1 第三方库

• 获取随机UA

pip install fake-useragent

• 连接数据库

$ pip3 install PyMySQL

• 发起请求

pip install requests

• 解析页面

pip install beautifulsoup4

• 进度条

pip install tqdm

2 爬取

2.1 初始化函数

• 新建爬虫类

class mySpider:

• 创建数据库连接和初始化url

# 初始化urldef __init__(self, url):self.url = url# 计数，请求一个页面的次数，初始值为1self.count = 1# 数据库连接对象self.db = pymysql.connect(host='localhost',port=3306,user='root',password='123456',database='test')# 创建游标对象self.cursor = self.db.cursor()

2.2 结束时关闭数据库

• 关闭数据库释放资源，方法运行完后调用。

# 结束断开数据库连接def __del__(self):self.cursor.close()self.db.close()print("关闭数据库！")

2.3 生成header

• 使用第三方库fake-useragent生成随机UA

# 获取一个headerdef getHeader(self):# 实例化ua对象ua = UserAgent()# 随机获取一个uaheaders = {'User-Agent': ua.random}return headers

2.4 获取请求body

• 注意有返回值的递归，要把返回值返回，回调时加return

def getBody(self, url, send_type, data):# 每次请求都随机停顿一些时间# time.sleep(random.randint(1, 2))# 在超时时间内，对于失败页面尝试请求三次if self.count <= 3:try:if send_type == 'get':res = requests.get(url=url, headers=self.getHeader(), params=data, timeout=2)elif send_type == 'post':res = requests.post(url=url, headers=self.getHeader(), data=data, timeout=2)else:print("未输入send_type，直接返回None")res = Nonereturn resexcept Exception as e:print(e)self.count += 1print(f"第{self.count}次，发起请求")# 再次调用自己,并把值返回，（注意要加return）return self.getBody(url, send_type, data)

2.5 解析异步json数据

• 解析异步json数据

def parseData(self, dataList):# 循环查看详情for row in tqdm(dataList, desc='爬取进度'):# 请求详情页urlurlDetail = f"https://www.baidu.com/CTMDS/pub/PUB010100.do?method=handle04&compId={row['companyId']}"# 发起请求# 每次请求都初始化一次self.countself.count = 1res = self.getBody(url=urlDetail, send_type='get', data={})if res is not None:# 解析htmlself.parseHtml(row=row, htmlText=res.text)else:print(f"{urlDetail}请求失败！")

2.6 使用BS4的find方法解析

• find_all() 方法用来搜索当前 tag 的所有子节点，并判断这些节点是否符合过滤条件，最后以列表形式将符合条件的内容返回，语法格式如下
  find_all( name , attrs , recursive , text , limit )
• 参数说明
• name：查找所有名字为 name 的 tag 标签，字符串对象会被自动忽略。
• attrs：按照属性名和属性值搜索 tag 标签，注意由于 class 是 Python 的关键字吗，所以要使用 “class_”。
• recursive：find_all() 会搜索 tag 的所有子孙节点，设置 recursive=False 可以只搜索 tag 的直接子节点。
• text：用来搜文档中的字符串内容，该参数可以接受字符串 、正则表达式 、列表、True。
• limit：由于 find_all() 会返回所有的搜索结果，这样会影响执行效率，通过 limit 参数可以限制返回结果的数量。

def parseHtml(self, row, htmlText):soup = BeautifulSoup(htmlText, 'html.parser')# 获取备案信息divList = soup.find_all('div', class_=['col-md-8'])divtextList = [re.sub(r'\s+', '', div.text) for div in divList]# 获取其他机构地址divListOther = soup.find_all('div', class_=['col-sm-8'])divtextListOther = [re.sub(r'\s+', '', div.text) for div in divListOther]otherOrgAdd = ','.join(divtextListOther)# 插入数据库companyId = row['companyId']linkTel = row['linkTel']recordNo = row['recordNo']areaName = row['areaName']linkMan = row['linkMan']address = row['address']compName = row['compName']recordStatus = row['recordStatus']cancelRecordTime = row.get('cancelRecordTime', '')compLevel = divtextList[2]recordTime = divtextList[6]sql1 = "insert INTO medical_register(company_id,area_name,record_no,comp_name,address,link_man,link_tel,record_status,comp_level,record_time,cancel_record_time,other_org_add) "sql2 = f"values('{companyId}','{areaName}','{recordNo}','{compName}','{address}','{linkMan}','{linkTel}','{recordStatus}','{compLevel}','{recordTime}','{cancelRecordTime}','{otherOrgAdd}')"sql3 = sql1 + sql2# 执行sqlself.cursor.execute(sql3)# 提交self.db.commit()# 获取备案专业和主要研究者信息tbody = soup.find('tbody')trList = tbody.find_all('tr')# 对tr循环获取tdfor tr in trList:tdList = tr.find_all('td')tdTextList = [td.text for td in tdList]tdTextList.insert(0, companyId)# print(tdTextList)# 插入数据库sql4 = "insert into medical_register_sub (company_id,professional_name,principal_investigator,job_title) values(%s,%s,%s,%s)"self.cursor.execute(sql4, tdTextList)# 提交到数据库self.db.commit()

