Upload demo/upload_used_bundle_to_hf.py with huggingface_hub

demo/upload_used_bundle_to_hf.py

@@ -0,0 +1,173 @@
+from __future__ import annotations
+
+import argparse
+import os
+from pathlib import Path
+
+from huggingface_hub import HfApi, snapshot_download
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+
+DEFAULT_BBOX_REPO_ID = "SynLayers/Bbox-caption-8b"
+DEFAULT_STAGE2_REPO_ID = "SynLayers/synlayers"
+DEFAULT_STAGE2_SOURCE_REPO_ID = DEFAULT_BBOX_REPO_ID
+
+BBOX_FILE_MAP = {
+    PROJECT_ROOT / "demo" / "model_card.md": "README.md",
+}
+
+STAGE2_FILE_MAP = {
+    PROJECT_ROOT / "demo" / "stage2_model_card.md": "README.md",
+    PROJECT_ROOT / "demo" / "__init__.py": "demo/__init__.py",
+    PROJECT_ROOT / "demo" / "app.py": "demo/app.py",
+    PROJECT_ROOT / "demo" / "hf_repo_assets.py": "demo/hf_repo_assets.py",
+    PROJECT_ROOT / "demo" / "README.md": "demo/README.md",
+    PROJECT_ROOT / "demo" / "publish_space.py": "demo/publish_space.py",
+    PROJECT_ROOT / "demo" / "real_world_pipeline.py": "demo/real_world_pipeline.py",
+    PROJECT_ROOT / "demo" / "requirements-hf-space.txt": "demo/requirements-hf-space.txt",
+    PROJECT_ROOT / "demo" / "upload_used_bundle_to_hf.py": "demo/upload_used_bundle_to_hf.py",
+    PROJECT_ROOT / "demo" / "infer" / "__init__.py": "demo/infer/__init__.py",
+    PROJECT_ROOT / "demo" / "infer" / "run_caption_bbox_infer.py": "demo/infer/run_caption_bbox_infer.py",
+    PROJECT_ROOT / "demo" / "infer" / "vlm_bbox_inference.py": "demo/infer/vlm_bbox_inference.py",
+    PROJECT_ROOT / "infer" / "__init__.py": "infer/__init__.py",
+    PROJECT_ROOT / "infer" / "common_infer.py": "infer/common_infer.py",
+    PROJECT_ROOT / "infer" / "infer.py": "infer/infer.py",
+    PROJECT_ROOT / "infer" / "infer.yaml": "infer/infer.yaml",
+    PROJECT_ROOT / "models" / "__init__.py": "models/__init__.py",
+    PROJECT_ROOT / "models" / "multiLayer_adapter.py": "models/multiLayer_adapter.py",
+    PROJECT_ROOT / "models" / "mmdit.py": "models/mmdit.py",
+    PROJECT_ROOT / "models" / "pipeline.py": "models/pipeline.py",
+    PROJECT_ROOT / "models" / "transp_vae.py": "models/transp_vae.py",
+    PROJECT_ROOT / "tools" / "__init__.py": "tools/__init__.py",
+    PROJECT_ROOT / "tools" / "tools.py": "tools/tools.py",
+    PROJECT_ROOT / "dataset_scaleup" / "dataset_construction.sh": "dataset/dataset_construction.sh",
+    PROJECT_ROOT / "dataset_scaleup" / "scaleup_api.py": "dataset/scaleup_api.py",
+    PROJECT_ROOT / "dataset_scaleup" / "scaleup_dataset.py": "dataset/scaleup_dataset.py",
+    PROJECT_ROOT / "dataset_scaleup" / "scaleup_utils.py": "dataset/scaleup_utils.py",
+    PROJECT_ROOT / "environment.yml": "environment.yml",
+}
+
+STAGE2_SOURCE_ASSET_FILES = {
+    "ckpt/trans_vae/0008000.pt": "ckpt/trans_vae/0008000.pt",
+    "ckpt/pre_trained_LoRA/pytorch_lora_weights.safetensors": "ckpt/pre_trained_LoRA/pytorch_lora_weights.safetensors",
+    "ckpt/prism_ft_LoRA/pytorch_lora_weights.safetensors": "ckpt/prism_ft_LoRA/pytorch_lora_weights.safetensors",
+}
+
+STAGE2_SOURCE_ASSET_FOLDERS = {
+    "SynLayers_checkpoints/FLUX.1-dev": "SynLayers_checkpoints/FLUX.1-dev",
+    "SynLayers_ckpt/step_120000": "SynLayers_ckpt/step_120000",
+    "SynLayers_checkpoints/FLUX.1-dev-Controlnet-Inpainting-Alpha": "SynLayers_checkpoints/FLUX.1-dev-Controlnet-Inpainting-Alpha",
+}
+
+
+def upload_file_map(api: HfApi, repo_id: str, repo_type: str, file_map: dict[Path, str]):
+    for local_path, remote_path in file_map.items():
+        if not local_path.exists():
+            print(f"Skipping missing file: {local_path}")
+            continue
