Add files using upload-large-folder tool

upload_huggingface.py

@@ -13,10 +13,12 @@ or fall back to an interactive login prompt from huggingface_hub.
 """
 
 import argparse
+import csv
 import os
 import sys
+from pathlib import Path
 
-from huggingface_hub import login, upload_folder
+from huggingface_hub import HfApi, login, upload_folder
 
 
 def parse_args() -> argparse.Namespace:
@@ -42,6 +44,21 @@ def parse_args() -> argparse.Namespace:
         default=None,
         help="Hugging Face token. If omitted, uses HF_TOKEN env or interactive login.",
     )
+    parser.add_argument(
+        "--labels",
+        default="label.txt",
+        help="Path to label file (TSV: image_path<TAB>label). Default: label.txt",
+    )
+    parser.add_argument(
+        "--metadata",
+        default="metadata.csv",
+        help="Output metadata CSV (file_name,text) for imagefolder. Default: metadata.csv",
+    )
+    parser.add_argument(
+        "--use-large-upload",
+        action="store_true",
+        help="Use upload_large_folder to handle big datasets (recommended for large folders).",
+    )
     return parser.parse_args()
 
 
@@ -55,12 +72,49 @@ def main() -> None:
     else:
         login()
 
-    upload_folder(
-        folder_path=args.folder,
-        repo_id=args.repo_id,
-        repo_type=args.repo_type,
-    )
-    print(f"Uploaded '{args.folder}' to {args.repo_type} repo '{args.repo_id}'.")
+    # Build metadata.csv from label.txt so Hugging Face uses explicit labels
+    labels_path = Path(args.labels)
+    metadata_path = Path(args.metadata)
+    if labels_path.exists():
+        base = Path(args.folder).resolve()
+        with labels_path.open("r", encoding="utf-8") as fin, metadata_path.open(
+            "w", encoding="utf-8", newline=""
+        ) as fout:
+            writer = csv.writer(fout)
+            writer.writerow(["file_name", "text"])
+            for line in fin:
+                line = line.rstrip("\n")
+                if not line:
+                    continue
+                try:
+                    rel_path, text = line.split("\t", 1)
+                except ValueError:
+                    print(f"Skipping malformed label line: {line}")
+                    continue
+                # Ensure path relative to the uploaded folder
+                rel_path = str(Path(rel_path))
+                writer.writerow([rel_path, text])
+        print(f"Wrote metadata to {metadata_path} using labels from {labels_path}")
+    else:
+        print(f"Label file not found at {labels_path}; skipping metadata build.")
+
+    if args.use_large_upload:
+        api = HfApi()
+        api.upload_large_folder(
+            folder_path=args.folder,
+            repo_id=args.repo_id,
+            repo_type=args.repo_type,
+        )
+        print(
+            f"Uploaded '{args.folder}' via upload_large_folder to {args.repo_type} repo '{args.repo_id}'."
+        )
+    else:
+        upload_folder(
+            folder_path=args.folder,
+            repo_id=args.repo_id,
+            repo_type=args.repo_type,
+        )
+        print(f"Uploaded '{args.folder}' to {args.repo_type} repo '{args.repo_id}'.")
 
 
 if __name__ == "__main__":