一个爬虫练习游戏：黑板课爬虫闯关

网址：http://www.heibanke.com/lesson/crawler_ex00/

大概用了一小天的时间吧，把这五关给过了，还挺好玩的。推荐小白玩玩。直接po代码了。

我装pytessetact总有问题，之后其实可以直接训练一个cnn来做的，但是现在先po代码了，晚上回来随便弄弄。然后我第五关验证码就是人工把图片show出来，然后人工识别输入结果，很脑残。不管了。

import requests
from lxml import etree
import re


def level_1():
    url_row = 'http://www.heibanke.com/lesson/crawler_ex00/'
    url = 'http://www.heibanke.com/lesson/crawler_ex00/10963/'

    while 1:
        try:
            html = etree.HTML(requests.get(url).text)
            a = html.xpath("//div[@class='col-xs-12 col-sm-10 col-md-8 col-lg-6']//h3")[0].text
            num = re.findall('下一个你需要输入的数字是(\d+)',a)[0]

            url = url_row+num+'/'
            print(url)
        except:
            break

def level_2():
    url = 'http://www.heibanke.com/lesson/crawler_ex01/'

    username = 'hi'
    password = 0
    while password<40:
        resp = requests.post(url, data={'username':username,'password':password})
        html = etree.HTML(resp.text)
        a = html.xpath("//div[@class='col-xs-12 col-sm-10 col-md-8 col-lg-6']//h3")[0].text
        if a=='您输入的密码错误, 请重新输入':
            password += 1
        else:
            print(password)
            break

def level_3():

    username = 'emma'
    password = 'emmaczw'
    # url = 'http://www.heibanke.com/lesson/crawler_ex02/'
    url = 'http://www.heibanke.com/accounts/login/?next=/lesson/crawler_ex02/'
    # url = 'http://www.heibanke.com/accounts/login'
    r_session = requests.Session()

    resp = r_session.get(url)
    # resp = r_session.post(url,data={'username':username,'password':password})
    # print(resp.text)
    token = resp.cookies.get('csrftoken')
    resp = r_session.post(url,data={'username':username,'password':password,'csrfmiddlewaretoken':token})
    url = resp.url
    print(url)
    username = 'hi'
    password = 20
    while password<40:
        r_session.get(url)
        token = r_session.cookies.get('csrftoken')
        resp = requests.post(url,data={'username':username,'password':password,'csrfmiddlewaretoken':token},cookies=r_session.cookies)
        print(resp.url)
        html = etree.HTML(resp.text)
        a = html.xpath("//div[@class='col-xs-12 col-sm-10 col-md-8 col-lg-6']//h3")[0].text
        if a=='您输入的密码错误, 请重新输入':
            password += 1
        else:
            print('correct password is: ',password)
            break

def level_4():
    import _thread
    import time
    password = [-1]*100
    sum_num = 0
    r_session = requests.Session()
    r_session.get('http://www.heibanke.com/lesson/crawler_ex03/pw_list/?page=1')
    # url = 'http://www.heibanke.com/lesson/crawler_ex03/pw_list/'
    url = 'http://www.heibanke.com/accounts/login/?next=/lesson/crawler_ex03/pw_list/?page='
    # url = 'http://www.heibanke.com/accounts/login/?next=/lesson/crawler_ex02/'
    r_session.cookies.get('csrftoken')

    def get_password(number):
        nonlocal sum_num
        nonlocal password
        url_temp = url+str(number)
        print(url_temp)
        resp = r_session.post(url_temp, data={'username': 'emma', 'password': 'emmaczw', 'csrfmiddlewaretoken': r_session.cookies.get('csrftoken')})
        html = etree.HTML(resp.text)
        pass_list = html.xpath("//table[@class='table table-striped']//tr")[1:]
        for k in pass_list:
            ks = k.xpath(".//td")
            index = int(ks[0].text)
            num = int(ks[1].text)
            if password[index-1] == -1:
                print(sum_num)
                sum_num += 1
                password[index-1] = num

    while sum_num<=98:
        for i in range(1,14):
            _thread.start_new_thread(get_password, (i,))
            time.sleep(1)
        time.sleep(2)

    for i in range(len(password)):
        if password[i] == -1:
            password[i] = '{}'
        else:
            password[i] = str(password[i])
    print(''.join(password))

def level_5():
    import pytesseract
    from PIL import Image
    def get_image(img_url):
        img = requests.get(img_url)
        if img.status_code == 200:
            print(type(img.content))
            with open('./img_temp.png','wb') as f:
                f.write(img.content)
            img = Image.open('./img_temp.png')
            img.show()
            img = input()
            # pytesseract.pytesseract.tesseract_cmd = "/anaconda3/lib/python3.6/site-packages/tesseract"
            # img = pytesseract.image_to_string(Image.open('./img_temp.png'), lang='eng',config='-psm 7')
            print(img)
        else:
            img = -1
        return img

    url_ = 'http://www.heibanke.com/lesson/crawler_ex04/'
    url = 'http://www.heibanke.com/accounts/login/?next=/lesson/crawler_ex04/'
    r_session = requests.Session()
    r_session.get(url)
    password = 14
    resp = r_session.post(url, data={'username': 'emma', 'password': 'emmaczw',
                                     'csrfmiddlewaretoken': r_session.cookies.get('csrftoken')})
    while password<40:
        img_url = 'http://www.heibanke.com/' + etree.HTML(resp.text).xpath("//div[@class='form-group']//img")[0].get(
            'src')
        number = get_image(img_url)
        print(img_url)
        print(img_url[-41:-1])
        if number == -1:
            print('没有找到图片')
            continue
        else:
            resp = r_session.post(url_, data={'username': 'hi', 'password': password,
                                             'csrfmiddlewaretoken': r_session.cookies.get('csrftoken'),
                                             'captcha_1':number,'captcha_0':img_url[-41:-1]})
            a = etree.HTML(resp.text).xpath("//div[@class='col-xs-12 col-sm-10 col-md-8 col-lg-6']//h3")[0].text
            if a=='您输入的密码错误, 请重新输入':
                password += 1
            else:
                print(password)
                break