.0'
+
+    GENERATOR = 'distlib (%s)' % __version__
+
+    MANDATORY_KEYS = {
+        'name': (),
+        'version': (),
+        'summary': ('legacy',),
+    }
+
+    INDEX_KEYS = ('name version license summary description author '
+                  'author_email keywords platform home_page classifiers '
+                  'download_url')
+
+    DEPENDENCY_KEYS = ('extras run_requires test_requires build_requires '
+                       'dev_requires provides meta_requires obsoleted_by '
+                       'supports_environments')
+
+    SYNTAX_VALIDATORS = {
+        'metadata_version': (METADATA_VERSION_MATCHER, ()),
+        'name': (NAME_MATCHER, ('legacy',)),
+        'version': (VERSION_MATCHER, ('legacy',)),
+        'summary': (SUMMARY_MATCHER, ('legacy',)),
+        'dynamic': (FIELDNAME_MATCHER, ('legacy',)),
+    }
+
+    __slots__ = ('_legacy', '_data', 'scheme')
+
+    def __init__(self, path=None, fileobj=None, mapping=None,
+                 scheme='default'):
+        if [path, fileobj, mapping].count(None) < 2:
+            raise TypeError('path, fileobj and mapping are exclusive')
+        self._legacy = None
+        self._data = None
+        self.scheme = scheme
+        #import pdb; pdb.set_trace()
+        if mapping is not None:
+            try:
+                self._validate_mapping(mapping, scheme)
+                self._data = mapping
+            except MetadataUnrecognizedVersionError:
+                self._legacy = LegacyMetadata(mapping=mapping, scheme=scheme)
+                self.validate()
+        else:
+            data = None
+            if path:
+                with open(path, 'rb') as f:
+                    data = f.read()
+            elif fileobj:
+                data = fileobj.read()
+            if data is None:
+                # Initialised with no args - to be added
+                self._data = {
+                    'metadata_version': self.METADATA_VERSION,
+                    'generator': self.GENERATOR,
+                }
+            else:
+                if not isinstance(data, text_type):
+                    data = data.decode('utf-8')
+                try:
+                    self._data = json.loads(data)
+                    self._validate_mapping(self._data, scheme)
+                except ValueError:
+                    # Note: MetadataUnrecognizedVersionError does not
+                    # inherit from ValueError (it's a DistlibException,
+                    # which should not inherit from ValueError).
+                    # The ValueError comes from the json.load - if that
+                    # succeeds and we get a validation error, we want
+                    # that to propagate
+                    self._legacy = LegacyMetadata(fileobj=StringIO(data),
+                                                  scheme=scheme)
+                    self.validate()
+
+    common_keys = set(('name', 'version', 'license', 'keywords', 'summary'))
+
+    none_list = (None, list)
+    none_dict = (None, dict)
+
+    mapped_keys = {
+        'run_requires': ('Requires-Dist', list),
+        'build_requires': ('Setup-Requires-Dist', list),
+        'dev_requires': none_list,
+        'test_requires': none_list,
+        'meta_requires': none_list,
+        'extras': ('Provides-Extra', list),
+        'modules': none_list,
+        'namespaces': none_list,
+        'exports': none_dict,
+        'commands': none_dict,
+        'classifiers': ('Classifier', list),
+        'source_url': ('Download-URL', None),
+        'metadata_version': ('Metadata-Version', None),
+    }
+
+    del none_list, none_dict
+
+    def __getattribute__(self, key):
+        common = object.__getattribute__(self, 'common_keys')
+        mapped = object.__getattribute__(self, 'mapped_keys')
+        if key in mapped:
+            lk, maker = mapped[key]
+            if self._legacy:
+                if lk is None:
+                    result = None if maker is None else maker()
+                else:
+                    result = self._legacy.get(lk)
+            else:
+                value = None if maker is None else maker()
+                if key not in ('commands', 'exports', 'modules', 'namespaces',
+                               'classifiers'):
+                    result = self._data.get(key, value)
+                else:
+                    # special cases for PEP 459
+                    sentinel = object()
+                    result = sentinel
+                    d = self._data.get('extensions')
+                    if d:
+                        if key == 'commands':
+                            result = d.get('python.commands', value)
+                        elif key == 'classifiers':
+                            d = d.get('python.details')
+                            if d:
+                                result = d.get(key, value)
+                        else:
+                            d = d.get('python.exports')
+                            if not d:
+                                d = self._data.get('python.exports')
+                            if d:
+                                result = d.get(key, value)
+                    if result is sentinel:
+                        result = value
+        elif key not in common:
+            result = object.__getattribute__(self, key)
+        elif self._legacy:
+            result = self._legacy.get(key)
+        else:
+            result = self._data.get(key)
+        return result
+
+    def _validate_value(self, key, value, scheme=None):
+        if key in self.SYNTAX_VALIDATORS:
+            pattern, exclusions = self.SYNTAX_VALIDATORS[key]
+            if (scheme or self.scheme) not in exclusions:
+                m = pattern.match(value)
+                if not m:
+                    raise MetadataInvalidError("'%s' is an invalid value for "
+                                               "the '%s' property" % (value,
+                                                                    key))
+
+    def __setattr__(self, key, value):
+        self._validate_value(key, value)
+        common = object.__getattribute__(self, 'common_keys')
+        mapped = object.__getattribute__(self, 'mapped_keys')
+        if key in mapped:
+            lk, _ = mapped[key]
+            if self._legacy:
+                if lk is None:
+                    raise NotImplementedError
+                self._legacy[lk] = value
+            elif key not in ('commands', 'exports', 'modules', 'namespaces',
+                             'classifiers'):
+                self._data[key] = value
+            else:
+                # special cases for PEP 459
+                d = self._data.setdefault('extensions', {})
+                if key == 'commands':
+                    d['python.commands'] = value
+                elif key == 'classifiers':
+                    d = d.setdefault('python.details', {})
+                    d[key] = value
+                else:
+                    d = d.setdefault('python.exports', {})
+                    d[key] = value
+        elif key not in common:
+            object.__setattr__(self, key, value)
+        else:
+            if key == 'keywords':
+                if isinstance(value, string_types):
+                    value = value.strip()
+                    if value:
+                        value = value.split()
+                    else:
+                        value = []
+            if self._legacy:
+                self._legacy[key] = value
+            else:
+                self._data[key] = value
+
+    @property
+    def name_and_version(self):
+        return _get_name_and_version(self.name, self.version, True)
+
+    @property
+    def provides(self):
+        if self._legacy:
+            result = self._legacy['Provides-Dist']
+        else:
+            result = self._data.setdefault('provides', [])
+        s = '%s (%s)' % (self.name, self.version)
+        if s not in result:
+            result.append(s)
+        return result
+
+    @provides.setter
+    def provides(self, value):
+        if self._legacy:
+            self._legacy['Provides-Dist'] = value
+        else:
+            self._data['provides'] = value
+
+    def get_requirements(self, reqts, extras=None, env=None):
+        """
+        Base method to get dependencies, given a set of extras
+        to satisfy and an optional environment context.
+        :param reqts: A list of sometimes-wanted dependencies,
+                      perhaps dependent on extras and environment.
+        :param extras: A list of optional components being requested.
+        :param env: An optional environment for marker evaluation.
+        """
+        if self._legacy:
+            result = reqts
+        else:
+            result = []
+            extras = get_extras(extras or [], self.extras)
+            for d in reqts:
+                if 'extra' not in d and 'environment' not in d:
+                    # unconditional
+                    include = True
+                else:
+                    if 'extra' not in d:
+                        # Not extra-dependent - only environment-dependent
+                        include = True
+                    else:
+                        include = d.get('extra') in extras
+                    if include:
+                        # Not excluded because of extras, check environment
+                        marker = d.get('environment')
+                        if marker:
+                            include = interpret(marker, env)
+                if include:
+                    result.extend(d['requires'])
+            for key in ('build', 'dev', 'test'):
+                e = ':%s:' % key
+                if e in extras:
+                    extras.remove(e)
+                    # A recursive call, but it should terminate since 'test'
+                    # has been removed from the extras
+                    reqts = self._data.get('%s_requires' % key, [])
+                    result.extend(self.get_requirements(reqts, extras=extras,
+                                                        env=env))
+        return result
+
+    @property
+    def dictionary(self):
+        if self._legacy:
+            return self._from_legacy()
+        return self._data
+
+    @property
+    def dependencies(self):
+        if self._legacy:
+            raise NotImplementedError
+        else:
+            return extract_by_key(self._data, self.DEPENDENCY_KEYS)
+
+    @dependencies.setter
+    def dependencies(self, value):
+        if self._legacy:
+            raise NotImplementedError
+        else:
+            self._data.update(value)
+
+    def _validate_mapping(self, mapping, scheme):
+        if mapping.get('metadata_version') != self.METADATA_VERSION:
+            raise MetadataUnrecognizedVersionError()
+        missing = []
+        for key, exclusions in self.MANDATORY_KEYS.items():
+            if key not in mapping:
+                if scheme not in exclusions:
+                    missing.append(key)
+        if missing:
+            msg = 'Missing metadata items: %s' % ', '.join(missing)
+            raise MetadataMissingError(msg)
+        for k, v in mapping.items():
+            self._validate_value(k, v, scheme)
+
+    def validate(self):
+        if self._legacy:
+            missing, warnings = self._legacy.check(True)
+            if missing or warnings:
+                logger.warning('Metadata: missing: %s, warnings: %s',
+                               missing, warnings)
+        else:
+            self._validate_mapping(self._data, self.scheme)
+
+    def todict(self):
+        if self._legacy:
+            return self._legacy.todict(True)
+        else:
+            result = extract_by_key(self._data, self.INDEX_KEYS)
+            return result
+
+    def _from_legacy(self):
+        assert self._legacy and not self._data
+        result = {
+            'metadata_version': self.METADATA_VERSION,
+            'generator': self.GENERATOR,
+        }
+        lmd = self._legacy.todict(True)     # skip missing ones
+        for k in ('name', 'version', 'license', 'summary', 'description',
+                  'classifier'):
+            if k in lmd:
+                if k == 'classifier':
+                    nk = 'classifiers'
+                else:
+                    nk = k
+                result[nk] = lmd[k]
+        kw = lmd.get('Keywords', [])
+        if kw == ['']:
+            kw = []
+        result['keywords'] = kw
+        keys = (('requires_dist', 'run_requires'),
+                ('setup_requires_dist', 'build_requires'))
+        for ok, nk in keys:
+            if ok in lmd and lmd[ok]:
+                result[nk] = [{'requires': lmd[ok]}]
+        result['provides'] = self.provides
+        author = {}
+        maintainer = {}
+        return result
+
+    LEGACY_MAPPING = {
+        'name': 'Name',
+        'version': 'Version',
+        ('extensions', 'python.details', 'license'): 'License',
+        'summary': 'Summary',
+        'description': 'Description',
+        ('extensions', 'python.project', 'project_urls', 'Home'): 'Home-page',
+        ('extensions', 'python.project', 'contacts', 0, 'name'): 'Author',
+        ('extensions', 'python.project', 'contacts', 0, 'email'): 'Author-email',
+        'source_url': 'Download-URL',
+        ('extensions', 'python.details', 'classifiers'): 'Classifier',
+    }
+
+    def _to_legacy(self):
+        def process_entries(entries):
+            reqts = set()
+            for e in entries:
+                extra = e.get('extra')
+                env = e.get('environment')
+                rlist = e['requires']
+                for r in rlist:
+                    if not env and not extra:
+                        reqts.add(r)
+                    else:
+                        marker = ''
+                        if extra:
+                            marker = 'extra == "%s"' % extra
+                        if env:
+                            if marker:
+                                marker = '(%s) and %s' % (env, marker)
+                            else:
+                                marker = env
+                        reqts.add(';'.join((r, marker)))
+            return reqts
+
+        assert self._data and not self._legacy
+        result = LegacyMetadata()
+        nmd = self._data
+        # import pdb; pdb.set_trace()
+        for nk, ok in self.LEGACY_MAPPING.items():
+            if not isinstance(nk, tuple):
+                if nk in nmd:
+                    result[ok] = nmd[nk]
+            else:
+                d = nmd
+                found = True
+                for k in nk:
+                    try:
+                        d = d[k]
+                    except (KeyError, IndexError):
+                        found = False
+                        break
+                if found:
+                    result[ok] = d
+        r1 = process_entries(self.run_requires + self.meta_requires)
+        r2 = process_entries(self.build_requires + self.dev_requires)
+        if self.extras:
+            result['Provides-Extra'] = sorted(self.extras)
+        result['Requires-Dist'] = sorted(r1)
+        result['Setup-Requires-Dist'] = sorted(r2)
+        # TODO: any other fields wanted
+        return result
+
+    def write(self, path=None, fileobj=None, legacy=False, skip_unknown=True):
+        if [path, fileobj].count(None) != 1:
+            raise ValueError('Exactly one of path and fileobj is needed')
+        self.validate()
+        if legacy:
+            if self._legacy:
+                legacy_md = self._legacy
+            else:
+                legacy_md = self._to_legacy()
+            if path:
+                legacy_md.write(path, skip_unknown=skip_unknown)
+            else:
+                legacy_md.write_file(fileobj, skip_unknown=skip_unknown)
+        else:
+            if self._legacy:
+                d = self._from_legacy()
+            else:
+                d = self._data
+            if fileobj:
+                json.dump(d, fileobj, ensure_ascii=True, indent=2,
+                          sort_keys=True)
+            else:
+                with codecs.open(path, 'w', 'utf-8') as f:
+                    json.dump(d, f, ensure_ascii=True, indent=2,
+                              sort_keys=True)
+
+    def add_requirements(self, requirements):
+        if self._legacy:
+            self._legacy.add_requirements(requirements)
+        else:
+            run_requires = self._data.setdefault('run_requires', [])
+            always = None
+            for entry in run_requires:
+                if 'environment' not in entry and 'extra' not in entry:
+                    always = entry
+                    break
+            if always is None:
+                always = { 'requires': requirements }
+                run_requires.insert(0, always)
+            else:
+                rset = set(always['requires']) | set(requirements)
+                always['requires'] = sorted(rset)
+
+    def __repr__(self):
+        name = self.name or '(no name)'
+        version = self.version or 'no version'
+        return '<%s %s %s (%s)>' % (self.__class__.__name__,
+                                    self.metadata_version, name, version)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/resources.py

@@ -0,0 +1,358 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2013-2017 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+from __future__ import unicode_literals
+
+import bisect
+import io
+import logging
+import os
+import pkgutil
+import sys
+import types
+import zipimport
+
+from . import DistlibException
+from .util import cached_property, get_cache_base, Cache
+
+logger = logging.getLogger(__name__)
+
+
+cache = None    # created when needed
+
+
+class ResourceCache(Cache):
+    def __init__(self, base=None):
+        if base is None:
+            # Use native string to avoid issues on 2.x: see Python #20140.
+            base = os.path.join(get_cache_base(), str('resource-cache'))
+        super(ResourceCache, self).__init__(base)
+
+    def is_stale(self, resource, path):
+        """
+        Is the cache stale for the given resource?
+
+        :param resource: The :class:`Resource` being cached.
+        :param path: The path of the resource in the cache.
+        :return: True if the cache is stale.
+        """
+        # Cache invalidation is a hard problem :-)
+        return True
+
+    def get(self, resource):
+        """
+        Get a resource into the cache,
+
+        :param resource: A :class:`Resource` instance.
+        :return: The pathname of the resource in the cache.
+        """
+        prefix, path = resource.finder.get_cache_info(resource)
+        if prefix is None:
+            result = path
+        else:
+            result = os.path.join(self.base, self.prefix_to_dir(prefix), path)
+            dirname = os.path.dirname(result)
+            if not os.path.isdir(dirname):
+                os.makedirs(dirname)
+            if not os.path.exists(result):
+                stale = True
+            else:
+                stale = self.is_stale(resource, path)
+            if stale:
+                # write the bytes of the resource to the cache location
+                with open(result, 'wb') as f:
+                    f.write(resource.bytes)
+        return result
+
+
+class ResourceBase(object):
+    def __init__(self, finder, name):
+        self.finder = finder
+        self.name = name
+
+
+class Resource(ResourceBase):
+    """
+    A class representing an in-package resource, such as a data file. This is
+    not normally instantiated by user code, but rather by a
+    :class:`ResourceFinder` which manages the resource.
+    """
+    is_container = False        # Backwards compatibility
+
+    def as_stream(self):
+        """
+        Get the resource as a stream.
+
+        This is not a property to make it obvious that it returns a new stream
+        each time.
+        """
+        return self.finder.get_stream(self)
+
+    @cached_property
+    def file_path(self):
+        global cache
+        if cache is None:
+            cache = ResourceCache()
+        return cache.get(self)
+
+    @cached_property
+    def bytes(self):
+        return self.finder.get_bytes(self)
+
+    @cached_property
+    def size(self):
+        return self.finder.get_size(self)
+
+
+class ResourceContainer(ResourceBase):
+    is_container = True     # Backwards compatibility
+
+    @cached_property
+    def resources(self):
+        return self.finder.get_resources(self)
+
+
+class ResourceFinder(object):
+    """
+    Resource finder for file system resources.
+    """
+
+    if sys.platform.startswith('java'):
+        skipped_extensions = ('.pyc', '.pyo', '.class')
+    else:
+        skipped_extensions = ('.pyc', '.pyo')
+
+    def __init__(self, module):
+        self.module = module
+        self.loader = getattr(module, '__loader__', None)
+        self.base = os.path.dirname(getattr(module, '__file__', ''))
+
+    def _adjust_path(self, path):
+        return os.path.realpath(path)
+
+    def _make_path(self, resource_name):
+        # Issue #50: need to preserve type of path on Python 2.x
+        # like os.path._get_sep
+        if isinstance(resource_name, bytes):    # should only happen on 2.x
+            sep = b'/'
+        else:
+            sep = '/'
+        parts = resource_name.split(sep)
+        parts.insert(0, self.base)
+        result = os.path.join(*parts)
+        return self._adjust_path(result)
+
+    def _find(self, path):
+        return os.path.exists(path)
+
+    def get_cache_info(self, resource):
+        return None, resource.path
+
+    def find(self, resource_name):
+        path = self._make_path(resource_name)
+        if not self._find(path):
+            result = None
+        else:
+            if self._is_directory(path):
+                result = ResourceContainer(self, resource_name)
+            else:
+                result = Resource(self, resource_name)
+            result.path = path
+        return result
+
+    def get_stream(self, resource):
+        return open(resource.path, 'rb')
+
+    def get_bytes(self, resource):
+        with open(resource.path, 'rb') as f:
+            return f.read()
+
+    def get_size(self, resource):
+        return os.path.getsize(resource.path)
+
+    def get_resources(self, resource):
+        def allowed(f):
+            return (f != '__pycache__' and not
+                    f.endswith(self.skipped_extensions))
+        return set([f for f in os.listdir(resource.path) if allowed(f)])
+
+    def is_container(self, resource):
+        return self._is_directory(resource.path)
+
+    _is_directory = staticmethod(os.path.isdir)
+
+    def iterator(self, resource_name):
+        resource = self.find(resource_name)
+        if resource is not None:
+            todo = [resource]
+            while todo:
+                resource = todo.pop(0)
+                yield resource
+                if resource.is_container:
+                    rname = resource.name
+                    for name in resource.resources:
+                        if not rname:
+                            new_name = name
+                        else:
+                            new_name = '/'.join([rname, name])
+                        child = self.find(new_name)
+                        if child.is_container:
+                            todo.append(child)
+                        else:
+                            yield child
+
+
+class ZipResourceFinder(ResourceFinder):
+    """
+    Resource finder for resources in .zip files.
+    """
+    def __init__(self, module):
+        super(ZipResourceFinder, self).__init__(module)
+        archive = self.loader.archive
+        self.prefix_len = 1 + len(archive)
+        # PyPy doesn't have a _files attr on zipimporter, and you can't set one
+        if hasattr(self.loader, '_files'):
+            self._files = self.loader._files
+        else:
+            self._files = zipimport._zip_directory_cache[archive]
+        self.index = sorted(self._files)
+
+    def _adjust_path(self, path):
+        return path
+
+    def _find(self, path):
+        path = path[self.prefix_len:]
+        if path in self._files:
+            result = True
+        else:
+            if path and path[-1] != os.sep:
+                path = path + os.sep
+            i = bisect.bisect(self.index, path)
+            try:
+                result = self.index[i].startswith(path)
+            except IndexError:
+                result = False
+        if not result:
+            logger.debug('_find failed: %r %r', path, self.loader.prefix)
+        else:
+            logger.debug('_find worked: %r %r', path, self.loader.prefix)
+        return result
+
+    def get_cache_info(self, resource):
+        prefix = self.loader.archive
+        path = resource.path[1 + len(prefix):]
+        return prefix, path
+
+    def get_bytes(self, resource):
+        return self.loader.get_data(resource.path)
+
+    def get_stream(self, resource):
+        return io.BytesIO(self.get_bytes(resource))
+
+    def get_size(self, resource):
+        path = resource.path[self.prefix_len:]
+        return self._files[path][3]
+
+    def get_resources(self, resource):
+        path = resource.path[self.prefix_len:]
+        if path and path[-1] != os.sep:
+            path += os.sep
+        plen = len(path)
+        result = set()
+        i = bisect.bisect(self.index, path)
+        while i < len(self.index):
+            if not self.index[i].startswith(path):
+                break
+            s = self.index[i][plen:]
+            result.add(s.split(os.sep, 1)[0])   # only immediate children
+            i += 1
+        return result
+
+    def _is_directory(self, path):
+        path = path[self.prefix_len:]
+        if path and path[-1] != os.sep:
+            path += os.sep
+        i = bisect.bisect(self.index, path)
+        try:
+            result = self.index[i].startswith(path)
+        except IndexError:
+            result = False
+        return result
+
+
+_finder_registry = {
+    type(None): ResourceFinder,
+    zipimport.zipimporter: ZipResourceFinder
+}
+
+try:
+    # In Python 3.6, _frozen_importlib -> _frozen_importlib_external
+    try:
+        import _frozen_importlib_external as _fi
+    except ImportError:
+        import _frozen_importlib as _fi
+    _finder_registry[_fi.SourceFileLoader] = ResourceFinder
+    _finder_registry[_fi.FileFinder] = ResourceFinder
+    # See issue #146
+    _finder_registry[_fi.SourcelessFileLoader] = ResourceFinder
+    del _fi
+except (ImportError, AttributeError):
+    pass
+
+
+def register_finder(loader, finder_maker):
+    _finder_registry[type(loader)] = finder_maker
+
+
+_finder_cache = {}
+
+
+def finder(package):
+    """
+    Return a resource finder for a package.
+    :param package: The name of the package.
+    :return: A :class:`ResourceFinder` instance for the package.
+    """
+    if package in _finder_cache:
+        result = _finder_cache[package]
+    else:
+        if package not in sys.modules:
+            __import__(package)
+        module = sys.modules[package]
+        path = getattr(module, '__path__', None)
+        if path is None:
+            raise DistlibException('You cannot get a finder for a module, '
+                                   'only for a package')
+        loader = getattr(module, '__loader__', None)
+        finder_maker = _finder_registry.get(type(loader))
+        if finder_maker is None:
+            raise DistlibException('Unable to locate finder for %r' % package)
+        result = finder_maker(module)
+        _finder_cache[package] = result
+    return result
+
+
+_dummy_module = types.ModuleType(str('__dummy__'))
+
+
+def finder_for_path(path):
+    """
+    Return a resource finder for a path, which should represent a container.
+
+    :param path: The path.
+    :return: A :class:`ResourceFinder` instance for the path.
+    """
+    result = None
+    # calls any path hooks, gets importer into cache
+    pkgutil.get_importer(path)
+    loader = sys.path_importer_cache.get(path)
+    finder = _finder_registry.get(type(loader))
+    if finder:
+        module = _dummy_module
+        module.__file__ = os.path.join(path, '')
+        module.__loader__ = loader
+        result = finder(module)
+    return result

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/scripts.py

@@ -0,0 +1,452 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2013-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+from io import BytesIO
+import logging
+import os
+import re
+import struct
+import sys
+import time
+from zipfile import ZipInfo
+
+from .compat import sysconfig, detect_encoding, ZipFile
+from .resources import finder
+from .util import (FileOperator, get_export_entry, convert_path,
+                   get_executable, get_platform, in_venv)
+
+logger = logging.getLogger(__name__)
+
+_DEFAULT_MANIFEST = '''
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0">
+ <assemblyIdentity version="1.0.0.0"
+ processorArchitecture="X86"
+ name="%s"
+ type="win32"/>
+
+ <!-- Identify the application security requirements. -->
+ <trustInfo xmlns="urn:schemas-microsoft-com:asm.v3">
+ <security>
+ <requestedPrivileges>
+ <requestedExecutionLevel level="asInvoker" uiAccess="false"/>
+ </requestedPrivileges>
+ </security>
+ </trustInfo>
+</assembly>'''.strip()
+
+# check if Python is called on the first line with this expression
+FIRST_LINE_RE = re.compile(b'^#!.*pythonw?[0-9.]*([ \t].*)?$')
+SCRIPT_TEMPLATE = r'''# -*- coding: utf-8 -*-
+import re
+import sys
+from %(module)s import %(import_name)s
+if __name__ == '__main__':
+    sys.argv[0] = re.sub(r'(-script\.pyw|\.exe)?$', '', sys.argv[0])
+    sys.exit(%(func)s())
+'''
+
+
+def enquote_executable(executable):
+    if ' ' in executable:
+        # make sure we quote only the executable in case of env
+        # for example /usr/bin/env "/dir with spaces/bin/jython"
+        # instead of "/usr/bin/env /dir with spaces/bin/jython"
+        # otherwise whole
+        if executable.startswith('/usr/bin/env '):
+            env, _executable = executable.split(' ', 1)
+            if ' ' in _executable and not _executable.startswith('"'):
+                executable = '%s "%s"' % (env, _executable)
+        else:
+            if not executable.startswith('"'):
+                executable = '"%s"' % executable
+    return executable
+
+
+# Keep the old name around (for now), as there is at least one project using it!
+_enquote_executable = enquote_executable
+
+
+class ScriptMaker(object):
+    """
+    A class to copy or create scripts from source scripts or callable
+    specifications.
+    """
+    script_template = SCRIPT_TEMPLATE
+
+    executable = None  # for shebangs
+
+    def __init__(self,
+                 source_dir,
+                 target_dir,
+                 add_launchers=True,
+                 dry_run=False,
+                 fileop=None):
+        self.source_dir = source_dir
+        self.target_dir = target_dir
+        self.add_launchers = add_launchers
+        self.force = False
+        self.clobber = False
+        # It only makes sense to set mode bits on POSIX.
+        self.set_mode = (os.name == 'posix') or (os.name == 'java'
+                                                 and os._name == 'posix')
+        self.variants = set(('', 'X.Y'))
+        self._fileop = fileop or FileOperator(dry_run)
+
+        self._is_nt = os.name == 'nt' or (os.name == 'java'
+                                          and os._name == 'nt')
+        self.version_info = sys.version_info
+
+    def _get_alternate_executable(self, executable, options):
+        if options.get('gui', False) and self._is_nt:  # pragma: no cover
+            dn, fn = os.path.split(executable)
+            fn = fn.replace('python', 'pythonw')
+            executable = os.path.join(dn, fn)
+        return executable
+
+    if sys.platform.startswith('java'):  # pragma: no cover
+
+        def _is_shell(self, executable):
+            """
+            Determine if the specified executable is a script
+            (contains a #! line)
+            """
+            try:
+                with open(executable) as fp:
+                    return fp.read(2) == '#!'
+            except (OSError, IOError):
+                logger.warning('Failed to open %s', executable)
+                return False
+
+        def _fix_jython_executable(self, executable):
+            if self._is_shell(executable):
+                # Workaround for Jython is not needed on Linux systems.
+                import java
+
+                if java.lang.System.getProperty('os.name') == 'Linux':
+                    return executable
+            elif executable.lower().endswith('jython.exe'):
+                # Use wrapper exe for Jython on Windows
+                return executable
+            return '/usr/bin/env %s' % executable
+
+    def _build_shebang(self, executable, post_interp):
+        """
+        Build a shebang line. In the simple case (on Windows, or a shebang line
+        which is not too long or contains spaces) use a simple formulation for
+        the shebang. Otherwise, use /bin/sh as the executable, with a contrived
+        shebang which allows the script to run either under Python or sh, using
+        suitable quoting. Thanks to Harald Nordgren for his input.
+
+        See also: http://www.in-ulm.de/~mascheck/various/shebang/#length
+                  https://hg.mozilla.org/mozilla-central/file/tip/mach
+        """
+        if os.name != 'posix':
+            simple_shebang = True
+        else:
+            # Add 3 for '#!' prefix and newline suffix.
+            shebang_length = len(executable) + len(post_interp) + 3
+            if sys.platform == 'darwin':
+                max_shebang_length = 512
+            else:
+                max_shebang_length = 127
+            simple_shebang = ((b' ' not in executable)
+                              and (shebang_length <= max_shebang_length))
+
+        if simple_shebang:
+            result = b'#!' + executable + post_interp + b'\n'
+        else:
+            result = b'#!/bin/sh\n'
+            result += b"'''exec' " + executable + post_interp + b' "$0" "$@"\n'
+            result += b"' '''"
+        return result
+
+    def _get_shebang(self, encoding, post_interp=b'', options=None):
+        enquote = True
+        if self.executable:
+            executable = self.executable
+            enquote = False  # assume this will be taken care of
+        elif not sysconfig.is_python_build():
+            executable = get_executable()
+        elif in_venv():  # pragma: no cover
+            executable = os.path.join(
+                sysconfig.get_path('scripts'),
+                'python%s' % sysconfig.get_config_var('EXE'))
+        else:  # pragma: no cover
+            if os.name == 'nt':
+                # for Python builds from source on Windows, no Python executables with
+                # a version suffix are created, so we use python.exe
+                executable = os.path.join(
+                    sysconfig.get_config_var('BINDIR'),
+                    'python%s' % (sysconfig.get_config_var('EXE')))
+            else:
+                executable = os.path.join(
+                    sysconfig.get_config_var('BINDIR'),
+                    'python%s%s' % (sysconfig.get_config_var('VERSION'),
+                                    sysconfig.get_config_var('EXE')))
+        if options:
+            executable = self._get_alternate_executable(executable, options)
+
+        if sys.platform.startswith('java'):  # pragma: no cover
+            executable = self._fix_jython_executable(executable)
+
+        # Normalise case for Windows - COMMENTED OUT
+        # executable = os.path.normcase(executable)
+        # N.B. The normalising operation above has been commented out: See
+        # issue #124. Although paths in Windows are generally case-insensitive,
+        # they aren't always. For example, a path containing a ẞ (which is a
+        # LATIN CAPITAL LETTER SHARP S - U+1E9E) is normcased to ß (which is a
+        # LATIN SMALL LETTER SHARP S' - U+00DF). The two are not considered by
+        # Windows as equivalent in path names.
+
+        # If the user didn't specify an executable, it may be necessary to
+        # cater for executable paths with spaces (not uncommon on Windows)
+        if enquote:
+            executable = enquote_executable(executable)
+        # Issue #51: don't use fsencode, since we later try to
+        # check that the shebang is decodable using utf-8.
+        executable = executable.encode('utf-8')
+        # in case of IronPython, play safe and enable frames support
+        if (sys.platform == 'cli' and '-X:Frames' not in post_interp
+                and '-X:FullFrames' not in post_interp):  # pragma: no cover
+            post_interp += b' -X:Frames'
+        shebang = self._build_shebang(executable, post_interp)
+        # Python parser starts to read a script using UTF-8 until
+        # it gets a #coding:xxx cookie. The shebang has to be the
+        # first line of a file, the #coding:xxx cookie cannot be
+        # written before. So the shebang has to be decodable from
+        # UTF-8.
+        try:
+            shebang.decode('utf-8')
+        except UnicodeDecodeError:  # pragma: no cover
+            raise ValueError('The shebang (%r) is not decodable from utf-8' %
+                             shebang)
+        # If the script is encoded to a custom encoding (use a
+        # #coding:xxx cookie), the shebang has to be decodable from
+        # the script encoding too.
+        if encoding != 'utf-8':
+            try:
+                shebang.decode(encoding)
+            except UnicodeDecodeError:  # pragma: no cover
+                raise ValueError('The shebang (%r) is not decodable '
+                                 'from the script encoding (%r)' %
+                                 (shebang, encoding))
+        return shebang
+
+    def _get_script_text(self, entry):
+        return self.script_template % dict(
+            module=entry.prefix,
+            import_name=entry.suffix.split('.')[0],
+            func=entry.suffix)
+
+    manifest = _DEFAULT_MANIFEST
+
+    def get_manifest(self, exename):
+        base = os.path.basename(exename)
+        return self.manifest % base
+
+    def _write_script(self, names, shebang, script_bytes, filenames, ext):
+        use_launcher = self.add_launchers and self._is_nt
+        linesep = os.linesep.encode('utf-8')
+        if not shebang.endswith(linesep):
+            shebang += linesep
+        if not use_launcher:
+            script_bytes = shebang + script_bytes
+        else:  # pragma: no cover
+            if ext == 'py':
+                launcher = self._get_launcher('t')
+            else:
+                launcher = self._get_launcher('w')
+            stream = BytesIO()
+            with ZipFile(stream, 'w') as zf:
+                source_date_epoch = os.environ.get('SOURCE_DATE_EPOCH')
+                if source_date_epoch:
+                    date_time = time.gmtime(int(source_date_epoch))[:6]
+                    zinfo = ZipInfo(filename='__main__.py',
+                                    date_time=date_time)
+                    zf.writestr(zinfo, script_bytes)
+                else:
+                    zf.writestr('__main__.py', script_bytes)
+            zip_data = stream.getvalue()
+            script_bytes = launcher + shebang + zip_data
+        for name in names:
+            outname = os.path.join(self.target_dir, name)
+            if use_launcher:  # pragma: no cover
+                n, e = os.path.splitext(outname)
+                if e.startswith('.py'):
+                    outname = n
+                outname = '%s.exe' % outname
+                try:
+                    self._fileop.write_binary_file(outname, script_bytes)
+                except Exception:
+                    # Failed writing an executable - it might be in use.
+                    logger.warning('Failed to write executable - trying to '
+                                   'use .deleteme logic')
+                    dfname = '%s.deleteme' % outname
+                    if os.path.exists(dfname):
+                        os.remove(dfname)  # Not allowed to fail here
+                    os.rename(outname, dfname)  # nor here
+                    self._fileop.write_binary_file(outname, script_bytes)
+                    logger.debug('Able to replace executable using '
+                                 '.deleteme logic')
+                    try:
+                        os.remove(dfname)
+                    except Exception:
+                        pass  # still in use - ignore error
+            else:
+                if self._is_nt and not outname.endswith(
+                        '.' + ext):  # pragma: no cover
+                    outname = '%s.%s' % (outname, ext)
+                if os.path.exists(outname) and not self.clobber:
+                    logger.warning('Skipping existing file %s', outname)
+                    continue
+                self._fileop.write_binary_file(outname, script_bytes)
+                if self.set_mode:
+                    self._fileop.set_executable_mode([outname])
+            filenames.append(outname)
+
+    variant_separator = '-'
+
+    def get_script_filenames(self, name):
+        result = set()
+        if '' in self.variants:
+            result.add(name)
+        if 'X' in self.variants:
+            result.add('%s%s' % (name, self.version_info[0]))
+        if 'X.Y' in self.variants:
+            result.add('%s%s%s.%s' %
+                       (name, self.variant_separator, self.version_info[0],
+                        self.version_info[1]))
+        return result
+
+    def _make_script(self, entry, filenames, options=None):
+        post_interp = b''
+        if options:
+            args = options.get('interpreter_args', [])
+            if args:
+                args = ' %s' % ' '.join(args)
+                post_interp = args.encode('utf-8')
+        shebang = self._get_shebang('utf-8', post_interp, options=options)
+        script = self._get_script_text(entry).encode('utf-8')
+        scriptnames = self.get_script_filenames(entry.name)
+        if options and options.get('gui', False):
+            ext = 'pyw'
+        else:
+            ext = 'py'
+        self._write_script(scriptnames, shebang, script, filenames, ext)
+
+    def _copy_script(self, script, filenames):
+        adjust = False
+        script = os.path.join(self.source_dir, convert_path(script))
+        outname = os.path.join(self.target_dir, os.path.basename(script))
+        if not self.force and not self._fileop.newer(script, outname):
+            logger.debug('not copying %s (up-to-date)', script)
+            return
+
+        # Always open the file, but ignore failures in dry-run mode --
+        # that way, we'll get accurate feedback if we can read the
+        # script.
+        try:
+            f = open(script, 'rb')
+        except IOError:  # pragma: no cover
+            if not self.dry_run:
+                raise
+            f = None
+        else:
+            first_line = f.readline()
+            if not first_line:  # pragma: no cover
+                logger.warning('%s is an empty file (skipping)', script)
+                return
+
+            match = FIRST_LINE_RE.match(first_line.replace(b'\r\n', b'\n'))
+            if match:
+                adjust = True
+                post_interp = match.group(1) or b''
+
+        if not adjust:
+            if f:
+                f.close()
+            self._fileop.copy_file(script, outname)
+            if self.set_mode:
+                self._fileop.set_executable_mode([outname])
+            filenames.append(outname)
+        else:
+            logger.info('copying and adjusting %s -> %s', script,
+                        self.target_dir)
+            if not self._fileop.dry_run:
+                encoding, lines = detect_encoding(f.readline)
+                f.seek(0)
+                shebang = self._get_shebang(encoding, post_interp)
+                if b'pythonw' in first_line:  # pragma: no cover
+                    ext = 'pyw'
+                else:
+                    ext = 'py'
+                n = os.path.basename(outname)
+                self._write_script([n], shebang, f.read(), filenames, ext)
+            if f:
+                f.close()
+
+    @property
+    def dry_run(self):
+        return self._fileop.dry_run
+
+    @dry_run.setter
+    def dry_run(self, value):
+        self._fileop.dry_run = value
+
+    if os.name == 'nt' or (os.name == 'java'
+                           and os._name == 'nt'):  # pragma: no cover
+        # Executable launcher support.
+        # Launchers are from https://bitbucket.org/vinay.sajip/simple_launcher/
+
+        def _get_launcher(self, kind):
+            if struct.calcsize('P') == 8:  # 64-bit
+                bits = '64'
+            else:
+                bits = '32'
+            platform_suffix = '-arm' if get_platform() == 'win-arm64' else ''
+            name = '%s%s%s.exe' % (kind, bits, platform_suffix)
+            # Issue 31: don't hardcode an absolute package name, but
+            # determine it relative to the current package
+            distlib_package = __name__.rsplit('.', 1)[0]
+            resource = finder(distlib_package).find(name)
+            if not resource:
+                msg = ('Unable to find resource %s in package %s' %
+                       (name, distlib_package))
+                raise ValueError(msg)
+            return resource.bytes
+
+    # Public API follows
+
+    def make(self, specification, options=None):
+        """
+        Make a script.
+
+        :param specification: The specification, which is either a valid export
+                              entry specification (to make a script from a
+                              callable) or a filename (to make a script by
+                              copying from a source location).
+        :param options: A dictionary of options controlling script generation.
+        :return: A list of all absolute pathnames written to.
+        """
+        filenames = []
+        entry = get_export_entry(specification)
+        if entry is None:
+            self._copy_script(specification, filenames)
+        else:
+            self._make_script(entry, filenames, options=options)
+        return filenames
+
+    def make_multiple(self, specifications, options=None):
+        """
+        Take a list of specifications and make scripts from them,
+        :param specifications: A list of specifications.
+        :return: A list of all absolute pathnames written to,
+        """
+        filenames = []
+        for specification in specifications:
+            filenames.extend(self.make(specification, options))
+        return filenames

