def get_captcha_login(self, response):
post_data = response.meta.get("post_data", "")
try:
from PIL import Image
except:
pass
with open('captcha.gif', 'wb') as f:
f.write(response.body)
f.close()
# 用pillow 的 Image 显示验证码
# 如果没有安装 pillow 到源代码所在的目录去找到验证码然后手动输入
from zheye import zheye
z = zheye()
positions = z.Recognize('captcha.gif')
position_arr = []
if len(positions)==2:
if positions[1][1] < positions[0][1]:
position_arr.append([positions[1][1], int(positions[1][0])])
position_arr.append([positions[0][1], int(positions[0][0])])
else:
position_arr.append([positions[0][1], positions[0][0]])
position_arr.append([positions[1][1], positions[1][0]])
else:
position_arr.append([positions[0][1], positions[0][0]])
print(positions)
# try:
# im = Image.open('captcha.jpg')
# im.show()
# im.close()
# except:
# print(u'请到 %s 目录找到captcha.jpg 手动输入' % os.path.abspath('captcha.jpg'))
# captcha = input("please input the captcha\n>")
# img_width = 0
# img_height = 0
# with Image.open("captcha.jpg") as img:
# img_width, img_height = img.size
if len(positions) == 2:
post_data["captcha"] = '{"img_size": [200, 44], "input_points": [[%.2f, %.f], [%.2f, %.f]]}' % (
position_arr[0][0] / 2, position_arr[0][1] / 2, position_arr[1][0] / 2, position_arr[1][1] / 2)
else:
post_data["captcha"] = '{"img_size": [200, 44], "input_points": [[%.2f, %.f]]}' % (
position_arr[0][0] / 2, position_arr[0][1] / 2)
post_data['captcha_type'] = "cn"
print (post_data)
return [scrapy.FormRequest(
url = "https://www.zhihu.com/login/phone_num",
formdata = post_data,
headers=self.headers,
callback=self.check_login
)]
def login(self):
try:
self.browser.maximize_window() # 最大化窗口
except: # 已最大化的情况 代码会出错 捕获错误
pass
while not self.cheeck_login():
self.browser.get('https://www.zhihu.com/signin') # 打开知乎登录页面
time.sleep(5)
move(914, 329) # 点击 帐号密码登录
click()
time.sleep(2)
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a") # 全选 然后输入账户密码
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(self.user_name)
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(self.password)
move(955, 566)
click()
if self.cheeck_login():
Cookies = self.browser.get_cookies() # 获取登录成功的cookie
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
cookie_dict[cookie['name']] = cookie['value']
# self.browser.close() # 暂时不关闭
return cookie_dict
else:
try:
english_captcha_element = self.browser.find_element_by_class_name("Captcha-englishImg") # 是否出现英文验证码
except:
english_captcha_element = None
try:
chinese_captcha_element = self.browser.find_element_by_class_name("Captcha-chineseImg") # 是否出现中文验证码
except:
chinese_captcha_element = None
if chinese_captcha_element: # 如果产生中文验证码
time.sleep(1)
ele_position = chinese_captcha_element.location # 获取节点坐标
x_relative = ele_position["x"] # x坐标
y_relative = ele_position["y"] # y坐标
browser_navigation_panel_height = self.browser.execute_script(
'return window.outerHeight - window.innerHeight;') # 浏览器上栏高度
browser_navigation_panel_height = 70
time.sleep(3)
base64_text = chinese_captcha_element.get_attribute("src") # 提取中文验证码节点的arc属性
code = base64_text.replace("data:image/jpg;base64,", "").replace("%0A", "") # 消除图片bs64编码中的无用符号
fh = open("yzm_cn.jpeg", "wb") # 保存文件
fh.write(base64.b64decode(code))
fh.close()
z = zheye()
positions = z.Recognize('yzm_cn.jpeg') # 使用者也 提取倒立文字坐标
last_position = []
if len(positions) == 2:
if positions[0][0] > positions[1][0]: # 按照顺序排列倒立文字坐标
last_position.append([positions[1][0], positions[1][1]])
last_position.append([positions[0][0], positions[0][1]])
else:
last_position.append([positions[0][0], positions[0][1]])
last_position.append([positions[1][0], positions[1][0]])
if len(positions) == 2:
first_position = [int(last_position[0][1] / 2) + x_relative, int(last_position[0][
0] / 2) + y_relative + browser_navigation_panel_height] # 实际页面中 倒立文字图片为正常图片缩放的一倍 所有坐标需要除2取整 来获得可以在页面中使用的坐标
second_position = [int(last_position[1][1] / 2) + x_relative,
int(last_position[1][0] / 2) + y_relative + browser_navigation_panel_height]
move(first_position[0], first_position[1]) # 坐标 起始点x坐标+倒立文字x坐标 起始点y坐标+浏览器地址栏高度+倒立文字y坐标
click()
move(second_position[0], second_position[1])
click()
else: # 如果只有一个倒立文字
last_position.append([positions[0][1], positions[0][1]])
first_position = [int(last_position[0][1] / 2) + x_relative,
int(last_position[0][0] / 2) + browser_navigation_panel_height + y_relative]
time.sleep(5)
move(first_position[0], first_position[1])
time.sleep(5)
click()
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a") # 全选 然后输入账户密码
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(self.user_name)
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(self.password)
move(954, 619)
click()
if english_captcha_element: # 如果产生英文验证码
time.sleep(1)
base64_text = english_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,", "").replace("%0A", "") # 消除图片bs64编码中的无用符号
fh = open("yzm_en.jpeg", "wb") # 保存文件
fh.write(base64.b64decode(code))
fh.close()
cjy_cli = chaojiying.Chaojiying_Client(settings.CJY_USERNAME, settings.CJY_PASSWORD, '96001')
im = open('yzm_en.jpeg', "rb").read()
json_data = cjy_cli.PostPic(im, 1902)
if json_data["err_no"] == 0:
print("识别成功！")
code = json_data["pic_str"]
print(json_data["pic_str"])
else:
print("识别失败，继续尝试！")
return
# while True: # 若识别失败 不停识别 直至成功
# if code == "": #
# code = Yundama.decode("yzm_en.jpeg", 5000, 60)
# time.sleep(0.5)
# else:
# break
self.browser.find_element_by_xpath('//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/label/input').send_keys(Keys.CONTROL + "a") # 找到英文验证码位置
self.browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/label/input').send_keys(code)
move(956, 600)
click()
time.sleep(5)
Cookies = self.browser.get_cookies() # 获取登录成功的cookie
cookie_dict = {}
for cookie in Cookies:
cookie_dict[cookie['name']] = cookie['value']
# self.browser.close() # 暂时不关闭
return cookie_dict
def login():
"""登录，需要输入用户名和密码
"""
login_url = 'https://www.zhihu.com/signin'
print('xxxxxxxx')
browser.get(login_url)
print('xxxxxxxx')
time.sleep(1)
print('open:' + login_url)
#username, password = '', ''
username, password = '******', 'shenzhenmao18'
# 输入用户名和密码
"""
while username.strip() == '':
username = raw_input('username:'******'':
password = getpass.getpass('passord:')
"""
browser.find_elements_by_class_name('SignFlow-tab')[1].click()
#time.sleep(3)
browser.find_element_by_name('username').send_keys(username)
print('username')
#time.sleep(3)
browser.find_element_by_name('password').send_keys(password)
print('password')
#time.sleep(3)
browser.find_element_by_class_name('SignFlow-submitButton').click()
time.sleep(3)
img = None
is_ch, is_en = False, False
try:
img = browser.find_element_by_class_name('Captcha-chineseImg')
is_ch = True
except Exception as e:
try:
img = browser.find_element_by_class_name('Captcha-englishImg')
is_en = True
except Exception as e:
print('img not found')
if img is None:
print('img not found')
exit("xxx")
print("size:")
print(img.size)
print('get img src')
imgsrc = img.get_attribute('src')
print(imgsrc)
src_prefix, base64_img = imgsrc.split(',')
