class item(ItemElement):
def klass(self):
return BiplanCalendarEventConcert() if self.env['is_concert'] else BiplanCalendarEventTheatre()
def condition(self):
return (self.el.xpath('./div') and CleanText('./div/a/img/@src')(self)[-1] != '/')
def validate(self, obj):
return (self.is_valid_event(obj, self.env['city'], self.env['categories']) and
self.is_event_in_valid_period(obj.start_date, self.env['date_from'], self.env['date_to']))
def is_valid_event(self, event, city, categories):
if city and city != '' and city.upper() != event.city.upper():
return False
if categories and len(categories) > 0 and event.category not in categories:
return False
return True
def is_event_in_valid_period(self, event_date, date_from, date_to):
if event_date >= date_from:
if not date_to:
return True
else:
if event_date <= date_to:
return True
return False
obj_id = Regexp(Link('./div/a'), '/(.*?).html')
obj_start_date = CombineDate(BiplanDate('div/div/b'), StartTime('div/div/b'))
obj_end_date = CombineDate(BiplanDate('div/div/b'), EndTime('.'))
obj_price = BiplanPrice('div/div/b')
obj_summary = CleanText("div/div/div/a/strong")
class get_event(ItemElement):
klass = BiplanCalendarEventConcert if Env(
'is_concert') else BiplanCalendarEventTheatre
def parse(self, el):
_div = "//div/div/div[@id='popup']"
div = el.xpath("%s" % _div)[0]
if self.obj.id:
event = self.obj
event.url = self.page.url
event.description = CleanHTML(
"%s/div/div[@class='presentation-popup']" % _div)(self)
raise SkipItem()
self.env['is_concert'] = (div.attrib['class'] != 'theatre-popup')
self.env['url'] = self.page.url
obj_id = Env('_id')
base = "//div[@id='popup']"
obj_price = BiplanPrice("%s/div/b" % base)
obj_start_date = CombineDate(BiplanDate("%s/div/b" % base),
StartTime("%s/div/b" % base))
obj_end_date = CombineDate(BiplanDate("%s/div/b" % base), EndTime("."))
obj_url = Env('url')
obj_summary = CleanText('%s/div/div/span' % base)
obj_description = CleanHTML('%s/div/div[@class="presentation-popup"]' %
base)
def obj_end_date(self):
end_time = Time(
Regexp(CleanText('//aside[@id="detail"]/ul/li[1]'),
r'- (\d{2}:\d{2})'))(self)
end_date = CombineDate(self._date, end_time)(self)
if end_date > self.obj_start_date():
end_date += timedelta(days=1)
return end_date
class get_event(ItemElement):
klass = AgendaDuLibreCalendarEvent
def parse(self, el):
self.env['url'] = self.page.url
obj_id = Env('_id')
obj_url = Env('url')
obj_summary = CleanText('//meta[@property="DC:title"]/@content')
obj_description = CleanHTML('//div[@class="description"]')
obj_location = CleanText('//p[@class="full_address"]/span[1]')
obj_city = CleanText('//meta[@property="geo:placename"]/@content')
obj_start_date = DateTime(CleanText('//meta[@property="DC:date"]/@content'))
obj_end_date = CombineDate(DateTime(CleanText('//meta[@property="DC:date"]/@content')),
EventEndDate('.'))
class get_event(ItemElement):
klass = RazibusCalendarEvent
obj_id = Env('_id')
obj_summary = CleanText('//h2[@itemprop="name"]')
obj_start_date = DateTime(
CleanText('//span[@itemprop="startDate"]/time/@datetime'))
obj_end_date = CombineDate(
DateTime(
CleanText('//span[@itemprop="startDate"]/time/@datetime')),
EndTime('.'))
obj_location = CleanText(
'//meta[@property="og:street-address"]/@content')
obj_city = CleanText('//meta[@property="og:locality"]/@content')
obj_url = CleanText('//meta[@property="og:url"]/@content')
obj_description = CleanHTML('//div[@itemprop="description"]')
class item(ItemElement):
klass = RazibusCalendarEvent
def validate(self, obj):
return (self.is_valid_event(obj, self.env['city'],
self.env['categories'])
and self.is_event_in_valid_period(
obj.start_date, self.env['date_from'],
self.env['date_to']))
def is_valid_event(self, event, city, categories):
if city and city != '' and city.upper() != event.city.upper():
return False
if categories and len(
categories) > 0 and event.category not in categories:
return False
return True
def is_event_in_valid_period(self, event_date, date_from, date_to):
if event_date >= date_from:
if not date_to:
return True
else:
if event_date <= date_to:
return True
return False
obj_id = Regexp(Link('./p/strong/a[@itemprop="url"]'),
'http://razibus.net/(.*).html')
obj_summary = CleanText('./p/strong/a[@itemprop="url"]')
obj_start_date = DateTime(
CleanText('./p/span[@itemprop="startDate"]/@content'))
obj_end_date = CombineDate(
DateTime(
CleanText('./p/span[@itemprop="startDate"]/@content')),
EndTime('.'))
obj_location = CleanText('./p/span[@itemprop="location"]/@content')
obj_city = CleanText('./p/span[@itemprop="location"]')
def obj_start_date(self):
start_time = Time(
Regexp(CleanText('//aside[@id="detail"]/ul/li[1]'),
r'(\d{2}:\d{2}) -'))(self)
return CombineDate(self._date, start_time)(self)