day01-某某客信息爬取

某某客信息爬取

import csv

import requests
from bs4 import BeautifulSoup


def get_net(url: str):
    res = requests.get(
        url=url,
        headers={
            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/101.0.0.0 Safari/537.36'
        }
    )
    return res.text


def save_file(html: str):
    all_film = []
    soup = BeautifulSoup(html, 'lxml')
    list1 = soup.select('#esfMain>section .list>div')
    # print(list1, len(list1))
    for div in list1:
        title = div.select_one('.property-content-title>h3').text
        home_type_ = div.select('.property-content-info>.property-content-info-text>span')
        home_type = ''
        for ty in home_type_:
            home_type += ty.text
        size = div.select('.property-content-info-text')[1].text.strip()
        towards = div.select('.property-content-info-text')[2].text.strip()
        try:
            layers = div.select('.property-content-info-text')[3].text.strip()
        except IndexError:
            layers = '无'
        try:
            put_time = div.select('.property-content-info-text')[4].text.strip()
        except IndexError:
            put_time = '无'
        address = div.select_one('.property-content-info-comm-name').text
        address1 = div.select_one('.property-content-info-comm-address>span').text
        address2 = div.select('.property-content-info-comm-address>span')[1].text
        address3 = div.select_one('.property-content-info-comm-address').text
        address_end = f'{address} {address1}-{address2}-{address3}'
        money = div.select_one('.property-price-total').text.replace(' ', '')
        unit_price = div.select_one('.property-price-average').text
        all_film.append([title, home_type, size, towards, layers, put_time, address_end, money, unit_price])
    writer.writerows(all_film)
    print('保存完成')


if __name__ == '__main__':
    f = open('files/安居客.csv', 'w', encoding='utf-8', newline='')
    writer = csv.writer(f)
    writer.writerow(['标题', '房屋类型', '房屋大小', '房屋朝向', '楼层', '建造时间', '具体地址', '总价', '单价（每平方米）'])
    i = 1
    while i <= 10:
        url = f'https://chengdu.anjuke.com/sale/p{i}/?from=HomePage_TopBar'
        res = get_net(url)
        save_file(res)
        i += 1
# https://chengdu.anjuke.com/sale/p5/?from=HomePage_TopBar