#imgdata = base64.b64decode(imgsrc[len('data:image/jpg;base64,'):])
missing_padding = len(base64_img) % 4
if missing_padding != 0:
base64_img += '=' * (4 - missing_padding)
base64_img = parse.unquote(base64_img)
imgdata = base64.b64decode(base64_img)
if is_ch:
pic_name = 'catcha_ch.jpg'
elif is_en:
pic_name = 'catcha_en.gif'
file = open(pic_name, 'wb')
file.write(imgdata)
file.close()
if is_ch:
z = zheye()
positions = z.Recognize(pic_name)
print(positions)
mouse_action = ActionChains(browser)
for pos in positions:
x, y = pos
print(x, y)
x, y = x / 2, y / 2
print(x, y)
mouse_action.move_to_element_with_offset(img, y,
x).click().perform()
time.sleep(2)
mouse_action.click().perform()
if is_en:
ocr(pic_name)
browser.find_element_by_class_name('SignFlow-submitButton').click()
time.sleep(1)
browser.find_element_by_class_name('SignFlow-submitButton').click()
def start_requests(self):
cookies = []
if os.path.exists(BASE_DIR + r"\cookies\zhihu.cookie"):
cookies = pickle.load(open(BASE_DIR + "/cookies/lagou.cookie", "rb"))
if not cookies:
from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.common.keys import Keys
chrome_option = Options()
chrome_option.add_argument('--disable-extensions')
chrome_option.add_experimental_option('debuggerAddress', '127.0.0.1:9222')
browser = webdriver.Chrome(executable_path=r"D:\scrapytest\ArticleSpider\venv\Scripts\chromedriver.exe",
chrome_options=chrome_option)
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_xpath('//*[@id="root"]/div/main/div/div/div[1]/div/form/div[1]/div[2]').click()
browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys("13643095504")
time.sleep(1)
browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
time.sleep(1)
browser.find_element_by_css_selector(".SignFlow-password input").send_keys("634498qxp@")
time.sleep(2)
# browser.find_element_by_css_selector(".Button SignFlow-submitButton.Button--primary Button--blue").click()
# browser.find_element_by_xpath('//*[@id = "root"]/div/main/div/div/div[1]/div/form/button').click()
move(678, 511)
click()
time.sleep(6)
login_success = False
while not login_success:
try:
notify_ele = browser.find_element_by_class_name("Popover PushNotifications AppHeader-notifications")
login_success = True
return [scrapy.Request(url=self.start_urls[0], dont_filter=True)]
except:
pass
try:
browser.maximize_window()
except:
pass
try:
english_captcha = browser.find_element_by_class_name('Captcha-englishImg')
except:
english_captcha = None
try:
chinese_captcha = browser.find_element_by_class_name('Captcha-chineseImg')
except:
chinese_captcha = None
if chinese_captcha:
ele_postion = chinese_captcha.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;'
)
base64_text = chinese_captcha.get_attribute("src")
import base64
code = base64_text.replace("data:image/jpg;base64,", '').replace("%0A", "")
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
position = z.Recognize('yzm_cn.jpeg')
last_position = []
if len(position) == 2:
if position[0][1] > position[1][1]:
last_position.append([position[1][1], position[1][0]])
last_position.append([position[0][1], position[0][0]])
else:
last_position.append([position[0][1], position[0][0]])
last_position.append([position[1][1], position[1][0]])
first_position = [int(last_position[0][0] / 2), int(last_position[0][1] / 2)]
second_position = [int(last_position[1][0] / 2), int(last_position[1][1] / 2)]
move(x_relative + first_position[0],
y_relative + browser_navigation_panel_height + first_position[1])
click()
time.sleep(2)
move(x_relative + second_position[0],
y_relative + browser_navigation_panel_height + second_position[1])
click()
# for url in self.start_urls:
# yield scrapy.Request(url, dont_filter=True, headers=self.headers)
else:
last_position.append([position[0][1], position[0][0]])
first_position = [int(last_position[0][0] / 2), int(last_position[0][1] / 2)]
move(x_relative + first_position[0],
y_relative + browser_navigation_panel_height + first_position[1])
click()
time.sleep(1)
move(663, 569)
click()
if english_captcha:
base64_text = english_captcha.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,', '').replace("%0A", "")
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from tools.yundama_requests import YDMHttp
yundama = YDMHttp("zzzzqxp", "634498qxp", 8954, "fd03eddd0dc7ebe6eb4ce5c00012bb31")
code = yundama.decode("yzm_en.jpeg", 5000, 60)
while True:
if code == "":
code = yundama.decode("yzm_en.jpeg", 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/div[1]/input').send_keys(
Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/div[1]/input').send_keys(code)
time.sleep(1)
move(663, 544)
click()
time.sleep(2)
time.sleep(1)
browser.get("https://www.zhihu.com/")
cookies = browser.get_cookies()
pickle.dump(cookies,open(r'D:\scrapytest\ArticleSpider\cookies\zhihu.cookie','wb'))
cookie_dict={}
for cookie in cookies:
cookie_dict[cookie["name"]]=cookie["value"]#cookie储存到本地后就可以在开始打开获取，就不用seleniun
return [scrapy.Request(url=self.start_urls[0],dont_filter=True,cookies=cookie_dict,headers=self.headers)]
time.sleep(1)
browser.get("https://www.zhihu.com/")
cookies = browser.get_cookies()
pickle.dump(cookies, open(BASE_DIR + r"\cookies\zhihu.cookie", 'wb'))
cookie_dict = {}
for cookie in cookies:
cookie_dict[cookie["name"]] = cookie["value"] # cookie储存到本地后就可以在开始打开获取，就不用seleniun
return [scrapy.Request(url=self.start_urls[0], dont_filter=True, cookies=cookie_dict, headers=self.headers)]
def start_requests(self):
"""
用selenium完成知乎自动登陆
"""
chrome_options = Options()
chrome_options.add_argument('--no-sandbox')
chrome_options.add_argument('--disable-extensions')
chrome_options.add_experimental_option('debuggerAddress',
'127.0.0.1:9222')
browser = webdriver.Chrome(
executable_path='/usr/local/chromedriver/chromedriver',
chrome_options=chrome_options)
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin")
try:
# 未登录知乎
time.sleep(3)
# 清空默认填充值,clear()方法失效，使用control+a全选再输入,全选之前先等待3秒让浏览器加载了默认填充值
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)').send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)').send_keys(
"15986744115")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys('**********')
# css捕获’登陆‘按钮并鼠标点击
browser.find_element_by_css_selector(
'button.Button:nth-child(5)').click()
time.sleep(5)
login_success = False
while not login_success:
try:
notify_element = browser.find_element_by_css_selector(
'.Zi--Bell > path:nth-child(1)')
login_success = True
except:
pass
try:
english_captcha_element = browser.find_element_by_class_name(
'Captcha-englishImg')
except:
english_captcha_element = None
try:
chinese_cptcha_element = browser.find_element_by_class_name(
'Captcha-chineseImg')
except:
chinese_cptcha_element = None
if chinese_cptcha_element:
element_position = chinese_cptcha_element.location
x_relative = element_position['x']
y_relative = element_position['y']
# browser_navigtion_panel_height = browser.execute_script(
# 'return window.outerHeight - window.innerHeight;'
# )
browser_navigtion_panel_height = 103
