python爬虫之selenium+打码平台识别验证码

1、常用的打码平台：超级鹰、打码兔等

2、打码平台在识别图形验证码和点触验证码上比较好用

　　（1）12306点触验证码

from selenium import webdriver
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.wait import WebDriverWait
from selenium.webdriver.common.by import By
from selenium.webdriver.common.action_chains import ActionChains
from chaojiying import Chaojiying_Client
from account import USERNAME12306, PASSWORD12306, USERNAMECJY, PASSWORDCJY
import time


class TrainLogin(object):
    def __init__(self, username, password):
        self.username = username
        self.password = password
        self.index_url = "https://www.12306.cn/index/"
        self.login_url = "https://kyfw.12306.cn/otn/resources/login.html"
        self.code_path = r"E:vscodePython超级鹰验证码识别chaojiying_Python12306.png"
        self.chrome = webdriver.Chrome(r"D:chromedriver_win32chromedriver.exe")
        self.chaojiying = Chaojiying_Client(USERNAMECJY, PASSWORDCJY, "903176")

    def login(self):
        self.pre_set()
        self.come_login_url()
        self.process_username_password()
        while True:
            self.process_captchar()
            if self.success_login():
                print("成功登陆!!!")
                break
        time.sleep(5)
        self.chrome.quit()

    def pre_set(self):
        self.chrome.implicitly_wait(1)
        self.chrome.maximize_window()

    def come_login_url(self):
        self.chrome.get(self.login_url)
        WebDriverWait(self.chrome, 5).until(EC.presence_of_element_located((By.LINK_TEXT, "账号登录")))
        self.chrome.find_element_by_link_text("账号登录").click()

    def process_username_password(self):
        self.chrome.find_element_by_id("J-userName").send_keys(self.username)
        time.sleep(1)
        self.chrome.find_element_by_id("J-password").send_keys(self.password)

    def process_captchar(self):
        self.chrome.find_element_by_class_name("login-pwd-code").screenshot(self.code_path)
        with open(self.code_path, "rb") as fp:
            img = fp.read()
        res = self.chaojiying.PostPic(img, 9004)["pic_str"]
        pos_ls = self.get_postion_ls(res)
        action_chains = ActionChains(self.chrome)
        for pos in pos_ls:
            action_chains.move_to_element_with_offset(self.chrome.find_element_by_class_name("login-pwd-code"), 
            pos[0], pos[1]).click().release().perform()
            time.sleep(0.5)
            action_chains = ActionChains(self.chrome)

    def success_login(self):
        self.chrome.find_element_by_id("J-login").click()
        try:
            WebDriverWait(self.chrome, 5).until(EC.text_to_be_present_in_element((By.CLASS_NAME, "welcome-name"), "XXX"))
        except:
            return False
        else:
            return True

    def get_postion_ls(self, pos_str):
        pos_ls = []
        for pa in pos_str.split("|"):
            pos = []
            for pb in pa.split(","):
                pos.append(int(pb))
            pos = tuple(pos)

            pos_ls.append(pos)
        return pos_ls


if __name__ == "__main__":
    loginer = TrainLogin(USERNAME12306, PASSWORD12306)
    loginer.login()

　　（2）哔哩哔哩点触验证码

from selenium import webdriver
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.wait import WebDriverWait
from selenium.webdriver.common.by import By
from selenium.webdriver.common.action_chains import ActionChains
from chaojiying import Chaojiying_Client
from account import USERNAMEBL, PASSWORDBL, USERNAMECJY, PASSWORDCJY
import time

class BiLiLogin(object):
    def __init__(self, username, password):
        self.username = username
        self.password = password
        self.chrome = webdriver.Chrome(r"D:chromedriver_win32chromedriver.exe")
        self.code_path = r"E:vscodePython超级鹰验证码识别chaojiying_Pythonili.png"
        self.login_url = "https://passport.bilibili.com/login"
        self.chaojiying = Chaojiying_Client(USERNAMECJY, PASSWORDCJY, "903176")


    def login(self):
        self.pre_set()
        self.come_login_win()
        self.process_username_password()
        while True:
            self.process_captchar()
            if self.is_success():
                print("登录成功!!!")
                break
        print(self.chrome.get_cookies())
        # 延迟十秒退出
        time.sleep(10)
        self.chrome.quit()


    def pre_set(self):
        self.chrome.implicitly_wait(1)
        self.chrome.maximize_window()
    
    def come_login_win(self):
        self.chrome.get(self.login_url)

    def process_username_password(self):
        self.chrome.find_element_by_id("login-username").send_keys(self.username)
        self.chrome.find_element_by_id("login-passwd").send_keys(self.password)
        self.chrome.find_element_by_class_name("btn-login").click()

    def process_captchar(self):
        WebDriverWait(self.chrome, 5).until(EC.presence_of_element_located((By.CLASS_NAME, "geetest_widget")))
        time.sleep(5)
        self.chrome.find_element_by_class_name("geetest_widget").screenshot(self.code_path)
        with open(self.code_path, "rb") as fp:
            img = fp.read()
        pos_str = self.chaojiying.PostPic(img, "9004")["pic_str"]
        pos_ls = self.get_postion_ls(pos_str)
        for pos in pos_ls:
            action_chain = ActionChains(self.chrome)
            action_chain.move_to_element_with_offset(self.chrome.find_element_by_class_name("geetest_widget"), pos[0], pos[1]).click().release().perform()
            time.sleep(0.5)
        self.chrome.find_element_by_class_name("geetest_commit").click()

    def is_success(self):
        try:
            WebDriverWait(self.chrome, 10).until(EC.presence_of_element_located((By.CLASS_NAME, "nickname")))
        except:
            return False
        else:
            return True
        

    def get_postion_ls(self, pos_str):
        pos_ls = []
        for pa in pos_str.split("|"):
            pos = []
            for pb in pa.split(","):
                pos.append(int(pb))
            pos = tuple(pos)

            pos_ls.append(pos)
        return pos_ls
        

if __name__ == "__main__":
    loginer = BiLiLogin(USERNAMEBL, PASSWORDBL)
    loginer.login()