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

[hybride] improve browser2 adaptation

Browse source
This commit is contained in:
Bezleputh 2014-03-13 19:23:32 +01:00 • committed by Romain Bignon
commit 9cd521bfde
3 changed files with 35 additions and 66 deletions
Download patch file Download diff file Expand all files Collapse all files

82
modules/hybride/pages.py
View file

@ -46,19 +46,25 @@ class CombineDate(Filter):
return datetime.combine(format_date(text), time.max)
class Description(Filter):
def filter(self, el):
description = ''
description_intro = el[0].xpath("div[@class='itemIntroText']/table/tbody/tr/td")
if description_intro and len(description_intro) > 0:
description += u'%s' % description_intro[0].text_content()
description_content = el[0].xpath("div[@class='itemFullText']/table/tbody/tr/td")
if description_content and len(description_content) > 0:
description += u'%s' % description_content[0].text_content()
return u'%s' % description
class ProgramPage(HTMLPage):
date_from = None
date_to = None
city = None
categories = None
def set_filters(self, date_from, date_to, city, categories):
self.date_from = date_from
self.date_to = date_to
self.city = city
self.categories = categories
@method
class list_events(ListElement):
item_xpath = '//div[@class="catItemView groupLeading"]'
@ -66,27 +72,23 @@ class ProgramPage(HTMLPage):
class item(ItemElement):
klass = HybrideCalendarEvent
def condition(self):
return self.check_date() and self.check_city() and self.check_category()
def validate(self, obj):
return self.check_date(obj) and self.check_city(obj) and self.check_category(obj)
def check_date(self):
date = self.el.xpath("div[@class='catItemHeader']/span[@class='catItemDateCreated']")[0]
event_date = format_date(date.text)
if self.page.date_from and event_date >= self.page.date_from:
if not self.page.date_to:
def check_date(self, obj):
if self.env['date_from'] and obj.start_date > self.env['date_from']:
if not self.env['date_to']:
return True
else:
if event_date <= self.page.date_to:
if obj.end_date < self.env['date_to']:
return True
return False
def check_city(self):
return (not self.page.city or (self.page.city and
self.page.city.upper() == HybrideCalendarEvent.get_city().upper())
)
def check_city(self, obj):
return (not self.env['city'] or self.env['city'].upper() == obj.city.upper())
def check_category(self):
return (not self.page.categories or HybrideCalendarEvent.get_category() in self.page.categories)
def check_category(self, obj):
return (not self.env['categories'] or obj.category in self.env['categories'])
class CheckId(Filter):
def filter(self, a_id):
@ -100,8 +102,6 @@ class ProgramPage(HTMLPage):
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')
obj_city = HybrideCalendarEvent.get_city()
obj_category = HybrideCalendarEvent.get_category()
class EventPage(HTMLPage):
@ -111,40 +111,18 @@ class EventPage(HTMLPage):
klass = HybrideCalendarEvent
def parse(self, el):
div = el.xpath("//div[@class='itemView']")[0]
if self.obj.id:
event = self.obj
event.url = self.page.url
event.description = self.get_description(div)
event.description = Description('//div[@class="itemView"]/div[@class="itemBody"]')(self)
raise SkipItem()
re_id = re.compile('http://www.lhybride.org/programme/item/(.*?)', re.DOTALL)
self.env['id'] = re_id.search(self.page.url).group(1)
self.env['url'] = self.page.url
self.env['description'] = self.get_description(div)
def get_description(self, div):
description = ''
description_intro = div.xpath("div[@class='itemBody']/div[@class='itemIntroText']/table/tbody/tr/td")
if description_intro and len(description_intro) > 0:
description += u'%s' % description_intro[0].text_content()
description_content = div.xpath("div[@class='itemBody']/div[@class='itemFullText']/table/tbody/tr/td")
if description_content and len(description_content) > 0:
description += u'%s' % description_content[0].text_content()
return u'%s' % description
obj_id = Env('id')
obj_id = Env('_id')
base = '//div[@class="itemView"]/div[@class="itemHeader"]'
obj_start_date = Date(CleanText('%s/span[@class="itemDateCreated"]' % base))
obj_end_date = CombineDate(CleanText('%s/span[@class="itemDateCreated"]' % base))
obj_summary = CleanText('%s/h2[@class="itemTitle"]' % base)
obj_city = HybrideCalendarEvent.get_city()
obj_category = HybrideCalendarEvent.get_category()
obj_url = Env('url')
obj_description = Env('description')
obj_description = Description('//div[@class="itemView"]/div[@class="itemBody"]')