Upload hf_backend/filename_utils.py with huggingface_hub

hf_backend/filename_utils.py

@@ -0,0 +1,52 @@
+from __future__ import annotations
+
+import re
+from pathlib import PurePosixPath
+from urllib.parse import unquote
+
+
+def normalize_source_filename(
+    name: str,
+    *,
+    default_stem: str = "downloaded_book",
+    default_extension: str = "",
+) -> str:
+    raw_name = str(name or "").strip()
+    decoded_name = unquote(raw_name)
+    if not decoded_name:
+        return _compose_filename(default_stem, default_extension)
+
+    basename = PurePosixPath(decoded_name).name or decoded_name
+    path = PurePosixPath(basename)
+    extension = path.suffix.lower() or _normalize_extension(default_extension)
+    stem = path.name[: -len(path.suffix)] if path.suffix else path.name
+
+    match = re.match(r"^(.+?)\s+--\s+", stem)
+    if match:
+        stem = match.group(1)
+
+    stem = stem.strip()
+    stem = re.sub(r"\s+", "_", stem)
+    stem = re.sub(r"[^a-zA-Z0-9_]", "_", stem)
+    stem = re.sub(r"_+", "_", stem)
+    stem = stem.strip("_")
+    if not stem:
+        stem = default_stem
+
+    return _compose_filename(stem, extension)
+
+
+def _normalize_extension(value: str) -> str:
+    extension = str(value or "").strip().lower()
+    if not extension:
+        return ""
+    if not extension.startswith("."):
+        extension = f".{extension}"
+    return extension
+
+
+def _compose_filename(stem: str, extension: str) -> str:
+    clean_stem = re.sub(r"[^a-zA-Z0-9_]", "_", str(stem or "").strip())
+    clean_stem = re.sub(r"_+", "_", clean_stem).strip("_") or "downloaded_book"
+    clean_extension = _normalize_extension(extension)
+    return f"{clean_stem}{clean_extension}"