def __find_similar_ad_from_pic(self, picture):
new_hash = phash(Image.open(urlopen(picture)))
hashes = [ad.picturehash for ad in Annonce.select()]
for old_hash in hashes:
if old_hash is not None and hex_to_hash(
old_hash) - new_hash < self.HASH_SIMILAR_TRESHOLD:
return Annonce.get(Annonce.picturehash == old_hash)
else:
return False
def search(parameters):
# Préparation des paramètres de la requête
payload = {
"limit": 35,
"limit_alu": 3,
"filters": {
"enums": {
"ad_type": ["offer"]
},
"category": {
"id": "10"
},
"location": {
"locations": []
},
"ranges": {
"square": {
"min": parameters['surface'][0],
"max": parameters['surface'][1]
},
"price": {
"min": parameters['price'][0],
"max": parameters['price'][1]
}
},
"keywords": {}
}
}
for cp in parameters['cities']:
payload['filters']['location']['locations'].append(
{'zipcode': str(cp[1])})
header = {'api_key': 'ba0c2dad52b3ec'}
request = requests.post("https://api.leboncoin.fr/finder/search",
json=payload,
headers=header)
data = request.json()
for ad in data['ads']:
try:
annonce = Annonce.get(id='lbc-' + str(ad['list_id']))
except:
_request = requests.get(
"https://api.leboncoin.fr/finder/classified/" +
str(ad['list_id']),
headers=header)
_data = _request.json()
rooms, surface = 0, 0
for param in _data.get('attributes'):
if param['key'] == 'rooms':
rooms = param['value']
if param['key'] == 'square':
surface = param['value'].replace(" m²", "")
annonce, created = Annonce.get_or_create(
id='lbc-' + str(_data.get('list_id')),
defaults={
'site':
"Leboncoin Pro" if ad['owner']['no_salesmen'] == False else
"Leboncoin Particulier",
'created':
datetime.strptime(_data.get('first_publication_date'),
"%Y-%m-%d %H:%M:%S"),
'title':
BeautifulSoup(_data.get('subject'), "lxml").text,
'description':
BeautifulSoup(
_data.get('body').replace("<br />", "\n"),
"lxml").text,
'telephone':
_data.get("phone"),
'price':
_data.get('price')[0],
'surface':
surface if surface.replace('.', '', 1).isdigit() else 0,
'rooms':
rooms,
'city':
_data.get('zipcode')
if _data.get('zipcode') is not None else '',
'link':
"https://www.leboncoin.fr/locations/%s.htm?ca=12_s" %
_data.get('list_id'),
'picture':
_data['images']['urls_large']
if 'urls_large' in _data['images'] else []
})
if created:
annonce.save()
def search(parameters):
# Préparation des paramètres de la requête
payload = {
'px_loyermin': parameters['price'][0],
'px_loyermax': parameters['price'][1],
'surfacemin': parameters['surface'][0],
'surfacemax': parameters['surface'][1],
# Si parameters['rooms'] = (2, 4) => "2,3,4"
'nbpieces': list(range(parameters['rooms'][0], parameters['rooms'][1] + 1)),
# Si parameters['bedrooms'] = (2, 4) => "2,3,4"
'nb_chambres': list(range(parameters['bedrooms'][0], parameters['bedrooms'][1] + 1)),
'ci': [int(cp[2]) for cp in parameters['cities']]
}
# Insertion des paramètres propres à LeBonCoin
payload.update(parameters['seloger'])
headers = {'user-agent': 'Dalvik/2.1.0 (Linux; U; Android 6.0.1; D5803 Build/MOB30M.Z1)'}
request = requests.get("http://ws-seloger.svc.groupe-seloger.com/search_4.0.xml", params=payload, headers=headers)
xml_root = ET.fromstring(request.text)
for annonceNode in xml_root.findall('annonces/annonce'):
# Seconde requête pour obtenir la description de l'annonce
# TODO: seulement si id absent de la base (limite les requetes api)
try:
annonce = Annonce.get(
id = 'seloger-' + annonceNode.find('idAnnonce').text
)
except:
_payload = {'noAudiotel': 1, 'idAnnonce': annonceNode.findtext('idAnnonce')}
_request = requests.get("http://ws-seloger.svc.groupe-seloger.com/annonceDetail_4.0.xml", params=_payload, headers=headers)
photos = list()
for photo in annonceNode.find("photos"):
photos.append(photo.findtext("stdUrl"))
# Une seule photo, la première
# break
annonce, created = Annonce.get_or_create(
id = 'seloger-' + annonceNode.find('idAnnonce').text,
site = 'SeLoger',
defaults = {
# SeLoger peut ne pas fournir de titre pour une annonce T_T
'title': "Appartement " + annonceNode.findtext('nbPiece') + " pièces" if annonceNode.findtext('titre') is None else annonceNode.findtext('titre'),
'description': ET.fromstring(_request.text).findtext("descriptif"),
'telephone': ET.fromstring(_request.text).findtext("contact/telephone"),
'created': datetime.strptime(annonceNode.findtext('dtCreation'), '%Y-%m-%dT%H:%M:%S'),
'price': annonceNode.find('prix').text,
'charges': annonceNode.find('charges').text,
'surface': annonceNode.find('surface').text,
'rooms': annonceNode.find('nbPiece').text,
'bedrooms': annonceNode.find('nbChambre').text,
'city': annonceNode.findtext('ville'),
'link': annonceNode.findtext('permaLien'),
'picture': photos
}
)
if created:
annonce.save()