Apartement Searcher

immo.py

import requests
import re
import json
import time
import urllib.parse
import logging
import sys


def countSite(url, regex, name):
    global sites
    try:
        page = requests.get(url)
    except:
        logging.error("An error occured %s " % sys.exc_info()[0])
        return []
    else :
        result = re.findall(regex, page.text)
        unique = []
        for link in result:
            if link not in unique:
                unique.append(link)
        new = []
        previous = len(sites[name]['offers'])
        for offer in unique:
            if offer not in sites[name]['offers']:
                if previous > 0:
                    new.append(offer)
                sites[name]['offers'].append(offer)
        removed = 0
        #for offer in sites[name]['offers']:
        #    if offer not in unique:
        #        removed += 1
        #        sites[name]['offers'].remove(offer)
        #        print("removed %s" % offer)
        if len(unique) == 0:
            message = "Aucune offre trouvée sur %s" % name
            url = "https://smsapi.free-mobile.fr/sendmsg?user=XXXXXXX&pass=XXXXXXXXX&msg=" + urllib.parse.quote(message)
            requests.get(url)
        logging.info("Found %s offers on %s (%s new, %s remove, %s previous)" % (len(unique), name, len(new), removed, previous))
        return new



if __name__ == "__main__":
    logging.basicConfig(filename='immo.log', level=logging.INFO, format='%(asctime)s Immo %(levelname)s: %(message)s')
    #logging.getLogger().addHandler(logging.StreamHandler())

    sites = {
        'CA': {
            'url': "https://www.ca-immobilier.fr/louer/recherche?sortby=undefined&codes=69000%3Alyon%3A69123%2C69100%3Avilleurbanne%3A69266%2C69200%3Av%C3%A9nissieux%3A69259%2C69500%3Abron%3A69029&sections=location&types=appartment&zones=&distance=0&displayMode=mosaic",
            're': r"<a role=\"link\" href=\"(/biens/location/\S+/\d+)\" class=\"tc-link\"",
            'prefix': "https://www.ca-immobilier.fr",
            'offers': []
        },
        'Citya': {
            'url': "https://www.citya.com/annonces/location/appartement/lyon-3e-arrondissement-69003,lyon-7e-arrondissement-69007,lyon-8e-arrondissement-69008?prixMax=750&surfaceMin=30",
            're': r"a href=\"(/annonces/location/appartement/.+)\" class=",
            'prefix': "https://www.citya.com",
            'offers': []
        },
        'Foncia': {
            'url': "https://fr.foncia.com/location/lyon-69--villeurbanne-69100--venissieux-69200/appartement/(params)/on/(surface_min)/30/(prix_max)/750/(pieces)/1--2/(parking)/1",
            're': r"<a href=\"(/location/\S+/appartement/\d+\.htm)\" class=\"TeaserOffer-ill\"",
            'prefix': "https://fr.foncia.com",
            'offers': []
        },
    }

    while True:
        for site, data in sites.items():
            for offer in countSite(data['url'], data['re'], site):
                link = data['prefix'] + offer
                logging.info("New offer appeared on %s : %s" % (site, link))
                message = "Nouvel appartement sur %s : %s" % (site, link)
                url = "https://smsapi.free-mobile.fr/sendmsg?user=XXXXXXXXX&pass=XXXXXXXXXXX&msg=" + urllib.parse.quote(message)
                requests.get(url)

        time.sleep(1 * 60)