| 4 |
|
# |
| 5 |
|
# $Id$ |
| 6 |
|
|
| 7 |
< |
from BeautifulSoup import BeautifulSoup |
| 7 |
> |
from BeautifulSoup import MinimalSoup |
| 8 |
> |
import copy |
| 9 |
|
import mechanize |
| 10 |
+ |
import re |
| 11 |
|
|
| 12 |
|
# XXX: hack to make Wells Fargo http-equiv redirects actually work |
| 13 |
|
mechanize._http.AbstractHeadParser.head_elems = tuple(list(mechanize._http.AbstractHeadParser.head_elems) + ['body']) |
| 14 |
|
|
| 15 |
+ |
class DuckSoup(MinimalSoup): |
| 16 |
+ |
MARKUP_MASSAGE = copy.copy(MinimalSoup.MARKUP_MASSAGE) |
| 17 |
+ |
|
| 18 |
+ |
MARKUP_MASSAGE.append((re.compile(r'''</?[a-z]+['"]\+['"][a-z]+'''), lambda match: '')) |
| 19 |
+ |
|
| 20 |
|
class Website(object): |
| 21 |
< |
Soup = BeautifulSoup |
| 21 |
> |
Soup = DuckSoup |
| 22 |
|
|
| 23 |
|
def __init__(self, debug): |
| 24 |
|
self.browser = mechanize.Browser() |