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/util.py

@@ -0,0 +1,2025 @@
+#
+# Copyright (C) 2012-2023 The Python Software Foundation.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+import codecs
+from collections import deque
+import contextlib
+import csv
+from glob import iglob as std_iglob
+import io
+import json
+import logging
+import os
+import py_compile
+import re
+import socket
+try:
+    import ssl
+except ImportError:  # pragma: no cover
+    ssl = None
+import subprocess
+import sys
+import tarfile
+import tempfile
+import textwrap
+
+try:
+    import threading
+except ImportError:  # pragma: no cover
+    import dummy_threading as threading
+import time
+
+from . import DistlibException
+from .compat import (string_types, text_type, shutil, raw_input, StringIO,
+                     cache_from_source, urlopen, urljoin, httplib, xmlrpclib,
+                     HTTPHandler, BaseConfigurator, valid_ident,
+                     Container, configparser, URLError, ZipFile, fsdecode,
+                     unquote, urlparse)
+
+logger = logging.getLogger(__name__)
+
+#
+# Requirement parsing code as per PEP 508
+#
+
+IDENTIFIER = re.compile(r'^([\w\.-]+)\s*')
+VERSION_IDENTIFIER = re.compile(r'^([\w\.*+-]+)\s*')
+COMPARE_OP = re.compile(r'^(<=?|>=?|={2,3}|[~!]=)\s*')
+MARKER_OP = re.compile(r'^((<=?)|(>=?)|={2,3}|[~!]=|in|not\s+in)\s*')
+OR = re.compile(r'^or\b\s*')
+AND = re.compile(r'^and\b\s*')
+NON_SPACE = re.compile(r'(\S+)\s*')
+STRING_CHUNK = re.compile(r'([\s\w\.{}()*+#:;,/?!~`@$%^&=|<>\[\]-]+)')
+
+
+def parse_marker(marker_string):
+    """
+    Parse a marker string and return a dictionary containing a marker expression.
+
+    The dictionary will contain keys "op", "lhs" and "rhs" for non-terminals in
+    the expression grammar, or strings. A string contained in quotes is to be
+    interpreted as a literal string, and a string not contained in quotes is a
+    variable (such as os_name).
+    """
+
+    def marker_var(remaining):
+        # either identifier, or literal string
+        m = IDENTIFIER.match(remaining)
+        if m:
+            result = m.groups()[0]
+            remaining = remaining[m.end():]
+        elif not remaining:
+            raise SyntaxError('unexpected end of input')
+        else:
+            q = remaining[0]
+            if q not in '\'"':
+                raise SyntaxError('invalid expression: %s' % remaining)
+            oq = '\'"'.replace(q, '')
+            remaining = remaining[1:]
+            parts = [q]
+            while remaining:
+                # either a string chunk, or oq, or q to terminate
+                if remaining[0] == q:
+                    break
+                elif remaining[0] == oq:
+                    parts.append(oq)
+                    remaining = remaining[1:]
+                else:
+                    m = STRING_CHUNK.match(remaining)
+                    if not m:
+                        raise SyntaxError('error in string literal: %s' %
+                                          remaining)
+                    parts.append(m.groups()[0])
+                    remaining = remaining[m.end():]
+            else:
+                s = ''.join(parts)
+                raise SyntaxError('unterminated string: %s' % s)
+            parts.append(q)
+            result = ''.join(parts)
+            remaining = remaining[1:].lstrip()  # skip past closing quote
+        return result, remaining
+
+    def marker_expr(remaining):
+        if remaining and remaining[0] == '(':
+            result, remaining = marker(remaining[1:].lstrip())
+            if remaining[0] != ')':
+                raise SyntaxError('unterminated parenthesis: %s' % remaining)
+            remaining = remaining[1:].lstrip()
+        else:
+            lhs, remaining = marker_var(remaining)
+            while remaining:
+                m = MARKER_OP.match(remaining)
+                if not m:
+                    break
+                op = m.groups()[0]
+                remaining = remaining[m.end():]
+                rhs, remaining = marker_var(remaining)
+                lhs = {'op': op, 'lhs': lhs, 'rhs': rhs}
+            result = lhs
+        return result, remaining
+
+    def marker_and(remaining):
+        lhs, remaining = marker_expr(remaining)
+        while remaining:
+            m = AND.match(remaining)
+            if not m:
+                break
+            remaining = remaining[m.end():]
+            rhs, remaining = marker_expr(remaining)
+            lhs = {'op': 'and', 'lhs': lhs, 'rhs': rhs}
+        return lhs, remaining
+
+    def marker(remaining):
+        lhs, remaining = marker_and(remaining)
+        while remaining:
+            m = OR.match(remaining)
+            if not m:
+                break
+            remaining = remaining[m.end():]
+            rhs, remaining = marker_and(remaining)
+            lhs = {'op': 'or', 'lhs': lhs, 'rhs': rhs}
+        return lhs, remaining
+
+    return marker(marker_string)
+
+
+def parse_requirement(req):
+    """
+    Parse a requirement passed in as a string. Return a Container
+    whose attributes contain the various parts of the requirement.
+    """
+    remaining = req.strip()
+    if not remaining or remaining.startswith('#'):
+        return None
+    m = IDENTIFIER.match(remaining)
+    if not m:
+        raise SyntaxError('name expected: %s' % remaining)
+    distname = m.groups()[0]
+    remaining = remaining[m.end():]
+    extras = mark_expr = versions = uri = None
+    if remaining and remaining[0] == '[':
+        i = remaining.find(']', 1)
+        if i < 0:
+            raise SyntaxError('unterminated extra: %s' % remaining)
+        s = remaining[1:i]
+        remaining = remaining[i + 1:].lstrip()
+        extras = []
+        while s:
+            m = IDENTIFIER.match(s)
+            if not m:
+                raise SyntaxError('malformed extra: %s' % s)
+            extras.append(m.groups()[0])
+            s = s[m.end():]
+            if not s:
+                break
+            if s[0] != ',':
+                raise SyntaxError('comma expected in extras: %s' % s)
+            s = s[1:].lstrip()
+        if not extras:
+            extras = None
+    if remaining:
+        if remaining[0] == '@':
+            # it's a URI
+            remaining = remaining[1:].lstrip()
+            m = NON_SPACE.match(remaining)
+            if not m:
+                raise SyntaxError('invalid URI: %s' % remaining)
+            uri = m.groups()[0]
+            t = urlparse(uri)
+            # there are issues with Python and URL parsing, so this test
+            # is a bit crude. See bpo-20271, bpo-23505. Python doesn't
+            # always parse invalid URLs correctly - it should raise
+            # exceptions for malformed URLs
+            if not (t.scheme and t.netloc):
+                raise SyntaxError('Invalid URL: %s' % uri)
+            remaining = remaining[m.end():].lstrip()
+        else:
+
+            def get_versions(ver_remaining):
+                """
+                Return a list of operator, version tuples if any are
+                specified, else None.
+                """
+                m = COMPARE_OP.match(ver_remaining)
+                versions = None
+                if m:
+                    versions = []
+                    while True:
+                        op = m.groups()[0]
+                        ver_remaining = ver_remaining[m.end():]
+                        m = VERSION_IDENTIFIER.match(ver_remaining)
+                        if not m:
+                            raise SyntaxError('invalid version: %s' %
+                                              ver_remaining)
+                        v = m.groups()[0]
+                        versions.append((op, v))
+                        ver_remaining = ver_remaining[m.end():]
+                        if not ver_remaining or ver_remaining[0] != ',':
+                            break
+                        ver_remaining = ver_remaining[1:].lstrip()
+                        # Some packages have a trailing comma which would break things
+                        # See issue #148
+                        if not ver_remaining:
+                            break
+                        m = COMPARE_OP.match(ver_remaining)
+                        if not m:
+                            raise SyntaxError('invalid constraint: %s' %
+                                              ver_remaining)
+                    if not versions:
+                        versions = None
+                return versions, ver_remaining
+
+            if remaining[0] != '(':
+                versions, remaining = get_versions(remaining)
+            else:
+                i = remaining.find(')', 1)
+                if i < 0:
+                    raise SyntaxError('unterminated parenthesis: %s' %
+                                      remaining)
+                s = remaining[1:i]
+                remaining = remaining[i + 1:].lstrip()
+                # As a special diversion from PEP 508, allow a version number
+                # a.b.c in parentheses as a synonym for ~= a.b.c (because this
+                # is allowed in earlier PEPs)
+                if COMPARE_OP.match(s):
+                    versions, _ = get_versions(s)
+                else:
+                    m = VERSION_IDENTIFIER.match(s)
+                    if not m:
+                        raise SyntaxError('invalid constraint: %s' % s)
+                    v = m.groups()[0]
+                    s = s[m.end():].lstrip()
+                    if s:
+                        raise SyntaxError('invalid constraint: %s' % s)
+                    versions = [('~=', v)]
+
+    if remaining:
+        if remaining[0] != ';':
+            raise SyntaxError('invalid requirement: %s' % remaining)
+        remaining = remaining[1:].lstrip()
+
+        mark_expr, remaining = parse_marker(remaining)
+
+    if remaining and remaining[0] != '#':
+        raise SyntaxError('unexpected trailing data: %s' % remaining)
+
+    if not versions:
+        rs = distname
+    else:
+        rs = '%s %s' % (distname, ', '.join(
+            ['%s %s' % con for con in versions]))
+    return Container(name=distname,
+                     extras=extras,
+                     constraints=versions,
+                     marker=mark_expr,
+                     url=uri,
+                     requirement=rs)
+
+
+def get_resources_dests(resources_root, rules):
+    """Find destinations for resources files"""
+
+    def get_rel_path(root, path):
+        # normalizes and returns a lstripped-/-separated path
+        root = root.replace(os.path.sep, '/')
+        path = path.replace(os.path.sep, '/')
+        assert path.startswith(root)
+        return path[len(root):].lstrip('/')
+
+    destinations = {}
+    for base, suffix, dest in rules:
+        prefix = os.path.join(resources_root, base)
+        for abs_base in iglob(prefix):
+            abs_glob = os.path.join(abs_base, suffix)
+            for abs_path in iglob(abs_glob):
+                resource_file = get_rel_path(resources_root, abs_path)
+                if dest is None:  # remove the entry if it was here
+                    destinations.pop(resource_file, None)
+                else:
+                    rel_path = get_rel_path(abs_base, abs_path)
+                    rel_dest = dest.replace(os.path.sep, '/').rstrip('/')
+                    destinations[resource_file] = rel_dest + '/' + rel_path
+    return destinations
+
+
+def in_venv():
+    if hasattr(sys, 'real_prefix'):
+        # virtualenv venvs
+        result = True
+    else:
+        # PEP 405 venvs
+        result = sys.prefix != getattr(sys, 'base_prefix', sys.prefix)
+    return result
+
+
+def get_executable():
+    # The __PYVENV_LAUNCHER__ dance is apparently no longer needed, as
+    # changes to the stub launcher mean that sys.executable always points
+    # to the stub on OS X
+    #    if sys.platform == 'darwin' and ('__PYVENV_LAUNCHER__'
+    #                                     in os.environ):
+    #        result =  os.environ['__PYVENV_LAUNCHER__']
+    #    else:
+    #        result = sys.executable
+    #    return result
+    # Avoid normcasing: see issue #143
+    # result = os.path.normcase(sys.executable)
+    result = sys.executable
+    if not isinstance(result, text_type):
+        result = fsdecode(result)
+    return result
+
+
+def proceed(prompt, allowed_chars, error_prompt=None, default=None):
+    p = prompt
+    while True:
+        s = raw_input(p)
+        p = prompt
+        if not s and default:
+            s = default
+        if s:
+            c = s[0].lower()
+            if c in allowed_chars:
+                break
+            if error_prompt:
+                p = '%c: %s\n%s' % (c, error_prompt, prompt)
+    return c
+
+
+def extract_by_key(d, keys):
+    if isinstance(keys, string_types):
+        keys = keys.split()
+    result = {}
+    for key in keys:
+        if key in d:
+            result[key] = d[key]
+    return result
+
+
+def read_exports(stream):
+    if sys.version_info[0] >= 3:
+        # needs to be a text stream
+        stream = codecs.getreader('utf-8')(stream)
+    # Try to load as JSON, falling back on legacy format
+    data = stream.read()
+    stream = StringIO(data)
+    try:
+        jdata = json.load(stream)
+        result = jdata['extensions']['python.exports']['exports']
+        for group, entries in result.items():
+            for k, v in entries.items():
+                s = '%s = %s' % (k, v)
+                entry = get_export_entry(s)
+                assert entry is not None
+                entries[k] = entry
+        return result
+    except Exception:
+        stream.seek(0, 0)
+
+    def read_stream(cp, stream):
+        if hasattr(cp, 'read_file'):
+            cp.read_file(stream)
+        else:
+            cp.readfp(stream)
+
+    cp = configparser.ConfigParser()
+    try:
+        read_stream(cp, stream)
+    except configparser.MissingSectionHeaderError:
+        stream.close()
+        data = textwrap.dedent(data)
+        stream = StringIO(data)
+        read_stream(cp, stream)
+
+    result = {}
+    for key in cp.sections():
+        result[key] = entries = {}
+        for name, value in cp.items(key):
+            s = '%s = %s' % (name, value)
+            entry = get_export_entry(s)
+            assert entry is not None
+            # entry.dist = self
+            entries[name] = entry
+    return result
+
+
+def write_exports(exports, stream):
+    if sys.version_info[0] >= 3:
+        # needs to be a text stream
+        stream = codecs.getwriter('utf-8')(stream)
+    cp = configparser.ConfigParser()
+    for k, v in exports.items():
+        # TODO check k, v for valid values
+        cp.add_section(k)
+        for entry in v.values():
+            if entry.suffix is None:
+                s = entry.prefix
+            else:
+                s = '%s:%s' % (entry.prefix, entry.suffix)
+            if entry.flags:
+                s = '%s [%s]' % (s, ', '.join(entry.flags))
+            cp.set(k, entry.name, s)
+    cp.write(stream)
+
+
+@contextlib.contextmanager
+def tempdir():
+    td = tempfile.mkdtemp()
+    try:
+        yield td
+    finally:
+        shutil.rmtree(td)
+
+
+@contextlib.contextmanager
+def chdir(d):
+    cwd = os.getcwd()
+    try:
+        os.chdir(d)
+        yield
+    finally:
+        os.chdir(cwd)
+
+
+@contextlib.contextmanager
+def socket_timeout(seconds=15):
+    cto = socket.getdefaulttimeout()
+    try:
+        socket.setdefaulttimeout(seconds)
+        yield
+    finally:
+        socket.setdefaulttimeout(cto)
+
+
+class cached_property(object):
+
+    def __init__(self, func):
+        self.func = func
+        # for attr in ('__name__', '__module__', '__doc__'):
+        #     setattr(self, attr, getattr(func, attr, None))
+
+    def __get__(self, obj, cls=None):
+        if obj is None:
+            return self
+        value = self.func(obj)
+        object.__setattr__(obj, self.func.__name__, value)
+        # obj.__dict__[self.func.__name__] = value = self.func(obj)
+        return value
+
+
+def convert_path(pathname):
+    """Return 'pathname' as a name that will work on the native filesystem.
+
+    The path is split on '/' and put back together again using the current
+    directory separator.  Needed because filenames in the setup script are
+    always supplied in Unix style, and have to be converted to the local
+    convention before we can actually use them in the filesystem.  Raises
+    ValueError on non-Unix-ish systems if 'pathname' either starts or
+    ends with a slash.
+    """
+    if os.sep == '/':
+        return pathname
+    if not pathname:
+        return pathname
+    if pathname[0] == '/':
+        raise ValueError("path '%s' cannot be absolute" % pathname)
+    if pathname[-1] == '/':
+        raise ValueError("path '%s' cannot end with '/'" % pathname)
+
+    paths = pathname.split('/')
+    while os.curdir in paths:
+        paths.remove(os.curdir)
+    if not paths:
+        return os.curdir
+    return os.path.join(*paths)
+
+
+class FileOperator(object):
+
+    def __init__(self, dry_run=False):
+        self.dry_run = dry_run
+        self.ensured = set()
+        self._init_record()
+
+    def _init_record(self):
+        self.record = False
+        self.files_written = set()
+        self.dirs_created = set()
+
+    def record_as_written(self, path):
+        if self.record:
+            self.files_written.add(path)
+
+    def newer(self, source, target):
+        """Tell if the target is newer than the source.
+
+        Returns true if 'source' exists and is more recently modified than
+        'target', or if 'source' exists and 'target' doesn't.
+
+        Returns false if both exist and 'target' is the same age or younger
+        than 'source'. Raise PackagingFileError if 'source' does not exist.
+
+        Note that this test is not very accurate: files created in the same
+        second will have the same "age".
+        """
+        if not os.path.exists(source):
+            raise DistlibException("file '%r' does not exist" %
+                                   os.path.abspath(source))
+        if not os.path.exists(target):
+            return True
+
+        return os.stat(source).st_mtime > os.stat(target).st_mtime
+
+    def copy_file(self, infile, outfile, check=True):
+        """Copy a file respecting dry-run and force flags.
+        """
+        self.ensure_dir(os.path.dirname(outfile))
+        logger.info('Copying %s to %s', infile, outfile)
+        if not self.dry_run:
+            msg = None
+            if check:
+                if os.path.islink(outfile):
+                    msg = '%s is a symlink' % outfile
+                elif os.path.exists(outfile) and not os.path.isfile(outfile):
+                    msg = '%s is a non-regular file' % outfile
+            if msg:
+                raise ValueError(msg + ' which would be overwritten')
+            shutil.copyfile(infile, outfile)
+        self.record_as_written(outfile)
+
+    def copy_stream(self, instream, outfile, encoding=None):
+        assert not os.path.isdir(outfile)
+        self.ensure_dir(os.path.dirname(outfile))
+        logger.info('Copying stream %s to %s', instream, outfile)
+        if not self.dry_run:
+            if encoding is None:
+                outstream = open(outfile, 'wb')
+            else:
+                outstream = codecs.open(outfile, 'w', encoding=encoding)
+            try:
+                shutil.copyfileobj(instream, outstream)
+            finally:
+                outstream.close()
+        self.record_as_written(outfile)
+
+    def write_binary_file(self, path, data):
+        self.ensure_dir(os.path.dirname(path))
+        if not self.dry_run:
+            if os.path.exists(path):
+                os.remove(path)
+            with open(path, 'wb') as f:
+                f.write(data)
+        self.record_as_written(path)
+
+    def write_text_file(self, path, data, encoding):
+        self.write_binary_file(path, data.encode(encoding))
+
+    def set_mode(self, bits, mask, files):
+        if os.name == 'posix' or (os.name == 'java' and os._name == 'posix'):
+            # Set the executable bits (owner, group, and world) on
+            # all the files specified.
+            for f in files:
+                if self.dry_run:
+                    logger.info("changing mode of %s", f)
+                else:
+                    mode = (os.stat(f).st_mode | bits) & mask
+                    logger.info("changing mode of %s to %o", f, mode)
+                    os.chmod(f, mode)
+
+    set_executable_mode = lambda s, f: s.set_mode(0o555, 0o7777, f)
+
+    def ensure_dir(self, path):
+        path = os.path.abspath(path)
+        if path not in self.ensured and not os.path.exists(path):
+            self.ensured.add(path)
+            d, f = os.path.split(path)
+            self.ensure_dir(d)
+            logger.info('Creating %s' % path)
+            if not self.dry_run:
+                os.mkdir(path)
+            if self.record:
+                self.dirs_created.add(path)
+
+    def byte_compile(self,
+                     path,
+                     optimize=False,
+                     force=False,
+                     prefix=None,
+                     hashed_invalidation=False):
+        dpath = cache_from_source(path, not optimize)
+        logger.info('Byte-compiling %s to %s', path, dpath)
+        if not self.dry_run:
+            if force or self.newer(path, dpath):
+                if not prefix:
+                    diagpath = None
+                else:
+                    assert path.startswith(prefix)
+                    diagpath = path[len(prefix):]
+            compile_kwargs = {}
+            if hashed_invalidation and hasattr(py_compile,
+                                               'PycInvalidationMode'):
+                compile_kwargs[
+                    'invalidation_mode'] = py_compile.PycInvalidationMode.CHECKED_HASH
+            py_compile.compile(path, dpath, diagpath, True,
+                               **compile_kwargs)  # raise error
+        self.record_as_written(dpath)
+        return dpath
+
+    def ensure_removed(self, path):
+        if os.path.exists(path):
+            if os.path.isdir(path) and not os.path.islink(path):
+                logger.debug('Removing directory tree at %s', path)
+                if not self.dry_run:
+                    shutil.rmtree(path)
+                if self.record:
+                    if path in self.dirs_created:
+                        self.dirs_created.remove(path)
+            else:
+                if os.path.islink(path):
+                    s = 'link'
+                else:
+                    s = 'file'
+                logger.debug('Removing %s %s', s, path)
+                if not self.dry_run:
+                    os.remove(path)
+                if self.record:
+                    if path in self.files_written:
+                        self.files_written.remove(path)
+
+    def is_writable(self, path):
+        result = False
+        while not result:
+            if os.path.exists(path):
+                result = os.access(path, os.W_OK)
+                break
+            parent = os.path.dirname(path)
+            if parent == path:
+                break
+            path = parent
+        return result
+
+    def commit(self):
+        """
+        Commit recorded changes, turn off recording, return
+        changes.
+        """
+        assert self.record
+        result = self.files_written, self.dirs_created
+        self._init_record()
+        return result
+
+    def rollback(self):
+        if not self.dry_run:
+            for f in list(self.files_written):
+                if os.path.exists(f):
+                    os.remove(f)
+            # dirs should all be empty now, except perhaps for
+            # __pycache__ subdirs
+            # reverse so that subdirs appear before their parents
+            dirs = sorted(self.dirs_created, reverse=True)
+            for d in dirs:
+                flist = os.listdir(d)
+                if flist:
+                    assert flist == ['__pycache__']
+                    sd = os.path.join(d, flist[0])
+                    os.rmdir(sd)
+                os.rmdir(d)  # should fail if non-empty
+        self._init_record()
+
+
+def resolve(module_name, dotted_path):
+    if module_name in sys.modules:
+        mod = sys.modules[module_name]
+    else:
+        mod = __import__(module_name)
+    if dotted_path is None:
+        result = mod
+    else:
+        parts = dotted_path.split('.')
+        result = getattr(mod, parts.pop(0))
+        for p in parts:
+            result = getattr(result, p)
+    return result
+
+
+class ExportEntry(object):
+
+    def __init__(self, name, prefix, suffix, flags):
+        self.name = name
+        self.prefix = prefix
+        self.suffix = suffix
+        self.flags = flags
+
+    @cached_property
+    def value(self):
+        return resolve(self.prefix, self.suffix)
+
+    def __repr__(self):  # pragma: no cover
+        return '<ExportEntry %s = %s:%s %s>' % (self.name, self.prefix,
+                                                self.suffix, self.flags)
+
+    def __eq__(self, other):
+        if not isinstance(other, ExportEntry):
+            result = False
+        else:
+            result = (self.name == other.name and self.prefix == other.prefix
+                      and self.suffix == other.suffix
+                      and self.flags == other.flags)
+        return result
+
+    __hash__ = object.__hash__
+
+
+ENTRY_RE = re.compile(
+    r'''(?P<name>([^\[]\S*))
+                      \s*=\s*(?P<callable>(\w+)([:\.]\w+)*)
+                      \s*(\[\s*(?P<flags>[\w-]+(=\w+)?(,\s*\w+(=\w+)?)*)\s*\])?
+                      ''', re.VERBOSE)
+
+
+def get_export_entry(specification):
+    m = ENTRY_RE.search(specification)
+    if not m:
+        result = None
+        if '[' in specification or ']' in specification:
+            raise DistlibException("Invalid specification "
+                                   "'%s'" % specification)
+    else:
+        d = m.groupdict()
+        name = d['name']
+        path = d['callable']
+        colons = path.count(':')
+        if colons == 0:
+            prefix, suffix = path, None
+        else:
+            if colons != 1:
+                raise DistlibException("Invalid specification "
+                                       "'%s'" % specification)
+            prefix, suffix = path.split(':')
+        flags = d['flags']
+        if flags is None:
+            if '[' in specification or ']' in specification:
+                raise DistlibException("Invalid specification "
+                                       "'%s'" % specification)
+            flags = []
+        else:
+            flags = [f.strip() for f in flags.split(',')]
+        result = ExportEntry(name, prefix, suffix, flags)
+    return result
+
+
+def get_cache_base(suffix=None):
+    """
+    Return the default base location for distlib caches. If the directory does
+    not exist, it is created. Use the suffix provided for the base directory,
+    and default to '.distlib' if it isn't provided.
+
+    On Windows, if LOCALAPPDATA is defined in the environment, then it is
+    assumed to be a directory, and will be the parent directory of the result.
+    On POSIX, and on Windows if LOCALAPPDATA is not defined, the user's home
+    directory - using os.expanduser('~') - will be the parent directory of
+    the result.
+
+    The result is just the directory '.distlib' in the parent directory as
+    determined above, or with the name specified with ``suffix``.
+    """
+    if suffix is None:
+        suffix = '.distlib'
+    if os.name == 'nt' and 'LOCALAPPDATA' in os.environ:
+        result = os.path.expandvars('$localappdata')
+    else:
+        # Assume posix, or old Windows
+        result = os.path.expanduser('~')
+    # we use 'isdir' instead of 'exists', because we want to
+    # fail if there's a file with that name
+    if os.path.isdir(result):
+        usable = os.access(result, os.W_OK)
+        if not usable:
+            logger.warning('Directory exists but is not writable: %s', result)
+    else:
+        try:
+            os.makedirs(result)
+            usable = True
+        except OSError:
+            logger.warning('Unable to create %s', result, exc_info=True)
+            usable = False
+    if not usable:
+        result = tempfile.mkdtemp()
+        logger.warning('Default location unusable, using %s', result)
+    return os.path.join(result, suffix)
+
+
+def path_to_cache_dir(path):
+    """
+    Convert an absolute path to a directory name for use in a cache.
+
+    The algorithm used is:
+
+    #. On Windows, any ``':'`` in the drive is replaced with ``'---'``.
+    #. Any occurrence of ``os.sep`` is replaced with ``'--'``.
+    #. ``'.cache'`` is appended.
+    """
+    d, p = os.path.splitdrive(os.path.abspath(path))
+    if d:
+        d = d.replace(':', '---')
+    p = p.replace(os.sep, '--')
+    return d + p + '.cache'
+
+
+def ensure_slash(s):
+    if not s.endswith('/'):
+        return s + '/'
+    return s
+
+
+def parse_credentials(netloc):
+    username = password = None
+    if '@' in netloc:
+        prefix, netloc = netloc.rsplit('@', 1)
+        if ':' not in prefix:
+            username = prefix
+        else:
+            username, password = prefix.split(':', 1)
+    if username:
+        username = unquote(username)
+    if password:
+        password = unquote(password)
+    return username, password, netloc
+
+
+def get_process_umask():
+    result = os.umask(0o22)
+    os.umask(result)
+    return result
+
+
+def is_string_sequence(seq):
+    result = True
+    i = None
+    for i, s in enumerate(seq):
+        if not isinstance(s, string_types):
+            result = False
+            break
+    assert i is not None
+    return result
+
+
+PROJECT_NAME_AND_VERSION = re.compile(
+    '([a-z0-9_]+([.-][a-z_][a-z0-9_]*)*)-'
+    '([a-z0-9_.+-]+)', re.I)
+PYTHON_VERSION = re.compile(r'-py(\d\.?\d?)')
+
+
+def split_filename(filename, project_name=None):
+    """
+    Extract name, version, python version from a filename (no extension)
+
+    Return name, version, pyver or None
+    """
+    result = None
+    pyver = None
+    filename = unquote(filename).replace(' ', '-')
+    m = PYTHON_VERSION.search(filename)
+    if m:
+        pyver = m.group(1)
+        filename = filename[:m.start()]
+    if project_name and len(filename) > len(project_name) + 1:
+        m = re.match(re.escape(project_name) + r'\b', filename)
+        if m:
+            n = m.end()
+            result = filename[:n], filename[n + 1:], pyver
+    if result is None:
+        m = PROJECT_NAME_AND_VERSION.match(filename)
+        if m:
+            result = m.group(1), m.group(3), pyver
+    return result
+
+
+# Allow spaces in name because of legacy dists like "Twisted Core"
+NAME_VERSION_RE = re.compile(r'(?P<name>[\w .-]+)\s*'
+                             r'\(\s*(?P<ver>[^\s)]+)\)$')
+
+
+def parse_name_and_version(p):
+    """
+    A utility method used to get name and version from a string.
+
+    From e.g. a Provides-Dist value.
+
+    :param p: A value in a form 'foo (1.0)'
+    :return: The name and version as a tuple.
+    """
+    m = NAME_VERSION_RE.match(p)
+    if not m:
+        raise DistlibException('Ill-formed name/version string: \'%s\'' % p)
+    d = m.groupdict()
+    return d['name'].strip().lower(), d['ver']
+
+
+def get_extras(requested, available):
+    result = set()
+    requested = set(requested or [])
+    available = set(available or [])
+    if '*' in requested:
+        requested.remove('*')
+        result |= available
+    for r in requested:
+        if r == '-':
+            result.add(r)
+        elif r.startswith('-'):
+            unwanted = r[1:]
+            if unwanted not in available:
+                logger.warning('undeclared extra: %s' % unwanted)
+            if unwanted in result:
+                result.remove(unwanted)
+        else:
+            if r not in available:
+                logger.warning('undeclared extra: %s' % r)
+            result.add(r)
+    return result
+
+
+#
+# Extended metadata functionality
+#
+
+
+def _get_external_data(url):
+    result = {}
+    try:
+        # urlopen might fail if it runs into redirections,
+        # because of Python issue #13696. Fixed in locators
+        # using a custom redirect handler.
+        resp = urlopen(url)
+        headers = resp.info()
+        ct = headers.get('Content-Type')
+        if not ct.startswith('application/json'):
+            logger.debug('Unexpected response for JSON request: %s', ct)
+        else:
+            reader = codecs.getreader('utf-8')(resp)
+            # data = reader.read().decode('utf-8')
+            # result = json.loads(data)
+            result = json.load(reader)
+    except Exception as e:
+        logger.exception('Failed to get external data for %s: %s', url, e)
+    return result
+
+
+_external_data_base_url = 'https://www.red-dove.com/pypi/projects/'
+
+
+def get_project_data(name):
+    url = '%s/%s/project.json' % (name[0].upper(), name)
+    url = urljoin(_external_data_base_url, url)
+    result = _get_external_data(url)
+    return result
+
+
+def get_package_data(name, version):
+    url = '%s/%s/package-%s.json' % (name[0].upper(), name, version)
+    url = urljoin(_external_data_base_url, url)
+    return _get_external_data(url)
+
+
+class Cache(object):
+    """
+    A class implementing a cache for resources that need to live in the file system
+    e.g. shared libraries. This class was moved from resources to here because it
+    could be used by other modules, e.g. the wheel module.
+    """
+
+    def __init__(self, base):
+        """
+        Initialise an instance.
+
+        :param base: The base directory where the cache should be located.
+        """
+        # we use 'isdir' instead of 'exists', because we want to
+        # fail if there's a file with that name
+        if not os.path.isdir(base):  # pragma: no cover
+            os.makedirs(base)
+        if (os.stat(base).st_mode & 0o77) != 0:
+            logger.warning('Directory \'%s\' is not private', base)
+        self.base = os.path.abspath(os.path.normpath(base))
+
+    def prefix_to_dir(self, prefix):
+        """
+        Converts a resource prefix to a directory name in the cache.
+        """
+        return path_to_cache_dir(prefix)
+
+    def clear(self):
+        """
+        Clear the cache.
+        """
+        not_removed = []
+        for fn in os.listdir(self.base):
+            fn = os.path.join(self.base, fn)
+            try:
+                if os.path.islink(fn) or os.path.isfile(fn):
+                    os.remove(fn)
+                elif os.path.isdir(fn):
+                    shutil.rmtree(fn)
+            except Exception:
+                not_removed.append(fn)
+        return not_removed
+
+
+class EventMixin(object):
+    """
+    A very simple publish/subscribe system.
+    """
+
+    def __init__(self):
+        self._subscribers = {}
+
+    def add(self, event, subscriber, append=True):
+        """
+        Add a subscriber for an event.
+
+        :param event: The name of an event.
+        :param subscriber: The subscriber to be added (and called when the
+                           event is published).
+        :param append: Whether to append or prepend the subscriber to an
+                       existing subscriber list for the event.
+        """
+        subs = self._subscribers
+        if event not in subs:
+            subs[event] = deque([subscriber])
+        else:
+            sq = subs[event]
+            if append:
+                sq.append(subscriber)
+            else:
+                sq.appendleft(subscriber)
+
+    def remove(self, event, subscriber):
+        """
+        Remove a subscriber for an event.
+
+        :param event: The name of an event.
+        :param subscriber: The subscriber to be removed.
+        """
+        subs = self._subscribers
+        if event not in subs:
+            raise ValueError('No subscribers: %r' % event)
+        subs[event].remove(subscriber)
+
+    def get_subscribers(self, event):
+        """
+        Return an iterator for the subscribers for an event.
+        :param event: The event to return subscribers for.
+        """
+        return iter(self._subscribers.get(event, ()))
+
+    def publish(self, event, *args, **kwargs):
+        """
+        Publish a event and return a list of values returned by its
+        subscribers.
+
+        :param event: The event to publish.
+        :param args: The positional arguments to pass to the event's
+                     subscribers.
+        :param kwargs: The keyword arguments to pass to the event's
+                       subscribers.
+        """
+        result = []
+        for subscriber in self.get_subscribers(event):
+            try:
+                value = subscriber(event, *args, **kwargs)
+            except Exception:
+                logger.exception('Exception during event publication')
+                value = None
+            result.append(value)
+        logger.debug('publish %s: args = %s, kwargs = %s, result = %s', event,
+                     args, kwargs, result)
+        return result
+
+
+#
+# Simple sequencing
+#
+class Sequencer(object):
+
+    def __init__(self):
+        self._preds = {}
+        self._succs = {}
+        self._nodes = set()  # nodes with no preds/succs
+
+    def add_node(self, node):
+        self._nodes.add(node)
+
+    def remove_node(self, node, edges=False):
+        if node in self._nodes:
+            self._nodes.remove(node)
+        if edges:
+            for p in set(self._preds.get(node, ())):
+                self.remove(p, node)
+            for s in set(self._succs.get(node, ())):
+                self.remove(node, s)
+            # Remove empties
+            for k, v in list(self._preds.items()):
+                if not v:
+                    del self._preds[k]
+            for k, v in list(self._succs.items()):
+                if not v:
+                    del self._succs[k]
+
+    def add(self, pred, succ):
+        assert pred != succ
+        self._preds.setdefault(succ, set()).add(pred)
+        self._succs.setdefault(pred, set()).add(succ)
+
+    def remove(self, pred, succ):
+        assert pred != succ
+        try:
+            preds = self._preds[succ]
+            succs = self._succs[pred]
+        except KeyError:  # pragma: no cover
+            raise ValueError('%r not a successor of anything' % succ)
+        try:
+            preds.remove(pred)
+            succs.remove(succ)
+        except KeyError:  # pragma: no cover
+            raise ValueError('%r not a successor of %r' % (succ, pred))
+
+    def is_step(self, step):
+        return (step in self._preds or step in self._succs
+                or step in self._nodes)
+
+    def get_steps(self, final):
+        if not self.is_step(final):
+            raise ValueError('Unknown: %r' % final)
+        result = []
+        todo = []
+        seen = set()
+        todo.append(final)
+        while todo:
+            step = todo.pop(0)
+            if step in seen:
+                # if a step was already seen,
+                # move it to the end (so it will appear earlier
+                # when reversed on return) ... but not for the
+                # final step, as that would be confusing for
+                # users
+                if step != final:
+                    result.remove(step)
+                    result.append(step)
+            else:
+                seen.add(step)
+                result.append(step)
+                preds = self._preds.get(step, ())
+                todo.extend(preds)
+        return reversed(result)
+
+    @property
+    def strong_connections(self):
+        # http://en.wikipedia.org/wiki/Tarjan%27s_strongly_connected_components_algorithm
+        index_counter = [0]
+        stack = []
+        lowlinks = {}
+        index = {}
+        result = []
+
+        graph = self._succs
+
+        def strongconnect(node):
+            # set the depth index for this node to the smallest unused index
+            index[node] = index_counter[0]
+            lowlinks[node] = index_counter[0]
+            index_counter[0] += 1
+            stack.append(node)
+
+            # Consider successors
+            try:
+                successors = graph[node]
+            except Exception:
+                successors = []
+            for successor in successors:
+                if successor not in lowlinks:
+                    # Successor has not yet been visited
+                    strongconnect(successor)
+                    lowlinks[node] = min(lowlinks[node], lowlinks[successor])
+                elif successor in stack:
+                    # the successor is in the stack and hence in the current
+                    # strongly connected component (SCC)
+                    lowlinks[node] = min(lowlinks[node], index[successor])
+
+            # If `node` is a root node, pop the stack and generate an SCC
+            if lowlinks[node] == index[node]:
+                connected_component = []
+
+                while True:
+                    successor = stack.pop()
+                    connected_component.append(successor)
+                    if successor == node:
+                        break
+                component = tuple(connected_component)
+                # storing the result
+                result.append(component)
+
+        for node in graph:
+            if node not in lowlinks:
+                strongconnect(node)
+
+        return result
+
+    @property
+    def dot(self):
+        result = ['digraph G {']
+        for succ in self._preds:
+            preds = self._preds[succ]
+            for pred in preds:
+                result.append('  %s -> %s;' % (pred, succ))
+        for node in self._nodes:
+            result.append('  %s;' % node)
+        result.append('}')
+        return '\n'.join(result)
+
+
+#
+# Unarchiving functionality for zip, tar, tgz, tbz, whl
+#
+
+ARCHIVE_EXTENSIONS = ('.tar.gz', '.tar.bz2', '.tar', '.zip', '.tgz', '.tbz',
+                      '.whl')
+
+
+def unarchive(archive_filename, dest_dir, format=None, check=True):
+
+    def check_path(path):
+        if not isinstance(path, text_type):
+            path = path.decode('utf-8')
+        p = os.path.abspath(os.path.join(dest_dir, path))
+        if not p.startswith(dest_dir) or p[plen] != os.sep:
+            raise ValueError('path outside destination: %r' % p)
+
+    dest_dir = os.path.abspath(dest_dir)
+    plen = len(dest_dir)
+    archive = None
+    if format is None:
+        if archive_filename.endswith(('.zip', '.whl')):
+            format = 'zip'
+        elif archive_filename.endswith(('.tar.gz', '.tgz')):
+            format = 'tgz'
+            mode = 'r:gz'
+        elif archive_filename.endswith(('.tar.bz2', '.tbz')):
+            format = 'tbz'
+            mode = 'r:bz2'
+        elif archive_filename.endswith('.tar'):
+            format = 'tar'
+            mode = 'r'
+        else:  # pragma: no cover
+            raise ValueError('Unknown format for %r' % archive_filename)
+    try:
+        if format == 'zip':
+            archive = ZipFile(archive_filename, 'r')
+            if check:
+                names = archive.namelist()
+                for name in names:
+                    check_path(name)
+        else:
+            archive = tarfile.open(archive_filename, mode)
+            if check:
+                names = archive.getnames()
+                for name in names:
+                    check_path(name)
+        if format != 'zip' and sys.version_info[0] < 3:
+            # See Python issue 17153. If the dest path contains Unicode,
+            # tarfile extraction fails on Python 2.x if a member path name
+            # contains non-ASCII characters - it leads to an implicit
+            # bytes -> unicode conversion using ASCII to decode.
+            for tarinfo in archive.getmembers():
+                if not isinstance(tarinfo.name, text_type):
+                    tarinfo.name = tarinfo.name.decode('utf-8')
+
+        # Limit extraction of dangerous items, if this Python
+        # allows it easily. If not, just trust the input.
+        # See: https://docs.python.org/3/library/tarfile.html#extraction-filters
+        def extraction_filter(member, path):
+            """Run tarfile.tar_filter, but raise the expected ValueError"""
+            # This is only called if the current Python has tarfile filters
+            try:
+                return tarfile.tar_filter(member, path)
+            except tarfile.FilterError as exc:
+                raise ValueError(str(exc))
+
+        archive.extraction_filter = extraction_filter
+
+        archive.extractall(dest_dir)
+
+    finally:
+        if archive:
+            archive.close()
+
+
+def zip_dir(directory):
+    """zip a directory tree into a BytesIO object"""
+    result = io.BytesIO()
+    dlen = len(directory)
+    with ZipFile(result, "w") as zf:
+        for root, dirs, files in os.walk(directory):
+            for name in files:
+                full = os.path.join(root, name)
+                rel = root[dlen:]
+                dest = os.path.join(rel, name)
+                zf.write(full, dest)
+    return result
+
+
+#
+# Simple progress bar
+#
+
+UNITS = ('', 'K', 'M', 'G', 'T', 'P')
+
+
+class Progress(object):
+    unknown = 'UNKNOWN'
+
+    def __init__(self, minval=0, maxval=100):
+        assert maxval is None or maxval >= minval
+        self.min = self.cur = minval
+        self.max = maxval
+        self.started = None
+        self.elapsed = 0
+        self.done = False
+
+    def update(self, curval):
+        assert self.min <= curval
+        assert self.max is None or curval <= self.max
+        self.cur = curval
+        now = time.time()
+        if self.started is None:
+            self.started = now
+        else:
+            self.elapsed = now - self.started
+
+    def increment(self, incr):
+        assert incr >= 0
+        self.update(self.cur + incr)
+
+    def start(self):
+        self.update(self.min)
+        return self
+
+    def stop(self):
+        if self.max is not None:
+            self.update(self.max)
+        self.done = True
+
+    @property
+    def maximum(self):
+        return self.unknown if self.max is None else self.max
+
+    @property
+    def percentage(self):
+        if self.done:
+            result = '100 %'
+        elif self.max is None:
+            result = ' ?? %'
+        else:
+            v = 100.0 * (self.cur - self.min) / (self.max - self.min)
+            result = '%3d %%' % v
+        return result
+
+    def format_duration(self, duration):
+        if (duration <= 0) and self.max is None or self.cur == self.min:
+            result = '??:??:??'
+        # elif duration < 1:
+        #     result = '--:--:--'
+        else:
+            result = time.strftime('%H:%M:%S', time.gmtime(duration))
+        return result
+
+    @property
+    def ETA(self):
+        if self.done:
+            prefix = 'Done'
+            t = self.elapsed
+            # import pdb; pdb.set_trace()
+        else:
+            prefix = 'ETA '
+            if self.max is None:
+                t = -1
+            elif self.elapsed == 0 or (self.cur == self.min):
+                t = 0
+            else:
+                # import pdb; pdb.set_trace()
+                t = float(self.max - self.min)
+                t /= self.cur - self.min
+                t = (t - 1) * self.elapsed
+        return '%s: %s' % (prefix, self.format_duration(t))
+
+    @property
+    def speed(self):
+        if self.elapsed == 0:
+            result = 0.0
+        else:
+            result = (self.cur - self.min) / self.elapsed
+        for unit in UNITS:
+            if result < 1000:
+                break
+            result /= 1000.0
+        return '%d %sB/s' % (result, unit)
+
+
+#
+# Glob functionality
+#
+
+RICH_GLOB = re.compile(r'\{([^}]*)\}')
+_CHECK_RECURSIVE_GLOB = re.compile(r'[^/\\,{]\*\*|\*\*[^/\\,}]')
+_CHECK_MISMATCH_SET = re.compile(r'^[^{]*\}|\{[^}]*$')
+
+
+def iglob(path_glob):
+    """Extended globbing function that supports ** and {opt1,opt2,opt3}."""
+    if _CHECK_RECURSIVE_GLOB.search(path_glob):
+        msg = """invalid glob %r: recursive glob "**" must be used alone"""
+        raise ValueError(msg % path_glob)
+    if _CHECK_MISMATCH_SET.search(path_glob):
+        msg = """invalid glob %r: mismatching set marker '{' or '}'"""
+        raise ValueError(msg % path_glob)
+    return _iglob(path_glob)
+
+
+def _iglob(path_glob):
+    rich_path_glob = RICH_GLOB.split(path_glob, 1)
+    if len(rich_path_glob) > 1:
+        assert len(rich_path_glob) == 3, rich_path_glob
+        prefix, set, suffix = rich_path_glob
+        for item in set.split(','):
+            for path in _iglob(''.join((prefix, item, suffix))):
+                yield path
+    else:
+        if '**' not in path_glob:
+            for item in std_iglob(path_glob):
+                yield item
+        else:
+            prefix, radical = path_glob.split('**', 1)
+            if prefix == '':
+                prefix = '.'
+            if radical == '':
+                radical = '*'
+            else:
+                # we support both
+                radical = radical.lstrip('/')
+                radical = radical.lstrip('\\')
+            for path, dir, files in os.walk(prefix):
+                path = os.path.normpath(path)
+                for fn in _iglob(os.path.join(path, radical)):
+                    yield fn
+
+
+if ssl:
+    from .compat import (HTTPSHandler as BaseHTTPSHandler, match_hostname,
+                         CertificateError)
+
+    #
+    # HTTPSConnection which verifies certificates/matches domains
+    #
+
+    class HTTPSConnection(httplib.HTTPSConnection):
+        ca_certs = None  # set this to the path to the certs file (.pem)
+        check_domain = True  # only used if ca_certs is not None
+
+        # noinspection PyPropertyAccess
+        def connect(self):
+            sock = socket.create_connection((self.host, self.port),
+                                            self.timeout)
+            if getattr(self, '_tunnel_host', False):
+                self.sock = sock
+                self._tunnel()
+
+            context = ssl.SSLContext(ssl.PROTOCOL_SSLv23)
+            if hasattr(ssl, 'OP_NO_SSLv2'):
+                context.options |= ssl.OP_NO_SSLv2
+            if getattr(self, 'cert_file', None):
+                context.load_cert_chain(self.cert_file, self.key_file)
+            kwargs = {}
+            if self.ca_certs:
+                context.verify_mode = ssl.CERT_REQUIRED
+                context.load_verify_locations(cafile=self.ca_certs)
+                if getattr(ssl, 'HAS_SNI', False):
+                    kwargs['server_hostname'] = self.host
+
+            self.sock = context.wrap_socket(sock, **kwargs)
+            if self.ca_certs and self.check_domain:
+                try:
+                    match_hostname(self.sock.getpeercert(), self.host)
+                    logger.debug('Host verified: %s', self.host)
+                except CertificateError:  # pragma: no cover
+                    self.sock.shutdown(socket.SHUT_RDWR)
+                    self.sock.close()
+                    raise
+
+    class HTTPSHandler(BaseHTTPSHandler):
+
+        def __init__(self, ca_certs, check_domain=True):
+            BaseHTTPSHandler.__init__(self)
+            self.ca_certs = ca_certs
+            self.check_domain = check_domain
+
+        def _conn_maker(self, *args, **kwargs):
+            """
+            This is called to create a connection instance. Normally you'd
+            pass a connection class to do_open, but it doesn't actually check for
+            a class, and just expects a callable. As long as we behave just as a
+            constructor would have, we should be OK. If it ever changes so that
+            we *must* pass a class, we'll create an UnsafeHTTPSConnection class
+            which just sets check_domain to False in the class definition, and
+            choose which one to pass to do_open.
+            """
+            result = HTTPSConnection(*args, **kwargs)
+            if self.ca_certs:
+                result.ca_certs = self.ca_certs
+                result.check_domain = self.check_domain
+            return result
+
+        def https_open(self, req):
+            try:
+                return self.do_open(self._conn_maker, req)
+            except URLError as e:
+                if 'certificate verify failed' in str(e.reason):
+                    raise CertificateError(
+                        'Unable to verify server certificate '
+                        'for %s' % req.host)
+                else:
+                    raise
+
+    #
+    # To prevent against mixing HTTP traffic with HTTPS (examples: A Man-In-The-
+    # Middle proxy using HTTP listens on port 443, or an index mistakenly serves
+    # HTML containing a http://xyz link when it should be https://xyz),
+    # you can use the following handler class, which does not allow HTTP traffic.
+    #
+    # It works by inheriting from HTTPHandler - so build_opener won't add a
+    # handler for HTTP itself.
+    #
+    class HTTPSOnlyHandler(HTTPSHandler, HTTPHandler):
+
+        def http_open(self, req):
+            raise URLError(
+                'Unexpected HTTP request on what should be a secure '
+                'connection: %s' % req)
+
+
+#
+# XML-RPC with timeouts
+#
+class Transport(xmlrpclib.Transport):
+
+    def __init__(self, timeout, use_datetime=0):
+        self.timeout = timeout
+        xmlrpclib.Transport.__init__(self, use_datetime)
+
+    def make_connection(self, host):
+        h, eh, x509 = self.get_host_info(host)
+        if not self._connection or host != self._connection[0]:
+            self._extra_headers = eh
+            self._connection = host, httplib.HTTPConnection(h)
+        return self._connection[1]
+
+
+if ssl:
+
+    class SafeTransport(xmlrpclib.SafeTransport):
+
+        def __init__(self, timeout, use_datetime=0):
+            self.timeout = timeout
+            xmlrpclib.SafeTransport.__init__(self, use_datetime)
+
+        def make_connection(self, host):
+            h, eh, kwargs = self.get_host_info(host)
+            if not kwargs:
+                kwargs = {}
+            kwargs['timeout'] = self.timeout
+            if not self._connection or host != self._connection[0]:
+                self._extra_headers = eh
+                self._connection = host, httplib.HTTPSConnection(
+                    h, None, **kwargs)
+            return self._connection[1]
+
+
+class ServerProxy(xmlrpclib.ServerProxy):
+
+    def __init__(self, uri, **kwargs):
+        self.timeout = timeout = kwargs.pop('timeout', None)
+        # The above classes only come into play if a timeout
+        # is specified
+        if timeout is not None:
+            # scheme = splittype(uri)  # deprecated as of Python 3.8
+            scheme = urlparse(uri)[0]
+            use_datetime = kwargs.get('use_datetime', 0)
+            if scheme == 'https':
+                tcls = SafeTransport
+            else:
+                tcls = Transport
+            kwargs['transport'] = t = tcls(timeout, use_datetime=use_datetime)
+            self.transport = t
+        xmlrpclib.ServerProxy.__init__(self, uri, **kwargs)
+
+
+#
+# CSV functionality. This is provided because on 2.x, the csv module can't
+# handle Unicode. However, we need to deal with Unicode in e.g. RECORD files.
+#
+
+
+def _csv_open(fn, mode, **kwargs):
+    if sys.version_info[0] < 3:
+        mode += 'b'
+    else:
+        kwargs['newline'] = ''
+        # Python 3 determines encoding from locale. Force 'utf-8'
+        # file encoding to match other forced utf-8 encoding
+        kwargs['encoding'] = 'utf-8'
+    return open(fn, mode, **kwargs)
+
+
+class CSVBase(object):
+    defaults = {
+        'delimiter': str(','),  # The strs are used because we need native
+        'quotechar': str('"'),  # str in the csv API (2.x won't take
+        'lineterminator': str('\n')  # Unicode)
+    }
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, *exc_info):
+        self.stream.close()
+
+
+class CSVReader(CSVBase):
+
+    def __init__(self, **kwargs):
+        if 'stream' in kwargs:
+            stream = kwargs['stream']
+            if sys.version_info[0] >= 3:
+                # needs to be a text stream
+                stream = codecs.getreader('utf-8')(stream)
+            self.stream = stream
+        else:
+            self.stream = _csv_open(kwargs['path'], 'r')
+        self.reader = csv.reader(self.stream, **self.defaults)
+
+    def __iter__(self):
+        return self
+
+    def next(self):
+        result = next(self.reader)
+        if sys.version_info[0] < 3:
+            for i, item in enumerate(result):
+                if not isinstance(item, text_type):
+                    result[i] = item.decode('utf-8')
+        return result
+
+    __next__ = next
+
+
+class CSVWriter(CSVBase):
+
+    def __init__(self, fn, **kwargs):
+        self.stream = _csv_open(fn, 'w')
+        self.writer = csv.writer(self.stream, **self.defaults)
+
+    def writerow(self, row):
+        if sys.version_info[0] < 3:
+            r = []
+            for item in row:
+                if isinstance(item, text_type):
+                    item = item.encode('utf-8')
+                r.append(item)
+            row = r
+        self.writer.writerow(row)
+
+
+#
+#   Configurator functionality
+#
+
+
+class Configurator(BaseConfigurator):
+
+    value_converters = dict(BaseConfigurator.value_converters)
+    value_converters['inc'] = 'inc_convert'
+
+    def __init__(self, config, base=None):
+        super(Configurator, self).__init__(config)
+        self.base = base or os.getcwd()
+
+    def configure_custom(self, config):
+
+        def convert(o):
+            if isinstance(o, (list, tuple)):
+                result = type(o)([convert(i) for i in o])
+            elif isinstance(o, dict):
+                if '()' in o:
+                    result = self.configure_custom(o)
+                else:
+                    result = {}
+                    for k in o:
+                        result[k] = convert(o[k])
+            else:
+                result = self.convert(o)
+            return result
+
+        c = config.pop('()')
+        if not callable(c):
+            c = self.resolve(c)
+        props = config.pop('.', None)
+        # Check for valid identifiers
+        args = config.pop('[]', ())
+        if args:
+            args = tuple([convert(o) for o in args])
+        items = [(k, convert(config[k])) for k in config if valid_ident(k)]
+        kwargs = dict(items)
+        result = c(*args, **kwargs)
+        if props:
+            for n, v in props.items():
+                setattr(result, n, convert(v))
+        return result
+
+    def __getitem__(self, key):
+        result = self.config[key]
+        if isinstance(result, dict) and '()' in result:
+            self.config[key] = result = self.configure_custom(result)
+        return result
+
+    def inc_convert(self, value):
+        """Default converter for the inc:// protocol."""
+        if not os.path.isabs(value):
+            value = os.path.join(self.base, value)
+        with codecs.open(value, 'r', encoding='utf-8') as f:
+            result = json.load(f)
+        return result
+
+
+class SubprocessMixin(object):
+    """
+    Mixin for running subprocesses and capturing their output
+    """
+
+    def __init__(self, verbose=False, progress=None):
+        self.verbose = verbose
+        self.progress = progress
+
+    def reader(self, stream, context):
+        """
+        Read lines from a subprocess' output stream and either pass to a progress
+        callable (if specified) or write progress information to sys.stderr.
+        """
+        progress = self.progress
+        verbose = self.verbose
+        while True:
+            s = stream.readline()
+            if not s:
+                break
+            if progress is not None:
+                progress(s, context)
+            else:
+                if not verbose:
+                    sys.stderr.write('.')
+                else:
+                    sys.stderr.write(s.decode('utf-8'))
+                sys.stderr.flush()
+        stream.close()
+
+    def run_command(self, cmd, **kwargs):
+        p = subprocess.Popen(cmd,
+                             stdout=subprocess.PIPE,
+                             stderr=subprocess.PIPE,
+                             **kwargs)
+        t1 = threading.Thread(target=self.reader, args=(p.stdout, 'stdout'))
+        t1.start()
+        t2 = threading.Thread(target=self.reader, args=(p.stderr, 'stderr'))
+        t2.start()
+        p.wait()
+        t1.join()
+        t2.join()
+        if self.progress is not None:
+            self.progress('done.', 'main')
+        elif self.verbose:
+            sys.stderr.write('done.\n')
+        return p
+
+
+def normalize_name(name):
+    """Normalize a python package name a la PEP 503"""
+    # https://www.python.org/dev/peps/pep-0503/#normalized-names
+    return re.sub('[-_.]+', '-', name).lower()
+
+
+# def _get_pypirc_command():
+# """
+# Get the distutils command for interacting with PyPI configurations.
+# :return: the command.
+# """
+# from distutils.core import Distribution
+# from distutils.config import PyPIRCCommand
+# d = Distribution()
+# return PyPIRCCommand(d)
+
+
+class PyPIRCFile(object):
+
+    DEFAULT_REPOSITORY = 'https://upload.pypi.org/legacy/'
+    DEFAULT_REALM = 'pypi'
+
+    def __init__(self, fn=None, url=None):
+        if fn is None:
+            fn = os.path.join(os.path.expanduser('~'), '.pypirc')
+        self.filename = fn
+        self.url = url
+
+    def read(self):
+        result = {}
+
+        if os.path.exists(self.filename):
+            repository = self.url or self.DEFAULT_REPOSITORY
+
+            config = configparser.RawConfigParser()
+            config.read(self.filename)
+            sections = config.sections()
+            if 'distutils' in sections:
+                # let's get the list of servers
+                index_servers = config.get('distutils', 'index-servers')
+                _servers = [
+                    server.strip() for server in index_servers.split('\n')
+                    if server.strip() != ''
+                ]
+                if _servers == []:
+                    # nothing set, let's try to get the default pypi
+                    if 'pypi' in sections:
+                        _servers = ['pypi']
+                else:
+                    for server in _servers:
+                        result = {'server': server}
+                        result['username'] = config.get(server, 'username')
+
+                        # optional params
+                        for key, default in (('repository',
+                                              self.DEFAULT_REPOSITORY),
+                                             ('realm', self.DEFAULT_REALM),
+                                             ('password', None)):
+                            if config.has_option(server, key):
+                                result[key] = config.get(server, key)
+                            else:
+                                result[key] = default
+
+                        # work around people having "repository" for the "pypi"
+                        # section of their config set to the HTTP (rather than
+                        # HTTPS) URL
+                        if (server == 'pypi' and repository
+                                in (self.DEFAULT_REPOSITORY, 'pypi')):
+                            result['repository'] = self.DEFAULT_REPOSITORY
+                        elif (result['server'] != repository
+                              and result['repository'] != repository):
+                            result = {}
+            elif 'server-login' in sections:
+                # old format
+                server = 'server-login'
+                if config.has_option(server, 'repository'):
+                    repository = config.get(server, 'repository')
+                else:
+                    repository = self.DEFAULT_REPOSITORY
+                result = {
+                    'username': config.get(server, 'username'),
+                    'password': config.get(server, 'password'),
+                    'repository': repository,
+                    'server': server,
+                    'realm': self.DEFAULT_REALM
+                }
+        return result
+
+    def update(self, username, password):
+        # import pdb; pdb.set_trace()
+        config = configparser.RawConfigParser()
+        fn = self.filename
+        config.read(fn)
+        if not config.has_section('pypi'):
+            config.add_section('pypi')
+        config.set('pypi', 'username', username)
+        config.set('pypi', 'password', password)
+        with open(fn, 'w') as f:
+            config.write(f)
+
+
+def _load_pypirc(index):
+    """
+    Read the PyPI access configuration as supported by distutils.
+    """
+    return PyPIRCFile(url=index.url).read()
+
+
+def _store_pypirc(index):
+    PyPIRCFile().update(index.username, index.password)
+
+
+#
+# get_platform()/get_host_platform() copied from Python 3.10.a0 source, with some minor
+# tweaks
+#
+
+
+def get_host_platform():
+    """Return a string that identifies the current platform.  This is used mainly to
+    distinguish platform-specific build directories and platform-specific built
+    distributions.  Typically includes the OS name and version and the
+    architecture (as supplied by 'os.uname()'), although the exact information
+    included depends on the OS; eg. on Linux, the kernel version isn't
+    particularly important.
+
+    Examples of returned values:
+       linux-i586
+       linux-alpha (?)
+       solaris-2.6-sun4u
+
+    Windows will return one of:
+       win-amd64 (64bit Windows on AMD64 (aka x86_64, Intel64, EM64T, etc)
+       win32 (all others - specifically, sys.platform is returned)
+
+    For other non-POSIX platforms, currently just returns 'sys.platform'.
+
+    """
+    if os.name == 'nt':
+        if 'amd64' in sys.version.lower():
+            return 'win-amd64'
+        if '(arm)' in sys.version.lower():
+            return 'win-arm32'
+        if '(arm64)' in sys.version.lower():
+            return 'win-arm64'
+        return sys.platform
+
+    # Set for cross builds explicitly
+    if "_PYTHON_HOST_PLATFORM" in os.environ:
+        return os.environ["_PYTHON_HOST_PLATFORM"]
+
+    if os.name != 'posix' or not hasattr(os, 'uname'):
+        # XXX what about the architecture? NT is Intel or Alpha,
+        # Mac OS is M68k or PPC, etc.
+        return sys.platform
+
+    # Try to distinguish various flavours of Unix
+
+    (osname, host, release, version, machine) = os.uname()
+
+    # Convert the OS name to lowercase, remove '/' characters, and translate
+    # spaces (for "Power Macintosh")
+    osname = osname.lower().replace('/', '')
+    machine = machine.replace(' ', '_').replace('/', '-')
+
+    if osname[:5] == 'linux':
+        # At least on Linux/Intel, 'machine' is the processor --
+        # i386, etc.
+        # XXX what about Alpha, SPARC, etc?
+        return "%s-%s" % (osname, machine)
+
+    elif osname[:5] == 'sunos':
+        if release[0] >= '5':  # SunOS 5 == Solaris 2
+            osname = 'solaris'
+            release = '%d.%s' % (int(release[0]) - 3, release[2:])
+            # We can't use 'platform.architecture()[0]' because a
+            # bootstrap problem. We use a dict to get an error
+            # if some suspicious happens.
+            bitness = {2147483647: '32bit', 9223372036854775807: '64bit'}
+            machine += '.%s' % bitness[sys.maxsize]
+        # fall through to standard osname-release-machine representation
+    elif osname[:3] == 'aix':
+        from _aix_support import aix_platform
+        return aix_platform()
+    elif osname[:6] == 'cygwin':
+        osname = 'cygwin'
+        rel_re = re.compile(r'[\d.]+', re.ASCII)
+        m = rel_re.match(release)
+        if m:
+            release = m.group()
+    elif osname[:6] == 'darwin':
+        import _osx_support
+        try:
+            from distutils import sysconfig
+        except ImportError:
+            import sysconfig
+        osname, release, machine = _osx_support.get_platform_osx(
+            sysconfig.get_config_vars(), osname, release, machine)
+
+    return '%s-%s-%s' % (osname, release, machine)
+
+
+_TARGET_TO_PLAT = {
+    'x86': 'win32',
+    'x64': 'win-amd64',
+    'arm': 'win-arm32',
+}
+
+
+def get_platform():
+    if os.name != 'nt':
+        return get_host_platform()
+    cross_compilation_target = os.environ.get('VSCMD_ARG_TGT_ARCH')
+    if cross_compilation_target not in _TARGET_TO_PLAT:
+        return get_host_platform()
+    return _TARGET_TO_PLAT[cross_compilation_target]

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/version.py

@@ -0,0 +1,751 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2012-2023 The Python Software Foundation.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+"""
+Implementation of a flexible versioning scheme providing support for PEP-440,
+setuptools-compatible and semantic versioning.
+"""
+
+import logging
+import re
+
+from .compat import string_types
+from .util import parse_requirement
+
+__all__ = ['NormalizedVersion', 'NormalizedMatcher',
+           'LegacyVersion', 'LegacyMatcher',
+           'SemanticVersion', 'SemanticMatcher',
+           'UnsupportedVersionError', 'get_scheme']
+
+logger = logging.getLogger(__name__)
+
+
+class UnsupportedVersionError(ValueError):
+    """This is an unsupported version."""
+    pass
+
+
+class Version(object):
+    def __init__(self, s):
+        self._string = s = s.strip()
+        self._parts = parts = self.parse(s)
+        assert isinstance(parts, tuple)
+        assert len(parts) > 0
+
+    def parse(self, s):
+        raise NotImplementedError('please implement in a subclass')
+
+    def _check_compatible(self, other):
+        if type(self) != type(other):
+            raise TypeError('cannot compare %r and %r' % (self, other))
+
+    def __eq__(self, other):
+        self._check_compatible(other)
+        return self._parts == other._parts
+
+    def __ne__(self, other):
+        return not self.__eq__(other)
+
+    def __lt__(self, other):
+        self._check_compatible(other)
+        return self._parts < other._parts
+
+    def __gt__(self, other):
+        return not (self.__lt__(other) or self.__eq__(other))
+
+    def __le__(self, other):
+        return self.__lt__(other) or self.__eq__(other)
+
+    def __ge__(self, other):
+        return self.__gt__(other) or self.__eq__(other)
+
+    # See http://docs.python.org/reference/datamodel#object.__hash__
+    def __hash__(self):
+        return hash(self._parts)
+
+    def __repr__(self):
+        return "%s('%s')" % (self.__class__.__name__, self._string)
+
+    def __str__(self):
+        return self._string
+
+    @property
+    def is_prerelease(self):
+        raise NotImplementedError('Please implement in subclasses.')
+
+
+class Matcher(object):
+    version_class = None
+
+    # value is either a callable or the name of a method
+    _operators = {
+        '<': lambda v, c, p: v < c,
+        '>': lambda v, c, p: v > c,
+        '<=': lambda v, c, p: v == c or v < c,
+        '>=': lambda v, c, p: v == c or v > c,
+        '==': lambda v, c, p: v == c,
+        '===': lambda v, c, p: v == c,
+        # by default, compatible => >=.
+        '~=': lambda v, c, p: v == c or v > c,
+        '!=': lambda v, c, p: v != c,
+    }
+
+    # this is a method only to support alternative implementations
+    # via overriding
+    def parse_requirement(self, s):
+        return parse_requirement(s)
+
+    def __init__(self, s):
+        if self.version_class is None:
+            raise ValueError('Please specify a version class')
+        self._string = s = s.strip()
+        r = self.parse_requirement(s)
+        if not r:
+            raise ValueError('Not valid: %r' % s)
+        self.name = r.name
+        self.key = self.name.lower()    # for case-insensitive comparisons
+        clist = []
+        if r.constraints:
+            # import pdb; pdb.set_trace()
+            for op, s in r.constraints:
+                if s.endswith('.*'):
+                    if op not in ('==', '!='):
+                        raise ValueError('\'.*\' not allowed for '
+                                         '%r constraints' % op)
+                    # Could be a partial version (e.g. for '2.*') which
+                    # won't parse as a version, so keep it as a string
+                    vn, prefix = s[:-2], True
+                    # Just to check that vn is a valid version
+                    self.version_class(vn)
+                else:
+                    # Should parse as a version, so we can create an
+                    # instance for the comparison
+                    vn, prefix = self.version_class(s), False
+                clist.append((op, vn, prefix))
+        self._parts = tuple(clist)
+
+    def match(self, version):
+        """
+        Check if the provided version matches the constraints.
+
+        :param version: The version to match against this instance.
+        :type version: String or :class:`Version` instance.
+        """
+        if isinstance(version, string_types):
+            version = self.version_class(version)
+        for operator, constraint, prefix in self._parts:
+            f = self._operators.get(operator)
+            if isinstance(f, string_types):
+                f = getattr(self, f)
+            if not f:
+                msg = ('%r not implemented '
+                       'for %s' % (operator, self.__class__.__name__))
+                raise NotImplementedError(msg)
+            if not f(version, constraint, prefix):
+                return False
+        return True
+
+    @property
+    def exact_version(self):
+        result = None
+        if len(self._parts) == 1 and self._parts[0][0] in ('==', '==='):
+            result = self._parts[0][1]
+        return result
+
+    def _check_compatible(self, other):
+        if type(self) != type(other) or self.name != other.name:
+            raise TypeError('cannot compare %s and %s' % (self, other))
+
+    def __eq__(self, other):
+        self._check_compatible(other)
+        return self.key == other.key and self._parts == other._parts
+
+    def __ne__(self, other):
+        return not self.__eq__(other)
+
+    # See http://docs.python.org/reference/datamodel#object.__hash__
+    def __hash__(self):
+        return hash(self.key) + hash(self._parts)
+
+    def __repr__(self):
+        return "%s(%r)" % (self.__class__.__name__, self._string)
+
+    def __str__(self):
+        return self._string
+
+
+PEP440_VERSION_RE = re.compile(r'^v?(\d+!)?(\d+(\.\d+)*)((a|alpha|b|beta|c|rc|pre|preview)(\d+)?)?'
+                               r'(\.(post|r|rev)(\d+)?)?([._-]?(dev)(\d+)?)?'
+                               r'(\+([a-zA-Z\d]+(\.[a-zA-Z\d]+)?))?$', re.I)
+
+
+def _pep_440_key(s):
+    s = s.strip()
+    m = PEP440_VERSION_RE.match(s)
+    if not m:
+        raise UnsupportedVersionError('Not a valid version: %s' % s)
+    groups = m.groups()
+    nums = tuple(int(v) for v in groups[1].split('.'))
+    while len(nums) > 1 and nums[-1] == 0:
+        nums = nums[:-1]
+
+    if not groups[0]:
+        epoch = 0
+    else:
+        epoch = int(groups[0][:-1])
+    pre = groups[4:6]
+    post = groups[7:9]
+    dev = groups[10:12]
+    local = groups[13]
+    if pre == (None, None):
+        pre = ()
+    else:
+        if pre[1] is None:
+            pre = pre[0], 0
+        else:
+            pre = pre[0], int(pre[1])
+    if post == (None, None):
+        post = ()
+    else:
+        if post[1] is None:
+            post = post[0], 0
+        else:
+            post = post[0], int(post[1])
+    if dev == (None, None):
+        dev = ()
+    else:
+        if dev[1] is None:
+            dev = dev[0], 0
+        else:
+            dev = dev[0], int(dev[1])
+    if local is None:
+        local = ()
+    else:
+        parts = []
+        for part in local.split('.'):
+            # to ensure that numeric compares as > lexicographic, avoid
+            # comparing them directly, but encode a tuple which ensures
+            # correct sorting
+            if part.isdigit():
+                part = (1, int(part))
+            else:
+                part = (0, part)
+            parts.append(part)
+        local = tuple(parts)
+    if not pre:
+        # either before pre-release, or final release and after
+        if not post and dev:
+            # before pre-release
+            pre = ('a', -1)     # to sort before a0
+        else:
+            pre = ('z',)        # to sort after all pre-releases
+    # now look at the state of post and dev.
+    if not post:
+        post = ('_',)   # sort before 'a'
+    if not dev:
+        dev = ('final',)
+
+    return epoch, nums, pre, post, dev, local
+
+
+_normalized_key = _pep_440_key
+
+
+class NormalizedVersion(Version):
+    """A rational version.
+
+    Good:
+        1.2         # equivalent to "1.2.0"
+        1.2.0
+        1.2a1
+        1.2.3a2
+        1.2.3b1
+        1.2.3c1
+        1.2.3.4
+        TODO: fill this out
+
+    Bad:
+        1           # minimum two numbers
+        1.2a        # release level must have a release serial
+        1.2.3b
+    """
+    def parse(self, s):
+        result = _normalized_key(s)
+        # _normalized_key loses trailing zeroes in the release
+        # clause, since that's needed to ensure that X.Y == X.Y.0 == X.Y.0.0
+        # However, PEP 440 prefix matching needs it: for example,
+        # (~= 1.4.5.0) matches differently to (~= 1.4.5.0.0).
+        m = PEP440_VERSION_RE.match(s)      # must succeed
+        groups = m.groups()
+        self._release_clause = tuple(int(v) for v in groups[1].split('.'))
+        return result
+
+    PREREL_TAGS = set(['a', 'b', 'c', 'rc', 'dev'])
+
+    @property
+    def is_prerelease(self):
+        return any(t[0] in self.PREREL_TAGS for t in self._parts if t)
+
+
+def _match_prefix(x, y):
+    x = str(x)
+    y = str(y)
+    if x == y:
+        return True
+    if not x.startswith(y):
+        return False
+    n = len(y)
+    return x[n] == '.'
+
+
+class NormalizedMatcher(Matcher):
+    version_class = NormalizedVersion
+
+    # value is either a callable or the name of a method
+    _operators = {
+        '~=': '_match_compatible',
+        '<': '_match_lt',
+        '>': '_match_gt',
+        '<=': '_match_le',
+        '>=': '_match_ge',
+        '==': '_match_eq',
+        '===': '_match_arbitrary',
+        '!=': '_match_ne',
+    }
+
+    def _adjust_local(self, version, constraint, prefix):
+        if prefix:
+            strip_local = '+' not in constraint and version._parts[-1]
+        else:
+            # both constraint and version are
+            # NormalizedVersion instances.
+            # If constraint does not have a local component,
+            # ensure the version doesn't, either.
+            strip_local = not constraint._parts[-1] and version._parts[-1]
+        if strip_local:
+            s = version._string.split('+', 1)[0]
+            version = self.version_class(s)
+        return version, constraint
+
+    def _match_lt(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        if version >= constraint:
+            return False
+        release_clause = constraint._release_clause
+        pfx = '.'.join([str(i) for i in release_clause])
+        return not _match_prefix(version, pfx)
+
+    def _match_gt(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        if version <= constraint:
+            return False
+        release_clause = constraint._release_clause
+        pfx = '.'.join([str(i) for i in release_clause])
+        return not _match_prefix(version, pfx)
+
+    def _match_le(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        return version <= constraint
+
+    def _match_ge(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        return version >= constraint
+
+    def _match_eq(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        if not prefix:
+            result = (version == constraint)
+        else:
+            result = _match_prefix(version, constraint)
+        return result
+
+    def _match_arbitrary(self, version, constraint, prefix):
+        return str(version) == str(constraint)
+
+    def _match_ne(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        if not prefix:
+            result = (version != constraint)
+        else:
+            result = not _match_prefix(version, constraint)
+        return result
+
+    def _match_compatible(self, version, constraint, prefix):
+        version, constraint = self._adjust_local(version, constraint, prefix)
+        if version == constraint:
+            return True
+        if version < constraint:
+            return False
+#        if not prefix:
+#            return True
+        release_clause = constraint._release_clause
+        if len(release_clause) > 1:
+            release_clause = release_clause[:-1]
+        pfx = '.'.join([str(i) for i in release_clause])
+        return _match_prefix(version, pfx)
+
+
+_REPLACEMENTS = (
+    (re.compile('[.+-]$'), ''),                     # remove trailing puncts
+    (re.compile(r'^[.](\d)'), r'0.\1'),             # .N -> 0.N at start
+    (re.compile('^[.-]'), ''),                      # remove leading puncts
+    (re.compile(r'^\((.*)\)$'), r'\1'),             # remove parentheses
+    (re.compile(r'^v(ersion)?\s*(\d+)'), r'\2'),    # remove leading v(ersion)
+    (re.compile(r'^r(ev)?\s*(\d+)'), r'\2'),        # remove leading v(ersion)
+    (re.compile('[.]{2,}'), '.'),                   # multiple runs of '.'
+    (re.compile(r'\b(alfa|apha)\b'), 'alpha'),      # misspelt alpha
+    (re.compile(r'\b(pre-alpha|prealpha)\b'),
+        'pre.alpha'),                               # standardise
+    (re.compile(r'\(beta\)$'), 'beta'),             # remove parentheses
+)
+
+_SUFFIX_REPLACEMENTS = (
+    (re.compile('^[:~._+-]+'), ''),                   # remove leading puncts
+    (re.compile('[,*")([\\]]'), ''),                  # remove unwanted chars
+    (re.compile('[~:+_ -]'), '.'),                    # replace illegal chars
+    (re.compile('[.]{2,}'), '.'),                   # multiple runs of '.'
+    (re.compile(r'\.$'), ''),                       # trailing '.'
+)
+
+_NUMERIC_PREFIX = re.compile(r'(\d+(\.\d+)*)')
+
+
+def _suggest_semantic_version(s):
+    """
+    Try to suggest a semantic form for a version for which
+    _suggest_normalized_version couldn't come up with anything.
+    """
+    result = s.strip().lower()
+    for pat, repl in _REPLACEMENTS:
+        result = pat.sub(repl, result)
+    if not result:
+        result = '0.0.0'
+
+    # Now look for numeric prefix, and separate it out from
+    # the rest.
+    # import pdb; pdb.set_trace()
+    m = _NUMERIC_PREFIX.match(result)
+    if not m:
+        prefix = '0.0.0'
+        suffix = result
+    else:
+        prefix = m.groups()[0].split('.')
+        prefix = [int(i) for i in prefix]
+        while len(prefix) < 3:
+            prefix.append(0)
+        if len(prefix) == 3:
+            suffix = result[m.end():]
+        else:
+            suffix = '.'.join([str(i) for i in prefix[3:]]) + result[m.end():]
+            prefix = prefix[:3]
+        prefix = '.'.join([str(i) for i in prefix])
+        suffix = suffix.strip()
+    if suffix:
+        # import pdb; pdb.set_trace()
+        # massage the suffix.
+        for pat, repl in _SUFFIX_REPLACEMENTS:
+            suffix = pat.sub(repl, suffix)
+
+    if not suffix:
+        result = prefix
+    else:
+        sep = '-' if 'dev' in suffix else '+'
+        result = prefix + sep + suffix
+    if not is_semver(result):
+        result = None
+    return result
+
+
+def _suggest_normalized_version(s):
+    """Suggest a normalized version close to the given version string.
+
+    If you have a version string that isn't rational (i.e. NormalizedVersion
+    doesn't like it) then you might be able to get an equivalent (or close)
+    rational version from this function.
+
+    This does a number of simple normalizations to the given string, based
+    on observation of versions currently in use on PyPI. Given a dump of
+    those version during PyCon 2009, 4287 of them:
+    - 2312 (53.93%) match NormalizedVersion without change
+      with the automatic suggestion
+    - 3474 (81.04%) match when using this suggestion method
+
+    @param s {str} An irrational version string.
+    @returns A rational version string, or None, if couldn't determine one.
+    """
+    try:
+        _normalized_key(s)
+        return s   # already rational
+    except UnsupportedVersionError:
+        pass
+
+    rs = s.lower()
+
+    # part of this could use maketrans
+    for orig, repl in (('-alpha', 'a'), ('-beta', 'b'), ('alpha', 'a'),
+                       ('beta', 'b'), ('rc', 'c'), ('-final', ''),
+                       ('-pre', 'c'),
+                       ('-release', ''), ('.release', ''), ('-stable', ''),
+                       ('+', '.'), ('_', '.'), (' ', ''), ('.final', ''),
+                       ('final', '')):
+        rs = rs.replace(orig, repl)
+
+    # if something ends with dev or pre, we add a 0
+    rs = re.sub(r"pre$", r"pre0", rs)
+    rs = re.sub(r"dev$", r"dev0", rs)
+
+    # if we have something like "b-2" or "a.2" at the end of the
+    # version, that is probably beta, alpha, etc
+    # let's remove the dash or dot
+    rs = re.sub(r"([abc]|rc)[\-\.](\d+)$", r"\1\2", rs)
+
+    # 1.0-dev-r371 -> 1.0.dev371
+    # 0.1-dev-r79 -> 0.1.dev79
+    rs = re.sub(r"[\-\.](dev)[\-\.]?r?(\d+)$", r".\1\2", rs)
+
+    # Clean: 2.0.a.3, 2.0.b1, 0.9.0~c1
+    rs = re.sub(r"[.~]?([abc])\.?", r"\1", rs)
+
+    # Clean: v0.3, v1.0
+    if rs.startswith('v'):
+        rs = rs[1:]
+
+    # Clean leading '0's on numbers.
+    # TODO: unintended side-effect on, e.g., "2003.05.09"
+    # PyPI stats: 77 (~2%) better
+    rs = re.sub(r"\b0+(\d+)(?!\d)", r"\1", rs)
+
+    # Clean a/b/c with no version. E.g. "1.0a" -> "1.0a0". Setuptools infers
+    # zero.
+    # PyPI stats: 245 (7.56%) better
+    rs = re.sub(r"(\d+[abc])$", r"\g<1>0", rs)
+
+    # the 'dev-rNNN' tag is a dev tag
+    rs = re.sub(r"\.?(dev-r|dev\.r)\.?(\d+)$", r".dev\2", rs)
+
+    # clean the - when used as a pre delimiter
+    rs = re.sub(r"-(a|b|c)(\d+)$", r"\1\2", rs)
+
+    # a terminal "dev" or "devel" can be changed into ".dev0"
+    rs = re.sub(r"[\.\-](dev|devel)$", r".dev0", rs)
+
+    # a terminal "dev" can be changed into ".dev0"
+    rs = re.sub(r"(?![\.\-])dev$", r".dev0", rs)
+
+    # a terminal "final" or "stable" can be removed
+    rs = re.sub(r"(final|stable)$", "", rs)
+
+    # The 'r' and the '-' tags are post release tags
+    #   0.4a1.r10       ->  0.4a1.post10
+    #   0.9.33-17222    ->  0.9.33.post17222
+    #   0.9.33-r17222   ->  0.9.33.post17222
+    rs = re.sub(r"\.?(r|-|-r)\.?(\d+)$", r".post\2", rs)
+
+    # Clean 'r' instead of 'dev' usage:
+    #   0.9.33+r17222   ->  0.9.33.dev17222
+    #   1.0dev123       ->  1.0.dev123
+    #   1.0.git123      ->  1.0.dev123
+    #   1.0.bzr123      ->  1.0.dev123
+    #   0.1a0dev.123    ->  0.1a0.dev123
+    # PyPI stats:  ~150 (~4%) better
+    rs = re.sub(r"\.?(dev|git|bzr)\.?(\d+)$", r".dev\2", rs)
+
+    # Clean '.pre' (normalized from '-pre' above) instead of 'c' usage:
+    #   0.2.pre1        ->  0.2c1
+    #   0.2-c1         ->  0.2c1
+    #   1.0preview123   ->  1.0c123
+    # PyPI stats: ~21 (0.62%) better
+    rs = re.sub(r"\.?(pre|preview|-c)(\d+)$", r"c\g<2>", rs)
+
+    # Tcl/Tk uses "px" for their post release markers
+    rs = re.sub(r"p(\d+)$", r".post\1", rs)
+
+    try:
+        _normalized_key(rs)
+    except UnsupportedVersionError:
+        rs = None
+    return rs
+
+#
+#   Legacy version processing (distribute-compatible)
+#
+
+
+_VERSION_PART = re.compile(r'([a-z]+|\d+|[\.-])', re.I)
+_VERSION_REPLACE = {
+    'pre': 'c',
+    'preview': 'c',
+    '-': 'final-',
+    'rc': 'c',
+    'dev': '@',
+    '': None,
+    '.': None,
+}
+
+
+def _legacy_key(s):
+    def get_parts(s):
+        result = []
+        for p in _VERSION_PART.split(s.lower()):
+            p = _VERSION_REPLACE.get(p, p)
+            if p:
+                if '0' <= p[:1] <= '9':
+                    p = p.zfill(8)
+                else:
+                    p = '*' + p
+                result.append(p)
+        result.append('*final')
+        return result
+
+    result = []
+    for p in get_parts(s):
+        if p.startswith('*'):
+            if p < '*final':
+                while result and result[-1] == '*final-':
+                    result.pop()
+            while result and result[-1] == '00000000':
+                result.pop()
+        result.append(p)
+    return tuple(result)
+
+
+class LegacyVersion(Version):
+    def parse(self, s):
+        return _legacy_key(s)
+
+    @property
+    def is_prerelease(self):
+        result = False
+        for x in self._parts:
+            if (isinstance(x, string_types) and x.startswith('*') and
+                    x < '*final'):
+                result = True
+                break
+        return result
+
+
+class LegacyMatcher(Matcher):
+    version_class = LegacyVersion
+
+    _operators = dict(Matcher._operators)
+    _operators['~='] = '_match_compatible'
+
+    numeric_re = re.compile(r'^(\d+(\.\d+)*)')
+
+    def _match_compatible(self, version, constraint, prefix):
+        if version < constraint:
+            return False
+        m = self.numeric_re.match(str(constraint))
+        if not m:
+            logger.warning('Cannot compute compatible match for version %s '
+                           ' and constraint %s', version, constraint)
+            return True
+        s = m.groups()[0]
+        if '.' in s:
+            s = s.rsplit('.', 1)[0]
+        return _match_prefix(version, s)
+
+#
+#   Semantic versioning
+#
+
+
+_SEMVER_RE = re.compile(r'^(\d+)\.(\d+)\.(\d+)'
+                        r'(-[a-z0-9]+(\.[a-z0-9-]+)*)?'
+                        r'(\+[a-z0-9]+(\.[a-z0-9-]+)*)?$', re.I)
+
+
+def is_semver(s):
+    return _SEMVER_RE.match(s)
+
+
+def _semantic_key(s):
+    def make_tuple(s, absent):
+        if s is None:
+            result = (absent,)
+        else:
+            parts = s[1:].split('.')
+            # We can't compare ints and strings on Python 3, so fudge it
+            # by zero-filling numeric values so simulate a numeric comparison
+            result = tuple([p.zfill(8) if p.isdigit() else p for p in parts])
+        return result
+
+    m = is_semver(s)
+    if not m:
+        raise UnsupportedVersionError(s)
+    groups = m.groups()
+    major, minor, patch = [int(i) for i in groups[:3]]
+    # choose the '|' and '*' so that versions sort correctly
+    pre, build = make_tuple(groups[3], '|'), make_tuple(groups[5], '*')
+    return (major, minor, patch), pre, build
+
+
+class SemanticVersion(Version):
+    def parse(self, s):
+        return _semantic_key(s)
+
+    @property
+    def is_prerelease(self):
+        return self._parts[1][0] != '|'
+
+
+class SemanticMatcher(Matcher):
+    version_class = SemanticVersion
+
+
+class VersionScheme(object):
+    def __init__(self, key, matcher, suggester=None):
+        self.key = key
+        self.matcher = matcher
+        self.suggester = suggester
+
+    def is_valid_version(self, s):
+        try:
+            self.matcher.version_class(s)
+            result = True
+        except UnsupportedVersionError:
+            result = False
+        return result
+
+    def is_valid_matcher(self, s):
+        try:
+            self.matcher(s)
+            result = True
+        except UnsupportedVersionError:
+            result = False
+        return result
+
+    def is_valid_constraint_list(self, s):
+        """
+        Used for processing some metadata fields
+        """
+        # See issue #140. Be tolerant of a single trailing comma.
+        if s.endswith(','):
+            s = s[:-1]
+        return self.is_valid_matcher('dummy_name (%s)' % s)
+
+    def suggest(self, s):
+        if self.suggester is None:
+            result = None
+        else:
+            result = self.suggester(s)
+        return result
+
+
+_SCHEMES = {
+    'normalized': VersionScheme(_normalized_key, NormalizedMatcher,
+                                _suggest_normalized_version),
+    'legacy': VersionScheme(_legacy_key, LegacyMatcher, lambda self, s: s),
+    'semantic': VersionScheme(_semantic_key, SemanticMatcher,
+                              _suggest_semantic_version),
+}
+
+_SCHEMES['default'] = _SCHEMES['normalized']
+
+
+def get_scheme(name):
+    if name not in _SCHEMES:
+        raise ValueError('unknown scheme name: %r' % name)
+    return _SCHEMES[name]

