Python程序设计作业20

2019302523-于米提江 2022夏-程序设计实训 2022-07-29 20:50:46

import requests
from bs4 import BeautifulSoup

def crawler(book_list):
    headers = {
        "User-Agent": "Mozilla/5.0 (compatible; MSIE 9.0; Windows Phone OS 7.5; Trident/5.0; IEMobile/9.0; HTC; Titan)",
    }
    id_list = []
    rank_list = []
    id_rank_dict = {}

    for page in range(1, 6):
        if len(id_list) < 3:
            base_url = 'https://www.qidian.com/rank/yuepiao?style=1&page={}'.format(page)
            response = requests.get(base_url, headers=headers)
            response.encoding = 'utf-8'
            html = response.text
            book_name = []
            book_rank = []
            htmls = BeautifulSoup(html, 'html.parser')
                # xpath守则
            for temp in htmls.find_all(class_='book-mid-info'):
                tempname = temp.find('a')
                book_name.append(tempname.string)
            for temp in htmls.find_all(class_='book-img-box'):
                tempnum = temp.span
                book_rank.append(tempnum.next_element)
            for i in book_list:
                k = ''
                j = 0
                if i in book_name:
                    for k in book_name:
                        if k == i:
                            id_list.append(i)
                            rank_list.append(int(book_rank[j]))
                            break
                        else:
                            j += 1
        elif len(id_list) == 3:
            id_rank_dict = dict(zip(id_list, rank_list))
            d_order = sorted(id_rank_dict.items(), key=lambda x: x[1], reverse=False)
            d_order[0] = list(d_order[0])
            d_order[1] = list(d_order[1])
            d_order[2] = list(d_order[2])
            d_order[0][1] = 1
            d_order[1][1] = 2
            d_order[2][1] = 3
            id_rank_dict = dict(d_order)
            return (id_rank_dict)
if __name__ == '__main__':
    book_list = ["大奉打更人", "明克街13号","灵境行者"]
    print(crawler(book_list))

...全文