weflow-cli 1.5.0

package/scripts/classify_daily.py

@@ -0,0 +1,411 @@
+#!/usr/bin/env python3
+"""
+后处理：广告清洗 → 兴趣主题深度摘要 → 分文件夹 → 重建README。
+
+前置: 已运行 biz_daily.py，md 文件中已含【主题】标签。
+
+用法: python scripts/classify_daily.py [date_dir] --api-key <key> [--interest AI]
+"""
+
+import sys, os, json, re, time, urllib.request, shutil
+from collections import Counter
+from pathlib import Path
+
+# 公共工具
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+from _utils import (
+    call_deepseek, parse_frontmatter, write_with_frontmatter,
+    DEFAULT_USER_PROFILE, generate_action_suggestion, load_config
+)
+
+OUTPUT_ROOT = 'output/biz-daily'
+TOPICS = ['AI', '学术', '新闻', '文学', '投资']
+
+AD_PATTERNS = [
+    re.compile(r'在小说阅读器读本章\s*'),
+    re.compile(r'在小说阅读器中沉浸阅读\s*'),
+    re.compile(r'去阅读\s*'),
+    re.compile(r'Scan to Follow\s*'),
+    re.compile(r'轻触阅读原文\s*'),
+    re.compile(r'预览时标签不可点\s*'),
+    re.compile(r'继续滑动看下一个\s*'),
+    re.compile(r'\[.*?\]\(javascript:void\(0\);\)'),
+    re.compile(r'\n{4,}'),
+]
+
+INTEREST_PROMPT = """对以下AI领域文章生成深度解析：
+
+标题：{title}
+来源：{source}
+
+正文：
+{content}
+
+请用markdown返回：
+### 核心观点
+（1-2句）
+
+### 关键细节
+- 点1
+- 点2
+- 点3
+
+### 启示
+（1句话）"""
+
+
+BATCH_ACTION_PROMPT = """你是读者的科研助手。基于以下文章列表，为读者筛选并生成可落地的行动建议。
+
+【读者定位】
+{profile}
+
+【文章列表】
+{articles}
+
+请完成以下任务：
+
+1. **筛选**：只保留对读者有实际落地价值的文章（排除纯新闻、娱乐、广告、无关内容）
+2. **评估**：对每篇保留的文章，判断相关度（高/中）并给出理由
+3. **生成建议**：为每篇保留的文章生成三级行动建议：
+   - **立即可做**：今天就能执行的具体动作
+   - **本周计划**：本周可以推进的中期动作
+   - **长期关注**：值得持续跟踪的方向（仅高相关度时输出）
+
+输出格式要求：
+
+```
+## 📋 今日行动清单（快速浏览）
+- [ ] 动作1（来自：文章标题）
+- [ ] 动作2（来自：文章标题）
+...
+
+---
+
+## 详细建议
+
+### 🔴 高相关度
+
+#### 1. 《文章标题》
+> 来源：公众号名 | 主题：AI/学术
+
+**相关度**：高 — 一句话解释原因
+
+**行动建议**
+- **立即可做**：xxx
+- **本周计划**：xxx
+- **长期关注**：xxx
+
+### 🟡 中相关度
+
+#### 1. 《文章标题》
+...
+```
+
+如果筛选后没有值得建议的文章，直接输出：「今日文章暂无直接可落地的行动建议，建议信息性阅读即可。」"""
+
+
+def generate_batch_action_suggestions(articles_data: list[dict], api_key: str, profile: str = '') -> str:
+    """批量生成行动建议。
+    
+    Args:
+        articles_data: 文章数据列表，每项包含 title, source, topic, summary, content
+        api_key: DeepSeek API key
+        profile: 用户定位描述
+        
+    Returns:
+        格式化的行动建议 markdown 文本
+    """
+    if not profile:
+        profile = DEFAULT_USER_PROFILE
+    
+    # 构建文章列表文本
+    articles_text = []
+    for i, a in enumerate(articles_data, 1):
+        articles_text.append(
+            f"[{i}] 《{a['title']}》\n"
+            f"    来源：{a['source']} | 主题：{a['topic']} | 相关度：{a.get('relevance', '中')}\n"
+            f"    摘要：{a['summary'][:200]}...\n"
+            f"    正文节选：{a['content'][:500]}...\n"
+        )
+    
+    prompt = BATCH_ACTION_PROMPT.format(
+        profile=profile,
+        articles='\n'.join(articles_text)
+    )
+    
+    return call_deepseek(prompt, api_key, max_tokens=4000, timeout=120)
+
+
+def clean_ads(text: str) -> str:
+    for pat in AD_PATTERNS:
+        text = pat.sub('', text)
+    text = re.sub(r'\n来源：[^\n]+\n编辑：[^\n]+\n校对：[^\n]+\n校审：[^\n]+', '', text)
+    text = re.sub(r'\n>/ [^\n]+', '', text)
+    text = re.sub(r'\n{4,}', '\n\n\n', text)
+    return text.strip()
+
+
+def extract_topic_from_file(content: str) -> str:
+    """Extract topic from YAML frontmatter, fallback to text parsing."""
