爬虫代码基础教学

from http.client import ResponseNotReady
import json
from unicodedata import name
import requests
from lxml import etree
import csv
import xlwt

# base_url = 'https://www.basketball-reference.com'
base_url = https://www.baidu.com
headers = {
    'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/101.0.4951.67 Safari/537.36',
}
response = requests.get(base_url, headers=headers)
html = etree.HTML(response.text)
url_list = html.xpath('//*[@id="site_menu"]/ul/li[2]/div/a/@href')
team_my_names = html.xpath('//*[@id="site_menu"]/ul/li[2]/div/a/text()')
teams_list = url_list
# print(teams_list)

for h in range(len(teams_list)):
    base_url = 'https://www.basketball-reference.com' + teams_list[h]
    headers = {
        'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/101.0.4951.67 Safari/537.36',
    }
    xls = xlwt.Workbook()
    sht1 = xls.add_sheet('Sheet1')
    response = requests.get(base_url, headers=headers)
    html = etree.HTML(response.text)
    url_list = html.xpath('//*[@id="roster"]/tbody/tr/td[1]/a/@href')
    names = html.xpath('//*[@id="roster"]/tbody/tr/td[1]/a')
    # print(url_list)
    # print(names)

    name_lst = []
    for j in range(len(names)):
        name_lst.append(names[j].text)
    names = name_lst

    person_list = []
    for j in range(len(url_list)):
        person_list.append('https://www.basketball-reference.com' + url_list[j])
    url_list = person_list

    print(team_my_names[h], end=": ")
    file_name = 'files/' + team_my_names[h] + ".xls"
    count = 0
    print()
    for i in range(len(url_list)):
        print(names[i], end=',')
        response = requests.get(url_list[i], headers=headers)
        html = etree.HTML(response.text)
        year_lst = html.xpath('//*[@id="per_game"]/tbody/tr/th/a/text()')
        team_name = html.xpath('//*[@id="per_game"]/tbody/tr/td[@data-stat="team_id"]//text()')
        for j in range(len(year_lst)):
            sht1.write(count, 0, names[i])
            sht1.write(count, 1, year_lst[j])
            sht1.write(count, 2, 'NULL' if len(team_name) < j else team_name[j])
            count += 1
    xls.save(file_name)
    print()