cppmanlite 0.1.1__tar.gz → 0.1.6__tar.gz

{cppmanlite-0.1.1 → cppmanlite-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cppmanlite
-Version: 0.1.1
+Version: 0.1.6
 Summary: Lightweight serverless C++ documentation lookup — pure Python, no C deps
 License-Expression: MIT
 Project-URL: Homepage, https://dive4dec.github.io/cppmanlite/

{cppmanlite-0.1.1 → cppmanlite-0.1.6}/cppmanlite/core.py

@@ -150,7 +150,7 @@ def list_pages(limit: int = 0) -> list[dict[str, str]]:
 # --------------------------------------------------------------------------- #
 
 _CONTENT_RE = re.compile(
-    r'<div id="mw-content-text"[^>]*>(.*?)(?:</div>\s*<!--|\Z)',
+    r'<div class="mw-content-ltr mw-parser-output"[^>]*>(.*?)(?:</div>\s*<!--|\Z)',
     re.DOTALL,
 )
 _SCRIPT_RE = re.compile(r"<script[^>]*>.*?</script>", re.DOTALL)
@@ -349,12 +349,64 @@ def _format_search_html(results: list[dict[str, str]]) -> str:
     )
 
 
-def _format_page_html(content: str) -> str:
+def _format_page_html(content: str, page_url: str = "") -> str:
+    """Format page content for Jupyter display with working links.
+
+    In a **trusted** notebook, Jupyter renders external https:// links
+    with target=\"_blank\" — clicking opens cppreference.com in a new tab.
+    In an untrusted notebook, the sanitizer strips external hrefs to '#'.
+
+    Links:
+    - Navigation links → absolute https://en.cppreference.com/w/... + target=\"_blank\"
+    - Edit links (<a href=\".../index.php?...action=edit\">) → unwrapped (text kept, <a> removed)
+    - Anchor links (href=\"#...\") → left as-is (in-page navigation)
+    """
+    from urllib.parse import urljoin
+
+    base_href = (
+        f"https://en.cppreference.com/w/{page_url}" if page_url
+        else "https://en.cppreference.com/w/"
+    )
+
+    # 1. Remove edit links entirely: <a ... href=".../index.php?...action=edit...">text</a> → text
+    content = re.sub(
+        r'<a [^>]*href="[^"]*index\.php[^"]*action=edit[^"]*"[^>]*>(.*?)</a>',
+        r'\1',
+        content,
+        flags=re.DOTALL,
+    )
+
+    # 2. Rewrite remaining hrefs
+    def _rewrite_href(m: re.Match) -> str:
+        href = m.group(1)
+
+        # Skip javascript: URLs — neutralize
+        if href.startswith("javascript:"):
+            return 'href="#" onclick="return false"'
+
+        # Skip anchors (in-page navigation) — keep as-is
+        if href.startswith("#"):
+            return f'href="{href}"'
+
+        # Resolve to absolute URL if needed
+        if not href.startswith("http://") and not href.startswith("https://"):
+            href = urljoin(base_href, href)
+
+        return f'href="{html.escape(href)}" target="_blank" rel="noopener"'
+
+    content = re.sub(r'href="([^"]*)"', _rewrite_href, content)
+
     return (
-        '<div style="max-height:600px;overflow:auto;'
+        '<div class="cppmanlite-content" '
+        'style="max-height:600px;overflow:auto;'
         'border:1px solid #ddd;padding:16px;font-size:14px">'
         + content
-        + "</div>"
+        + '</div>'
+        + '<p style="font-size:11px;color:#888;margin-top:4px">'
+        + 'Links open cppreference.com in a new tab. '
+        + 'If links don\'t work, trust this notebook: '
+        + '<b>File → Trust Notebook</b>'
+        + '</p>'
     )
 
 
@@ -395,7 +447,7 @@ def _man_sync(query: str) -> None:
     if _is_jupyter():
         from IPython.display import HTML, display
 
-        display(HTML(_format_page_html(content)))
+        display(HTML(_format_page_html(content, page_url=url)))
     else:
         # Terminal: print formatted text
         print(f"\n{'=' * 80}\n{title}\n{'=' * 80}\n")
@@ -414,7 +466,7 @@ async def _man_async(query: str) -> None:
     if _is_jupyter():
         from IPython.display import HTML, display
 
-        display(HTML(_format_page_html(content)))
+        display(HTML(_format_page_html(content, page_url=url)))
     else:
         # Pyodide console / terminal
         print(f"\n{'=' * 80}\n{title}\n{'=' * 80}\n")

{cppmanlite-0.1.1 → cppmanlite-0.1.6}/cppmanlite.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cppmanlite
-Version: 0.1.1
+Version: 0.1.6
 Summary: Lightweight serverless C++ documentation lookup — pure Python, no C deps
 License-Expression: MIT
 Project-URL: Homepage, https://dive4dec.github.io/cppmanlite/

{cppmanlite-0.1.1 → cppmanlite-0.1.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "cppmanlite"
-version = "0.1.1"
+version = "0.1.6"
 description = "Lightweight serverless C++ documentation lookup — pure Python, no C deps"
 readme = "README.md"
 license = "MIT"