Added support for <meta http-equiv="Content-Type" content="..."> in inject_meta_charset filter (Python only) along with a few tests.

tbroyer · tbroyer · commit 0dbe1542e9b5 · 2007-06-20T14:29:11.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40782
diff --git a/src/filters/inject_meta_charset.py b/src/filters/inject_meta_charset.py
@@ -18,11 +18,22 @@ def __iter__(self):
 
             elif type == "EmptyTag":
                 if token["name"].lower() == "meta":
-                   # replace charset with actual encoding
-                   for i,(name,value) in enumerate(token["data"]):
-                       if name == 'charset':
-                          token["data"][i] = (token["data"][i][0], self.encoding)
-                          meta_found = True
+                   # replace charset with actual encoding
+                   has_http_equiv_content_type = False
+                   content_index = -1
+                   for i,(name,value) in enumerate(token["data"]):
+                       if name.lower() == 'charset':
+                          token["data"][i] = (u'charset', self.encoding)
+                          meta_found = True
+                          break
+                       elif name == 'http-equiv' and value.lower() == 'content-type':
+                           has_http_equiv_content_type = True
+                       elif name == 'content':
+                           content_index = i
+                   else:
+                       if has_http_equiv_content_type and content_index >= 0:
+                           token["data"][content_index] = (u'content', u'text/html; charset=%s' % self.encoding)
+                           meta_found = True
 
                 elif token["name"].lower() == "head" and not meta_found:
                     # insert meta into empty head
