pptList/scripts/clean_blank_template_assets.py

from __future__ import annotations

import json
import shutil
from pathlib import Path

from extract_music_template_assets import make_pptx


ROOT = Path(__file__).resolve().parents[1]
ASSET_ROOT = (ROOT / "public" / "ai-courseware-templates" / "extracted").resolve()
MANIFEST = ASSET_ROOT / "manifest.json"

# 视觉复核后的拒绝清单：纯底色或装饰覆盖率过低，缩略图中等同空白。
REJECTED = {
    "dark-lute": {2, 3, 5, 6, 7, 8},
    "spring-outing": {7, 8, 9},
    "ink-jiangnan": {1, 7, 8, 9, 10},
    "rhythm-train": {2, 3, 5},
    "sunset-drum": {10},
    "ballroom-silhouette": {1},
    "color-theory": {1},
}

# 清理后仅剩封面，没有内容页能力，整套下架。
DROPPED_TEMPLATES = {"floral-jasmine"}


def checked_template_dir(template_id: str) -> Path:
    target = (ASSET_ROOT / template_id).resolve()
    if target.parent != ASSET_ROOT:
        raise RuntimeError(f"拒绝处理资产目录之外的路径：{target}")
    return target


def main() -> None:
    manifest = json.loads(MANIFEST.read_text(encoding="utf-8"))
    kept_items: list[dict[str, object]] = []
    removed_files: list[str] = []

    for item in manifest["items"]:
        template_id = str(item["id"])
        target_dir = checked_template_dir(template_id)
        if template_id in DROPPED_TEMPLATES:
            if target_dir.exists():
                shutil.rmtree(target_dir)
                removed_files.append(f"{template_id}/（整套）")
            continue

        rejected = REJECTED.get(template_id, set())
        old_backgrounds = list(item.get("backgrounds", []))
        old_sources = list(item.get("roleSourceSlides", []))
        backgrounds: list[str] = []
        sources: list[int] = []
        for index, background in enumerate(old_backgrounds):
            path = target_dir / Path(str(background)).name
            asset_number = int(path.stem.rsplit("-", 1)[-1])
            if asset_number in rejected:
                if path.exists():
                    path.unlink()
                    removed_files.append(f"{template_id}/{path.name}")
                continue
            if not path.exists():
                raise FileNotFoundError(f"有效模板背景缺失：{path}")
            backgrounds.append(str(background))
            if index < len(old_sources):
                sources.append(old_sources[index])

        if not backgrounds:
            raise RuntimeError(f"模板 {template_id} 清理后没有可用背景")
        template_file = target_dir / f"{template_id}.pptx"
        make_pptx([target_dir / Path(path).name for path in backgrounds], template_file)
        item["backgrounds"] = backgrounds
        item["roleSourceSlides"] = sources
        item["usableBackgroundCount"] = len(backgrounds)
        kept_items.append(item)

    MANIFEST.write_text(json.dumps({"items": kept_items}, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps({"removed": removed_files, "templates": len(kept_items)}, ensure_ascii=False, indent=2))


if __name__ == "__main__":
    main()