scripts/delete-hf.py

#!/usr/bin/env python3
"""Hugging Face 资源删除工具 — 支持删除 Space / Dataset / Model / Storage / 文件。

用法:
    # 删除整个仓库
    python scripts/delete-hf.py repo <repo_id> --type space
    python scripts/delete-hf.py repo <repo_id> --type dataset
    python scripts/delete-hf.py repo <repo_id> --type model

    # 删除仓库中的文件（支持通配符）
    python scripts/delete-hf.py files <repo_id> --pattern "backups/*"

    # 删除 Space 持久存储
    python scripts/delete-hf.py storage <space_id>

    # 交互确认 + token 参数
    python scripts/delete-hf.py repo my-space --type space --yes
    python scripts/delete-hf.py files my-backup --pattern "*.old" --token hf_xxx
"""

import argparse
import sys
import time

from huggingface_hub import HfApi

_BATCH_SIZE = 100


def _confirm(prompt: str, auto_yes: bool) -> bool:
    if auto_yes:
        return True
    ans = input(f"{prompt} (yes/NO): ").strip().lower()
    return ans in ("yes", "y")


def cmd_delete_repo(api: HfApi, args: argparse.Namespace) -> None:
    """删除整个仓库（Space / Dataset / Model）。"""
    repo_id = args.repo_id
    repo_type = args.type

    print(f"待删除仓库: {repo_id} (type={repo_type})")
    print("  此操作不可恢复！")

    if not _confirm(f"确认删除仓库 '{repo_id}'？", args.yes):
        print("[INFO] 已取消")
        return

    try:
        api.delete_repo(repo_id=repo_id, repo_type=repo_type)
        print(f"✓ 仓库 '{repo_id}' 已删除")
    except Exception as e:
        print(f"✗ 删除失败: {e}", file=sys.stderr)
        sys.exit(1)


def cmd_delete_files(api: HfApi, args: argparse.Namespace) -> None:
    """根据通配符模式批量删除仓库中的文件。"""
    repo_id = args.repo_id
    patterns = args.patterns
    repo_type = args.type

    # 先列出匹配的文件，供用户预览
    try:
        all_files = api.list_repo_files(repo_id=repo_id, repo_type=repo_type)
    except Exception as e:
        print(f"[ERROR] 无法列出文件: {e}", file=sys.stderr)
        sys.exit(1)

    import fnmatch
    matched = []
    for f in all_files:
        for pat in patterns:
            if fnmatch.fnmatch(f, pat):
                matched.append(f)
                break

    if not matched:
        print("[INFO] 没有匹配的文件")
        return

    print(f"匹配到 {len(matched)} 个文件:")
    for f in sorted(matched):
        print(f"  - {f}")

    if not _confirm(f"确认删除这 {len(matched)} 个文件？", args.yes):
        print("[INFO] 已取消")
        return

    # 按通配符模式分批删除
    total = len(patterns)
    total_batches = (total + _BATCH_SIZE - 1) // _BATCH_SIZE
    deleted_total = 0

    for batch_idx in range(total_batches):
        start = batch_idx * _BATCH_SIZE
        end = min(start + _BATCH_SIZE, total)
        batch = patterns[start:end]

        try:
            api.delete_files(
                repo_id=repo_id,
                repo_type=repo_type,
                delete_patterns=batch,
                commit_message=(
                    f"delete-hf: batch {batch_idx + 1}/{total_batches} "
                    f"({len(batch)} patterns)"
                ),
            )
            deleted_total += len(batch)
            print(f"  ✓ 第 {batch_idx + 1}/{total_batches} 批: "
                  f"删除 {len(batch)} 个通配符模式")
            if batch_idx + 1 < total_batches:
                time.sleep(1)
        except Exception as e:
            print(f"  ✗ 第 {batch_idx + 1}/{total_batches} 批失败: {e}",
                  file=sys.stderr)
            print(f"    已删除 {deleted_total}/{total} 个模式后中断")
            sys.exit(1)

    print(f"✓ 所有匹配文件已删除（{len(matched)} 个文件，{total} 个模式）")


def cmd_delete_storage(api: HfApi, args: argparse.Namespace) -> None:
    """删除 Space 的持久存储。"""
    space_id = args.space_id

    print(f"待删除存储的 Space: {space_id}")
    print("  此操作不可恢复！所有持久存储数据将被清除。")

    if not _confirm(f"确认删除 '{space_id}' 的持久存储？", args.yes):
        print("[INFO] 已取消")
        return

    try:
        result = api.delete_space_storage(repo_id=space_id)
        print(f"✓ Space '{space_id}' 的持久存储已删除")
        if result:
            print(f"  当前运行时状态: {result}")
    except Exception as e:
        print(f"✗ 删除失败: {e}", file=sys.stderr)
        if "no persistent storage" in str(e).lower():
            print("  [HINT] 该 Space 可能未启用持久存储", file=sys.stderr)
        sys.exit(1)


def main():
    parser = argparse.ArgumentParser(
        description="Hugging Face 资源删除工具"
    )
    parser.add_argument("--token", default=None, help="HF API token")
    parser.add_argument("--yes", "-y", action="store_true",
                        help="跳过确认提示")

    sub = parser.add_subparsers(dest="command", required=True,
                                 help="子命令")

    # --- repo ---
    p_repo = sub.add_parser("repo", help="删除整个仓库")
    p_repo.add_argument("repo_id", help="仓库 ID，如 user/space-name")
    p_repo.add_argument("--type", dest="type", required=True,
                        choices=["space", "dataset", "model"],
                        help="仓库类型")
    p_repo.set_defaults(func=cmd_delete_repo)

    # --- files ---
    p_files = sub.add_parser("files", help="批量删除仓库中的文件")
    p_files.add_argument("repo_id", help="仓库 ID")
    p_files.add_argument("--pattern", dest="patterns",
                         required=True, action="append",
                         help="通配符模式（可多次使用）")
    p_files.add_argument("--type", dest="type", default="dataset",
                         choices=["dataset", "model", "space"],
                         help="仓库类型（默认: dataset）")
    p_files.set_defaults(func=cmd_delete_files)

    # --- storage ---
    p_storage = sub.add_parser("storage", help="删除 Space 持久存储")
    p_storage.add_argument("space_id", help="Space ID，如 user/space-name")
    p_storage.set_defaults(func=cmd_delete_storage)

    args = parser.parse_args()

    # 创建 API 实例
    api = HfApi(token=args.token)

    # 执行对应子命令
    args.func(api, args)


if __name__ == "__main__":
    main()