internal-web-reader 0.2.2__tar.gz → 0.3.0__tar.gz

{internal_web_reader-0.2.2 → internal_web_reader-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: internal-web-reader
-Version: 0.2.2
+Version: 0.3.0
 Summary: MCP Server - let Claude Code read internal web pages via browser cookies
 Author: zhanghaoran18006
 License-Expression: MIT
@@ -37,7 +37,7 @@ claude mcp add -s user internal-web-reader -- internal-web-reader
 在 Claude Code 里给链接：
 
 ```
-> 看看 https://doc.autohome.com.cn/docapi/page/share/share_1Hx8joJJIfo
+> 看看 https://wiki.yourcompany.com/pages/xxx
 ```
 
 第一次访问需要登录的网站，会弹出 Chrome 窗口。登录一下，之后就不用再登了。

{internal_web_reader-0.2.2 → internal_web_reader-0.3.0}/README.md

@@ -19,7 +19,7 @@ claude mcp add -s user internal-web-reader -- internal-web-reader
 在 Claude Code 里给链接：
 
 ```
-> 看看 https://doc.autohome.com.cn/docapi/page/share/share_1Hx8joJJIfo
+> 看看 https://wiki.yourcompany.com/pages/xxx
 ```
 
 第一次访问需要登录的网站，会弹出 Chrome 窗口。登录一下，之后就不用再登了。

{internal_web_reader-0.2.2 → internal_web_reader-0.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "internal-web-reader"
-version = "0.2.2"
+version = "0.3.0"
 description = "MCP Server - let Claude Code read internal web pages via browser cookies"
 readme = "README.md"
 requires-python = ">=3.10"

internal_web_reader-0.3.0/src/internal_web_reader/__main__.py

@@ -0,0 +1,54 @@
+"""MCP Server 入口。"""
+
+import atexit
+import logging
+
+from mcp.server.fastmcp import FastMCP
+
+from .reader import read_page, close_browser
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="[web-reader] %(message)s",
+    handlers=[logging.StreamHandler()],
+)
+
+mcp = FastMCP("internal-web-reader")
+
+
+@mcp.tool()
+async def read_web_page(url: str) -> str:
+    """[READ WEB PAGES] Open any URL in a local Chrome browser and return Markdown content.
+    USE THIS TOOL for all web page reading, especially internal/intranet sites
+    (GitLab, Wiki, docs requiring VPN/SSO). Do NOT use the built-in Fetch tool for web pages.
+    First visit to a login-required site will pop up a browser window for you to log in.
+    Session is saved automatically for future visits.
+    """
+    page = await read_page(url)
+
+    lines = [f"# {page.title}", f"> {page.url}", "", page.content]
+    if 0 < len(page.links) <= 30:
+        lines += ["", "---", "## Links"]
+        for lk in page.links:
+            lines.append(f"- [{lk['text']}]({lk['href']})")
+    return "\n".join(lines)
+
+
+@mcp.tool()
+async def list_web_links(url: str) -> str:
+    """List all links on a web page for navigation."""
+    page = await read_page(url)
+    if not page.links:
+        return f"No links found on {url}."
+    lines = [f"# {page.title} - Links", f"> {len(page.links)} links", ""]
+    for lk in page.links:
+        lines.append(f"- [{lk['text']}]({lk['href']})")
+    return "\n".join(lines)
+
+
+def main():
+    mcp.run()
+
+
+if __name__ == "__main__":
+    main()

{internal_web_reader-0.2.2 → internal_web_reader-0.3.0}/src/internal_web_reader/reader.py

@@ -1,5 +1,6 @@
-"""用 Playwright 启动浏览器读网页。登录一次，session 自动保存。"""
+"""用 Playwright (async) 启动浏览器读网页。登录一次，session 自动保存。"""
 
+import asyncio
 import logging
 import os
 import sys
@@ -39,22 +40,21 @@ _pw = None
 
 
 def _stderr(msg: str):
-    """写 stderr，保证在 MCP stdio 模式下用户能看到提示。"""
     sys.stderr.write(f"\n{'='*60}\n{msg}\n{'='*60}\n")
     sys.stderr.flush()
 
 
-def _ensure_browser():
+async def _ensure_browser():
     global _context, _pw
     if _context is not None:
         return _context
 