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/distlib/wheel.py

@@ -0,0 +1,1099 @@
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2013-2023 Vinay Sajip.
+# Licensed to the Python Software Foundation under a contributor agreement.
+# See LICENSE.txt and CONTRIBUTORS.txt.
+#
+from __future__ import unicode_literals
+
+import base64
+import codecs
+import datetime
+from email import message_from_file
+import hashlib
+import json
+import logging
+import os
+import posixpath
+import re
+import shutil
+import sys
+import tempfile
+import zipfile
+
+from . import __version__, DistlibException
+from .compat import sysconfig, ZipFile, fsdecode, text_type, filter
+from .database import InstalledDistribution
+from .metadata import Metadata, WHEEL_METADATA_FILENAME, LEGACY_METADATA_FILENAME
+from .util import (FileOperator, convert_path, CSVReader, CSVWriter, Cache,
+                   cached_property, get_cache_base, read_exports, tempdir,
+                   get_platform)
+from .version import NormalizedVersion, UnsupportedVersionError
+
+logger = logging.getLogger(__name__)
+
+cache = None  # created when needed
+
+if hasattr(sys, 'pypy_version_info'):  # pragma: no cover
+    IMP_PREFIX = 'pp'
+elif sys.platform.startswith('java'):  # pragma: no cover
+    IMP_PREFIX = 'jy'
+elif sys.platform == 'cli':  # pragma: no cover
+    IMP_PREFIX = 'ip'
+else:
+    IMP_PREFIX = 'cp'
+
+VER_SUFFIX = sysconfig.get_config_var('py_version_nodot')
+if not VER_SUFFIX:  # pragma: no cover
+    VER_SUFFIX = '%s%s' % sys.version_info[:2]
+PYVER = 'py' + VER_SUFFIX
+IMPVER = IMP_PREFIX + VER_SUFFIX
+
+ARCH = get_platform().replace('-', '_').replace('.', '_')
+
+ABI = sysconfig.get_config_var('SOABI')
+if ABI and ABI.startswith('cpython-'):
+    ABI = ABI.replace('cpython-', 'cp').split('-')[0]
+else:
+
+    def _derive_abi():
+        parts = ['cp', VER_SUFFIX]
+        if sysconfig.get_config_var('Py_DEBUG'):
+            parts.append('d')
+        if IMP_PREFIX == 'cp':
+            vi = sys.version_info[:2]
+            if vi < (3, 8):
+                wpm = sysconfig.get_config_var('WITH_PYMALLOC')
+                if wpm is None:
+                    wpm = True
+                if wpm:
+                    parts.append('m')
+                if vi < (3, 3):
+                    us = sysconfig.get_config_var('Py_UNICODE_SIZE')
+                    if us == 4 or (us is None and sys.maxunicode == 0x10FFFF):
+                        parts.append('u')
+        return ''.join(parts)
+
+    ABI = _derive_abi()
+    del _derive_abi
+
+FILENAME_RE = re.compile(
+    r'''
+(?P<nm>[^-]+)
+-(?P<vn>\d+[^-]*)
+(-(?P<bn>\d+[^-]*))?
+-(?P<py>\w+\d+(\.\w+\d+)*)
+-(?P<bi>\w+)
+-(?P<ar>\w+(\.\w+)*)
+\.whl$
+''', re.IGNORECASE | re.VERBOSE)
+
+NAME_VERSION_RE = re.compile(
+    r'''
+(?P<nm>[^-]+)
+-(?P<vn>\d+[^-]*)
+(-(?P<bn>\d+[^-]*))?$
+''', re.IGNORECASE | re.VERBOSE)
+
+SHEBANG_RE = re.compile(br'\s*#![^\r\n]*')
+SHEBANG_DETAIL_RE = re.compile(br'^(\s*#!("[^"]+"|\S+))\s+(.*)$')
+SHEBANG_PYTHON = b'#!python'
+SHEBANG_PYTHONW = b'#!pythonw'
+
+if os.sep == '/':
+    to_posix = lambda o: o
+else:
+    to_posix = lambda o: o.replace(os.sep, '/')
+
+if sys.version_info[0] < 3:
+    import imp
+else:
+    imp = None
+    import importlib.machinery
+    import importlib.util
+
+
+def _get_suffixes():
+    if imp:
+        return [s[0] for s in imp.get_suffixes()]
+    else:
+        return importlib.machinery.EXTENSION_SUFFIXES
+
+
+def _load_dynamic(name, path):
+    # https://docs.python.org/3/library/importlib.html#importing-a-source-file-directly
+    if imp:
+        return imp.load_dynamic(name, path)
+    else:
+        spec = importlib.util.spec_from_file_location(name, path)
+        module = importlib.util.module_from_spec(spec)
+        sys.modules[name] = module
+        spec.loader.exec_module(module)
+        return module
+
+
+class Mounter(object):
+
+    def __init__(self):
+        self.impure_wheels = {}
+        self.libs = {}
+
+    def add(self, pathname, extensions):
+        self.impure_wheels[pathname] = extensions
+        self.libs.update(extensions)
+
+    def remove(self, pathname):
+        extensions = self.impure_wheels.pop(pathname)
+        for k, v in extensions:
+            if k in self.libs:
+                del self.libs[k]
+
+    def find_module(self, fullname, path=None):
+        if fullname in self.libs:
+            result = self
+        else:
+            result = None
+        return result
+
+    def load_module(self, fullname):
+        if fullname in sys.modules:
+            result = sys.modules[fullname]
+        else:
+            if fullname not in self.libs:
+                raise ImportError('unable to find extension for %s' % fullname)
+            result = _load_dynamic(fullname, self.libs[fullname])
+            result.__loader__ = self
+            parts = fullname.rsplit('.', 1)
+            if len(parts) > 1:
+                result.__package__ = parts[0]
+        return result
+
+
+_hook = Mounter()
+
+
+class Wheel(object):
+    """
+    Class to build and install from Wheel files (PEP 427).
+    """
+
+    wheel_version = (1, 1)
+    hash_kind = 'sha256'
+
+    def __init__(self, filename=None, sign=False, verify=False):
+        """
+        Initialise an instance using a (valid) filename.
+        """
+        self.sign = sign
+        self.should_verify = verify
+        self.buildver = ''
+        self.pyver = [PYVER]
+        self.abi = ['none']
+        self.arch = ['any']
+        self.dirname = os.getcwd()
+        if filename is None:
+            self.name = 'dummy'
+            self.version = '0.1'
+            self._filename = self.filename
+        else:
+            m = NAME_VERSION_RE.match(filename)
+            if m:
+                info = m.groupdict('')
+                self.name = info['nm']
+                # Reinstate the local version separator
+                self.version = info['vn'].replace('_', '-')
+                self.buildver = info['bn']
+                self._filename = self.filename
+            else:
+                dirname, filename = os.path.split(filename)
+                m = FILENAME_RE.match(filename)
+                if not m:
+                    raise DistlibException('Invalid name or '
+                                           'filename: %r' % filename)
+                if dirname:
+                    self.dirname = os.path.abspath(dirname)
+                self._filename = filename
+                info = m.groupdict('')
+                self.name = info['nm']
+                self.version = info['vn']
+                self.buildver = info['bn']
+                self.pyver = info['py'].split('.')
+                self.abi = info['bi'].split('.')
+                self.arch = info['ar'].split('.')
+
+    @property
+    def filename(self):
+        """
+        Build and return a filename from the various components.
+        """
+        if self.buildver:
+            buildver = '-' + self.buildver
+        else:
+            buildver = ''
+        pyver = '.'.join(self.pyver)
+        abi = '.'.join(self.abi)
+        arch = '.'.join(self.arch)
+        # replace - with _ as a local version separator
+        version = self.version.replace('-', '_')
+        return '%s-%s%s-%s-%s-%s.whl' % (self.name, version, buildver, pyver,
+                                         abi, arch)
+
+    @property
+    def exists(self):
+        path = os.path.join(self.dirname, self.filename)
+        return os.path.isfile(path)
+
+    @property
+    def tags(self):
+        for pyver in self.pyver:
+            for abi in self.abi:
+                for arch in self.arch:
+                    yield pyver, abi, arch
+
+    @cached_property
+    def metadata(self):
+        pathname = os.path.join(self.dirname, self.filename)
+        name_ver = '%s-%s' % (self.name, self.version)
+        info_dir = '%s.dist-info' % name_ver
+        wrapper = codecs.getreader('utf-8')
+        with ZipFile(pathname, 'r') as zf:
+            self.get_wheel_metadata(zf)
+            # wv = wheel_metadata['Wheel-Version'].split('.', 1)
+            # file_version = tuple([int(i) for i in wv])
+            # if file_version < (1, 1):
+            # fns = [WHEEL_METADATA_FILENAME, METADATA_FILENAME,
+            # LEGACY_METADATA_FILENAME]
+            # else:
+            # fns = [WHEEL_METADATA_FILENAME, METADATA_FILENAME]
+            fns = [WHEEL_METADATA_FILENAME, LEGACY_METADATA_FILENAME]
+            result = None
+            for fn in fns:
+                try:
+                    metadata_filename = posixpath.join(info_dir, fn)
+                    with zf.open(metadata_filename) as bf:
+                        wf = wrapper(bf)
+                        result = Metadata(fileobj=wf)
+                        if result:
+                            break
+                except KeyError:
+                    pass
+            if not result:
+                raise ValueError('Invalid wheel, because metadata is '
+                                 'missing: looked in %s' % ', '.join(fns))
+        return result
+
+    def get_wheel_metadata(self, zf):
+        name_ver = '%s-%s' % (self.name, self.version)
+        info_dir = '%s.dist-info' % name_ver
+        metadata_filename = posixpath.join(info_dir, 'WHEEL')
+        with zf.open(metadata_filename) as bf:
+            wf = codecs.getreader('utf-8')(bf)
+            message = message_from_file(wf)
+        return dict(message)
+
+    @cached_property
+    def info(self):
+        pathname = os.path.join(self.dirname, self.filename)
+        with ZipFile(pathname, 'r') as zf:
+            result = self.get_wheel_metadata(zf)
+        return result
+
+    def process_shebang(self, data):
+        m = SHEBANG_RE.match(data)
+        if m:
+            end = m.end()
+            shebang, data_after_shebang = data[:end], data[end:]
+            # Preserve any arguments after the interpreter
+            if b'pythonw' in shebang.lower():
+                shebang_python = SHEBANG_PYTHONW
+            else:
+                shebang_python = SHEBANG_PYTHON
+            m = SHEBANG_DETAIL_RE.match(shebang)
+            if m:
+                args = b' ' + m.groups()[-1]
+            else:
+                args = b''
+            shebang = shebang_python + args
+            data = shebang + data_after_shebang
+        else:
+            cr = data.find(b'\r')
+            lf = data.find(b'\n')
+            if cr < 0 or cr > lf:
+                term = b'\n'
+            else:
+                if data[cr:cr + 2] == b'\r\n':
+                    term = b'\r\n'
+                else:
+                    term = b'\r'
+            data = SHEBANG_PYTHON + term + data
+        return data
+
+    def get_hash(self, data, hash_kind=None):
+        if hash_kind is None:
+            hash_kind = self.hash_kind
+        try:
+            hasher = getattr(hashlib, hash_kind)
+        except AttributeError:
+            raise DistlibException('Unsupported hash algorithm: %r' %
+                                   hash_kind)
+        result = hasher(data).digest()
+        result = base64.urlsafe_b64encode(result).rstrip(b'=').decode('ascii')
+        return hash_kind, result
+
+    def write_record(self, records, record_path, archive_record_path):
+        records = list(records)  # make a copy, as mutated
+        records.append((archive_record_path, '', ''))
+        with CSVWriter(record_path) as writer:
+            for row in records:
+                writer.writerow(row)
+
+    def write_records(self, info, libdir, archive_paths):
+        records = []
+        distinfo, info_dir = info
+        # hasher = getattr(hashlib, self.hash_kind)
+        for ap, p in archive_paths:
+            with open(p, 'rb') as f:
+                data = f.read()
+            digest = '%s=%s' % self.get_hash(data)
+            size = os.path.getsize(p)
+            records.append((ap, digest, size))
+
+        p = os.path.join(distinfo, 'RECORD')
+        ap = to_posix(os.path.join(info_dir, 'RECORD'))
+        self.write_record(records, p, ap)
+        archive_paths.append((ap, p))
+
+    def build_zip(self, pathname, archive_paths):
+        with ZipFile(pathname, 'w', zipfile.ZIP_DEFLATED) as zf:
+            for ap, p in archive_paths:
+                logger.debug('Wrote %s to %s in wheel', p, ap)
+                zf.write(p, ap)
+
+    def build(self, paths, tags=None, wheel_version=None):
+        """
+        Build a wheel from files in specified paths, and use any specified tags
+        when determining the name of the wheel.
+        """
+        if tags is None:
+            tags = {}
+
+        libkey = list(filter(lambda o: o in paths, ('purelib', 'platlib')))[0]
+        if libkey == 'platlib':
+            is_pure = 'false'
+            default_pyver = [IMPVER]
+            default_abi = [ABI]
+            default_arch = [ARCH]
+        else:
+            is_pure = 'true'
+            default_pyver = [PYVER]
+            default_abi = ['none']
+            default_arch = ['any']
+
+        self.pyver = tags.get('pyver', default_pyver)
+        self.abi = tags.get('abi', default_abi)
+        self.arch = tags.get('arch', default_arch)
+
+        libdir = paths[libkey]
+
+        name_ver = '%s-%s' % (self.name, self.version)
+        data_dir = '%s.data' % name_ver
+        info_dir = '%s.dist-info' % name_ver
+
+        archive_paths = []
+
+        # First, stuff which is not in site-packages
+        for key in ('data', 'headers', 'scripts'):
+            if key not in paths:
+                continue
+            path = paths[key]
+            if os.path.isdir(path):
+                for root, dirs, files in os.walk(path):
+                    for fn in files:
+                        p = fsdecode(os.path.join(root, fn))
+                        rp = os.path.relpath(p, path)
+                        ap = to_posix(os.path.join(data_dir, key, rp))
+                        archive_paths.append((ap, p))
+                        if key == 'scripts' and not p.endswith('.exe'):
+                            with open(p, 'rb') as f:
+                                data = f.read()
+                            data = self.process_shebang(data)
+                            with open(p, 'wb') as f:
+                                f.write(data)
+
+        # Now, stuff which is in site-packages, other than the
+        # distinfo stuff.
+        path = libdir
+        distinfo = None
+        for root, dirs, files in os.walk(path):
+            if root == path:
+                # At the top level only, save distinfo for later
+                # and skip it for now
+                for i, dn in enumerate(dirs):
+                    dn = fsdecode(dn)
+                    if dn.endswith('.dist-info'):
+                        distinfo = os.path.join(root, dn)
+                        del dirs[i]
+                        break
+                assert distinfo, '.dist-info directory expected, not found'
+
+            for fn in files:
+                # comment out next suite to leave .pyc files in
+                if fsdecode(fn).endswith(('.pyc', '.pyo')):
+                    continue
+                p = os.path.join(root, fn)
+                rp = to_posix(os.path.relpath(p, path))
+                archive_paths.append((rp, p))
+
+        # Now distinfo. Assumed to be flat, i.e. os.listdir is enough.
+        files = os.listdir(distinfo)
+        for fn in files:
+            if fn not in ('RECORD', 'INSTALLER', 'SHARED', 'WHEEL'):
+                p = fsdecode(os.path.join(distinfo, fn))
+                ap = to_posix(os.path.join(info_dir, fn))
+                archive_paths.append((ap, p))
+
+        wheel_metadata = [
+            'Wheel-Version: %d.%d' % (wheel_version or self.wheel_version),
+            'Generator: distlib %s' % __version__,
+            'Root-Is-Purelib: %s' % is_pure,
+        ]
+        for pyver, abi, arch in self.tags:
+            wheel_metadata.append('Tag: %s-%s-%s' % (pyver, abi, arch))
+        p = os.path.join(distinfo, 'WHEEL')
+        with open(p, 'w') as f:
+            f.write('\n'.join(wheel_metadata))
+        ap = to_posix(os.path.join(info_dir, 'WHEEL'))
+        archive_paths.append((ap, p))
+
+        # sort the entries by archive path. Not needed by any spec, but it
+        # keeps the archive listing and RECORD tidier than they would otherwise
+        # be. Use the number of path segments to keep directory entries together,
+        # and keep the dist-info stuff at the end.
+        def sorter(t):
+            ap = t[0]
+            n = ap.count('/')
+            if '.dist-info' in ap:
+                n += 10000
+            return (n, ap)
+
+        archive_paths = sorted(archive_paths, key=sorter)
+
+        # Now, at last, RECORD.
+        # Paths in here are archive paths - nothing else makes sense.
+        self.write_records((distinfo, info_dir), libdir, archive_paths)
+        # Now, ready to build the zip file
+        pathname = os.path.join(self.dirname, self.filename)
+        self.build_zip(pathname, archive_paths)
+        return pathname
+
+    def skip_entry(self, arcname):
+        """
+        Determine whether an archive entry should be skipped when verifying
+        or installing.
+        """
+        # The signature file won't be in RECORD,
+        # and we  don't currently don't do anything with it
+        # We also skip directories, as they won't be in RECORD
+        # either. See:
+        #
+        # https://github.com/pypa/wheel/issues/294
+        # https://github.com/pypa/wheel/issues/287
+        # https://github.com/pypa/wheel/pull/289
+        #
+        return arcname.endswith(('/', '/RECORD.jws'))
+
+    def install(self, paths, maker, **kwargs):
+        """
+        Install a wheel to the specified paths. If kwarg ``warner`` is
+        specified, it should be a callable, which will be called with two
+        tuples indicating the wheel version of this software and the wheel
+        version in the file, if there is a discrepancy in the versions.
+        This can be used to issue any warnings to raise any exceptions.
+        If kwarg ``lib_only`` is True, only the purelib/platlib files are
+        installed, and the headers, scripts, data and dist-info metadata are
+        not written. If kwarg ``bytecode_hashed_invalidation`` is True, written
+        bytecode will try to use file-hash based invalidation (PEP-552) on
+        supported interpreter versions (CPython 2.7+).
+
+        The return value is a :class:`InstalledDistribution` instance unless
+        ``options.lib_only`` is True, in which case the return value is ``None``.
+        """
+
+        dry_run = maker.dry_run
+        warner = kwargs.get('warner')
+        lib_only = kwargs.get('lib_only', False)
+        bc_hashed_invalidation = kwargs.get('bytecode_hashed_invalidation',
+                                            False)
+
+        pathname = os.path.join(self.dirname, self.filename)
+        name_ver = '%s-%s' % (self.name, self.version)
+        data_dir = '%s.data' % name_ver
+        info_dir = '%s.dist-info' % name_ver
+
+        metadata_name = posixpath.join(info_dir, LEGACY_METADATA_FILENAME)
+        wheel_metadata_name = posixpath.join(info_dir, 'WHEEL')
+        record_name = posixpath.join(info_dir, 'RECORD')
+
+        wrapper = codecs.getreader('utf-8')
+
+        with ZipFile(pathname, 'r') as zf:
+            with zf.open(wheel_metadata_name) as bwf:
+                wf = wrapper(bwf)
+                message = message_from_file(wf)
+            wv = message['Wheel-Version'].split('.', 1)
+            file_version = tuple([int(i) for i in wv])
+            if (file_version != self.wheel_version) and warner:
+                warner(self.wheel_version, file_version)
+
+            if message['Root-Is-Purelib'] == 'true':
+                libdir = paths['purelib']
+            else:
+                libdir = paths['platlib']
+
+            records = {}
+            with zf.open(record_name) as bf:
+                with CSVReader(stream=bf) as reader:
+                    for row in reader:
+                        p = row[0]
+                        records[p] = row
+
+            data_pfx = posixpath.join(data_dir, '')
+            info_pfx = posixpath.join(info_dir, '')
+            script_pfx = posixpath.join(data_dir, 'scripts', '')
+
+            # make a new instance rather than a copy of maker's,
+            # as we mutate it
+            fileop = FileOperator(dry_run=dry_run)
+            fileop.record = True  # so we can rollback if needed
+
+            bc = not sys.dont_write_bytecode  # Double negatives. Lovely!
+
+            outfiles = []  # for RECORD writing
+
+            # for script copying/shebang processing
+            workdir = tempfile.mkdtemp()
+            # set target dir later
+            # we default add_launchers to False, as the
+            # Python Launcher should be used instead
+            maker.source_dir = workdir
+            maker.target_dir = None
+            try:
+                for zinfo in zf.infolist():
+                    arcname = zinfo.filename
+                    if isinstance(arcname, text_type):
+                        u_arcname = arcname
+                    else:
+                        u_arcname = arcname.decode('utf-8')
+                    if self.skip_entry(u_arcname):
+                        continue
+                    row = records[u_arcname]
+                    if row[2] and str(zinfo.file_size) != row[2]:
+                        raise DistlibException('size mismatch for '
+                                               '%s' % u_arcname)
+                    if row[1]:
+                        kind, value = row[1].split('=', 1)
+                        with zf.open(arcname) as bf:
+                            data = bf.read()
+                        _, digest = self.get_hash(data, kind)
+                        if digest != value:
+                            raise DistlibException('digest mismatch for '
+                                                   '%s' % arcname)
+
+                    if lib_only and u_arcname.startswith((info_pfx, data_pfx)):
+                        logger.debug('lib_only: skipping %s', u_arcname)
+                        continue
+                    is_script = (u_arcname.startswith(script_pfx)
+                                 and not u_arcname.endswith('.exe'))
+
+                    if u_arcname.startswith(data_pfx):
+                        _, where, rp = u_arcname.split('/', 2)
+                        outfile = os.path.join(paths[where], convert_path(rp))
+                    else:
+                        # meant for site-packages.
+                        if u_arcname in (wheel_metadata_name, record_name):
+                            continue
+                        outfile = os.path.join(libdir, convert_path(u_arcname))
+                    if not is_script:
+                        with zf.open(arcname) as bf:
+                            fileop.copy_stream(bf, outfile)
+                        # Issue #147: permission bits aren't preserved. Using
+                        # zf.extract(zinfo, libdir) should have worked, but didn't,
+                        # see https://www.thetopsites.net/article/53834422.shtml
+                        # So ... manually preserve permission bits as given in zinfo
+                        if os.name == 'posix':
+                            # just set the normal permission bits
+                            os.chmod(outfile,
+                                     (zinfo.external_attr >> 16) & 0x1FF)
+                        outfiles.append(outfile)
+                        # Double check the digest of the written file
+                        if not dry_run and row[1]:
+                            with open(outfile, 'rb') as bf:
+                                data = bf.read()
+                                _, newdigest = self.get_hash(data, kind)
+                                if newdigest != digest:
+                                    raise DistlibException('digest mismatch '
+                                                           'on write for '
+                                                           '%s' % outfile)
+                        if bc and outfile.endswith('.py'):
+                            try:
+                                pyc = fileop.byte_compile(
+                                    outfile,
+                                    hashed_invalidation=bc_hashed_invalidation)
+                                outfiles.append(pyc)
+                            except Exception:
+                                # Don't give up if byte-compilation fails,
+                                # but log it and perhaps warn the user
+                                logger.warning('Byte-compilation failed',
+                                               exc_info=True)
+                    else:
+                        fn = os.path.basename(convert_path(arcname))
+                        workname = os.path.join(workdir, fn)
+                        with zf.open(arcname) as bf:
+                            fileop.copy_stream(bf, workname)
+
+                        dn, fn = os.path.split(outfile)
+                        maker.target_dir = dn
+                        filenames = maker.make(fn)
+                        fileop.set_executable_mode(filenames)
+                        outfiles.extend(filenames)
+
+                if lib_only:
+                    logger.debug('lib_only: returning None')
+                    dist = None
+                else:
+                    # Generate scripts
+
+                    # Try to get pydist.json so we can see if there are
+                    # any commands to generate. If this fails (e.g. because
+                    # of a legacy wheel), log a warning but don't give up.
+                    commands = None
+                    file_version = self.info['Wheel-Version']
+                    if file_version == '1.0':
+                        # Use legacy info
+                        ep = posixpath.join(info_dir, 'entry_points.txt')
+                        try:
+                            with zf.open(ep) as bwf:
+                                epdata = read_exports(bwf)
+                            commands = {}
+                            for key in ('console', 'gui'):
+                                k = '%s_scripts' % key
+                                if k in epdata:
+                                    commands['wrap_%s' % key] = d = {}
+                                    for v in epdata[k].values():
+                                        s = '%s:%s' % (v.prefix, v.suffix)
+                                        if v.flags:
+                                            s += ' [%s]' % ','.join(v.flags)
+                                        d[v.name] = s
+                        except Exception:
+                            logger.warning('Unable to read legacy script '
+                                           'metadata, so cannot generate '
+                                           'scripts')
+                    else:
+                        try:
+                            with zf.open(metadata_name) as bwf:
+                                wf = wrapper(bwf)
+                                commands = json.load(wf).get('extensions')
+                                if commands:
+                                    commands = commands.get('python.commands')
+                        except Exception:
+                            logger.warning('Unable to read JSON metadata, so '
+                                           'cannot generate scripts')
+                    if commands:
+                        console_scripts = commands.get('wrap_console', {})
+                        gui_scripts = commands.get('wrap_gui', {})
+                        if console_scripts or gui_scripts:
+                            script_dir = paths.get('scripts', '')
+                            if not os.path.isdir(script_dir):
+                                raise ValueError('Valid script path not '
+                                                 'specified')
+                            maker.target_dir = script_dir
+                            for k, v in console_scripts.items():
+                                script = '%s = %s' % (k, v)
+                                filenames = maker.make(script)
+                                fileop.set_executable_mode(filenames)
+
+                            if gui_scripts:
+                                options = {'gui': True}
+                                for k, v in gui_scripts.items():
+                                    script = '%s = %s' % (k, v)
+                                    filenames = maker.make(script, options)
+                                    fileop.set_executable_mode(filenames)
+
+                    p = os.path.join(libdir, info_dir)
+                    dist = InstalledDistribution(p)
+
+                    # Write SHARED
+                    paths = dict(paths)  # don't change passed in dict
+                    del paths['purelib']
+                    del paths['platlib']
+                    paths['lib'] = libdir
+                    p = dist.write_shared_locations(paths, dry_run)
+                    if p:
+                        outfiles.append(p)
+
+                    # Write RECORD
+                    dist.write_installed_files(outfiles, paths['prefix'],
+                                               dry_run)
+                return dist
+            except Exception:  # pragma: no cover
+                logger.exception('installation failed.')
+                fileop.rollback()
+                raise
+            finally:
+                shutil.rmtree(workdir)
+
+    def _get_dylib_cache(self):
+        global cache
+        if cache is None:
+            # Use native string to avoid issues on 2.x: see Python #20140.
+            base = os.path.join(get_cache_base(), str('dylib-cache'),
+                                '%s.%s' % sys.version_info[:2])
+            cache = Cache(base)
+        return cache
+
+    def _get_extensions(self):
+        pathname = os.path.join(self.dirname, self.filename)
+        name_ver = '%s-%s' % (self.name, self.version)
+        info_dir = '%s.dist-info' % name_ver
+        arcname = posixpath.join(info_dir, 'EXTENSIONS')
+        wrapper = codecs.getreader('utf-8')
+        result = []
+        with ZipFile(pathname, 'r') as zf:
+            try:
+                with zf.open(arcname) as bf:
+                    wf = wrapper(bf)
+                    extensions = json.load(wf)
+                    cache = self._get_dylib_cache()
+                    prefix = cache.prefix_to_dir(pathname)
+                    cache_base = os.path.join(cache.base, prefix)
+                    if not os.path.isdir(cache_base):
+                        os.makedirs(cache_base)
+                    for name, relpath in extensions.items():
+                        dest = os.path.join(cache_base, convert_path(relpath))
+                        if not os.path.exists(dest):
+                            extract = True
+                        else:
+                            file_time = os.stat(dest).st_mtime
+                            file_time = datetime.datetime.fromtimestamp(
+                                file_time)
+                            info = zf.getinfo(relpath)
+                            wheel_time = datetime.datetime(*info.date_time)
+                            extract = wheel_time > file_time
+                        if extract:
+                            zf.extract(relpath, cache_base)
+                        result.append((name, dest))
+            except KeyError:
+                pass
+        return result
+
+    def is_compatible(self):
+        """
+        Determine if a wheel is compatible with the running system.
+        """
+        return is_compatible(self)
+
+    def is_mountable(self):
+        """
+        Determine if a wheel is asserted as mountable by its metadata.
+        """
+        return True  # for now - metadata details TBD
+
+    def mount(self, append=False):
+        pathname = os.path.abspath(os.path.join(self.dirname, self.filename))
+        if not self.is_compatible():
+            msg = 'Wheel %s not compatible with this Python.' % pathname
+            raise DistlibException(msg)
+        if not self.is_mountable():
+            msg = 'Wheel %s is marked as not mountable.' % pathname
+            raise DistlibException(msg)
+        if pathname in sys.path:
+            logger.debug('%s already in path', pathname)
+        else:
+            if append:
+                sys.path.append(pathname)
+            else:
+                sys.path.insert(0, pathname)
+            extensions = self._get_extensions()
+            if extensions:
+                if _hook not in sys.meta_path:
+                    sys.meta_path.append(_hook)
+                _hook.add(pathname, extensions)
+
+    def unmount(self):
+        pathname = os.path.abspath(os.path.join(self.dirname, self.filename))
+        if pathname not in sys.path:
+            logger.debug('%s not in path', pathname)
+        else:
+            sys.path.remove(pathname)
+            if pathname in _hook.impure_wheels:
+                _hook.remove(pathname)
+            if not _hook.impure_wheels:
+                if _hook in sys.meta_path:
+                    sys.meta_path.remove(_hook)
+
+    def verify(self):
+        pathname = os.path.join(self.dirname, self.filename)
+        name_ver = '%s-%s' % (self.name, self.version)
+        # data_dir = '%s.data' % name_ver
+        info_dir = '%s.dist-info' % name_ver
+
+        # metadata_name = posixpath.join(info_dir, LEGACY_METADATA_FILENAME)
+        wheel_metadata_name = posixpath.join(info_dir, 'WHEEL')
+        record_name = posixpath.join(info_dir, 'RECORD')
+
+        wrapper = codecs.getreader('utf-8')
+
+        with ZipFile(pathname, 'r') as zf:
+            with zf.open(wheel_metadata_name) as bwf:
+                wf = wrapper(bwf)
+                message_from_file(wf)
+            # wv = message['Wheel-Version'].split('.', 1)
+            # file_version = tuple([int(i) for i in wv])
+            # TODO version verification
+
+            records = {}
+            with zf.open(record_name) as bf:
+                with CSVReader(stream=bf) as reader:
+                    for row in reader:
+                        p = row[0]
+                        records[p] = row
+
+            for zinfo in zf.infolist():
+                arcname = zinfo.filename
+                if isinstance(arcname, text_type):
+                    u_arcname = arcname
+                else:
+                    u_arcname = arcname.decode('utf-8')
+                # See issue #115: some wheels have .. in their entries, but
+                # in the filename ... e.g. __main__..py ! So the check is
+                # updated to look for .. in the directory portions
+                p = u_arcname.split('/')
+                if '..' in p:
+                    raise DistlibException('invalid entry in '
+                                           'wheel: %r' % u_arcname)
+
+                if self.skip_entry(u_arcname):
+                    continue
+                row = records[u_arcname]
+                if row[2] and str(zinfo.file_size) != row[2]:
+                    raise DistlibException('size mismatch for '
+                                           '%s' % u_arcname)
+                if row[1]:
+                    kind, value = row[1].split('=', 1)
+                    with zf.open(arcname) as bf:
+                        data = bf.read()
+                    _, digest = self.get_hash(data, kind)
+                    if digest != value:
+                        raise DistlibException('digest mismatch for '
+                                               '%s' % arcname)
+
+    def update(self, modifier, dest_dir=None, **kwargs):
+        """
+        Update the contents of a wheel in a generic way. The modifier should
+        be a callable which expects a dictionary argument: its keys are
+        archive-entry paths, and its values are absolute filesystem paths
+        where the contents the corresponding archive entries can be found. The
+        modifier is free to change the contents of the files pointed to, add
+        new entries and remove entries, before returning. This method will
+        extract the entire contents of the wheel to a temporary location, call
+        the modifier, and then use the passed (and possibly updated)
+        dictionary to write a new wheel. If ``dest_dir`` is specified, the new
+        wheel is written there -- otherwise, the original wheel is overwritten.
+
+        The modifier should return True if it updated the wheel, else False.
+        This method returns the same value the modifier returns.
+        """
+
+        def get_version(path_map, info_dir):
+            version = path = None
+            key = '%s/%s' % (info_dir, LEGACY_METADATA_FILENAME)
+            if key not in path_map:
+                key = '%s/PKG-INFO' % info_dir
+            if key in path_map:
+                path = path_map[key]
+                version = Metadata(path=path).version
+            return version, path
+
+        def update_version(version, path):
+            updated = None
+            try:
+                NormalizedVersion(version)
+                i = version.find('-')
+                if i < 0:
+                    updated = '%s+1' % version
+                else:
+                    parts = [int(s) for s in version[i + 1:].split('.')]
+                    parts[-1] += 1
+                    updated = '%s+%s' % (version[:i], '.'.join(
+                        str(i) for i in parts))
+            except UnsupportedVersionError:
+                logger.debug(
+                    'Cannot update non-compliant (PEP-440) '
+                    'version %r', version)
+            if updated:
+                md = Metadata(path=path)
+                md.version = updated
+                legacy = path.endswith(LEGACY_METADATA_FILENAME)
+                md.write(path=path, legacy=legacy)
+                logger.debug('Version updated from %r to %r', version, updated)
+
+        pathname = os.path.join(self.dirname, self.filename)
+        name_ver = '%s-%s' % (self.name, self.version)
+        info_dir = '%s.dist-info' % name_ver
+        record_name = posixpath.join(info_dir, 'RECORD')
+        with tempdir() as workdir:
+            with ZipFile(pathname, 'r') as zf:
+                path_map = {}
+                for zinfo in zf.infolist():
+                    arcname = zinfo.filename
+                    if isinstance(arcname, text_type):
+                        u_arcname = arcname
+                    else:
+                        u_arcname = arcname.decode('utf-8')
+                    if u_arcname == record_name:
+                        continue
+                    if '..' in u_arcname:
+                        raise DistlibException('invalid entry in '
+                                               'wheel: %r' % u_arcname)
+                    zf.extract(zinfo, workdir)
+                    path = os.path.join(workdir, convert_path(u_arcname))
+                    path_map[u_arcname] = path
+
+            # Remember the version.
+            original_version, _ = get_version(path_map, info_dir)
+            # Files extracted. Call the modifier.
+            modified = modifier(path_map, **kwargs)
+            if modified:
+                # Something changed - need to build a new wheel.
+                current_version, path = get_version(path_map, info_dir)
+                if current_version and (current_version == original_version):
+                    # Add or update local version to signify changes.
+                    update_version(current_version, path)
+                # Decide where the new wheel goes.
+                if dest_dir is None:
+                    fd, newpath = tempfile.mkstemp(suffix='.whl',
+                                                   prefix='wheel-update-',
+                                                   dir=workdir)
+                    os.close(fd)
+                else:
+                    if not os.path.isdir(dest_dir):
+                        raise DistlibException('Not a directory: %r' %
+                                               dest_dir)
+                    newpath = os.path.join(dest_dir, self.filename)
+                archive_paths = list(path_map.items())
+                distinfo = os.path.join(workdir, info_dir)
+                info = distinfo, info_dir
+                self.write_records(info, workdir, archive_paths)
+                self.build_zip(newpath, archive_paths)
+                if dest_dir is None:
+                    shutil.copyfile(newpath, pathname)
+        return modified
+
+
+def _get_glibc_version():
+    import platform
+    ver = platform.libc_ver()
+    result = []
+    if ver[0] == 'glibc':
+        for s in ver[1].split('.'):
+            result.append(int(s) if s.isdigit() else 0)
+        result = tuple(result)
+    return result
+
+
+def compatible_tags():
+    """
+    Return (pyver, abi, arch) tuples compatible with this Python.
+    """
+    versions = [VER_SUFFIX]
+    major = VER_SUFFIX[0]
+    for minor in range(sys.version_info[1] - 1, -1, -1):
+        versions.append(''.join([major, str(minor)]))
+
+    abis = []
+    for suffix in _get_suffixes():
+        if suffix.startswith('.abi'):
+            abis.append(suffix.split('.', 2)[1])
+    abis.sort()
+    if ABI != 'none':
+        abis.insert(0, ABI)
+    abis.append('none')
+    result = []
+
+    arches = [ARCH]
+    if sys.platform == 'darwin':
+        m = re.match(r'(\w+)_(\d+)_(\d+)_(\w+)$', ARCH)
+        if m:
+            name, major, minor, arch = m.groups()
+            minor = int(minor)
+            matches = [arch]
+            if arch in ('i386', 'ppc'):
+                matches.append('fat')
+            if arch in ('i386', 'ppc', 'x86_64'):
+                matches.append('fat3')
+            if arch in ('ppc64', 'x86_64'):
+                matches.append('fat64')
+            if arch in ('i386', 'x86_64'):
+                matches.append('intel')
+            if arch in ('i386', 'x86_64', 'intel', 'ppc', 'ppc64'):
+                matches.append('universal')
+            while minor >= 0:
+                for match in matches:
+                    s = '%s_%s_%s_%s' % (name, major, minor, match)
+                    if s != ARCH:  # already there
+                        arches.append(s)
+                minor -= 1
+
+    # Most specific - our Python version, ABI and arch
+    for abi in abis:
+        for arch in arches:
+            result.append((''.join((IMP_PREFIX, versions[0])), abi, arch))
+            # manylinux
+            if abi != 'none' and sys.platform.startswith('linux'):
+                arch = arch.replace('linux_', '')
+                parts = _get_glibc_version()
+                if len(parts) == 2:
+                    if parts >= (2, 5):
+                        result.append((''.join((IMP_PREFIX, versions[0])), abi,
+                                       'manylinux1_%s' % arch))
+                    if parts >= (2, 12):
+                        result.append((''.join((IMP_PREFIX, versions[0])), abi,
+                                       'manylinux2010_%s' % arch))
+                    if parts >= (2, 17):
+                        result.append((''.join((IMP_PREFIX, versions[0])), abi,
+                                       'manylinux2014_%s' % arch))
+                    result.append(
+                        (''.join((IMP_PREFIX, versions[0])), abi,
+                         'manylinux_%s_%s_%s' % (parts[0], parts[1], arch)))
+
+    # where no ABI / arch dependency, but IMP_PREFIX dependency
+    for i, version in enumerate(versions):
+        result.append((''.join((IMP_PREFIX, version)), 'none', 'any'))
+        if i == 0:
+            result.append((''.join((IMP_PREFIX, version[0])), 'none', 'any'))
+
+    # no IMP_PREFIX, ABI or arch dependency
+    for i, version in enumerate(versions):
+        result.append((''.join(('py', version)), 'none', 'any'))
+        if i == 0:
+            result.append((''.join(('py', version[0])), 'none', 'any'))
+
+    return set(result)
+
+
+COMPATIBLE_TAGS = compatible_tags()
+
+del compatible_tags
+
+
+def is_compatible(wheel, tags=None):
+    if not isinstance(wheel, Wheel):
+        wheel = Wheel(wheel)  # assume it's a filename
+    result = False
+    if tags is None:
+        tags = COMPATIBLE_TAGS
+    for ver, abi, arch in tags:
+        if ver in wheel.pyver and abi in wheel.abi and arch in wheel.arch:
+            result = True
+            break
+    return result

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/__init__.py

@@ -0,0 +1,322 @@
+# module pyparsing.py
+#
+# Copyright (c) 2003-2022  Paul T. McGuire
+#
+# Permission is hereby granted, free of charge, to any person obtaining
+# a copy of this software and associated documentation files (the
+# "Software"), to deal in the Software without restriction, including
+# without limitation the rights to use, copy, modify, merge, publish,
+# distribute, sublicense, and/or sell copies of the Software, and to
+# permit persons to whom the Software is furnished to do so, subject to
+# the following conditions:
+#
+# The above copyright notice and this permission notice shall be
+# included in all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+# IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+# CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+# TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+# SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+#
+
+__doc__ = """
+pyparsing module - Classes and methods to define and execute parsing grammars
+=============================================================================
+
+The pyparsing module is an alternative approach to creating and
+executing simple grammars, vs. the traditional lex/yacc approach, or the
+use of regular expressions.  With pyparsing, you don't need to learn
+a new syntax for defining grammars or matching expressions - the parsing
+module provides a library of classes that you use to construct the
+grammar directly in Python.
+
+Here is a program to parse "Hello, World!" (or any greeting of the form
+``"<salutation>, <addressee>!"``), built up using :class:`Word`,
+:class:`Literal`, and :class:`And` elements
+(the :meth:`'+'<ParserElement.__add__>` operators create :class:`And` expressions,
+and the strings are auto-converted to :class:`Literal` expressions)::
+
+    from pip._vendor.pyparsing import Word, alphas
+
+    # define grammar of a greeting
+    greet = Word(alphas) + "," + Word(alphas) + "!"
+
+    hello = "Hello, World!"
+    print(hello, "->", greet.parse_string(hello))
+
+The program outputs the following::
+
+    Hello, World! -> ['Hello', ',', 'World', '!']
+
+The Python representation of the grammar is quite readable, owing to the
+self-explanatory class names, and the use of :class:`'+'<And>`,
+:class:`'|'<MatchFirst>`, :class:`'^'<Or>` and :class:`'&'<Each>` operators.
+
+The :class:`ParseResults` object returned from
+:class:`ParserElement.parse_string` can be
+accessed as a nested list, a dictionary, or an object with named
+attributes.
+
+The pyparsing module handles some of the problems that are typically
+vexing when writing text parsers:
+
+  - extra or missing whitespace (the above program will also handle
+    "Hello,World!", "Hello  ,  World  !", etc.)
+  - quoted strings
+  - embedded comments
+
+
+Getting Started -
+-----------------
+Visit the classes :class:`ParserElement` and :class:`ParseResults` to
+see the base classes that most other pyparsing
+classes inherit from. Use the docstrings for examples of how to:
+
+ - construct literal match expressions from :class:`Literal` and
+   :class:`CaselessLiteral` classes
+ - construct character word-group expressions using the :class:`Word`
+   class
+ - see how to create repetitive expressions using :class:`ZeroOrMore`
+   and :class:`OneOrMore` classes
+ - use :class:`'+'<And>`, :class:`'|'<MatchFirst>`, :class:`'^'<Or>`,
+   and :class:`'&'<Each>` operators to combine simple expressions into
+   more complex ones
+ - associate names with your parsed results using
+   :class:`ParserElement.set_results_name`
+ - access the parsed data, which is returned as a :class:`ParseResults`
+   object
+ - find some helpful expression short-cuts like :class:`DelimitedList`
+   and :class:`one_of`
+ - find more useful common expressions in the :class:`pyparsing_common`
+   namespace class
+"""
+from typing import NamedTuple
+
+
+class version_info(NamedTuple):
+    major: int
+    minor: int
+    micro: int
+    releaselevel: str
+    serial: int
+
+    @property
+    def __version__(self):
+        return (
+            f"{self.major}.{self.minor}.{self.micro}"
+            + (
+                f"{'r' if self.releaselevel[0] == 'c' else ''}{self.releaselevel[0]}{self.serial}",
+                "",
+            )[self.releaselevel == "final"]
+        )
+
+    def __str__(self):
+        return f"{__name__} {self.__version__} / {__version_time__}"
+
+    def __repr__(self):
+        return f"{__name__}.{type(self).__name__}({', '.join('{}={!r}'.format(*nv) for nv in zip(self._fields, self))})"
+
+
+__version_info__ = version_info(3, 1, 0, "final", 1)
+__version_time__ = "18 Jun 2023 14:05 UTC"
+__version__ = __version_info__.__version__
+__versionTime__ = __version_time__
+__author__ = "Paul McGuire <ptmcg.gm+pyparsing@gmail.com>"
+
+from .util import *
+from .exceptions import *
+from .actions import *
+from .core import __diag__, __compat__
+from .results import *
+from .core import *  # type: ignore[misc, assignment]
+from .core import _builtin_exprs as core_builtin_exprs
+from .helpers import *  # type: ignore[misc, assignment]
+from .helpers import _builtin_exprs as helper_builtin_exprs
+
+from .unicode import unicode_set, UnicodeRangeList, pyparsing_unicode as unicode
+from .testing import pyparsing_test as testing
+from .common import (
+    pyparsing_common as common,
+    _builtin_exprs as common_builtin_exprs,
+)
+
+# define backward compat synonyms
+if "pyparsing_unicode" not in globals():
+    pyparsing_unicode = unicode  # type: ignore[misc]
+if "pyparsing_common" not in globals():
+    pyparsing_common = common  # type: ignore[misc]
+if "pyparsing_test" not in globals():
+    pyparsing_test = testing  # type: ignore[misc]
+
+core_builtin_exprs += common_builtin_exprs + helper_builtin_exprs
+
+
+__all__ = [
+    "__version__",
+    "__version_time__",
+    "__author__",
+    "__compat__",
+    "__diag__",
+    "And",
+    "AtLineStart",
+    "AtStringStart",
+    "CaselessKeyword",
+    "CaselessLiteral",
+    "CharsNotIn",
+    "CloseMatch",
+    "Combine",
+    "DelimitedList",
+    "Dict",
+    "Each",
+    "Empty",
+    "FollowedBy",
+    "Forward",
+    "GoToColumn",
+    "Group",
+    "IndentedBlock",
+    "Keyword",
+    "LineEnd",
+    "LineStart",
+    "Literal",
+    "Located",
+    "PrecededBy",
+    "MatchFirst",
+    "NoMatch",
+    "NotAny",
+    "OneOrMore",
+    "OnlyOnce",
+    "OpAssoc",
+    "Opt",
+    "Optional",
+    "Or",
+    "ParseBaseException",
+    "ParseElementEnhance",
+    "ParseException",
+    "ParseExpression",
+    "ParseFatalException",
+    "ParseResults",
+    "ParseSyntaxException",
+    "ParserElement",
+    "PositionToken",
+    "QuotedString",
+    "RecursiveGrammarException",
+    "Regex",
+    "SkipTo",
+    "StringEnd",
+    "StringStart",
+    "Suppress",
+    "Token",
+    "TokenConverter",
+    "White",
+    "Word",
+    "WordEnd",
+    "WordStart",
+    "ZeroOrMore",
+    "Char",
+    "alphanums",
+    "alphas",
+    "alphas8bit",
+    "any_close_tag",
+    "any_open_tag",
+    "autoname_elements",
+    "c_style_comment",
+    "col",
+    "common_html_entity",
+    "condition_as_parse_action",
+    "counted_array",
+    "cpp_style_comment",
+    "dbl_quoted_string",
+    "dbl_slash_comment",
+    "delimited_list",
+    "dict_of",
+    "empty",
+    "hexnums",
+    "html_comment",
+    "identchars",
+    "identbodychars",
+    "infix_notation",
+    "java_style_comment",
+    "line",
+    "line_end",
+    "line_start",
+    "lineno",
+    "make_html_tags",
+    "make_xml_tags",
+    "match_only_at_col",
+    "match_previous_expr",
+    "match_previous_literal",
+    "nested_expr",
+    "null_debug_action",
+    "nums",
+    "one_of",
+    "original_text_for",
+    "printables",
+    "punc8bit",
+    "pyparsing_common",
+    "pyparsing_test",
+    "pyparsing_unicode",
+    "python_style_comment",
+    "quoted_string",
+    "remove_quotes",
+    "replace_with",
+    "replace_html_entity",
+    "rest_of_line",
+    "sgl_quoted_string",
+    "srange",
+    "string_end",
+    "string_start",
+    "token_map",
+    "trace_parse_action",
+    "ungroup",
+    "unicode_set",
+    "unicode_string",
+    "with_attribute",
+    "with_class",
+    # pre-PEP8 compatibility names
+    "__versionTime__",
+    "anyCloseTag",
+    "anyOpenTag",
+    "cStyleComment",
+    "commonHTMLEntity",
+    "conditionAsParseAction",
+    "countedArray",
+    "cppStyleComment",
+    "dblQuotedString",
+    "dblSlashComment",
+    "delimitedList",
+    "dictOf",
+    "htmlComment",
+    "indentedBlock",
+    "infixNotation",
+    "javaStyleComment",
+    "lineEnd",
+    "lineStart",
+    "locatedExpr",
+    "makeHTMLTags",
+    "makeXMLTags",
+    "matchOnlyAtCol",
+    "matchPreviousExpr",
+    "matchPreviousLiteral",
+    "nestedExpr",
+    "nullDebugAction",
+    "oneOf",
+    "opAssoc",
+    "originalTextFor",
+    "pythonStyleComment",
+    "quotedString",
+    "removeQuotes",
+    "replaceHTMLEntity",
+    "replaceWith",
+    "restOfLine",
+    "sglQuotedString",
+    "stringEnd",
+    "stringStart",
+    "tokenMap",
+    "traceParseAction",
+    "unicodeString",
+    "withAttribute",
+    "withClass",
+]

