def get_housing(self):
div = self.parser.select(self.document.getroot(), 'div#annonce_detail',
1)
housing = Housing(self.url.split('-')[-1])
parts = div.find('h1').text.split(' - ')
housing.title = parts[0].strip()
housing.cost = Decimal(parts[1].strip(u' \t\u20ac\xa0€\n\r').replace(
'.', '').replace(',', '.'))
housing.currency = u'€'
m = re.match('(\w+) (.+) (\d+)\xa0m\xb2 (.*)', housing.title)
if m:
housing.area = Decimal(m.group(3))
housing.date = housing.station = housing.location = housing.phone = NotAvailable
metro = div.cssselect('p.metro')
if len(metro) > 0:
housing.station = metro[0].text.strip()
p = div.cssselect('p.annonce-detail-texte')[0]
b = p.findall('b')
if len(b) > 0:
housing.text = b[0].tail.strip()
housing.location = unicode(b[0].text)
if len(b) > 1:
housing.phone = b[1].text
else:
housing.text = p.text.strip()
housing.details = NotAvailable
housing.photos = NotAvailable
return housing
def get_housing(self):
div = self.parser.select(self.document.getroot(), 'div#annonce_detail', 1)
housing = Housing(self.url.split('-')[-1])
parts = div.find('h1').text.split(' - ')
housing.title = parts[0].strip()
housing.cost = Decimal(parts[1].strip(u' \t\u20ac\xa0€\n\r').replace('.', '').replace(',', '.'))
housing.currency = u'€'
m = re.match('(\w+) (.+) (\d+)\xa0m\xb2 (.*)', housing.title)
if m:
housing.area = Decimal(m.group(3))
housing.date = housing.station = housing.location = housing.phone = NotAvailable
metro = div.cssselect('p.metro')
if len(metro) > 0:
housing.station = metro[0].text.strip()
p = div.cssselect('p.annonce-detail-texte')[0]
b = p.findall('b')
if len(b) > 0:
housing.text = b[0].tail.strip()
housing.location = unicode(b[0].text)
if len(b) > 1:
housing.phone = b[1].text
else:
housing.text = p.text.strip()
housing.details = NotAvailable
housing.photos = NotAvailable
return housing
def iter_housings(self):
for div in self.document.getroot().cssselect('div.annonce-resume'):
a = div.cssselect('td.lien-annonce')[0].find('a')
if a is None:
# not a real announce.
continue
id = a.attrib['href'].split('-')[-1]
housing = Housing(id)
housing.title = a.text.strip()
m = re.match('(\w+) (.+) (\d+)\xa0m\xb2 (.*)', housing.title)
if m:
housing.area = Decimal(m.group(3))
housing.cost = Decimal(
div.cssselect('td.prix')[0].text.strip(
u' \t\u20ac\xa0€\n\r').replace('.', '').replace(',', '.'))
housing.currency = u'€'
m = self.DATE_RE.match(
div.cssselect('p.date-publication')[0].text.strip())
if m:
date = m.group(1)
for fr, en in self.MONTHS.iteritems():
date = date.replace(fr, en)
housing.date = parse_date(date)
metro = div.cssselect('p.metro')
if len(metro) > 0:
housing.station = unicode(metro[0].text.strip())
else:
housing.station = NotAvailable
p = div.cssselect('p.annonce-resume-texte')[0]
b = p.findall('b')
if len(b) > 0:
housing.text = b[0].tail.strip()
housing.location = unicode(b[0].text)
else:
housing.text = p.text.strip()
housing.photos = NotAvailable
yield housing
def get_housing(self, housing=None):
if housing is None:
housing = Housing(self.groups[0])
details = self.document.getroot().xpath('//detailannonce')[0]
if details.find('titre') is None:
return None
housing.title = unicode(details.find('titre').text)
housing.text = details.find('descriptif').text.strip()
housing.cost = Decimal(details.find('prix').text)
housing.currency = u'€'
housing.date = parse_date(details.find('dtfraicheur').text)
housing.area = Decimal(details.find('surface').text)
housing.phone = unicode(details.find('contact').find('telephone').text)
try:
housing.station = unicode(details.find('proximite').text)
except AttributeError:
housing.station = NotAvailable
housing.location = details.find('adresse').text
if not housing.location and details.find('quartier') is not None:
housing.location = unicode(details.find('quartier').text)
if not housing.location:
housing.location = NotAvailable
housing.photos = []
for photo in details.xpath('./photos/photo'):
if photo.find('bigurl').text:
url = photo.find('bigurl').text
else:
url = photo.find('stdurl').text
housing.photos.append(HousingPhoto(unicode(url)))
housing.details = {}