-    from playwright.sync_api import sync_playwright
+    from playwright.async_api import async_playwright
 
-    _pw = sync_playwright().start()
+    _pw = await async_playwright().start()
     os.makedirs(_USER_DATA_DIR, exist_ok=True)
 
-    _context = _pw.chromium.launch_persistent_context(
+    _context = await _pw.chromium.launch_persistent_context(
         _USER_DATA_DIR,
         headless=False,
         channel="chrome",
@@ -65,80 +65,65 @@ def _ensure_browser():
     return _context
 
 
-def _is_login_page(page) -> bool:
-    url_lower = page.url.lower()
-    login_keywords = ["/login", "/signin", "/sso", "/cas", "/oauth", "/auth", "passport"]
-    if any(k in url_lower for k in login_keywords):
-        return True
-    if page.query_selector('input[type="password"]'):
-        return True
-    return False
+def _is_login_url(url: str) -> bool:
+    url_lower = url.lower()
+    keywords = ["/login", "/signin", "/sso", "/cas", "/oauth", "/auth", "passport"]
+    return any(k in url_lower for k in keywords)
 
 
-def _wait_for_login(page, timeout_ms: int = 300_000):
-    """
-    检测到登录页后，在页面上注入提示，并等待用户完成登录。
-    """
+async def _wait_for_login(page, timeout_s: int = 300):
     _stderr(
         "  需要登录！请切换到弹出的浏览器窗口完成登录。\n"
         "  登录成功后会自动继续，不用回来操作终端。"
     )
 
-    # 在页面上注入一个醒目的提示
-    page.evaluate("""() => {
-        const banner = document.createElement('div');
-        banner.id = '__iwr_login_hint__';
-        banner.style.cssText = 'position:fixed;top:0;left:0;right:0;z-index:999999;'
+    await page.evaluate("""() => {
+        const b = document.createElement('div');
+        b.id = '__iwr__';
+        b.style.cssText = 'position:fixed;top:0;left:0;right:0;z-index:999999;'
             + 'background:#ff6b35;color:white;padding:16px;text-align:center;'
             + 'font-size:18px;font-weight:bold;font-family:sans-serif;'
             + 'box-shadow:0 4px 12px rgba(0,0,0,0.3);';
-        banner.textContent = 'Internal Web Reader: Please log in here, then wait a few seconds...';
-        document.body.prepend(banner);
+        b.textContent = 'Internal Web Reader: Please log in here. Auto-continuing after login...';
+        document.body.prepend(b);
     }""")
 
-    # 等待：密码框消失 + URL 不再包含 login 关键词
-    import time
-    start = time.time()
-    while (time.time() - start) * 1000 < timeout_ms:
-        time.sleep(2)
+    elapsed = 0
+    while elapsed < timeout_s:
+        await asyncio.sleep(2)
+        elapsed += 2
+
         current_url = page.url.lower()
-        has_pw = page.query_selector('input[type="password"]')
-
-        # 如果 URL 不再含 login 关键词 且 没有密码框 → 登录成功
-        login_in_url = any(k in current_url for k in ["/login", "/signin", "/sso", "/cas", "/oauth", "passport"])
-        if not login_in_url and not has_pw:
-            # 移除提示
-            page.evaluate("""() => {
-                const el = document.getElementById('__iwr_login_hint__');
-                if (el) el.remove();
-            }""")
+        has_pw = await page.query_selector('input[type="password"]')
+
+        if not _is_login_url(current_url) and not has_pw:
+            await page.evaluate("() => { const e = document.getElementById('__iwr__'); if(e) e.remove(); }")
             _stderr("  登录成功！正在读取页面...")
-            page.wait_for_load_state("domcontentloaded", timeout=30_000)
+            await page.wait_for_load_state("domcontentloaded", timeout=30_000)
             return True
 
-    _stderr("  登录等待超时（5分钟），尝试继续读取当前页面...")
+    _stderr("  登录等待超时，尝试继续...")
     return False
 
 
-def read_page(url: str, timeout: float = 60) -> PageResult:
-    ctx = _ensure_browser()
-    page = ctx.pages[0] if ctx.pages else ctx.new_page()
+async def read_page(url: str, timeout: float = 60) -> PageResult:
+    ctx = await _ensure_browser()
+    page = ctx.pages[0] if ctx.pages else await ctx.new_page()
 