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/actions.py

@@ -0,0 +1,217 @@
+# actions.py
+
+from .exceptions import ParseException
+from .util import col, replaced_by_pep8
+
+
+class OnlyOnce:
+    """
+    Wrapper for parse actions, to ensure they are only called once.
+    """
+
+    def __init__(self, method_call):
+        from .core import _trim_arity
+
+        self.callable = _trim_arity(method_call)
+        self.called = False
+
+    def __call__(self, s, l, t):
+        if not self.called:
+            results = self.callable(s, l, t)
+            self.called = True
+            return results
+        raise ParseException(s, l, "OnlyOnce obj called multiple times w/out reset")
+
+    def reset(self):
+        """
+        Allow the associated parse action to be called once more.
+        """
+
+        self.called = False
+
+
+def match_only_at_col(n):
+    """
+    Helper method for defining parse actions that require matching at
+    a specific column in the input text.
+    """
+
+    def verify_col(strg, locn, toks):
+        if col(locn, strg) != n:
+            raise ParseException(strg, locn, f"matched token not at column {n}")
+
+    return verify_col
+
+
+def replace_with(repl_str):
+    """
+    Helper method for common parse actions that simply return
+    a literal value.  Especially useful when used with
+    :class:`transform_string<ParserElement.transform_string>` ().
+
+    Example::
+
+        num = Word(nums).set_parse_action(lambda toks: int(toks[0]))
+        na = one_of("N/A NA").set_parse_action(replace_with(math.nan))
+        term = na | num
+
+        term[1, ...].parse_string("324 234 N/A 234") # -> [324, 234, nan, 234]
+    """
+    return lambda s, l, t: [repl_str]
+
+
+def remove_quotes(s, l, t):
+    """
+    Helper parse action for removing quotation marks from parsed
+    quoted strings.
+
+    Example::
+
+        # by default, quotation marks are included in parsed results
+        quoted_string.parse_string("'Now is the Winter of our Discontent'") # -> ["'Now is the Winter of our Discontent'"]
+
+        # use remove_quotes to strip quotation marks from parsed results
+        quoted_string.set_parse_action(remove_quotes)
+        quoted_string.parse_string("'Now is the Winter of our Discontent'") # -> ["Now is the Winter of our Discontent"]
+    """
+    return t[0][1:-1]
+
+
+def with_attribute(*args, **attr_dict):
+    """
+    Helper to create a validating parse action to be used with start
+    tags created with :class:`make_xml_tags` or
+    :class:`make_html_tags`. Use ``with_attribute`` to qualify
+    a starting tag with a required attribute value, to avoid false
+    matches on common tags such as ``<TD>`` or ``<DIV>``.
+
+    Call ``with_attribute`` with a series of attribute names and
+    values. Specify the list of filter attributes names and values as:
+
+    - keyword arguments, as in ``(align="right")``, or
+    - as an explicit dict with ``**`` operator, when an attribute
+      name is also a Python reserved word, as in ``**{"class":"Customer", "align":"right"}``
+    - a list of name-value tuples, as in ``(("ns1:class", "Customer"), ("ns2:align", "right"))``
+
+    For attribute names with a namespace prefix, you must use the second
+    form.  Attribute names are matched insensitive to upper/lower case.
+
+    If just testing for ``class`` (with or without a namespace), use
+    :class:`with_class`.
+
+    To verify that the attribute exists, but without specifying a value,
+    pass ``with_attribute.ANY_VALUE`` as the value.
+
+    Example::
+
+        html = '''
+            <div>
+            Some text
+            <div type="grid">1 4 0 1 0</div>
+            <div type="graph">1,3 2,3 1,1</div>
+            <div>this has no type</div>
+            </div>
+
+        '''
+        div,div_end = make_html_tags("div")
+
+        # only match div tag having a type attribute with value "grid"
+        div_grid = div().set_parse_action(with_attribute(type="grid"))
+        grid_expr = div_grid + SkipTo(div | div_end)("body")
+        for grid_header in grid_expr.search_string(html):
+            print(grid_header.body)
+
+        # construct a match with any div tag having a type attribute, regardless of the value
+        div_any_type = div().set_parse_action(with_attribute(type=with_attribute.ANY_VALUE))
+        div_expr = div_any_type + SkipTo(div | div_end)("body")
+        for div_header in div_expr.search_string(html):
+            print(div_header.body)
+
+    prints::
+
+        1 4 0 1 0
+
+        1 4 0 1 0
+        1,3 2,3 1,1
+    """
+    if args:
+        attrs = args[:]
+    else:
+        attrs = attr_dict.items()
+    attrs = [(k, v) for k, v in attrs]
+
+    def pa(s, l, tokens):
+        for attrName, attrValue in attrs:
+            if attrName not in tokens:
+                raise ParseException(s, l, "no matching attribute " + attrName)
+            if attrValue != with_attribute.ANY_VALUE and tokens[attrName] != attrValue:
+                raise ParseException(
+                    s,
+                    l,
+                    f"attribute {attrName!r} has value {tokens[attrName]!r}, must be {attrValue!r}",
+                )
+
+    return pa
+
+
+with_attribute.ANY_VALUE = object()  # type: ignore [attr-defined]
+
+
+def with_class(classname, namespace=""):
+    """
+    Simplified version of :class:`with_attribute` when
+    matching on a div class - made difficult because ``class`` is
+    a reserved word in Python.
+
+    Example::
+
+        html = '''
+            <div>
+            Some text
+            <div class="grid">1 4 0 1 0</div>
+            <div class="graph">1,3 2,3 1,1</div>
+            <div>this &lt;div&gt; has no class</div>
+            </div>
+
+        '''
+        div,div_end = make_html_tags("div")
+        div_grid = div().set_parse_action(with_class("grid"))
+
+        grid_expr = div_grid + SkipTo(div | div_end)("body")
+        for grid_header in grid_expr.search_string(html):
+            print(grid_header.body)
+
+        div_any_type = div().set_parse_action(with_class(withAttribute.ANY_VALUE))
+        div_expr = div_any_type + SkipTo(div | div_end)("body")
+        for div_header in div_expr.search_string(html):
+            print(div_header.body)
+
+    prints::
+
+        1 4 0 1 0
+
+        1 4 0 1 0
+        1,3 2,3 1,1
+    """
+    classattr = f"{namespace}:class" if namespace else "class"
+    return with_attribute(**{classattr: classname})
+
+
+# pre-PEP8 compatibility symbols
+# fmt: off
+@replaced_by_pep8(replace_with)
+def replaceWith(): ...
+
+@replaced_by_pep8(remove_quotes)
+def removeQuotes(): ...
+
+@replaced_by_pep8(with_attribute)
+def withAttribute(): ...
+
+@replaced_by_pep8(with_class)
+def withClass(): ...
+
+@replaced_by_pep8(match_only_at_col)
+def matchOnlyAtCol(): ...
+
+# fmt: on

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/common.py

@@ -0,0 +1,432 @@
+# common.py
+from .core import *
+from .helpers import DelimitedList, any_open_tag, any_close_tag
+from datetime import datetime
+
+
+# some other useful expressions - using lower-case class name since we are really using this as a namespace
+class pyparsing_common:
+    """Here are some common low-level expressions that may be useful in
+    jump-starting parser development:
+
+    - numeric forms (:class:`integers<integer>`, :class:`reals<real>`,
+      :class:`scientific notation<sci_real>`)
+    - common :class:`programming identifiers<identifier>`
+    - network addresses (:class:`MAC<mac_address>`,
+      :class:`IPv4<ipv4_address>`, :class:`IPv6<ipv6_address>`)
+    - ISO8601 :class:`dates<iso8601_date>` and
+      :class:`datetime<iso8601_datetime>`
+    - :class:`UUID<uuid>`
+    - :class:`comma-separated list<comma_separated_list>`
+    - :class:`url`
+
+    Parse actions:
+
+    - :class:`convert_to_integer`
+    - :class:`convert_to_float`
+    - :class:`convert_to_date`
+    - :class:`convert_to_datetime`
+    - :class:`strip_html_tags`
+    - :class:`upcase_tokens`
+    - :class:`downcase_tokens`
+
+    Example::
+
+        pyparsing_common.number.run_tests('''
+            # any int or real number, returned as the appropriate type
+            100
+            -100
+            +100
+            3.14159
+            6.02e23
+            1e-12
+            ''')
+
+        pyparsing_common.fnumber.run_tests('''
+            # any int or real number, returned as float
+            100
+            -100
+            +100
+            3.14159
+            6.02e23
+            1e-12
+            ''')
+
+        pyparsing_common.hex_integer.run_tests('''
+            # hex numbers
+            100
+            FF
+            ''')
+
+        pyparsing_common.fraction.run_tests('''
+            # fractions
+            1/2
+            -3/4
+            ''')
+
+        pyparsing_common.mixed_integer.run_tests('''
+            # mixed fractions
+            1
+            1/2
+            -3/4
+            1-3/4
+            ''')
+
+        import uuid
+        pyparsing_common.uuid.set_parse_action(token_map(uuid.UUID))
+        pyparsing_common.uuid.run_tests('''
+            # uuid
+            12345678-1234-5678-1234-567812345678
+            ''')
+
+    prints::
+
+        # any int or real number, returned as the appropriate type
+        100
+        [100]
+
+        -100
+        [-100]
+
+        +100
+        [100]
+
+        3.14159
+        [3.14159]
+
+        6.02e23
+        [6.02e+23]
+
+        1e-12
+        [1e-12]
+
+        # any int or real number, returned as float
+        100
+        [100.0]
+
+        -100
+        [-100.0]
+
+        +100
+        [100.0]
+
+        3.14159
+        [3.14159]
+
+        6.02e23
+        [6.02e+23]
+
+        1e-12
+        [1e-12]
+
+        # hex numbers
+        100
+        [256]
+
+        FF
+        [255]
+
+        # fractions
+        1/2
+        [0.5]
+
+        -3/4
+        [-0.75]
+
+        # mixed fractions
+        1
+        [1]
+
+        1/2
+        [0.5]
+
+        -3/4
+        [-0.75]
+
+        1-3/4
+        [1.75]
+
+        # uuid
+        12345678-1234-5678-1234-567812345678
+        [UUID('12345678-1234-5678-1234-567812345678')]
+    """
+
+    convert_to_integer = token_map(int)
+    """
+    Parse action for converting parsed integers to Python int
+    """
+
+    convert_to_float = token_map(float)
+    """
+    Parse action for converting parsed numbers to Python float
+    """
+
+    integer = Word(nums).set_name("integer").set_parse_action(convert_to_integer)
+    """expression that parses an unsigned integer, returns an int"""
+
+    hex_integer = (
+        Word(hexnums).set_name("hex integer").set_parse_action(token_map(int, 16))
+    )
+    """expression that parses a hexadecimal integer, returns an int"""
+
+    signed_integer = (
+        Regex(r"[+-]?\d+")
+        .set_name("signed integer")
+        .set_parse_action(convert_to_integer)
+    )
+    """expression that parses an integer with optional leading sign, returns an int"""
+
+    fraction = (
+        signed_integer().set_parse_action(convert_to_float)
+        + "/"
+        + signed_integer().set_parse_action(convert_to_float)
+    ).set_name("fraction")
+    """fractional expression of an integer divided by an integer, returns a float"""
+    fraction.add_parse_action(lambda tt: tt[0] / tt[-1])
+
+    mixed_integer = (
+        fraction | signed_integer + Opt(Opt("-").suppress() + fraction)
+    ).set_name("fraction or mixed integer-fraction")
+    """mixed integer of the form 'integer - fraction', with optional leading integer, returns float"""
+    mixed_integer.add_parse_action(sum)
+
+    real = (
+        Regex(r"[+-]?(?:\d+\.\d*|\.\d+)")
+        .set_name("real number")
+        .set_parse_action(convert_to_float)
+    )
+    """expression that parses a floating point number and returns a float"""
+
+    sci_real = (
+        Regex(r"[+-]?(?:\d+(?:[eE][+-]?\d+)|(?:\d+\.\d*|\.\d+)(?:[eE][+-]?\d+)?)")
+        .set_name("real number with scientific notation")
+        .set_parse_action(convert_to_float)
+    )
+    """expression that parses a floating point number with optional
+    scientific notation and returns a float"""
+
+    # streamlining this expression makes the docs nicer-looking
+    number = (sci_real | real | signed_integer).setName("number").streamline()
+    """any numeric expression, returns the corresponding Python type"""
+
+    fnumber = (
+        Regex(r"[+-]?\d+\.?\d*([eE][+-]?\d+)?")
+        .set_name("fnumber")
+        .set_parse_action(convert_to_float)
+    )
+    """any int or real number, returned as float"""
+
+    identifier = Word(identchars, identbodychars).set_name("identifier")
+    """typical code identifier (leading alpha or '_', followed by 0 or more alphas, nums, or '_')"""
+
+    ipv4_address = Regex(
+        r"(25[0-5]|2[0-4][0-9]|1?[0-9]{1,2})(\.(25[0-5]|2[0-4][0-9]|1?[0-9]{1,2})){3}"
+    ).set_name("IPv4 address")
+    "IPv4 address (``0.0.0.0 - 255.255.255.255``)"
+
+    _ipv6_part = Regex(r"[0-9a-fA-F]{1,4}").set_name("hex_integer")
+    _full_ipv6_address = (_ipv6_part + (":" + _ipv6_part) * 7).set_name(
+        "full IPv6 address"
+    )
+    _short_ipv6_address = (
+        Opt(_ipv6_part + (":" + _ipv6_part) * (0, 6))
+        + "::"
+        + Opt(_ipv6_part + (":" + _ipv6_part) * (0, 6))
+    ).set_name("short IPv6 address")
+    _short_ipv6_address.add_condition(
+        lambda t: sum(1 for tt in t if pyparsing_common._ipv6_part.matches(tt)) < 8
+    )
+    _mixed_ipv6_address = ("::ffff:" + ipv4_address).set_name("mixed IPv6 address")
+    ipv6_address = Combine(
+        (_full_ipv6_address | _mixed_ipv6_address | _short_ipv6_address).set_name(
+            "IPv6 address"
+        )
+    ).set_name("IPv6 address")
+    "IPv6 address (long, short, or mixed form)"
+
+    mac_address = Regex(
+        r"[0-9a-fA-F]{2}([:.-])[0-9a-fA-F]{2}(?:\1[0-9a-fA-F]{2}){4}"
+    ).set_name("MAC address")
+    "MAC address xx:xx:xx:xx:xx (may also have '-' or '.' delimiters)"
+
+    @staticmethod
+    def convert_to_date(fmt: str = "%Y-%m-%d"):
+        """
+        Helper to create a parse action for converting parsed date string to Python datetime.date
+
+        Params -
+        - fmt - format to be passed to datetime.strptime (default= ``"%Y-%m-%d"``)
+
+        Example::
+
+            date_expr = pyparsing_common.iso8601_date.copy()
+            date_expr.set_parse_action(pyparsing_common.convert_to_date())
+            print(date_expr.parse_string("1999-12-31"))
+
+        prints::
+
+            [datetime.date(1999, 12, 31)]
+        """
+
+        def cvt_fn(ss, ll, tt):
+            try:
+                return datetime.strptime(tt[0], fmt).date()
+            except ValueError as ve:
+                raise ParseException(ss, ll, str(ve))
+
+        return cvt_fn
+
+    @staticmethod
+    def convert_to_datetime(fmt: str = "%Y-%m-%dT%H:%M:%S.%f"):
+        """Helper to create a parse action for converting parsed
+        datetime string to Python datetime.datetime
+
+        Params -
+        - fmt - format to be passed to datetime.strptime (default= ``"%Y-%m-%dT%H:%M:%S.%f"``)
+
+        Example::
+
+            dt_expr = pyparsing_common.iso8601_datetime.copy()
+            dt_expr.set_parse_action(pyparsing_common.convert_to_datetime())
+            print(dt_expr.parse_string("1999-12-31T23:59:59.999"))
+
+        prints::
+
+            [datetime.datetime(1999, 12, 31, 23, 59, 59, 999000)]
+        """
+
+        def cvt_fn(s, l, t):
+            try:
+                return datetime.strptime(t[0], fmt)
+            except ValueError as ve:
+                raise ParseException(s, l, str(ve))
+
+        return cvt_fn
+
+    iso8601_date = Regex(
+        r"(?P<year>\d{4})(?:-(?P<month>\d\d)(?:-(?P<day>\d\d))?)?"
+    ).set_name("ISO8601 date")
+    "ISO8601 date (``yyyy-mm-dd``)"
+
+    iso8601_datetime = Regex(
+        r"(?P<year>\d{4})-(?P<month>\d\d)-(?P<day>\d\d)[T ](?P<hour>\d\d):(?P<minute>\d\d)(:(?P<second>\d\d(\.\d*)?)?)?(?P<tz>Z|[+-]\d\d:?\d\d)?"
+    ).set_name("ISO8601 datetime")
+    "ISO8601 datetime (``yyyy-mm-ddThh:mm:ss.s(Z|+-00:00)``) - trailing seconds, milliseconds, and timezone optional; accepts separating ``'T'`` or ``' '``"
+
+    uuid = Regex(r"[0-9a-fA-F]{8}(-[0-9a-fA-F]{4}){3}-[0-9a-fA-F]{12}").set_name("UUID")
+    "UUID (``xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx``)"
+
+    _html_stripper = any_open_tag.suppress() | any_close_tag.suppress()
+
+    @staticmethod
+    def strip_html_tags(s: str, l: int, tokens: ParseResults):
+        """Parse action to remove HTML tags from web page HTML source
+
+        Example::
+
+            # strip HTML links from normal text
+            text = '<td>More info at the <a href="https://github.com/pyparsing/pyparsing/wiki">pyparsing</a> wiki page</td>'
+            td, td_end = make_html_tags("TD")
+            table_text = td + SkipTo(td_end).set_parse_action(pyparsing_common.strip_html_tags)("body") + td_end
+            print(table_text.parse_string(text).body)
+
+        Prints::
+
+            More info at the pyparsing wiki page
+        """
+        return pyparsing_common._html_stripper.transform_string(tokens[0])
+
+    _commasepitem = (
+        Combine(
+            OneOrMore(
+                ~Literal(",")
+                + ~LineEnd()
+                + Word(printables, exclude_chars=",")
+                + Opt(White(" \t") + ~FollowedBy(LineEnd() | ","))
+            )
+        )
+        .streamline()
+        .set_name("commaItem")
+    )
+    comma_separated_list = DelimitedList(
+        Opt(quoted_string.copy() | _commasepitem, default="")
+    ).set_name("comma separated list")
+    """Predefined expression of 1 or more printable words or quoted strings, separated by commas."""
+
+    upcase_tokens = staticmethod(token_map(lambda t: t.upper()))
+    """Parse action to convert tokens to upper case."""
+
+    downcase_tokens = staticmethod(token_map(lambda t: t.lower()))
+    """Parse action to convert tokens to lower case."""
+
+    # fmt: off
+    url = Regex(
+        # https://mathiasbynens.be/demo/url-regex
+        # https://gist.github.com/dperini/729294
+        r"(?P<url>" +
+        # protocol identifier (optional)
+        # short syntax // still required
+        r"(?:(?:(?P<scheme>https?|ftp):)?\/\/)" +
+        # user:pass BasicAuth (optional)
+        r"(?:(?P<auth>\S+(?::\S*)?)@)?" +
+        r"(?P<host>" +
+        # IP address exclusion
+        # private & local networks
+        r"(?!(?:10|127)(?:\.\d{1,3}){3})" +
+        r"(?!(?:169\.254|192\.168)(?:\.\d{1,3}){2})" +
+        r"(?!172\.(?:1[6-9]|2\d|3[0-1])(?:\.\d{1,3}){2})" +
+        # IP address dotted notation octets
+        # excludes loopback network 0.0.0.0
+        # excludes reserved space >= 224.0.0.0
+        # excludes network & broadcast addresses
+        # (first & last IP address of each class)
+        r"(?:[1-9]\d?|1\d\d|2[01]\d|22[0-3])" +
+        r"(?:\.(?:1?\d{1,2}|2[0-4]\d|25[0-5])){2}" +
+        r"(?:\.(?:[1-9]\d?|1\d\d|2[0-4]\d|25[0-4]))" +
+        r"|" +
+        # host & domain names, may end with dot
+        # can be replaced by a shortest alternative
+        # (?![-_])(?:[-\w\u00a1-\uffff]{0,63}[^-_]\.)+
+        r"(?:" +
+        r"(?:" +
+        r"[a-z0-9\u00a1-\uffff]" +
+        r"[a-z0-9\u00a1-\uffff_-]{0,62}" +
+        r")?" +
+        r"[a-z0-9\u00a1-\uffff]\." +
+        r")+" +
+        # TLD identifier name, may end with dot
+        r"(?:[a-z\u00a1-\uffff]{2,}\.?)" +
+        r")" +
+        # port number (optional)
+        r"(:(?P<port>\d{2,5}))?" +
+        # resource path (optional)
+        r"(?P<path>\/[^?# ]*)?" +
+        # query string (optional)
+        r"(\?(?P<query>[^#]*))?" +
+        # fragment (optional)
+        r"(#(?P<fragment>\S*))?" +
+        r")"
+    ).set_name("url")
+    """URL (http/https/ftp scheme)"""
+    # fmt: on
+
+    # pre-PEP8 compatibility names
+    convertToInteger = convert_to_integer
+    """Deprecated - use :class:`convert_to_integer`"""
+    convertToFloat = convert_to_float
+    """Deprecated - use :class:`convert_to_float`"""
+    convertToDate = convert_to_date
+    """Deprecated - use :class:`convert_to_date`"""
+    convertToDatetime = convert_to_datetime
+    """Deprecated - use :class:`convert_to_datetime`"""
+    stripHTMLTags = strip_html_tags
+    """Deprecated - use :class:`strip_html_tags`"""
+    upcaseTokens = upcase_tokens
+    """Deprecated - use :class:`upcase_tokens`"""
+    downcaseTokens = downcase_tokens
+    """Deprecated - use :class:`downcase_tokens`"""
+
+
+_builtin_exprs = [
+    v for v in vars(pyparsing_common).values() if isinstance(v, ParserElement)
+]

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/diagram/__init__.py

@@ -0,0 +1,656 @@
+# mypy: ignore-errors
+import railroad
+from pip._vendor import pyparsing
+import typing
+from typing import (
+    List,
+    NamedTuple,
+    Generic,
+    TypeVar,
+    Dict,
+    Callable,
+    Set,
+    Iterable,
+)
+from jinja2 import Template
+from io import StringIO
+import inspect
+
+
+jinja2_template_source = """\
+{% if not embed %}
+<!DOCTYPE html>
+<html>
+<head>
+{% endif %}
+    {% if not head %}
+        <style>
+            .railroad-heading {
+                font-family: monospace;
+            }
+        </style>
+    {% else %}
+        {{ head | safe }}
+    {% endif %}
+{% if not embed %}
+</head>
+<body>
+{% endif %}
+{{ body | safe }}
+{% for diagram in diagrams %}
+    <div class="railroad-group">
+        <h1 class="railroad-heading">{{ diagram.title }}</h1>
+        <div class="railroad-description">{{ diagram.text }}</div>
+        <div class="railroad-svg">
+            {{ diagram.svg }}
+        </div>
+    </div>
+{% endfor %}
+{% if not embed %}
+</body>
+</html>
+{% endif %}
+"""
+
+template = Template(jinja2_template_source)
+
+# Note: ideally this would be a dataclass, but we're supporting Python 3.5+ so we can't do this yet
+NamedDiagram = NamedTuple(
+    "NamedDiagram",
+    [("name", str), ("diagram", typing.Optional[railroad.DiagramItem]), ("index", int)],
+)
+"""
+A simple structure for associating a name with a railroad diagram
+"""
+
+T = TypeVar("T")
+
+
+class EachItem(railroad.Group):
+    """
+    Custom railroad item to compose a:
+    - Group containing a
+      - OneOrMore containing a
+        - Choice of the elements in the Each
+    with the group label indicating that all must be matched
+    """
+
+    all_label = "[ALL]"
+
+    def __init__(self, *items):
+        choice_item = railroad.Choice(len(items) - 1, *items)
+        one_or_more_item = railroad.OneOrMore(item=choice_item)
+        super().__init__(one_or_more_item, label=self.all_label)
+
+
+class AnnotatedItem(railroad.Group):
+    """
+    Simple subclass of Group that creates an annotation label
+    """
+
+    def __init__(self, label: str, item):
+        super().__init__(item=item, label="[{}]".format(label) if label else label)
+
+
+class EditablePartial(Generic[T]):
+    """
+    Acts like a functools.partial, but can be edited. In other words, it represents a type that hasn't yet been
+    constructed.
+    """
+
+    # We need this here because the railroad constructors actually transform the data, so can't be called until the
+    # entire tree is assembled
+
+    def __init__(self, func: Callable[..., T], args: list, kwargs: dict):
+        self.func = func
+        self.args = args
+        self.kwargs = kwargs
+
+    @classmethod
+    def from_call(cls, func: Callable[..., T], *args, **kwargs) -> "EditablePartial[T]":
+        """
+        If you call this function in the same way that you would call the constructor, it will store the arguments
+        as you expect. For example EditablePartial.from_call(Fraction, 1, 3)() == Fraction(1, 3)
+        """
+        return EditablePartial(func=func, args=list(args), kwargs=kwargs)
+
+    @property
+    def name(self):
+        return self.kwargs["name"]
+
+    def __call__(self) -> T:
+        """
+        Evaluate the partial and return the result
+        """
+        args = self.args.copy()
+        kwargs = self.kwargs.copy()
+
+        # This is a helpful hack to allow you to specify varargs parameters (e.g. *args) as keyword args (e.g.
+        # args=['list', 'of', 'things'])
+        arg_spec = inspect.getfullargspec(self.func)
+        if arg_spec.varargs in self.kwargs:
+            args += kwargs.pop(arg_spec.varargs)
+
+        return self.func(*args, **kwargs)
+
+
+def railroad_to_html(diagrams: List[NamedDiagram], embed=False, **kwargs) -> str:
+    """
+    Given a list of NamedDiagram, produce a single HTML string that visualises those diagrams
+    :params kwargs: kwargs to be passed in to the template
+    """
+    data = []
+    for diagram in diagrams:
+        if diagram.diagram is None:
+            continue
+        io = StringIO()
+        try:
+            css = kwargs.get('css')
+            diagram.diagram.writeStandalone(io.write, css=css)
+        except AttributeError:
+            diagram.diagram.writeSvg(io.write)
+        title = diagram.name
+        if diagram.index == 0:
+            title += " (root)"
+        data.append({"title": title, "text": "", "svg": io.getvalue()})
+
+    return template.render(diagrams=data, embed=embed, **kwargs)
+
+
+def resolve_partial(partial: "EditablePartial[T]") -> T:
+    """
+    Recursively resolves a collection of Partials into whatever type they are
+    """
+    if isinstance(partial, EditablePartial):
+        partial.args = resolve_partial(partial.args)
+        partial.kwargs = resolve_partial(partial.kwargs)
+        return partial()
+    elif isinstance(partial, list):
+        return [resolve_partial(x) for x in partial]
+    elif isinstance(partial, dict):
+        return {key: resolve_partial(x) for key, x in partial.items()}
+    else:
+        return partial
+
+
+def to_railroad(
+    element: pyparsing.ParserElement,
+    diagram_kwargs: typing.Optional[dict] = None,
+    vertical: int = 3,
+    show_results_names: bool = False,
+    show_groups: bool = False,
+) -> List[NamedDiagram]:
+    """
+    Convert a pyparsing element tree into a list of diagrams. This is the recommended entrypoint to diagram
+    creation if you want to access the Railroad tree before it is converted to HTML
+    :param element: base element of the parser being diagrammed
+    :param diagram_kwargs: kwargs to pass to the Diagram() constructor
+    :param vertical: (optional) - int - limit at which number of alternatives should be
+       shown vertically instead of horizontally
+    :param show_results_names - bool to indicate whether results name annotations should be
+       included in the diagram
+    :param show_groups - bool to indicate whether groups should be highlighted with an unlabeled
+       surrounding box
+    """
+    # Convert the whole tree underneath the root
+    lookup = ConverterState(diagram_kwargs=diagram_kwargs or {})
+    _to_diagram_element(
+        element,
+        lookup=lookup,
+        parent=None,
+        vertical=vertical,
+        show_results_names=show_results_names,
+        show_groups=show_groups,
+    )
+
+    root_id = id(element)
+    # Convert the root if it hasn't been already
+    if root_id in lookup:
+        if not element.customName:
+            lookup[root_id].name = ""
+        lookup[root_id].mark_for_extraction(root_id, lookup, force=True)
+
+    # Now that we're finished, we can convert from intermediate structures into Railroad elements
+    diags = list(lookup.diagrams.values())
+    if len(diags) > 1:
+        # collapse out duplicate diags with the same name
+        seen = set()
+        deduped_diags = []
+        for d in diags:
+            # don't extract SkipTo elements, they are uninformative as subdiagrams
+            if d.name == "...":
+                continue
+            if d.name is not None and d.name not in seen:
+                seen.add(d.name)
+                deduped_diags.append(d)
+        resolved = [resolve_partial(partial) for partial in deduped_diags]
+    else:
+        # special case - if just one diagram, always display it, even if
+        # it has no name
+        resolved = [resolve_partial(partial) for partial in diags]
+    return sorted(resolved, key=lambda diag: diag.index)
+
+
+def _should_vertical(
+    specification: int, exprs: Iterable[pyparsing.ParserElement]
+) -> bool:
+    """
+    Returns true if we should return a vertical list of elements
+    """
+    if specification is None:
+        return False
+    else:
+        return len(_visible_exprs(exprs)) >= specification
+
+
+class ElementState:
+    """
+    State recorded for an individual pyparsing Element
+    """
+
+    # Note: this should be a dataclass, but we have to support Python 3.5
+    def __init__(
+        self,
+        element: pyparsing.ParserElement,
+        converted: EditablePartial,
+        parent: EditablePartial,
+        number: int,
+        name: str = None,
+        parent_index: typing.Optional[int] = None,
+    ):
+        #: The pyparsing element that this represents
+        self.element: pyparsing.ParserElement = element
+        #: The name of the element
+        self.name: typing.Optional[str] = name
+        #: The output Railroad element in an unconverted state
+        self.converted: EditablePartial = converted
+        #: The parent Railroad element, which we store so that we can extract this if it's duplicated
+        self.parent: EditablePartial = parent
+        #: The order in which we found this element, used for sorting diagrams if this is extracted into a diagram
+        self.number: int = number
+        #: The index of this inside its parent
+        self.parent_index: typing.Optional[int] = parent_index
+        #: If true, we should extract this out into a subdiagram
+        self.extract: bool = False
+        #: If true, all of this element's children have been filled out
+        self.complete: bool = False
+
+    def mark_for_extraction(
+        self, el_id: int, state: "ConverterState", name: str = None, force: bool = False
+    ):
+        """
+        Called when this instance has been seen twice, and thus should eventually be extracted into a sub-diagram
+        :param el_id: id of the element
+        :param state: element/diagram state tracker
+        :param name: name to use for this element's text
+        :param force: If true, force extraction now, regardless of the state of this. Only useful for extracting the
+        root element when we know we're finished
+        """
+        self.extract = True
+
+        # Set the name
+        if not self.name:
+            if name:
+                # Allow forcing a custom name
+                self.name = name
+            elif self.element.customName:
+                self.name = self.element.customName
+            else:
+                self.name = ""
+
+        # Just because this is marked for extraction doesn't mean we can do it yet. We may have to wait for children
+        # to be added
+        # Also, if this is just a string literal etc, don't bother extracting it
+        if force or (self.complete and _worth_extracting(self.element)):
+            state.extract_into_diagram(el_id)
+
+
+class ConverterState:
+    """
+    Stores some state that persists between recursions into the element tree
+    """
+
+    def __init__(self, diagram_kwargs: typing.Optional[dict] = None):
+        #: A dictionary mapping ParserElements to state relating to them
+        self._element_diagram_states: Dict[int, ElementState] = {}
+        #: A dictionary mapping ParserElement IDs to subdiagrams generated from them
+        self.diagrams: Dict[int, EditablePartial[NamedDiagram]] = {}
+        #: The index of the next unnamed element
+        self.unnamed_index: int = 1
+        #: The index of the next element. This is used for sorting
+        self.index: int = 0
+        #: Shared kwargs that are used to customize the construction of diagrams
+        self.diagram_kwargs: dict = diagram_kwargs or {}
+        self.extracted_diagram_names: Set[str] = set()
+
+    def __setitem__(self, key: int, value: ElementState):
+        self._element_diagram_states[key] = value
+
+    def __getitem__(self, key: int) -> ElementState:
+        return self._element_diagram_states[key]
+
+    def __delitem__(self, key: int):
+        del self._element_diagram_states[key]
+
+    def __contains__(self, key: int):
+        return key in self._element_diagram_states
+
+    def generate_unnamed(self) -> int:
+        """
+        Generate a number used in the name of an otherwise unnamed diagram
+        """
+        self.unnamed_index += 1
+        return self.unnamed_index
+
+    def generate_index(self) -> int:
+        """
+        Generate a number used to index a diagram
+        """
+        self.index += 1
+        return self.index
+
+    def extract_into_diagram(self, el_id: int):
+        """
+        Used when we encounter the same token twice in the same tree. When this
+        happens, we replace all instances of that token with a terminal, and
+        create a new subdiagram for the token
+        """
+        position = self[el_id]
+
+        # Replace the original definition of this element with a regular block
+        if position.parent:
+            ret = EditablePartial.from_call(railroad.NonTerminal, text=position.name)
+            if "item" in position.parent.kwargs:
+                position.parent.kwargs["item"] = ret
+            elif "items" in position.parent.kwargs:
+                position.parent.kwargs["items"][position.parent_index] = ret
+
+        # If the element we're extracting is a group, skip to its content but keep the title
+        if position.converted.func == railroad.Group:
+            content = position.converted.kwargs["item"]
+        else:
+            content = position.converted
+
+        self.diagrams[el_id] = EditablePartial.from_call(
+            NamedDiagram,
+            name=position.name,
+            diagram=EditablePartial.from_call(
+                railroad.Diagram, content, **self.diagram_kwargs
+            ),
+            index=position.number,
+        )
+
+        del self[el_id]
+
+
+def _worth_extracting(element: pyparsing.ParserElement) -> bool:
+    """
+    Returns true if this element is worth having its own sub-diagram. Simply, if any of its children
+    themselves have children, then its complex enough to extract
+    """
+    children = element.recurse()
+    return any(child.recurse() for child in children)
+
+
+def _apply_diagram_item_enhancements(fn):
+    """
+    decorator to ensure enhancements to a diagram item (such as results name annotations)
+    get applied on return from _to_diagram_element (we do this since there are several
+    returns in _to_diagram_element)
+    """
+
+    def _inner(
+        element: pyparsing.ParserElement,
+        parent: typing.Optional[EditablePartial],
+        lookup: ConverterState = None,
+        vertical: int = None,
+        index: int = 0,
+        name_hint: str = None,
+        show_results_names: bool = False,
+        show_groups: bool = False,
+    ) -> typing.Optional[EditablePartial]:
+        ret = fn(
+            element,
+            parent,
+            lookup,
+            vertical,
+            index,
+            name_hint,
+            show_results_names,
+            show_groups,
+        )
+
+        # apply annotation for results name, if present
+        if show_results_names and ret is not None:
+            element_results_name = element.resultsName
+            if element_results_name:
+                # add "*" to indicate if this is a "list all results" name
+                element_results_name += "" if element.modalResults else "*"
+                ret = EditablePartial.from_call(
+                    railroad.Group, item=ret, label=element_results_name
+                )
+
+        return ret
+
+    return _inner
+
+
+def _visible_exprs(exprs: Iterable[pyparsing.ParserElement]):
+    non_diagramming_exprs = (
+        pyparsing.ParseElementEnhance,
+        pyparsing.PositionToken,
+        pyparsing.And._ErrorStop,
+    )
+    return [
+        e
+        for e in exprs
+        if not (e.customName or e.resultsName or isinstance(e, non_diagramming_exprs))
+    ]
+
+
+@_apply_diagram_item_enhancements
+def _to_diagram_element(
+    element: pyparsing.ParserElement,
+    parent: typing.Optional[EditablePartial],
+    lookup: ConverterState = None,
+    vertical: int = None,
+    index: int = 0,
+    name_hint: str = None,
+    show_results_names: bool = False,
+    show_groups: bool = False,
+) -> typing.Optional[EditablePartial]:
+    """
+    Recursively converts a PyParsing Element to a railroad Element
+    :param lookup: The shared converter state that keeps track of useful things
+    :param index: The index of this element within the parent
+    :param parent: The parent of this element in the output tree
+    :param vertical: Controls at what point we make a list of elements vertical. If this is an integer (the default),
+    it sets the threshold of the number of items before we go vertical. If True, always go vertical, if False, never
+    do so
+    :param name_hint: If provided, this will override the generated name
+    :param show_results_names: bool flag indicating whether to add annotations for results names
+    :returns: The converted version of the input element, but as a Partial that hasn't yet been constructed
+    :param show_groups: bool flag indicating whether to show groups using bounding box
+    """
+    exprs = element.recurse()
+    name = name_hint or element.customName or element.__class__.__name__
+
+    # Python's id() is used to provide a unique identifier for elements
+    el_id = id(element)
+
+    element_results_name = element.resultsName
+
+    # Here we basically bypass processing certain wrapper elements if they contribute nothing to the diagram
+    if not element.customName:
+        if isinstance(
+            element,
+            (
+                # pyparsing.TokenConverter,
+                # pyparsing.Forward,
+                pyparsing.Located,
+            ),
+        ):
+            # However, if this element has a useful custom name, and its child does not, we can pass it on to the child
+            if exprs:
+                if not exprs[0].customName:
+                    propagated_name = name
+                else:
+                    propagated_name = None
+
+                return _to_diagram_element(
+                    element.expr,
+                    parent=parent,
+                    lookup=lookup,
+                    vertical=vertical,
+                    index=index,
+                    name_hint=propagated_name,
+                    show_results_names=show_results_names,
+                    show_groups=show_groups,
+                )
+
+    # If the element isn't worth extracting, we always treat it as the first time we say it
+    if _worth_extracting(element):
+        if el_id in lookup:
+            # If we've seen this element exactly once before, we are only just now finding out that it's a duplicate,
+            # so we have to extract it into a new diagram.
+            looked_up = lookup[el_id]
+            looked_up.mark_for_extraction(el_id, lookup, name=name_hint)
+            ret = EditablePartial.from_call(railroad.NonTerminal, text=looked_up.name)
+            return ret
+
+        elif el_id in lookup.diagrams:
+            # If we have seen the element at least twice before, and have already extracted it into a subdiagram, we
+            # just put in a marker element that refers to the sub-diagram
+            ret = EditablePartial.from_call(
+                railroad.NonTerminal, text=lookup.diagrams[el_id].kwargs["name"]
+            )
+            return ret
+
+    # Recursively convert child elements
+    # Here we find the most relevant Railroad element for matching pyparsing Element
+    # We use ``items=[]`` here to hold the place for where the child elements will go once created
+    if isinstance(element, pyparsing.And):
+        # detect And's created with ``expr*N`` notation - for these use a OneOrMore with a repeat
+        # (all will have the same name, and resultsName)
+        if not exprs:
+            return None
+        if len(set((e.name, e.resultsName) for e in exprs)) == 1:
+            ret = EditablePartial.from_call(
+                railroad.OneOrMore, item="", repeat=str(len(exprs))
+            )
+        elif _should_vertical(vertical, exprs):
+            ret = EditablePartial.from_call(railroad.Stack, items=[])
+        else:
+            ret = EditablePartial.from_call(railroad.Sequence, items=[])
+    elif isinstance(element, (pyparsing.Or, pyparsing.MatchFirst)):
+        if not exprs:
+            return None
+        if _should_vertical(vertical, exprs):
+            ret = EditablePartial.from_call(railroad.Choice, 0, items=[])
+        else:
+            ret = EditablePartial.from_call(railroad.HorizontalChoice, items=[])
+    elif isinstance(element, pyparsing.Each):
+        if not exprs:
+            return None
+        ret = EditablePartial.from_call(EachItem, items=[])
+    elif isinstance(element, pyparsing.NotAny):
+        ret = EditablePartial.from_call(AnnotatedItem, label="NOT", item="")
+    elif isinstance(element, pyparsing.FollowedBy):
+        ret = EditablePartial.from_call(AnnotatedItem, label="LOOKAHEAD", item="")
+    elif isinstance(element, pyparsing.PrecededBy):
+        ret = EditablePartial.from_call(AnnotatedItem, label="LOOKBEHIND", item="")
+    elif isinstance(element, pyparsing.Group):
+        if show_groups:
+            ret = EditablePartial.from_call(AnnotatedItem, label="", item="")
+        else:
+            ret = EditablePartial.from_call(railroad.Group, label="", item="")
+    elif isinstance(element, pyparsing.TokenConverter):
+        label = type(element).__name__.lower()
+        if label == "tokenconverter":
+            ret = EditablePartial.from_call(railroad.Sequence, items=[])
+        else:
+            ret = EditablePartial.from_call(AnnotatedItem, label=label, item="")
+    elif isinstance(element, pyparsing.Opt):
+        ret = EditablePartial.from_call(railroad.Optional, item="")
+    elif isinstance(element, pyparsing.OneOrMore):
+        ret = EditablePartial.from_call(railroad.OneOrMore, item="")
+    elif isinstance(element, pyparsing.ZeroOrMore):
+        ret = EditablePartial.from_call(railroad.ZeroOrMore, item="")
+    elif isinstance(element, pyparsing.Group):
+        ret = EditablePartial.from_call(
+            railroad.Group, item=None, label=element_results_name
+        )
+    elif isinstance(element, pyparsing.Empty) and not element.customName:
+        # Skip unnamed "Empty" elements
+        ret = None
+    elif isinstance(element, pyparsing.ParseElementEnhance):
+        ret = EditablePartial.from_call(railroad.Sequence, items=[])
+    elif len(exprs) > 0 and not element_results_name:
+        ret = EditablePartial.from_call(railroad.Group, item="", label=name)
+    elif len(exprs) > 0:
+        ret = EditablePartial.from_call(railroad.Sequence, items=[])
+    else:
+        terminal = EditablePartial.from_call(railroad.Terminal, element.defaultName)
+        ret = terminal
+
+    if ret is None:
+        return
+
+    # Indicate this element's position in the tree so we can extract it if necessary
+    lookup[el_id] = ElementState(
+        element=element,
+        converted=ret,
+        parent=parent,
+        parent_index=index,
+        number=lookup.generate_index(),
+    )
+    if element.customName:
+        lookup[el_id].mark_for_extraction(el_id, lookup, element.customName)
+
+    i = 0
+    for expr in exprs:
+        # Add a placeholder index in case we have to extract the child before we even add it to the parent
+        if "items" in ret.kwargs:
+            ret.kwargs["items"].insert(i, None)
+
+        item = _to_diagram_element(
+            expr,
+            parent=ret,
+            lookup=lookup,
+            vertical=vertical,
+            index=i,
+            show_results_names=show_results_names,
+            show_groups=show_groups,
+        )
+
+        # Some elements don't need to be shown in the diagram
+        if item is not None:
+            if "item" in ret.kwargs:
+                ret.kwargs["item"] = item
+            elif "items" in ret.kwargs:
+                # If we've already extracted the child, don't touch this index, since it's occupied by a nonterminal
+                ret.kwargs["items"][i] = item
+                i += 1
+        elif "items" in ret.kwargs:
+            # If we're supposed to skip this element, remove it from the parent
+            del ret.kwargs["items"][i]
+
+    # If all this items children are none, skip this item
+    if ret and (
+        ("items" in ret.kwargs and len(ret.kwargs["items"]) == 0)
+        or ("item" in ret.kwargs and ret.kwargs["item"] is None)
+    ):
+        ret = EditablePartial.from_call(railroad.Terminal, name)
+
+    # Mark this element as "complete", ie it has all of its children
+    if el_id in lookup:
+        lookup[el_id].complete = True
+
+    if el_id in lookup and lookup[el_id].extract and lookup[el_id].complete:
+        lookup.extract_into_diagram(el_id)
+        if ret is not None:
+            ret = EditablePartial.from_call(
+                railroad.NonTerminal, text=lookup.diagrams[el_id].kwargs["name"]
+            )
+
+    return ret

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/exceptions.py

@@ -0,0 +1,299 @@
+# exceptions.py
+
+import re
+import sys
+import typing
+
+from .util import (
+    col,
+    line,
+    lineno,
+    _collapse_string_to_ranges,
+    replaced_by_pep8,
+)
+from .unicode import pyparsing_unicode as ppu
+
+
+class ExceptionWordUnicode(ppu.Latin1, ppu.LatinA, ppu.LatinB, ppu.Greek, ppu.Cyrillic):
+    pass
+
+
+_extract_alphanums = _collapse_string_to_ranges(ExceptionWordUnicode.alphanums)
+_exception_word_extractor = re.compile("([" + _extract_alphanums + "]{1,16})|.")
+
+
+class ParseBaseException(Exception):
+    """base exception class for all parsing runtime exceptions"""
+
+    loc: int
+    msg: str
+    pstr: str
+    parser_element: typing.Any  # "ParserElement"
+    args: typing.Tuple[str, int, typing.Optional[str]]
+
+    __slots__ = (
+        "loc",
+        "msg",
+        "pstr",
+        "parser_element",
+        "args",
+    )
+
+    # Performance tuning: we construct a *lot* of these, so keep this
+    # constructor as small and fast as possible
+    def __init__(
+        self,
+        pstr: str,
+        loc: int = 0,
+        msg: typing.Optional[str] = None,
+        elem=None,
+    ):
+        self.loc = loc
+        if msg is None:
+            self.msg = pstr
+            self.pstr = ""
+        else:
+            self.msg = msg
+            self.pstr = pstr
+        self.parser_element = elem
+        self.args = (pstr, loc, msg)
+
+    @staticmethod
+    def explain_exception(exc, depth=16):
+        """
+        Method to take an exception and translate the Python internal traceback into a list
+        of the pyparsing expressions that caused the exception to be raised.
+
+        Parameters:
+
+        - exc - exception raised during parsing (need not be a ParseException, in support
+          of Python exceptions that might be raised in a parse action)
+        - depth (default=16) - number of levels back in the stack trace to list expression
+          and function names; if None, the full stack trace names will be listed; if 0, only
+          the failing input line, marker, and exception string will be shown
+
+        Returns a multi-line string listing the ParserElements and/or function names in the
+        exception's stack trace.
+        """
+        import inspect
+        from .core import ParserElement
+
+        if depth is None:
+            depth = sys.getrecursionlimit()
+        ret = []
+        if isinstance(exc, ParseBaseException):
+            ret.append(exc.line)
+            ret.append(" " * (exc.column - 1) + "^")
+        ret.append(f"{type(exc).__name__}: {exc}")
+
+        if depth > 0:
+            callers = inspect.getinnerframes(exc.__traceback__, context=depth)
+            seen = set()
+            for i, ff in enumerate(callers[-depth:]):
+                frm = ff[0]
+
+                f_self = frm.f_locals.get("self", None)
+                if isinstance(f_self, ParserElement):
+                    if not frm.f_code.co_name.startswith(
+                        ("parseImpl", "_parseNoCache")
+                    ):
+                        continue
+                    if id(f_self) in seen:
+                        continue
+                    seen.add(id(f_self))
+
+                    self_type = type(f_self)
+                    ret.append(
+                        f"{self_type.__module__}.{self_type.__name__} - {f_self}"
+                    )
+
+                elif f_self is not None:
+                    self_type = type(f_self)
+                    ret.append(f"{self_type.__module__}.{self_type.__name__}")
+
+                else:
+                    code = frm.f_code
+                    if code.co_name in ("wrapper", "<module>"):
+                        continue
+
+                    ret.append(code.co_name)
+
+                depth -= 1
+                if not depth:
+                    break
+
+        return "\n".join(ret)
+
+    @classmethod
+    def _from_exception(cls, pe):
+        """
+        internal factory method to simplify creating one type of ParseException
+        from another - avoids having __init__ signature conflicts among subclasses
+        """
+        return cls(pe.pstr, pe.loc, pe.msg, pe.parser_element)
+
+    @property
+    def line(self) -> str:
+        """
+        Return the line of text where the exception occurred.
+        """
+        return line(self.loc, self.pstr)
+
+    @property
+    def lineno(self) -> int:
+        """
+        Return the 1-based line number of text where the exception occurred.
+        """
+        return lineno(self.loc, self.pstr)
+
+    @property
+    def col(self) -> int:
+        """
+        Return the 1-based column on the line of text where the exception occurred.
+        """
+        return col(self.loc, self.pstr)
+
+    @property
+    def column(self) -> int:
+        """
+        Return the 1-based column on the line of text where the exception occurred.
+        """
+        return col(self.loc, self.pstr)
+
+    # pre-PEP8 compatibility
+    @property
+    def parserElement(self):
+        return self.parser_element
+
+    @parserElement.setter
+    def parserElement(self, elem):
+        self.parser_element = elem
+
+    def __str__(self) -> str:
+        if self.pstr:
+            if self.loc >= len(self.pstr):
+                foundstr = ", found end of text"
+            else:
+                # pull out next word at error location
+                found_match = _exception_word_extractor.match(self.pstr, self.loc)
+                if found_match is not None:
+                    found = found_match.group(0)
+                else:
+                    found = self.pstr[self.loc : self.loc + 1]
+                foundstr = (", found %r" % found).replace(r"\\", "\\")
+        else:
+            foundstr = ""
+        return f"{self.msg}{foundstr}  (at char {self.loc}), (line:{self.lineno}, col:{self.column})"
+
+    def __repr__(self):
+        return str(self)
+
+    def mark_input_line(
+        self, marker_string: typing.Optional[str] = None, *, markerString: str = ">!<"
+    ) -> str:
+        """
+        Extracts the exception line from the input string, and marks
+        the location of the exception with a special symbol.
+        """
+        markerString = marker_string if marker_string is not None else markerString
+        line_str = self.line
+        line_column = self.column - 1
+        if markerString:
+            line_str = "".join(
+                (line_str[:line_column], markerString, line_str[line_column:])
+            )
+        return line_str.strip()
+
+    def explain(self, depth=16) -> str:
+        """
+        Method to translate the Python internal traceback into a list
+        of the pyparsing expressions that caused the exception to be raised.
+
+        Parameters:
+
+        - depth (default=16) - number of levels back in the stack trace to list expression
+          and function names; if None, the full stack trace names will be listed; if 0, only
+          the failing input line, marker, and exception string will be shown
+
+        Returns a multi-line string listing the ParserElements and/or function names in the
+        exception's stack trace.
+
+        Example::
+
+            expr = pp.Word(pp.nums) * 3
+            try:
+                expr.parse_string("123 456 A789")
+            except pp.ParseException as pe:
+                print(pe.explain(depth=0))
+
+        prints::
+
+            123 456 A789
+                    ^
+            ParseException: Expected W:(0-9), found 'A'  (at char 8), (line:1, col:9)
+
+        Note: the diagnostic output will include string representations of the expressions
+        that failed to parse. These representations will be more helpful if you use `set_name` to
+        give identifiable names to your expressions. Otherwise they will use the default string
+        forms, which may be cryptic to read.
+
+        Note: pyparsing's default truncation of exception tracebacks may also truncate the
+        stack of expressions that are displayed in the ``explain`` output. To get the full listing
+        of parser expressions, you may have to set ``ParserElement.verbose_stacktrace = True``
+        """
+        return self.explain_exception(self, depth)
+
+    # fmt: off
+    @replaced_by_pep8(mark_input_line)
+    def markInputline(self): ...
+    # fmt: on
+
+
+class ParseException(ParseBaseException):
+    """
+    Exception thrown when a parse expression doesn't match the input string
+
+    Example::
+
+        try:
+            Word(nums).set_name("integer").parse_string("ABC")
+        except ParseException as pe:
+            print(pe)
+            print("column: {}".format(pe.column))
+
+    prints::
+
+       Expected integer (at char 0), (line:1, col:1)
+        column: 1
+
+    """
+
+
+class ParseFatalException(ParseBaseException):
+    """
+    User-throwable exception thrown when inconsistent parse content
+    is found; stops all parsing immediately
+    """
+
+
+class ParseSyntaxException(ParseFatalException):
+    """
+    Just like :class:`ParseFatalException`, but thrown internally
+    when an :class:`ErrorStop<And._ErrorStop>` ('-' operator) indicates
+    that parsing is to stop immediately because an unbacktrackable
+    syntax error has been found.
+    """
+
+
+class RecursiveGrammarException(Exception):
+    """
+    Exception thrown by :class:`ParserElement.validate` if the
+    grammar could be left-recursive; parser may need to enable
+    left recursion using :class:`ParserElement.enable_left_recursion<ParserElement.enable_left_recursion>`
+    """
+
+    def __init__(self, parseElementList):
+        self.parseElementTrace = parseElementList
+
+    def __str__(self) -> str:
+        return f"RecursiveGrammarException: {self.parseElementTrace}"

