一不小心，我爬取了100萬條微博評論
- 2020 年 3 月 10 日
- 筆記
↑ 關注 + 星標 ~ 有趣的不像個技術號
每晚九點，我們準時相約
鄭重聲明：本項目及所有相關文章，僅用於經驗技術交流，禁止將相關技術應用到不正當途徑，因為濫用技術產生的風險與本人無關
大家好，我是朱小五
今天給大家分享一篇文章用來學習，是關於微博評論的爬蟲。
作者宗傑，已經在源碼關鍵處做了注釋。
所以這可能是「凹凸數據」有史以來代碼最多、文字最少的一篇了
（溫馨提醒：完整代碼地址在文末！文末！）
下面見證奇蹟的時刻：
1、引入庫
import time import base64 import rsa import binascii import requests import re from PIL import Image import random from urllib.parse import quote_plus import http.cookiejar as cookielib import csv import os
2、一些全局變量的設置
comment_path = 'comment' agent = 'mozilla/5.0 (windowS NT 10.0; win64; x64) appLewEbkit/537.36 (KHTML, likE gecko) chrome/71.0.3578.98 safari/537.36' headers = {'User-Agent': agent}
3、創立目錄作為存放數據的
if not os.path.exists(comment_path): os.mkdir(comment_path)
4、登陸類的創立
class WeiboLogin(object): """ 通過登錄 weibo.com 然後跳轉到 m.weibo.cn """ # 初始化數據 def __init__(self, user, password, cookie_path): super(WeiboLogin, self).__init__() self.user = user self.password = password self.session = requests.Session() self.cookie_path = cookie_path # LWPCookieJar是python中管理cookie的工具，可以將cookie保存到文件，或者在文件中讀取cookie數據到程序 self.session.cookies = cookielib.LWPCookieJar(filename=self.cookie_path) self.index_url = "http://weibo.com/login.php" self.session.get(self.index_url, headers=headers, timeout=2) self.postdata = dict() def get_su(self): """ 對 email 地址和手機號碼 先 javascript 中 encodeURIComponent 對應 Python 3 中的是 urllib.parse.quote_plus 然後在 base64 加密後decode """ username_quote = quote_plus(self.user) username_base64 = base64.b64encode(username_quote.encode("utf-8")) return username_base64.decode("utf-8") # 預登陸獲得 servertime, nonce, pubkey, rsakv def get_server_data(self, su): """與原來的相比，微博的登錄從 v1.4.18 升級到了 v1.4.19 這裡使用了 URL 拼接的方式，也可以用 Params 參數傳遞的方式 """ pre_url = "http://login.sina.com.cn/sso/prelogin.php?entry=weibo&callback=sinaSSOController.preloginCallBack&su=" pre_url = pre_url + su + "&rsakt=mod&checkpin=1&client=ssologin.js(v1.4.19)&_=" pre_url = pre_url + str(int(time.time() * 1000)) pre_data_res = self.session.get(pre_url, headers=headers) # print("*"*50) # print(pre_data_res.text) # print("*" * 50) sever_data = eval(pre_data_res.content.decode("utf-8").replace("sinaSSOController.preloginCallBack", '')) return sever_data def get_password(self, servertime, nonce, pubkey): """對密碼進行 RSA 的加密""" rsaPublickey = int(pubkey, 16) key = rsa.PublicKey(rsaPublickey, 65537) # 創建公鑰 message = str(servertime) + 't' + str(nonce) + 'n' + str(self.password) # 拼接明文js加密文件中得到 message = message.encode("utf-8") passwd = rsa.encrypt(message, key) # 加密 passwd = binascii.b2a_hex(passwd) # 將加密信息轉換為16進制。 return passwd def get_cha(self, pcid): """獲取驗證碼，並且用PIL打開， 1. 如果本機安裝了圖片查看軟件，也可以用 os.subprocess 的打開驗證碼 2. 可以改寫此函數接入打碼平台。 """ cha_url = "https://login.sina.com.cn/cgi/pin.php?r=" cha_url = cha_url + str(int(random.random() * 100000000)) + "&s=0&p=" cha_url = cha_url + pcid cha_page = self.session.get(cha_url, headers=headers) with open("cha.jpg", 'wb') as f: f.write(cha_page.content) f.close() try: im = Image.open("cha.jpg") im.show() im.close() except Exception as e: print(u"請到當前目錄下，找到驗證碼後輸入") def pre_login(self): # su 是加密後的用戶名 su = self.get_su() sever_data = self.get_server_data(su) servertime = sever_data["servertime"] nonce = sever_data['nonce'] rsakv = sever_data["rsakv"] pubkey = sever_data["pubkey"] showpin = sever_data["showpin"] # 這個參數的意義待探索 password_secret = self.get_password(servertime, nonce, pubkey) self.postdata = { 'entry': 'weibo', 'gateway': '1', 'from': '', 'savestate': '7', 'useticket': '1', 'pagerefer': "https://passport.weibo.com", 'vsnf': '1', 'su': su, 'service': 'miniblog', 'servertime': servertime, 'nonce': nonce, 'pwencode': 'rsa2', 'rsakv': rsakv, 'sp': password_secret, 'sr': '1366*768', 'encoding': 'UTF-8', 'prelt': '115', "cdult": "38", 'url': 'http://weibo.com/ajaxlogin.php?framelogin=1&callback=parent.sinaSSOController.feedBackUrlCallBack', 'returntype': 'TEXT' # 這裡是 TEXT 和 META 選擇，具體含義待探索 } return sever_data def login(self): # 先不輸入驗證碼登錄測試 try: sever_data = self.pre_login() login_url = 'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.19)&_' login_url = login_url + str(time.time() * 1000) login_page = self.session.post(login_url, data=self.postdata, headers=headers) ticket_js = login_page.json() ticket = ticket_js["ticket"] except Exception as e: sever_data = self.pre_login() login_url = 'https://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.4.19)&_' login_url = login_url + str(time.time() * 1000) pcid = sever_data["pcid"] self.get_cha(pcid) self.postdata['door'] = input(u"請輸入驗證碼") login_page = self.session.post(login_url, data=self.postdata, headers=headers) ticket_js = login_page.json() ticket = ticket_js["ticket"] # 以下內容是 處理登錄跳轉鏈接 save_pa = r'==-(d+)-' ssosavestate = int(re.findall(save_pa, ticket)[0]) + 3600 * 7 jump_ticket_params = { "callback": "sinaSSOController.callbackLoginStatus", "ticket": ticket, "ssosavestate": str(ssosavestate), "client": "ssologin.js(v1.4.19)", "_": str(time.time() * 1000), } jump_url = "https://passport.weibo.com/wbsso/login" jump_headers = { "Host": "passport.weibo.com", "Referer": "https://weibo.com/", "User-Agent": headers["User-Agent"] } jump_login = self.session.get(jump_url, params=jump_ticket_params, headers=jump_headers) uuid = jump_login.text uuid_pa = r'"uniqueid":"(.*?)"' uuid_res = re.findall(uuid_pa, uuid, re.S)[0] web_weibo_url = "http://weibo.com/%s/profile?topnav=1&wvr=6&is_all=1" % uuid_res weibo_page = self.session.get(web_weibo_url, headers=headers) # print(weibo_page.content.decode("utf-8") Mheaders = { "Host": "login.sina.com.cn", "User-Agent": agent } # m.weibo.cn 登錄的 url 拼接 _rand = str(time.time()) mParams = { "url": "https://m.weibo.cn/", "_rand": _rand, "gateway": "1", "service": "sinawap", "entry": "sinawap", "useticket": "1", "returntype": "META", "sudaref": "", "_client_version": "0.6.26", } murl = "https://login.sina.com.cn/sso/login.php" mhtml = self.session.get(murl, params=mParams, headers=Mheaders) mhtml.encoding = mhtml.apparent_encoding mpa = r'replace((.*?));' mres = re.findall(mpa, mhtml.text) # 關鍵的跳轉步驟，這裡不出問題，基本就成功了。 