weflow-cli 1.5.0 → 1.6.1

package/scripts/vault_rag.py

@@ -1,139 +1,140 @@
-#!/usr/bin/env python3
-"""
-Vault RAG 对话 — 基于 Vault 知识库（文章+概念+笔记）的问答。
-
-用法:
-  python scripts/vault_rag.py "深度学习在遥感中的应用"
-  python scripts/vault_rag.py "最近读了什么文章" --json
-"""
-
-import sys, os, json, argparse, re
-from pathlib import Path
-
-SCRIPTS_DIR = os.path.dirname(os.path.abspath(__file__))
-PROJECT_ROOT = os.path.dirname(SCRIPTS_DIR)
-DEFAULT_VAULT = os.path.join(PROJECT_ROOT, 'output', 'wechat-vault')
-DEFAULT_BIZ = os.path.join(PROJECT_ROOT, 'output', 'biz-daily')
-
-sys.path.insert(0, SCRIPTS_DIR)
-from _utils import call_deepseek
-
-
-def collect_context(vault: str, biz_daily: str, question: str, top_k: int) -> list[dict]:
-    """从 Vault + biz-daily 检索相关上下文。"""
-    results = []
-    search_terms = question.lower().split()
-
-    # 1. 概念页（权重最高）
-    concepts_dir = Path(vault) / 'Wiki' / 'Concepts'
-    if concepts_dir.is_dir():
-        for md in concepts_dir.glob('*.md'):
-            try:
-                text = md.read_text(encoding='utf-8')[:2000]
-            except Exception:
-                continue
-            score = sum(text.lower().count(t) for t in search_terms)
-            if score > 0:
-                results.append({'source': 'concept', 'title': md.stem, 'content': text, 'score': score})
-
-    # 2. 阅读笔记
-    notes_dir = Path(vault) / 'Notes'
-    if notes_dir.is_dir():
-        for md in notes_dir.rglob('*.md'):
-            try:
-                text = md.read_text(encoding='utf-8')[:1500]
-            except Exception:
-                continue
-            score = sum(text.lower().count(t) for t in search_terms)
-            # 标题匹配加分
-            if any(t in md.stem.lower() for t in search_terms):
-                score += 15
-            if score > 0:
-                results.append({'source': 'note', 'title': md.stem[:60], 'content': text, 'score': score})
-
-    # 3. 最新文章
-    biz_path = Path(biz_daily)
-    for date_dir in sorted(biz_path.glob('20*'), reverse=True)[:7]:  # 最近 7 天
-        for md in date_dir.rglob('*.md'):
-            if md.name == 'README.md' or 'README' in md.name:
-                continue
-            try:
-                text = md.read_text(encoding='utf-8')[:1000]
-            except Exception:
-                continue
-            score = sum(text.lower().count(t) for t in search_terms)
-            if score > 0:
-                results.append({'source': 'article', 'title': md.stem[:60], 'content': text, 'score': score})
-
-    results.sort(key=lambda x: x['score'], reverse=True)
-    return results[:top_k]
-
-
-def build_prompt(question: str, context: list[dict]) -> str:
-    sources_text = ''
-    for i, ctx in enumerate(context):
-        label = {'concept': '概念', 'note': '笔记', 'article': '文章'}.get(ctx['source'], ctx['source'])
-        sources_text += f'\n[{i+1}] ({label}) {ctx["title"]}\n{ctx["content"][:600]}\n'
-
-    return f'''你是个人知识助手，基于用户笔记库回答问题。
-
-知识库内容：
-{sources_text}
-
-用户问题：{question}
-
-请综合知识库内容回答。如果知识库中没有相关信息，如实说明。'''
-
-
-def main():
-    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
-    parser = argparse.ArgumentParser(description='Vault RAG 对话')
-    parser.add_argument('question', nargs='?', default='', help='问题')
-    parser.add_argument('--top-k', default='8', help='检索条数')
-    parser.add_argument('--vault', default=DEFAULT_VAULT, help='Vault 路径')
-    parser.add_argument('--biz-daily', default=DEFAULT_BIZ, help='biz-daily 路径')
-    parser.add_argument('--api-key', default='', help='DeepSeek API key')
-    parser.add_argument('--json', action='store_true', help='JSON 输出')
-    args = parser.parse_args()
-
-    if not args.question:
-        print('请提供问题')
-        sys.exit(1)
-
-    api_key = args.api_key or os.environ.get('DEEPSEEK_API_KEY', '')
-    if not api_key:
-        print('[ERROR] 缺少 API Key')
-        sys.exit(1)
-
-    top_k = int(args.top_k)
-    context = collect_context(args.vault, args.biz_daily, args.question, top_k)
-
-    if args.json:
-        output = {'question': args.question, 'sources': [{'source': c['source'], 'title': c['title'], 'score': c['score']} for c in context]}
-        if context:
-            prompt = build_prompt(args.question, context)
-            output['answer'] = call_deepseek(prompt, api_key, max_tokens=800)
-        else:
-            output['answer'] = '知识库中没有相关信息'
-        print(json.dumps(output, ensure_ascii=False, indent=2))
-        return
-
-    print(f'🔍 "{args.question}"\n')
-    if not context:
-        print('未找到相关内容')
-        return
-
-    print(f'检索到 {len(context)} 条相关内容:\n')
-    for i, c in enumerate(context):
-        icon = {'concept': '🧠', 'note': '📝', 'article': '📄'}.get(c['source'], '📎')
-        print(f'  {i+1}. {icon} [{c["source"]}] {c["title"]}')
-
-    prompt = build_prompt(args.question, context)
-    print(f'\n{"="*50}')
-    answer = call_deepseek(prompt, api_key, max_tokens=800)
-    print(answer)
-    print(f'{"="*50}')
-
-
-if __name__ == '__main__':
-    main()
+#!/usr/bin/env python3
+"""
+Vault RAG 对话 — 基于 Vault 知识库（文章+概念+笔记）的问答。
+
+用法:
+  python scripts/vault_rag.py "深度学习在遥感中的应用"
+  python scripts/vault_rag.py "最近读了什么文章" --json
+"""
+
+import sys, os, json, argparse, re
+from pathlib import Path
+
+SCRIPTS_DIR = os.path.dirname(os.path.abspath(__file__))
+PROJECT_ROOT = os.path.dirname(SCRIPTS_DIR)
+DEFAULT_VAULT = os.path.join(PROJECT_ROOT, 'output', 'wechat-vault')
+DEFAULT_BIZ = os.path.join(PROJECT_ROOT, 'output', 'biz-daily')
+
+sys.path.insert(0, SCRIPTS_DIR)
+from _utils import call_deepseek
+
+
+def collect_context(vault: str, biz_daily: str, question: str, top_k: int) -> list[dict]:
+    """从 Vault + biz-daily 检索相关上下文。"""
+    results = []
+    search_terms = question.lower().split()
+
+    # 1. 概念页（权重最高）
+    concepts_dir = Path(vault) / 'Wiki' / 'Concepts'
+    if concepts_dir.is_dir():
+        for md in concepts_dir.glob('*.md'):
+            try:
+                text = md.read_text(encoding='utf-8')[:2000]
+            except Exception:
+                continue
+            score = sum(text.lower().count(t) for t in search_terms)
+            if score > 0:
+                results.append({'source': 'concept', 'title': md.stem, 'content': text, 'score': score})
+
+    # 2. 阅读笔记
+    notes_dir = Path(vault) / 'Notes'
+    if notes_dir.is_dir():
+        for md in notes_dir.rglob('*.md'):
+            try:
+                text = md.read_text(encoding='utf-8')[:1500]
+            except Exception:
+                continue
+            score = sum(text.lower().count(t) for t in search_terms)
+            # 标题匹配加分
+            if any(t in md.stem.lower() for t in search_terms):
+                score += 15
+            if score > 0:
+                results.append({'source': 'note', 'title': md.stem[:60], 'content': text, 'score': score})
+
+    # 3. 最新文章
+    biz_path = Path(biz_daily)
+    for date_dir in sorted(biz_path.glob('20*'), reverse=True)[:7]:  # 最近 7 天
+        for md in date_dir.rglob('*.md'):
+            if md.name == 'README.md' or 'README' in md.name:
+                continue
+            try:
+                text = md.read_text(encoding='utf-8')[:1000]
+            except Exception:
+                continue
+            score = sum(text.lower().count(t) for t in search_terms)
+            if score > 0:
+                results.append({'source': 'article', 'title': md.stem[:60], 'content': text, 'score': score})
+
+    results.sort(key=lambda x: x['score'], reverse=True)
+    return results[:top_k]
+
+
+def build_prompt(question: str, context: list[dict]) -> str:
+    sources_text = ''
+    for i, ctx in enumerate(context):
+        label = {'concept': '概念', 'note': '笔记', 'article': '文章'}.get(ctx['source'], ctx['source'])
+        sources_text += f'\n[{i+1}] ({label}) {ctx["title"]}\n{ctx["content"][:600]}\n'
+
+    return f'''你是个人知识助手，基于用户笔记库回答问题。
+
+知识库内容：
+{sources_text}
+
+用户问题：{question}
+
+请综合知识库内容回答。如果知识库中没有相关信息，如实说明。'''
+
+
+def main():
+    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+    parser = argparse.ArgumentParser(description='Vault RAG 对话')
