|
| 1 | +import os |
| 2 | +import re |
| 3 | +import time |
| 4 | +import requests |
| 5 | +from pymongo import MongoClient |
| 6 | +from info import rent_type, city_info |
| 7 | + |
| 8 | + |
| 9 | +class Rent(object): |
| 10 | + """ |
| 11 | + 初始化函数，获取租房类型（整租、合租）、要爬取的城市分区信息以及连接mongodb数据库 |
| 12 | + """ |
| 13 | + def __init__(self): |
| 14 | + self.rent_type = rent_type |
| 15 | + self.city_info = city_info |
| 16 | + |
| 17 | + host = os.environ.get('MONGODB_HOST', '127.0.0.1') # 本地数据库 |
| 18 | + port = os.environ.get('MONGODB_PORT', '27017') # 数据库端口 |
| 19 | + mongo_url = 'mongodb://{}:{}'.format(host, port) |
| 20 | + mongo_db = os.environ.get('MONGODB_DATABASE', 'Lianjia') |
| 21 | + client = MongoClient(mongo_url) |
| 22 | + self.db = client[mongo_db] |
| 23 | + self.db['zufang'].create_index('m_url', unique=True) # 以m端链接为主键进行去重 |
| 24 | + |
| 25 | + def get_data(self): |
| 26 | + """ |
| 27 | + 爬取不同租房类型、不同城市各区域的租房信息 |
| 28 | + :return: None |
| 29 | + """ |
| 30 | + for ty, type_code in self.rent_type.items(): # 整租、合租 |
| 31 | + for city, info in self.city_info.items(): # 城市、城市各区的信息 |
| 32 | + for dist, dist_py in info[2].items(): # 各区及其拼音 |
| 33 | + res_bc = requests.get('https://m.lianjia.com/chuzu/{}/zufang/{}/'.format(info[1], dist_py)) |
| 34 | + pa_bc = r"data-type=\"bizcircle\" data-key=\"(.*)\" class=\"oneline \">" |
| 35 | + bc_list = re.findall(pa_bc, res_bc.text) |
| 36 | + self._write_bc(bc_list) |
| 37 | + bc_list = self._read_bc() # 先爬取各区的商圈，最终以各区商圈来爬数据，如果按区爬，每区最多只能获得2000条数据 |
| 38 | + |
| 39 | + if len(bc_list) > 0: |
| 40 | + for bc_name in bc_list: |
| 41 | + idx = 0 |
| 42 | + has_more = 1 |
| 43 | + while has_more: |
| 44 | + try: |
| 45 | + url = 'https://app.api.lianjia.com/Rentplat/v1/house/list?city_id={}&condition={}' \ |
| 46 | + '/rt{}&limit=30&offset={}&request_ts={}&scene=list'.format(info[0], |
| 47 | + bc_name, |
| 48 | + type_code, |
| 49 | + idx*30, |
| 50 | + int(time.time())) |
| 51 | + res = requests.get(url=url, timeout=10) |
| 52 | + print('成功爬取{}市{}-{}的{}第{}页数据！'.format(city, dist, bc_name, ty, idx+1)) |
| 53 | + item = {'city': city, 'type': ty, 'dist': dist} |
| 54 | + self._parse_record(res.json()['data']['list'], item) |
| 55 | + |
| 56 | + total = res.json()['data']['total'] |
| 57 | + idx += 1 |
| 58 | + if total/30 <= idx: |
| 59 | + has_more = 0 |
| 60 | + # time.sleep(random.random()) |
| 61 | + except: |
| 62 | + print('链接访问不成功，正在重试！') |
| 63 | + |
| 64 | + def _parse_record(self, data, item): |
| 65 | + """ |
| 66 | + 解析函数，用于解析爬回来的response的json数据 |
| 67 | + :param data: 一个包含房源数据的列表 |
| 68 | + :param item: 传递字典 |
| 69 | + :return: None |
| 70 | + """ |
| 71 | + if len(data) > 0: |
| 72 | + for rec in data: |
| 73 | + item['bedroom_num'] = rec.get('frame_bedroom_num') |
| 74 | + item['hall_num'] = rec.get('frame_hall_num') |
| 75 | + item['bathroom_num'] = rec.get('frame_bathroom_num') |
| 76 | + item['rent_area'] = rec.get('rent_area') |
| 77 | + item['house_title'] = rec.get('house_title') |
| 78 | + item['resblock_name'] = rec.get('resblock_name') |
| 79 | + item['bizcircle_name'] = rec.get('bizcircle_name') |
| 80 | + item['layout'] = rec.get('layout') |
| 81 | + item['rent_price_listing'] = rec.get('rent_price_listing') |
| 82 | + item['house_tag'] = self._parse_house_tags(rec.get('house_tags')) |
| 83 | + item['frame_orientation'] = rec.get('frame_orientation') |
| 84 | + item['m_url'] = rec.get('m_url') |
| 85 | + item['rent_price_unit'] = rec.get('rent_price_unit') |
| 86 | + |
| 87 | + try: |
| 88 | + res2 = requests.get(item['m_url'], timeout=5) |
| 89 | + pa_lon = r"longitude: '(.*)'," |
| 90 | + pa_lat = r"latitude: '(.*)'" |
| 91 | + pa_distance = r"<span class=\"fr\">(\d*)米</span>" |
| 92 | + item['longitude'] = re.findall(pa_lon, res2.text)[0] |
| 93 | + item['latitude'] = re.findall(pa_lat, res2.text)[0] |
| 94 | + distance = re.findall(pa_distance, res2.text) |
| 95 | + if len(distance) > 0: |
| 96 | + item['distance'] = distance[0] |
| 97 | + else: |
| 98 | + item['distance'] = None |
| 99 | + except: |
| 100 | + item['longitude'] = None |
| 101 | + item['latitude'] = None |
| 102 | + item['distance'] = None |
| 103 | + |
| 104 | + self.db['zufang'].update_one({'m_url': item['m_url']}, {'$set': item}, upsert=True) |
| 105 | + print('成功保存数据:{}!'.format(item)) |
| 106 | + |
| 107 | + @staticmethod |
| 108 | + def _parse_house_tags(house_tag): |
| 109 | + """ |
| 110 | + 处理house_tags字段，相当于数据清洗 |
| 111 | + :param house_tag: house_tags字段的数据 |
| 112 | + :return: 处理后的house_tags |
| 113 | + """ |
| 114 | + if len(house_tag) > 0: |
| 115 | + st = '' |
| 116 | + for tag in house_tag: |
| 117 | + st += tag.get('name') + ' ' |
| 118 | + return st.strip() |
| 119 | + |
| 120 | + @staticmethod |
| 121 | + def _write_bc(bc_list): |
| 122 | + """ |
| 123 | + 把爬取的商圈写入txt，为了整个爬取过程更加可控 |
| 124 | + :param bc_list: 商圈list |
| 125 | + :return: None |
| 126 | + """ |
| 127 | + with open('bc_list.txt', 'w') as f: |
| 128 | + for bc in bc_list: |
| 129 | + f.write(bc+'\n') |
| 130 | + |
| 131 | + @staticmethod |
| 132 | + def _read_bc(): |
| 133 | + """ |
| 134 | + 读入商圈 |
| 135 | + :return: None |
| 136 | + """ |
| 137 | + with open('bc_list.txt', 'r') as f: |
| 138 | + return [bc.strip() for bc in f.readlines()] |
| 139 | + |
| 140 | + |
| 141 | +if __name__ == '__main__': |
| 142 | + rent = Rent() |
| 143 | + rent.get_data() |
0 commit comments