Mheaders["Host"] = "passport.weibo.cn" self.session.get(eval(mres[0]), headers=Mheaders) mlogin = self.session.get(eval(mres[0]), headers=Mheaders) # print(mlogin.status_code) # 進過幾次 頁面跳轉後，m.weibo.cn 登錄成功，下次測試是否登錄成功 Mheaders["Host"] = "m.weibo.cn" Set_url = "https://m.weibo.cn" pro = self.session.get(Set_url, headers=Mheaders) pa_login = r'isLogin":true,' login_res = re.findall(pa_login, pro.text) # print(login_res) # 可以通過 session.cookies 對 cookies 進行下一步相關操作 self.session.cookies.save() # print("*"*50) # print(self.cookie_path)
5、定義cookie的加載和信息的重定義
def get_cookies(): # 加載cookie cookies = cookielib.LWPCookieJar("Cookie.txt") cookies.load(ignore_discard=True, ignore_expires=True) # 將cookie轉換成字典 cookie_dict = requests.utils.dict_from_cookiejar(cookies) return cookie_dict def info_parser(data): id,time,text = data['id'],data['created_at'],data['text'] user = data['user'] uid,username,following,followed,gender = user['id'],user['screen_name'],user['follow_count'],user['followers_count'],user['gender'] return { 'wid':id, 'time':time, 'text':text, 'uid':uid, 'username':username, 'following':following, 'followed':followed, 'gender':gender }
6、開始爬
def start_crawl(cookie_dict,id): base_url = 'https://m.weibo.cn/comments/hotflow?id={}&mid={}&max_id_type=0' next_url = 'https://m.weibo.cn/comments/hotflow?id={}&mid={}&max_id={}&max_id_type={}' page = 1 id_type = 0 comment_count = 0 requests_count = 1 res = requests.get(url=base_url.format(id,id), headers=headers,cookies=cookie_dict) while True: print('parse page {}'.format(page)) page += 1 try: data = res.json()['data'] wdata = [] max_id = data['max_id'] for c in data['data']: comment_count += 1 row = info_parser(c) wdata.append(info_parser(c)) if c.get('comments', None): temp = [] for cc in c.get('comments'): temp.append(info_parser(cc)) wdata.append(info_parser(cc)) comment_count += 1 row['comments'] = temp print(row) with open('{}/{}.csv'.format(comment_path, id), mode='a+', encoding='utf-8-sig', newline='') as f: writer = csv.writer(f) for d in wdata: writer.writerow([d['wid'],d['time'],d['text'],d['uid'],d['username'],d['following'],d['followed'],d['gender']]) time.sleep(3) except: print(res.text) id_type += 1 print('評論總數: {}'.format(comment_count)) res = requests.get(url=next_url.format(id, id, max_id,id_type), headers=headers,cookies=cookie_dict) requests_count += 1 if requests_count%50==0: print(id_type) print(res.status_code)
7、主函數
if __name__ == '__main__': username = "18100000000" # 用戶名（註冊的手機號） password = "123456" # 密碼 cookie_path = "Cookie.txt" # 保存cookie 的文件名稱 id = '4477416430959369' # 爬取微博的 id WeiboLogin(username, password, cookie_path).login() with open('{}/{}.csv'.format(comment_path, id), mode='w', encoding='utf-8-sig', newline='') as f: writer = csv.writer(f) writer.writerow(['wid', 'time', 'text', 'uid', 'username', 'following', 'followed', 'gender']) start_crawl(get_cookies(), id)
8、獲取id
你需要獲得想要找的微博id，那麼對於小白來說怎麼找id呢？看圖說話
首先找到你想爬的微博，這裡以微博故事為例，在瀏覽器內按下F12，並且點擊評論按鈕
點擊『網絡』，找到一條像圖中的get請求。查看它的參數，mid就是它的id