base64_text = chinese_cptcha_element.get_attribute('src')
code = base64_text.replace('data:image/jpg;base64,',
'').replace('%0A', '')
fh = open('yzm_cn.jpeg', 'wb')
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
try:
positions = z.Recognize('yzm_cn.jpeg')
except:
time.sleep(3)
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)'
).send_keys("15986744115")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys('29q82q8t7q')
# css捕获’登陆‘按钮并鼠标点击
browser.find_element_by_css_selector(
'button.Button:nth-child(5)').click()
time.sleep(5)
continue
last_position = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0] / 2),
int(last_position[1][1] / 2)
]
time.sleep(1)
pyautogui.moveTo(x_relative + first_position[0],
y_relative +
browser_navigtion_panel_height +
first_position[1],
duration=0.5)
pyautogui.click()
time.sleep(1)
pyautogui.moveTo(x_relative + second_position[0],
y_relative +
browser_navigtion_panel_height +
second_position[1],
duration=0.5)
pyautogui.click()
else:
last_position.append(
[positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
time.sleep(1)
pyautogui.moveTo(x_relative + first_position[0],
y_relative +
browser_navigtion_panel_height +
first_position[1],
duration=0.5)
pyautogui.click()
if english_captcha_element:
base64_text = english_captcha_element.get_attribute('src')
code = base64_text.replace('data:image/jpg;base64,',
'').replace('%0A', '').replace(
r'\n', '')
fh = open('yzm_en.jpeg', 'wb')
fh.write(base64.b64decode(code))
fh.close()
from tool.yundama_requests import YDMHttp
yundama = YDMHttp('915263031', 'slzkszkbjz', 7346,
'8da3657e176de5e60bce4bf449eae130')
code = yundama.decode('yzm_en.jpeg', 5000, 60)
while True:
if code == '':
code = yundama.decode('yzm_en.jpeg', 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(code)
if english_captcha_element or chinese_cptcha_element:
# 再次输入用户名和密码
time.sleep(3)
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'.SignFlow-accountInput > input:nth-child(1)'
).send_keys("15986744115")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
'div.SignFlowInput:nth-child(1) > div:nth-child(1) > input:nth-child(1)'
).send_keys('29q82q8t7q')
# css捕获’登陆‘按钮并鼠标点击
browser.find_element_by_css_selector(
'button.Button:nth-child(5)').click()
time.sleep(5)
except:
# 已登录 不做任何操作
pass
cookies = browser.get_cookies()
cookie_dict = {}
for cookie in cookies:
cookie_dict[cookie['name']] = cookie['value']
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
def login(self):
import time
try:
self.browser.maximize_window() # 将窗口最大化防止定位错误
except Exception as e:
pass
while not self.check_login():
self.browser.get("https://www.zhihu.com/signin")
#选择用户密码登陆元素并点击
time.sleep(5)
login_element = self.browser.find_element_by_css_selector(".SignFlow-tabs div.SignFlow-tab")
login_element.click()
browser_navigation_panel_height = self.browser.execute_script('return window.outerHeight - window.innerHeight;')
time.sleep(5)
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
self.user_name)
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(
self.pass_word)
self.browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(15)
from mouse import move, click
print("判断登录是否成功")
if self.check_login():
break
try:
# 查询是否有英文验证码
english_captcha_element = self.browser.find_element_by_class_name("Captcha-englishImg")
except:
english_captcha_element = None
try:
# 查询是否有中文验证码
chinese_captcha_element = self.browser.find_element_by_class_name("Captcha-chineseImg")
except:
chinese_captcha_element = None
if chinese_captcha_element:
y_relative_coord = chinese_captcha_element.location['y']
y_absolute_coord = y_relative_coord + browser_navigation_panel_height
x_absolute_coord = chinese_captcha_element.location['x']
"""
保存图片
1. 通过保存base64编码
2. 通过crop方法
"""
# 1. 通过保存base64编码
base64_text = chinese_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,', '').replace("%0A", "")
# print code
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize("yzm_cn.jpeg")
pos_arr = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
pos_arr.append([positions[1][1], positions[1][0]])
pos_arr.append([positions[0][1], positions[0][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
pos_arr.append([positions[1][1], positions[1][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
if len(positions) == 2:
first_point = [int(pos_arr[0][0] / 2), int(pos_arr[0][1] / 2)]
second_point = [int(pos_arr[1][0] / 2), int(pos_arr[1][1] / 2)]
move((x_absolute_coord + first_point[0]), y_absolute_coord + first_point[1])
click()
move((x_absolute_coord + second_point[0]), y_absolute_coord + second_point[1])
click()
else:
first_point = [int(pos_arr[0][0] / 2), int(pos_arr[0][1] / 2)]
move((x_absolute_coord + first_point[0]), y_absolute_coord + first_point[1])
click()
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
self.user_name)
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(
self.pass_word)
self.browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
self.browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
if english_captcha_element:
# 2. 通过crop方法
# from pil import Image
# image = Image.open(path)
# image = image.crop((locations["x"], locations["y"], locations["x"] + image_size["width"],
# locations["y"] + image_size["height"])) # defines crop points
#
# rgb_im = image.convert('RGB')
# rgb_im.save("D:/ImoocProjects/python_scrapy/coding-92/ArticleSpider/tools/image/yzm.jpeg",
# 'jpeg') # saves new cropped image
# # 1. 通过保存base64编码
base64_text = english_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,', '').replace("%0A", "")
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
cjy_cli = chaojiying.Chaojiying_Client(settings.CJY_USERNAME, settings.CJY_PASSWORD, '96001')
im = open("yzm_en.jpeg", 'rb').read()
json_data = cjy_cli.PostPic(im, 1902)
if json_data["err_no"] == 0:
print("识别成功！")
code = json_data["pic_str"]
print("英文验证码： {code}".format(code=code))
else:
print("识别失败，继续尝试！")
return
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input').send_keys(code)
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
self.user_name)
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
self.browser.find_element_by_css_selector(".SignFlow-password input").send_keys(
self.pass_word)
submit_ele = self.browser.find_element_by_css_selector(".Button.SignFlow-submitButton")
self.browser.find_element_by_css_selector(".Button.SignFlow-submitButton").click()
# 等待登录成功后加载个人中心信息
time.sleep(10)
Cookies = self.browser.get_cookies()
print(Cookies)
cookie_dict = {}
for cookie in Cookies:
cookie_dict[cookie['name']] = cookie['value']
self.browser.close()
return cookie_dict
def start_requests(self):
# 手动启动本地Chrome
from selenium.webdriver.chrome.options import Options
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
chrome_options.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
# 建立浏览器对象
browser = webdriver.Chrome(
executable_path="/usr/local/bin/chromedriver",
chrome_options=chrome_options)
try:
# 窗口最大化
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin")
