Workaround for pydoc generated module link ending with `.html`

[Christian Boos]

... and this is annoying since we can't just truncate a trailing .html in load_object(), as that method takes a dot-separated module path.

The solution would be to clean up the generated title HTML.

Example:

tracpydoc/tracpydoc.py

@@ -17 +17 @@
 class TracDoc(pydoc.HTMLDoc):
 
     _cleanup_re = re.compile(r'(?:bg)?color="[^"]+"')
+    _cleanup_heading_re = re.compile(r'href="([^"]+).html"')
 
     def __init__(self, env):
         self.env = env
@@ -45 +46 @@
         return '.'.join(links)
 
     def heading(self, *args):
-        return self._cleanup('heading', *args)
+        return re.sub(self._cleanup_heading_re, r'href="\1"',
+                      self._cleanup('heading', *args))
 
     def section(self, *args):
         return self._cleanup('section', *args)

(this is on top of my previous patch, posted on the Trac ML, but you see the kind of change needed).

[Alec Thomas]

Thanks Christian, I've applied both of these patches. Out of curiosity, where are you seeing the spurious .html's?