return j
-def parse_html(filename, charset=None):
+def parse_html(filename, charset=None, log=None):
infile = open(filename, 'r')
try:
root = BadDeclParser(infile, fromEncoding=charset)
except TypeError:
+ if log: log("TypeError")
return None
finally:
infile.close()
try:
head = root.html.head
except AttributeError:
+ if log: log("No HTML in root or no HEAD in HTML")
return None
if head is None:
meta = head.find(_find_contenttype, recursive=False)
if meta:
- __charset = meta.get("content").lower().split('charset=')[1].split(';')[0]
- meta_charset = _charset == __charset
+ try:
+ __charset = meta.get("content").lower().split('charset=')[1].split(';')[0]
+ except IndexError: # No charset in the META Content-Type
+ meta_charset = False
+ else:
+ meta_charset = _charset == __charset
else:
meta_charset = False