+    fm, _ = parse_frontmatter(content)
+    if 'topic' in fm and fm['topic'] in TOPICS:
+        return fm['topic']
+    # Fallback: legacy text-based parsing
+    m = re.search(r'> 主题：(\S+)', content)
+    if m: return m.group(1)
+    return '学术'
+
+
+def extract_tags_from_file(content: str) -> list[str]:
+    """Extract tags from YAML frontmatter."""
+    fm, _ = parse_frontmatter(content)
+    return fm.get('tags', [])
+
+
+def main():
+    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+    import argparse
+    parser = argparse.ArgumentParser()
+    parser.add_argument('date_dir', nargs='?', help='日期目录')
+    parser.add_argument('--api-key', help='DeepSeek API key')
+    parser.add_argument('--interest', default='AI', help='兴趣主题，默认 AI')
+    parser.add_argument('--profile', default='', help='用户定位描述（默认使用环境科学研究生画像）')
+    parser.add_argument('--skip-action', action='store_true', help='跳过行动建议生成')
+    args = parser.parse_args()
+
+    config = load_config()
+    api_key = args.api_key or os.environ.get('DEEPSEEK_API_KEY', '') or config.get('deepseekApiKey', '')
+    if not api_key:
+        print('[ERROR] 需要 DeepSeek API key。请通过 --api-key、环境变量 DEEPSEEK_API_KEY 或 ~/.weflow-cli/config.json 中的 deepseekApiKey 提供')
+        sys.exit(1)
+    
+    user_profile = args.profile if args.profile else DEFAULT_USER_PROFILE
+
+    if args.date_dir:
+        base = Path(OUTPUT_ROOT) / args.date_dir
+    else:
+        dirs = sorted(Path(OUTPUT_ROOT).glob('202*'), reverse=True)
+        if not dirs: print('[ERROR] 未找到日报目录'); sys.exit(1)
+        base = dirs[0]
+
+    md_files = sorted([f for f in base.rglob('*.md') if f.name != 'README.md'])
+    print(f'目录: {base}')
+    print(f'文章: {len(md_files)} 篇 | 兴趣: {args.interest}\n')
+
+    # ====== Step 1: Clean ads ======
+    print('=== Step 1: 广告清洗 ===')
+    cleaned = 0
+    for fpath in md_files:
+        with open(fpath, 'r', encoding='utf-8') as f:
+            content = f.read()
+        new_content = clean_ads(content)
+        if len(new_content) != len(content):
+            cleaned += 1
+        with open(fpath, 'w', encoding='utf-8') as f:
+            f.write(new_content)
+    print(f'  清洗 {cleaned} 篇\n')
+
+    # ====== Step 2: Read topics from md (already set by biz_daily) ======
+    print('=== Step 2: 读取主题 ===')
+    topic_map = {}
+    for fpath in md_files:
+        with open(fpath, 'r', encoding='utf-8') as f:
+            content = f.read()
+        topic = extract_topic_from_file(content)
+        if topic not in TOPICS:
+            for t in TOPICS:
+                if t in topic: topic = t; break
+            else: topic = '学术'
+        topic_map[fpath] = topic
+
+    dist = Counter(topic_map.values())
+    print(f'  分布: {dict(dist)}\n')
+
+    # ====== Step 3: Deep summary for interest ======
+    interest_files = [f for f, t in topic_map.items() if t == args.interest]
+    if interest_files:
+        print(f'=== Step 3: [{args.interest}] 深度摘要 ({len(interest_files)}篇) ===')
+        for i, fpath in enumerate(interest_files):
+            with open(fpath, 'r', encoding='utf-8') as f:
+                content = f.read()
+            title = re.search(r'^# (.+)', content, re.MULTILINE)
+            source = re.search(r'来源：(.+)', content)
+            title = title.group(1) if title else ''
+            source = source.group(1).strip() if source else ''
+            body_match = re.search(r'## 正文\n\n(.+)', content, re.DOTALL)
+            body = body_match.group(1)[:5000] if body_match else content[500:5500]
+
+            try:
+                prompt = INTEREST_PROMPT.format(title=title, source=source, content=body)
+                deep = call_deepseek(prompt, api_key, max_tokens=2000)
+                content = re.sub(
+                    r'## AI 摘要\n\n.+?(?=\n\n---|\Z)',
+                    f'## 深度解析\n\n{deep}',
+                    content, flags=re.DOTALL,
