Minor patch

stamparm · stamparm · commit c19a283434ad · 2013-03-26T20:06:50.000+01:00
diff --git a/lib/request/basic.py b/lib/request/basic.py
@@ -254,7 +254,14 @@ def decodePage(page, contentEncoding, contentType):
 
         # e.g. &#8217;&#8230;&#8482;
         if "&#" in page:
-            page = re.sub(r"&#(\d+);", lambda _: unichr(int(_.group(1))), page)
+            def _(match):
+                retVal = match.group(0)
+                try:
+                    retVal = unichr(int(match.group(1)))
+                except ValueError:
+                    pass
+                return retVal
+            page = re.sub(r"&#(\d+);", _, page)
 
         # e.g. &zeta;
         page = re.sub(r"&([^;]+);", lambda _: unichr(htmlEntities[_.group(1)]) if htmlEntities.get(_.group(1), 0) > 255 else _.group(0), page)