+    parser.add_argument('question', nargs='?', default='', help='问题')
+    parser.add_argument('--top-k', default='8', help='检索条数')
+    parser.add_argument('--vault', default=DEFAULT_VAULT, help='Vault 路径')
+    parser.add_argument('--biz-daily', default=DEFAULT_BIZ, help='biz-daily 路径')
+    parser.add_argument('--api-key', default='', help='DeepSeek API key')
+    parser.add_argument('--json', action='store_true', help='JSON 输出')
+    args = parser.parse_args()
+
+    question = args.question or os.environ.get('WEFLOW_VAULT_QUESTION', '')
+    if not question:
+        print('请提供问题')
+        sys.exit(1)
+
+    api_key = args.api_key or os.environ.get('DEEPSEEK_API_KEY', '')
+    if not api_key:
+        print('[ERROR] 缺少 API Key')
+        sys.exit(1)
+
+    top_k = int(args.top_k)
+    context = collect_context(args.vault, args.biz_daily, question, top_k)
+
+    if args.json:
+        output = {'question': question, 'sources': [{'source': c['source'], 'title': c['title'], 'score': c['score']} for c in context]}
+        if context:
+            prompt = build_prompt(question, context)
+            output['answer'] = call_deepseek(prompt, api_key, max_tokens=800)
+        else:
+            output['answer'] = '知识库中没有相关信息'
+        print(json.dumps(output, ensure_ascii=False, indent=2))
+        return
+
+    print(f'🔍 "{question}"\n')
+    if not context:
+        print('未找到相关内容')
+        return
+
+    print(f'检索到 {len(context)} 条相关内容:\n')
+    for i, c in enumerate(context):
+        icon = {'concept': '🧠', 'note': '📝', 'article': '📄'}.get(c['source'], '📎')
+        print(f'  {i+1}. {icon} [{c["source"]}] {c["title"]}')
+
+    prompt = build_prompt(question, context)
+    print(f'\n{"="*50}')
+    answer = call_deepseek(prompt, api_key, max_tokens=800)
+    print(answer)
+    print(f'{"="*50}')
+
+
+if __name__ == '__main__':
+    main()

package/scripts/vault_search.py

@@ -1,141 +1,141 @@
-#!/usr/bin/env python3
-"""
-Vault 全局搜索 — 关键词 + AI 排序，覆盖文章/概念/笔记。
-
-用法:
-  python scripts/vault_search.py "遥感反演" --top-k 10
-  python scripts/vault_search.py "空气污染" --type article --days 30
-"""
-
-import sys, os, json, argparse, re
-from pathlib import Path
-from datetime import datetime, timedelta
-
-SCRIPTS_DIR = os.path.dirname(os.path.abspath(__file__))
-PROJECT_ROOT = os.path.dirname(SCRIPTS_DIR)
-DEFAULT_VAULT = os.path.join(PROJECT_ROOT, 'output', 'wechat-vault')
-DEFAULT_BIZ = os.path.join(PROJECT_ROOT, 'output', 'biz-daily')
-
-
-def search_text(query: str, files: list[tuple], top_k: int) -> list[dict]:
-    """关键词搜索 + 简单排序。"""
-    terms = query.lower().split()
-    scored = []
-    for filepath, category, meta in files:
-        try:
-            text = filepath.read_text(encoding='utf-8')[:2000]
-        except Exception:
-            continue
-        text_lower = text.lower()
-        # 计分：标题匹配 > 关键词频次
-        score = 0
-        title = meta.get('title', '')
-        if any(t in title.lower() for t in terms):
-            score += 10
-        for t in terms:
-            score += text_lower.count(t)
-        if score > 0:
-            scored.append({'file': str(filepath), 'category': category, 'title': title,
-                           'score': score, 'snippet': text[:300]})
-    scored.sort(key=lambda x: x['score'], reverse=True)
-    return scored[:top_k]
-
-
-def collect_files(vault: str, biz_daily: str, search_type: str, days: int) -> list[tuple]:
-    """收集可搜索文件。"""
-    files = []
-    cutoff = datetime.now() - timedelta(days=days)
-
-    def parse_fm(text):
-        if not text.startswith('---'):
-            return {}
-        end = text.find('---', 3)
-        if end == -1:
-            return {}
-        fm = {}
-        for line in text[3:end].strip().split('\n'):
-            if ':' in line:
-                k, _, v = line.partition(':')
-                fm[k.strip()] = v.strip().strip('"').strip("'")
-        return fm
-
-    if search_type in ('all', 'article'):
-        # biz-daily 文章
-        biz_dir = Path(biz_daily)