# 浏览器执行js代码获取浏览器工具栏高度
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
# browser_navigation_panel_height = 71
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL, "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"18612345678")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL, "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("123456")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
# 等待加载完成
time.sleep(5)
# 判断是否登录成功
login_success = False
while not login_success:
try:
# 如果能找到某个元素证明登录成功
notify_element = browser.find_element_by_class_name(
"Popover PushNotifications AppHeader-notifications")
login_success = True
except:
pass
try:
# 查询是否有英文验证码
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg")
except:
english_captcha_element = None
try:
# 查询是否有中文验证码
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg")
except:
chinese_captcha_element = None
if chinese_captcha_element:
y_relative_coord = chinese_captcha_element.location['y']
y_absolute_coord = y_relative_coord + browser_navigation_panel_height
x_absolute_coord = chinese_captcha_element.location['x']
# x_absolute_coord = 842
# y_absolute_coord = 428
"""
保存图片
1. 通过保存base64编码
2. 通过crop方法
"""
# 1. 通过保存base64编码
base64_text = chinese_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
# print code
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize("yzm_cn.jpeg")
pos_arr = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
pos_arr.append([positions[1][1], positions[1][0]])
pos_arr.append([positions[0][1], positions[0][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
pos_arr.append([positions[1][1], positions[1][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
if len(positions) == 2:
# 保存到本地的图片长宽是原图的两倍
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
second_point = [
int(pos_arr[1][0] / 2),
int(pos_arr[1][1] / 2)
]
# move((x_absolute_coord + first_point[0]), y_absolute_coord + first_point[1])
# click()
#
# move((x_absolute_coord + second_point[0]), y_absolute_coord + second_point[1])
# click()
pyautogui.moveTo((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
pyautogui.click()
else:
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
# move((x_absolute_coord + first_point[0]), y_absolute_coord + first_point[1])
# click()
pyautogui.moveTo((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
pyautogui.click()
# 输入验证码后重新登录
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
def start_requests(self):
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
# chrome_options.add_experimental_option("excludeSwitches", ['enable-automation'])
chrome_options.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(chrome_options=chrome_options)
try:
browser.maximize_window() # 将窗口最大化防止定位错误
except:
pass
browser.get("https://www.zhihu.com/signin")
# logo_element = browser.find_element_by_class_name("SignFlowHeader")
# y_relative_coord = logo_element.location['y']
#此处一定不要将浏览器放大 会造成高度获取失败！！！
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
time.sleep(5)
browser.implicitly_wait(5)
#用来测试验证码，特意输入错误的密码
#
# try:
# tabs = browser.find_elements_by_css_selector(
# ".SignFlow-tabs div")
# tabs[1].click()
# except:
# pass
# browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(Keys.CONTROL + "a")
# browser.find_element_by_css_selector(".SignFlow-accountInput.Input-wrapper input").send_keys(
# "17628040175")
#
# browser.find_element_by_css_selector(".SignFlow-password input").send_keys(Keys.CONTROL + "a")
# browser.find_element_by_css_selector(".SignFlow-password input").send_keys(
# "2014@ltg1")
# browser.find_element_by_css_selector(
# ".Button.SignFlow-submitButton").click()
# browser.implicitly_wait(5)
# 先判断是否登录成功
login_success = False
while not login_success:
try:
notify_element = browser.find_element_by_class_name(
"AppHeader-profile")
login_success = True
except:
pass
try:
tabs = browser.find_elements_by_css_selector(
".SignFlow-tabs div")
tabs[1].click()
except:
pass
try:
#查询是否有英文验证码
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg")
base64_text = english_captcha_element.get_attribute("src")
#'data:image/jpg;base64,null'
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
if code == 'null':
english_captcha_element = None
except:
english_captcha_element = None
try:
# 查询是否有中文验证码
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg")
base64_text = chinese_captcha_element.get_attribute("src")
#'data:image/jpg;base64,null'
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
if code == 'null':
chinese_captcha_element = None
except:
chinese_captcha_element = None
if not english_captcha_element and not chinese_captcha_element and not login_success:
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"17628040175")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("2014@ltg")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
browser.implicitly_wait(5)
if chinese_captcha_element:
y_relative_coord = chinese_captcha_element.location['y']
y_absolute_coord = y_relative_coord + browser_navigation_panel_height
x_absolute_coord = chinese_captcha_element.location['x']
# x_absolute_coord = 842
# y_absolute_coord = 428
"""
保存图片
1. 通过保存base64编码
2. 通过crop方法
"""
# 1. 通过保存base64编码
# base64_text = chinese_captcha_element.get_attribute("src")
# #'data:image/jpg;base64,null'
# code = base64_text.replace('data:image/jpg;base64,', '').replace("%0A", "")
# print code
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize("yzm_cn.jpeg")
pos_arr = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
pos_arr.append([positions[1][1], positions[1][0]])
pos_arr.append([positions[0][1], positions[0][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
pos_arr.append([positions[1][1], positions[1][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
if len(positions) == 2:
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
second_point = [
int(pos_arr[1][0] / 2),
int(pos_arr[1][1] / 2)
]
move((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
click()
move((x_absolute_coord + second_point[0]),
y_absolute_coord + second_point[1])
click()
else:
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
move((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"17628040175")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("2014@ltg")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
if english_captcha_element:
# 2. 通过crop方法
# from pil import Image
# image = Image.open(path)
# image = image.crop((locations["x"], locations["y"], locations["x"] + image_size["width"],
# locations["y"] + image_size["height"])) # defines crop points
#
# rgb_im = image.convert('RGB')
# rgb_im.save("D:/ImoocProjects/python_scrapy/coding-92/ArticleSpider/tools/image/yzm.jpeg",
# 'jpeg') # saves new cropped image
# # 1. 通过保存base64编码
# base64_text = english_captcha_element.get_attribute("src")
# code = base64_text.replace('data:image/jpg;base64,', '').replace("%0A", "")
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