+                )
+                # Update frontmatter: mark as enhanced
+                fm, body = parse_frontmatter(content)
+                if fm:
+                    fm['enhanced'] = 'true'
+                    write_with_frontmatter(str(fpath), fm, body)
+                else:
+                    with open(fpath, 'w', encoding='utf-8') as f:
+                        f.write(content)
+                print(f'  [{i+1}/{len(interest_files)}] {title[:50]}')
+                time.sleep(0.3)
+            except Exception as e:
+                print(f'  [{i+1}] ERR: {e}')
+
+    # ====== Step 4: Generate action suggestions (batch) ======
+    if not args.skip_action:
+        print(f'\n=== Step 4: 生成行动建议 ===')
+        # 只处理 AI 和 学术 主题的文章
+        candidate_files = [f for f, t in topic_map.items() if t in ['AI', '学术']]
+        if candidate_files:
+            articles_data = []
+            skipped_articles = []
+            for fpath in candidate_files:
+                with open(fpath, 'r', encoding='utf-8') as f:
+                    content = f.read()
+                
+                fm, body = parse_frontmatter(content)
+                title = fm.get('title', '') if fm else ''
+                source = fm.get('source', '') if fm else ''
+                topic = fm.get('topic', '') if fm else ''
+                relevance = fm.get('relevance', '中') if fm else '中'
+                
+                # 提取摘要
+                summary_match = re.search(r'## AI 摘要\n\n(.+?)(?=\n\n---|\n\n## |\Z)', content, re.DOTALL)
+                summary = summary_match.group(1).strip() if summary_match else ''
+                
+                # 提取正文
+                body_match = re.search(r'## 正文\n\n(.+)', content, re.DOTALL)
+                body_text = body_match.group(1)[:3000] if body_match else content[:3000]
+                
+                articles_data.append({
+                    'title': title,
+                    'source': source,
+                    'topic': topic,
+                    'relevance': relevance,
+                    'summary': summary,
+                    'content': body_text,
+                })
+            
+            # 收集被跳过的文章（新闻/文学/投资类）
+            for fpath, topic in topic_map.items():
+                if topic not in ['AI', '学术']:
+                    with open(fpath, 'r', encoding='utf-8') as f:
+                        content = f.read()
+                    fm, _ = parse_frontmatter(content)
+                    title = fm.get('title', '') if fm else ''
+                    source = fm.get('source', '') if fm else ''
+                    skipped_articles.append({'title': title, 'source': source, 'topic': topic})
+            
+            try:
+                print(f'  候选文章: {len(articles_data)} 篇 (AI/学术)')
+                print(f'  跳过文章: {len(skipped_articles)} 篇 (新闻/文学/投资)')
+                action_content = generate_batch_action_suggestions(articles_data, api_key, user_profile)
+                
+                # 添加被跳过的文章列表
+                if skipped_articles:
+                    action_content += '\n\n---\n\n## 被跳过的文章（主题不符）\n\n'
+                    for a in skipped_articles:
+                        action_content += f"- 《{a['title']}》| 来源：{a['source']} | 主题：{a['topic']}\n"
+                
+                # 写入行动建议文件
+                action_file = base / '行动建议.md'
+                with open(action_file, 'w', encoding='utf-8') as f:
+                    f.write(f'# 行动建议 — {base.name}\n\n')
+                    f.write(f'> 基于定位：{user_profile[:80]}...\n\n')
+                    f.write(action_content)
+                print(f'  ✓ 已生成: {action_file}')
+            except Exception as e:
+                print(f'  [WARN] 行动建议生成失败: {e}')
+        else:
+            print('  无 AI/学术 类文章，跳过行动建议')
+    else:
+        print(f'\n=== Step 4: 跳过行动建议 (--skip-action) ===')