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/helpers.py

@@ -0,0 +1,1100 @@
+# helpers.py
+import html.entities
+import re
+import sys
+import typing
+
+from . import __diag__
+from .core import *
+from .util import (
+    _bslash,
+    _flatten,
+    _escape_regex_range_chars,
+    replaced_by_pep8,
+)
+
+
+#
+# global helpers
+#
+def counted_array(
+    expr: ParserElement,
+    int_expr: typing.Optional[ParserElement] = None,
+    *,
+    intExpr: typing.Optional[ParserElement] = None,
+) -> ParserElement:
+    """Helper to define a counted list of expressions.
+
+    This helper defines a pattern of the form::
+
+        integer expr expr expr...
+
+    where the leading integer tells how many expr expressions follow.
+    The matched tokens returns the array of expr tokens as a list - the
+    leading count token is suppressed.
+
+    If ``int_expr`` is specified, it should be a pyparsing expression
+    that produces an integer value.
+
+    Example::
+
+        counted_array(Word(alphas)).parse_string('2 ab cd ef')  # -> ['ab', 'cd']
+
+        # in this parser, the leading integer value is given in binary,
+        # '10' indicating that 2 values are in the array
+        binary_constant = Word('01').set_parse_action(lambda t: int(t[0], 2))
+        counted_array(Word(alphas), int_expr=binary_constant).parse_string('10 ab cd ef')  # -> ['ab', 'cd']
+
+        # if other fields must be parsed after the count but before the
+        # list items, give the fields results names and they will
+        # be preserved in the returned ParseResults:
+        count_with_metadata = integer + Word(alphas)("type")
+        typed_array = counted_array(Word(alphanums), int_expr=count_with_metadata)("items")
+        result = typed_array.parse_string("3 bool True True False")
+        print(result.dump())
+
+        # prints
+        # ['True', 'True', 'False']
+        # - items: ['True', 'True', 'False']
+        # - type: 'bool'
+    """
+    intExpr = intExpr or int_expr
+    array_expr = Forward()
+
+    def count_field_parse_action(s, l, t):
+        nonlocal array_expr
+        n = t[0]
+        array_expr <<= (expr * n) if n else Empty()
+        # clear list contents, but keep any named results
+        del t[:]
+
+    if intExpr is None:
+        intExpr = Word(nums).set_parse_action(lambda t: int(t[0]))
+    else:
+        intExpr = intExpr.copy()
+    intExpr.set_name("arrayLen")
+    intExpr.add_parse_action(count_field_parse_action, call_during_try=True)
+    return (intExpr + array_expr).set_name("(len) " + str(expr) + "...")
+
+
+def match_previous_literal(expr: ParserElement) -> ParserElement:
+    """Helper to define an expression that is indirectly defined from
+    the tokens matched in a previous expression, that is, it looks for
+    a 'repeat' of a previous expression.  For example::
+
+        first = Word(nums)
+        second = match_previous_literal(first)
+        match_expr = first + ":" + second
+
+    will match ``"1:1"``, but not ``"1:2"``.  Because this
+    matches a previous literal, will also match the leading
+    ``"1:1"`` in ``"1:10"``. If this is not desired, use
+    :class:`match_previous_expr`. Do *not* use with packrat parsing
+    enabled.
+    """
+    rep = Forward()
+
+    def copy_token_to_repeater(s, l, t):
+        if t:
+            if len(t) == 1:
+                rep << t[0]
+            else:
+                # flatten t tokens
+                tflat = _flatten(t.as_list())
+                rep << And(Literal(tt) for tt in tflat)
+        else:
+            rep << Empty()
+
+    expr.add_parse_action(copy_token_to_repeater, callDuringTry=True)
+    rep.set_name("(prev) " + str(expr))
+    return rep
+
+
+def match_previous_expr(expr: ParserElement) -> ParserElement:
+    """Helper to define an expression that is indirectly defined from
+    the tokens matched in a previous expression, that is, it looks for
+    a 'repeat' of a previous expression.  For example::
+
+        first = Word(nums)
+        second = match_previous_expr(first)
+        match_expr = first + ":" + second
+
+    will match ``"1:1"``, but not ``"1:2"``.  Because this
+    matches by expressions, will *not* match the leading ``"1:1"``
+    in ``"1:10"``; the expressions are evaluated first, and then
+    compared, so ``"1"`` is compared with ``"10"``. Do *not* use
+    with packrat parsing enabled.
+    """
+    rep = Forward()
+    e2 = expr.copy()
+    rep <<= e2
+
+    def copy_token_to_repeater(s, l, t):
+        matchTokens = _flatten(t.as_list())
+
+        def must_match_these_tokens(s, l, t):
+            theseTokens = _flatten(t.as_list())
+            if theseTokens != matchTokens:
+                raise ParseException(
+                    s, l, f"Expected {matchTokens}, found{theseTokens}"
+                )
+
+        rep.set_parse_action(must_match_these_tokens, callDuringTry=True)
+
+    expr.add_parse_action(copy_token_to_repeater, callDuringTry=True)
+    rep.set_name("(prev) " + str(expr))
+    return rep
+
+
+def one_of(
+    strs: Union[typing.Iterable[str], str],
+    caseless: bool = False,
+    use_regex: bool = True,
+    as_keyword: bool = False,
+    *,
+    useRegex: bool = True,
+    asKeyword: bool = False,
+) -> ParserElement:
+    """Helper to quickly define a set of alternative :class:`Literal` s,
+    and makes sure to do longest-first testing when there is a conflict,
+    regardless of the input order, but returns
+    a :class:`MatchFirst` for best performance.
+
+    Parameters:
+
+    - ``strs`` - a string of space-delimited literals, or a collection of
+      string literals
+    - ``caseless`` - treat all literals as caseless - (default= ``False``)
+    - ``use_regex`` - as an optimization, will
+      generate a :class:`Regex` object; otherwise, will generate
+      a :class:`MatchFirst` object (if ``caseless=True`` or ``as_keyword=True``, or if
+      creating a :class:`Regex` raises an exception) - (default= ``True``)
+    - ``as_keyword`` - enforce :class:`Keyword`-style matching on the
+      generated expressions - (default= ``False``)
+    - ``asKeyword`` and ``useRegex`` are retained for pre-PEP8 compatibility,
+      but will be removed in a future release
+
+    Example::
+
+        comp_oper = one_of("< = > <= >= !=")
+        var = Word(alphas)
+        number = Word(nums)
+        term = var | number
+        comparison_expr = term + comp_oper + term
+        print(comparison_expr.search_string("B = 12  AA=23 B<=AA AA>12"))
+
+    prints::
+
+        [['B', '=', '12'], ['AA', '=', '23'], ['B', '<=', 'AA'], ['AA', '>', '12']]
+    """
+    asKeyword = asKeyword or as_keyword
+    useRegex = useRegex and use_regex
+
+    if (
+        isinstance(caseless, str_type)
+        and __diag__.warn_on_multiple_string_args_to_oneof
+    ):
+        warnings.warn(
+            "More than one string argument passed to one_of, pass"
+            " choices as a list or space-delimited string",
+            stacklevel=2,
+        )
+
+    if caseless:
+        isequal = lambda a, b: a.upper() == b.upper()
+        masks = lambda a, b: b.upper().startswith(a.upper())
+        parseElementClass = CaselessKeyword if asKeyword else CaselessLiteral
+    else:
+        isequal = lambda a, b: a == b
+        masks = lambda a, b: b.startswith(a)
+        parseElementClass = Keyword if asKeyword else Literal
+
+    symbols: List[str] = []
+    if isinstance(strs, str_type):
+        strs = typing.cast(str, strs)
+        symbols = strs.split()
+    elif isinstance(strs, Iterable):
+        symbols = list(strs)
+    else:
+        raise TypeError("Invalid argument to one_of, expected string or iterable")
+    if not symbols:
+        return NoMatch()
+
+    # reorder given symbols to take care to avoid masking longer choices with shorter ones
+    # (but only if the given symbols are not just single characters)
+    if any(len(sym) > 1 for sym in symbols):
+        i = 0
+        while i < len(symbols) - 1:
+            cur = symbols[i]
+            for j, other in enumerate(symbols[i + 1 :]):
+                if isequal(other, cur):
+                    del symbols[i + j + 1]
+                    break
+                elif masks(cur, other):
+                    del symbols[i + j + 1]
+                    symbols.insert(i, other)
+                    break
+            else:
+                i += 1
+
+    if useRegex:
+        re_flags: int = re.IGNORECASE if caseless else 0
+
+        try:
+            if all(len(sym) == 1 for sym in symbols):
+                # symbols are just single characters, create range regex pattern
+                patt = f"[{''.join(_escape_regex_range_chars(sym) for sym in symbols)}]"
+            else:
+                patt = "|".join(re.escape(sym) for sym in symbols)
+
+            # wrap with \b word break markers if defining as keywords
+            if asKeyword:
+                patt = rf"\b(?:{patt})\b"
+
+            ret = Regex(patt, flags=re_flags).set_name(" | ".join(symbols))
+
+            if caseless:
+                # add parse action to return symbols as specified, not in random
+                # casing as found in input string
+                symbol_map = {sym.lower(): sym for sym in symbols}
+                ret.add_parse_action(lambda s, l, t: symbol_map[t[0].lower()])
+
+            return ret
+
+        except re.error:
+            warnings.warn(
+                "Exception creating Regex for one_of, building MatchFirst", stacklevel=2
+            )
+
+    # last resort, just use MatchFirst
+    return MatchFirst(parseElementClass(sym) for sym in symbols).set_name(
+        " | ".join(symbols)
+    )
+
+
+def dict_of(key: ParserElement, value: ParserElement) -> ParserElement:
+    """Helper to easily and clearly define a dictionary by specifying
+    the respective patterns for the key and value.  Takes care of
+    defining the :class:`Dict`, :class:`ZeroOrMore`, and
+    :class:`Group` tokens in the proper order.  The key pattern
+    can include delimiting markers or punctuation, as long as they are
+    suppressed, thereby leaving the significant key text.  The value
+    pattern can include named results, so that the :class:`Dict` results
+    can include named token fields.
+
+    Example::
+
+        text = "shape: SQUARE posn: upper left color: light blue texture: burlap"
+        attr_expr = (label + Suppress(':') + OneOrMore(data_word, stop_on=label).set_parse_action(' '.join))
+        print(attr_expr[1, ...].parse_string(text).dump())
+
+        attr_label = label
+        attr_value = Suppress(':') + OneOrMore(data_word, stop_on=label).set_parse_action(' '.join)
+
+        # similar to Dict, but simpler call format
+        result = dict_of(attr_label, attr_value).parse_string(text)
+        print(result.dump())
+        print(result['shape'])
+        print(result.shape)  # object attribute access works too
+        print(result.as_dict())
+
+    prints::
+
+        [['shape', 'SQUARE'], ['posn', 'upper left'], ['color', 'light blue'], ['texture', 'burlap']]
+        - color: 'light blue'
+        - posn: 'upper left'
+        - shape: 'SQUARE'
+        - texture: 'burlap'
+        SQUARE
+        SQUARE
+        {'color': 'light blue', 'shape': 'SQUARE', 'posn': 'upper left', 'texture': 'burlap'}
+    """
+    return Dict(OneOrMore(Group(key + value)))
+
+
+def original_text_for(
+    expr: ParserElement, as_string: bool = True, *, asString: bool = True
+) -> ParserElement:
+    """Helper to return the original, untokenized text for a given
+    expression.  Useful to restore the parsed fields of an HTML start
+    tag into the raw tag text itself, or to revert separate tokens with
+    intervening whitespace back to the original matching input text. By
+    default, returns a string containing the original parsed text.
+
+    If the optional ``as_string`` argument is passed as
+    ``False``, then the return value is
+    a :class:`ParseResults` containing any results names that
+    were originally matched, and a single token containing the original
+    matched text from the input string.  So if the expression passed to
+    :class:`original_text_for` contains expressions with defined
+    results names, you must set ``as_string`` to ``False`` if you
+    want to preserve those results name values.
+
+    The ``asString`` pre-PEP8 argument is retained for compatibility,
+    but will be removed in a future release.
+
+    Example::
+
+        src = "this is test <b> bold <i>text</i> </b> normal text "
+        for tag in ("b", "i"):
+            opener, closer = make_html_tags(tag)
+            patt = original_text_for(opener + ... + closer)
+            print(patt.search_string(src)[0])
+
+    prints::
+
+        ['<b> bold <i>text</i> </b>']
+        ['<i>text</i>']
+    """
+    asString = asString and as_string
+
+    locMarker = Empty().set_parse_action(lambda s, loc, t: loc)
+    endlocMarker = locMarker.copy()
+    endlocMarker.callPreparse = False
+    matchExpr = locMarker("_original_start") + expr + endlocMarker("_original_end")
+    if asString:
+        extractText = lambda s, l, t: s[t._original_start : t._original_end]
+    else:
+
+        def extractText(s, l, t):
+            t[:] = [s[t.pop("_original_start") : t.pop("_original_end")]]
+
+    matchExpr.set_parse_action(extractText)
+    matchExpr.ignoreExprs = expr.ignoreExprs
+    matchExpr.suppress_warning(Diagnostics.warn_ungrouped_named_tokens_in_collection)
+    return matchExpr
+
+
+def ungroup(expr: ParserElement) -> ParserElement:
+    """Helper to undo pyparsing's default grouping of And expressions,
+    even if all but one are non-empty.
+    """
+    return TokenConverter(expr).add_parse_action(lambda t: t[0])
+
+
+def locatedExpr(expr: ParserElement) -> ParserElement:
+    """
+    (DEPRECATED - future code should use the :class:`Located` class)
+    Helper to decorate a returned token with its starting and ending
+    locations in the input string.
+
+    This helper adds the following results names:
+
+    - ``locn_start`` - location where matched expression begins
+    - ``locn_end`` - location where matched expression ends
+    - ``value`` - the actual parsed results
+
+    Be careful if the input text contains ``<TAB>`` characters, you
+    may want to call :class:`ParserElement.parse_with_tabs`
+
+    Example::
+
+        wd = Word(alphas)
+        for match in locatedExpr(wd).search_string("ljsdf123lksdjjf123lkkjj1222"):
+            print(match)
+
+    prints::
+
+        [[0, 'ljsdf', 5]]
+        [[8, 'lksdjjf', 15]]
+        [[18, 'lkkjj', 23]]
+    """
+    locator = Empty().set_parse_action(lambda ss, ll, tt: ll)
+    return Group(
+        locator("locn_start")
+        + expr("value")
+        + locator.copy().leaveWhitespace()("locn_end")
+    )
+
+
+def nested_expr(
+    opener: Union[str, ParserElement] = "(",
+    closer: Union[str, ParserElement] = ")",
+    content: typing.Optional[ParserElement] = None,
+    ignore_expr: ParserElement = quoted_string(),
+    *,
+    ignoreExpr: ParserElement = quoted_string(),
+) -> ParserElement:
+    """Helper method for defining nested lists enclosed in opening and
+    closing delimiters (``"("`` and ``")"`` are the default).
+
+    Parameters:
+
+    - ``opener`` - opening character for a nested list
+      (default= ``"("``); can also be a pyparsing expression
+    - ``closer`` - closing character for a nested list
+      (default= ``")"``); can also be a pyparsing expression
+    - ``content`` - expression for items within the nested lists
+      (default= ``None``)
+    - ``ignore_expr`` - expression for ignoring opening and closing delimiters
+      (default= :class:`quoted_string`)
+    - ``ignoreExpr`` - this pre-PEP8 argument is retained for compatibility
+      but will be removed in a future release
+
+    If an expression is not provided for the content argument, the
+    nested expression will capture all whitespace-delimited content
+    between delimiters as a list of separate values.
+
+    Use the ``ignore_expr`` argument to define expressions that may
+    contain opening or closing characters that should not be treated as
+    opening or closing characters for nesting, such as quoted_string or
+    a comment expression.  Specify multiple expressions using an
+    :class:`Or` or :class:`MatchFirst`. The default is
+    :class:`quoted_string`, but if no expressions are to be ignored, then
+    pass ``None`` for this argument.
+
+    Example::
+
+        data_type = one_of("void int short long char float double")
+        decl_data_type = Combine(data_type + Opt(Word('*')))
+        ident = Word(alphas+'_', alphanums+'_')
+        number = pyparsing_common.number
+        arg = Group(decl_data_type + ident)
+        LPAR, RPAR = map(Suppress, "()")
+
+        code_body = nested_expr('{', '}', ignore_expr=(quoted_string | c_style_comment))
+
+        c_function = (decl_data_type("type")
+                      + ident("name")
+                      + LPAR + Opt(DelimitedList(arg), [])("args") + RPAR
+                      + code_body("body"))
+        c_function.ignore(c_style_comment)
+
+        source_code = '''
+            int is_odd(int x) {
+                return (x%2);
+            }
+
+            int dec_to_hex(char hchar) {
+                if (hchar >= '0' && hchar <= '9') {
+                    return (ord(hchar)-ord('0'));
+                } else {
+                    return (10+ord(hchar)-ord('A'));
+                }
+            }
+        '''
+        for func in c_function.search_string(source_code):
+            print("%(name)s (%(type)s) args: %(args)s" % func)
+
+
+    prints::
+
+        is_odd (int) args: [['int', 'x']]
+        dec_to_hex (int) args: [['char', 'hchar']]
+    """
+    if ignoreExpr != ignore_expr:
+        ignoreExpr = ignore_expr if ignoreExpr == quoted_string() else ignoreExpr
+    if opener == closer:
+        raise ValueError("opening and closing strings cannot be the same")
+    if content is None:
+        if isinstance(opener, str_type) and isinstance(closer, str_type):
+            opener = typing.cast(str, opener)
+            closer = typing.cast(str, closer)
+            if len(opener) == 1 and len(closer) == 1:
+                if ignoreExpr is not None:
+                    content = Combine(
+                        OneOrMore(
+                            ~ignoreExpr
+                            + CharsNotIn(
+                                opener + closer + ParserElement.DEFAULT_WHITE_CHARS,
+                                exact=1,
+                            )
+                        )
+                    ).set_parse_action(lambda t: t[0].strip())
+                else:
+                    content = empty.copy() + CharsNotIn(
+                        opener + closer + ParserElement.DEFAULT_WHITE_CHARS
+                    ).set_parse_action(lambda t: t[0].strip())
+            else:
+                if ignoreExpr is not None:
+                    content = Combine(
+                        OneOrMore(
+                            ~ignoreExpr
+                            + ~Literal(opener)
+                            + ~Literal(closer)
+                            + CharsNotIn(ParserElement.DEFAULT_WHITE_CHARS, exact=1)
+                        )
+                    ).set_parse_action(lambda t: t[0].strip())
+                else:
+                    content = Combine(
+                        OneOrMore(
+                            ~Literal(opener)
+                            + ~Literal(closer)
+                            + CharsNotIn(ParserElement.DEFAULT_WHITE_CHARS, exact=1)
+                        )
+                    ).set_parse_action(lambda t: t[0].strip())
+        else:
+            raise ValueError(
+                "opening and closing arguments must be strings if no content expression is given"
+            )
+    ret = Forward()
+    if ignoreExpr is not None:
+        ret <<= Group(
+            Suppress(opener) + ZeroOrMore(ignoreExpr | ret | content) + Suppress(closer)
+        )
+    else:
+        ret <<= Group(Suppress(opener) + ZeroOrMore(ret | content) + Suppress(closer))
+    ret.set_name("nested %s%s expression" % (opener, closer))
+    return ret
+
+
+def _makeTags(tagStr, xml, suppress_LT=Suppress("<"), suppress_GT=Suppress(">")):
+    """Internal helper to construct opening and closing tag expressions, given a tag name"""
+    if isinstance(tagStr, str_type):
+        resname = tagStr
+        tagStr = Keyword(tagStr, caseless=not xml)
+    else:
+        resname = tagStr.name
+
+    tagAttrName = Word(alphas, alphanums + "_-:")
+    if xml:
+        tagAttrValue = dbl_quoted_string.copy().set_parse_action(remove_quotes)
+        openTag = (
+            suppress_LT
+            + tagStr("tag")
+            + Dict(ZeroOrMore(Group(tagAttrName + Suppress("=") + tagAttrValue)))
+            + Opt("/", default=[False])("empty").set_parse_action(
+                lambda s, l, t: t[0] == "/"
+            )
+            + suppress_GT
+        )
+    else:
+        tagAttrValue = quoted_string.copy().set_parse_action(remove_quotes) | Word(
+            printables, exclude_chars=">"
+        )
+        openTag = (
+            suppress_LT
+            + tagStr("tag")
+            + Dict(
+                ZeroOrMore(
+                    Group(
+                        tagAttrName.set_parse_action(lambda t: t[0].lower())
+                        + Opt(Suppress("=") + tagAttrValue)
+                    )
+                )
+            )
+            + Opt("/", default=[False])("empty").set_parse_action(
+                lambda s, l, t: t[0] == "/"
+            )
+            + suppress_GT
+        )
+    closeTag = Combine(Literal("</") + tagStr + ">", adjacent=False)
+
+    openTag.set_name("<%s>" % resname)
+    # add start<tagname> results name in parse action now that ungrouped names are not reported at two levels
+    openTag.add_parse_action(
+        lambda t: t.__setitem__(
+            "start" + "".join(resname.replace(":", " ").title().split()), t.copy()
+        )
+    )
+    closeTag = closeTag(
+        "end" + "".join(resname.replace(":", " ").title().split())
+    ).set_name("</%s>" % resname)
+    openTag.tag = resname
+    closeTag.tag = resname
+    openTag.tag_body = SkipTo(closeTag())
+    return openTag, closeTag
+
+
+def make_html_tags(
+    tag_str: Union[str, ParserElement]
+) -> Tuple[ParserElement, ParserElement]:
+    """Helper to construct opening and closing tag expressions for HTML,
+    given a tag name. Matches tags in either upper or lower case,
+    attributes with namespaces and with quoted or unquoted values.
+
+    Example::
+
+        text = '<td>More info at the <a href="https://github.com/pyparsing/pyparsing/wiki">pyparsing</a> wiki page</td>'
+        # make_html_tags returns pyparsing expressions for the opening and
+        # closing tags as a 2-tuple
+        a, a_end = make_html_tags("A")
+        link_expr = a + SkipTo(a_end)("link_text") + a_end
+
+        for link in link_expr.search_string(text):
+            # attributes in the <A> tag (like "href" shown here) are
+            # also accessible as named results
+            print(link.link_text, '->', link.href)
+
+    prints::
+
+        pyparsing -> https://github.com/pyparsing/pyparsing/wiki
+    """
+    return _makeTags(tag_str, False)
+
+
+def make_xml_tags(
+    tag_str: Union[str, ParserElement]
+) -> Tuple[ParserElement, ParserElement]:
+    """Helper to construct opening and closing tag expressions for XML,
+    given a tag name. Matches tags only in the given upper/lower case.
+
+    Example: similar to :class:`make_html_tags`
+    """
+    return _makeTags(tag_str, True)
+
+
+any_open_tag: ParserElement
+any_close_tag: ParserElement
+any_open_tag, any_close_tag = make_html_tags(
+    Word(alphas, alphanums + "_:").set_name("any tag")
+)
+
+_htmlEntityMap = {k.rstrip(";"): v for k, v in html.entities.html5.items()}
+common_html_entity = Regex("&(?P<entity>" + "|".join(_htmlEntityMap) + ");").set_name(
+    "common HTML entity"
+)
+
+
+def replace_html_entity(s, l, t):
+    """Helper parser action to replace common HTML entities with their special characters"""
+    return _htmlEntityMap.get(t.entity)
+
+
+class OpAssoc(Enum):
+    """Enumeration of operator associativity
+    - used in constructing InfixNotationOperatorSpec for :class:`infix_notation`"""
+
+    LEFT = 1
+    RIGHT = 2
+
+
+InfixNotationOperatorArgType = Union[
+    ParserElement, str, Tuple[Union[ParserElement, str], Union[ParserElement, str]]
+]
+InfixNotationOperatorSpec = Union[
+    Tuple[
+        InfixNotationOperatorArgType,
+        int,
+        OpAssoc,
+        typing.Optional[ParseAction],
+    ],
+    Tuple[
+        InfixNotationOperatorArgType,
+        int,
+        OpAssoc,
+    ],
+]
+
+
+def infix_notation(
+    base_expr: ParserElement,
+    op_list: List[InfixNotationOperatorSpec],
+    lpar: Union[str, ParserElement] = Suppress("("),
+    rpar: Union[str, ParserElement] = Suppress(")"),
+) -> ParserElement:
+    """Helper method for constructing grammars of expressions made up of
+    operators working in a precedence hierarchy.  Operators may be unary
+    or binary, left- or right-associative.  Parse actions can also be
+    attached to operator expressions. The generated parser will also
+    recognize the use of parentheses to override operator precedences
+    (see example below).
+
+    Note: if you define a deep operator list, you may see performance
+    issues when using infix_notation. See
+    :class:`ParserElement.enable_packrat` for a mechanism to potentially
+    improve your parser performance.
+
+    Parameters:
+
+    - ``base_expr`` - expression representing the most basic operand to
+      be used in the expression
+    - ``op_list`` - list of tuples, one for each operator precedence level
+      in the expression grammar; each tuple is of the form ``(op_expr,
+      num_operands, right_left_assoc, (optional)parse_action)``, where:
+
+      - ``op_expr`` is the pyparsing expression for the operator; may also
+        be a string, which will be converted to a Literal; if ``num_operands``
+        is 3, ``op_expr`` is a tuple of two expressions, for the two
+        operators separating the 3 terms
+      - ``num_operands`` is the number of terms for this operator (must be 1,
+        2, or 3)
+      - ``right_left_assoc`` is the indicator whether the operator is right
+        or left associative, using the pyparsing-defined constants
+        ``OpAssoc.RIGHT`` and ``OpAssoc.LEFT``.
+      - ``parse_action`` is the parse action to be associated with
+        expressions matching this operator expression (the parse action
+        tuple member may be omitted); if the parse action is passed
+        a tuple or list of functions, this is equivalent to calling
+        ``set_parse_action(*fn)``
+        (:class:`ParserElement.set_parse_action`)
+    - ``lpar`` - expression for matching left-parentheses; if passed as a
+      str, then will be parsed as ``Suppress(lpar)``. If lpar is passed as
+      an expression (such as ``Literal('(')``), then it will be kept in
+      the parsed results, and grouped with them. (default= ``Suppress('(')``)
+    - ``rpar`` - expression for matching right-parentheses; if passed as a
+      str, then will be parsed as ``Suppress(rpar)``. If rpar is passed as
+      an expression (such as ``Literal(')')``), then it will be kept in
+      the parsed results, and grouped with them. (default= ``Suppress(')')``)
+
+    Example::
+
+        # simple example of four-function arithmetic with ints and
+        # variable names
+        integer = pyparsing_common.signed_integer
+        varname = pyparsing_common.identifier
+
+        arith_expr = infix_notation(integer | varname,
+            [
+            ('-', 1, OpAssoc.RIGHT),
+            (one_of('* /'), 2, OpAssoc.LEFT),
+            (one_of('+ -'), 2, OpAssoc.LEFT),
+            ])
+
+        arith_expr.run_tests('''
+            5+3*6
+            (5+3)*6
+            -2--11
+            ''', full_dump=False)
+
+    prints::
+
+        5+3*6
+        [[5, '+', [3, '*', 6]]]
+
+        (5+3)*6
+        [[[5, '+', 3], '*', 6]]
+
+        (5+x)*y
+        [[[5, '+', 'x'], '*', 'y']]
+
+        -2--11
+        [[['-', 2], '-', ['-', 11]]]
+    """
+
+    # captive version of FollowedBy that does not do parse actions or capture results names
+    class _FB(FollowedBy):
+        def parseImpl(self, instring, loc, doActions=True):
+            self.expr.try_parse(instring, loc)
+            return loc, []
+
+    _FB.__name__ = "FollowedBy>"
+
+    ret = Forward()
+    if isinstance(lpar, str):
+        lpar = Suppress(lpar)
+    if isinstance(rpar, str):
+        rpar = Suppress(rpar)
+
+    # if lpar and rpar are not suppressed, wrap in group
+    if not (isinstance(rpar, Suppress) and isinstance(rpar, Suppress)):
+        lastExpr = base_expr | Group(lpar + ret + rpar)
+    else:
+        lastExpr = base_expr | (lpar + ret + rpar)
+
+    arity: int
+    rightLeftAssoc: opAssoc
+    pa: typing.Optional[ParseAction]
+    opExpr1: ParserElement
+    opExpr2: ParserElement
+    for i, operDef in enumerate(op_list):
+        opExpr, arity, rightLeftAssoc, pa = (operDef + (None,))[:4]  # type: ignore[assignment]
+        if isinstance(opExpr, str_type):
+            opExpr = ParserElement._literalStringClass(opExpr)
+        opExpr = typing.cast(ParserElement, opExpr)
+        if arity == 3:
+            if not isinstance(opExpr, (tuple, list)) or len(opExpr) != 2:
+                raise ValueError(
+                    "if numterms=3, opExpr must be a tuple or list of two expressions"
+                )
+            opExpr1, opExpr2 = opExpr
+            term_name = f"{opExpr1}{opExpr2} term"
+        else:
+            term_name = f"{opExpr} term"
+
+        if not 1 <= arity <= 3:
+            raise ValueError("operator must be unary (1), binary (2), or ternary (3)")
+
+        if rightLeftAssoc not in (OpAssoc.LEFT, OpAssoc.RIGHT):
+            raise ValueError("operator must indicate right or left associativity")
+
+        thisExpr: ParserElement = Forward().set_name(term_name)
+        thisExpr = typing.cast(Forward, thisExpr)
+        if rightLeftAssoc is OpAssoc.LEFT:
+            if arity == 1:
+                matchExpr = _FB(lastExpr + opExpr) + Group(lastExpr + opExpr[1, ...])
+            elif arity == 2:
+                if opExpr is not None:
+                    matchExpr = _FB(lastExpr + opExpr + lastExpr) + Group(
+                        lastExpr + (opExpr + lastExpr)[1, ...]
+                    )
+                else:
+                    matchExpr = _FB(lastExpr + lastExpr) + Group(lastExpr[2, ...])
+            elif arity == 3:
+                matchExpr = _FB(
+                    lastExpr + opExpr1 + lastExpr + opExpr2 + lastExpr
+                ) + Group(lastExpr + OneOrMore(opExpr1 + lastExpr + opExpr2 + lastExpr))
+        elif rightLeftAssoc is OpAssoc.RIGHT:
+            if arity == 1:
+                # try to avoid LR with this extra test
+                if not isinstance(opExpr, Opt):
+                    opExpr = Opt(opExpr)
+                matchExpr = _FB(opExpr.expr + thisExpr) + Group(opExpr + thisExpr)
+            elif arity == 2:
+                if opExpr is not None:
+                    matchExpr = _FB(lastExpr + opExpr + thisExpr) + Group(
+                        lastExpr + (opExpr + thisExpr)[1, ...]
+                    )
+                else:
+                    matchExpr = _FB(lastExpr + thisExpr) + Group(
+                        lastExpr + thisExpr[1, ...]
+                    )
+            elif arity == 3:
+                matchExpr = _FB(
+                    lastExpr + opExpr1 + thisExpr + opExpr2 + thisExpr
+                ) + Group(lastExpr + opExpr1 + thisExpr + opExpr2 + thisExpr)
+        if pa:
+            if isinstance(pa, (tuple, list)):
+                matchExpr.set_parse_action(*pa)
+            else:
+                matchExpr.set_parse_action(pa)
+        thisExpr <<= (matchExpr | lastExpr).setName(term_name)
+        lastExpr = thisExpr
+    ret <<= lastExpr
+    return ret
+
+
+def indentedBlock(blockStatementExpr, indentStack, indent=True, backup_stacks=[]):
+    """
+    (DEPRECATED - use :class:`IndentedBlock` class instead)
+    Helper method for defining space-delimited indentation blocks,
+    such as those used to define block statements in Python source code.
+
+    Parameters:
+
+    - ``blockStatementExpr`` - expression defining syntax of statement that
+      is repeated within the indented block
+    - ``indentStack`` - list created by caller to manage indentation stack
+      (multiple ``statementWithIndentedBlock`` expressions within a single
+      grammar should share a common ``indentStack``)
+    - ``indent`` - boolean indicating whether block must be indented beyond
+      the current level; set to ``False`` for block of left-most statements
+      (default= ``True``)
+
+    A valid block must contain at least one ``blockStatement``.
+
+    (Note that indentedBlock uses internal parse actions which make it
+    incompatible with packrat parsing.)
+
+    Example::
+
+        data = '''
+        def A(z):
+          A1
+          B = 100
+          G = A2
+          A2
+          A3
+        B
+        def BB(a,b,c):
+          BB1
+          def BBA():
+            bba1
+            bba2
+            bba3
+        C
+        D
+        def spam(x,y):
+             def eggs(z):
+                 pass
+        '''
+
+
+        indentStack = [1]
+        stmt = Forward()
+
+        identifier = Word(alphas, alphanums)
+        funcDecl = ("def" + identifier + Group("(" + Opt(delimitedList(identifier)) + ")") + ":")
+        func_body = indentedBlock(stmt, indentStack)
+        funcDef = Group(funcDecl + func_body)
+
+        rvalue = Forward()
+        funcCall = Group(identifier + "(" + Opt(delimitedList(rvalue)) + ")")
+        rvalue << (funcCall | identifier | Word(nums))
+        assignment = Group(identifier + "=" + rvalue)
+        stmt << (funcDef | assignment | identifier)
+
+        module_body = stmt[1, ...]
+
+        parseTree = module_body.parseString(data)
+        parseTree.pprint()
+
+    prints::
+
+        [['def',
+          'A',
+          ['(', 'z', ')'],
+          ':',
+          [['A1'], [['B', '=', '100']], [['G', '=', 'A2']], ['A2'], ['A3']]],
+         'B',
+         ['def',
+          'BB',
+          ['(', 'a', 'b', 'c', ')'],
+          ':',
+          [['BB1'], [['def', 'BBA', ['(', ')'], ':', [['bba1'], ['bba2'], ['bba3']]]]]],
+         'C',
+         'D',
+         ['def',
+          'spam',
+          ['(', 'x', 'y', ')'],
+          ':',
+          [[['def', 'eggs', ['(', 'z', ')'], ':', [['pass']]]]]]]
+    """
+    backup_stacks.append(indentStack[:])
+
+    def reset_stack():
+        indentStack[:] = backup_stacks[-1]
+
+    def checkPeerIndent(s, l, t):
+        if l >= len(s):
+            return
+        curCol = col(l, s)
+        if curCol != indentStack[-1]:
+            if curCol > indentStack[-1]:
+                raise ParseException(s, l, "illegal nesting")
+            raise ParseException(s, l, "not a peer entry")
+
+    def checkSubIndent(s, l, t):
+        curCol = col(l, s)
+        if curCol > indentStack[-1]:
+            indentStack.append(curCol)
+        else:
+            raise ParseException(s, l, "not a subentry")
+
+    def checkUnindent(s, l, t):
+        if l >= len(s):
+            return
+        curCol = col(l, s)
+        if not (indentStack and curCol in indentStack):
+            raise ParseException(s, l, "not an unindent")
+        if curCol < indentStack[-1]:
+            indentStack.pop()
+
+    NL = OneOrMore(LineEnd().set_whitespace_chars("\t ").suppress())
+    INDENT = (Empty() + Empty().set_parse_action(checkSubIndent)).set_name("INDENT")
+    PEER = Empty().set_parse_action(checkPeerIndent).set_name("")
+    UNDENT = Empty().set_parse_action(checkUnindent).set_name("UNINDENT")
+    if indent:
+        smExpr = Group(
+            Opt(NL)
+            + INDENT
+            + OneOrMore(PEER + Group(blockStatementExpr) + Opt(NL))
+            + UNDENT
+        )
+    else:
+        smExpr = Group(
+            Opt(NL)
+            + OneOrMore(PEER + Group(blockStatementExpr) + Opt(NL))
+            + Opt(UNDENT)
+        )
+
+    # add a parse action to remove backup_stack from list of backups
+    smExpr.add_parse_action(
+        lambda: backup_stacks.pop(-1) and None if backup_stacks else None
+    )
+    smExpr.set_fail_action(lambda a, b, c, d: reset_stack())
+    blockStatementExpr.ignore(_bslash + LineEnd())
+    return smExpr.set_name("indented block")
+
+
+# it's easy to get these comment structures wrong - they're very common, so may as well make them available
+c_style_comment = Combine(Regex(r"/\*(?:[^*]|\*(?!/))*") + "*/").set_name(
+    "C style comment"
+)
+"Comment of the form ``/* ... */``"
+
+html_comment = Regex(r"<!--[\s\S]*?-->").set_name("HTML comment")
+"Comment of the form ``<!-- ... -->``"
+
+rest_of_line = Regex(r".*").leave_whitespace().set_name("rest of line")
+dbl_slash_comment = Regex(r"//(?:\\\n|[^\n])*").set_name("// comment")
+"Comment of the form ``// ... (to end of line)``"
+
+cpp_style_comment = Combine(
+    Regex(r"/\*(?:[^*]|\*(?!/))*") + "*/" | dbl_slash_comment
+).set_name("C++ style comment")
+"Comment of either form :class:`c_style_comment` or :class:`dbl_slash_comment`"
+
+java_style_comment = cpp_style_comment
+"Same as :class:`cpp_style_comment`"
+
+python_style_comment = Regex(r"#.*").set_name("Python style comment")
+"Comment of the form ``# ... (to end of line)``"
+
+
+# build list of built-in expressions, for future reference if a global default value
+# gets updated
+_builtin_exprs: List[ParserElement] = [
+    v for v in vars().values() if isinstance(v, ParserElement)
+]
+
+
+# compatibility function, superseded by DelimitedList class
+def delimited_list(
+    expr: Union[str, ParserElement],
+    delim: Union[str, ParserElement] = ",",
+    combine: bool = False,
+    min: typing.Optional[int] = None,
+    max: typing.Optional[int] = None,
+    *,
+    allow_trailing_delim: bool = False,
+) -> ParserElement:
+    """(DEPRECATED - use :class:`DelimitedList` class)"""
+    return DelimitedList(
+        expr, delim, combine, min, max, allow_trailing_delim=allow_trailing_delim
+    )
+
+
+# pre-PEP8 compatible names
+# fmt: off
+opAssoc = OpAssoc
+anyOpenTag = any_open_tag
+anyCloseTag = any_close_tag
+commonHTMLEntity = common_html_entity
+cStyleComment = c_style_comment
+htmlComment = html_comment
+restOfLine = rest_of_line
+dblSlashComment = dbl_slash_comment
+cppStyleComment = cpp_style_comment
+javaStyleComment = java_style_comment
+pythonStyleComment = python_style_comment
+
+@replaced_by_pep8(DelimitedList)
+def delimitedList(): ...
+
+@replaced_by_pep8(DelimitedList)
+def delimited_list(): ...
+
+@replaced_by_pep8(counted_array)
+def countedArray(): ...
+
+@replaced_by_pep8(match_previous_literal)
+def matchPreviousLiteral(): ...
+
+@replaced_by_pep8(match_previous_expr)
+def matchPreviousExpr(): ...
+
+@replaced_by_pep8(one_of)
+def oneOf(): ...
+
+@replaced_by_pep8(dict_of)
+def dictOf(): ...
+
+@replaced_by_pep8(original_text_for)
+def originalTextFor(): ...
+
+@replaced_by_pep8(nested_expr)
+def nestedExpr(): ...
+
+@replaced_by_pep8(make_html_tags)
+def makeHTMLTags(): ...
+
+@replaced_by_pep8(make_xml_tags)
+def makeXMLTags(): ...
+
+@replaced_by_pep8(replace_html_entity)
+def replaceHTMLEntity(): ...
+
+@replaced_by_pep8(infix_notation)
+def infixNotation(): ...
+# fmt: on

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/results.py

@@ -0,0 +1,796 @@
+# results.py
+from collections.abc import (
+    MutableMapping,
+    Mapping,
+    MutableSequence,
+    Iterator,
+    Sequence,
+    Container,
+)
+import pprint
+from typing import Tuple, Any, Dict, Set, List
+
+str_type: Tuple[type, ...] = (str, bytes)
+_generator_type = type((_ for _ in ()))
+
+
+class _ParseResultsWithOffset:
+    tup: Tuple["ParseResults", int]
+    __slots__ = ["tup"]
+
+    def __init__(self, p1: "ParseResults", p2: int):
+        self.tup: Tuple[ParseResults, int] = (p1, p2)
+
+    def __getitem__(self, i):
+        return self.tup[i]
+
+    def __getstate__(self):
+        return self.tup
+
+    def __setstate__(self, *args):
+        self.tup = args[0]
+
+
+class ParseResults:
+    """Structured parse results, to provide multiple means of access to
+    the parsed data:
+
+    - as a list (``len(results)``)
+    - by list index (``results[0], results[1]``, etc.)
+    - by attribute (``results.<results_name>`` - see :class:`ParserElement.set_results_name`)
+
+    Example::
+
+        integer = Word(nums)
+        date_str = (integer.set_results_name("year") + '/'
+                    + integer.set_results_name("month") + '/'
+                    + integer.set_results_name("day"))
+        # equivalent form:
+        # date_str = (integer("year") + '/'
+        #             + integer("month") + '/'
+        #             + integer("day"))
+
+        # parse_string returns a ParseResults object
+        result = date_str.parse_string("1999/12/31")
+
+        def test(s, fn=repr):
+            print(f"{s} -> {fn(eval(s))}")
+        test("list(result)")
+        test("result[0]")
+        test("result['month']")
+        test("result.day")
+        test("'month' in result")
+        test("'minutes' in result")
+        test("result.dump()", str)
+
+    prints::
+
+        list(result) -> ['1999', '/', '12', '/', '31']
+        result[0] -> '1999'
+        result['month'] -> '12'
+        result.day -> '31'
+        'month' in result -> True
+        'minutes' in result -> False
+        result.dump() -> ['1999', '/', '12', '/', '31']
+        - day: '31'
+        - month: '12'
+        - year: '1999'
+    """
+
+    _null_values: Tuple[Any, ...] = (None, [], ())
+
+    _name: str
+    _parent: "ParseResults"
+    _all_names: Set[str]
+    _modal: bool
+    _toklist: List[Any]
+    _tokdict: Dict[str, Any]
+
+    __slots__ = (
+        "_name",
+        "_parent",
+        "_all_names",
+        "_modal",
+        "_toklist",
+        "_tokdict",
+    )
+
+    class List(list):
+        """
+        Simple wrapper class to distinguish parsed list results that should be preserved
+        as actual Python lists, instead of being converted to :class:`ParseResults`::
+
+            LBRACK, RBRACK = map(pp.Suppress, "[]")
+            element = pp.Forward()
+            item = ppc.integer
+            element_list = LBRACK + pp.DelimitedList(element) + RBRACK
+
+            # add parse actions to convert from ParseResults to actual Python collection types
+            def as_python_list(t):
+                return pp.ParseResults.List(t.as_list())
+            element_list.add_parse_action(as_python_list)
+
+            element <<= item | element_list
+
+            element.run_tests('''
+                100
+                [2,3,4]
+                [[2, 1],3,4]
+                [(2, 1),3,4]
+                (2,3,4)
+                ''', post_parse=lambda s, r: (r[0], type(r[0])))
+
+        prints::
+
+            100
+            (100, <class 'int'>)
+
+            [2,3,4]
+            ([2, 3, 4], <class 'list'>)
+
+            [[2, 1],3,4]
+            ([[2, 1], 3, 4], <class 'list'>)
+
+        (Used internally by :class:`Group` when `aslist=True`.)
+        """
+
+        def __new__(cls, contained=None):
+            if contained is None:
+                contained = []
+
+            if not isinstance(contained, list):
+                raise TypeError(
+                    f"{cls.__name__} may only be constructed with a list, not {type(contained).__name__}"
+                )
+
+            return list.__new__(cls)
+
+    def __new__(cls, toklist=None, name=None, **kwargs):
+        if isinstance(toklist, ParseResults):
+            return toklist
+        self = object.__new__(cls)
+        self._name = None
+        self._parent = None
+        self._all_names = set()
+
+        if toklist is None:
+            self._toklist = []
+        elif isinstance(toklist, (list, _generator_type)):
+            self._toklist = (
+                [toklist[:]]
+                if isinstance(toklist, ParseResults.List)
+                else list(toklist)
+            )
+        else:
+            self._toklist = [toklist]
+        self._tokdict = dict()
+        return self
+
+    # Performance tuning: we construct a *lot* of these, so keep this
+    # constructor as small and fast as possible
+    def __init__(
+        self, toklist=None, name=None, asList=True, modal=True, isinstance=isinstance
+    ):
+        self._tokdict: Dict[str, _ParseResultsWithOffset]
+        self._modal = modal
+        if name is not None and name != "":
+            if isinstance(name, int):
+                name = str(name)
+            if not modal:
+                self._all_names = {name}
+            self._name = name
+            if toklist not in self._null_values:
+                if isinstance(toklist, (str_type, type)):
+                    toklist = [toklist]
+                if asList:
+                    if isinstance(toklist, ParseResults):
+                        self[name] = _ParseResultsWithOffset(
+                            ParseResults(toklist._toklist), 0
+                        )
+                    else:
+                        self[name] = _ParseResultsWithOffset(
+                            ParseResults(toklist[0]), 0
+                        )
+                    self[name]._name = name
+                else:
+                    try:
+                        self[name] = toklist[0]
+                    except (KeyError, TypeError, IndexError):
+                        if toklist is not self:
+                            self[name] = toklist
+                        else:
+                            self._name = name
+
+    def __getitem__(self, i):
+        if isinstance(i, (int, slice)):
+            return self._toklist[i]
+        else:
+            if i not in self._all_names:
+                return self._tokdict[i][-1][0]
+            else:
+                return ParseResults([v[0] for v in self._tokdict[i]])
+
+    def __setitem__(self, k, v, isinstance=isinstance):
+        if isinstance(v, _ParseResultsWithOffset):
+            self._tokdict[k] = self._tokdict.get(k, list()) + [v]
+            sub = v[0]
+        elif isinstance(k, (int, slice)):
+            self._toklist[k] = v
+            sub = v
+        else:
+            self._tokdict[k] = self._tokdict.get(k, list()) + [
+                _ParseResultsWithOffset(v, 0)
+            ]
+            sub = v
+        if isinstance(sub, ParseResults):
+            sub._parent = self
+
+    def __delitem__(self, i):
+        if isinstance(i, (int, slice)):
+            mylen = len(self._toklist)
+            del self._toklist[i]
+
+            # convert int to slice
+            if isinstance(i, int):
+                if i < 0:
+                    i += mylen
+                i = slice(i, i + 1)
+            # get removed indices
+            removed = list(range(*i.indices(mylen)))
+            removed.reverse()
+            # fixup indices in token dictionary
+            for name, occurrences in self._tokdict.items():
+                for j in removed:
+                    for k, (value, position) in enumerate(occurrences):
+                        occurrences[k] = _ParseResultsWithOffset(
+                            value, position - (position > j)
+                        )
+        else:
+            del self._tokdict[i]
+
+    def __contains__(self, k) -> bool:
+        return k in self._tokdict
+
+    def __len__(self) -> int:
+        return len(self._toklist)
+
+    def __bool__(self) -> bool:
+        return not not (self._toklist or self._tokdict)
+
+    def __iter__(self) -> Iterator:
+        return iter(self._toklist)
+
+    def __reversed__(self) -> Iterator:
+        return iter(self._toklist[::-1])
+
+    def keys(self):
+        return iter(self._tokdict)
+
+    def values(self):
+        return (self[k] for k in self.keys())
+
+    def items(self):
+        return ((k, self[k]) for k in self.keys())
+
+    def haskeys(self) -> bool:
+        """
+        Since ``keys()`` returns an iterator, this method is helpful in bypassing
+        code that looks for the existence of any defined results names."""
+        return not not self._tokdict
+
+    def pop(self, *args, **kwargs):
+        """
+        Removes and returns item at specified index (default= ``last``).
+        Supports both ``list`` and ``dict`` semantics for ``pop()``. If
+        passed no argument or an integer argument, it will use ``list``
+        semantics and pop tokens from the list of parsed tokens. If passed
+        a non-integer argument (most likely a string), it will use ``dict``
+        semantics and pop the corresponding value from any defined results
+        names. A second default return value argument is supported, just as in
+        ``dict.pop()``.
+
+        Example::
+
+            numlist = Word(nums)[...]
+            print(numlist.parse_string("0 123 321")) # -> ['0', '123', '321']
+
+            def remove_first(tokens):
+                tokens.pop(0)
+            numlist.add_parse_action(remove_first)
+            print(numlist.parse_string("0 123 321")) # -> ['123', '321']
+
+            label = Word(alphas)
+            patt = label("LABEL") + Word(nums)[1, ...]
+            print(patt.parse_string("AAB 123 321").dump())
+
+            # Use pop() in a parse action to remove named result (note that corresponding value is not
+            # removed from list form of results)
+            def remove_LABEL(tokens):
+                tokens.pop("LABEL")
+                return tokens
+            patt.add_parse_action(remove_LABEL)
+            print(patt.parse_string("AAB 123 321").dump())
+
+        prints::
+
+            ['AAB', '123', '321']
+            - LABEL: 'AAB'
+
+            ['AAB', '123', '321']
+        """
+        if not args:
+            args = [-1]
+        for k, v in kwargs.items():
+            if k == "default":
+                args = (args[0], v)
+            else:
+                raise TypeError(f"pop() got an unexpected keyword argument {k!r}")
+        if isinstance(args[0], int) or len(args) == 1 or args[0] in self:
+            index = args[0]
+            ret = self[index]
+            del self[index]
+            return ret
+        else:
+            defaultvalue = args[1]
+            return defaultvalue
+
+    def get(self, key, default_value=None):
+        """
+        Returns named result matching the given key, or if there is no
+        such name, then returns the given ``default_value`` or ``None`` if no
+        ``default_value`` is specified.
+
+        Similar to ``dict.get()``.
+
+        Example::
+
+            integer = Word(nums)
+            date_str = integer("year") + '/' + integer("month") + '/' + integer("day")
+
+            result = date_str.parse_string("1999/12/31")
+            print(result.get("year")) # -> '1999'
+            print(result.get("hour", "not specified")) # -> 'not specified'
+            print(result.get("hour")) # -> None
+        """
+        if key in self:
+            return self[key]
+        else:
+            return default_value
+
+    def insert(self, index, ins_string):
+        """
+        Inserts new element at location index in the list of parsed tokens.
+
+        Similar to ``list.insert()``.
+
+        Example::
+
+            numlist = Word(nums)[...]
+            print(numlist.parse_string("0 123 321")) # -> ['0', '123', '321']
+
+            # use a parse action to insert the parse location in the front of the parsed results
+            def insert_locn(locn, tokens):
+                tokens.insert(0, locn)
+            numlist.add_parse_action(insert_locn)
+            print(numlist.parse_string("0 123 321")) # -> [0, '0', '123', '321']
+        """
+        self._toklist.insert(index, ins_string)
+        # fixup indices in token dictionary
+        for name, occurrences in self._tokdict.items():
+            for k, (value, position) in enumerate(occurrences):
+                occurrences[k] = _ParseResultsWithOffset(
+                    value, position + (position > index)
+                )
+
+    def append(self, item):
+        """
+        Add single element to end of ``ParseResults`` list of elements.
+
+        Example::
+
+            numlist = Word(nums)[...]
+            print(numlist.parse_string("0 123 321")) # -> ['0', '123', '321']
+
+            # use a parse action to compute the sum of the parsed integers, and add it to the end
+            def append_sum(tokens):
+                tokens.append(sum(map(int, tokens)))
+            numlist.add_parse_action(append_sum)
+            print(numlist.parse_string("0 123 321")) # -> ['0', '123', '321', 444]
+        """
+        self._toklist.append(item)
+
+    def extend(self, itemseq):
+        """
+        Add sequence of elements to end of ``ParseResults`` list of elements.
+
+        Example::
+
+            patt = Word(alphas)[1, ...]
+
+            # use a parse action to append the reverse of the matched strings, to make a palindrome
+            def make_palindrome(tokens):
+                tokens.extend(reversed([t[::-1] for t in tokens]))
+                return ''.join(tokens)
+            patt.add_parse_action(make_palindrome)
+            print(patt.parse_string("lskdj sdlkjf lksd")) # -> 'lskdjsdlkjflksddsklfjkldsjdksl'
+        """
+        if isinstance(itemseq, ParseResults):
+            self.__iadd__(itemseq)
+        else:
+            self._toklist.extend(itemseq)
+
+    def clear(self):
+        """
+        Clear all elements and results names.
+        """
+        del self._toklist[:]
+        self._tokdict.clear()
+
+    def __getattr__(self, name):
+        try:
+            return self[name]
+        except KeyError:
+            if name.startswith("__"):
+                raise AttributeError(name)
+            return ""
+
+    def __add__(self, other: "ParseResults") -> "ParseResults":
+        ret = self.copy()
+        ret += other
+        return ret
+
+    def __iadd__(self, other: "ParseResults") -> "ParseResults":
+        if not other:
+            return self
+
+        if other._tokdict:
+            offset = len(self._toklist)
+            addoffset = lambda a: offset if a < 0 else a + offset
+            otheritems = other._tokdict.items()
+            otherdictitems = [
+                (k, _ParseResultsWithOffset(v[0], addoffset(v[1])))
+                for k, vlist in otheritems
+                for v in vlist
+            ]
+            for k, v in otherdictitems:
+                self[k] = v
+                if isinstance(v[0], ParseResults):
+                    v[0]._parent = self
+
+        self._toklist += other._toklist
+        self._all_names |= other._all_names
+        return self
+
+    def __radd__(self, other) -> "ParseResults":
+        if isinstance(other, int) and other == 0:
+            # useful for merging many ParseResults using sum() builtin
+            return self.copy()
+        else:
+            # this may raise a TypeError - so be it
+            return other + self
+
+    def __repr__(self) -> str:
+        return f"{type(self).__name__}({self._toklist!r}, {self.as_dict()})"
+
+    def __str__(self) -> str:
+        return (
+            "["
+            + ", ".join(
+                [
+                    str(i) if isinstance(i, ParseResults) else repr(i)
+                    for i in self._toklist
+                ]
+            )
+            + "]"
+        )
+
+    def _asStringList(self, sep=""):
+        out = []
+        for item in self._toklist:
+            if out and sep:
+                out.append(sep)
+            if isinstance(item, ParseResults):
+                out += item._asStringList()
+            else:
+                out.append(str(item))
+        return out
+
+    def as_list(self) -> list:
+        """
+        Returns the parse results as a nested list of matching tokens, all converted to strings.
+
+        Example::
+
+            patt = Word(alphas)[1, ...]
+            result = patt.parse_string("sldkj lsdkj sldkj")
+            # even though the result prints in string-like form, it is actually a pyparsing ParseResults
+            print(type(result), result) # -> <class 'pyparsing.ParseResults'> ['sldkj', 'lsdkj', 'sldkj']
+
+            # Use as_list() to create an actual list
+            result_list = result.as_list()
+            print(type(result_list), result_list) # -> <class 'list'> ['sldkj', 'lsdkj', 'sldkj']
+        """
+        return [
+            res.as_list() if isinstance(res, ParseResults) else res
+            for res in self._toklist
+        ]
+
+    def as_dict(self) -> dict:
+        """
+        Returns the named parse results as a nested dictionary.
+
+        Example::
+
+            integer = Word(nums)
+            date_str = integer("year") + '/' + integer("month") + '/' + integer("day")
+
+            result = date_str.parse_string('12/31/1999')
+            print(type(result), repr(result)) # -> <class 'pyparsing.ParseResults'> (['12', '/', '31', '/', '1999'], {'day': [('1999', 4)], 'year': [('12', 0)], 'month': [('31', 2)]})
+
+            result_dict = result.as_dict()
+            print(type(result_dict), repr(result_dict)) # -> <class 'dict'> {'day': '1999', 'year': '12', 'month': '31'}
+
+            # even though a ParseResults supports dict-like access, sometime you just need to have a dict
+            import json
+            print(json.dumps(result)) # -> Exception: TypeError: ... is not JSON serializable
+            print(json.dumps(result.as_dict())) # -> {"month": "31", "day": "1999", "year": "12"}
+        """
+
+        def to_item(obj):
+            if isinstance(obj, ParseResults):
+                return obj.as_dict() if obj.haskeys() else [to_item(v) for v in obj]
+            else:
+                return obj
+
+        return dict((k, to_item(v)) for k, v in self.items())
+
+    def copy(self) -> "ParseResults":
+        """
+        Returns a new shallow copy of a :class:`ParseResults` object. `ParseResults`
+        items contained within the source are shared with the copy. Use
+        :class:`ParseResults.deepcopy()` to create a copy with its own separate
+        content values.
+        """
+        ret = ParseResults(self._toklist)
+        ret._tokdict = self._tokdict.copy()
+        ret._parent = self._parent
+        ret._all_names |= self._all_names
+        ret._name = self._name
+        return ret
+
+    def deepcopy(self) -> "ParseResults":
+        """
+        Returns a new deep copy of a :class:`ParseResults` object.
+        """
+        ret = self.copy()
+        # replace values with copies if they are of known mutable types
+        for i, obj in enumerate(self._toklist):
+            if isinstance(obj, ParseResults):
+                self._toklist[i] = obj.deepcopy()
+            elif isinstance(obj, (str, bytes)):
+                pass
+            elif isinstance(obj, MutableMapping):
+                self._toklist[i] = dest = type(obj)()
+                for k, v in obj.items():
+                    dest[k] = v.deepcopy() if isinstance(v, ParseResults) else v
+            elif isinstance(obj, Container):
+                self._toklist[i] = type(obj)(
+                    v.deepcopy() if isinstance(v, ParseResults) else v for v in obj
+                )
+        return ret
+
+    def get_name(self):
+        r"""
+        Returns the results name for this token expression. Useful when several
+        different expressions might match at a particular location.
+
+        Example::
+
+            integer = Word(nums)
+            ssn_expr = Regex(r"\d\d\d-\d\d-\d\d\d\d")
+            house_number_expr = Suppress('#') + Word(nums, alphanums)
+            user_data = (Group(house_number_expr)("house_number")
+                        | Group(ssn_expr)("ssn")
+                        | Group(integer)("age"))
+            user_info = user_data[1, ...]
+
+            result = user_info.parse_string("22 111-22-3333 #221B")
+            for item in result:
+                print(item.get_name(), ':', item[0])
+
+        prints::
+
+            age : 22
+            ssn : 111-22-3333
+            house_number : 221B
+        """
+        if self._name:
+            return self._name
+        elif self._parent:
+            par: "ParseResults" = self._parent
+            parent_tokdict_items = par._tokdict.items()
+            return next(
+                (
+                    k
+                    for k, vlist in parent_tokdict_items
+                    for v, loc in vlist
+                    if v is self
+                ),
+                None,
+            )
+        elif (
+            len(self) == 1
+            and len(self._tokdict) == 1
+            and next(iter(self._tokdict.values()))[0][1] in (0, -1)
+        ):
+            return next(iter(self._tokdict.keys()))
+        else:
+            return None
+
+    def dump(self, indent="", full=True, include_list=True, _depth=0) -> str:
+        """
+        Diagnostic method for listing out the contents of
+        a :class:`ParseResults`. Accepts an optional ``indent`` argument so
+        that this string can be embedded in a nested display of other data.
+
+        Example::
+
+            integer = Word(nums)
+            date_str = integer("year") + '/' + integer("month") + '/' + integer("day")
+
+            result = date_str.parse_string('1999/12/31')
+            print(result.dump())
+
+        prints::
+
+            ['1999', '/', '12', '/', '31']
+            - day: '31'
+            - month: '12'
+            - year: '1999'
+        """
+        out = []
+        NL = "\n"
+        out.append(indent + str(self.as_list()) if include_list else "")
+
+        if full:
+            if self.haskeys():
+                items = sorted((str(k), v) for k, v in self.items())
+                for k, v in items:
+                    if out:
+                        out.append(NL)
+                    out.append(f"{indent}{('  ' * _depth)}- {k}: ")
+                    if isinstance(v, ParseResults):
+                        if v:
+                            out.append(
+                                v.dump(
+                                    indent=indent,
+                                    full=full,
+                                    include_list=include_list,
+                                    _depth=_depth + 1,
+                                )
+                            )
+                        else:
+                            out.append(str(v))
+                    else:
+                        out.append(repr(v))
+            if any(isinstance(vv, ParseResults) for vv in self):
+                v = self
+                for i, vv in enumerate(v):
+                    if isinstance(vv, ParseResults):
+                        out.append(
+                            "\n{}{}[{}]:\n{}{}{}".format(
+                                indent,
+                                ("  " * (_depth)),
+                                i,
+                                indent,
+                                ("  " * (_depth + 1)),
+                                vv.dump(
+                                    indent=indent,
+                                    full=full,
+                                    include_list=include_list,
+                                    _depth=_depth + 1,
+                                ),
+                            )
+                        )
+                    else:
+                        out.append(
+                            "\n%s%s[%d]:\n%s%s%s"
+                            % (
+                                indent,
+                                ("  " * (_depth)),
+                                i,
+                                indent,
+                                ("  " * (_depth + 1)),
+                                str(vv),
+                            )
+                        )
+
+        return "".join(out)
+
+    def pprint(self, *args, **kwargs):
+        """
+        Pretty-printer for parsed results as a list, using the
+        `pprint <https://docs.python.org/3/library/pprint.html>`_ module.
+        Accepts additional positional or keyword args as defined for
+        `pprint.pprint <https://docs.python.org/3/library/pprint.html#pprint.pprint>`_ .
+
+        Example::
+
+            ident = Word(alphas, alphanums)
+            num = Word(nums)
+            func = Forward()
+            term = ident | num | Group('(' + func + ')')
+            func <<= ident + Group(Optional(DelimitedList(term)))
+            result = func.parse_string("fna a,b,(fnb c,d,200),100")
+            result.pprint(width=40)
+
+        prints::
+
+            ['fna',
+             ['a',
+              'b',
+              ['(', 'fnb', ['c', 'd', '200'], ')'],
+              '100']]
+        """
+        pprint.pprint(self.as_list(), *args, **kwargs)
+
+    # add support for pickle protocol
+    def __getstate__(self):
+        return (
+            self._toklist,
+            (
+                self._tokdict.copy(),
+                None,
+                self._all_names,
+                self._name,
+            ),
+        )
+
+    def __setstate__(self, state):
+        self._toklist, (self._tokdict, par, inAccumNames, self._name) = state
+        self._all_names = set(inAccumNames)
+        self._parent = None
+
+    def __getnewargs__(self):
+        return self._toklist, self._name
+
+    def __dir__(self):
+        return dir(type(self)) + list(self.keys())
+
+    @classmethod
+    def from_dict(cls, other, name=None) -> "ParseResults":
+        """
+        Helper classmethod to construct a ``ParseResults`` from a ``dict``, preserving the
+        name-value relations as results names. If an optional ``name`` argument is
+        given, a nested ``ParseResults`` will be returned.
+        """
+
+        def is_iterable(obj):
+            try:
+                iter(obj)
+            except Exception:
+                return False
+            # str's are iterable, but in pyparsing, we don't want to iterate over them
+            else:
+                return not isinstance(obj, str_type)
+
+        ret = cls([])
+        for k, v in other.items():
+            if isinstance(v, Mapping):
+                ret += cls.from_dict(v, name=k)
+            else:
+                ret += cls([v], name=k, asList=is_iterable(v))
+        if name is not None:
+            ret = cls([ret], name=name)
+        return ret
+
+    asList = as_list
+    """Deprecated - use :class:`as_list`"""
+    asDict = as_dict
+    """Deprecated - use :class:`as_dict`"""
+    getName = get_name
+    """Deprecated - use :class:`get_name`"""
+
+
+MutableMapping.register(ParseResults)
+MutableSequence.register(ParseResults)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/testing.py