# from tools.yundama_requests import YDMHttp
# yundama = YDMHttp("da_ge_da1", "dageda", 3129, "40d5ad41c047179fc797631e3b9c3025")
# code = yundama.decode("yzm_en.jpeg", 5000, 60)
# while True:
# if code == "":
# code = yundama.decode("yzm_en.jpeg", 5000, 60)
# else:
# break
chaojiying = Chaojiying_Client(
'yiqieanran01', '1qazxsw23edc',
'904611') # 用户中心>>软件ID 生成一个替换 96001
im = open('yzm_en.jpeg',
'rb').read() # 本地图片文件路径 来替换 a.jpg 有时WIN系统须要//
code = chaojiying.PostPic(im, 1902)['pic_str']
while True:
if code == '':
code = chaojiying.PostPic(im, 1902)['pic_str']
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(code)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"17628040175")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("2014@ltg")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
browser.implicitly_wait(5)
try:
notify_element = browser.find_element_by_class_name(
"AppHeader-profile")
login_success = True
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
import pickle
f = open(
'C:\\LTG\\code\\ArticleSpider\\ArticleSpider\\cookies\\zhihu.cookie',
'wb')
pickle.dump(Cookies, f)
f.close()
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
# f = open('C:\\LTG\\code\\ArticleSpider\\ArticleSpider\\cookies\\' + cookie['name'] + '.zhihu', 'wb')
# pickle.dump(cookie, f)
# f.close()
cookie_dict[cookie['name']] = cookie['value']
# browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
except:
pass
print("yes")
def get_pwd_login(self):
t = random.uniform(0.5, 1)
pwd_login = self.driver.find_element_by_xpath(
'//div[@class="SignFlow-tab"]')
pwd_login.click()
time.sleep(1)
username = self.driver.find_element_by_xpath(
'//div[@class="SignFlow-account"]/div/label/input')
username.send_keys('账号')
time.sleep(t)
password = self.driver.find_element_by_xpath(
'//div[@class="SignFlow-password"]/div/label/input')
password.send_keys('密码')
time.sleep(t)
login_button = self.driver.find_element_by_xpath(
'//button[@class="Button SignFlow-submitButton Button--primary Button--blue"]'
)
login_button.click()
time.sleep(t)
try:
image_code = self.driver.find_element_by_xpath(
'//div[@class="Captcha-englishContainer"]/img').get_attribute(
'src')
print(image_code)
if 'null' not in str(image_code):
print('******************')
'''
# 截取整个网页图片
self.driver.save_screenshot(r'./Images/zh_full.png')
# 定位验证码位置
image_code_input = self.driver.find_element_by_xpath('//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/div[1]/input')
# 获取验证码x,y轴 尺寸
location = image_code_input.location
# 获取验证码的长和宽
size = image_code_input.size
# 截取的验证码位置坐标
left, top, right, bottom = location['x'], location['y'], location['x']+size['width'], location['y'] + size['height']
image_full = Image.open(r'./Images/zh_full.png')
# 使用Image的crop函数，从截图中再次截取我们需要的区域
image_code_pic = image_full.crop((left, top, right, bottom))
'''
print('^^^^^^^^^^^^^^^^^')
# 保存截取的验证码图片
request.urlretrieve(image_code, './Images/zh_code.png')
time.sleep(1)
# 调用打码平台解析验证码， 缺少代码 如果打码平台解析错误需要做出判断并重新调用，且设置重调次数
image_code_cont = identify(r'./Images/zh_code.png')
print(image_code_cont)
# 定位验证码填写位置
image_code_input = self.driver.find_element_by_xpath(
'//div[@class="Captcha SignFlow-captchaContainer"]/div/div/label/input'
)
image_code_input.send_keys(image_code_cont)
except Exception as e:
if '//div[@class="Captcha-englishContainer"]' in str(e):
pass
else:
print(e)
try:
zh_code_cn = self.driver.find_element_by_xpath(
'//div[@class="Captcha-chineseContainer"]/img').get_attribute(
'src')
# print(zh_code_cn)
if 'null' not in str(zh_code_cn):
'''
request.urlretrieve(zh_code_cn, './Images/zh_code_cn.png')
time.sleep(5) # 睡5秒-->人工点击（目前还没有好的解决方法）
'''
# 定位抓取文字验证码图片
chinese_captcha_element = self.driver.find_element_by_class_name(
"Captcha-chineseImg")
ele_postion = chinese_captcha_element.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
browser_navigation_panel_height = 70
base64_text = chinese_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
fh = open("./Images/zh_code_cn.png", "wb")
fh.write(base64.b64decode(code))
fh.close()
# 调用zheye库
from zheye import zheye
from mouse import move, click
z = zheye()
positions = z.Recognize('./Images/zh_code_cn.png')
print(positions)
last_position = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0] / 2),
int(last_position[1][1] / 2)
]
# +15 是x轴坐标上右移15，+30是y轴下移30
move(
x_relative + first_position[0] + 15,
y_relative + browser_navigation_panel_height +
first_position[1] + 30)
click()
move(
x_relative + second_position[0] + 15,
y_relative + browser_navigation_panel_height +
second_position[1] + 30)
click()
else:
last_position.append([positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
move(
x_relative + first_position[0] + 15,
y_relative + browser_navigation_panel_height +
first_position[1] + 30)
click()
print(last_position)
except Exception as e:
if '//div[@class="Captcha-chineseContainer"]' in str(e):
pass
else:
print(e)
time.sleep(3)
login_button = self.driver.find_element_by_xpath(
'//button[@class="Button SignFlow-submitButton Button--primary Button--blue"]'
)
# login_button.click()
ActionChains(
self.driver).move_to_element(login_button).click().perform()
time.sleep(2)
print('------------')
获取cookie
cookie = {}
cookies = self.driver.get_cookies()
for i in cookies:
a = i['name']
b = i['value']
cookie[a] = b
return cookie
def start_requests(self):
from selenium import webdriver
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.chrome.options import Options
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
#chrome_options.add_experimental_option("debuggerAddress", "127.0.0.1:9222")
#browser = webdriver.Chrome(executable_path="E:/chromedriver/chromedriver_win32/chromedriver.exe", chrome_options=chrome_options)
browser = webdriver.Chrome(
executable_path=
"E:/chromedriver/chromedriver_win32/chromedriver.exe")
import time
try:
browser.maximize_window() #将窗口最大化防止定位错误
except:
pass
browser.get("https://www.zhihu.com/signin")
logo_element = browser.find_element_by_class_name("SignFlowHeader")
# y_relative_coord = logo_element.location['y']
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
browser_navigation_panel_height = 71
time.sleep(5)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys("xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(15)
from mouse import move, click
# move(800, 400 ,True)
# actions = ActionChains(browser)
# actions.move_to_element(browser.find_element_by_css_selector(
# ".Button.SignFlow-submitButton"))
# actions.click(browser.find_element_by_css_selector(
# ".Button.SignFlow-submitButton"))
# actions.perform()
# actions.move_to_element_with_offset(browser.find_element_by_css_selector(
# ".Button.SignFlow-submitButton"), 30, 30).perform()
#chrome的版本问题有两种解决方案
#1. 自己启动chrome(推荐) 可以防止chromedriver被识别，因为chromedriver出现的一些js变量可以被服务器识别出来
#2. 使用chrome60(版本)
# 先判断是否登录成功
login_success = False
while not login_success:
try:
notify_element = browser.find_element_by_class_name(
"Popover PushNotifications AppHeader-notifications")
login_success = True
except:
pass
try:
#查询是否有英文验证码
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg")
