Spaces:
Running on Zero
Running on Zero
Create hf_blender_extract.py
Browse files
src/data/hf_blender_extract.py
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# src/data/hf_blender_extract.py
|
| 2 |
+
import os
|
| 3 |
+
import sys
|
| 4 |
+
import time
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
|
| 7 |
+
def _parse(argv):
|
| 8 |
+
# argv comes after `--`
|
| 9 |
+
args = {"input": None, "output_dir": None, "target_count": 50000}
|
| 10 |
+
it = iter(argv)
|
| 11 |
+
for k in it:
|
| 12 |
+
if k == "--input":
|
| 13 |
+
args["input"] = next(it)
|
| 14 |
+
elif k == "--output_dir":
|
| 15 |
+
args["output_dir"] = next(it)
|
| 16 |
+
elif k == "--target_count":
|
| 17 |
+
args["target_count"] = int(next(it))
|
| 18 |
+
if not args["input"] or not args["output_dir"]:
|
| 19 |
+
raise SystemExit("Usage: --input <file> --output_dir <dir> [--target_count N]")
|
| 20 |
+
return args
|
| 21 |
+
|
| 22 |
+
def main():
|
| 23 |
+
argv = sys.argv
|
| 24 |
+
if "--" in argv:
|
| 25 |
+
argv = argv[argv.index("--") + 1 :]
|
| 26 |
+
else:
|
| 27 |
+
argv = []
|
| 28 |
+
args = _parse(argv)
|
| 29 |
+
|
| 30 |
+
# Ensure output dir
|
| 31 |
+
out = Path(args["output_dir"])
|
| 32 |
+
out.mkdir(parents=True, exist_ok=True)
|
| 33 |
+
|
| 34 |
+
from src.data.extract import extract_builtin, get_files
|
| 35 |
+
|
| 36 |
+
files = get_files(
|
| 37 |
+
data_name="raw_data.npz",
|
| 38 |
+
inputs=str(args["input"]),
|
| 39 |
+
input_dataset_dir=None,
|
| 40 |
+
output_dataset_dir=str(out),
|
| 41 |
+
force_override=True,
|
| 42 |
+
warning=False,
|
| 43 |
+
)
|
| 44 |
+
if not files:
|
| 45 |
+
raise RuntimeError("No files to extract")
|
| 46 |
+
|
| 47 |
+
timestamp = str(int(time.time()))
|
| 48 |
+
extract_builtin(
|
| 49 |
+
output_folder=str(out),
|
| 50 |
+
target_count=int(args["target_count"]),
|
| 51 |
+
num_runs=1,
|
| 52 |
+
id=0,
|
| 53 |
+
time=timestamp,
|
| 54 |
+
files=files,
|
| 55 |
+
)
|
| 56 |
+
|
| 57 |
+
if __name__ == "__main__":
|
| 58 |
+
main()
|