[hybride] site changed

This commit is contained in:
Bezleputh 2014-12-09 10:19:13 +01:00 committed by Florent
commit 5945d5d539
2 changed files with 10 additions and 29 deletions

View file

@ -29,8 +29,8 @@ class HybrideBrowser(PagesBrowser):
PROFILE = Firefox() PROFILE = Firefox()
BASEURL = 'http://www.lhybride.org' BASEURL = 'http://www.lhybride.org'
program_page = URL('/programme.html', ProgramPage) program_page = URL('programme.html', ProgramPage)
event_page = URL('/programme/item/(?P<_id>.*)', EventPage) event_page = URL('programme/item/(?P<_id>.*)', EventPage)
def list_events(self, date_from, date_to=None, city=None, categories=None): def list_events(self, date_from, date_to=None, city=None, categories=None):
return self.program_page.stay_or_go().list_events(date_from=date_from, return self.program_page.stay_or_go().list_events(date_from=date_from,

View file

@ -17,15 +17,13 @@
# You should have received a copy of the GNU Affero General Public License # You should have received a copy of the GNU Affero General Public License
# along with weboob. If not, see <http://www.gnu.org/licenses/>. # along with weboob. If not, see <http://www.gnu.org/licenses/>.
from datetime import time, datetime
from .calendar import HybrideCalendarEvent from .calendar import HybrideCalendarEvent
import weboob.tools.date as date_util import weboob.tools.date as date_util
import re
from weboob.browser.pages import HTMLPage from weboob.browser.pages import HTMLPage
from weboob.browser.elements import ItemElement, SkipItem, ListElement, method from weboob.browser.elements import ItemElement, ListElement, method
from weboob.browser.filters.standard import Filter, CleanText, Env, Format, BrowserURL from weboob.browser.filters.standard import Filter, CleanText, Env, Format, BrowserURL, Regexp
from weboob.browser.filters.html import CleanHTML from weboob.browser.filters.html import CleanHTML
from weboob.browser.filters.html import Link from weboob.browser.filters.html import Link
@ -35,16 +33,11 @@ class Date(Filter):
return date_util.parse_french_date(text) return date_util.parse_french_date(text)
class CombineDate(Filter):
def filter(self, text):
return datetime.combine(date_util.parse_french_date(text), time.max)
class ProgramPage(HTMLPage): class ProgramPage(HTMLPage):
@method @method
class list_events(ListElement): class list_events(ListElement):
item_xpath = '//div[@class="catItemView groupLeading"]' item_xpath = '//div[@class="itemContainer itemContainerLast"]'
class item(ItemElement): class item(ItemElement):
klass = HybrideCalendarEvent klass = HybrideCalendarEvent
@ -67,18 +60,9 @@ class ProgramPage(HTMLPage):
def check_category(self, obj): def check_category(self, obj):
return (not self.env['categories'] or obj.category in self.env['categories']) return (not self.env['categories'] or obj.category in self.env['categories'])
class CheckId(Filter): obj_id = Regexp(Link('div/div[@class="catItemHeader"]/h3[@class="catItemTitle"]/a'), '/programme/item/(.*?).html')
def filter(self, a_id): obj_start_date = Date(CleanText('div/div[@class="catItemHeader"]/span[@class="catItemDateCreated"]'))
re_id = re.compile('/programme/item/(.*?).html', re.DOTALL) obj_summary = CleanText('div/div[@class="catItemHeader"]/h3[@class="catItemTitle"]/a')
_id = re_id.search(a_id).group(1)
if _id:
return _id
raise SkipItem()
obj_id = CheckId(Link('div[@class="catItemHeader"]/h3[@class="catItemTitle"]/a'))
obj_start_date = Date(CleanText('div[@class="catItemHeader"]/span[@class="catItemDateCreated"]'))
obj_end_date = CombineDate(CleanText('div[@class="catItemHeader"]/span[@class="catItemDateCreated"]'))
obj_summary = CleanText('div[@class="catItemHeader"]/h3[@class="catItemTitle"]/a')
class EventPage(HTMLPage): class EventPage(HTMLPage):
@ -88,11 +72,8 @@ class EventPage(HTMLPage):
klass = HybrideCalendarEvent klass = HybrideCalendarEvent
obj_id = Env('_id') obj_id = Env('_id')
base = '//div[@class="itemView"]/div[@class="itemHeader"]' obj_start_date = Date(CleanText('//span[@class="itemDateCreated"]'))
obj_start_date = Date(CleanText('%s/span[@class="itemDateCreated"]' % base)) obj_summary = CleanText('//h2[@class="itemTitle"]')
obj_end_date = CombineDate(CleanText('%s/span[@class="itemDateCreated"]' % base))
obj_summary = CleanText('%s/h2[@class="itemTitle"]' % base)
obj_url = Env('url')
obj_description = Format('%s\n%s', obj_description = Format('%s\n%s',
CleanHTML('//div[@class="itemIntroText"]'), CleanHTML('//div[@class="itemIntroText"]'),
CleanHTML('//div[@class="itemFullText"]')) CleanHTML('//div[@class="itemFullText"]'))