force UTF-8 encoding with the standard mechanism
This commit is contained in:
parent
7f10865215
commit
8798312e04
1 changed files with 1 additions and 8 deletions
|
|
@ -19,8 +19,6 @@
|
|||
|
||||
|
||||
import calendar
|
||||
from StringIO import StringIO
|
||||
import lxml.html as html
|
||||
from datetime import datetime
|
||||
from decimal import Decimal
|
||||
|
||||
|
|
@ -38,12 +36,7 @@ class FormatDate(Filter):
|
|||
|
||||
|
||||
class BadUTF8Page(HTMLPage):
|
||||
def __init__(self, browser, response, *args, **kwargs):
|
||||
# XXX it is volontary the parent class of HTMLPage's constructor which
|
||||
# is called, but that's ugly.
|
||||
super(HTMLPage, self).__init__(browser, response, *args, **kwargs)
|
||||
parser = html.HTMLParser(encoding='UTF-8')
|
||||
self.doc = html.parse(StringIO(response.content), parser)
|
||||
ENCODING = 'UTF-8'
|
||||
|
||||
|
||||
class DetailsPage(LoggedPage, BadUTF8Page):
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue