Initial commit: book-mcp 教辅处理 MCP server + Arco Vue 管理端

后端 (book_mcp_server.py, 335 行, 15 个 MCP 工具):
- create_book / add_document / mark_html_ready / mark_json_ready / mark_failed
- update_book_status / soft_delete_book / retry_failed
- get_book / list_books / get_document / list_documents / get_document_url
- upload_to_rustfs / health
- 走 JSON-RPC + stateless_http + json_response（前端可直接调）
- launchd 守护 / 6 秒拉起（macOS）
- systemd 守护脚本（Linux/codelab）

存储:
- MongoDB (mongo:8.0 容器) - book/book_document/document_file 三表
- RustFS 对象存储（fs.code-lab.cn/scan/）- book/<id>/docs/<doc_id>/{docx,html,json}
- 幂等设计: zip_sha256 唯一 + (book_id+original_path) 唯一

前端 (Vue 3 + TS + Arco Design):
- 列表页 /book/list - 搜索 + 状态过滤 + 进度
- 详情页 /book/:id - 文档折叠 + 文件下载
- MCP 客户端 src/api/mcp.ts (JSON-RPC over fetch)

部署:
- deploy/install.sh (codelab 一键装 backend + systemd)
- deploy/install-frontend.sh (vite dev / 生产构建)
- docs/deploy.md (运维手册)

测试:
- tests/test_basic.py (sanity 7 项)
- tests/test_e2e.py (完整 e2e: docx→HTML→入库→上传→校验)

services/wordparse/__init__.py

@@ -0,0 +1,18 @@
+"""wordparse 子包：docx → HTML 工具集
+
+子模块:
+  cleaner  — 修脏 docx XML
+  post     — HTTP 调 docx4j
+  render   — HTML 美化（MathJax / 样式 / 导航 / 统计）
+  convert  — 高级入口（convert_one / batch_convert）
+"""
+from .cleaner import clean_docx
+from .post import post_docx, health
+from .render import clean_html, stats, copy_mathjax_to
+from .convert import convert_one, batch_convert
+
+__all__ = [
+    'clean_docx', 'post_docx', 'health',
+    'clean_html', 'stats', 'copy_mathjax_to',
+    'convert_one', 'batch_convert',
+]

services/wordparse/cleaner.py

@@ -0,0 +1,76 @@
+"""docx XML 清洗：去脏值（单位 / NaN / 浮点）以避开 docx4j 8.3.9 JAXB 解析失败。
+
+参考 wordparse_server.py 的 _clean_docx_xml，独立成模块方便测试。
+"""
+import re
+import os
+import zipfile
+
+XML_DECL = re.compile(r'<\?xml[^?]*\?>')
+
+
+def _fix_text(txt: str) -> str:
+    """修三类问题：
+      1. 带单位的数字属性："210mm" / "25.4cm" → "0"
+      2. NaN / Inf / -Inf / 空字符串 → "0"
+      3. 浮点属性 → 整数（JAXB BigInteger 不接小数）
+    """
+    decl = XML_DECL.search(txt)
+    decl_str = decl.group(0) if decl else '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+    if decl:
+        txt = XML_DECL.sub('__XMLDECL__', txt, count=1)
+
+    # 1. 单位
+    txt = re.sub(r'="(-?[0-9]*\.?[0-9]+)\s*(mm|cm|pt|px|in|em|%)"', r'="0"', txt, flags=re.I)
+    # 2. NaN / Inf
+    txt = re.sub(r'="(NaN|nan|Inf|-Inf|inf|-inf)"', '="0"', txt, flags=re.I)
+    # 3. 空
+    txt = re.sub(r'=""', '="0"', txt)
+    # 4. 浮点 → 整数
+    def _num(m):
+        try:
+            f = float(m.group(1))
+            if f != f or f == float('inf') or f == float('-inf'):
+                return '="0"'
+            return f'="{int(f)}"'
+        except Exception:
+            return m.group(0)
+
+    txt = re.sub(r'="(-?[0-9]+\.[0-9]+(?:[eE][-+]?[0-9]+)?)"', _num, txt)
+
+    if decl:
+        txt = txt.replace('__XMLDECL__', decl_str, 1)
+    return txt
+
+
+def clean_docx(path: str) -> str:
+    """清洗 docx 里的 word/document.xml，幂等。失败回退原文件。"""
+    marker = path + '.cleaned'
+    if os.path.exists(marker):
+        return path
+
+    try:
+        with zipfile.ZipFile(path, 'r') as zin:
+            entries = []
+            changed = False
+            for item in zin.infolist():
+                data = zin.read(item.filename)
+                if item.filename == 'word/document.xml':
+                    raw = data.decode('utf-8', errors='replace')
+                    new = _fix_text(raw)
+                    if new != raw:
+                        data = new.encode('utf-8')
+                        changed = True
+                entries.append((item, data))
+        if not changed:
+            open(marker, 'w').close()
+            return path
+        tmp = path + '.tmp'
+        with zipfile.ZipFile(tmp, 'w', zipfile.ZIP_DEFLATED) as zout:
+            for item, data in entries:
+                zout.writestr(item, data)
+        os.replace(tmp, path)
+        open(marker, 'w').close()
+        return path
+    except Exception:
+        return path