+
+    # ====== Step 5: Move to topic folders ======
+    print(f'\n=== Step 5: 重建目录 ===')
+    # 保护收藏文件夹和状态文件
+    fav_dir = base / '收藏'
+    fav_state = base / '.fav_state.json'
+    fav_backup = None
+    if fav_dir.exists():
+        import tempfile as _tmp
+        fav_backup = Path(_tmp.mkdtemp()) / '收藏'
+        shutil.copytree(str(fav_dir), str(fav_backup))
+        shutil.rmtree(str(fav_dir))
+    for topic in TOPICS:
+        (base / topic).mkdir(exist_ok=True)
+    for fpath, topic in topic_map.items():
+        try:
+            dest = base / topic / fpath.name
+            if fpath.parent != dest.parent:
+                shutil.move(str(fpath), str(dest))
+        except:
+            pass
+    # 恢复收藏
+    if fav_backup and fav_backup.exists():
+        shutil.copytree(str(fav_backup), str(fav_dir))
+        shutil.rmtree(str(fav_backup.parent))
+
+    # ====== Step 6: README ======
+    # Preserve briefing block from biz_daily README
+    briefing_block = []
+    old_readme = base / 'README.md'
+    if old_readme.exists():
+        try:
+            with open(old_readme, 'r', encoding='utf-8') as f:
+                for line in f:
+                    if '📋' in line and '简报' in line:
+                        briefing_block.append(line.rstrip())
+                        for line in f:
+                            if line.startswith('>'):
+                                stripped = line.rstrip()
+                                if stripped != '>':  # skip empty blockquote lines
+                                    briefing_block.append(stripped)
+                            else:
+                                break
+                        break
+        except:
+            pass
+
+    action_badge = ' ✅' if (base / '行动建议.md').exists() else ''
+    lines = [
+        f'# 公众号日报 — {base.name}',
+        '',
+    ]
+    if briefing_block:
+        lines.extend(briefing_block)
+        lines.append('')
+    lines.extend([
+        f'共 {len(md_files)} 篇 | 兴趣: {args.interest}',
+        '',
+        f'**[📋 查看行动建议](./行动建议.md){action_badge}** — 基于你的定位生成的可落地建议',
+        ''
+    ])
+    for topic in TOPICS:
+        files = sorted((base / topic).glob('*.md'))
+        if not files: continue
+        fire = ' 🔥' if topic == args.interest else ''
+        lines.append(f'## {topic}{fire} ({len(files)}篇)')
+        lines.append('')
+        for j, f in enumerate(files):
+            with open(f, 'r', encoding='utf-8') as fh:
+                t = fh.readline().strip('#').strip()
+            lines.append(f'{j+1}. [{t}](./{topic}/{f.name})')
+        lines.append('')
+
+    with open(base / 'README.md', 'w', encoding='utf-8') as f:
+        f.write('\n'.join(lines))
+
+    print(f'\n✓ 完成！')
+    for topic in TOPICS:
+        n = len(list((base / topic).glob('*.md')))
+        if n: print(f'  {topic}: {n} 篇')
+
+
+if __name__ == '__main__':
+    main()

package/scripts/compile_wiki.py

@@ -0,0 +1,258 @@
+#!/usr/bin/env python3
+"""
+概念图谱编译 — 扫描文章的 [[Wikilinks]] → 聚合 → DeepSeek 生成概念页。
+
+用法:
+  python scripts/compile_wiki.py --api-key <key> [--limit 20] [--source output/biz-daily]
+"""
+import sys, os, json, re, time, hashlib
+from pathlib import Path
+from collections import defaultdict, Counter
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+from _utils import call_deepseek, parse_frontmatter, write_with_frontmatter
+
+# Default paths
+SOURCE_ROOT = 'output/biz-daily'
+OUTPUT_ROOT = 'output/wechat-vault/Wiki/Concepts'
+
+CONCEPT_PROMPT = """为概念生成 Wiki 知识页。
+
+概念名：{name}
+
+参考来源（来自公众号文章）：
+{references}
+
+请按格式返回：
+【定义】
+（1-2句话定义这个概念）
+
+【关键要点】
+- 要点1
+- 要点2
+- 要点3
+
+【标签】
+tag1, tag2, tag3
+
+【相关概念】
+概念A, 概念B, 概念C
+
+要求：定义精准，要点简洁（每条≤30字），标签2-3个，相关概念2-4个。"""
+
+
+def scan_articles(source_dir: str) -> list[dict]:
+    """Scan all .md files, extract frontmatter + wikilinks."""