2.7 写入口函数

• 这里pageSize直接干到最大，懂的都懂！

def run(self):try:# 拿第一页的数据data = {'pageSize': 1350, 'curPage': 1}# 每次请求都初始化一次self.countself.count = 1res = self.getBody(url=self.url, send_type='post', data=data)if res is not None:# 加载为jsonjsonRes = json.loads(res.text)# 查看响应状态码status = jsonRes['success']# 如果状态为Trueif status == True:# 获取数据dataList = jsonRes['data']# 处理数据self.parseData(dataList=dataList)else:print(f"{self.url}请求失败")except Exception as e:print('发生错误！', e)

2.8 调用

• 调用

if __name__ == '__main__':spider = mySpider('https://www.baidu.com/CTMDS/pub/PUB010100.do?method=handle05')spider.run()

3 完整代码

• 完整代码

import requests
from bs4 import BeautifulSoup
from fake_useragent import UserAgent
import time
import random
import json
import re
import pymysql
from tqdm import tqdmclass mySpider:# 初始化urldef __init__(self, url):self.url = url# 计数，请求一个页面的次数，初始值为1self.count = 1# 数据库连接对象self.db = pymysql.connect(host='localhost',port=3306,user='root',password='logicfeng',database='test2')# 创建游标对象self.cursor = self.db.cursor()# 结束断开数据库连接def __del__(self):self.cursor.close()self.db.close()print("关闭数据库！")# 获取一个headerdef getHeader(self):# 实例化ua对象ua = UserAgent()# 随机获取一个uaheaders = {'User-Agent': ua.random}return headers# 获取请求bodydef getBody(self, url, send_type, data):# 每次请求都随机停顿一些时间# time.sleep(random.randint(1, 2))# 在超时时间内，对于失败页面尝试请求三次if self.count <= 3:try:if send_type == 'get':res = requests.get(url=url, headers=self.getHeader(), params=data, timeout=2)elif send_type == 'post':res = requests.post(url=url, headers=self.getHeader(), data=data, timeout=2)else:print("未输入send_type，直接返回None")res = Nonereturn resexcept Exception as e:print(e)self.count += 1print(f"第{self.count}次，发起请求")# 再次调用自己,并把值返回，（注意要加return）return self.getBody(url, send_type, data)# 解析bodydef parseData(self, dataList):# 循环查看详情for row in tqdm(dataList, desc='爬取进度'):# 请求详情页urlurlDetail = f"https://www.baidu.com/CTMDS/pub/PUB010100.do?method=handle04&compId={row['companyId']}"# 发起请求# 每次请求都初始化一次self.countself.count = 1res = self.getBody(url=urlDetail, send_type='get', data={})if res is not None:# 解析htmlself.parseHtml(row=row, htmlText=res.text)else:print(f"{urlDetail}请求失败！")# 解析页面def parseHtml(self, row, htmlText):soup = BeautifulSoup(htmlText, 'html.parser')# 获取备案信息divList = soup.find_all('div', class_=['col-md-8'])divtextList = [re.sub(r'\s+', '', div.text) for div in divList]# 获取其他机构地址divListOther = soup.find_all('div', class_=['col-sm-8'])divtextListOther = [re.sub(r'\s+', '', div.text) for div in divListOther]otherOrgAdd = ','.join(divtextListOther)# 插入数据库companyId = row['companyId']linkTel = row['linkTel']recordNo = row['recordNo']areaName = row['areaName']linkMan = row['linkMan']address = row['address']compName = row['compName']recordStatus = row['recordStatus']cancelRecordTime = row.get('cancelRecordTime', '')compLevel = divtextList[2]recordTime = divtextList[6]sql1 = "insert INTO medical_register(company_id,area_name,record_no,comp_name,address,link_man,link_tel,record_status,comp_level,record_time,cancel_record_time,other_org_add) "sql2 = f"values('{companyId}','{areaName}','{recordNo}','{compName}','{address}','{linkMan}','{linkTel}','{recordStatus}','{compLevel}','{recordTime}','{cancelRecordTime}','{otherOrgAdd}')"sql3 = sql1 + sql2# 执行sqlself.cursor.execute(sql3)# 提交self.db.commit()# 获取备案专业和主要研究者信息tbody = soup.find('tbody')trList = tbody.find_all('tr')# 对tr循环获取tdfor tr in trList:tdList = tr.find_all('td')tdTextList = [td.text for td in tdList]tdTextList.insert(0, companyId)# print(tdTextList)# 插入数据库sql4 = "insert into medical_register_sub (company_id,professional_name,principal_investigator,job_title) values(%s,%s,%s,%s)"self.cursor.execute(sql4, tdTextList)# 提交到数据库self.db.commit()# 入口函数def run(self):try:# 拿第一页的数据data = {'pageSize': 1350, 'curPage': 1}# 每次请求都初始化一次self.countself.count = 1res = self.getBody(url=self.url, send_type='post', data=data)if res is not None:# 加载为jsonjsonRes = json.loads(res.text)# 查看响应状态码status = jsonRes['success']# 如果状态为Trueif status == True:# 获取数据dataList = jsonRes['data']# 处理数据self.parseData(dataList=dataList)else:print(f"{self.url}请求失败")except Exception as e:print('发生错误！', e)if __name__ == '__main__':spider = mySpider('https://www.百度.com/CTMDS/pub/PUB010100.do?method=handle05')spider.run()