except:
english_captcha_element = None
try:
# 查询是否有中文验证码
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg")
except:
chinese_captcha_element = None
if chinese_captcha_element:
y_relative_coord = chinese_captcha_element.location['y']
y_absolute_coord = y_relative_coord + browser_navigation_panel_height
x_absolute_coord = chinese_captcha_element.location['x']
# x_absolute_coord = 842
# y_absolute_coord = 428
"""
保存图片
1. 通过保存base64编码
2. 通过crop方法
"""
# 1. 通过保存base64编码
base64_text = chinese_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
# print code
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize("yzm_cn.jpeg")
pos_arr = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
pos_arr.append([positions[1][1], positions[1][0]])
pos_arr.append([positions[0][1], positions[0][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
pos_arr.append([positions[1][1], positions[1][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
if len(positions) == 2:
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
second_point = [
int(pos_arr[1][0] / 2),
int(pos_arr[1][1] / 2)
]
move((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
click()
move((x_absolute_coord + second_point[0]),
y_absolute_coord + second_point[1])
click()
else:
first_point = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
]
move((x_absolute_coord + first_point[0]),
y_absolute_coord + first_point[1])
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
if english_captcha_element:
# 2. 通过crop方法
# from pil import Image
# image = Image.open(path)
# image = image.crop((locations["x"], locations["y"], locations["x"] + image_size["width"],
# locations["y"] + image_size["height"])) # defines crop points
#
# rgb_im = image.convert('RGB')
# rgb_im.save("D:/ImoocProjects/python_scrapy/coding-92/ArticleSpider/tools/image/yzm.jpeg",
# 'jpeg') # saves new cropped image
# # 1. 通过保存base64编码
base64_text = english_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from tools.yundama_requests import YDMHttp
yundama = YDMHttp("da_ge_da1", "dageda", 3129,
"40d5ad41c047179fc797631e3b9c3025")
code = yundama.decode("yzm_en.jpeg", 5000, 60)
while True:
if code == "":
code = yundama.decode("yzm_en.jpeg", 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(code)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
submit_ele = browser.find_element_by_css_selector(
".Button.SignFlow-submitButton")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(10)
try:
notify_element = browser.find_element_by_class_name(
"Popover PushNotifications AppHeader-notifications")
login_success = True
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
f = open(
'./ArticleSpider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
except:
pass
print("yes")
def start_requests(self):
# cookies = pickle.load(open("C:/Users/Administrator/ZhihuSpider/cookies/zhihu.cookie", "rb"))
# cookie_dict = {}
# for cookie in cookies:
# cookie_dict[cookie["name"]] = cookie["value"]
chromeOption = Options()
chromeOption.add_argument("--disable-extensions")
chromeOption.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(
executable_path="I:/chromedriver/chromedriver.exe",
chrome_options=chromeOption)
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_css_selector(
".SignFlow-tabs div:nth-child(2)").click()
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"13247598671")
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys("156416421727av.")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(10)
logon_success = False # 登录状态
while not logon_success:
# 登陆成功
try:
# notify_ele = browser.find_element_by_class_name("Popover PushNotifications AppHeader-notifications")
notify_ele = browser.find_element_by_id("Popover17-toggle")
logon_success = True
except:
pass
# 登录失败出现中文验证码
try:
chinese_captcha_ele = browser.find_element_by_class_name(
"Captcha-chineseImg")
except:
chinese_captcha_ele = None
# 登录失败出现英文验证码
try:
english_captcha_ele = browser.find_element_by_class_name(
"Captcha-englishImg")
except:
english_captcha_ele = None
# 处理中文验证码
if chinese_captcha_ele:
ele_postion = chinese_captcha_ele.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
js = "return (window.outerHeight - window.innerHeight);"
browser_navigation_panel_height = browser.execute_script(js)
# 获取中文验证码图片（base64）解码，并写入二进制文件
base64_text = chinese_captcha_ele.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
f = open("captcha_cn.jpeg", "wb")
f.write(base64.b64decode(code))
f.close()
# 自动识别中文验证码
z = zheye()
positions = z.Recognize('captcha_cn.jpeg')
last_position = []
# 有两个倒立文字
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_pos = [
int(last_position[0][0]) / 2,
int(last_position[0][1]) / 2
]
second_pos = [
int(last_position[1][0]) / 2,
int(last_position[1][1]) / 2
]
move(
x_relative + first_pos[0], y_relative + first_pos[1] +
browser_navigation_panel_height)
click()
time.sleep(3)
move(
x_relative + second_pos[0], y_relative +
second_pos[1] + browser_navigation_panel_height)
click()
# 只有一个倒立文字
else:
last_position.append([positions[0][1], positions[0][0]])
first_pos = [
int(last_position[0][0]) / 2,
int(last_position[0][1]) / 2
]
move(
x_relative + first_pos[0], y_relative + first_pos[1] +
browser_navigation_panel_height)
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"13247598671")
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys("156416421727av")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
# 处理英文验证码
if english_captcha_ele:
base64_text = english_captcha_ele.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
f = open("captcha_en.jpeg", "wb")
f.write(base64.b64decode(code))
f.close()
# 使用在线云打码
yundama = YDMHttp('swartz2324', '156416421727av.', 9561,
'd9acf25c4e7f52926f4008a55b4080a3')
captcha_result = yundama.decode('captcha_en.jpeg', 5000, 60)
while True:
time.sleep(1)
if captcha_result == '':
captcha_result = yundama.decode(
'captcha_en.jpeg', 5000, 60)
else:
break
browser.find_element_by_css_selector(
".Captcha.SignFlow-captchaContainer .SignFlowInput .Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".Captcha.SignFlow-captchaContainer .SignFlowInput .Input-wrapper input"
).send_keys(captcha_result)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"13247598671")
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password .SignFlowInput .Input-wrapper input"
).send_keys("156416421727av")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(3)
# 登录成功后把cookies保存到本地
cookies = browser.get_cookies()
pickle.dump(
cookies,
open("C:/Users/Administrator/ZhihuSpider/cookies/zhihu.cookie",
"wb"))
cookie_dict = {}
for cookie in cookies:
cookie_dict[cookie["name"]] = cookie["value"]
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
def judge_login(self, response):
self.browser.get('https://www.zhihu.com/signin')
if self.__is_login():
return [scrapy.Request(url=self.start_urls[0], dont_filter=True)]
action = ActionChains(self.browser)
self.browser.find_element_by_css_selector(
'.SignFlow-tabs .SignFlow-tab:nth-child(2)').click()