@@ -0,0 +1,331 @@
+# testing.py
+
+from contextlib import contextmanager
+import typing
+
+from .core import (
+    ParserElement,
+    ParseException,
+    Keyword,
+    __diag__,
+    __compat__,
+)
+
+
+class pyparsing_test:
+    """
+    namespace class for classes useful in writing unit tests
+    """
+
+    class reset_pyparsing_context:
+        """
+        Context manager to be used when writing unit tests that modify pyparsing config values:
+        - packrat parsing
+        - bounded recursion parsing
+        - default whitespace characters.
+        - default keyword characters
+        - literal string auto-conversion class
+        - __diag__ settings
+
+        Example::
+
+            with reset_pyparsing_context():
+                # test that literals used to construct a grammar are automatically suppressed
+                ParserElement.inlineLiteralsUsing(Suppress)
+
+                term = Word(alphas) | Word(nums)
+                group = Group('(' + term[...] + ')')
+
+                # assert that the '()' characters are not included in the parsed tokens
+                self.assertParseAndCheckList(group, "(abc 123 def)", ['abc', '123', 'def'])
+
+            # after exiting context manager, literals are converted to Literal expressions again
+        """
+
+        def __init__(self):
+            self._save_context = {}
+
+        def save(self):
+            self._save_context["default_whitespace"] = ParserElement.DEFAULT_WHITE_CHARS
+            self._save_context["default_keyword_chars"] = Keyword.DEFAULT_KEYWORD_CHARS
+
+            self._save_context[
+                "literal_string_class"
+            ] = ParserElement._literalStringClass
+
+            self._save_context["verbose_stacktrace"] = ParserElement.verbose_stacktrace
+
+            self._save_context["packrat_enabled"] = ParserElement._packratEnabled
+            if ParserElement._packratEnabled:
+                self._save_context[
+                    "packrat_cache_size"
+                ] = ParserElement.packrat_cache.size
+            else:
+                self._save_context["packrat_cache_size"] = None
+            self._save_context["packrat_parse"] = ParserElement._parse
+            self._save_context[
+                "recursion_enabled"
+            ] = ParserElement._left_recursion_enabled
+
+            self._save_context["__diag__"] = {
+                name: getattr(__diag__, name) for name in __diag__._all_names
+            }
+
+            self._save_context["__compat__"] = {
+                "collect_all_And_tokens": __compat__.collect_all_And_tokens
+            }
+
+            return self
+
+        def restore(self):
+            # reset pyparsing global state
+            if (
+                ParserElement.DEFAULT_WHITE_CHARS
+                != self._save_context["default_whitespace"]
+            ):
+                ParserElement.set_default_whitespace_chars(
+                    self._save_context["default_whitespace"]
+                )
+
+            ParserElement.verbose_stacktrace = self._save_context["verbose_stacktrace"]
+
+            Keyword.DEFAULT_KEYWORD_CHARS = self._save_context["default_keyword_chars"]
+            ParserElement.inlineLiteralsUsing(
+                self._save_context["literal_string_class"]
+            )
+
+            for name, value in self._save_context["__diag__"].items():
+                (__diag__.enable if value else __diag__.disable)(name)
+
+            ParserElement._packratEnabled = False
+            if self._save_context["packrat_enabled"]:
+                ParserElement.enable_packrat(self._save_context["packrat_cache_size"])
+            else:
+                ParserElement._parse = self._save_context["packrat_parse"]
+            ParserElement._left_recursion_enabled = self._save_context[
+                "recursion_enabled"
+            ]
+
+            __compat__.collect_all_And_tokens = self._save_context["__compat__"]
+
+            return self
+
+        def copy(self):
+            ret = type(self)()
+            ret._save_context.update(self._save_context)
+            return ret
+
+        def __enter__(self):
+            return self.save()
+
+        def __exit__(self, *args):
+            self.restore()
+
+    class TestParseResultsAsserts:
+        """
+        A mixin class to add parse results assertion methods to normal unittest.TestCase classes.
+        """
+
+        def assertParseResultsEquals(
+            self, result, expected_list=None, expected_dict=None, msg=None
+        ):
+            """
+            Unit test assertion to compare a :class:`ParseResults` object with an optional ``expected_list``,
+            and compare any defined results names with an optional ``expected_dict``.
+            """
+            if expected_list is not None:
+                self.assertEqual(expected_list, result.as_list(), msg=msg)
+            if expected_dict is not None:
+                self.assertEqual(expected_dict, result.as_dict(), msg=msg)
+
+        def assertParseAndCheckList(
+            self, expr, test_string, expected_list, msg=None, verbose=True
+        ):
+            """
+            Convenience wrapper assert to test a parser element and input string, and assert that
+            the resulting ``ParseResults.asList()`` is equal to the ``expected_list``.
+            """
+            result = expr.parse_string(test_string, parse_all=True)
+            if verbose:
+                print(result.dump())
+            else:
+                print(result.as_list())
+            self.assertParseResultsEquals(result, expected_list=expected_list, msg=msg)
+
+        def assertParseAndCheckDict(
+            self, expr, test_string, expected_dict, msg=None, verbose=True
+        ):
+            """
+            Convenience wrapper assert to test a parser element and input string, and assert that
+            the resulting ``ParseResults.asDict()`` is equal to the ``expected_dict``.
+            """
+            result = expr.parse_string(test_string, parseAll=True)
+            if verbose:
+                print(result.dump())
+            else:
+                print(result.as_list())
+            self.assertParseResultsEquals(result, expected_dict=expected_dict, msg=msg)
+
+        def assertRunTestResults(
+            self, run_tests_report, expected_parse_results=None, msg=None
+        ):
+            """
+            Unit test assertion to evaluate output of ``ParserElement.runTests()``. If a list of
+            list-dict tuples is given as the ``expected_parse_results`` argument, then these are zipped
+            with the report tuples returned by ``runTests`` and evaluated using ``assertParseResultsEquals``.
+            Finally, asserts that the overall ``runTests()`` success value is ``True``.
+
+            :param run_tests_report: tuple(bool, [tuple(str, ParseResults or Exception)]) returned from runTests
+            :param expected_parse_results (optional): [tuple(str, list, dict, Exception)]
+            """
+            run_test_success, run_test_results = run_tests_report
+
+            if expected_parse_results is not None:
+                merged = [
+                    (*rpt, expected)
+                    for rpt, expected in zip(run_test_results, expected_parse_results)
+                ]
+                for test_string, result, expected in merged:
+                    # expected should be a tuple containing a list and/or a dict or an exception,
+                    # and optional failure message string
+                    # an empty tuple will skip any result validation
+                    fail_msg = next(
+                        (exp for exp in expected if isinstance(exp, str)), None
+                    )
+                    expected_exception = next(
+                        (
+                            exp
+                            for exp in expected
+                            if isinstance(exp, type) and issubclass(exp, Exception)
+                        ),
+                        None,
+                    )
+                    if expected_exception is not None:
+                        with self.assertRaises(
+                            expected_exception=expected_exception, msg=fail_msg or msg
+                        ):
+                            if isinstance(result, Exception):
+                                raise result
+                    else:
+                        expected_list = next(
+                            (exp for exp in expected if isinstance(exp, list)), None
+                        )
+                        expected_dict = next(
+                            (exp for exp in expected if isinstance(exp, dict)), None
+                        )
+                        if (expected_list, expected_dict) != (None, None):
+                            self.assertParseResultsEquals(
+                                result,
+                                expected_list=expected_list,
+                                expected_dict=expected_dict,
+                                msg=fail_msg or msg,
+                            )
+                        else:
+                            # warning here maybe?
+                            print(f"no validation for {test_string!r}")
+
+            # do this last, in case some specific test results can be reported instead
+            self.assertTrue(
+                run_test_success, msg=msg if msg is not None else "failed runTests"
+            )
+
+        @contextmanager
+        def assertRaisesParseException(self, exc_type=ParseException, msg=None):
+            with self.assertRaises(exc_type, msg=msg):
+                yield
+
+    @staticmethod
+    def with_line_numbers(
+        s: str,
+        start_line: typing.Optional[int] = None,
+        end_line: typing.Optional[int] = None,
+        expand_tabs: bool = True,
+        eol_mark: str = "|",
+        mark_spaces: typing.Optional[str] = None,
+        mark_control: typing.Optional[str] = None,
+    ) -> str:
+        """
+        Helpful method for debugging a parser - prints a string with line and column numbers.
+        (Line and column numbers are 1-based.)
+
+        :param s: tuple(bool, str - string to be printed with line and column numbers
+        :param start_line: int - (optional) starting line number in s to print (default=1)
+        :param end_line: int - (optional) ending line number in s to print (default=len(s))
+        :param expand_tabs: bool - (optional) expand tabs to spaces, to match the pyparsing default
+        :param eol_mark: str - (optional) string to mark the end of lines, helps visualize trailing spaces (default="|")
+        :param mark_spaces: str - (optional) special character to display in place of spaces
+        :param mark_control: str - (optional) convert non-printing control characters to a placeholding
+                                 character; valid values:
+                                 - "unicode" - replaces control chars with Unicode symbols, such as "␍" and "␊"
+                                 - any single character string - replace control characters with given string
+                                 - None (default) - string is displayed as-is
+
+        :return: str - input string with leading line numbers and column number headers
+        """
+        if expand_tabs:
+            s = s.expandtabs()
+        if mark_control is not None:
+            mark_control = typing.cast(str, mark_control)
+            if mark_control == "unicode":
+                transtable_map = {
+                    c: u for c, u in zip(range(0, 33), range(0x2400, 0x2433))
+                }
+                transtable_map[127] = 0x2421
+                tbl = str.maketrans(transtable_map)
+                eol_mark = ""
+            else:
+                ord_mark_control = ord(mark_control)
+                tbl = str.maketrans(
+                    {c: ord_mark_control for c in list(range(0, 32)) + [127]}
+                )
+            s = s.translate(tbl)
+        if mark_spaces is not None and mark_spaces != " ":
+            if mark_spaces == "unicode":
+                tbl = str.maketrans({9: 0x2409, 32: 0x2423})
+                s = s.translate(tbl)
+            else:
+                s = s.replace(" ", mark_spaces)
+        if start_line is None:
+            start_line = 1
+        if end_line is None:
+            end_line = len(s)
+        end_line = min(end_line, len(s))
+        start_line = min(max(1, start_line), end_line)
+
+        if mark_control != "unicode":
+            s_lines = s.splitlines()[start_line - 1 : end_line]
+        else:
+            s_lines = [line + "␊" for line in s.split("␊")[start_line - 1 : end_line]]
+        if not s_lines:
+            return ""
+
+        lineno_width = len(str(end_line))
+        max_line_len = max(len(line) for line in s_lines)
+        lead = " " * (lineno_width + 1)
+        if max_line_len >= 99:
+            header0 = (
+                lead
+                + "".join(
+                    f"{' ' * 99}{(i + 1) % 100}"
+                    for i in range(max(max_line_len // 100, 1))
+                )
+                + "\n"
+            )
+        else:
+            header0 = ""
+        header1 = (
+            header0
+            + lead
+            + "".join(f"         {(i + 1) % 10}" for i in range(-(-max_line_len // 10)))
+            + "\n"
+        )
+        header2 = lead + "1234567890" * (-(-max_line_len // 10)) + "\n"
+        return (
+            header1
+            + header2
+            + "\n".join(
+                f"{i:{lineno_width}d}:{line}{eol_mark}"
+                for i, line in enumerate(s_lines, start=start_line)
+            )
+            + "\n"
+        )

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/unicode.py

@@ -0,0 +1,361 @@
+# unicode.py
+
+import sys
+from itertools import filterfalse
+from typing import List, Tuple, Union
+
+
+class _lazyclassproperty:
+    def __init__(self, fn):
+        self.fn = fn
+        self.__doc__ = fn.__doc__
+        self.__name__ = fn.__name__
+
+    def __get__(self, obj, cls):
+        if cls is None:
+            cls = type(obj)
+        if not hasattr(cls, "_intern") or any(
+            cls._intern is getattr(superclass, "_intern", [])
+            for superclass in cls.__mro__[1:]
+        ):
+            cls._intern = {}
+        attrname = self.fn.__name__
+        if attrname not in cls._intern:
+            cls._intern[attrname] = self.fn(cls)
+        return cls._intern[attrname]
+
+
+UnicodeRangeList = List[Union[Tuple[int, int], Tuple[int]]]
+
+
+class unicode_set:
+    """
+    A set of Unicode characters, for language-specific strings for
+    ``alphas``, ``nums``, ``alphanums``, and ``printables``.
+    A unicode_set is defined by a list of ranges in the Unicode character
+    set, in a class attribute ``_ranges``. Ranges can be specified using
+    2-tuples or a 1-tuple, such as::
+
+        _ranges = [
+            (0x0020, 0x007e),
+            (0x00a0, 0x00ff),
+            (0x0100,),
+            ]
+
+    Ranges are left- and right-inclusive. A 1-tuple of (x,) is treated as (x, x).
+
+    A unicode set can also be defined using multiple inheritance of other unicode sets::
+
+        class CJK(Chinese, Japanese, Korean):
+            pass
+    """
+
+    _ranges: UnicodeRangeList = []
+
+    @_lazyclassproperty
+    def _chars_for_ranges(cls):
+        ret = []
+        for cc in cls.__mro__:
+            if cc is unicode_set:
+                break
+            for rr in getattr(cc, "_ranges", ()):
+                ret.extend(range(rr[0], rr[-1] + 1))
+        return [chr(c) for c in sorted(set(ret))]
+
+    @_lazyclassproperty
+    def printables(cls):
+        """all non-whitespace characters in this range"""
+        return "".join(filterfalse(str.isspace, cls._chars_for_ranges))
+
+    @_lazyclassproperty
+    def alphas(cls):
+        """all alphabetic characters in this range"""
+        return "".join(filter(str.isalpha, cls._chars_for_ranges))
+
+    @_lazyclassproperty
+    def nums(cls):
+        """all numeric digit characters in this range"""
+        return "".join(filter(str.isdigit, cls._chars_for_ranges))
+
+    @_lazyclassproperty
+    def alphanums(cls):
+        """all alphanumeric characters in this range"""
+        return cls.alphas + cls.nums
+
+    @_lazyclassproperty
+    def identchars(cls):
+        """all characters in this range that are valid identifier characters, plus underscore '_'"""
+        return "".join(
+            sorted(
+                set(
+                    "".join(filter(str.isidentifier, cls._chars_for_ranges))
+                    + "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyzªµº"
+                    + "ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ"
+                    + "_"
+                )
+            )
+        )
+
+    @_lazyclassproperty
+    def identbodychars(cls):
+        """
+        all characters in this range that are valid identifier body characters,
+        plus the digits 0-9, and · (Unicode MIDDLE DOT)
+        """
+        return "".join(
+            sorted(
+                set(
+                    cls.identchars
+                    + "0123456789·"
+                    + "".join(
+                        [c for c in cls._chars_for_ranges if ("_" + c).isidentifier()]
+                    )
+                )
+            )
+        )
+
+    @_lazyclassproperty
+    def identifier(cls):
+        """
+        a pyparsing Word expression for an identifier using this range's definitions for
+        identchars and identbodychars
+        """
+        from pip._vendor.pyparsing import Word
+
+        return Word(cls.identchars, cls.identbodychars)
+
+
+class pyparsing_unicode(unicode_set):
+    """
+    A namespace class for defining common language unicode_sets.
+    """
+
+    # fmt: off
+
+    # define ranges in language character sets
+    _ranges: UnicodeRangeList = [
+        (0x0020, sys.maxunicode),
+    ]
+
+    class BasicMultilingualPlane(unicode_set):
+        """Unicode set for the Basic Multilingual Plane"""
+        _ranges: UnicodeRangeList = [
+            (0x0020, 0xFFFF),
+        ]
+
+    class Latin1(unicode_set):
+        """Unicode set for Latin-1 Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0020, 0x007E),
+            (0x00A0, 0x00FF),
+        ]
+
+    class LatinA(unicode_set):
+        """Unicode set for Latin-A Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0100, 0x017F),
+        ]
+
+    class LatinB(unicode_set):
+        """Unicode set for Latin-B Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0180, 0x024F),
+        ]
+
+    class Greek(unicode_set):
+        """Unicode set for Greek Unicode Character Ranges"""
+        _ranges: UnicodeRangeList = [
+            (0x0342, 0x0345),
+            (0x0370, 0x0377),
+            (0x037A, 0x037F),
+            (0x0384, 0x038A),
+            (0x038C,),
+            (0x038E, 0x03A1),
+            (0x03A3, 0x03E1),
+            (0x03F0, 0x03FF),
+            (0x1D26, 0x1D2A),
+            (0x1D5E,),
+            (0x1D60,),
+            (0x1D66, 0x1D6A),
+            (0x1F00, 0x1F15),
+            (0x1F18, 0x1F1D),
+            (0x1F20, 0x1F45),
+            (0x1F48, 0x1F4D),
+            (0x1F50, 0x1F57),
+            (0x1F59,),
+            (0x1F5B,),
+            (0x1F5D,),
+            (0x1F5F, 0x1F7D),
+            (0x1F80, 0x1FB4),
+            (0x1FB6, 0x1FC4),
+            (0x1FC6, 0x1FD3),
+            (0x1FD6, 0x1FDB),
+            (0x1FDD, 0x1FEF),
+            (0x1FF2, 0x1FF4),
+            (0x1FF6, 0x1FFE),
+            (0x2129,),
+            (0x2719, 0x271A),
+            (0xAB65,),
+            (0x10140, 0x1018D),
+            (0x101A0,),
+            (0x1D200, 0x1D245),
+            (0x1F7A1, 0x1F7A7),
+        ]
+
+    class Cyrillic(unicode_set):
+        """Unicode set for Cyrillic Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0400, 0x052F),
+            (0x1C80, 0x1C88),
+            (0x1D2B,),
+            (0x1D78,),
+            (0x2DE0, 0x2DFF),
+            (0xA640, 0xA672),
+            (0xA674, 0xA69F),
+            (0xFE2E, 0xFE2F),
+        ]
+
+    class Chinese(unicode_set):
+        """Unicode set for Chinese Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x2E80, 0x2E99),
+            (0x2E9B, 0x2EF3),
+            (0x31C0, 0x31E3),
+            (0x3400, 0x4DB5),
+            (0x4E00, 0x9FEF),
+            (0xA700, 0xA707),
+            (0xF900, 0xFA6D),
+            (0xFA70, 0xFAD9),
+            (0x16FE2, 0x16FE3),
+            (0x1F210, 0x1F212),
+            (0x1F214, 0x1F23B),
+            (0x1F240, 0x1F248),
+            (0x20000, 0x2A6D6),
+            (0x2A700, 0x2B734),
+            (0x2B740, 0x2B81D),
+            (0x2B820, 0x2CEA1),
+            (0x2CEB0, 0x2EBE0),
+            (0x2F800, 0x2FA1D),
+        ]
+
+    class Japanese(unicode_set):
+        """Unicode set for Japanese Unicode Character Range, combining Kanji, Hiragana, and Katakana ranges"""
+
+        class Kanji(unicode_set):
+            "Unicode set for Kanji Unicode Character Range"
+            _ranges: UnicodeRangeList = [
+                (0x4E00, 0x9FBF),
+                (0x3000, 0x303F),
+            ]
+
+        class Hiragana(unicode_set):
+            """Unicode set for Hiragana Unicode Character Range"""
+            _ranges: UnicodeRangeList = [
+                (0x3041, 0x3096),
+                (0x3099, 0x30A0),
+                (0x30FC,),
+                (0xFF70,),
+                (0x1B001,),
+                (0x1B150, 0x1B152),
+                (0x1F200,),
+            ]
+
+        class Katakana(unicode_set):
+            """Unicode set for Katakana  Unicode Character Range"""
+            _ranges: UnicodeRangeList = [
+                (0x3099, 0x309C),
+                (0x30A0, 0x30FF),
+                (0x31F0, 0x31FF),
+                (0x32D0, 0x32FE),
+                (0xFF65, 0xFF9F),
+                (0x1B000,),
+                (0x1B164, 0x1B167),
+                (0x1F201, 0x1F202),
+                (0x1F213,),
+            ]
+
+        漢字 = Kanji
+        カタカナ = Katakana
+        ひらがな = Hiragana
+
+        _ranges = (
+            Kanji._ranges
+            + Hiragana._ranges
+            + Katakana._ranges
+        )
+
+    class Hangul(unicode_set):
+        """Unicode set for Hangul (Korean) Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x1100, 0x11FF),
+            (0x302E, 0x302F),
+            (0x3131, 0x318E),
+            (0x3200, 0x321C),
+            (0x3260, 0x327B),
+            (0x327E,),
+            (0xA960, 0xA97C),
+            (0xAC00, 0xD7A3),
+            (0xD7B0, 0xD7C6),
+            (0xD7CB, 0xD7FB),
+            (0xFFA0, 0xFFBE),
+            (0xFFC2, 0xFFC7),
+            (0xFFCA, 0xFFCF),
+            (0xFFD2, 0xFFD7),
+            (0xFFDA, 0xFFDC),
+        ]
+
+    Korean = Hangul
+
+    class CJK(Chinese, Japanese, Hangul):
+        """Unicode set for combined Chinese, Japanese, and Korean (CJK) Unicode Character Range"""
+
+    class Thai(unicode_set):
+        """Unicode set for Thai Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0E01, 0x0E3A),
+            (0x0E3F, 0x0E5B)
+        ]
+
+    class Arabic(unicode_set):
+        """Unicode set for Arabic Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0600, 0x061B),
+            (0x061E, 0x06FF),
+            (0x0700, 0x077F),
+        ]
+
+    class Hebrew(unicode_set):
+        """Unicode set for Hebrew Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0591, 0x05C7),
+            (0x05D0, 0x05EA),
+            (0x05EF, 0x05F4),
+            (0xFB1D, 0xFB36),
+            (0xFB38, 0xFB3C),
+            (0xFB3E,),
+            (0xFB40, 0xFB41),
+            (0xFB43, 0xFB44),
+            (0xFB46, 0xFB4F),
+        ]
+
+    class Devanagari(unicode_set):
+        """Unicode set for Devanagari Unicode Character Range"""
+        _ranges: UnicodeRangeList = [
+            (0x0900, 0x097F),
+            (0xA8E0, 0xA8FF)
+        ]
+
+    BMP = BasicMultilingualPlane
+
+    # add language identifiers using language Unicode
+    العربية = Arabic
+    中文 = Chinese
+    кириллица = Cyrillic
+    Ελληνικά = Greek
+    עִברִית = Hebrew
+    日本語 = Japanese
+    한국어 = Korean
+    ไทย = Thai
+    देवनागरी = Devanagari
+
+    # fmt: on

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/pyparsing/util.py

@@ -0,0 +1,284 @@
+# util.py
+import inspect
+import warnings
+import types
+import collections
+import itertools
+from functools import lru_cache, wraps
+from typing import Callable, List, Union, Iterable, TypeVar, cast
+
+_bslash = chr(92)
+C = TypeVar("C", bound=Callable)
+
+
+class __config_flags:
+    """Internal class for defining compatibility and debugging flags"""
+
+    _all_names: List[str] = []
+    _fixed_names: List[str] = []
+    _type_desc = "configuration"
+
+    @classmethod
+    def _set(cls, dname, value):
+        if dname in cls._fixed_names:
+            warnings.warn(
+                f"{cls.__name__}.{dname} {cls._type_desc} is {str(getattr(cls, dname)).upper()}"
+                f" and cannot be overridden",
+                stacklevel=3,
+            )
+            return
+        if dname in cls._all_names:
+            setattr(cls, dname, value)
+        else:
+            raise ValueError(f"no such {cls._type_desc} {dname!r}")
+
+    enable = classmethod(lambda cls, name: cls._set(name, True))
+    disable = classmethod(lambda cls, name: cls._set(name, False))
+
+
+@lru_cache(maxsize=128)
+def col(loc: int, strg: str) -> int:
+    """
+    Returns current column within a string, counting newlines as line separators.
+    The first column is number 1.
+
+    Note: the default parsing behavior is to expand tabs in the input string
+    before starting the parsing process.  See
+    :class:`ParserElement.parse_string` for more
+    information on parsing strings containing ``<TAB>`` s, and suggested
+    methods to maintain a consistent view of the parsed string, the parse
+    location, and line and column positions within the parsed string.
+    """
+    s = strg
+    return 1 if 0 < loc < len(s) and s[loc - 1] == "\n" else loc - s.rfind("\n", 0, loc)
+
+
+@lru_cache(maxsize=128)
+def lineno(loc: int, strg: str) -> int:
+    """Returns current line number within a string, counting newlines as line separators.
+    The first line is number 1.
+
+    Note - the default parsing behavior is to expand tabs in the input string
+    before starting the parsing process.  See :class:`ParserElement.parse_string`
+    for more information on parsing strings containing ``<TAB>`` s, and
+    suggested methods to maintain a consistent view of the parsed string, the
+    parse location, and line and column positions within the parsed string.
+    """
+    return strg.count("\n", 0, loc) + 1
+
+
+@lru_cache(maxsize=128)
+def line(loc: int, strg: str) -> str:
+    """
+    Returns the line of text containing loc within a string, counting newlines as line separators.
+    """
+    last_cr = strg.rfind("\n", 0, loc)
+    next_cr = strg.find("\n", loc)
+    return strg[last_cr + 1 : next_cr] if next_cr >= 0 else strg[last_cr + 1 :]
+
+
+class _UnboundedCache:
+    def __init__(self):
+        cache = {}
+        cache_get = cache.get
+        self.not_in_cache = not_in_cache = object()
+
+        def get(_, key):
+            return cache_get(key, not_in_cache)
+
+        def set_(_, key, value):
+            cache[key] = value
+
+        def clear(_):
+            cache.clear()
+
+        self.size = None
+        self.get = types.MethodType(get, self)
+        self.set = types.MethodType(set_, self)
+        self.clear = types.MethodType(clear, self)
+
+
+class _FifoCache:
+    def __init__(self, size):
+        self.not_in_cache = not_in_cache = object()
+        cache = {}
+        keyring = [object()] * size
+        cache_get = cache.get
+        cache_pop = cache.pop
+        keyiter = itertools.cycle(range(size))
+
+        def get(_, key):
+            return cache_get(key, not_in_cache)
+
+        def set_(_, key, value):
+            cache[key] = value
+            i = next(keyiter)
+            cache_pop(keyring[i], None)
+            keyring[i] = key
+
+        def clear(_):
+            cache.clear()
+            keyring[:] = [object()] * size
+
+        self.size = size
+        self.get = types.MethodType(get, self)
+        self.set = types.MethodType(set_, self)
+        self.clear = types.MethodType(clear, self)
+
+
+class LRUMemo:
+    """
+    A memoizing mapping that retains `capacity` deleted items
+
+    The memo tracks retained items by their access order; once `capacity` items
+    are retained, the least recently used item is discarded.
+    """
+
+    def __init__(self, capacity):
+        self._capacity = capacity
+        self._active = {}
+        self._memory = collections.OrderedDict()
+
+    def __getitem__(self, key):
+        try:
+            return self._active[key]
+        except KeyError:
+            self._memory.move_to_end(key)
+            return self._memory[key]
+
+    def __setitem__(self, key, value):
+        self._memory.pop(key, None)
+        self._active[key] = value
+
+    def __delitem__(self, key):
+        try:
+            value = self._active.pop(key)
+        except KeyError:
+            pass
+        else:
+            while len(self._memory) >= self._capacity:
+                self._memory.popitem(last=False)
+            self._memory[key] = value
+
+    def clear(self):
+        self._active.clear()
+        self._memory.clear()
+
+
+class UnboundedMemo(dict):
+    """
+    A memoizing mapping that retains all deleted items
+    """
+
+    def __delitem__(self, key):
+        pass
+
+
+def _escape_regex_range_chars(s: str) -> str:
+    # escape these chars: ^-[]
+    for c in r"\^-[]":
+        s = s.replace(c, _bslash + c)
+    s = s.replace("\n", r"\n")
+    s = s.replace("\t", r"\t")
+    return str(s)
+
+
+def _collapse_string_to_ranges(
+    s: Union[str, Iterable[str]], re_escape: bool = True
+) -> str:
+    def is_consecutive(c):
+        c_int = ord(c)
+        is_consecutive.prev, prev = c_int, is_consecutive.prev
+        if c_int - prev > 1:
+            is_consecutive.value = next(is_consecutive.counter)
+        return is_consecutive.value
+
+    is_consecutive.prev = 0  # type: ignore [attr-defined]
+    is_consecutive.counter = itertools.count()  # type: ignore [attr-defined]
+    is_consecutive.value = -1  # type: ignore [attr-defined]
+
+    def escape_re_range_char(c):
+        return "\\" + c if c in r"\^-][" else c
+
+    def no_escape_re_range_char(c):
+        return c
+
+    if not re_escape:
+        escape_re_range_char = no_escape_re_range_char
+
+    ret = []
+    s = "".join(sorted(set(s)))
+    if len(s) > 3:
+        for _, chars in itertools.groupby(s, key=is_consecutive):
+            first = last = next(chars)
+            last = collections.deque(
+                itertools.chain(iter([last]), chars), maxlen=1
+            ).pop()
+            if first == last:
+                ret.append(escape_re_range_char(first))
+            else:
+                sep = "" if ord(last) == ord(first) + 1 else "-"
+                ret.append(
+                    f"{escape_re_range_char(first)}{sep}{escape_re_range_char(last)}"
+                )
+    else:
+        ret = [escape_re_range_char(c) for c in s]
+
+    return "".join(ret)
+
+
+def _flatten(ll: list) -> list:
+    ret = []
+    for i in ll:
+        if isinstance(i, list):
+            ret.extend(_flatten(i))
+        else:
+            ret.append(i)
+    return ret
+
+
+def _make_synonym_function(compat_name: str, fn: C) -> C:
+    # In a future version, uncomment the code in the internal _inner() functions
+    # to begin emitting DeprecationWarnings.
+
+    # Unwrap staticmethod/classmethod
+    fn = getattr(fn, "__func__", fn)
+
+    # (Presence of 'self' arg in signature is used by explain_exception() methods, so we take
+    # some extra steps to add it if present in decorated function.)
+    if "self" == list(inspect.signature(fn).parameters)[0]:
+
+        @wraps(fn)
+        def _inner(self, *args, **kwargs):
+            # warnings.warn(
+            #     f"Deprecated - use {fn.__name__}", DeprecationWarning, stacklevel=3
+            # )
+            return fn(self, *args, **kwargs)
+
+    else:
+
+        @wraps(fn)
+        def _inner(*args, **kwargs):
+            # warnings.warn(
+            #     f"Deprecated - use {fn.__name__}", DeprecationWarning, stacklevel=3
+            # )
+            return fn(*args, **kwargs)
+
+    _inner.__doc__ = f"""Deprecated - use :class:`{fn.__name__}`"""
+    _inner.__name__ = compat_name
+    _inner.__annotations__ = fn.__annotations__
+    if isinstance(fn, types.FunctionType):
+        _inner.__kwdefaults__ = fn.__kwdefaults__
+    elif isinstance(fn, type) and hasattr(fn, "__init__"):
+        _inner.__kwdefaults__ = fn.__init__.__kwdefaults__
+    else:
+        _inner.__kwdefaults__ = None
+    _inner.__qualname__ = fn.__qualname__
+    return cast(C, _inner)
+
+
+def replaced_by_pep8(fn: C) -> Callable[[Callable], C]:
+    """
+    Decorator for pre-PEP8 compatibility synonyms, to link them to the new function.
+    """
+    return lambda other: _make_synonym_function(other.__name__, fn)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/__init__.py

@@ -0,0 +1,102 @@
+"""
+Python HTTP library with thread-safe connection pooling, file post support, user friendly, and more
+"""
+from __future__ import absolute_import
+
+# Set default logging handler to avoid "No handler found" warnings.
+import logging
+import warnings
+from logging import NullHandler
+
+from . import exceptions
+from ._version import __version__
+from .connectionpool import HTTPConnectionPool, HTTPSConnectionPool, connection_from_url
+from .filepost import encode_multipart_formdata
+from .poolmanager import PoolManager, ProxyManager, proxy_from_url
+from .response import HTTPResponse
+from .util.request import make_headers
+from .util.retry import Retry
+from .util.timeout import Timeout
+from .util.url import get_host
+
+# === NOTE TO REPACKAGERS AND VENDORS ===
+# Please delete this block, this logic is only
+# for urllib3 being distributed via PyPI.
+# See: https://github.com/urllib3/urllib3/issues/2680
+try:
+    import urllib3_secure_extra  # type: ignore # noqa: F401
+except ImportError:
+    pass
+else:
+    warnings.warn(
+        "'urllib3[secure]' extra is deprecated and will be removed "
+        "in a future release of urllib3 2.x. Read more in this issue: "
+        "https://github.com/urllib3/urllib3/issues/2680",
+        category=DeprecationWarning,
+        stacklevel=2,
+    )
+
+__author__ = "Andrey Petrov (andrey.petrov@shazow.net)"
+__license__ = "MIT"
+__version__ = __version__
+
+__all__ = (
+    "HTTPConnectionPool",
+    "HTTPSConnectionPool",
+    "PoolManager",
+    "ProxyManager",
+    "HTTPResponse",
+    "Retry",
+    "Timeout",
+    "add_stderr_logger",
+    "connection_from_url",
+    "disable_warnings",
+    "encode_multipart_formdata",
+    "get_host",
+    "make_headers",
+    "proxy_from_url",
+)
+
+logging.getLogger(__name__).addHandler(NullHandler())
+
+
+def add_stderr_logger(level=logging.DEBUG):
+    """
+    Helper for quickly adding a StreamHandler to the logger. Useful for
+    debugging.
+
+    Returns the handler after adding it.
+    """
+    # This method needs to be in this __init__.py to get the __name__ correct
+    # even if urllib3 is vendored within another package.
+    logger = logging.getLogger(__name__)
+    handler = logging.StreamHandler()
+    handler.setFormatter(logging.Formatter("%(asctime)s %(levelname)s %(message)s"))
+    logger.addHandler(handler)
+    logger.setLevel(level)
+    logger.debug("Added a stderr logging handler to logger: %s", __name__)
+    return handler
+
+
+# ... Clean up.
+del NullHandler
+
+
+# All warning filters *must* be appended unless you're really certain that they
+# shouldn't be: otherwise, it's very hard for users to use most Python
+# mechanisms to silence them.
+# SecurityWarning's always go off by default.
+warnings.simplefilter("always", exceptions.SecurityWarning, append=True)
+# SubjectAltNameWarning's should go off once per host
+warnings.simplefilter("default", exceptions.SubjectAltNameWarning, append=True)
+# InsecurePlatformWarning's don't vary between requests, so we keep it default.
+warnings.simplefilter("default", exceptions.InsecurePlatformWarning, append=True)
+# SNIMissingWarnings should go off only once.
+warnings.simplefilter("default", exceptions.SNIMissingWarning, append=True)
+
+
+def disable_warnings(category=exceptions.HTTPWarning):
+    """
+    Helper for quickly disabling all urllib3 warnings.
+    """
+    warnings.simplefilter("ignore", category)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/_collections.py

@@ -0,0 +1,337 @@
+from __future__ import absolute_import
+
+try:
+    from collections.abc import Mapping, MutableMapping
+except ImportError:
+    from collections import Mapping, MutableMapping
+try:
+    from threading import RLock
+except ImportError:  # Platform-specific: No threads available
+
+    class RLock:
+        def __enter__(self):
+            pass
+
+        def __exit__(self, exc_type, exc_value, traceback):
+            pass
+
+
+from collections import OrderedDict
+
+from .exceptions import InvalidHeader
+from .packages import six
+from .packages.six import iterkeys, itervalues
+
+__all__ = ["RecentlyUsedContainer", "HTTPHeaderDict"]
+
+
+_Null = object()
+
+
+class RecentlyUsedContainer(MutableMapping):
+    """
+    Provides a thread-safe dict-like container which maintains up to
+    ``maxsize`` keys while throwing away the least-recently-used keys beyond
+    ``maxsize``.
+
+    :param maxsize:
+        Maximum number of recent elements to retain.
+
+    :param dispose_func:
+        Every time an item is evicted from the container,
+        ``dispose_func(value)`` is called.  Callback which will get called
+    """
+
+    ContainerCls = OrderedDict
+
+    def __init__(self, maxsize=10, dispose_func=None):
+        self._maxsize = maxsize
+        self.dispose_func = dispose_func
+
+        self._container = self.ContainerCls()
+        self.lock = RLock()
+
+    def __getitem__(self, key):
+        # Re-insert the item, moving it to the end of the eviction line.
+        with self.lock:
+            item = self._container.pop(key)
+            self._container[key] = item
+            return item
+
+    def __setitem__(self, key, value):
+        evicted_value = _Null
+        with self.lock:
+            # Possibly evict the existing value of 'key'
+            evicted_value = self._container.get(key, _Null)
+            self._container[key] = value
+
+            # If we didn't evict an existing value, we might have to evict the
+            # least recently used item from the beginning of the container.
+            if len(self._container) > self._maxsize:
+                _key, evicted_value = self._container.popitem(last=False)
+
+        if self.dispose_func and evicted_value is not _Null:
+            self.dispose_func(evicted_value)
+
+    def __delitem__(self, key):
+        with self.lock:
+            value = self._container.pop(key)
+
+        if self.dispose_func:
+            self.dispose_func(value)
+
+    def __len__(self):
+        with self.lock:
+            return len(self._container)
+
+    def __iter__(self):
+        raise NotImplementedError(
+            "Iteration over this class is unlikely to be threadsafe."
+        )
+
+    def clear(self):
+        with self.lock:
+            # Copy pointers to all values, then wipe the mapping
+            values = list(itervalues(self._container))
+            self._container.clear()
+
+        if self.dispose_func:
+            for value in values:
+                self.dispose_func(value)
+
+    def keys(self):
+        with self.lock:
+            return list(iterkeys(self._container))
+
+
+class HTTPHeaderDict(MutableMapping):
+    """
+    :param headers:
+        An iterable of field-value pairs. Must not contain multiple field names
+        when compared case-insensitively.
+
+    :param kwargs:
+        Additional field-value pairs to pass in to ``dict.update``.
+
+    A ``dict`` like container for storing HTTP Headers.
+
+    Field names are stored and compared case-insensitively in compliance with
+    RFC 7230. Iteration provides the first case-sensitive key seen for each
+    case-insensitive pair.
+
+    Using ``__setitem__`` syntax overwrites fields that compare equal
+    case-insensitively in order to maintain ``dict``'s api. For fields that
+    compare equal, instead create a new ``HTTPHeaderDict`` and use ``.add``
+    in a loop.
+
+    If multiple fields that are equal case-insensitively are passed to the
+    constructor or ``.update``, the behavior is undefined and some will be
+    lost.
+
+    >>> headers = HTTPHeaderDict()
+    >>> headers.add('Set-Cookie', 'foo=bar')
+    >>> headers.add('set-cookie', 'baz=quxx')
+    >>> headers['content-length'] = '7'
+    >>> headers['SET-cookie']
+    'foo=bar, baz=quxx'
+    >>> headers['Content-Length']
+    '7'
+    """
+
+    def __init__(self, headers=None, **kwargs):
+        super(HTTPHeaderDict, self).__init__()
+        self._container = OrderedDict()
+        if headers is not None:
+            if isinstance(headers, HTTPHeaderDict):
+                self._copy_from(headers)
+            else:
+                self.extend(headers)
+        if kwargs:
+            self.extend(kwargs)
+
+    def __setitem__(self, key, val):
+        self._container[key.lower()] = [key, val]
+        return self._container[key.lower()]
+
+    def __getitem__(self, key):
+        val = self._container[key.lower()]
+        return ", ".join(val[1:])
+
+    def __delitem__(self, key):
+        del self._container[key.lower()]
+
+    def __contains__(self, key):
+        return key.lower() in self._container
+
+    def __eq__(self, other):
+        if not isinstance(other, Mapping) and not hasattr(other, "keys"):
+            return False
+        if not isinstance(other, type(self)):
+            other = type(self)(other)
+        return dict((k.lower(), v) for k, v in self.itermerged()) == dict(
+            (k.lower(), v) for k, v in other.itermerged()
+        )
+
+    def __ne__(self, other):
+        return not self.__eq__(other)
+
+    if six.PY2:  # Python 2
+        iterkeys = MutableMapping.iterkeys
+        itervalues = MutableMapping.itervalues
+
+    __marker = object()
+
+    def __len__(self):
+        return len(self._container)
+
+    def __iter__(self):
+        # Only provide the originally cased names
+        for vals in self._container.values():
+            yield vals[0]
+
+    def pop(self, key, default=__marker):
+        """D.pop(k[,d]) -> v, remove specified key and return the corresponding value.
+        If key is not found, d is returned if given, otherwise KeyError is raised.
+        """
+        # Using the MutableMapping function directly fails due to the private marker.
+        # Using ordinary dict.pop would expose the internal structures.
+        # So let's reinvent the wheel.
+        try:
+            value = self[key]
+        except KeyError:
+            if default is self.__marker:
+                raise
+            return default
+        else:
+            del self[key]
+            return value
+
+    def discard(self, key):
+        try:
+            del self[key]
+        except KeyError:
+            pass
+
+    def add(self, key, val):
+        """Adds a (name, value) pair, doesn't overwrite the value if it already
+        exists.
+
+        >>> headers = HTTPHeaderDict(foo='bar')
+        >>> headers.add('Foo', 'baz')
+        >>> headers['foo']
+        'bar, baz'
+        """
+        key_lower = key.lower()
+        new_vals = [key, val]
+        # Keep the common case aka no item present as fast as possible
+        vals = self._container.setdefault(key_lower, new_vals)
+        if new_vals is not vals:
+            vals.append(val)
+
+    def extend(self, *args, **kwargs):
+        """Generic import function for any type of header-like object.
+        Adapted version of MutableMapping.update in order to insert items
+        with self.add instead of self.__setitem__
+        """
+        if len(args) > 1:
+            raise TypeError(
+                "extend() takes at most 1 positional "
+                "arguments ({0} given)".format(len(args))
+            )
+        other = args[0] if len(args) >= 1 else ()
+
+        if isinstance(other, HTTPHeaderDict):
+            for key, val in other.iteritems():
+                self.add(key, val)
+        elif isinstance(other, Mapping):
+            for key in other:
+                self.add(key, other[key])
+        elif hasattr(other, "keys"):
+            for key in other.keys():
+                self.add(key, other[key])
+        else:
+            for key, value in other:
+                self.add(key, value)
+
+        for key, value in kwargs.items():
+            self.add(key, value)
+
+    def getlist(self, key, default=__marker):
+        """Returns a list of all the values for the named field. Returns an
+        empty list if the key doesn't exist."""
+        try:
+            vals = self._container[key.lower()]
+        except KeyError:
+            if default is self.__marker:
+                return []
+            return default
+        else:
+            return vals[1:]
+
+    # Backwards compatibility for httplib
+    getheaders = getlist
+    getallmatchingheaders = getlist
+    iget = getlist
+
+    # Backwards compatibility for http.cookiejar
+    get_all = getlist
+
+    def __repr__(self):
+        return "%s(%s)" % (type(self).__name__, dict(self.itermerged()))
+
+    def _copy_from(self, other):
+        for key in other:
+            val = other.getlist(key)
+            if isinstance(val, list):
+                # Don't need to convert tuples
+                val = list(val)
+            self._container[key.lower()] = [key] + val
+
+    def copy(self):
+        clone = type(self)()
+        clone._copy_from(self)
+        return clone
+
+    def iteritems(self):
+        """Iterate over all header lines, including duplicate ones."""
+        for key in self:
+            vals = self._container[key.lower()]
+            for val in vals[1:]:
+                yield vals[0], val
+
+    def itermerged(self):
+        """Iterate over all headers, merging duplicate ones together."""
+        for key in self:
+            val = self._container[key.lower()]
+            yield val[0], ", ".join(val[1:])
+
+    def items(self):
+        return list(self.iteritems())
+
+    @classmethod
+    def from_httplib(cls, message):  # Python 2
+        """Read headers from a Python 2 httplib message object."""
+        # python2.7 does not expose a proper API for exporting multiheaders
+        # efficiently. This function re-reads raw lines from the message
+        # object and extracts the multiheaders properly.
+        obs_fold_continued_leaders = (" ", "\t")
+        headers = []
+
+        for line in message.headers:
+            if line.startswith(obs_fold_continued_leaders):
+                if not headers:
+                    # We received a header line that starts with OWS as described
+                    # in RFC-7230 S3.2.4. This indicates a multiline header, but
+                    # there exists no previous header to which we can attach it.
+                    raise InvalidHeader(
+                        "Header continuation with no previous header: %s" % line
+                    )
+                else:
+                    key, value = headers[-1]
+                    headers[-1] = (key, value + " " + line.strip())
+                    continue
+
+            key, value = line.split(":", 1)
+            headers.append((key, value.strip()))
+
+        return cls(headers)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/_version.py

@@ -0,0 +1,2 @@
+# This file is protected via CODEOWNERS
+__version__ = "1.26.17"

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/connection.py

@@ -0,0 +1,572 @@
+from __future__ import absolute_import
+
+import datetime
+import logging
+import os
+import re
+import socket
+import warnings
+from socket import error as SocketError
+from socket import timeout as SocketTimeout
+
+from .packages import six
+from .packages.six.moves.http_client import HTTPConnection as _HTTPConnection
+from .packages.six.moves.http_client import HTTPException  # noqa: F401
+from .util.proxy import create_proxy_ssl_context
+
+try:  # Compiled with SSL?
+    import ssl
+
+    BaseSSLError = ssl.SSLError
+except (ImportError, AttributeError):  # Platform-specific: No SSL.
+    ssl = None
+
+    class BaseSSLError(BaseException):
+        pass
+
+
+try:
+    # Python 3: not a no-op, we're adding this to the namespace so it can be imported.
+    ConnectionError = ConnectionError
+except NameError:
+    # Python 2
+    class ConnectionError(Exception):
+        pass
+
+
+try:  # Python 3:
+    # Not a no-op, we're adding this to the namespace so it can be imported.
+    BrokenPipeError = BrokenPipeError
+except NameError:  # Python 2:
+
+    class BrokenPipeError(Exception):
+        pass
+
+
+from ._collections import HTTPHeaderDict  # noqa (historical, removed in v2)
+from ._version import __version__
+from .exceptions import (
+    ConnectTimeoutError,
+    NewConnectionError,
+    SubjectAltNameWarning,
+    SystemTimeWarning,
+)
+from .util import SKIP_HEADER, SKIPPABLE_HEADERS, connection
+from .util.ssl_ import (
+    assert_fingerprint,
+    create_urllib3_context,
+    is_ipaddress,
+    resolve_cert_reqs,
+    resolve_ssl_version,
+    ssl_wrap_socket,
+)
+from .util.ssl_match_hostname import CertificateError, match_hostname
+
+log = logging.getLogger(__name__)
+
+port_by_scheme = {"http": 80, "https": 443}
+
+# When it comes time to update this value as a part of regular maintenance
+# (ie test_recent_date is failing) update it to ~6 months before the current date.
+RECENT_DATE = datetime.date(2022, 1, 1)
+
+_CONTAINS_CONTROL_CHAR_RE = re.compile(r"[^-!#$%&'*+.^_`|~0-9a-zA-Z]")
+
+
+class HTTPConnection(_HTTPConnection, object):
+    """
+    Based on :class:`http.client.HTTPConnection` but provides an extra constructor
+    backwards-compatibility layer between older and newer Pythons.
+
+    Additional keyword parameters are used to configure attributes of the connection.
+    Accepted parameters include:
+
+    - ``strict``: See the documentation on :class:`urllib3.connectionpool.HTTPConnectionPool`
+    - ``source_address``: Set the source address for the current connection.
+    - ``socket_options``: Set specific options on the underlying socket. If not specified, then
+      defaults are loaded from ``HTTPConnection.default_socket_options`` which includes disabling
+      Nagle's algorithm (sets TCP_NODELAY to 1) unless the connection is behind a proxy.
+
+      For example, if you wish to enable TCP Keep Alive in addition to the defaults,
+      you might pass:
+
+      .. code-block:: python
+
+         HTTPConnection.default_socket_options + [
+             (socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1),
+         ]
+
+      Or you may want to disable the defaults by passing an empty list (e.g., ``[]``).
+    """
+
+    default_port = port_by_scheme["http"]
+
+    #: Disable Nagle's algorithm by default.
+    #: ``[(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)]``
+    default_socket_options = [(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)]
+
+    #: Whether this connection verifies the host's certificate.
+    is_verified = False
+
+    #: Whether this proxy connection (if used) verifies the proxy host's
+    #: certificate.
+    proxy_is_verified = None
+
+    def __init__(self, *args, **kw):
+        if not six.PY2:
+            kw.pop("strict", None)
+
+        # Pre-set source_address.
+        self.source_address = kw.get("source_address")
+
+        #: The socket options provided by the user. If no options are
+        #: provided, we use the default options.
+        self.socket_options = kw.pop("socket_options", self.default_socket_options)
+
+        # Proxy options provided by the user.
+        self.proxy = kw.pop("proxy", None)
+        self.proxy_config = kw.pop("proxy_config", None)
+
+        _HTTPConnection.__init__(self, *args, **kw)
+
+    @property
+    def host(self):
+        """
+        Getter method to remove any trailing dots that indicate the hostname is an FQDN.
+
+        In general, SSL certificates don't include the trailing dot indicating a
+        fully-qualified domain name, and thus, they don't validate properly when
+        checked against a domain name that includes the dot. In addition, some
+        servers may not expect to receive the trailing dot when provided.
+
+        However, the hostname with trailing dot is critical to DNS resolution; doing a
+        lookup with the trailing dot will properly only resolve the appropriate FQDN,
+        whereas a lookup without a trailing dot will search the system's search domain
+        list. Thus, it's important to keep the original host around for use only in
+        those cases where it's appropriate (i.e., when doing DNS lookup to establish the
+        actual TCP connection across which we're going to send HTTP requests).
+        """
+        return self._dns_host.rstrip(".")
+
+    @host.setter
+    def host(self, value):
+        """
+        Setter for the `host` property.
+
+        We assume that only urllib3 uses the _dns_host attribute; httplib itself
+        only uses `host`, and it seems reasonable that other libraries follow suit.
+        """
+        self._dns_host = value
+
+    def _new_conn(self):
+        """Establish a socket connection and set nodelay settings on it.
+
+        :return: New socket connection.
+        """
+        extra_kw = {}
+        if self.source_address:
+            extra_kw["source_address"] = self.source_address
+
+        if self.socket_options:
+            extra_kw["socket_options"] = self.socket_options
+
+        try:
+            conn = connection.create_connection(
+                (self._dns_host, self.port), self.timeout, **extra_kw
+            )
+
+        except SocketTimeout:
+            raise ConnectTimeoutError(
+                self,
+                "Connection to %s timed out. (connect timeout=%s)"
+                % (self.host, self.timeout),
+            )
+
+        except SocketError as e:
+            raise NewConnectionError(
+                self, "Failed to establish a new connection: %s" % e
+            )
+
+        return conn
+
+    def _is_using_tunnel(self):
+        # Google App Engine's httplib does not define _tunnel_host
+        return getattr(self, "_tunnel_host", None)
+
+    def _prepare_conn(self, conn):
+        self.sock = conn
+        if self._is_using_tunnel():
+            # TODO: Fix tunnel so it doesn't depend on self.sock state.
+            self._tunnel()
+            # Mark this connection as not reusable
+            self.auto_open = 0
+
+    def connect(self):
+        conn = self._new_conn()
+        self._prepare_conn(conn)
+
+    def putrequest(self, method, url, *args, **kwargs):
+        """ """
+        # Empty docstring because the indentation of CPython's implementation
+        # is broken but we don't want this method in our documentation.
+        match = _CONTAINS_CONTROL_CHAR_RE.search(method)
+        if match:
+            raise ValueError(
+                "Method cannot contain non-token characters %r (found at least %r)"
+                % (method, match.group())
+            )
+
+        return _HTTPConnection.putrequest(self, method, url, *args, **kwargs)
+
+    def putheader(self, header, *values):
+        """ """
+        if not any(isinstance(v, str) and v == SKIP_HEADER for v in values):
+            _HTTPConnection.putheader(self, header, *values)
+        elif six.ensure_str(header.lower()) not in SKIPPABLE_HEADERS:
+            raise ValueError(
+                "urllib3.util.SKIP_HEADER only supports '%s'"
+                % ("', '".join(map(str.title, sorted(SKIPPABLE_HEADERS))),)
+            )
+
+    def request(self, method, url, body=None, headers=None):
+        # Update the inner socket's timeout value to send the request.
+        # This only triggers if the connection is re-used.
+        if getattr(self, "sock", None) is not None:
+            self.sock.settimeout(self.timeout)
+
+        if headers is None:
+            headers = {}
+        else:
+            # Avoid modifying the headers passed into .request()
+            headers = headers.copy()
+        if "user-agent" not in (six.ensure_str(k.lower()) for k in headers):
+            headers["User-Agent"] = _get_default_user_agent()
+        super(HTTPConnection, self).request(method, url, body=body, headers=headers)
+
+    def request_chunked(self, method, url, body=None, headers=None):
+        """
+        Alternative to the common request method, which sends the
+        body with chunked encoding and not as one block
+        """
+        headers = headers or {}
+        header_keys = set([six.ensure_str(k.lower()) for k in headers])
+        skip_accept_encoding = "accept-encoding" in header_keys
+        skip_host = "host" in header_keys
+        self.putrequest(
+            method, url, skip_accept_encoding=skip_accept_encoding, skip_host=skip_host
+        )
+        if "user-agent" not in header_keys:
+            self.putheader("User-Agent", _get_default_user_agent())
+        for header, value in headers.items():
+            self.putheader(header, value)
+        if "transfer-encoding" not in header_keys:
+            self.putheader("Transfer-Encoding", "chunked")
+        self.endheaders()
+
+        if body is not None:
+            stringish_types = six.string_types + (bytes,)
+            if isinstance(body, stringish_types):
+                body = (body,)
+            for chunk in body:
+                if not chunk:
+                    continue
+                if not isinstance(chunk, bytes):
+                    chunk = chunk.encode("utf8")
+                len_str = hex(len(chunk))[2:]
+                to_send = bytearray(len_str.encode())
+                to_send += b"\r\n"
+                to_send += chunk
+                to_send += b"\r\n"
+                self.send(to_send)
+
+        # After the if clause, to always have a closed body
+        self.send(b"0\r\n\r\n")
+
+
+class HTTPSConnection(HTTPConnection):
+    """
+    Many of the parameters to this constructor are passed to the underlying SSL
+    socket by means of :py:func:`urllib3.util.ssl_wrap_socket`.
+    """
+
+    default_port = port_by_scheme["https"]
+
+    cert_reqs = None
+    ca_certs = None
+    ca_cert_dir = None
+    ca_cert_data = None
+    ssl_version = None
+    assert_fingerprint = None
+    tls_in_tls_required = False
+
+    def __init__(
+        self,
+        host,
+        port=None,
+        key_file=None,
+        cert_file=None,
+        key_password=None,
+        strict=None,
+        timeout=socket._GLOBAL_DEFAULT_TIMEOUT,
+        ssl_context=None,
+        server_hostname=None,
+        **kw
+    ):
+
+        HTTPConnection.__init__(self, host, port, strict=strict, timeout=timeout, **kw)
+
+        self.key_file = key_file
+        self.cert_file = cert_file
+        self.key_password = key_password
+        self.ssl_context = ssl_context
+        self.server_hostname = server_hostname
+
+        # Required property for Google AppEngine 1.9.0 which otherwise causes
+        # HTTPS requests to go out as HTTP. (See Issue #356)
+        self._protocol = "https"
+
+    def set_cert(
+        self,
+        key_file=None,
+        cert_file=None,
+        cert_reqs=None,
+        key_password=None,
+        ca_certs=None,
+        assert_hostname=None,
+        assert_fingerprint=None,
+        ca_cert_dir=None,
+        ca_cert_data=None,
+    ):
+        """
+        This method should only be called once, before the connection is used.
+        """
+        # If cert_reqs is not provided we'll assume CERT_REQUIRED unless we also
+        # have an SSLContext object in which case we'll use its verify_mode.
+        if cert_reqs is None:
+            if self.ssl_context is not None:
+                cert_reqs = self.ssl_context.verify_mode
+            else:
+                cert_reqs = resolve_cert_reqs(None)
+
+        self.key_file = key_file
+        self.cert_file = cert_file
+        self.cert_reqs = cert_reqs
+        self.key_password = key_password
+        self.assert_hostname = assert_hostname
+        self.assert_fingerprint = assert_fingerprint
+        self.ca_certs = ca_certs and os.path.expanduser(ca_certs)
+        self.ca_cert_dir = ca_cert_dir and os.path.expanduser(ca_cert_dir)
+        self.ca_cert_data = ca_cert_data
+
+    def connect(self):
+        # Add certificate verification
+        self.sock = conn = self._new_conn()
+        hostname = self.host
+        tls_in_tls = False
+
+        if self._is_using_tunnel():
+            if self.tls_in_tls_required:
+                self.sock = conn = self._connect_tls_proxy(hostname, conn)
+                tls_in_tls = True
+
+            # Calls self._set_hostport(), so self.host is
+            # self._tunnel_host below.
+            self._tunnel()
+            # Mark this connection as not reusable
+            self.auto_open = 0
+
+            # Override the host with the one we're requesting data from.
+            hostname = self._tunnel_host
+
+        server_hostname = hostname
+        if self.server_hostname is not None:
+            server_hostname = self.server_hostname
+
+        is_time_off = datetime.date.today() < RECENT_DATE
+        if is_time_off:
+            warnings.warn(
+                (
+                    "System time is way off (before {0}). This will probably "
+                    "lead to SSL verification errors"
+                ).format(RECENT_DATE),
+                SystemTimeWarning,
+            )
+
+        # Wrap socket using verification with the root certs in
+        # trusted_root_certs
+        default_ssl_context = False
+        if self.ssl_context is None:
+            default_ssl_context = True
+            self.ssl_context = create_urllib3_context(
+                ssl_version=resolve_ssl_version(self.ssl_version),
+                cert_reqs=resolve_cert_reqs(self.cert_reqs),
+            )
+
+        context = self.ssl_context
+        context.verify_mode = resolve_cert_reqs(self.cert_reqs)
+
+        # Try to load OS default certs if none are given.
+        # Works well on Windows (requires Python3.4+)
+        if (
+            not self.ca_certs
+            and not self.ca_cert_dir
+            and not self.ca_cert_data
+            and default_ssl_context
+            and hasattr(context, "load_default_certs")
+        ):
+            context.load_default_certs()
+
+        self.sock = ssl_wrap_socket(
+            sock=conn,
+            keyfile=self.key_file,
+            certfile=self.cert_file,
+            key_password=self.key_password,
+            ca_certs=self.ca_certs,
+            ca_cert_dir=self.ca_cert_dir,
+            ca_cert_data=self.ca_cert_data,
+            server_hostname=server_hostname,
+            ssl_context=context,
+            tls_in_tls=tls_in_tls,
+        )
+
+        # If we're using all defaults and the connection
+        # is TLSv1 or TLSv1.1 we throw a DeprecationWarning
+        # for the host.
+        if (
+            default_ssl_context
+            and self.ssl_version is None
+            and hasattr(self.sock, "version")
+            and self.sock.version() in {"TLSv1", "TLSv1.1"}
+        ):
+            warnings.warn(
+                "Negotiating TLSv1/TLSv1.1 by default is deprecated "
+                "and will be disabled in urllib3 v2.0.0. Connecting to "
+                "'%s' with '%s' can be enabled by explicitly opting-in "
+                "with 'ssl_version'" % (self.host, self.sock.version()),
+                DeprecationWarning,
+            )
+
+        if self.assert_fingerprint:
+            assert_fingerprint(
+                self.sock.getpeercert(binary_form=True), self.assert_fingerprint
+            )
+        elif (
+            context.verify_mode != ssl.CERT_NONE
+            and not getattr(context, "check_hostname", False)
+            and self.assert_hostname is not False
+        ):
+            # While urllib3 attempts to always turn off hostname matching from
+            # the TLS library, this cannot always be done. So we check whether
+            # the TLS Library still thinks it's matching hostnames.
+            cert = self.sock.getpeercert()
+            if not cert.get("subjectAltName", ()):
+                warnings.warn(
+                    (
+                        "Certificate for {0} has no `subjectAltName`, falling back to check for a "
+                        "`commonName` for now. This feature is being removed by major browsers and "
+                        "deprecated by RFC 2818. (See https://github.com/urllib3/urllib3/issues/497 "
+                        "for details.)".format(hostname)
+                    ),
+                    SubjectAltNameWarning,
+                )
+            _match_hostname(cert, self.assert_hostname or server_hostname)
+
+        self.is_verified = (
+            context.verify_mode == ssl.CERT_REQUIRED
+            or self.assert_fingerprint is not None
+        )
+
+    def _connect_tls_proxy(self, hostname, conn):
+        """
+        Establish a TLS connection to the proxy using the provided SSL context.
+        """
+        proxy_config = self.proxy_config
+        ssl_context = proxy_config.ssl_context
+        if ssl_context:
+            # If the user provided a proxy context, we assume CA and client
+            # certificates have already been set
+            return ssl_wrap_socket(
+                sock=conn,
+                server_hostname=hostname,
+                ssl_context=ssl_context,
+            )
+
+        ssl_context = create_proxy_ssl_context(
+            self.ssl_version,
+            self.cert_reqs,
+            self.ca_certs,
+            self.ca_cert_dir,
+            self.ca_cert_data,
+        )
+
+        # If no cert was provided, use only the default options for server
+        # certificate validation
+        socket = ssl_wrap_socket(
+            sock=conn,
+            ca_certs=self.ca_certs,
+            ca_cert_dir=self.ca_cert_dir,
+            ca_cert_data=self.ca_cert_data,
+            server_hostname=hostname,
+            ssl_context=ssl_context,
+        )
+
+        if ssl_context.verify_mode != ssl.CERT_NONE and not getattr(
+            ssl_context, "check_hostname", False
+        ):
+            # While urllib3 attempts to always turn off hostname matching from
+            # the TLS library, this cannot always be done. So we check whether
+            # the TLS Library still thinks it's matching hostnames.
+            cert = socket.getpeercert()
+            if not cert.get("subjectAltName", ()):
+                warnings.warn(
+                    (
+                        "Certificate for {0} has no `subjectAltName`, falling back to check for a "
+                        "`commonName` for now. This feature is being removed by major browsers and "
+                        "deprecated by RFC 2818. (See https://github.com/urllib3/urllib3/issues/497 "
+                        "for details.)".format(hostname)
+                    ),
+                    SubjectAltNameWarning,
+                )
+            _match_hostname(cert, hostname)
+
+        self.proxy_is_verified = ssl_context.verify_mode == ssl.CERT_REQUIRED
+        return socket
+
+
+def _match_hostname(cert, asserted_hostname):
+    # Our upstream implementation of ssl.match_hostname()
+    # only applies this normalization to IP addresses so it doesn't
+    # match DNS SANs so we do the same thing!
+    stripped_hostname = asserted_hostname.strip("u[]")
+    if is_ipaddress(stripped_hostname):
+        asserted_hostname = stripped_hostname
+
+    try:
+        match_hostname(cert, asserted_hostname)
+    except CertificateError as e:
+        log.warning(
+            "Certificate did not match expected hostname: %s. Certificate: %s",
+            asserted_hostname,
+            cert,
+        )
+        # Add cert to exception and reraise so client code can inspect
+        # the cert when catching the exception, if they want to
+        e._peer_cert = cert
+        raise
+
+
+def _get_default_user_agent():
+    return "python-urllib3/%s" % __version__
+
+
+class DummyConnection(object):
+    """Used to detect a failed ConnectionCls import."""
+
+    pass
+
+
+if not ssl:
+    HTTPSConnection = DummyConnection  # noqa: F811
+
+
+VerifiedHTTPSConnection = HTTPSConnection