-        for date_dir in sorted(biz_dir.glob('20*'), reverse=True):
-            try:
-                dir_date = datetime.strptime(date_dir.name, '%Y-%m-%d')
-                if dir_date < cutoff:
-                    continue
-            except ValueError:
-                continue
-            for md in date_dir.rglob('*.md'):
-                if md.name == 'README.md':
-                    continue
-                try:
-                    fm = parse_fm(md.read_text(encoding='utf-8'))
-                except Exception:
-                    fm = {}
-                files.append((md, 'article', fm))
-
-    if search_type in ('all', 'concept'):
-        vault_path = Path(vault)
-        concepts_dir = vault_path / 'Wiki' / 'Concepts'
-        if concepts_dir.is_dir():
-            for md in concepts_dir.glob('*.md'):
-                try:
-                    fm = parse_fm(md.read_text(encoding='utf-8'))
-                except Exception:
-                    fm = {}
-                files.append((md, 'concept', fm))
-
-    if search_type in ('all', 'note'):
-        vault_path = Path(vault)
-        notes_dir = vault_path / 'Notes'
-        if notes_dir.is_dir():
-            for md in notes_dir.rglob('*.md'):
-                try:
-                    fm = parse_fm(md.read_text(encoding='utf-8'))
-                except Exception:
-                    fm = {}
-                files.append((md, 'note', fm))
-
-    return files
-
-
-def main():
-    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
-    parser = argparse.ArgumentParser(description='Vault 全局搜索')
-    parser.add_argument('query', nargs='?', default='', help='搜索关键词')
-    parser.add_argument('--top-k', default='10', help='返回数量')
-    parser.add_argument('--type', default='all', choices=['all', 'article', 'concept', 'note'])
-    parser.add_argument('--days', type=int, default=90, help='搜索天数范围')
-    parser.add_argument('--vault', default=DEFAULT_VAULT, help='Vault 路径')
-    parser.add_argument('--biz-daily', default=DEFAULT_BIZ, help='biz-daily 路径')
-    parser.add_argument('--json', action='store_true', help='JSON 输出')
-    args = parser.parse_args()
-
-    if not args.query:
-        print('请提供搜索关键词')
-        sys.exit(1)
-
-    files = collect_files(args.vault, args.biz_daily, args.type, args.days)
-    results = search_text(args.query, files, int(args.top_k))
-
-    if args.json:
-        print(json.dumps(results, ensure_ascii=False, indent=2))
-        return
-
-    print(f'🔍 搜索 "{args.query}" 找到 {len(results)} 条 (类型: {args.type}, {args.days}天内)\n')
-    for i, r in enumerate(results):
-        icon = {'article': '📄', 'concept': '🧠', 'note': '📝'}.get(r['category'], '📎')
-        num = str(i + 1).rjust(2)
-        print(f'{num}. {icon} {r["title"] or r["file"].split("/")[-1]}')
-        print(f'   {r["category"]} | 相关度: {r["score"]}')
-        snippet = re.sub(r'\s+', ' ', r['snippet'][:120])
-        print(f'   {snippet}')
-        print()
-
-
-if __name__ == '__main__':
-    main()
+#!/usr/bin/env python3
+"""
+Vault 全局搜索 — 关键词 + AI 排序，覆盖文章/概念/笔记。
+
+用法:
+  python scripts/vault_search.py "遥感反演" --top-k 10
+  python scripts/vault_search.py "空气污染" --type article --days 30
+"""
+
+import sys, os, json, argparse, re
+from pathlib import Path
+from datetime import datetime, timedelta
+
+SCRIPTS_DIR = os.path.dirname(os.path.abspath(__file__))
+PROJECT_ROOT = os.path.dirname(SCRIPTS_DIR)
+DEFAULT_VAULT = os.path.join(PROJECT_ROOT, 'output', 'wechat-vault')
+DEFAULT_BIZ = os.path.join(PROJECT_ROOT, 'output', 'biz-daily')
+
+
+def search_text(query: str, files: list[tuple], top_k: int) -> list[dict]:
+    """关键词搜索 + 简单排序。"""
+    terms = query.lower().split()
+    scored = []
+    for filepath, category, meta in files:
+        try:
+            text = filepath.read_text(encoding='utf-8')[:2000]
+        except Exception:
+            continue
+        text_lower = text.lower()
+        # 计分：标题匹配 > 关键词频次
+        score = 0
+        title = meta.get('title', '')
+        if any(t in title.lower() for t in terms):
+            score += 10