services/wordparse/convert.py

@@ -0,0 +1,91 @@
+"""docx → HTML 高层入口（集成清洗 + 转码 + 美化）"""
+import os
+import re
+import zipfile
+from typing import Callable
+
+from .cleaner import clean_docx
+from .post import post_docx
+from .render import clean_html, stats as render_stats, copy_mathjax_to
+
+
+def _zip_name(info) -> str:
+    """zip 条目名：带 UTF-8 标志位(0x800)直接用；否则按 cp437→GBK 还原"""
+    if info.flag_bits // 2048 % 2:
+        return info.filename
+    try:
+        return info.filename.encode('cp437').decode('gbk')
+    except Exception:
+        return info.filename
+
+
+def convert_one(docx_path: str, auto_clean: bool = True) -> dict:
+    """单份 docx → HTML 字符串 + stats
+
+    Returns: {ok, html, stats, error?}
+    """
+    try:
+        if auto_clean:
+            docx_path = clean_docx(docx_path)
+        raw = post_docx(docx_path)
+        html = clean_html(raw, mathjax=True)
+        return {'ok': True, 'html': html, 'stats': render_stats(html)}
+    except Exception as e:
+        return {'ok': False, 'error': str(e)[:300]}
+
+
+def batch_convert(path: str, out_dir: str, auto_clean: bool = True,
+                  progress_cb: Callable | None = None) -> dict:
+    """目录或 zip → 每份 docx 转 HTML 落 out_dir
+
+    Returns: {count, errors[{file, error}], files[{name, group, stats}]}
+    """
+    os.makedirs(out_dir, exist_ok=True)
+    copy_mathjax_to(out_dir)
+
+    items: list[tuple[str, str]] = []
+    if os.path.isdir(path):
+        for f in sorted(os.listdir(path)):
+            if f.lower().endswith('.docx'):
+                items.append((f, os.path.join(path, f)))
+    else:
+        with zipfile.ZipFile(path) as z:
+            for i in z.infolist():
+                if i.is_dir() or i.filename.startswith('__MACOSX/') or not i.filename.lower().endswith('.docx'):
+                    continue
+                parts = [p for p in _zip_name(i).split('/') if p]
+                rel = '/'.join(parts[1:]) if len(parts) > 1 else parts[-1]
+                # 不解压到磁盘，直接 post bytes
+                items.append((rel, zipfile.ZipFile(path).read(i)))
+
+    files, errors = [], []
+    for rel, src in items:
+        try:
+            if isinstance(src, bytes):
+                # 写临时文件
+                tmp = f'/tmp/_wpc_{abs(hash(rel)) & 0xFFFFFFFF:x}.docx'
+                with open(tmp, 'wb') as f:
+                    f.write(src)
+                src = tmp
+            r = convert_one(src, auto_clean=auto_clean)
+            if r['ok']:
+                html = r['html']
+                rel_html = re.sub(r'\.docx$', '', rel, flags=re.I) + '.html'
+                out = os.path.join(out_dir, rel_html)
+                os.makedirs(os.path.dirname(out) or out_dir, exist_ok=True)
+                with open(out, 'w', encoding='utf-8') as f:
+                    f.write(html)
+                files.append({
+                    'name': rel_html,
+                    'group': os.path.dirname(rel_html),
+                    **r['stats'],
+                })
+            else:
+                errors.append({'file': rel, 'error': r.get('error', '?')})
+        except Exception as e:
+            errors.append({'file': rel, 'error': str(e)[:200]})
+
+        if progress_cb:
+            progress_cb(len(files) + len(errors), len(items))
+
+    return {'count': len(files), 'errors': errors, 'files': files}

services/wordparse/post.py

@@ -0,0 +1,46 @@
+"""docx4j HTTP 客户端 — POST 一份 docx，返回 HTML 字符串"""
+import os
+import uuid
+import json
+import urllib.request
+from db.database import load_config
+
+
+def _config():
+    return load_config()['wordparse']
+
+
+def post_docx(path: str, timeout: int | None = None) -> str:
+    """返回 HTML 字符串（原始，未做 MathJax/导航注入）"""
+    cfg = _config()
+    url = cfg['url']
+    timeout = timeout or cfg.get('timeout', 600)
+
+    boundary = '----WP' + uuid.uuid4().hex
+    with open(path, 'rb') as f:
+        data = f.read()
+    fn = os.path.basename(path)
+    body = (
+        f'--{boundary}\r\n'.encode()
+        + f'Content-Disposition: form-data; name="file"; filename="{fn}"\r\n'.encode('utf-8')
+        + b'Content-Type: application/vnd.openxmlformats-officedocument.wordprocessingml.document\r\n\r\n'
+        + data
+        + b'\r\n' + f'--{boundary}--\r\n'.encode()
+    )
+    req = urllib.request.Request(url, data=body, headers={
+        'Content-Type': f'multipart/form-data; boundary={boundary}',
+        'Content-Length': str(len(body)),
+    })
+    with urllib.request.urlopen(req, timeout=timeout) as r:
+        resp = json.loads(r.read().decode('utf-8'))
+    return resp.get('content') or ''
+
+
+def health() -> bool:
+    cfg = _config()
+    try:
+        with urllib.request.urlopen(cfg['health_url'], timeout=5) as r:
+            j = json.loads(r.read().decode('utf-8'))
+            return j.get('status') == 'ok'
+    except Exception:
+        return False

