- _charset = root.originalEncoding
- try:
- title = head.title.string.encode(_charset)
- except AttributeError:
- title = '' # HEAD but no TITLE
+ title = head.title
+ if (title is None) and (html is not head):
+ # Some sites put TITLE in HTML outside of HEAD
+ title = html.title
+
+ if title is None:
+ # Lookup TITLE in the root
+ title = root.title
+
+ if title is None:
+ return None
+
+ if title.string:
+ title = title.string
+ else:
+ parts = []
+ for part in title:
+ if not isinstance(part, basestring):
+ part = unicode(part)
+ parts.append(part.strip())
+ title = ''.join(parts)
+
+ meta = head.find(_find_contenttype, recursive=False)
+ if meta:
+ try:
+ meta_content = meta.get("content")
+ if meta_content:
+ __charset = meta_content.lower().split('charset=')[1].split(';')[0]
+ else:
+ __charset = False
+ except IndexError: # No charset in the META Content-Type
+ meta_charset = False
+ else:
+ meta_charset = _charset == __charset
+ else:
+ meta_charset = False
+
+ if _charset or meta_charset:
+ title = title.encode(_charset or meta_charset)