nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[Apec][Adecco]use html2text to fill description content

Browse source
This commit is contained in:
Bezleputh 2013-12-13 12:27:36 +01:00
commit 06310579d6
2 changed files with 6 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
modules/adecco/pages.py
View file

@ -19,6 +19,7 @@
from weboob.tools.browser import BasePage from weboob.tools.browser import BasePage
from weboob.tools.misc import html2text
from .job import AdeccoJobAdvert from .job import AdeccoJobAdvert
import datetime import datetime
import re import re
@ -84,5 +85,6 @@ class AdvertPage(BasePage):
advert.pay = u'%s' % spans[2].text advert.pay = u'%s' % spans[2].text
advert.contract_type = u'%s' % spans[3].text advert.contract_type = u'%s' % spans[3].text
advert.url = url advert.url = url
advert.description = self.document.getroot().xpath("//div[@class='descriptionContainer']/p")[0].text_content() description = self.document.getroot().xpath("//div[@class='descriptionContainer']/p")[0]
advert.description = html2text(self.parser.tostring(description))
return advert return advert

4
modules/apec/pages.py
View file

@ -19,6 +19,7 @@
from weboob.tools.browser import BasePage from weboob.tools.browser import BasePage
from weboob.tools.misc import html2text
import dateutil.parser import dateutil.parser
import re import re
@ -52,7 +53,8 @@ class AdvertPage(BasePage):
advert = ApecJobAdvert(_id) advert = ApecJobAdvert(_id)
advert.title = re_id_title.search(url).group(2).replace('-', ' ') advert.title = re_id_title.search(url).group(2).replace('-', ' ')
advert.description = self.document.getroot().xpath("//div[@class='contentWithDashedBorderTop marginTop boxContent']/div")[0].text_content() description = self.document.getroot().xpath("//div[@class='contentWithDashedBorderTop marginTop boxContent']/div")[0]
advert.description = html2text(self.parser.tostring(description))
advert.job_name = advert.title advert.job_name = advert.title