+        for t in terms:
+            score += text_lower.count(t)
+        if score > 0:
+            scored.append({'file': str(filepath), 'category': category, 'title': title,
+                           'score': score, 'snippet': text[:300]})
+    scored.sort(key=lambda x: x['score'], reverse=True)
+    return scored[:top_k]
+
+
+def collect_files(vault: str, biz_daily: str, search_type: str, days: int) -> list[tuple]:
+    """收集可搜索文件。"""
+    files = []
+    cutoff = datetime.now() - timedelta(days=days)
+
+    def parse_fm(text):
+        if not text.startswith('---'):
+            return {}
+        end = text.find('---', 3)
+        if end == -1:
+            return {}
+        fm = {}
+        for line in text[3:end].strip().split('\n'):
+            if ':' in line:
+                k, _, v = line.partition(':')
+                fm[k.strip()] = v.strip().strip('"').strip("'")
+        return fm
+
+    if search_type in ('all', 'article'):
+        # biz-daily 文章
+        biz_dir = Path(biz_daily)
+        for date_dir in sorted(biz_dir.glob('20*'), reverse=True):
+            try:
+                dir_date = datetime.strptime(date_dir.name, '%Y-%m-%d')
+                if dir_date < cutoff:
+                    continue
+            except ValueError:
+                continue
+            for md in date_dir.rglob('*.md'):
+                if md.name == 'README.md':
+                    continue
+                try:
+                    fm = parse_fm(md.read_text(encoding='utf-8'))
+                except Exception:
+                    fm = {}
+                files.append((md, 'article', fm))
+
+    if search_type in ('all', 'concept'):
+        vault_path = Path(vault)
+        concepts_dir = vault_path / 'Wiki' / 'Concepts'
+        if concepts_dir.is_dir():
+            for md in concepts_dir.glob('*.md'):
+                try:
+                    fm = parse_fm(md.read_text(encoding='utf-8'))
+                except Exception:
+                    fm = {}
+                files.append((md, 'concept', fm))
+
+    if search_type in ('all', 'note'):
+        vault_path = Path(vault)
+        notes_dir = vault_path / 'Notes'
+        if notes_dir.is_dir():
+            for md in notes_dir.rglob('*.md'):
+                try:
+                    fm = parse_fm(md.read_text(encoding='utf-8'))
+                except Exception:
+                    fm = {}
+                files.append((md, 'note', fm))
+
+    return files
+
+
+def main():
+    sys.stdout.reconfigure(encoding='utf-8', errors='replace')
+    parser = argparse.ArgumentParser(description='Vault 全局搜索')
+    parser.add_argument('query', nargs='?', default='', help='搜索关键词')
+    parser.add_argument('--top-k', default='10', help='返回数量')
+    parser.add_argument('--type', default='all', choices=['all', 'article', 'concept', 'note'])
+    parser.add_argument('--days', type=int, default=90, help='搜索天数范围')
+    parser.add_argument('--vault', default=DEFAULT_VAULT, help='Vault 路径')
+    parser.add_argument('--biz-daily', default=DEFAULT_BIZ, help='biz-daily 路径')
+    parser.add_argument('--json', action='store_true', help='JSON 输出')
+    args = parser.parse_args()
+
+    if not args.query:
+        print('请提供搜索关键词')
+        sys.exit(1)
+
+    files = collect_files(args.vault, args.biz_daily, args.type, args.days)
+    results = search_text(args.query, files, int(args.top_k))
+
+    if args.json:
+        print(json.dumps(results, ensure_ascii=False, indent=2))
+        return
+
+    print(f'🔍 搜索 "{args.query}" 找到 {len(results)} 条 (类型: {args.type}, {args.days}天内)\n')
+    for i, r in enumerate(results):
+        icon = {'article': '📄', 'concept': '🧠', 'note': '📝'}.get(r['category'], '📎')
+        num = str(i + 1).rjust(2)
+        print(f'{num}. {icon} {r["title"] or r["file"].split("/")[-1]}')
+        print(f'   {r["category"]} | 相关度: {r["score"]}')
+        snippet = re.sub(r'\s+', ' ', r['snippet'][:120])
+        print(f'   {snippet}')
+        print()
+
+
+if __name__ == '__main__':
+    main()