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/connectionpool.py

@@ -0,0 +1,1132 @@
+from __future__ import absolute_import
+
+import errno
+import logging
+import re
+import socket
+import sys
+import warnings
+from socket import error as SocketError
+from socket import timeout as SocketTimeout
+
+from .connection import (
+    BaseSSLError,
+    BrokenPipeError,
+    DummyConnection,
+    HTTPConnection,
+    HTTPException,
+    HTTPSConnection,
+    VerifiedHTTPSConnection,
+    port_by_scheme,
+)
+from .exceptions import (
+    ClosedPoolError,
+    EmptyPoolError,
+    HeaderParsingError,
+    HostChangedError,
+    InsecureRequestWarning,
+    LocationValueError,
+    MaxRetryError,
+    NewConnectionError,
+    ProtocolError,
+    ProxyError,
+    ReadTimeoutError,
+    SSLError,
+    TimeoutError,
+)
+from .packages import six
+from .packages.six.moves import queue
+from .request import RequestMethods
+from .response import HTTPResponse
+from .util.connection import is_connection_dropped
+from .util.proxy import connection_requires_http_tunnel
+from .util.queue import LifoQueue
+from .util.request import set_file_position
+from .util.response import assert_header_parsing
+from .util.retry import Retry
+from .util.ssl_match_hostname import CertificateError
+from .util.timeout import Timeout
+from .util.url import Url, _encode_target
+from .util.url import _normalize_host as normalize_host
+from .util.url import get_host, parse_url
+
+try:  # Platform-specific: Python 3
+    import weakref
+
+    weakref_finalize = weakref.finalize
+except AttributeError:  # Platform-specific: Python 2
+    from .packages.backports.weakref_finalize import weakref_finalize
+
+xrange = six.moves.xrange
+
+log = logging.getLogger(__name__)
+
+_Default = object()
+
+
+# Pool objects
+class ConnectionPool(object):
+    """
+    Base class for all connection pools, such as
+    :class:`.HTTPConnectionPool` and :class:`.HTTPSConnectionPool`.
+
+    .. note::
+       ConnectionPool.urlopen() does not normalize or percent-encode target URIs
+       which is useful if your target server doesn't support percent-encoded
+       target URIs.
+    """
+
+    scheme = None
+    QueueCls = LifoQueue
+
+    def __init__(self, host, port=None):
+        if not host:
+            raise LocationValueError("No host specified.")
+
+        self.host = _normalize_host(host, scheme=self.scheme)
+        self._proxy_host = host.lower()
+        self.port = port
+
+    def __str__(self):
+        return "%s(host=%r, port=%r)" % (type(self).__name__, self.host, self.port)
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.close()
+        # Return False to re-raise any potential exceptions
+        return False
+
+    def close(self):
+        """
+        Close all pooled connections and disable the pool.
+        """
+        pass
+
+
+# This is taken from http://hg.python.org/cpython/file/7aaba721ebc0/Lib/socket.py#l252
+_blocking_errnos = {errno.EAGAIN, errno.EWOULDBLOCK}
+
+
+class HTTPConnectionPool(ConnectionPool, RequestMethods):
+    """
+    Thread-safe connection pool for one host.
+
+    :param host:
+        Host used for this HTTP Connection (e.g. "localhost"), passed into
+        :class:`http.client.HTTPConnection`.
+
+    :param port:
+        Port used for this HTTP Connection (None is equivalent to 80), passed
+        into :class:`http.client.HTTPConnection`.
+
+    :param strict:
+        Causes BadStatusLine to be raised if the status line can't be parsed
+        as a valid HTTP/1.0 or 1.1 status line, passed into
+        :class:`http.client.HTTPConnection`.
+
+        .. note::
+           Only works in Python 2. This parameter is ignored in Python 3.
+
+    :param timeout:
+        Socket timeout in seconds for each individual connection. This can
+        be a float or integer, which sets the timeout for the HTTP request,
+        or an instance of :class:`urllib3.util.Timeout` which gives you more
+        fine-grained control over request timeouts. After the constructor has
+        been parsed, this is always a `urllib3.util.Timeout` object.
+
+    :param maxsize:
+        Number of connections to save that can be reused. More than 1 is useful
+        in multithreaded situations. If ``block`` is set to False, more
+        connections will be created but they will not be saved once they've
+        been used.
+
+    :param block:
+        If set to True, no more than ``maxsize`` connections will be used at
+        a time. When no free connections are available, the call will block
+        until a connection has been released. This is a useful side effect for
+        particular multithreaded situations where one does not want to use more
+        than maxsize connections per host to prevent flooding.
+
+    :param headers:
+        Headers to include with all requests, unless other headers are given
+        explicitly.
+
+    :param retries:
+        Retry configuration to use by default with requests in this pool.
+
+    :param _proxy:
+        Parsed proxy URL, should not be used directly, instead, see
+        :class:`urllib3.ProxyManager`
+
+    :param _proxy_headers:
+        A dictionary with proxy headers, should not be used directly,
+        instead, see :class:`urllib3.ProxyManager`
+
+    :param \\**conn_kw:
+        Additional parameters are used to create fresh :class:`urllib3.connection.HTTPConnection`,
+        :class:`urllib3.connection.HTTPSConnection` instances.
+    """
+
+    scheme = "http"
+    ConnectionCls = HTTPConnection
+    ResponseCls = HTTPResponse
+
+    def __init__(
+        self,
+        host,
+        port=None,
+        strict=False,
+        timeout=Timeout.DEFAULT_TIMEOUT,
+        maxsize=1,
+        block=False,
+        headers=None,
+        retries=None,
+        _proxy=None,
+        _proxy_headers=None,
+        _proxy_config=None,
+        **conn_kw
+    ):
+        ConnectionPool.__init__(self, host, port)
+        RequestMethods.__init__(self, headers)
+
+        self.strict = strict
+
+        if not isinstance(timeout, Timeout):
+            timeout = Timeout.from_float(timeout)
+
+        if retries is None:
+            retries = Retry.DEFAULT
+
+        self.timeout = timeout
+        self.retries = retries
+
+        self.pool = self.QueueCls(maxsize)
+        self.block = block
+
+        self.proxy = _proxy
+        self.proxy_headers = _proxy_headers or {}
+        self.proxy_config = _proxy_config
+
+        # Fill the queue up so that doing get() on it will block properly
+        for _ in xrange(maxsize):
+            self.pool.put(None)
+
+        # These are mostly for testing and debugging purposes.
+        self.num_connections = 0
+        self.num_requests = 0
+        self.conn_kw = conn_kw
+
+        if self.proxy:
+            # Enable Nagle's algorithm for proxies, to avoid packet fragmentation.
+            # We cannot know if the user has added default socket options, so we cannot replace the
+            # list.
+            self.conn_kw.setdefault("socket_options", [])
+
+            self.conn_kw["proxy"] = self.proxy
+            self.conn_kw["proxy_config"] = self.proxy_config
+
+        # Do not pass 'self' as callback to 'finalize'.
+        # Then the 'finalize' would keep an endless living (leak) to self.
+        # By just passing a reference to the pool allows the garbage collector
+        # to free self if nobody else has a reference to it.
+        pool = self.pool
+
+        # Close all the HTTPConnections in the pool before the
+        # HTTPConnectionPool object is garbage collected.
+        weakref_finalize(self, _close_pool_connections, pool)
+
+    def _new_conn(self):
+        """
+        Return a fresh :class:`HTTPConnection`.
+        """
+        self.num_connections += 1
+        log.debug(
+            "Starting new HTTP connection (%d): %s:%s",
+            self.num_connections,
+            self.host,
+            self.port or "80",
+        )
+
+        conn = self.ConnectionCls(
+            host=self.host,
+            port=self.port,
+            timeout=self.timeout.connect_timeout,
+            strict=self.strict,
+            **self.conn_kw
+        )
+        return conn
+
+    def _get_conn(self, timeout=None):
+        """
+        Get a connection. Will return a pooled connection if one is available.
+
+        If no connections are available and :prop:`.block` is ``False``, then a
+        fresh connection is returned.
+
+        :param timeout:
+            Seconds to wait before giving up and raising
+            :class:`urllib3.exceptions.EmptyPoolError` if the pool is empty and
+            :prop:`.block` is ``True``.
+        """
+        conn = None
+        try:
+            conn = self.pool.get(block=self.block, timeout=timeout)
+
+        except AttributeError:  # self.pool is None
+            raise ClosedPoolError(self, "Pool is closed.")
+
+        except queue.Empty:
+            if self.block:
+                raise EmptyPoolError(
+                    self,
+                    "Pool reached maximum size and no more connections are allowed.",
+                )
+            pass  # Oh well, we'll create a new connection then
+
+        # If this is a persistent connection, check if it got disconnected
+        if conn and is_connection_dropped(conn):
+            log.debug("Resetting dropped connection: %s", self.host)
+            conn.close()
+            if getattr(conn, "auto_open", 1) == 0:
+                # This is a proxied connection that has been mutated by
+                # http.client._tunnel() and cannot be reused (since it would
+                # attempt to bypass the proxy)
+                conn = None
+
+        return conn or self._new_conn()
+
+    def _put_conn(self, conn):
+        """
+        Put a connection back into the pool.
+
+        :param conn:
+            Connection object for the current host and port as returned by
+            :meth:`._new_conn` or :meth:`._get_conn`.
+
+        If the pool is already full, the connection is closed and discarded
+        because we exceeded maxsize. If connections are discarded frequently,
+        then maxsize should be increased.
+
+        If the pool is closed, then the connection will be closed and discarded.
+        """
+        try:
+            self.pool.put(conn, block=False)
+            return  # Everything is dandy, done.
+        except AttributeError:
+            # self.pool is None.
+            pass
+        except queue.Full:
+            # This should never happen if self.block == True
+            log.warning(
+                "Connection pool is full, discarding connection: %s. Connection pool size: %s",
+                self.host,
+                self.pool.qsize(),
+            )
+        # Connection never got put back into the pool, close it.
+        if conn:
+            conn.close()
+
+    def _validate_conn(self, conn):
+        """
+        Called right before a request is made, after the socket is created.
+        """
+        pass
+
+    def _prepare_proxy(self, conn):
+        # Nothing to do for HTTP connections.
+        pass
+
+    def _get_timeout(self, timeout):
+        """Helper that always returns a :class:`urllib3.util.Timeout`"""
+        if timeout is _Default:
+            return self.timeout.clone()
+
+        if isinstance(timeout, Timeout):
+            return timeout.clone()
+        else:
+            # User passed us an int/float. This is for backwards compatibility,
+            # can be removed later
+            return Timeout.from_float(timeout)
+
+    def _raise_timeout(self, err, url, timeout_value):
+        """Is the error actually a timeout? Will raise a ReadTimeout or pass"""
+
+        if isinstance(err, SocketTimeout):
+            raise ReadTimeoutError(
+                self, url, "Read timed out. (read timeout=%s)" % timeout_value
+            )
+
+        # See the above comment about EAGAIN in Python 3. In Python 2 we have
+        # to specifically catch it and throw the timeout error
+        if hasattr(err, "errno") and err.errno in _blocking_errnos:
+            raise ReadTimeoutError(
+                self, url, "Read timed out. (read timeout=%s)" % timeout_value
+            )
+
+        # Catch possible read timeouts thrown as SSL errors. If not the
+        # case, rethrow the original. We need to do this because of:
+        # http://bugs.python.org/issue10272
+        if "timed out" in str(err) or "did not complete (read)" in str(
+            err
+        ):  # Python < 2.7.4
+            raise ReadTimeoutError(
+                self, url, "Read timed out. (read timeout=%s)" % timeout_value
+            )
+
+    def _make_request(
+        self, conn, method, url, timeout=_Default, chunked=False, **httplib_request_kw
+    ):
+        """
+        Perform a request on a given urllib connection object taken from our
+        pool.
+
+        :param conn:
+            a connection from one of our connection pools
+
+        :param timeout:
+            Socket timeout in seconds for the request. This can be a
+            float or integer, which will set the same timeout value for
+            the socket connect and the socket read, or an instance of
+            :class:`urllib3.util.Timeout`, which gives you more fine-grained
+            control over your timeouts.
+        """
+        self.num_requests += 1
+
+        timeout_obj = self._get_timeout(timeout)
+        timeout_obj.start_connect()
+        conn.timeout = Timeout.resolve_default_timeout(timeout_obj.connect_timeout)
+
+        # Trigger any extra validation we need to do.
+        try:
+            self._validate_conn(conn)
+        except (SocketTimeout, BaseSSLError) as e:
+            # Py2 raises this as a BaseSSLError, Py3 raises it as socket timeout.
+            self._raise_timeout(err=e, url=url, timeout_value=conn.timeout)
+            raise
+
+        # conn.request() calls http.client.*.request, not the method in
+        # urllib3.request. It also calls makefile (recv) on the socket.
+        try:
+            if chunked:
+                conn.request_chunked(method, url, **httplib_request_kw)
+            else:
+                conn.request(method, url, **httplib_request_kw)
+
+        # We are swallowing BrokenPipeError (errno.EPIPE) since the server is
+        # legitimately able to close the connection after sending a valid response.
+        # With this behaviour, the received response is still readable.
+        except BrokenPipeError:
+            # Python 3
+            pass
+        except IOError as e:
+            # Python 2 and macOS/Linux
+            # EPIPE and ESHUTDOWN are BrokenPipeError on Python 2, and EPROTOTYPE is needed on macOS
+            # https://erickt.github.io/blog/2014/11/19/adventures-in-debugging-a-potential-osx-kernel-bug/
+            if e.errno not in {
+                errno.EPIPE,
+                errno.ESHUTDOWN,
+                errno.EPROTOTYPE,
+            }:
+                raise
+
+        # Reset the timeout for the recv() on the socket
+        read_timeout = timeout_obj.read_timeout
+
+        # App Engine doesn't have a sock attr
+        if getattr(conn, "sock", None):
+            # In Python 3 socket.py will catch EAGAIN and return None when you
+            # try and read into the file pointer created by http.client, which
+            # instead raises a BadStatusLine exception. Instead of catching
+            # the exception and assuming all BadStatusLine exceptions are read
+            # timeouts, check for a zero timeout before making the request.
+            if read_timeout == 0:
+                raise ReadTimeoutError(
+                    self, url, "Read timed out. (read timeout=%s)" % read_timeout
+                )
+            if read_timeout is Timeout.DEFAULT_TIMEOUT:
+                conn.sock.settimeout(socket.getdefaulttimeout())
+            else:  # None or a value
+                conn.sock.settimeout(read_timeout)
+
+        # Receive the response from the server
+        try:
+            try:
+                # Python 2.7, use buffering of HTTP responses
+                httplib_response = conn.getresponse(buffering=True)
+            except TypeError:
+                # Python 3
+                try:
+                    httplib_response = conn.getresponse()
+                except BaseException as e:
+                    # Remove the TypeError from the exception chain in
+                    # Python 3 (including for exceptions like SystemExit).
+                    # Otherwise it looks like a bug in the code.
+                    six.raise_from(e, None)
+        except (SocketTimeout, BaseSSLError, SocketError) as e:
+            self._raise_timeout(err=e, url=url, timeout_value=read_timeout)
+            raise
+
+        # AppEngine doesn't have a version attr.
+        http_version = getattr(conn, "_http_vsn_str", "HTTP/?")
+        log.debug(
+            '%s://%s:%s "%s %s %s" %s %s',
+            self.scheme,
+            self.host,
+            self.port,
+            method,
+            url,
+            http_version,
+            httplib_response.status,
+            httplib_response.length,
+        )
+
+        try:
+            assert_header_parsing(httplib_response.msg)
+        except (HeaderParsingError, TypeError) as hpe:  # Platform-specific: Python 3
+            log.warning(
+                "Failed to parse headers (url=%s): %s",
+                self._absolute_url(url),
+                hpe,
+                exc_info=True,
+            )
+
+        return httplib_response
+
+    def _absolute_url(self, path):
+        return Url(scheme=self.scheme, host=self.host, port=self.port, path=path).url
+
+    def close(self):
+        """
+        Close all pooled connections and disable the pool.
+        """
+        if self.pool is None:
+            return
+        # Disable access to the pool
+        old_pool, self.pool = self.pool, None
+
+        # Close all the HTTPConnections in the pool.
+        _close_pool_connections(old_pool)
+
+    def is_same_host(self, url):
+        """
+        Check if the given ``url`` is a member of the same host as this
+        connection pool.
+        """
+        if url.startswith("/"):
+            return True
+
+        # TODO: Add optional support for socket.gethostbyname checking.
+        scheme, host, port = get_host(url)
+        if host is not None:
+            host = _normalize_host(host, scheme=scheme)
+
+        # Use explicit default port for comparison when none is given
+        if self.port and not port:
+            port = port_by_scheme.get(scheme)
+        elif not self.port and port == port_by_scheme.get(scheme):
+            port = None
+
+        return (scheme, host, port) == (self.scheme, self.host, self.port)
+
+    def urlopen(
+        self,
+        method,
+        url,
+        body=None,
+        headers=None,
+        retries=None,
+        redirect=True,
+        assert_same_host=True,
+        timeout=_Default,
+        pool_timeout=None,
+        release_conn=None,
+        chunked=False,
+        body_pos=None,
+        **response_kw
+    ):
+        """
+        Get a connection from the pool and perform an HTTP request. This is the
+        lowest level call for making a request, so you'll need to specify all
+        the raw details.
+
+        .. note::
+
+           More commonly, it's appropriate to use a convenience method provided
+           by :class:`.RequestMethods`, such as :meth:`request`.
+
+        .. note::
+
+           `release_conn` will only behave as expected if
+           `preload_content=False` because we want to make
+           `preload_content=False` the default behaviour someday soon without
+           breaking backwards compatibility.
+
+        :param method:
+            HTTP request method (such as GET, POST, PUT, etc.)
+
+        :param url:
+            The URL to perform the request on.
+
+        :param body:
+            Data to send in the request body, either :class:`str`, :class:`bytes`,
+            an iterable of :class:`str`/:class:`bytes`, or a file-like object.
+
+        :param headers:
+            Dictionary of custom headers to send, such as User-Agent,
+            If-None-Match, etc. If None, pool headers are used. If provided,
+            these headers completely replace any pool-specific headers.
+
+        :param retries:
+            Configure the number of retries to allow before raising a
+            :class:`~urllib3.exceptions.MaxRetryError` exception.
+
+            Pass ``None`` to retry until you receive a response. Pass a
+            :class:`~urllib3.util.retry.Retry` object for fine-grained control
+            over different types of retries.
+            Pass an integer number to retry connection errors that many times,
+            but no other types of errors. Pass zero to never retry.
+
+            If ``False``, then retries are disabled and any exception is raised
+            immediately. Also, instead of raising a MaxRetryError on redirects,
+            the redirect response will be returned.
+
+        :type retries: :class:`~urllib3.util.retry.Retry`, False, or an int.
+
+        :param redirect:
+            If True, automatically handle redirects (status codes 301, 302,
+            303, 307, 308). Each redirect counts as a retry. Disabling retries
+            will disable redirect, too.
+
+        :param assert_same_host:
+            If ``True``, will make sure that the host of the pool requests is
+            consistent else will raise HostChangedError. When ``False``, you can
+            use the pool on an HTTP proxy and request foreign hosts.
+
+        :param timeout:
+            If specified, overrides the default timeout for this one
+            request. It may be a float (in seconds) or an instance of
+            :class:`urllib3.util.Timeout`.
+
+        :param pool_timeout:
+            If set and the pool is set to block=True, then this method will
+            block for ``pool_timeout`` seconds and raise EmptyPoolError if no
+            connection is available within the time period.
+
+        :param release_conn:
+            If False, then the urlopen call will not release the connection
+            back into the pool once a response is received (but will release if
+            you read the entire contents of the response such as when
+            `preload_content=True`). This is useful if you're not preloading
+            the response's content immediately. You will need to call
+            ``r.release_conn()`` on the response ``r`` to return the connection
+            back into the pool. If None, it takes the value of
+            ``response_kw.get('preload_content', True)``.
+
+        :param chunked:
+            If True, urllib3 will send the body using chunked transfer
+            encoding. Otherwise, urllib3 will send the body using the standard
+            content-length form. Defaults to False.
+
+        :param int body_pos:
+            Position to seek to in file-like body in the event of a retry or
+            redirect. Typically this won't need to be set because urllib3 will
+            auto-populate the value when needed.
+
+        :param \\**response_kw:
+            Additional parameters are passed to
+            :meth:`urllib3.response.HTTPResponse.from_httplib`
+        """
+
+        parsed_url = parse_url(url)
+        destination_scheme = parsed_url.scheme
+
+        if headers is None:
+            headers = self.headers
+
+        if not isinstance(retries, Retry):
+            retries = Retry.from_int(retries, redirect=redirect, default=self.retries)
+
+        if release_conn is None:
+            release_conn = response_kw.get("preload_content", True)
+
+        # Check host
+        if assert_same_host and not self.is_same_host(url):
+            raise HostChangedError(self, url, retries)
+
+        # Ensure that the URL we're connecting to is properly encoded
+        if url.startswith("/"):
+            url = six.ensure_str(_encode_target(url))
+        else:
+            url = six.ensure_str(parsed_url.url)
+
+        conn = None
+
+        # Track whether `conn` needs to be released before
+        # returning/raising/recursing. Update this variable if necessary, and
+        # leave `release_conn` constant throughout the function. That way, if
+        # the function recurses, the original value of `release_conn` will be
+        # passed down into the recursive call, and its value will be respected.
+        #
+        # See issue #651 [1] for details.
+        #
+        # [1] <https://github.com/urllib3/urllib3/issues/651>
+        release_this_conn = release_conn
+
+        http_tunnel_required = connection_requires_http_tunnel(
+            self.proxy, self.proxy_config, destination_scheme
+        )
+
+        # Merge the proxy headers. Only done when not using HTTP CONNECT. We
+        # have to copy the headers dict so we can safely change it without those
+        # changes being reflected in anyone else's copy.
+        if not http_tunnel_required:
+            headers = headers.copy()
+            headers.update(self.proxy_headers)
+
+        # Must keep the exception bound to a separate variable or else Python 3
+        # complains about UnboundLocalError.
+        err = None
+
+        # Keep track of whether we cleanly exited the except block. This
+        # ensures we do proper cleanup in finally.
+        clean_exit = False
+
+        # Rewind body position, if needed. Record current position
+        # for future rewinds in the event of a redirect/retry.
+        body_pos = set_file_position(body, body_pos)
+
+        try:
+            # Request a connection from the queue.
+            timeout_obj = self._get_timeout(timeout)
+            conn = self._get_conn(timeout=pool_timeout)
+
+            conn.timeout = timeout_obj.connect_timeout
+
+            is_new_proxy_conn = self.proxy is not None and not getattr(
+                conn, "sock", None
+            )
+            if is_new_proxy_conn and http_tunnel_required:
+                self._prepare_proxy(conn)
+
+            # Make the request on the httplib connection object.
+            httplib_response = self._make_request(
+                conn,
+                method,
+                url,
+                timeout=timeout_obj,
+                body=body,
+                headers=headers,
+                chunked=chunked,
+            )
+
+            # If we're going to release the connection in ``finally:``, then
+            # the response doesn't need to know about the connection. Otherwise
+            # it will also try to release it and we'll have a double-release
+            # mess.
+            response_conn = conn if not release_conn else None
+
+            # Pass method to Response for length checking
+            response_kw["request_method"] = method
+
+            # Import httplib's response into our own wrapper object
+            response = self.ResponseCls.from_httplib(
+                httplib_response,
+                pool=self,
+                connection=response_conn,
+                retries=retries,
+                **response_kw
+            )
+
+            # Everything went great!
+            clean_exit = True
+
+        except EmptyPoolError:
+            # Didn't get a connection from the pool, no need to clean up
+            clean_exit = True
+            release_this_conn = False
+            raise
+
+        except (
+            TimeoutError,
+            HTTPException,
+            SocketError,
+            ProtocolError,
+            BaseSSLError,
+            SSLError,
+            CertificateError,
+        ) as e:
+            # Discard the connection for these exceptions. It will be
+            # replaced during the next _get_conn() call.
+            clean_exit = False
+
+            def _is_ssl_error_message_from_http_proxy(ssl_error):
+                # We're trying to detect the message 'WRONG_VERSION_NUMBER' but
+                # SSLErrors are kinda all over the place when it comes to the message,
+                # so we try to cover our bases here!
+                message = " ".join(re.split("[^a-z]", str(ssl_error).lower()))
+                return (
+                    "wrong version number" in message or "unknown protocol" in message
+                )
+
+            # Try to detect a common user error with proxies which is to
+            # set an HTTP proxy to be HTTPS when it should be 'http://'
+            # (ie {'http': 'http://proxy', 'https': 'https://proxy'})
+            # Instead we add a nice error message and point to a URL.
+            if (
+                isinstance(e, BaseSSLError)
+                and self.proxy
+                and _is_ssl_error_message_from_http_proxy(e)
+                and conn.proxy
+                and conn.proxy.scheme == "https"
+            ):
+                e = ProxyError(
+                    "Your proxy appears to only use HTTP and not HTTPS, "
+                    "try changing your proxy URL to be HTTP. See: "
+                    "https://urllib3.readthedocs.io/en/1.26.x/advanced-usage.html"
+                    "#https-proxy-error-http-proxy",
+                    SSLError(e),
+                )
+            elif isinstance(e, (BaseSSLError, CertificateError)):
+                e = SSLError(e)
+            elif isinstance(e, (SocketError, NewConnectionError)) and self.proxy:
+                e = ProxyError("Cannot connect to proxy.", e)
+            elif isinstance(e, (SocketError, HTTPException)):
+                e = ProtocolError("Connection aborted.", e)
+
+            retries = retries.increment(
+                method, url, error=e, _pool=self, _stacktrace=sys.exc_info()[2]
+            )
+            retries.sleep()
+
+            # Keep track of the error for the retry warning.
+            err = e
+
+        finally:
+            if not clean_exit:
+                # We hit some kind of exception, handled or otherwise. We need
+                # to throw the connection away unless explicitly told not to.
+                # Close the connection, set the variable to None, and make sure
+                # we put the None back in the pool to avoid leaking it.
+                conn = conn and conn.close()
+                release_this_conn = True
+
+            if release_this_conn:
+                # Put the connection back to be reused. If the connection is
+                # expired then it will be None, which will get replaced with a
+                # fresh connection during _get_conn.
+                self._put_conn(conn)
+
+        if not conn:
+            # Try again
+            log.warning(
+                "Retrying (%r) after connection broken by '%r': %s", retries, err, url
+            )
+            return self.urlopen(
+                method,
+                url,
+                body,
+                headers,
+                retries,
+                redirect,
+                assert_same_host,
+                timeout=timeout,
+                pool_timeout=pool_timeout,
+                release_conn=release_conn,
+                chunked=chunked,
+                body_pos=body_pos,
+                **response_kw
+            )
+
+        # Handle redirect?
+        redirect_location = redirect and response.get_redirect_location()
+        if redirect_location:
+            if response.status == 303:
+                method = "GET"
+
+            try:
+                retries = retries.increment(method, url, response=response, _pool=self)
+            except MaxRetryError:
+                if retries.raise_on_redirect:
+                    response.drain_conn()
+                    raise
+                return response
+
+            response.drain_conn()
+            retries.sleep_for_retry(response)
+            log.debug("Redirecting %s -> %s", url, redirect_location)
+            return self.urlopen(
+                method,
+                redirect_location,
+                body,
+                headers,
+                retries=retries,
+                redirect=redirect,
+                assert_same_host=assert_same_host,
+                timeout=timeout,
+                pool_timeout=pool_timeout,
+                release_conn=release_conn,
+                chunked=chunked,
+                body_pos=body_pos,
+                **response_kw
+            )
+
+        # Check if we should retry the HTTP response.
+        has_retry_after = bool(response.headers.get("Retry-After"))
+        if retries.is_retry(method, response.status, has_retry_after):
+            try:
+                retries = retries.increment(method, url, response=response, _pool=self)
+            except MaxRetryError:
+                if retries.raise_on_status:
+                    response.drain_conn()
+                    raise
+                return response
+
+            response.drain_conn()
+            retries.sleep(response)
+            log.debug("Retry: %s", url)
+            return self.urlopen(
+                method,
+                url,
+                body,
+                headers,
+                retries=retries,
+                redirect=redirect,
+                assert_same_host=assert_same_host,
+                timeout=timeout,
+                pool_timeout=pool_timeout,
+                release_conn=release_conn,
+                chunked=chunked,
+                body_pos=body_pos,
+                **response_kw
+            )
+
+        return response
+
+
+class HTTPSConnectionPool(HTTPConnectionPool):
+    """
+    Same as :class:`.HTTPConnectionPool`, but HTTPS.
+
+    :class:`.HTTPSConnection` uses one of ``assert_fingerprint``,
+    ``assert_hostname`` and ``host`` in this order to verify connections.
+    If ``assert_hostname`` is False, no verification is done.
+
+    The ``key_file``, ``cert_file``, ``cert_reqs``, ``ca_certs``,
+    ``ca_cert_dir``, ``ssl_version``, ``key_password`` are only used if :mod:`ssl`
+    is available and are fed into :meth:`urllib3.util.ssl_wrap_socket` to upgrade
+    the connection socket into an SSL socket.
+    """
+
+    scheme = "https"
+    ConnectionCls = HTTPSConnection
+
+    def __init__(
+        self,
+        host,
+        port=None,
+        strict=False,
+        timeout=Timeout.DEFAULT_TIMEOUT,
+        maxsize=1,
+        block=False,
+        headers=None,
+        retries=None,
+        _proxy=None,
+        _proxy_headers=None,
+        key_file=None,
+        cert_file=None,
+        cert_reqs=None,
+        key_password=None,
+        ca_certs=None,
+        ssl_version=None,
+        assert_hostname=None,
+        assert_fingerprint=None,
+        ca_cert_dir=None,
+        **conn_kw
+    ):
+
+        HTTPConnectionPool.__init__(
+            self,
+            host,
+            port,
+            strict,
+            timeout,
+            maxsize,
+            block,
+            headers,
+            retries,
+            _proxy,
+            _proxy_headers,
+            **conn_kw
+        )
+
+        self.key_file = key_file
+        self.cert_file = cert_file
+        self.cert_reqs = cert_reqs
+        self.key_password = key_password
+        self.ca_certs = ca_certs
+        self.ca_cert_dir = ca_cert_dir
+        self.ssl_version = ssl_version
+        self.assert_hostname = assert_hostname
+        self.assert_fingerprint = assert_fingerprint
+
+    def _prepare_conn(self, conn):
+        """
+        Prepare the ``connection`` for :meth:`urllib3.util.ssl_wrap_socket`
+        and establish the tunnel if proxy is used.
+        """
+
+        if isinstance(conn, VerifiedHTTPSConnection):
+            conn.set_cert(
+                key_file=self.key_file,
+                key_password=self.key_password,
+                cert_file=self.cert_file,
+                cert_reqs=self.cert_reqs,
+                ca_certs=self.ca_certs,
+                ca_cert_dir=self.ca_cert_dir,
+                assert_hostname=self.assert_hostname,
+                assert_fingerprint=self.assert_fingerprint,
+            )
+            conn.ssl_version = self.ssl_version
+        return conn
+
+    def _prepare_proxy(self, conn):
+        """
+        Establishes a tunnel connection through HTTP CONNECT.
+
+        Tunnel connection is established early because otherwise httplib would
+        improperly set Host: header to proxy's IP:port.
+        """
+
+        conn.set_tunnel(self._proxy_host, self.port, self.proxy_headers)
+
+        if self.proxy.scheme == "https":
+            conn.tls_in_tls_required = True
+
+        conn.connect()
+
+    def _new_conn(self):
+        """
+        Return a fresh :class:`http.client.HTTPSConnection`.
+        """
+        self.num_connections += 1
+        log.debug(
+            "Starting new HTTPS connection (%d): %s:%s",
+            self.num_connections,
+            self.host,
+            self.port or "443",
+        )
+
+        if not self.ConnectionCls or self.ConnectionCls is DummyConnection:
+            raise SSLError(
+                "Can't connect to HTTPS URL because the SSL module is not available."
+            )
+
+        actual_host = self.host
+        actual_port = self.port
+        if self.proxy is not None:
+            actual_host = self.proxy.host
+            actual_port = self.proxy.port
+
+        conn = self.ConnectionCls(
+            host=actual_host,
+            port=actual_port,
+            timeout=self.timeout.connect_timeout,
+            strict=self.strict,
+            cert_file=self.cert_file,
+            key_file=self.key_file,
+            key_password=self.key_password,
+            **self.conn_kw
+        )
+
+        return self._prepare_conn(conn)
+
+    def _validate_conn(self, conn):
+        """
+        Called right before a request is made, after the socket is created.
+        """
+        super(HTTPSConnectionPool, self)._validate_conn(conn)
+
+        # Force connect early to allow us to validate the connection.
+        if not getattr(conn, "sock", None):  # AppEngine might not have  `.sock`
+            conn.connect()
+
+        if not conn.is_verified:
+            warnings.warn(
+                (
+                    "Unverified HTTPS request is being made to host '%s'. "
+                    "Adding certificate verification is strongly advised. See: "
+                    "https://urllib3.readthedocs.io/en/1.26.x/advanced-usage.html"
+                    "#ssl-warnings" % conn.host
+                ),
+                InsecureRequestWarning,
+            )
+
+        if getattr(conn, "proxy_is_verified", None) is False:
+            warnings.warn(
+                (
+                    "Unverified HTTPS connection done to an HTTPS proxy. "
+                    "Adding certificate verification is strongly advised. See: "
+                    "https://urllib3.readthedocs.io/en/1.26.x/advanced-usage.html"
+                    "#ssl-warnings"
+                ),
+                InsecureRequestWarning,
+            )
+
+
+def connection_from_url(url, **kw):
+    """
+    Given a url, return an :class:`.ConnectionPool` instance of its host.
+
+    This is a shortcut for not having to parse out the scheme, host, and port
+    of the url before creating an :class:`.ConnectionPool` instance.
+
+    :param url:
+        Absolute URL string that must include the scheme. Port is optional.
+
+    :param \\**kw:
+        Passes additional parameters to the constructor of the appropriate
+        :class:`.ConnectionPool`. Useful for specifying things like
+        timeout, maxsize, headers, etc.
+
+    Example::
+
+        >>> conn = connection_from_url('http://google.com/')
+        >>> r = conn.request('GET', '/')
+    """
+    scheme, host, port = get_host(url)
+    port = port or port_by_scheme.get(scheme, 80)
+    if scheme == "https":
+        return HTTPSConnectionPool(host, port=port, **kw)
+    else:
+        return HTTPConnectionPool(host, port=port, **kw)
+
+
+def _normalize_host(host, scheme):
+    """
+    Normalize hosts for comparisons and use with sockets.
+    """
+
+    host = normalize_host(host, scheme)
+
+    # httplib doesn't like it when we include brackets in IPv6 addresses
+    # Specifically, if we include brackets but also pass the port then
+    # httplib crazily doubles up the square brackets on the Host header.
+    # Instead, we need to make sure we never pass ``None`` as the port.
+    # However, for backward compatibility reasons we can't actually
+    # *assert* that.  See http://bugs.python.org/issue28539
+    if host.startswith("[") and host.endswith("]"):
+        host = host[1:-1]
+    return host
+
+
+def _close_pool_connections(pool):
+    """Drains a queue of connections and closes each one."""
+    try:
+        while True:
+            conn = pool.get(block=False)
+            if conn:
+                conn.close()
+    except queue.Empty:
+        pass  # Done.

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/exceptions.py

@@ -0,0 +1,323 @@
+from __future__ import absolute_import
+
+from .packages.six.moves.http_client import IncompleteRead as httplib_IncompleteRead
+
+# Base Exceptions
+
+
+class HTTPError(Exception):
+    """Base exception used by this module."""
+
+    pass
+
+
+class HTTPWarning(Warning):
+    """Base warning used by this module."""
+
+    pass
+
+
+class PoolError(HTTPError):
+    """Base exception for errors caused within a pool."""
+
+    def __init__(self, pool, message):
+        self.pool = pool
+        HTTPError.__init__(self, "%s: %s" % (pool, message))
+
+    def __reduce__(self):
+        # For pickling purposes.
+        return self.__class__, (None, None)
+
+
+class RequestError(PoolError):
+    """Base exception for PoolErrors that have associated URLs."""
+
+    def __init__(self, pool, url, message):
+        self.url = url
+        PoolError.__init__(self, pool, message)
+
+    def __reduce__(self):
+        # For pickling purposes.
+        return self.__class__, (None, self.url, None)
+
+
+class SSLError(HTTPError):
+    """Raised when SSL certificate fails in an HTTPS connection."""
+
+    pass
+
+
+class ProxyError(HTTPError):
+    """Raised when the connection to a proxy fails."""
+
+    def __init__(self, message, error, *args):
+        super(ProxyError, self).__init__(message, error, *args)
+        self.original_error = error
+
+
+class DecodeError(HTTPError):
+    """Raised when automatic decoding based on Content-Type fails."""
+
+    pass
+
+
+class ProtocolError(HTTPError):
+    """Raised when something unexpected happens mid-request/response."""
+
+    pass
+
+
+#: Renamed to ProtocolError but aliased for backwards compatibility.
+ConnectionError = ProtocolError
+
+
+# Leaf Exceptions
+
+
+class MaxRetryError(RequestError):
+    """Raised when the maximum number of retries is exceeded.
+
+    :param pool: The connection pool
+    :type pool: :class:`~urllib3.connectionpool.HTTPConnectionPool`
+    :param string url: The requested Url
+    :param exceptions.Exception reason: The underlying error
+
+    """
+
+    def __init__(self, pool, url, reason=None):
+        self.reason = reason
+
+        message = "Max retries exceeded with url: %s (Caused by %r)" % (url, reason)
+
+        RequestError.__init__(self, pool, url, message)
+
+
+class HostChangedError(RequestError):
+    """Raised when an existing pool gets a request for a foreign host."""
+
+    def __init__(self, pool, url, retries=3):
+        message = "Tried to open a foreign host with url: %s" % url
+        RequestError.__init__(self, pool, url, message)
+        self.retries = retries
+
+
+class TimeoutStateError(HTTPError):
+    """Raised when passing an invalid state to a timeout"""
+
+    pass
+
+
+class TimeoutError(HTTPError):
+    """Raised when a socket timeout error occurs.
+
+    Catching this error will catch both :exc:`ReadTimeoutErrors
+    <ReadTimeoutError>` and :exc:`ConnectTimeoutErrors <ConnectTimeoutError>`.
+    """
+
+    pass
+
+
+class ReadTimeoutError(TimeoutError, RequestError):
+    """Raised when a socket timeout occurs while receiving data from a server"""
+
+    pass
+
+
+# This timeout error does not have a URL attached and needs to inherit from the
+# base HTTPError
+class ConnectTimeoutError(TimeoutError):
+    """Raised when a socket timeout occurs while connecting to a server"""
+
+    pass
+
+
+class NewConnectionError(ConnectTimeoutError, PoolError):
+    """Raised when we fail to establish a new connection. Usually ECONNREFUSED."""
+
+    pass
+
+
+class EmptyPoolError(PoolError):
+    """Raised when a pool runs out of connections and no more are allowed."""
+
+    pass
+
+
+class ClosedPoolError(PoolError):
+    """Raised when a request enters a pool after the pool has been closed."""
+
+    pass
+
+
+class LocationValueError(ValueError, HTTPError):
+    """Raised when there is something wrong with a given URL input."""
+
+    pass
+
+
+class LocationParseError(LocationValueError):
+    """Raised when get_host or similar fails to parse the URL input."""
+
+    def __init__(self, location):
+        message = "Failed to parse: %s" % location
+        HTTPError.__init__(self, message)
+
+        self.location = location
+
+
+class URLSchemeUnknown(LocationValueError):
+    """Raised when a URL input has an unsupported scheme."""
+
+    def __init__(self, scheme):
+        message = "Not supported URL scheme %s" % scheme
+        super(URLSchemeUnknown, self).__init__(message)
+
+        self.scheme = scheme
+
+
+class ResponseError(HTTPError):
+    """Used as a container for an error reason supplied in a MaxRetryError."""
+
+    GENERIC_ERROR = "too many error responses"
+    SPECIFIC_ERROR = "too many {status_code} error responses"
+
+
+class SecurityWarning(HTTPWarning):
+    """Warned when performing security reducing actions"""
+
+    pass
+
+
+class SubjectAltNameWarning(SecurityWarning):
+    """Warned when connecting to a host with a certificate missing a SAN."""
+
+    pass
+
+
+class InsecureRequestWarning(SecurityWarning):
+    """Warned when making an unverified HTTPS request."""
+
+    pass
+
+
+class SystemTimeWarning(SecurityWarning):
+    """Warned when system time is suspected to be wrong"""
+
+    pass
+
+
+class InsecurePlatformWarning(SecurityWarning):
+    """Warned when certain TLS/SSL configuration is not available on a platform."""
+
+    pass
+
+
+class SNIMissingWarning(HTTPWarning):
+    """Warned when making a HTTPS request without SNI available."""
+
+    pass
+
+
+class DependencyWarning(HTTPWarning):
+    """
+    Warned when an attempt is made to import a module with missing optional
+    dependencies.
+    """
+
+    pass
+
+
+class ResponseNotChunked(ProtocolError, ValueError):
+    """Response needs to be chunked in order to read it as chunks."""
+
+    pass
+
+
+class BodyNotHttplibCompatible(HTTPError):
+    """
+    Body should be :class:`http.client.HTTPResponse` like
+    (have an fp attribute which returns raw chunks) for read_chunked().
+    """
+
+    pass
+
+
+class IncompleteRead(HTTPError, httplib_IncompleteRead):
+    """
+    Response length doesn't match expected Content-Length
+
+    Subclass of :class:`http.client.IncompleteRead` to allow int value
+    for ``partial`` to avoid creating large objects on streamed reads.
+    """
+
+    def __init__(self, partial, expected):
+        super(IncompleteRead, self).__init__(partial, expected)
+
+    def __repr__(self):
+        return "IncompleteRead(%i bytes read, %i more expected)" % (
+            self.partial,
+            self.expected,
+        )
+
+
+class InvalidChunkLength(HTTPError, httplib_IncompleteRead):
+    """Invalid chunk length in a chunked response."""
+
+    def __init__(self, response, length):
+        super(InvalidChunkLength, self).__init__(
+            response.tell(), response.length_remaining
+        )
+        self.response = response
+        self.length = length
+
+    def __repr__(self):
+        return "InvalidChunkLength(got length %r, %i bytes read)" % (
+            self.length,
+            self.partial,
+        )
+
+
+class InvalidHeader(HTTPError):
+    """The header provided was somehow invalid."""
+
+    pass
+
+
+class ProxySchemeUnknown(AssertionError, URLSchemeUnknown):
+    """ProxyManager does not support the supplied scheme"""
+
+    # TODO(t-8ch): Stop inheriting from AssertionError in v2.0.
+
+    def __init__(self, scheme):
+        # 'localhost' is here because our URL parser parses
+        # localhost:8080 -> scheme=localhost, remove if we fix this.
+        if scheme == "localhost":
+            scheme = None
+        if scheme is None:
+            message = "Proxy URL had no scheme, should start with http:// or https://"
+        else:
+            message = (
+                "Proxy URL had unsupported scheme %s, should use http:// or https://"
+                % scheme
+            )
+        super(ProxySchemeUnknown, self).__init__(message)
+
+
+class ProxySchemeUnsupported(ValueError):
+    """Fetching HTTPS resources through HTTPS proxies is unsupported"""
+
+    pass
+
+
+class HeaderParsingError(HTTPError):
+    """Raised by assert_header_parsing, but we convert it to a log.warning statement."""
+
+    def __init__(self, defects, unparsed_data):
+        message = "%s, unparsed data: %r" % (defects or "Unknown", unparsed_data)
+        super(HeaderParsingError, self).__init__(message)
+
+
+class UnrewindableBodyError(HTTPError):
+    """urllib3 encountered an error when trying to rewind a body"""
+
+    pass

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/fields.py

@@ -0,0 +1,274 @@
+from __future__ import absolute_import
+
+import email.utils
+import mimetypes
+import re
+
+from .packages import six
+
+
+def guess_content_type(filename, default="application/octet-stream"):
+    """
+    Guess the "Content-Type" of a file.
+
+    :param filename:
+        The filename to guess the "Content-Type" of using :mod:`mimetypes`.
+    :param default:
+        If no "Content-Type" can be guessed, default to `default`.
+    """
+    if filename:
+        return mimetypes.guess_type(filename)[0] or default
+    return default
+
+
+def format_header_param_rfc2231(name, value):
+    """
+    Helper function to format and quote a single header parameter using the
+    strategy defined in RFC 2231.
+
+    Particularly useful for header parameters which might contain
+    non-ASCII values, like file names. This follows
+    `RFC 2388 Section 4.4 <https://tools.ietf.org/html/rfc2388#section-4.4>`_.
+
+    :param name:
+        The name of the parameter, a string expected to be ASCII only.
+    :param value:
+        The value of the parameter, provided as ``bytes`` or `str``.
+    :ret:
+        An RFC-2231-formatted unicode string.
+    """
+    if isinstance(value, six.binary_type):
+        value = value.decode("utf-8")
+
+    if not any(ch in value for ch in '"\\\r\n'):
+        result = u'%s="%s"' % (name, value)
+        try:
+            result.encode("ascii")
+        except (UnicodeEncodeError, UnicodeDecodeError):
+            pass
+        else:
+            return result
+
+    if six.PY2:  # Python 2:
+        value = value.encode("utf-8")
+
+    # encode_rfc2231 accepts an encoded string and returns an ascii-encoded
+    # string in Python 2 but accepts and returns unicode strings in Python 3
+    value = email.utils.encode_rfc2231(value, "utf-8")
+    value = "%s*=%s" % (name, value)
+
+    if six.PY2:  # Python 2:
+        value = value.decode("utf-8")
+
+    return value
+
+
+_HTML5_REPLACEMENTS = {
+    u"\u0022": u"%22",
+    # Replace "\" with "\\".
+    u"\u005C": u"\u005C\u005C",
+}
+
+# All control characters from 0x00 to 0x1F *except* 0x1B.
+_HTML5_REPLACEMENTS.update(
+    {
+        six.unichr(cc): u"%{:02X}".format(cc)
+        for cc in range(0x00, 0x1F + 1)
+        if cc not in (0x1B,)
+    }
+)
+
+
+def _replace_multiple(value, needles_and_replacements):
+    def replacer(match):
+        return needles_and_replacements[match.group(0)]
+
+    pattern = re.compile(
+        r"|".join([re.escape(needle) for needle in needles_and_replacements.keys()])
+    )
+
+    result = pattern.sub(replacer, value)
+
+    return result
+
+
+def format_header_param_html5(name, value):
+    """
+    Helper function to format and quote a single header parameter using the
+    HTML5 strategy.
+
+    Particularly useful for header parameters which might contain
+    non-ASCII values, like file names. This follows the `HTML5 Working Draft
+    Section 4.10.22.7`_ and matches the behavior of curl and modern browsers.
+
+    .. _HTML5 Working Draft Section 4.10.22.7:
+        https://w3c.github.io/html/sec-forms.html#multipart-form-data
+
+    :param name:
+        The name of the parameter, a string expected to be ASCII only.
+    :param value:
+        The value of the parameter, provided as ``bytes`` or `str``.
+    :ret:
+        A unicode string, stripped of troublesome characters.
+    """
+    if isinstance(value, six.binary_type):
+        value = value.decode("utf-8")
+
+    value = _replace_multiple(value, _HTML5_REPLACEMENTS)
+
+    return u'%s="%s"' % (name, value)
+
+
+# For backwards-compatibility.
+format_header_param = format_header_param_html5
+
+
+class RequestField(object):
+    """
+    A data container for request body parameters.
+
+    :param name:
+        The name of this request field. Must be unicode.
+    :param data:
+        The data/value body.
+    :param filename:
+        An optional filename of the request field. Must be unicode.
+    :param headers:
+        An optional dict-like object of headers to initially use for the field.
+    :param header_formatter:
+        An optional callable that is used to encode and format the headers. By
+        default, this is :func:`format_header_param_html5`.
+    """
+
+    def __init__(
+        self,
+        name,
+        data,
+        filename=None,
+        headers=None,
+        header_formatter=format_header_param_html5,
+    ):
+        self._name = name
+        self._filename = filename
+        self.data = data
+        self.headers = {}
+        if headers:
+            self.headers = dict(headers)
+        self.header_formatter = header_formatter
+
+    @classmethod
+    def from_tuples(cls, fieldname, value, header_formatter=format_header_param_html5):
+        """
+        A :class:`~urllib3.fields.RequestField` factory from old-style tuple parameters.
+
+        Supports constructing :class:`~urllib3.fields.RequestField` from
+        parameter of key/value strings AND key/filetuple. A filetuple is a
+        (filename, data, MIME type) tuple where the MIME type is optional.
+        For example::
+
+            'foo': 'bar',
+            'fakefile': ('foofile.txt', 'contents of foofile'),
+            'realfile': ('barfile.txt', open('realfile').read()),
+            'typedfile': ('bazfile.bin', open('bazfile').read(), 'image/jpeg'),
+            'nonamefile': 'contents of nonamefile field',
+
+        Field names and filenames must be unicode.
+        """
+        if isinstance(value, tuple):
+            if len(value) == 3:
+                filename, data, content_type = value
+            else:
+                filename, data = value
+                content_type = guess_content_type(filename)
+        else:
+            filename = None
+            content_type = None
+            data = value
+
+        request_param = cls(
+            fieldname, data, filename=filename, header_formatter=header_formatter
+        )
+        request_param.make_multipart(content_type=content_type)
+
+        return request_param
+
+    def _render_part(self, name, value):
+        """
+        Overridable helper function to format a single header parameter. By
+        default, this calls ``self.header_formatter``.
+
+        :param name:
+            The name of the parameter, a string expected to be ASCII only.
+        :param value:
+            The value of the parameter, provided as a unicode string.
+        """
+
+        return self.header_formatter(name, value)
+
+    def _render_parts(self, header_parts):
+        """
+        Helper function to format and quote a single header.
+
+        Useful for single headers that are composed of multiple items. E.g.,
+        'Content-Disposition' fields.
+
+        :param header_parts:
+            A sequence of (k, v) tuples or a :class:`dict` of (k, v) to format
+            as `k1="v1"; k2="v2"; ...`.
+        """
+        parts = []
+        iterable = header_parts
+        if isinstance(header_parts, dict):
+            iterable = header_parts.items()
+
+        for name, value in iterable:
+            if value is not None:
+                parts.append(self._render_part(name, value))
+
+        return u"; ".join(parts)
+
+    def render_headers(self):
+        """
+        Renders the headers for this request field.
+        """
+        lines = []
+
+        sort_keys = ["Content-Disposition", "Content-Type", "Content-Location"]
+        for sort_key in sort_keys:
+            if self.headers.get(sort_key, False):
+                lines.append(u"%s: %s" % (sort_key, self.headers[sort_key]))
+
+        for header_name, header_value in self.headers.items():
+            if header_name not in sort_keys:
+                if header_value:
+                    lines.append(u"%s: %s" % (header_name, header_value))
+
+        lines.append(u"\r\n")
+        return u"\r\n".join(lines)
+
+    def make_multipart(
+        self, content_disposition=None, content_type=None, content_location=None
+    ):
+        """
+        Makes this request field into a multipart request field.
+
+        This method overrides "Content-Disposition", "Content-Type" and
+        "Content-Location" headers to the request parameter.
+
+        :param content_type:
+            The 'Content-Type' of the request body.
+        :param content_location:
+            The 'Content-Location' of the request body.
+
+        """
+        self.headers["Content-Disposition"] = content_disposition or u"form-data"
+        self.headers["Content-Disposition"] += u"; ".join(
+            [
+                u"",
+                self._render_parts(
+                    ((u"name", self._name), (u"filename", self._filename))
+                ),
+            ]
+        )
+        self.headers["Content-Type"] = content_type
+        self.headers["Content-Location"] = content_location

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/filepost.py

@@ -0,0 +1,98 @@
+from __future__ import absolute_import
+
+import binascii
+import codecs
+import os
+from io import BytesIO
+
+from .fields import RequestField
+from .packages import six
+from .packages.six import b
+
+writer = codecs.lookup("utf-8")[3]
+
+
+def choose_boundary():
+    """
+    Our embarrassingly-simple replacement for mimetools.choose_boundary.
+    """
+    boundary = binascii.hexlify(os.urandom(16))
+    if not six.PY2:
+        boundary = boundary.decode("ascii")
+    return boundary
+
+
+def iter_field_objects(fields):
+    """
+    Iterate over fields.
+
+    Supports list of (k, v) tuples and dicts, and lists of
+    :class:`~urllib3.fields.RequestField`.
+
+    """
+    if isinstance(fields, dict):
+        i = six.iteritems(fields)
+    else:
+        i = iter(fields)
+
+    for field in i:
+        if isinstance(field, RequestField):
+            yield field
+        else:
+            yield RequestField.from_tuples(*field)
+
+
+def iter_fields(fields):
+    """
+    .. deprecated:: 1.6
+
+    Iterate over fields.
+
+    The addition of :class:`~urllib3.fields.RequestField` makes this function
+    obsolete. Instead, use :func:`iter_field_objects`, which returns
+    :class:`~urllib3.fields.RequestField` objects.
+
+    Supports list of (k, v) tuples and dicts.
+    """
+    if isinstance(fields, dict):
+        return ((k, v) for k, v in six.iteritems(fields))
+
+    return ((k, v) for k, v in fields)
+
+
+def encode_multipart_formdata(fields, boundary=None):
+    """
+    Encode a dictionary of ``fields`` using the multipart/form-data MIME format.
+
+    :param fields:
+        Dictionary of fields or list of (key, :class:`~urllib3.fields.RequestField`).
+
+    :param boundary:
+        If not specified, then a random boundary will be generated using
+        :func:`urllib3.filepost.choose_boundary`.
+    """
+    body = BytesIO()
+    if boundary is None:
+        boundary = choose_boundary()
+
+    for field in iter_field_objects(fields):
+        body.write(b("--%s\r\n" % (boundary)))
+
+        writer(body).write(field.render_headers())
+        data = field.data
+
+        if isinstance(data, int):
+            data = str(data)  # Backwards compatibility
+
+        if isinstance(data, six.text_type):
+            writer(body).write(data)
+        else:
+            body.write(data)
+
+        body.write(b"\r\n")
+
+    body.write(b("--%s--\r\n" % (boundary)))
+
+    content_type = str("multipart/form-data; boundary=%s" % boundary)
+
+    return body.getvalue(), content_type

