internal-web-reader 0.2.1__tar.gz → 0.2.2__tar.gz

{internal_web_reader-0.2.1 → internal_web_reader-0.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: internal-web-reader
-Version: 0.2.1
+Version: 0.2.2
 Summary: MCP Server - let Claude Code read internal web pages via browser cookies
 Author: zhanghaoran18006
 License-Expression: MIT

{internal_web_reader-0.2.1 → internal_web_reader-0.2.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "internal-web-reader"
-version = "0.2.1"
+version = "0.2.2"
 description = "MCP Server - let Claude Code read internal web pages via browser cookies"
 readme = "README.md"
 requires-python = ">=3.10"

{internal_web_reader-0.2.1 → internal_web_reader-0.2.2}/src/internal_web_reader/reader.py

@@ -2,9 +2,8 @@
 
 import logging
 import os
-import re
+import sys
 from dataclasses import dataclass, field
-from pathlib import Path
 from urllib.parse import urljoin
 
 from bs4 import BeautifulSoup
@@ -20,7 +19,6 @@ _H2T.protect_links = True
 _H2T.wrap_links = False
 _H2T.unicode_snob = True
 
-# 持久化浏览器 profile 目录
 _USER_DATA_DIR = os.path.join(
     os.environ.get("APPDATA", os.path.expanduser("~")),
     "internal-web-reader",
@@ -36,15 +34,18 @@ class PageResult:
     links: list[dict] = field(default_factory=list)
 
 
-# 全局浏览器实例（进程内复用）
-_browser = None
 _context = None
 _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():
-    """启动 Playwright 浏览器（进程内单例）。"""
-    global _browser, _context, _pw
+    global _context, _pw
     if _context is not None:
         return _context
 
@@ -61,58 +62,89 @@ def _ensure_browser():
         locale="zh-CN",
         args=["--disable-blink-features=AutomationControlled"],
     )
-    log.info("浏览器已启动")
     return _context
 
 
 def _is_login_page(page) -> bool:
-    """判断当前页面是否是登录页。"""
-    url = page.url.lower()
-    if any(k in url for k in ("/login", "/signin", "/sso", "/auth", "/cas/", "/oauth")):
+    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
-    # 检查页面是否有密码输入框
-    pw_inputs = page.query_selector_all('input[type="password"]')
-    if pw_inputs and len(pw_inputs) > 0:
+    if page.query_selector('input[type="password"]'):
         return True
     return False
 
 
+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。
+    """
+    _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;'
+            + '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);
+    }""")
+
+    # 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)
+        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();
+            }""")
+            _stderr("  登录成功！正在读取页面...")
+            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...")
+    return False
+
+
 def read_page(url: str, timeout: float = 60) -> PageResult:
-    """用浏览器打开 URL，返回清洗后的 Markdown。"""
     ctx = _ensure_browser()
-
-    # 用现有 page 或新建
     page = ctx.pages[0] if ctx.pages else ctx.new_page()
 
+    # 让浏览器窗口获得焦点
+    try:
+        page.bring_to_front()
+    except Exception:
+        pass
+
     page.goto(url, wait_until="domcontentloaded", timeout=timeout * 1000)
 
-    # 如果是登录页，等用户手动登录
+    # 登录检测
     if _is_login_page(page):
-        log.info("检测到登录页面，请在浏览器窗口中完成登录...")
-        # 等 URL 变化（登录成功后会跳转）或者最多等 120 秒
-        try:
-            page.wait_for_url(
-                lambda u: u != page.url and "login" not in u.lower(),
-                timeout=120_000,
-            )
-            # 等页面加载完
-            page.wait_for_load_state("domcontentloaded", timeout=30_000)
-            log.info(f"登录成功，当前页面: {page.url}")
-        except Exception:
-            log.warning("等待登录超时，尝试继续...")
+        _wait_for_login(page)
 
-    # 获取页面内容
     final_url = page.url
     html = page.content()
     title = page.title() or url
 
     soup = BeautifulSoup(html, "html.parser")
 
-    # 去噪
     for tag in soup.find_all(["script", "style", "noscript", "iframe", "svg", "nav", "footer", "aside"]):
         tag.decompose()
 
-    # 找主体
     main = None
     for sel in ("article", "main", '[role="main"]', ".content", ".article-content", ".markdown-body", "#content", "#main"):
         el = soup.select_one(sel)
@@ -125,14 +157,13 @@ def read_page(url: str, timeout: float = 60) -> PageResult:
     markdown = _H2T.handle(str(target)).strip()
 
     if len(markdown) > 200_000:
-        markdown = markdown[:200_000] + "\n\n... [已截断]"
+        markdown = markdown[:200_000] + "\n\n... [truncated]"
 
     return PageResult(url=final_url, title=title, content=markdown, links=links)
 
 
 def close_browser():
-    """关闭浏览器（进程退出时调用）。"""
-    global _browser, _context, _pw
+    global _context, _pw
     if _context:
         try:
             _context.close()

{internal_web_reader-0.2.1 → internal_web_reader-0.2.2}/src/internal_web_reader.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: internal-web-reader
-Version: 0.2.1
+Version: 0.2.2
 Summary: MCP Server - let Claude Code read internal web pages via browser cookies
 Author: zhanghaoran18006
 License-Expression: MIT