|
@@ -24,6 +24,7 @@ BS_HTML_BODY_PARSE_CONFIG = {
|
24
|
24
|
'tag_blacklist': ["script", "style", "blockquote"],
|
25
|
25
|
'class_blacklist': ['moz-cite-prefix','gmail_extra','gmail_quote',
|
26
|
26
|
'yahoo_quoted'],
|
|
27
|
+ 'id_blacklist': ['reply-intro'],
|
27
|
28
|
'tag_whitelist': ['a', 'b', 'strong', 'i', 'br', 'ul', 'li', 'ol',
|
28
|
29
|
'em', 'i', 'u',
|
29
|
30
|
'thead', 'tr', 'td', 'tbody', 'table', 'p', 'pre'],
|
|
@@ -94,6 +95,10 @@ class DecodedMail(object):
|
94
|
95
|
for elem in config['class_blacklist']:
|
95
|
96
|
if elem in tag.attrs['class']:
|
96
|
97
|
tag.extract()
|
|
98
|
+ elif 'id' in tag.attrs:
|
|
99
|
+ for elem in config['id_blacklist']:
|
|
100
|
+ if elem in tag.attrs['id']:
|
|
101
|
+ tag.extract()
|
97
|
102
|
elif tag.name.lower() in config['tag_whitelist']:
|
98
|
103
|
attrs = dict(tag.attrs)
|
99
|
104
|
for attr in attrs:
|