for detail in details.xpath('./details/detail'):
housing.details[detail.find(
'libelle').text.strip()] = detail.find('valeur').text or 'N/A'
housing.details['Reference'] = details.find('reference').text
return housing
def get_housing(self, housing=None):
if housing is None:
housing = Housing(self.groups[0])
details = self.document.getroot().xpath('//detailannonce')[0]
if details.find('titre') is None:
return None
housing.title = unicode(details.find('titre').text)
housing.text = details.find('descriptif').text.strip()
housing.cost = Decimal(details.find('prix').text)
housing.currency = u'€'
housing.date = parse_date(details.find('dtfraicheur').text)
housing.area = Decimal(details.find('surface').text)
housing.phone = unicode(details.find('contact').find('telephone').text)
try:
housing.station = unicode(details.find('proximite').text)
except AttributeError:
housing.station = NotAvailable
housing.location = details.find('adresse').text
if not housing.location and details.find('quartier') is not None:
housing.location = unicode(details.find('quartier').text)
if not housing.location:
housing.location = NotAvailable
housing.photos = []
for photo in details.xpath('./photos/photo'):
if photo.find('bigurl').text:
url = photo.find('bigurl').text
else:
url = photo.find('stdurl').text
housing.photos.append(HousingPhoto(unicode(url)))
housing.details = {}
for detail in details.xpath('./details/detail'):
housing.details[detail.find('libelle').text.strip()] = detail.find('valeur').text or 'N/A'
housing.details['Reference'] = details.find('reference').text
return housing
def iter_housings(self):
for div in self.document.getroot().cssselect('div.annonce-resume'):
a = div.cssselect('td.lien-annonce')[0].find('a')
if a is None:
# not a real announce.
continue
id = a.attrib['href'].split('-')[-1]
housing = Housing(id)
housing.title = a.text.strip()
m = re.match('(\w+) (.+) (\d+)\xa0m\xb2 (.*)', housing.title)
if m:
housing.area = Decimal(m.group(3))
housing.cost = Decimal(div.cssselect('td.prix')[0].text.strip(u' \t\u20ac\xa0€\n\r').replace('.', '').replace(',', '.'))
housing.currency = u'€'
m = self.DATE_RE.match(div.cssselect('p.date-publication')[0].text.strip())
if m:
date = m.group(1)
for fr, en in self.MONTHS.iteritems():
date = date.replace(fr, en)
housing.date = parse_date(date)
metro = div.cssselect('p.metro')
if len(metro) > 0:
housing.station = unicode(metro[0].text.strip())
else:
housing.station = NotAvailable
p = div.cssselect('p.annonce-resume-texte')[0]
b = p.findall('b')
if len(b) > 0:
housing.text = b[0].tail.strip()
housing.location = unicode(b[0].text)
else:
housing.text = p.text.strip()
housing.photos = NotAvailable
yield housing
def iter_housings(self):
for a in self.document.getroot().xpath('//annonce'):
housing = Housing(a.find('idannonce').text)
housing.title = unicode(a.find('titre').text)
housing.date = parse_date(a.find('dtfraicheur').text)
housing.cost = Decimal(a.find('prix').text)
housing.currency = u'€'
housing.area = Decimal(a.find('surface').text)
housing.text = unicode(a.find('descriptif').text.strip())
housing.location = unicode(a.find('ville').text)
try:
housing.station = unicode(a.find('proximite').text)
except AttributeError:
housing.station = NotAvailable
housing.photos = []
for photo in a.xpath('./photos/photo'):
url = unicode(photo.find('stdurl').text)
housing.photos.append(HousingPhoto(url))
yield housing
def iter_housings(self):
for a in self.document.getroot().xpath('//annonce'):
housing = Housing(a.find('idannonce').text)
housing.title = unicode(a.find('titre').text)
housing.date = parse_date(a.find('dtfraicheur').text)
housing.cost = Decimal(a.find('prix').text)
housing.currency = u'€'
housing.area = Decimal(a.find('surface').text)
housing.text = unicode(a.find('descriptif').text.strip())
housing.location = unicode(a.find('ville').text)
try:
housing.station = unicode(a.find('proximite').text)
except AttributeError:
housing.station = NotAvailable
housing.photos = []
for photo in a.xpath('./photos/photo'):
url = unicode(photo.find('stdurl').text)
housing.photos.append(HousingPhoto(url))
yield housing