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/poolmanager.py

@@ -0,0 +1,537 @@
+from __future__ import absolute_import
+
+import collections
+import functools
+import logging
+
+from ._collections import RecentlyUsedContainer
+from .connectionpool import HTTPConnectionPool, HTTPSConnectionPool, port_by_scheme
+from .exceptions import (
+    LocationValueError,
+    MaxRetryError,
+    ProxySchemeUnknown,
+    ProxySchemeUnsupported,
+    URLSchemeUnknown,
+)
+from .packages import six
+from .packages.six.moves.urllib.parse import urljoin
+from .request import RequestMethods
+from .util.proxy import connection_requires_http_tunnel
+from .util.retry import Retry
+from .util.url import parse_url
+
+__all__ = ["PoolManager", "ProxyManager", "proxy_from_url"]
+
+
+log = logging.getLogger(__name__)
+
+SSL_KEYWORDS = (
+    "key_file",
+    "cert_file",
+    "cert_reqs",
+    "ca_certs",
+    "ssl_version",
+    "ca_cert_dir",
+    "ssl_context",
+    "key_password",
+    "server_hostname",
+)
+
+# All known keyword arguments that could be provided to the pool manager, its
+# pools, or the underlying connections. This is used to construct a pool key.
+_key_fields = (
+    "key_scheme",  # str
+    "key_host",  # str
+    "key_port",  # int
+    "key_timeout",  # int or float or Timeout
+    "key_retries",  # int or Retry
+    "key_strict",  # bool
+    "key_block",  # bool
+    "key_source_address",  # str
+    "key_key_file",  # str
+    "key_key_password",  # str
+    "key_cert_file",  # str
+    "key_cert_reqs",  # str
+    "key_ca_certs",  # str
+    "key_ssl_version",  # str
+    "key_ca_cert_dir",  # str
+    "key_ssl_context",  # instance of ssl.SSLContext or urllib3.util.ssl_.SSLContext
+    "key_maxsize",  # int
+    "key_headers",  # dict
+    "key__proxy",  # parsed proxy url
+    "key__proxy_headers",  # dict
+    "key__proxy_config",  # class
+    "key_socket_options",  # list of (level (int), optname (int), value (int or str)) tuples
+    "key__socks_options",  # dict
+    "key_assert_hostname",  # bool or string
+    "key_assert_fingerprint",  # str
+    "key_server_hostname",  # str
+)
+
+#: The namedtuple class used to construct keys for the connection pool.
+#: All custom key schemes should include the fields in this key at a minimum.
+PoolKey = collections.namedtuple("PoolKey", _key_fields)
+
+_proxy_config_fields = ("ssl_context", "use_forwarding_for_https")
+ProxyConfig = collections.namedtuple("ProxyConfig", _proxy_config_fields)
+
+
+def _default_key_normalizer(key_class, request_context):
+    """
+    Create a pool key out of a request context dictionary.
+
+    According to RFC 3986, both the scheme and host are case-insensitive.
+    Therefore, this function normalizes both before constructing the pool
+    key for an HTTPS request. If you wish to change this behaviour, provide
+    alternate callables to ``key_fn_by_scheme``.
+
+    :param key_class:
+        The class to use when constructing the key. This should be a namedtuple
+        with the ``scheme`` and ``host`` keys at a minimum.
+    :type  key_class: namedtuple
+    :param request_context:
+        A dictionary-like object that contain the context for a request.
+    :type  request_context: dict
+
+    :return: A namedtuple that can be used as a connection pool key.
+    :rtype:  PoolKey
+    """
+    # Since we mutate the dictionary, make a copy first
+    context = request_context.copy()
+    context["scheme"] = context["scheme"].lower()
+    context["host"] = context["host"].lower()
+
+    # These are both dictionaries and need to be transformed into frozensets
+    for key in ("headers", "_proxy_headers", "_socks_options"):
+        if key in context and context[key] is not None:
+            context[key] = frozenset(context[key].items())
+
+    # The socket_options key may be a list and needs to be transformed into a
+    # tuple.
+    socket_opts = context.get("socket_options")
+    if socket_opts is not None:
+        context["socket_options"] = tuple(socket_opts)
+
+    # Map the kwargs to the names in the namedtuple - this is necessary since
+    # namedtuples can't have fields starting with '_'.
+    for key in list(context.keys()):
+        context["key_" + key] = context.pop(key)
+
+    # Default to ``None`` for keys missing from the context
+    for field in key_class._fields:
+        if field not in context:
+            context[field] = None
+
+    return key_class(**context)
+
+
+#: A dictionary that maps a scheme to a callable that creates a pool key.
+#: This can be used to alter the way pool keys are constructed, if desired.
+#: Each PoolManager makes a copy of this dictionary so they can be configured
+#: globally here, or individually on the instance.
+key_fn_by_scheme = {
+    "http": functools.partial(_default_key_normalizer, PoolKey),
+    "https": functools.partial(_default_key_normalizer, PoolKey),
+}
+
+pool_classes_by_scheme = {"http": HTTPConnectionPool, "https": HTTPSConnectionPool}
+
+
+class PoolManager(RequestMethods):
+    """
+    Allows for arbitrary requests while transparently keeping track of
+    necessary connection pools for you.
+
+    :param num_pools:
+        Number of connection pools to cache before discarding the least
+        recently used pool.
+
+    :param headers:
+        Headers to include with all requests, unless other headers are given
+        explicitly.
+
+    :param \\**connection_pool_kw:
+        Additional parameters are used to create fresh
+        :class:`urllib3.connectionpool.ConnectionPool` instances.
+
+    Example::
+
+        >>> manager = PoolManager(num_pools=2)
+        >>> r = manager.request('GET', 'http://google.com/')
+        >>> r = manager.request('GET', 'http://google.com/mail')
+        >>> r = manager.request('GET', 'http://yahoo.com/')
+        >>> len(manager.pools)
+        2
+
+    """
+
+    proxy = None
+    proxy_config = None
+
+    def __init__(self, num_pools=10, headers=None, **connection_pool_kw):
+        RequestMethods.__init__(self, headers)
+        self.connection_pool_kw = connection_pool_kw
+        self.pools = RecentlyUsedContainer(num_pools)
+
+        # Locally set the pool classes and keys so other PoolManagers can
+        # override them.
+        self.pool_classes_by_scheme = pool_classes_by_scheme
+        self.key_fn_by_scheme = key_fn_by_scheme.copy()
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.clear()
+        # Return False to re-raise any potential exceptions
+        return False
+
+    def _new_pool(self, scheme, host, port, request_context=None):
+        """
+        Create a new :class:`urllib3.connectionpool.ConnectionPool` based on host, port, scheme, and
+        any additional pool keyword arguments.
+
+        If ``request_context`` is provided, it is provided as keyword arguments
+        to the pool class used. This method is used to actually create the
+        connection pools handed out by :meth:`connection_from_url` and
+        companion methods. It is intended to be overridden for customization.
+        """
+        pool_cls = self.pool_classes_by_scheme[scheme]
+        if request_context is None:
+            request_context = self.connection_pool_kw.copy()
+
+        # Although the context has everything necessary to create the pool,
+        # this function has historically only used the scheme, host, and port
+        # in the positional args. When an API change is acceptable these can
+        # be removed.
+        for key in ("scheme", "host", "port"):
+            request_context.pop(key, None)
+
+        if scheme == "http":
+            for kw in SSL_KEYWORDS:
+                request_context.pop(kw, None)
+
+        return pool_cls(host, port, **request_context)
+
+    def clear(self):
+        """
+        Empty our store of pools and direct them all to close.
+
+        This will not affect in-flight connections, but they will not be
+        re-used after completion.
+        """
+        self.pools.clear()
+
+    def connection_from_host(self, host, port=None, scheme="http", pool_kwargs=None):
+        """
+        Get a :class:`urllib3.connectionpool.ConnectionPool` based on the host, port, and scheme.
+
+        If ``port`` isn't given, it will be derived from the ``scheme`` using
+        ``urllib3.connectionpool.port_by_scheme``. If ``pool_kwargs`` is
+        provided, it is merged with the instance's ``connection_pool_kw``
+        variable and used to create the new connection pool, if one is
+        needed.
+        """
+
+        if not host:
+            raise LocationValueError("No host specified.")
+
+        request_context = self._merge_pool_kwargs(pool_kwargs)
+        request_context["scheme"] = scheme or "http"
+        if not port:
+            port = port_by_scheme.get(request_context["scheme"].lower(), 80)
+        request_context["port"] = port
+        request_context["host"] = host
+
+        return self.connection_from_context(request_context)
+
+    def connection_from_context(self, request_context):
+        """
+        Get a :class:`urllib3.connectionpool.ConnectionPool` based on the request context.
+
+        ``request_context`` must at least contain the ``scheme`` key and its
+        value must be a key in ``key_fn_by_scheme`` instance variable.
+        """
+        scheme = request_context["scheme"].lower()
+        pool_key_constructor = self.key_fn_by_scheme.get(scheme)
+        if not pool_key_constructor:
+            raise URLSchemeUnknown(scheme)
+        pool_key = pool_key_constructor(request_context)
+
+        return self.connection_from_pool_key(pool_key, request_context=request_context)
+
+    def connection_from_pool_key(self, pool_key, request_context=None):
+        """
+        Get a :class:`urllib3.connectionpool.ConnectionPool` based on the provided pool key.
+
+        ``pool_key`` should be a namedtuple that only contains immutable
+        objects. At a minimum it must have the ``scheme``, ``host``, and
+        ``port`` fields.
+        """
+        with self.pools.lock:
+            # If the scheme, host, or port doesn't match existing open
+            # connections, open a new ConnectionPool.
+            pool = self.pools.get(pool_key)
+            if pool:
+                return pool
+
+            # Make a fresh ConnectionPool of the desired type
+            scheme = request_context["scheme"]
+            host = request_context["host"]
+            port = request_context["port"]
+            pool = self._new_pool(scheme, host, port, request_context=request_context)
+            self.pools[pool_key] = pool
+
+        return pool
+
+    def connection_from_url(self, url, pool_kwargs=None):
+        """
+        Similar to :func:`urllib3.connectionpool.connection_from_url`.
+
+        If ``pool_kwargs`` is not provided and a new pool needs to be
+        constructed, ``self.connection_pool_kw`` is used to initialize
+        the :class:`urllib3.connectionpool.ConnectionPool`. If ``pool_kwargs``
+        is provided, it is used instead. Note that if a new pool does not
+        need to be created for the request, the provided ``pool_kwargs`` are
+        not used.
+        """
+        u = parse_url(url)
+        return self.connection_from_host(
+            u.host, port=u.port, scheme=u.scheme, pool_kwargs=pool_kwargs
+        )
+
+    def _merge_pool_kwargs(self, override):
+        """
+        Merge a dictionary of override values for self.connection_pool_kw.
+
+        This does not modify self.connection_pool_kw and returns a new dict.
+        Any keys in the override dictionary with a value of ``None`` are
+        removed from the merged dictionary.
+        """
+        base_pool_kwargs = self.connection_pool_kw.copy()
+        if override:
+            for key, value in override.items():
+                if value is None:
+                    try:
+                        del base_pool_kwargs[key]
+                    except KeyError:
+                        pass
+                else:
+                    base_pool_kwargs[key] = value
+        return base_pool_kwargs
+
+    def _proxy_requires_url_absolute_form(self, parsed_url):
+        """
+        Indicates if the proxy requires the complete destination URL in the
+        request.  Normally this is only needed when not using an HTTP CONNECT
+        tunnel.
+        """
+        if self.proxy is None:
+            return False
+
+        return not connection_requires_http_tunnel(
+            self.proxy, self.proxy_config, parsed_url.scheme
+        )
+
+    def _validate_proxy_scheme_url_selection(self, url_scheme):
+        """
+        Validates that were not attempting to do TLS in TLS connections on
+        Python2 or with unsupported SSL implementations.
+        """
+        if self.proxy is None or url_scheme != "https":
+            return
+
+        if self.proxy.scheme != "https":
+            return
+
+        if six.PY2 and not self.proxy_config.use_forwarding_for_https:
+            raise ProxySchemeUnsupported(
+                "Contacting HTTPS destinations through HTTPS proxies "
+                "'via CONNECT tunnels' is not supported in Python 2"
+            )
+
+    def urlopen(self, method, url, redirect=True, **kw):
+        """
+        Same as :meth:`urllib3.HTTPConnectionPool.urlopen`
+        with custom cross-host redirect logic and only sends the request-uri
+        portion of the ``url``.
+
+        The given ``url`` parameter must be absolute, such that an appropriate
+        :class:`urllib3.connectionpool.ConnectionPool` can be chosen for it.
+        """
+        u = parse_url(url)
+        self._validate_proxy_scheme_url_selection(u.scheme)
+
+        conn = self.connection_from_host(u.host, port=u.port, scheme=u.scheme)
+
+        kw["assert_same_host"] = False
+        kw["redirect"] = False
+
+        if "headers" not in kw:
+            kw["headers"] = self.headers.copy()
+
+        if self._proxy_requires_url_absolute_form(u):
+            response = conn.urlopen(method, url, **kw)
+        else:
+            response = conn.urlopen(method, u.request_uri, **kw)
+
+        redirect_location = redirect and response.get_redirect_location()
+        if not redirect_location:
+            return response
+
+        # Support relative URLs for redirecting.
+        redirect_location = urljoin(url, redirect_location)
+
+        # RFC 7231, Section 6.4.4
+        if response.status == 303:
+            method = "GET"
+
+        retries = kw.get("retries")
+        if not isinstance(retries, Retry):
+            retries = Retry.from_int(retries, redirect=redirect)
+
+        # Strip headers marked as unsafe to forward to the redirected location.
+        # Check remove_headers_on_redirect to avoid a potential network call within
+        # conn.is_same_host() which may use socket.gethostbyname() in the future.
+        if retries.remove_headers_on_redirect and not conn.is_same_host(
+            redirect_location
+        ):
+            headers = list(six.iterkeys(kw["headers"]))
+            for header in headers:
+                if header.lower() in retries.remove_headers_on_redirect:
+                    kw["headers"].pop(header, None)
+
+        try:
+            retries = retries.increment(method, url, response=response, _pool=conn)
+        except MaxRetryError:
+            if retries.raise_on_redirect:
+                response.drain_conn()
+                raise
+            return response
+
+        kw["retries"] = retries
+        kw["redirect"] = redirect
+
+        log.info("Redirecting %s -> %s", url, redirect_location)
+
+        response.drain_conn()
+        return self.urlopen(method, redirect_location, **kw)
+
+
+class ProxyManager(PoolManager):
+    """
+    Behaves just like :class:`PoolManager`, but sends all requests through
+    the defined proxy, using the CONNECT method for HTTPS URLs.
+
+    :param proxy_url:
+        The URL of the proxy to be used.
+
+    :param proxy_headers:
+        A dictionary containing headers that will be sent to the proxy. In case
+        of HTTP they are being sent with each request, while in the
+        HTTPS/CONNECT case they are sent only once. Could be used for proxy
+        authentication.
+
+    :param proxy_ssl_context:
+        The proxy SSL context is used to establish the TLS connection to the
+        proxy when using HTTPS proxies.
+
+    :param use_forwarding_for_https:
+        (Defaults to False) If set to True will forward requests to the HTTPS
+        proxy to be made on behalf of the client instead of creating a TLS
+        tunnel via the CONNECT method. **Enabling this flag means that request
+        and response headers and content will be visible from the HTTPS proxy**
+        whereas tunneling keeps request and response headers and content
+        private.  IP address, target hostname, SNI, and port are always visible
+        to an HTTPS proxy even when this flag is disabled.
+
+    Example:
+        >>> proxy = urllib3.ProxyManager('http://localhost:3128/')
+        >>> r1 = proxy.request('GET', 'http://google.com/')
+        >>> r2 = proxy.request('GET', 'http://httpbin.org/')
+        >>> len(proxy.pools)
+        1
+        >>> r3 = proxy.request('GET', 'https://httpbin.org/')
+        >>> r4 = proxy.request('GET', 'https://twitter.com/')
+        >>> len(proxy.pools)
+        3
+
+    """
+
+    def __init__(
+        self,
+        proxy_url,
+        num_pools=10,
+        headers=None,
+        proxy_headers=None,
+        proxy_ssl_context=None,
+        use_forwarding_for_https=False,
+        **connection_pool_kw
+    ):
+
+        if isinstance(proxy_url, HTTPConnectionPool):
+            proxy_url = "%s://%s:%i" % (
+                proxy_url.scheme,
+                proxy_url.host,
+                proxy_url.port,
+            )
+        proxy = parse_url(proxy_url)
+
+        if proxy.scheme not in ("http", "https"):
+            raise ProxySchemeUnknown(proxy.scheme)
+
+        if not proxy.port:
+            port = port_by_scheme.get(proxy.scheme, 80)
+            proxy = proxy._replace(port=port)
+
+        self.proxy = proxy
+        self.proxy_headers = proxy_headers or {}
+        self.proxy_ssl_context = proxy_ssl_context
+        self.proxy_config = ProxyConfig(proxy_ssl_context, use_forwarding_for_https)
+
+        connection_pool_kw["_proxy"] = self.proxy
+        connection_pool_kw["_proxy_headers"] = self.proxy_headers
+        connection_pool_kw["_proxy_config"] = self.proxy_config
+
+        super(ProxyManager, self).__init__(num_pools, headers, **connection_pool_kw)
+
+    def connection_from_host(self, host, port=None, scheme="http", pool_kwargs=None):
+        if scheme == "https":
+            return super(ProxyManager, self).connection_from_host(
+                host, port, scheme, pool_kwargs=pool_kwargs
+            )
+
+        return super(ProxyManager, self).connection_from_host(
+            self.proxy.host, self.proxy.port, self.proxy.scheme, pool_kwargs=pool_kwargs
+        )
+
+    def _set_proxy_headers(self, url, headers=None):
+        """
+        Sets headers needed by proxies: specifically, the Accept and Host
+        headers. Only sets headers not provided by the user.
+        """
+        headers_ = {"Accept": "*/*"}
+
+        netloc = parse_url(url).netloc
+        if netloc:
+            headers_["Host"] = netloc
+
+        if headers:
+            headers_.update(headers)
+        return headers_
+
+    def urlopen(self, method, url, redirect=True, **kw):
+        "Same as HTTP(S)ConnectionPool.urlopen, ``url`` must be absolute."
+        u = parse_url(url)
+        if not connection_requires_http_tunnel(self.proxy, self.proxy_config, u.scheme):
+            # For connections using HTTP CONNECT, httplib sets the necessary
+            # headers on the CONNECT to the proxy. If we're not using CONNECT,
+            # we'll definitely need to set 'Host' at the very least.
+            headers = kw.get("headers", self.headers)
+            kw["headers"] = self._set_proxy_headers(url, headers)
+
+        return super(ProxyManager, self).urlopen(method, url, redirect=redirect, **kw)
+
+
+def proxy_from_url(url, **kw):
+    return ProxyManager(proxy_url=url, **kw)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/request.py

@@ -0,0 +1,191 @@
+from __future__ import absolute_import
+
+import sys
+
+from .filepost import encode_multipart_formdata
+from .packages import six
+from .packages.six.moves.urllib.parse import urlencode
+
+__all__ = ["RequestMethods"]
+
+
+class RequestMethods(object):
+    """
+    Convenience mixin for classes who implement a :meth:`urlopen` method, such
+    as :class:`urllib3.HTTPConnectionPool` and
+    :class:`urllib3.PoolManager`.
+
+    Provides behavior for making common types of HTTP request methods and
+    decides which type of request field encoding to use.
+
+    Specifically,
+
+    :meth:`.request_encode_url` is for sending requests whose fields are
+    encoded in the URL (such as GET, HEAD, DELETE).
+
+    :meth:`.request_encode_body` is for sending requests whose fields are
+    encoded in the *body* of the request using multipart or www-form-urlencoded
+    (such as for POST, PUT, PATCH).
+
+    :meth:`.request` is for making any kind of request, it will look up the
+    appropriate encoding format and use one of the above two methods to make
+    the request.
+
+    Initializer parameters:
+
+    :param headers:
+        Headers to include with all requests, unless other headers are given
+        explicitly.
+    """
+
+    _encode_url_methods = {"DELETE", "GET", "HEAD", "OPTIONS"}
+
+    def __init__(self, headers=None):
+        self.headers = headers or {}
+
+    def urlopen(
+        self,
+        method,
+        url,
+        body=None,
+        headers=None,
+        encode_multipart=True,
+        multipart_boundary=None,
+        **kw
+    ):  # Abstract
+        raise NotImplementedError(
+            "Classes extending RequestMethods must implement "
+            "their own ``urlopen`` method."
+        )
+
+    def request(self, method, url, fields=None, headers=None, **urlopen_kw):
+        """
+        Make a request using :meth:`urlopen` with the appropriate encoding of
+        ``fields`` based on the ``method`` used.
+
+        This is a convenience method that requires the least amount of manual
+        effort. It can be used in most situations, while still having the
+        option to drop down to more specific methods when necessary, such as
+        :meth:`request_encode_url`, :meth:`request_encode_body`,
+        or even the lowest level :meth:`urlopen`.
+        """
+        method = method.upper()
+
+        urlopen_kw["request_url"] = url
+
+        if method in self._encode_url_methods:
+            return self.request_encode_url(
+                method, url, fields=fields, headers=headers, **urlopen_kw
+            )
+        else:
+            return self.request_encode_body(
+                method, url, fields=fields, headers=headers, **urlopen_kw
+            )
+
+    def request_encode_url(self, method, url, fields=None, headers=None, **urlopen_kw):
+        """
+        Make a request using :meth:`urlopen` with the ``fields`` encoded in
+        the url. This is useful for request methods like GET, HEAD, DELETE, etc.
+        """
+        if headers is None:
+            headers = self.headers
+
+        extra_kw = {"headers": headers}
+        extra_kw.update(urlopen_kw)
+
+        if fields:
+            url += "?" + urlencode(fields)
+
+        return self.urlopen(method, url, **extra_kw)
+
+    def request_encode_body(
+        self,
+        method,
+        url,
+        fields=None,
+        headers=None,
+        encode_multipart=True,
+        multipart_boundary=None,
+        **urlopen_kw
+    ):
+        """
+        Make a request using :meth:`urlopen` with the ``fields`` encoded in
+        the body. This is useful for request methods like POST, PUT, PATCH, etc.
+
+        When ``encode_multipart=True`` (default), then
+        :func:`urllib3.encode_multipart_formdata` is used to encode
+        the payload with the appropriate content type. Otherwise
+        :func:`urllib.parse.urlencode` is used with the
+        'application/x-www-form-urlencoded' content type.
+
+        Multipart encoding must be used when posting files, and it's reasonably
+        safe to use it in other times too. However, it may break request
+        signing, such as with OAuth.
+
+        Supports an optional ``fields`` parameter of key/value strings AND
+        key/filetuple. A filetuple is a (filename, data, MIME type) tuple where
+        the MIME type is optional. For example::
+
+            fields = {
+                'foo': 'bar',
+                'fakefile': ('foofile.txt', 'contents of foofile'),
+                'realfile': ('barfile.txt', open('realfile').read()),
+                'typedfile': ('bazfile.bin', open('bazfile').read(),
+                              'image/jpeg'),
+                'nonamefile': 'contents of nonamefile field',
+            }
+
+        When uploading a file, providing a filename (the first parameter of the
+        tuple) is optional but recommended to best mimic behavior of browsers.
+
+        Note that if ``headers`` are supplied, the 'Content-Type' header will
+        be overwritten because it depends on the dynamic random boundary string
+        which is used to compose the body of the request. The random boundary
+        string can be explicitly set with the ``multipart_boundary`` parameter.
+        """
+        if headers is None:
+            headers = self.headers
+
+        extra_kw = {"headers": {}}
+
+        if fields:
+            if "body" in urlopen_kw:
+                raise TypeError(
+                    "request got values for both 'fields' and 'body', can only specify one."
+                )
+
+            if encode_multipart:
+                body, content_type = encode_multipart_formdata(
+                    fields, boundary=multipart_boundary
+                )
+            else:
+                body, content_type = (
+                    urlencode(fields),
+                    "application/x-www-form-urlencoded",
+                )
+
+            extra_kw["body"] = body
+            extra_kw["headers"] = {"Content-Type": content_type}
+
+        extra_kw["headers"].update(headers)
+        extra_kw.update(urlopen_kw)
+
+        return self.urlopen(method, url, **extra_kw)
+
+
+if not six.PY2:
+
+    class RequestModule(sys.modules[__name__].__class__):
+        def __call__(self, *args, **kwargs):
+            """
+            If user tries to call this module directly urllib3 v2.x style raise an error to the user
+            suggesting they may need urllib3 v2
+            """
+            raise TypeError(
+                "'module' object is not callable\n"
+                "urllib3.request() method is not supported in this release, "
+                "upgrade to urllib3 v2 to use it\n"
+                "see https://urllib3.readthedocs.io/en/stable/v2-migration-guide.html"
+            )
+
+    sys.modules[__name__].__class__ = RequestModule

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/response.py

@@ -0,0 +1,879 @@
+from __future__ import absolute_import
+
+import io
+import logging
+import sys
+import warnings
+import zlib
+from contextlib import contextmanager
+from socket import error as SocketError
+from socket import timeout as SocketTimeout
+
+brotli = None
+
+from . import util
+from ._collections import HTTPHeaderDict
+from .connection import BaseSSLError, HTTPException
+from .exceptions import (
+    BodyNotHttplibCompatible,
+    DecodeError,
+    HTTPError,
+    IncompleteRead,
+    InvalidChunkLength,
+    InvalidHeader,
+    ProtocolError,
+    ReadTimeoutError,
+    ResponseNotChunked,
+    SSLError,
+)
+from .packages import six
+from .util.response import is_fp_closed, is_response_to_head
+
+log = logging.getLogger(__name__)
+
+
+class DeflateDecoder(object):
+    def __init__(self):
+        self._first_try = True
+        self._data = b""
+        self._obj = zlib.decompressobj()
+
+    def __getattr__(self, name):
+        return getattr(self._obj, name)
+
+    def decompress(self, data):
+        if not data:
+            return data
+
+        if not self._first_try:
+            return self._obj.decompress(data)
+
+        self._data += data
+        try:
+            decompressed = self._obj.decompress(data)
+            if decompressed:
+                self._first_try = False
+                self._data = None
+            return decompressed
+        except zlib.error:
+            self._first_try = False
+            self._obj = zlib.decompressobj(-zlib.MAX_WBITS)
+            try:
+                return self.decompress(self._data)
+            finally:
+                self._data = None
+
+
+class GzipDecoderState(object):
+
+    FIRST_MEMBER = 0
+    OTHER_MEMBERS = 1
+    SWALLOW_DATA = 2
+
+
+class GzipDecoder(object):
+    def __init__(self):
+        self._obj = zlib.decompressobj(16 + zlib.MAX_WBITS)
+        self._state = GzipDecoderState.FIRST_MEMBER
+
+    def __getattr__(self, name):
+        return getattr(self._obj, name)
+
+    def decompress(self, data):
+        ret = bytearray()
+        if self._state == GzipDecoderState.SWALLOW_DATA or not data:
+            return bytes(ret)
+        while True:
+            try:
+                ret += self._obj.decompress(data)
+            except zlib.error:
+                previous_state = self._state
+                # Ignore data after the first error
+                self._state = GzipDecoderState.SWALLOW_DATA
+                if previous_state == GzipDecoderState.OTHER_MEMBERS:
+                    # Allow trailing garbage acceptable in other gzip clients
+                    return bytes(ret)
+                raise
+            data = self._obj.unused_data
+            if not data:
+                return bytes(ret)
+            self._state = GzipDecoderState.OTHER_MEMBERS
+            self._obj = zlib.decompressobj(16 + zlib.MAX_WBITS)
+
+
+if brotli is not None:
+
+    class BrotliDecoder(object):
+        # Supports both 'brotlipy' and 'Brotli' packages
+        # since they share an import name. The top branches
+        # are for 'brotlipy' and bottom branches for 'Brotli'
+        def __init__(self):
+            self._obj = brotli.Decompressor()
+            if hasattr(self._obj, "decompress"):
+                self.decompress = self._obj.decompress
+            else:
+                self.decompress = self._obj.process
+
+        def flush(self):
+            if hasattr(self._obj, "flush"):
+                return self._obj.flush()
+            return b""
+
+
+class MultiDecoder(object):
+    """
+    From RFC7231:
+        If one or more encodings have been applied to a representation, the
+        sender that applied the encodings MUST generate a Content-Encoding
+        header field that lists the content codings in the order in which
+        they were applied.
+    """
+
+    def __init__(self, modes):
+        self._decoders = [_get_decoder(m.strip()) for m in modes.split(",")]
+
+    def flush(self):
+        return self._decoders[0].flush()
+
+    def decompress(self, data):
+        for d in reversed(self._decoders):
+            data = d.decompress(data)
+        return data
+
+
+def _get_decoder(mode):
+    if "," in mode:
+        return MultiDecoder(mode)
+
+    if mode == "gzip":
+        return GzipDecoder()
+
+    if brotli is not None and mode == "br":
+        return BrotliDecoder()
+
+    return DeflateDecoder()
+
+
+class HTTPResponse(io.IOBase):
+    """
+    HTTP Response container.
+
+    Backwards-compatible with :class:`http.client.HTTPResponse` but the response ``body`` is
+    loaded and decoded on-demand when the ``data`` property is accessed.  This
+    class is also compatible with the Python standard library's :mod:`io`
+    module, and can hence be treated as a readable object in the context of that
+    framework.
+
+    Extra parameters for behaviour not present in :class:`http.client.HTTPResponse`:
+
+    :param preload_content:
+        If True, the response's body will be preloaded during construction.
+
+    :param decode_content:
+        If True, will attempt to decode the body based on the
+        'content-encoding' header.
+
+    :param original_response:
+        When this HTTPResponse wrapper is generated from an :class:`http.client.HTTPResponse`
+        object, it's convenient to include the original for debug purposes. It's
+        otherwise unused.
+
+    :param retries:
+        The retries contains the last :class:`~urllib3.util.retry.Retry` that
+        was used during the request.
+
+    :param enforce_content_length:
+        Enforce content length checking. Body returned by server must match
+        value of Content-Length header, if present. Otherwise, raise error.
+    """
+
+    CONTENT_DECODERS = ["gzip", "deflate"]
+    if brotli is not None:
+        CONTENT_DECODERS += ["br"]
+    REDIRECT_STATUSES = [301, 302, 303, 307, 308]
+
+    def __init__(
+        self,
+        body="",
+        headers=None,
+        status=0,
+        version=0,
+        reason=None,
+        strict=0,
+        preload_content=True,
+        decode_content=True,
+        original_response=None,
+        pool=None,
+        connection=None,
+        msg=None,
+        retries=None,
+        enforce_content_length=False,
+        request_method=None,
+        request_url=None,
+        auto_close=True,
+    ):
+
+        if isinstance(headers, HTTPHeaderDict):
+            self.headers = headers
+        else:
+            self.headers = HTTPHeaderDict(headers)
+        self.status = status
+        self.version = version
+        self.reason = reason
+        self.strict = strict
+        self.decode_content = decode_content
+        self.retries = retries
+        self.enforce_content_length = enforce_content_length
+        self.auto_close = auto_close
+
+        self._decoder = None
+        self._body = None
+        self._fp = None
+        self._original_response = original_response
+        self._fp_bytes_read = 0
+        self.msg = msg
+        self._request_url = request_url
+
+        if body and isinstance(body, (six.string_types, bytes)):
+            self._body = body
+
+        self._pool = pool
+        self._connection = connection
+
+        if hasattr(body, "read"):
+            self._fp = body
+
+        # Are we using the chunked-style of transfer encoding?
+        self.chunked = False
+        self.chunk_left = None
+        tr_enc = self.headers.get("transfer-encoding", "").lower()
+        # Don't incur the penalty of creating a list and then discarding it
+        encodings = (enc.strip() for enc in tr_enc.split(","))
+        if "chunked" in encodings:
+            self.chunked = True
+
+        # Determine length of response
+        self.length_remaining = self._init_length(request_method)
+
+        # If requested, preload the body.
+        if preload_content and not self._body:
+            self._body = self.read(decode_content=decode_content)
+
+    def get_redirect_location(self):
+        """
+        Should we redirect and where to?
+
+        :returns: Truthy redirect location string if we got a redirect status
+            code and valid location. ``None`` if redirect status and no
+            location. ``False`` if not a redirect status code.
+        """
+        if self.status in self.REDIRECT_STATUSES:
+            return self.headers.get("location")
+
+        return False
+
+    def release_conn(self):
+        if not self._pool or not self._connection:
+            return
+
+        self._pool._put_conn(self._connection)
+        self._connection = None
+
+    def drain_conn(self):
+        """
+        Read and discard any remaining HTTP response data in the response connection.
+
+        Unread data in the HTTPResponse connection blocks the connection from being released back to the pool.
+        """
+        try:
+            self.read()
+        except (HTTPError, SocketError, BaseSSLError, HTTPException):
+            pass
+
+    @property
+    def data(self):
+        # For backwards-compat with earlier urllib3 0.4 and earlier.
+        if self._body:
+            return self._body
+
+        if self._fp:
+            return self.read(cache_content=True)
+
+    @property
+    def connection(self):
+        return self._connection
+
+    def isclosed(self):
+        return is_fp_closed(self._fp)
+
+    def tell(self):
+        """
+        Obtain the number of bytes pulled over the wire so far. May differ from
+        the amount of content returned by :meth:``urllib3.response.HTTPResponse.read``
+        if bytes are encoded on the wire (e.g, compressed).
+        """
+        return self._fp_bytes_read
+
+    def _init_length(self, request_method):
+        """
+        Set initial length value for Response content if available.
+        """
+        length = self.headers.get("content-length")
+
+        if length is not None:
+            if self.chunked:
+                # This Response will fail with an IncompleteRead if it can't be
+                # received as chunked. This method falls back to attempt reading
+                # the response before raising an exception.
+                log.warning(
+                    "Received response with both Content-Length and "
+                    "Transfer-Encoding set. This is expressly forbidden "
+                    "by RFC 7230 sec 3.3.2. Ignoring Content-Length and "
+                    "attempting to process response as Transfer-Encoding: "
+                    "chunked."
+                )
+                return None
+
+            try:
+                # RFC 7230 section 3.3.2 specifies multiple content lengths can
+                # be sent in a single Content-Length header
+                # (e.g. Content-Length: 42, 42). This line ensures the values
+                # are all valid ints and that as long as the `set` length is 1,
+                # all values are the same. Otherwise, the header is invalid.
+                lengths = set([int(val) for val in length.split(",")])
+                if len(lengths) > 1:
+                    raise InvalidHeader(
+                        "Content-Length contained multiple "
+                        "unmatching values (%s)" % length
+                    )
+                length = lengths.pop()
+            except ValueError:
+                length = None
+            else:
+                if length < 0:
+                    length = None
+
+        # Convert status to int for comparison
+        # In some cases, httplib returns a status of "_UNKNOWN"
+        try:
+            status = int(self.status)
+        except ValueError:
+            status = 0
+
+        # Check for responses that shouldn't include a body
+        if status in (204, 304) or 100 <= status < 200 or request_method == "HEAD":
+            length = 0
+
+        return length
+
+    def _init_decoder(self):
+        """
+        Set-up the _decoder attribute if necessary.
+        """
+        # Note: content-encoding value should be case-insensitive, per RFC 7230
+        # Section 3.2
+        content_encoding = self.headers.get("content-encoding", "").lower()
+        if self._decoder is None:
+            if content_encoding in self.CONTENT_DECODERS:
+                self._decoder = _get_decoder(content_encoding)
+            elif "," in content_encoding:
+                encodings = [
+                    e.strip()
+                    for e in content_encoding.split(",")
+                    if e.strip() in self.CONTENT_DECODERS
+                ]
+                if len(encodings):
+                    self._decoder = _get_decoder(content_encoding)
+
+    DECODER_ERROR_CLASSES = (IOError, zlib.error)
+    if brotli is not None:
+        DECODER_ERROR_CLASSES += (brotli.error,)
+
+    def _decode(self, data, decode_content, flush_decoder):
+        """
+        Decode the data passed in and potentially flush the decoder.
+        """
+        if not decode_content:
+            return data
+
+        try:
+            if self._decoder:
+                data = self._decoder.decompress(data)
+        except self.DECODER_ERROR_CLASSES as e:
+            content_encoding = self.headers.get("content-encoding", "").lower()
+            raise DecodeError(
+                "Received response with content-encoding: %s, but "
+                "failed to decode it." % content_encoding,
+                e,
+            )
+        if flush_decoder:
+            data += self._flush_decoder()
+
+        return data
+
+    def _flush_decoder(self):
+        """
+        Flushes the decoder. Should only be called if the decoder is actually
+        being used.
+        """
+        if self._decoder:
+            buf = self._decoder.decompress(b"")
+            return buf + self._decoder.flush()
+
+        return b""
+
+    @contextmanager
+    def _error_catcher(self):
+        """
+        Catch low-level python exceptions, instead re-raising urllib3
+        variants, so that low-level exceptions are not leaked in the
+        high-level api.
+
+        On exit, release the connection back to the pool.
+        """
+        clean_exit = False
+
+        try:
+            try:
+                yield
+
+            except SocketTimeout:
+                # FIXME: Ideally we'd like to include the url in the ReadTimeoutError but
+                # there is yet no clean way to get at it from this context.
+                raise ReadTimeoutError(self._pool, None, "Read timed out.")
+
+            except BaseSSLError as e:
+                # FIXME: Is there a better way to differentiate between SSLErrors?
+                if "read operation timed out" not in str(e):
+                    # SSL errors related to framing/MAC get wrapped and reraised here
+                    raise SSLError(e)
+
+                raise ReadTimeoutError(self._pool, None, "Read timed out.")
+
+            except (HTTPException, SocketError) as e:
+                # This includes IncompleteRead.
+                raise ProtocolError("Connection broken: %r" % e, e)
+
+            # If no exception is thrown, we should avoid cleaning up
+            # unnecessarily.
+            clean_exit = True
+        finally:
+            # If we didn't terminate cleanly, we need to throw away our
+            # connection.
+            if not clean_exit:
+                # The response may not be closed but we're not going to use it
+                # anymore so close it now to ensure that the connection is
+                # released back to the pool.
+                if self._original_response:
+                    self._original_response.close()
+
+                # Closing the response may not actually be sufficient to close
+                # everything, so if we have a hold of the connection close that
+                # too.
+                if self._connection:
+                    self._connection.close()
+
+            # If we hold the original response but it's closed now, we should
+            # return the connection back to the pool.
+            if self._original_response and self._original_response.isclosed():
+                self.release_conn()
+
+    def _fp_read(self, amt):
+        """
+        Read a response with the thought that reading the number of bytes
+        larger than can fit in a 32-bit int at a time via SSL in some
+        known cases leads to an overflow error that has to be prevented
+        if `amt` or `self.length_remaining` indicate that a problem may
+        happen.
+
+        The known cases:
+          * 3.8 <= CPython < 3.9.7 because of a bug
+            https://github.com/urllib3/urllib3/issues/2513#issuecomment-1152559900.
+          * urllib3 injected with pyOpenSSL-backed SSL-support.
+          * CPython < 3.10 only when `amt` does not fit 32-bit int.
+        """
+        assert self._fp
+        c_int_max = 2 ** 31 - 1
+        if (
+            (
+                (amt and amt > c_int_max)
+                or (self.length_remaining and self.length_remaining > c_int_max)
+            )
+            and not util.IS_SECURETRANSPORT
+            and (util.IS_PYOPENSSL or sys.version_info < (3, 10))
+        ):
+            buffer = io.BytesIO()
+            # Besides `max_chunk_amt` being a maximum chunk size, it
+            # affects memory overhead of reading a response by this
+            # method in CPython.
+            # `c_int_max` equal to 2 GiB - 1 byte is the actual maximum
+            # chunk size that does not lead to an overflow error, but
+            # 256 MiB is a compromise.
+            max_chunk_amt = 2 ** 28
+            while amt is None or amt != 0:
+                if amt is not None:
+                    chunk_amt = min(amt, max_chunk_amt)
+                    amt -= chunk_amt
+                else:
+                    chunk_amt = max_chunk_amt
+                data = self._fp.read(chunk_amt)
+                if not data:
+                    break
+                buffer.write(data)
+                del data  # to reduce peak memory usage by `max_chunk_amt`.
+            return buffer.getvalue()
+        else:
+            # StringIO doesn't like amt=None
+            return self._fp.read(amt) if amt is not None else self._fp.read()
+
+    def read(self, amt=None, decode_content=None, cache_content=False):
+        """
+        Similar to :meth:`http.client.HTTPResponse.read`, but with two additional
+        parameters: ``decode_content`` and ``cache_content``.
+
+        :param amt:
+            How much of the content to read. If specified, caching is skipped
+            because it doesn't make sense to cache partial content as the full
+            response.
+
+        :param decode_content:
+            If True, will attempt to decode the body based on the
+            'content-encoding' header.
+
+        :param cache_content:
+            If True, will save the returned data such that the same result is
+            returned despite of the state of the underlying file object. This
+            is useful if you want the ``.data`` property to continue working
+            after having ``.read()`` the file object. (Overridden if ``amt`` is
+            set.)
+        """
+        self._init_decoder()
+        if decode_content is None:
+            decode_content = self.decode_content
+
+        if self._fp is None:
+            return
+
+        flush_decoder = False
+        fp_closed = getattr(self._fp, "closed", False)
+
+        with self._error_catcher():
+            data = self._fp_read(amt) if not fp_closed else b""
+            if amt is None:
+                flush_decoder = True
+            else:
+                cache_content = False
+                if (
+                    amt != 0 and not data
+                ):  # Platform-specific: Buggy versions of Python.
+                    # Close the connection when no data is returned
+                    #
+                    # This is redundant to what httplib/http.client _should_
+                    # already do.  However, versions of python released before
+                    # December 15, 2012 (http://bugs.python.org/issue16298) do
+                    # not properly close the connection in all cases. There is
+                    # no harm in redundantly calling close.
+                    self._fp.close()
+                    flush_decoder = True
+                    if self.enforce_content_length and self.length_remaining not in (
+                        0,
+                        None,
+                    ):
+                        # This is an edge case that httplib failed to cover due
+                        # to concerns of backward compatibility. We're
+                        # addressing it here to make sure IncompleteRead is
+                        # raised during streaming, so all calls with incorrect
+                        # Content-Length are caught.
+                        raise IncompleteRead(self._fp_bytes_read, self.length_remaining)
+
+        if data:
+            self._fp_bytes_read += len(data)
+            if self.length_remaining is not None:
+                self.length_remaining -= len(data)
+
+            data = self._decode(data, decode_content, flush_decoder)
+
+            if cache_content:
+                self._body = data
+
+        return data
+
+    def stream(self, amt=2 ** 16, decode_content=None):
+        """
+        A generator wrapper for the read() method. A call will block until
+        ``amt`` bytes have been read from the connection or until the
+        connection is closed.
+
+        :param amt:
+            How much of the content to read. The generator will return up to
+            much data per iteration, but may return less. This is particularly
+            likely when using compressed data. However, the empty string will
+            never be returned.
+
+        :param decode_content:
+            If True, will attempt to decode the body based on the
+            'content-encoding' header.
+        """
+        if self.chunked and self.supports_chunked_reads():
+            for line in self.read_chunked(amt, decode_content=decode_content):
+                yield line
+        else:
+            while not is_fp_closed(self._fp):
+                data = self.read(amt=amt, decode_content=decode_content)
+
+                if data:
+                    yield data
+
+    @classmethod
+    def from_httplib(ResponseCls, r, **response_kw):
+        """
+        Given an :class:`http.client.HTTPResponse` instance ``r``, return a
+        corresponding :class:`urllib3.response.HTTPResponse` object.
+
+        Remaining parameters are passed to the HTTPResponse constructor, along
+        with ``original_response=r``.
+        """
+        headers = r.msg
+
+        if not isinstance(headers, HTTPHeaderDict):
+            if six.PY2:
+                # Python 2.7
+                headers = HTTPHeaderDict.from_httplib(headers)
+            else:
+                headers = HTTPHeaderDict(headers.items())
+
+        # HTTPResponse objects in Python 3 don't have a .strict attribute
+        strict = getattr(r, "strict", 0)
+        resp = ResponseCls(
+            body=r,
+            headers=headers,
+            status=r.status,
+            version=r.version,
+            reason=r.reason,
+            strict=strict,
+            original_response=r,
+            **response_kw
+        )
+        return resp
+
+    # Backwards-compatibility methods for http.client.HTTPResponse
+    def getheaders(self):
+        warnings.warn(
+            "HTTPResponse.getheaders() is deprecated and will be removed "
+            "in urllib3 v2.1.0. Instead access HTTPResponse.headers directly.",
+            category=DeprecationWarning,
+            stacklevel=2,
+        )
+        return self.headers
+
+    def getheader(self, name, default=None):
+        warnings.warn(
+            "HTTPResponse.getheader() is deprecated and will be removed "
+            "in urllib3 v2.1.0. Instead use HTTPResponse.headers.get(name, default).",
+            category=DeprecationWarning,
+            stacklevel=2,
+        )
+        return self.headers.get(name, default)
+
+    # Backwards compatibility for http.cookiejar
+    def info(self):
+        return self.headers
+
+    # Overrides from io.IOBase
+    def close(self):
+        if not self.closed:
+            self._fp.close()
+
+        if self._connection:
+            self._connection.close()
+
+        if not self.auto_close:
+            io.IOBase.close(self)
+
+    @property
+    def closed(self):
+        if not self.auto_close:
+            return io.IOBase.closed.__get__(self)
+        elif self._fp is None:
+            return True
+        elif hasattr(self._fp, "isclosed"):
+            return self._fp.isclosed()
+        elif hasattr(self._fp, "closed"):
+            return self._fp.closed
+        else:
+            return True
+
+    def fileno(self):
+        if self._fp is None:
+            raise IOError("HTTPResponse has no file to get a fileno from")
+        elif hasattr(self._fp, "fileno"):
+            return self._fp.fileno()
+        else:
+            raise IOError(
+                "The file-like object this HTTPResponse is wrapped "
+                "around has no file descriptor"
+            )
+
+    def flush(self):
+        if (
+            self._fp is not None
+            and hasattr(self._fp, "flush")
+            and not getattr(self._fp, "closed", False)
+        ):
+            return self._fp.flush()
+
+    def readable(self):
+        # This method is required for `io` module compatibility.
+        return True
+
+    def readinto(self, b):
+        # This method is required for `io` module compatibility.
+        temp = self.read(len(b))
+        if len(temp) == 0:
+            return 0
+        else:
+            b[: len(temp)] = temp
+            return len(temp)
+
+    def supports_chunked_reads(self):
+        """
+        Checks if the underlying file-like object looks like a
+        :class:`http.client.HTTPResponse` object. We do this by testing for
+        the fp attribute. If it is present we assume it returns raw chunks as
+        processed by read_chunked().
+        """
+        return hasattr(self._fp, "fp")
+
+    def _update_chunk_length(self):
+        # First, we'll figure out length of a chunk and then
+        # we'll try to read it from socket.
+        if self.chunk_left is not None:
+            return
+        line = self._fp.fp.readline()
+        line = line.split(b";", 1)[0]
+        try:
+            self.chunk_left = int(line, 16)
+        except ValueError:
+            # Invalid chunked protocol response, abort.
+            self.close()
+            raise InvalidChunkLength(self, line)
+
+    def _handle_chunk(self, amt):
+        returned_chunk = None
+        if amt is None:
+            chunk = self._fp._safe_read(self.chunk_left)
+            returned_chunk = chunk
+            self._fp._safe_read(2)  # Toss the CRLF at the end of the chunk.
+            self.chunk_left = None
+        elif amt < self.chunk_left:
+            value = self._fp._safe_read(amt)
+            self.chunk_left = self.chunk_left - amt
+            returned_chunk = value
+        elif amt == self.chunk_left:
+            value = self._fp._safe_read(amt)
+            self._fp._safe_read(2)  # Toss the CRLF at the end of the chunk.
+            self.chunk_left = None
+            returned_chunk = value
+        else:  # amt > self.chunk_left
+            returned_chunk = self._fp._safe_read(self.chunk_left)
+            self._fp._safe_read(2)  # Toss the CRLF at the end of the chunk.
+            self.chunk_left = None
+        return returned_chunk
+
+    def read_chunked(self, amt=None, decode_content=None):
+        """
+        Similar to :meth:`HTTPResponse.read`, but with an additional
+        parameter: ``decode_content``.
+
+        :param amt:
+            How much of the content to read. If specified, caching is skipped
+            because it doesn't make sense to cache partial content as the full
+            response.
+
+        :param decode_content:
+            If True, will attempt to decode the body based on the
+            'content-encoding' header.
+        """
+        self._init_decoder()
+        # FIXME: Rewrite this method and make it a class with a better structured logic.
+        if not self.chunked:
+            raise ResponseNotChunked(
+                "Response is not chunked. "
+                "Header 'transfer-encoding: chunked' is missing."
+            )
+        if not self.supports_chunked_reads():
+            raise BodyNotHttplibCompatible(
+                "Body should be http.client.HTTPResponse like. "
+                "It should have have an fp attribute which returns raw chunks."
+            )
+
+        with self._error_catcher():
+            # Don't bother reading the body of a HEAD request.
+            if self._original_response and is_response_to_head(self._original_response):
+                self._original_response.close()
+                return
+
+            # If a response is already read and closed
+            # then return immediately.
+            if self._fp.fp is None:
+                return
+
+            while True:
+                self._update_chunk_length()
+                if self.chunk_left == 0:
+                    break
+                chunk = self._handle_chunk(amt)
+                decoded = self._decode(
+                    chunk, decode_content=decode_content, flush_decoder=False
+                )
+                if decoded:
+                    yield decoded
+
+            if decode_content:
+                # On CPython and PyPy, we should never need to flush the
+                # decoder. However, on Jython we *might* need to, so
+                # lets defensively do it anyway.
+                decoded = self._flush_decoder()
+                if decoded:  # Platform-specific: Jython.
+                    yield decoded
+
+            # Chunk content ends with \r\n: discard it.
+            while True:
+                line = self._fp.fp.readline()
+                if not line:
+                    # Some sites may not end with '\r\n'.
+                    break
+                if line == b"\r\n":
+                    break
+
+            # We read everything; close the "file".
+            if self._original_response:
+                self._original_response.close()
+
+    def geturl(self):
+        """
+        Returns the URL that was the source of this response.
+        If the request that generated this response redirected, this method
+        will return the final redirect location.
+        """
+        if self.retries is not None and len(self.retries.history):
+            return self.retries.history[-1].redirect_location
+        else:
+            return self._request_url
+
+    def __iter__(self):
+        buffer = []
+        for chunk in self.stream(decode_content=True):
+            if b"\n" in chunk:
+                chunk = chunk.split(b"\n")
+                yield b"".join(buffer) + chunk[0] + b"\n"
+                for x in chunk[1:-1]:
+                    yield x + b"\n"
+                if chunk[-1]:
+                    buffer = [chunk[-1]]
+                else:
+                    buffer = []
+            else:
+                buffer.append(chunk)
+        if buffer:
+            yield b"".join(buffer)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/util/__init__.py

@@ -0,0 +1,49 @@
+from __future__ import absolute_import
+
+# For backwards compatibility, provide imports that used to be here.
+from .connection import is_connection_dropped
+from .request import SKIP_HEADER, SKIPPABLE_HEADERS, make_headers
+from .response import is_fp_closed
+from .retry import Retry
+from .ssl_ import (
+    ALPN_PROTOCOLS,
+    HAS_SNI,
+    IS_PYOPENSSL,
+    IS_SECURETRANSPORT,
+    PROTOCOL_TLS,
+    SSLContext,
+    assert_fingerprint,
+    resolve_cert_reqs,
+    resolve_ssl_version,
+    ssl_wrap_socket,
+)
+from .timeout import Timeout, current_time
+from .url import Url, get_host, parse_url, split_first
+from .wait import wait_for_read, wait_for_write
+
+__all__ = (
+    "HAS_SNI",
+    "IS_PYOPENSSL",
+    "IS_SECURETRANSPORT",
+    "SSLContext",
+    "PROTOCOL_TLS",
+    "ALPN_PROTOCOLS",
+    "Retry",
+    "Timeout",
+    "Url",
+    "assert_fingerprint",
+    "current_time",
+    "is_connection_dropped",
+    "is_fp_closed",
+    "get_host",
+    "parse_url",
+    "make_headers",
+    "resolve_cert_reqs",
+    "resolve_ssl_version",
+    "split_first",
+    "ssl_wrap_socket",
+    "wait_for_read",
+    "wait_for_write",
+    "SKIP_HEADER",
+    "SKIPPABLE_HEADERS",
+)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/util/connection.py

@@ -0,0 +1,149 @@
+from __future__ import absolute_import
+
+import socket
+
+from ..contrib import _appengine_environ
+from ..exceptions import LocationParseError
+from ..packages import six
+from .wait import NoWayToWaitForSocketError, wait_for_read
+
+
+def is_connection_dropped(conn):  # Platform-specific
+    """
+    Returns True if the connection is dropped and should be closed.
+
+    :param conn:
+        :class:`http.client.HTTPConnection` object.
+
+    Note: For platforms like AppEngine, this will always return ``False`` to
+    let the platform handle connection recycling transparently for us.
+    """
+    sock = getattr(conn, "sock", False)
+    if sock is False:  # Platform-specific: AppEngine
+        return False
+    if sock is None:  # Connection already closed (such as by httplib).
+        return True
+    try:
+        # Returns True if readable, which here means it's been dropped
+        return wait_for_read(sock, timeout=0.0)
+    except NoWayToWaitForSocketError:  # Platform-specific: AppEngine
+        return False
+
+
+# This function is copied from socket.py in the Python 2.7 standard
+# library test suite. Added to its signature is only `socket_options`.
+# One additional modification is that we avoid binding to IPv6 servers
+# discovered in DNS if the system doesn't have IPv6 functionality.
+def create_connection(
+    address,
+    timeout=socket._GLOBAL_DEFAULT_TIMEOUT,
+    source_address=None,
+    socket_options=None,
+):
+    """Connect to *address* and return the socket object.
+
+    Convenience function.  Connect to *address* (a 2-tuple ``(host,
+    port)``) and return the socket object.  Passing the optional
+    *timeout* parameter will set the timeout on the socket instance
+    before attempting to connect.  If no *timeout* is supplied, the
+    global default timeout setting returned by :func:`socket.getdefaulttimeout`
+    is used.  If *source_address* is set it must be a tuple of (host, port)
+    for the socket to bind as a source address before making the connection.
+    An host of '' or port 0 tells the OS to use the default.
+    """
+
+    host, port = address
+    if host.startswith("["):
+        host = host.strip("[]")
+    err = None
+
+    # Using the value from allowed_gai_family() in the context of getaddrinfo lets
+    # us select whether to work with IPv4 DNS records, IPv6 records, or both.
+    # The original create_connection function always returns all records.
+    family = allowed_gai_family()
+
+    try:
+        host.encode("idna")
+    except UnicodeError:
+        return six.raise_from(
+            LocationParseError(u"'%s', label empty or too long" % host), None
+        )
+
+    for res in socket.getaddrinfo(host, port, family, socket.SOCK_STREAM):
+        af, socktype, proto, canonname, sa = res
+        sock = None
+        try:
+            sock = socket.socket(af, socktype, proto)
+
+            # If provided, set socket level options before connecting.
+            _set_socket_options(sock, socket_options)
+
+            if timeout is not socket._GLOBAL_DEFAULT_TIMEOUT:
+                sock.settimeout(timeout)
+            if source_address:
+                sock.bind(source_address)
+            sock.connect(sa)
+            return sock
+
+        except socket.error as e:
+            err = e
+            if sock is not None:
+                sock.close()
+                sock = None
+
+    if err is not None:
+        raise err
+
+    raise socket.error("getaddrinfo returns an empty list")
+
+
+def _set_socket_options(sock, options):
+    if options is None:
+        return
+
+    for opt in options:
+        sock.setsockopt(*opt)
+
+
+def allowed_gai_family():
+    """This function is designed to work in the context of
+    getaddrinfo, where family=socket.AF_UNSPEC is the default and
+    will perform a DNS search for both IPv6 and IPv4 records."""
+
+    family = socket.AF_INET
+    if HAS_IPV6:
+        family = socket.AF_UNSPEC
+    return family
+
+
+def _has_ipv6(host):
+    """Returns True if the system can bind an IPv6 address."""
+    sock = None
+    has_ipv6 = False
+
+    # App Engine doesn't support IPV6 sockets and actually has a quota on the
+    # number of sockets that can be used, so just early out here instead of
+    # creating a socket needlessly.
+    # See https://github.com/urllib3/urllib3/issues/1446
+    if _appengine_environ.is_appengine_sandbox():
+        return False
+
+    if socket.has_ipv6:
+        # has_ipv6 returns true if cPython was compiled with IPv6 support.
+        # It does not tell us if the system has IPv6 support enabled. To
+        # determine that we must bind to an IPv6 address.
+        # https://github.com/urllib3/urllib3/pull/611
+        # https://bugs.python.org/issue658327
+        try:
+            sock = socket.socket(socket.AF_INET6)
+            sock.bind((host, 0))
+            has_ipv6 = True
+        except Exception:
+            pass
+
+    if sock:
+        sock.close()
+    return has_ipv6
+
+
+HAS_IPV6 = _has_ipv6("::1")

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35/lib/python3.12/site-packages/pip/_vendor/urllib3/util/proxy.py

@@ -0,0 +1,57 @@
+from .ssl_ import create_urllib3_context, resolve_cert_reqs, resolve_ssl_version
+
+
+def connection_requires_http_tunnel(
+    proxy_url=None, proxy_config=None, destination_scheme=None
+):
+    """
+    Returns True if the connection requires an HTTP CONNECT through the proxy.
+
+    :param URL proxy_url:
+        URL of the proxy.
+    :param ProxyConfig proxy_config:
+        Proxy configuration from poolmanager.py
+    :param str destination_scheme:
+        The scheme of the destination. (i.e https, http, etc)
+    """
+    # If we're not using a proxy, no way to use a tunnel.
+    if proxy_url is None:
+        return False
+
+    # HTTP destinations never require tunneling, we always forward.
+    if destination_scheme == "http":
+        return False
+
+    # Support for forwarding with HTTPS proxies and HTTPS destinations.
+    if (
+        proxy_url.scheme == "https"
+        and proxy_config
+        and proxy_config.use_forwarding_for_https
+    ):
+        return False
+
+    # Otherwise always use a tunnel.
+    return True
+
+
+def create_proxy_ssl_context(
+    ssl_version, cert_reqs, ca_certs=None, ca_cert_dir=None, ca_cert_data=None
+):
+    """
+    Generates a default proxy ssl context if one hasn't been provided by the
+    user.
+    """
+    ssl_context = create_urllib3_context(
+        ssl_version=resolve_ssl_version(ssl_version),
+        cert_reqs=resolve_cert_reqs(cert_reqs),
+    )
+
+    if (
+        not ca_certs
+        and not ca_cert_dir
+        and not ca_cert_data
+        and hasattr(ssl_context, "load_default_certs")
+    ):
+        ssl_context.load_default_certs()
+
+    return ssl_context