-    # 让浏览器窗口获得焦点
     try:
-        page.bring_to_front()
+        await page.bring_to_front()
     except Exception:
         pass
 
-    page.goto(url, wait_until="domcontentloaded", timeout=timeout * 1000)
+    await page.goto(url, wait_until="domcontentloaded", timeout=timeout * 1000)
 
     # 登录检测
-    if _is_login_page(page):
-        _wait_for_login(page)
+    if _is_login_url(page.url) or await page.query_selector('input[type="password"]'):
+        await _wait_for_login(page)
 
     final_url = page.url
-    html = page.content()
-    title = page.title() or url
+    html = await page.content()
+    title = await page.title() or url
 
     soup = BeautifulSoup(html, "html.parser")
 
@@ -162,16 +147,16 @@ def read_page(url: str, timeout: float = 60) -> PageResult:
     return PageResult(url=final_url, title=title, content=markdown, links=links)
 
 
-def close_browser():
+async def close_browser():
     global _context, _pw
     if _context:
         try:
-            _context.close()
+            await _context.close()
         except Exception:
             pass
     if _pw:
         try:
-            _pw.stop()
+            await _pw.stop()
         except Exception:
             pass
     _context = None

{internal_web_reader-0.2.2 → internal_web_reader-0.3.0}/src/internal_web_reader.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: internal-web-reader
-Version: 0.2.2
+Version: 0.3.0
 Summary: MCP Server - let Claude Code read internal web pages via browser cookies
 Author: zhanghaoran18006
 License-Expression: MIT
@@ -37,7 +37,7 @@ claude mcp add -s user internal-web-reader -- internal-web-reader
 在 Claude Code 里给链接：
 
 ```
-> 看看 https://doc.autohome.com.cn/docapi/page/share/share_1Hx8joJJIfo
+> 看看 https://wiki.yourcompany.com/pages/xxx
 ```
 
 第一次访问需要登录的网站，会弹出 Chrome 窗口。登录一下，之后就不用再登了。

internal_web_reader-0.2.2/src/internal_web_reader/__main__.py

@@ -1,56 +0,0 @@
-"""MCP Server 入口。用 Playwright 浏览器读网页，登录一次自动保存。"""
-
-import atexit
-import logging
-
-from mcp.server.fastmcp import FastMCP
-
-from .reader import read_page, close_browser
-
-logging.basicConfig(
-    level=logging.INFO,
-    format="[web-reader] %(message)s",
-    handlers=[logging.StreamHandler()],
-)
-log = logging.getLogger("internal_web_reader")
-
-atexit.register(close_browser)
-
-mcp = FastMCP("internal-web-reader")
-
-
-@mcp.tool()
-def read_web_page(url: str) -> str:
-    """【读网页用这个】通过本地浏览器打开任意 URL 并返回 Markdown 内容。
-    专门用于访问公司内网页面（GitLab、Wiki、文档中心等需要 VPN 的网站）。
-    也适用于任何公开网页。首次访问需登录的站点会弹出浏览器窗口完成登录，
-    之后 session 自动保存。优先使用此工具而非内置 Fetch 来读取网页。
-    """
-    page = read_page(url)
-
-    lines = [f"# {page.title}", f"> {page.url}", "", page.content]
-    if 0 < len(page.links) <= 30:
-        lines += ["", "---", "## 页面链接"]
-        for lk in page.links:
-            lines.append(f"- [{lk['text']}]({lk['href']})")
-    return "\n".join(lines)
-
-
-@mcp.tool()
-def list_web_links(url: str) -> str:
-    """列出网页上的所有链接，用于浏览内网站点结构。"""
-    page = read_page(url)
-    if not page.links:
-        return f"页面 {url} 上未找到链接。"
-    lines = [f"# {page.title} 的链接", f"> {len(page.links)} 个", ""]
-    for lk in page.links:
-        lines.append(f"- [{lk['text']}]({lk['href']})")
-    return "\n".join(lines)
-
-
-def main():
-    mcp.run()
-
-
-if __name__ == "__main__":
-    main()