+    articles = []
+    for md_file in Path(source_dir).rglob('*.md'):
+        if md_file.name == 'README.md':
+            continue
+        try:
+            with open(md_file, 'r', encoding='utf-8') as f:
+                content = f.read()
+        except:
+            continue
+
+        fm, body = parse_frontmatter(content)
+
+        # Extract [[wikilinks]] with optional descriptions
+        wiki_pattern = re.findall(r'\[\[([^\]]+)\]\](?:\s*—?\s*([^\n]+))?', body)
+        wikilinks = [(name.strip(), desc.strip()) for name, desc in wiki_pattern]
+
+        if not wikilinks:
+            continue
+
+        articles.append({
+            'file': str(md_file.relative_to(source_dir)),
+            'title': fm.get('title', md_file.stem),
+            'source': fm.get('source', ''),
+            'topic': fm.get('topic', ''),
+            'tags': fm.get('tags', []),
+            'summary': _extract_summary(body),
+            'wikilinks': wikilinks,
+        })
+    return articles
+
+
+def _extract_summary(body: str) -> str:
+    """Extract the AI summary section from article body."""
+    m = re.search(r'## (?:AI 摘要|深度解析)\n\n(.+?)(?=\n\n##|\n\n---|\Z)', body, re.DOTALL)
+    if m:
+        return m.group(1).strip()[:500]
+    # Fallback: first paragraph after metadata
+    lines = body.strip().split('\n')
+    for line in lines:
+        line = line.strip()
+        if line and not line.startswith('#') and not line.startswith('>') and not line.startswith('-'):
+            return line[:200]
+    return ''
+
+
+def aggregate_concepts(articles: list[dict]) -> dict[str, list[dict]]:
+    """Aggregate wikilinks: concept_name -> [articles that reference it]."""
+    concept_map = defaultdict(list)
+    for art in articles:
+        seen = set()
+        for name, desc in art['wikilinks']:
+            if name in seen:
+                continue
+            seen.add(name)
+            concept_map[name].append({
+                'title': art['title'],
+                'source': art['source'],
+                'summary': art['summary'],
+                'desc': desc,
+                'file': art['file'],
+            })
+    return dict(concept_map)
+
+
+def generate_concept(name: str, refs: list[dict], api_key: str) -> str | None:
+    """Call DeepSeek to generate a concept Wiki page."""
+    # Build references section
+    ref_lines = []
+    for r in refs[:5]:  # max 5 references
+        ref_lines.append(f'- [{r["title"]}]（{r["source"]}）：{r["summary"][:150]}')
+    ref_text = '\n'.join(ref_lines) if ref_lines else '(无详细信息)'
+
+    prompt = CONCEPT_PROMPT.format(name=name, references=ref_text)
+    try:
+        response = call_deepseek(prompt, api_key, max_tokens=800)
+    except Exception as e:
+        print(f'  [ERR] {name}: {e}')
+        return None
+
+    # Parse response
+    definition_m = re.search(r'【定义】\s*(.+?)(?=\n【|$)', response, re.DOTALL)
+    points_m = re.search(r'【关键要点】\s*(.+?)(?=\n【|$)', response, re.DOTALL)
+    tags_m = re.search(r'【标签】\s*(.+)', response)
+    related_m = re.search(r'【相关概念】\s*(.+)', response)
+
+    definition = definition_m.group(1).strip() if definition_m else ''
+    points = points_m.group(1).strip() if points_m else ''
+    tags = [t.strip() for t in tags_m.group(1).split(',')] if tags_m else []
+    related = [r.strip() for r in related_m.group(1).split(',')] if related_m else []
+
+    # Build markdown body
+    body_parts = [f'# {name}\n']
+    if definition:
+        body_parts.append(f'{definition}\n\n')
+    if points:
+        body_parts.append('## 关键要点\n\n')
+        body_parts.append(points + '\n\n')
+    if related:
+        body_parts.append('## 相关概念\n\n')
+        for rc in related:
+            body_parts.append(f'- [[{rc}]]\n')
+        body_parts.append('\n')
+    body_parts.append('## 来源\n\n')
+    for r in refs[:5]:
+        body_parts.append(f'- [[{r["file"]}]] — {r["title"]}\n')
+
+    # Frontmatter
+    source_files = [r['file'] for r in refs[:5]]
+    today = time.strftime('%Y-%m-%d')
+    fm = {
+        'title': f'"{name}"',
+        'type': 'concept',
+        'tags': tags,
+        'created': today,
+        'sources': source_files,
+    }
+
+    return fm, ''.join(body_parts)
+
+
+def main():
+    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+    import argparse
+    parser = argparse.ArgumentParser(description='概念图谱编译')
+    parser.add_argument('--api-key', help='DeepSeek API key (或环境变量 DEEPSEEK_API_KEY)')
+    parser.add_argument('--limit', type=int, default=20, help='最多生成概念数 (默认20)')
+    parser.add_argument('--source', default=SOURCE_ROOT, help='文章目录')
+    parser.add_argument('--output', default=OUTPUT_ROOT, help='概念页输出目录')
+    args = parser.parse_args()
+
+    api_key = args.api_key or os.environ.get('DEEPSEEK_API_KEY', '')
+    if not api_key:
+        print('[ERROR] 需要 DeepSeek API key')
+        sys.exit(1)
+
+    source_dir = args.source
+    out_dir = Path(args.output)
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    # Step 1: Scan
+    print(f'=== Step 1: 扫描文章 ===')
+    articles = scan_articles(source_dir)
+    print(f'  找到 {len(articles)} 篇带 wikilinks 的文章')
+
+    if not articles:
+        print('No articles with wikilinks found. Exiting.')
+        return
+
+    # Step 2: Aggregate
+    print(f'\n=== Step 2: 聚合概念 ===')
+    concept_map = aggregate_concepts(articles)
+    ranked = sorted(concept_map.items(), key=lambda x: len(x[1]), reverse=True)
+    print(f'  共 {len(ranked)} 个概念（限制 TOP {args.limit}）')
+    for i, (name, refs) in enumerate(ranked[:10]):
+        print(f'  {i+1}. [[{name}]] — {len(refs)} 篇文章引用')
+
+    # Step 3: Generate
+    print(f'\n=== Step 3: AI 生成概念页 ===')
+    top_concepts = ranked[:args.limit]
+    generated = 0
+
+    # Load existing concepts to skip
+    skipped = 0
+    for name, refs in top_concepts:
+        safe_name = re.sub(r'[\\/:*?"<>|]', '_', name)[:60]
+        out_file = out_dir / f'{safe_name}.md'
+
+        if out_file.exists():
+            skipped += 1
+            print(f'  [SKIP] {name} (已存在)')
+            continue
+
+        print(f'  [{generated+1}/{args.limit}] {name} ({len(refs)} 引用)...')
+        result = generate_concept(name, refs, api_key)
+        if result:
+            fm, body = result
+            write_with_frontmatter(str(out_file), fm, body)
+            generated += 1
+            time.sleep(0.5)
+
+    if skipped:
+        print(f'  跳过 {skipped} 个已有概念')
+    print(f'  生成 {generated} 个新概念')
+
+    # Step 4: Index
+    print(f'\n=== Step 4: 生成索引 ===')
+    concept_files = sorted(out_dir.glob('*.md'))
+    index_lines = [
+        '# 概念索引',
+        '',
+        f'共 {len(concept_files)} 个概念 | 生成时间：{time.strftime("%Y-%m-%d %H:%M")}',
+        '',
+        '| # | 概念 | 引用数 |',
+        '|---|------|--------|',
+    ]
+    for i, cf in enumerate(concept_files):
+        with open(cf, 'r', encoding='utf-8') as f:
+            content = f.read()
+        fm, _ = parse_frontmatter(content)
+        title = fm.get('title', cf.stem)
+        count = len(concept_map.get(title, []))
+        index_lines.append(f'| {i+1} | [[{title}]] | {count} |')
+
+    index_path = out_dir.parent / '00-Overview.md'
+    with open(index_path, 'w', encoding='utf-8') as f:
+        f.write('\n'.join(index_lines) + '\n')
+
+    print(f'  索引: {index_path}')
+    print(f'\n✓ 完成！共 {len(concept_files)} 个概念页')
+
+
+if __name__ == '__main__':
+    main()