| from urllib.parse import urlparse, parse_qs, unquote | |
| def parse_link(link: str) -> dict: | |
| """ | |
| Parse a Hugging Face URL or shorthand string. | |
| Supports URLs with keywords "resolve", "blob", or "tree". | |
| Returns a dictionary with keys: | |
| - repo: e.g., "username/repo" | |
| - revision: if present (e.g., "main") | |
| - subfolder: if present (the path inside the repo) | |
| - file: if present (the file name for file downloads) | |
| """ | |
| parsed_url = urlparse(link) | |
| if parsed_url.scheme: | |
| path_parts = [unquote(p) for p in parsed_url.path.strip("/").split("/")] | |
| else: | |
| path_parts = [unquote(p) for p in link.strip("/").split("/")] | |
| result = {} | |
| if len(path_parts) >= 2: | |
| result["repo"] = f"{path_parts[0]}/{path_parts[1]}" | |
| else: | |
| raise ValueError("Link does not contain repository information.") | |
| if "resolve" in path_parts: | |
| idx = path_parts.index("resolve") | |
| if len(path_parts) > idx + 1: | |
| result["revision"] = path_parts[idx+1] | |
| if len(path_parts) > idx + 2: | |
| remaining = path_parts[idx+2:] | |
| if remaining: | |
| if len(remaining) > 1: | |
| result["subfolder"] = "/".join(remaining[:-1]) | |
| result["file"] = remaining[-1] | |
| elif "blob" in path_parts: | |
| idx = path_parts.index("blob") | |
| if len(path_parts) > idx + 1: | |
| result["revision"] = path_parts[idx+1] | |
| if len(path_parts) > idx + 2: | |
| remaining = path_parts[idx+2:] | |
| if remaining: | |
| if len(remaining) > 1: | |
| result["subfolder"] = "/".join(remaining[:-1]) | |
| result["file"] = remaining[-1] | |
| elif "tree" in path_parts: | |
| idx = path_parts.index("tree") | |
| if len(path_parts) > idx + 1: | |
| result["revision"] = path_parts[idx+1] | |
| if len(path_parts) > idx + 2: | |
| result["subfolder"] = "/".join(path_parts[idx+2:]) | |
| else: | |
| if len(path_parts) > 2: | |
| if "." in path_parts[-1]: | |
| result["subfolder"] = "/".join(path_parts[2:-1]) | |
| result["file"] = path_parts[-1] | |
| else: | |
| result["subfolder"] = "/".join(path_parts[2:]) | |
| return result | |