+        print(f"Uploading file {local_path} -> {remote_path}")
+        api.upload_file(
+            path_or_fileobj=str(local_path),
+            path_in_repo=remote_path,
+            repo_id=repo_id,
+            repo_type=repo_type,
+        )
+
+
+def download_stage2_source_bundle(source_repo_id: str, token: str | None) -> Path:
+    allow_patterns = list(STAGE2_SOURCE_ASSET_FILES.keys()) + [
+        f"{remote_path}/**" for remote_path in STAGE2_SOURCE_ASSET_FOLDERS
+    ]
+    local_root = snapshot_download(
+        repo_id=source_repo_id,
+        repo_type="model",
+        allow_patterns=allow_patterns,
+        token=token,
+    )
+    return Path(local_root)
+
+
+def upload_stage2_asset_files(
+    api: HfApi,
+    repo_id: str,
+    repo_type: str,
+    source_root: Path,
+):
+    for local_rel_path, remote_path in STAGE2_SOURCE_ASSET_FILES.items():
+        local_path = source_root / local_rel_path
+        if not local_path.exists():
+            print(f"Skipping missing asset: {local_path}")
+            continue
+        print(f"Uploading asset {local_path} -> {remote_path}")
+        api.upload_file(
+            path_or_fileobj=str(local_path),
+            path_in_repo=remote_path,
+            repo_id=repo_id,
+            repo_type=repo_type,
+        )
+
+
+def upload_stage2_asset_folders(
+    api: HfApi,
+    repo_id: str,
+    repo_type: str,
+    source_root: Path,
+):
+    for local_rel_path, remote_path in STAGE2_SOURCE_ASSET_FOLDERS.items():
+        local_path = source_root / local_rel_path
+        if not local_path.exists():
+            print(f"Skipping missing folder: {local_path}")
+            continue
+        print(f"Uploading folder {local_path} -> {remote_path}")
+        api.upload_folder(
+            folder_path=str(local_path),
+            path_in_repo=remote_path,
+            repo_id=repo_id,
+            repo_type=repo_type,
+        )
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Upload either the Stage 1 bbox repo card or the Stage 2 SynLayers bundle."
+    )
+    parser.add_argument("--bundle", type=str, default="bbox", choices=["bbox", "stage2"])
+    parser.add_argument("--repo-id", type=str, default=None)
+    parser.add_argument("--repo-type", type=str, default="model", choices=["model"])
+    parser.add_argument(
+        "--source-repo-id",
+        type=str,
+        default=os.environ.get("SYNLAYERS_STAGE2_SOURCE_REPO", DEFAULT_STAGE2_SOURCE_REPO_ID),
+        help="Source model repo that currently stores the Stage 2 SynLayers assets.",
+    )
+    parser.add_argument("--token", type=str, default=os.environ.get("HF_TOKEN"))
+    args = parser.parse_args()
+
+    if not args.token:
+        raise ValueError("Missing Hugging Face token. Pass --token or set HF_TOKEN.")
+
+    api = HfApi(token=args.token)
+    repo_id = args.repo_id or (
+        DEFAULT_BBOX_REPO_ID if args.bundle == "bbox" else DEFAULT_STAGE2_REPO_ID
+    )
+    api.create_repo(repo_id=repo_id, repo_type=args.repo_type, exist_ok=True)
+
+    if args.bundle == "bbox":
+        upload_file_map(api, repo_id, args.repo_type, BBOX_FILE_MAP)
+    else:
+        source_root = download_stage2_source_bundle(args.source_repo_id, args.token)
+        upload_file_map(api, repo_id, args.repo_type, STAGE2_FILE_MAP)
+        upload_stage2_asset_files(api, repo_id, args.repo_type, source_root)
+        upload_stage2_asset_folders(
+            api,
+            repo_id,
+            args.repo_type,
+            source_root,
+        )
+
+    print("")
+    print(f"Finished uploading the {args.bundle} bundle to https://huggingface.co/{repo_id}")
+
+
+if __name__ == "__main__":
+    main()