services/wordparse/render.py

@@ -0,0 +1,70 @@
+"""HTML 美化：注入 MathJax / 样式 / 导航 / 统计"""
+import re
+import os
+import shutil
+import urllib.request
+
+# --- 常量 ---
+NAV = ('<div class="nav" style="position:sticky;top:0;background:#f8fafc;border-bottom:1px solid #e2e8f0;'
+       'padding:8px 0;margin:0 0 16px;font-size:14px"><a href="index.html" '
+       'style="color:#2563eb;text-decoration:none">← 返回索引</a></div>')
+
+STYLE = ('<meta name="viewport" content="width=device-width, initial-scale=1">'
+         '<style>body{max-width:900px;margin:0 auto;padding:24px 18px;font-family:-apple-system,'
+         '"PingFang SC","Microsoft YaHei",serif;font-size:16px;line-height:1.9;color:#1a1a1a;background:#fff}'
+         'img{max-width:100%;height:auto}table{border-collapse:collapse}td,th{border:1px solid #ccc;padding:4px 6px}'
+         'mjx-container{overflow-x:auto;overflow-y:hidden;max-width:100%}</style>')
+
+MATHJAX_TAG = ('<script>window.MathJax={tex:{inlineMath:[["\\\\(","\\\\)"]],displayMath:[["\\\\[","\\\\]"]]},'
+               'svg:{fontCache:"local"},options:{enableMenu:false}};</script>'
+               '<script src="tex-svg.js"></script>')
+
+ASSETS_DIR = os.path.expanduser('~/.hermes/mcp/assets')
+MATHJAX_FILE = os.path.join(ASSETS_DIR, 'tex-svg.js')
+MATHJAX_MIRROR = 'https://registry.npmmirror.com/mathjax/3.2.2/files/es5/tex-svg.js'
+
+
+def _fix_math(m):
+    inner = (m.group(1).replace('&nbsp;', ' ').replace('&amp;', '&')
+                      .replace('&lt;', '<').replace('&gt;', '>').replace('&#160;', ' '))
+    return r'\(' + re.sub(r'[ \t]{2,}', ' ', inner).strip() + r'\)'
+
+
+def clean_html(content: str, mathjax: bool = True) -> str:
+    """原始 HTML → 加 MathJax + 导航 + 样式"""
+    c = re.sub(r'^<\?xml[^>]*\?>\s*', '', content)
+    c = re.sub(r'<!DOCTYPE[^>]*>', '<!DOCTYPE html>', c, count=1)
+    c = re.sub(r'\\\((.*?)\\\)', _fix_math, c, flags=re.S)
+    inject = STYLE + (MATHJAX_TAG if mathjax else '')
+    c = c.replace('</head>', inject + '</head>', 1) if '</head>' in c else inject + c
+    m = re.search(r'<body[^>]*>', c, re.I)
+    c = (c[:m.end()] + NAV + c[m.end():]) if m else (NAV + c)
+    return c
+
+
+def stats(html: str) -> dict:
+    """HTML 统计：字符 / 公式 / 图片 / 未包裹 LaTeX"""
+    stripped = re.sub(r'\\\(.*?\\\)', '', html, flags=re.S)
+    return {
+        'chars': len(html),
+        'formulas': len(re.findall(r'\\\(', html)),
+        'images': html.count('<img'),
+        'unwrapped_latex': len(re.findall(
+            r'\\(?:dfrac|frac|sqrt|left|begin\{array)', stripped)),
+    }
+
+
+def ensure_mathjax() -> str:
+    """下载/复用 MathJax 离线文件 → 返回路径"""
+    os.makedirs(ASSETS_DIR, exist_ok=True)
+    if not os.path.exists(MATHJAX_FILE) or os.path.getsize(MATHJAX_FILE) < 500_000:
+        urllib.request.urlretrieve(MATHJAX_MIRROR, MATHJAX_FILE)
+    return MATHJAX_FILE
+
+
+def copy_mathjax_to(out_dir: str) -> str:
+    """把 MathJax 复制到 out_dir，返回目标路径"""
+    src = ensure_mathjax()
+    dst = os.path.join(out_dir, 'tex-svg.js')
+    shutil.copy(src, dst)
+    return dst