Fix regexp

This commit is contained in:
Florent Fourcot 2013-07-04 21:12:32 +02:00
commit bbd4a0f357
2 changed files with 3 additions and 3 deletions

View file

@ -33,7 +33,7 @@ class AdeccoBrowser(BaseBrowser):
PAGES = {
'%s://%s/trouver-un-emploi/Pages/Offres-d-emploi.aspx\?keywords=(.*?)' % (PROTOCOL, DOMAIN): SearchPage,
'%s://%s/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)?$' % (PROTOCOL, DOMAIN): AdvertPage,
'%s://%s/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)$' % (PROTOCOL, DOMAIN): AdvertPage,
}
def search_job(self, pattern):

View file

@ -30,7 +30,7 @@ __all__ = ['SearchPage', 'AdvertPage']
class SearchPage(BasePage):
def iter_job_adverts(self):
locale.setlocale(locale.LC_TIME, 'fr_FR.UTF-8')
re_id = re.compile('http://www.adecco.fr/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)?$', re.DOTALL)
re_id = re.compile('http://www.adecco.fr/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)$', re.DOTALL)
adverts = []
@ -55,7 +55,7 @@ class SearchPage(BasePage):
class AdvertPage(BasePage):
def get_job_advert(self, url, advert):
re_id = re.compile('http://www.adecco.fr/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)?$', re.DOTALL)
re_id = re.compile('http://www.adecco.fr/trouver-un-emploi/Pages/Details-de-l-Offre/(.*?)/(.*?).aspx\?IOF=(.*?)$', re.DOTALL)
if advert is None:
_id = u'%s/%s/%s' % (re_id.search(url).group(1), re_id.search(url).group(2), re_id.search(url).group(3))
advert = AdeccoJobAdvert(_id)