def filter(self, el):
index = 1 if len(el) > 1 else 0
content = CleanText.clean(CleanText('.', ['HORAIRES'])(el[index]))
a_time = content.split(' - ')[-2]
regexp = re.compile(ur'(?P<hh>\d+)h?(?P<mm>\d+)')
m = regexp.search(a_time)
return time(int(m.groupdict()['hh'] or 0), int(m.groupdict()['mm'] or 0))
def filter(self, el):
index = 1 if len(el) > 1 else 0
content = CleanText.clean(CleanText('.', ['HORAIRES'])(el[index]))
a_price = content.split(' - ')[-1]
parsed_price = re.findall(r"\d*\,\d+|\d+", " ".join(a_price))
if parsed_price and len(parsed_price) > 0:
return float(parsed_price[0].replace(',', '.'))
return float(0)
def filter(self, el):
content = CleanText.clean(CleanText(CleanHTML('.'), ['*'])(el[0]))
a_date = content[0:content.index(' - ')]
for fr, en in date_util.DATE_TRANSLATE_FR:
a_date = fr.sub(en, a_date)
try:
_month = datetime.strptime(a_date, "%A %d %B").month
if (datetime.now().month > _month):
a_date += u' %i' % (datetime.now().year + 1)
else:
a_date += u' %i' % (datetime.now().year)
except ValueError:
pass
return datetime.strptime(a_date, "%A %d %B %Y")
def next_page(self):
try:
form = self.page.get_form('//form[@id="paginationForm"]')
except FormNotFound:
return
text = CleanText.clean(form.el)
m = re.search(u"(\d+) / (\d+)", text or "", flags=re.MULTILINE)
if not m:
return
cur = int(m.group(1))
last = int(m.group(2))
if cur == last:
return
form["page"] = str(cur + 1)
return form.request