def parse(self, el):
json_content = Regexp(CleanText('//script'),
"var ava_data = ({.+?});")(self)
json_content = json_content.replace("logged", "\"logged\"")
json_content = json_content.replace("lengthcarrousel",
"\"lengthcarrousel\"")
json_content = json_content.replace("products", "\"products\"")
json_content = json_content.replace(
"// // ANNONCES_SIMILAIRE / RECO", "")
self.house_json_datas = json.loads(json_content)['products'][0]
def obj_rdate(self):
s = Regexp(Field('raw'),
' (\d{2}/\d{2}/\d{2}) | (?!NUM) (\d{6}) ',
default=NotAvailable)(self)
if not s:
return Field('date')(self)
s = s.replace('/', '')
return Date(dayfirst=True).filter(
'%s%s%s%s%s' % (s[:2], '-', s[2:4], '-', s[4:]))
def obj_rdate(self):
if self.obj.rdate:
# Transaction.Raw may have already set it
return self.obj.rdate
s = Regexp(Field('raw'), ' (\d{2}/\d{2}/\d{2}) | (?!NUM) (\d{6}) ', default=NotAvailable)(self)
if not s:
return Field('date')(self)
s = s.replace('/', '')
return Date(dayfirst=True).filter('%s%s%s%s%s' % (s[:2], '-', s[2:4], '-', s[4:]))
def obj_rdate(self):
if self.obj.rdate:
# Transaction.Raw may have already set it
return self.obj.rdate
s = Regexp(Field('raw'), ' (\d{2}/\d{2}/\d{2}) | (?!NUM) (\d{6}) ', default=NotAvailable)(self)
if not s:
return Field('date')(self)
s = s.replace('/', '')
# Sometimes the user enters an invalid date 16/17/19 for example
return Date(dayfirst=True, default=NotAvailable).filter('%s%s%s%s%s' % (s[:2], '-', s[2:4], '-', s[4:]))
def obj_rdate(self):
if self.obj.rdate:
# Transaction.Raw may have already set it
return self.obj.rdate
s = Regexp(Field('raw'),
' (\d{2}/\d{2}/\d{2}) | (?!NUM) (\d{6}) ',
default=NotAvailable)(self)
if not s:
return Field('date')(self)
s = s.replace('/', '')
# Sometimes the user enters an invalid date 16/17/19 for example
return Date(dayfirst=True, default=NotAvailable).filter(
'%s-%s-%s' % (s[:2], s[2:4], s[4:]))
def obj_url(self):
url = Regexp(AbsoluteLink('//div[has-class("torrentinfo")]//div[has-class("dltorrent")]//a[text()="Download torrent"]'), '(^.*)\?.*', '\\1')(self)
return url.replace('http://', 'https://')
def obj_url(self):
url = Regexp(AbsoluteLink('.//div[has-class("tt-name")]/a[1]'), '(^.*)\?.*', '\\1')(self)
return url.replace('http://', 'https://')
def obj_rdate(self):
s = Regexp(Field('raw'), ' (\d{2}/\d{2}/\d{2}) | (?!NUM) (\d{6}) ', default=NotAvailable)(self)
if not s:
return Field('date')(self)
s = s.replace('/', '')
return Date(dayfirst=True).filter('%s%s%s%s%s' % (s[:2], '-', s[2:4], '-', s[4:]))
def obj_split_path(self):
_id = Regexp(CleanText('./@href'),
'/\w{2}/(.*)',
default=u'accueil')(self)
return [SITE.CREATIVE.get('id')] + [_id.replace('/', '^')]
def obj_split_path(self):
_id = Regexp(CleanText('./@href'), '/\w{2}/(.*)', default=u'accueil')(self)
return [SITE.CREATIVE.get('id')] + [_id.replace('/', '^')]