self.__inputLoginInfo(password=ZHIHU_PASSWORD + '1')
time.sleep(3)
login_success = False
while not login_success:
try:
if self.__is_login():
login_success = True
except:
pass
try:
english_captcha_element = self.browser.find_element_by_class_name(
'Captcha-englishImg')
except:
english_captcha_element = None
try:
chinese_captcha_element = self.browser.find_element_by_class_name(
'Captcha-chineseImg')
except:
chinese_captcha_element = None
if chinese_captcha_element:
if action._actions:
action._actions = []
self.browser.execute_script(
'document.querySelectorAll(".Button.ChineseCaptchaPoint").forEach(function(el){el.click()});')
# chinese_captcha_point = self.browser.find_element_by_class_name('')
# ele_position = chinese_captcha_element.location
# x_relative = ele_position['x']
# y_relative = ele_position['y']
# self.browser_navigation_panel_height = self.browser.execute_script(
# 'return window.outerHeight - window.innerHeight'
# )
base64_text = chinese_captcha_element.get_attribute('src')
import base64
code = base64_text.replace(
'data:image/jpg;base64,', '').replace('%0A', '')
fh = open('yzm_cn.jpeg', 'wb')
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize('yzm_cn.jpeg')
last_position = []
# 格式化倒立文字的坐标（zheye默认坐标返回为y,x的格式）
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2), int(last_position[0][1] / 2)]
second_position = [
int(last_position[1][0] / 2), int(last_position[1][1] / 2)]
action.move_to_element_with_offset(
chinese_captcha_element, first_position[0], first_position[1]).click()
action.move_to_element_with_offset(
chinese_captcha_element, second_position[0], second_position[1]).click().perform()
else:
last_position.append([positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2), int(last_position[0][1] / 2)]
action.move_to_element_with_offset(
chinese_captcha_element, first_position[0], first_position[1]).click().perform()
print(last_position)
self.__inputLoginInfo()
time.sleep(3)
if english_captcha_element:
base64_text = english_captcha_element.get_attribute('src')
import base64
code = base64_text.replace(
'data:image/jpg;base64,', '').replace('%0A', '')
fh = open('yzm_en.jpeg', 'wb')
fh.write(base64.b64decode(code))
fh.close()
from tools.fateadm_api import FateadmApi
from ScrapyDemo.settings import (
ff_app_id, ff_app_key, ff_pd_id, ff_pd_key)
api = FateadmApi(ff_app_id, ff_app_key, ff_pd_id, ff_pd_key)
pred_type = "30400"
# 返回详细识别结果
rsp = api.PredictFromFile(pred_type, 'yzm_en.jpeg')
while True:
if not rsp or rsp.ret_code != 0 or not rsp.pred_rsp or not rsp.pred_rsp.value:
rsp = api.PredictFromFile(pred_type, 'yzm_en.jpeg')
else:
break
self.__clear_input(
'.Captcha.SignFlow-captchaContainer input[name="captcha"]')
self.browser.find_element_by_css_selector(
'.Captcha.SignFlow-captchaContainer input[name="captcha"]').send_keys(rsp.pred_rsp.value)
self.__inputLoginInfo()
time.sleep(3)
cookies = self.browser.get_cookies()
pickle.dump(cookies, open(
'./cookies/zhihu.cookie', 'wb'))
cookie_dict = {}
for cookie in cookies:
cookie_dict[cookie['name']] = cookie['value']
return [scrapy.Request(url=self.start_urls[0], dont_filter=True, cookies=cookie_dict)]
def start_requests(self):
from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.common.keys import Keys
chrome_option = Options()
chrome_option.add_argument("--disable-extensions")
chrome_option.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(
executable_path=
"E:/chromedriver/chromedriver_win32/chromedriver.exe",
chrome_options=chrome_option)
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys("xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(10)
login_success = False
if login_success:
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
f = open(
'./ArticleSpider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
while not login_success:
try:
notify_ele = browser.find_element_by_class_name(
"Popover PushNotifications AppHeader-notifications")
login_success = True
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
f = open(
'./ArticleSpider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
except:
pass
try:
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg")
except:
english_captcha_element = None
try:
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg")
except:
chinese_captcha_element = None
if chinese_captcha_element:
ele_postion = chinese_captcha_element.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
base64_text = chinese_captcha_element.get_attribute("src")
import base64
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize('yzm_cn.jpeg')
last_position = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0] / 2),
int(last_position[1][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
browser_navigation_panel_height + first_position[1])
click()
time.sleep(3)
move(
x_relative + second_position[0], y_relative +
browser_navigation_panel_height + second_position[1])
click()
else:
last_position.append([positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
browser_navigation_panel_height + first_position[1])
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"18782902568")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("admin1234")
move(911, 643)
click()
if english_captcha_element:
base64_text = english_captcha_element.get_attribute("src")
import base64
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from tools.yundama_requests import YDMHttp
yundama = YDMHttp("xxx", "xxx", 3129, "xxx")
code = yundama.decode("yzm_en.jpeg", 5000, 60)
while True:
if code == "":
code = yundama.decode("yzm_en.jpeg", 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(code)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"xxx")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
move(895, 603)
click()
def selenuim_parse(self, url):
""" 使用selenium模拟登录知乎网站 """
# 若用文件保存cookie则取消一下注释，并注释if self.statu_code == 401 or len(self.account_item) < 5:
# 使用文件保持则无法使用自定义账号管理池类，需将相关内容注释，即与self.account_item相关内容
# if not bool(os.path.exists(COOKIES_PATH+'/cookies.json')):
# with open(COOKIES_PATH + '/cookies.json', 'w') as f:
# pass
# with open(COOKIES_PATH+'/cookies.json', 'r') as f:
# try:
# cookie = json.load(f)
# except:
# cookie = None
# if not bool(cookie):
if self.statu_code == 401 or len(self.account_item) < 5:
options = Options()
options.add_argument('--disable-extensions')
# 将调试selenium的端口设置为9222
options.add_experimental_option('debuggerAddress',
'127.0.0.1:9222')
browser = webdriver.Chrome(executable_path=CHROMEDRIVER_PATH,
chrome_options=options)
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin?next=%2F")
browser.find_element_by_xpath(
'//div[not(contains(text(),"免")) and contains(text(),"密码")]'
).click()
browser.find_element_by_xpath(
'//input[contains(@placeholder,"手机号或邮箱")]').send_keys(
Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//input[contains(@placeholder,"手机号或邮箱")]').send_keys(
self.account_item[1])
browser.find_element_by_xpath(
'//input[contains(@placeholder,"密码")]').send_keys(
Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//input[contains(@placeholder,"密码")]').send_keys(
self.account_item[2])
browser.find_element_by_xpath('//button[@type="submit"]').click()
time.sleep(3)
login_flag = False
while not login_flag:
try:
browser.find_element_by_xpath(
'//button[contains(text(),"提问")]')
cookies = browser.get_cookies()
cookies_file = {
item['name']: item['value']
for item in cookies
}
self.redis.update_cookies(self.account_item[0],
self.account_item[1],
self.account_item[2],
self.account_item[3],
cookies_file)
# with open(COOKIES_PATH+'/cookies.json', 'w') as f:
# f.write(json.dumps(cookies_file, ensure_ascii=False))
browser.close()
login_flag = True
return scrapy.Request(url=url,
callback=self.parse,
headers=self.headers,
cookies=cookies_file)
except:
pass
try:
eng_yzm = browser.find_element_by_xpath(
'//img[@class="Captcha-englishImg"]')
except:
eng_yzm = None
try:
chn_yzm = browser.find_element_by_xpath(
'//img[@class="Captcha-chineseImg"]')
except:
chn_yzm = None
if bool(eng_yzm):
# 使用菲菲打码平台解析英文数字的验证码
img_yzm_str = eng_yzm.get_attribute('src')
img_yzm = img_yzm_str.replace('data:image/jpg;base64,',
'').replace("%0A", "")
with open('eng_yzm.jpeg', 'wb') as img_file:
img_file.write(base64.b64decode(img_yzm))
feifei = FateadmApi(FEIFEI_USERNAME, FEIFEI_PASSWORD,
FEIFEI_PD_USER, FEIFEI_PD_PASSWORD)
code = feifei.PredictFromFileExtend(
"30400", "eng_yzm.jpeg")
browser.find_element_by_xpath(
'//input[@placeholder="验证码"]').send_keys(code)
time.sleep(1)
browser.find_element_by_xpath(
'//button[@type="submit"]').click()
time.sleep(3)
if bool(chn_yzm):
# 使用zheye模块识别倒立汉字验证码
img_yzm_str = chn_yzm.get_attribute('src')
img_yzm = img_yzm_str.replace('data:image/jpg;base64,',
'').replace("%0A", "")
with open('chn_yzm.jpeg', 'wb') as img_file:
img_file.write(base64.b64decode(img_yzm))
img_location = chn_yzm.location
# 获取浏览器工具栏的高度
browser_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
z = zheye()
# zheye模块识别后返回一个装有二元组的列表，形如：[(y,x)]
hz = z.Recognize('chn_yzm.jpeg')
if len(hz) == 2:
x1 = int(hz[0][1]) // 2 + img_location['x']
y1 = int(
hz[0][0]) // 2 + img_location['y'] + browser_height
move(x1, y1)
click()
time.sleep(3)
x2 = int(hz[1][1]) // 2 + img_location['x']
y2 = int(
hz[1][0]) // 2 + img_location['y'] + browser_height
move(x2, y2)
click()
else:
x1 = int(hz[0][1]) // 2 + img_location['x']
y1 = int(
hz[0][0]) // 2 + img_location['y'] + browser_height
move(x1, y1)
click()
time.sleep(1)
browser.find_element_by_xpath(
'//button[@type="submit"]').click()
time.sleep(3)
else:
return scrapy.Request(url=url,
callback=self.parse,
headers=self.headers,
cookies=self.account_item[4])
def login(self):
self.brower = webdriver.Chrome(
executable_path="webdriver/chromedriver.exe",
chrome_options=self.chrome_options)
self.brower.get("https://www.zhihu.com/signin")
while not self.check_login():
login_elemet = self.brower.find_element_by_css_selector(
"#root > div > main > div > div > div > div.SignContainer-content > div > form > div.SignFlow-tabs > div:nth-child(2)"
)
login_elemet.click()
self.brower.find_element_by_xpath(
"//input[@name='username']").send_keys(Keys.CONTROL + "a")
self.brower.find_element_by_xpath(
"//input[@name='username']").send_keys(self.uname)
self.brower.find_element_by_xpath(
"//input[@name='password']").send_keys(Keys.CONTROL + "a")
self.brower.find_element_by_xpath(
"//input[@name='password']").send_keys(self.passwd)
botton = self.brower.find_element_by_css_selector(
".Button.SignFlow-submitButton.Button--primary.Button--blue")
try:
chimgcode = "null"
chinese_captcha_element = self.brower.find_element_by_class_name(
"Captcha-chineseImg")
base64_text = chinese_captcha_element.get_attribute("src")
chimgcode = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
except:
pass
try:
enimgcode = "null"
englsh_captcha_element = self.brower.find_element_by_class_name(
"Captcha-englishImg")
base64_text = englsh_captcha_element.get_attribute("src")
enimgcode = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
except:
pass
if chimgcode != "null":
with open("captcha/chincaptcha.jpeg", "wb") as f:
f.write(base64.b64decode(chimgcode))
self.brower.maximize_window()
ele_postion = chinese_captcha_element.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
brower_navigation_panel_height = self.brower.execute_script(
"return window.outerHeight - window.innerHeight;")
z = zheye()
positions = z.Recognize('captcha/chincaptcha.jpeg')
last_position = []
if len(positions) == 2:
last_position.append([positions[0][1], positions[0][0]])
last_position.append([positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0] / 2),
int(last_position[1][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
brower_navigation_panel_height + first_position[1])
click()
move(
x_relative + second_position[0], y_relative +
brower_navigation_panel_height + second_position[1])
click()
elif len(positions) == 1:
last_position.append([positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
brower_navigation_panel_height + first_position[1])
click()
elif enimgcode != "null":
with open("captcha/engcaptcha.jpeg", "wb") as f:
f.write(base64.b64decode(enimgcode))
chaojiyingapi = Chaojiying_Client(self.settings.CJY_PASSWD,
self.settings.CJY_UNAME,
"906234")
im = open('captcha/engcaptcha.jpeg', 'rb').read()
json_result = chaojiyingapi.PostPic(im, 1902)
if json_result["err_no"] == 0:
pic_str = json_result["pic_str"]
print("识别码：{pic_str}".format(pic_str=pic_str))
else:
print("识别失败")
pic_str = "null"
self.brower.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(Keys.CONTROL + "a")
self.brower.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(pic_str)
botton.click()
cookie = self.brower.get_cookies()
cookie_dict = {}
for co in cookie:
cookie_dict[co['name']] = co['value']
self.brower.close()
return cookie_dict
def start_requests(self):
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
chrome_options.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(
'C:/Users/孙佩豪/AppData/Local/Google/Chrome/Application/chromedriver.exe',
chrome_options=chrome_options)
try:
browser.maximize_window() # 最大化窗口
except: # 已最大化的情况 代码会出错 捕获错误
pass
browser.get('https://www.zhihu.com/signin') # 打开知乎登录页面
time.sleep(2)
# browser.find_element_by_xpath('//div[@class="SignFlow-tabs"]/div[2]').click() # 点击帐号密码登录
login_success = False
try:
notify_ele = browser.find_element_by_xpath(
'//div[@class="Popover PushNotifications AppHeader-notifications"]'
) # 是否登录成功
login_success = True
except:
pass
if not login_success:
move(914, 329) # 点击 帐号密码登录
click()
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a") # 全选 然后输入账户密码
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"15292060685")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("qq1362441")
move(955, 566)
click()
click()
# browser.find_element_by_xpath('//button[@class="Button SignFlow-submitButton Button--primary Button--blue"]').click() # 点击登录按钮
login_success = False
while not login_success:
try:
time.sleep(1)
notify_ele = browser.find_element_by_xpath(
'//div[@class="Popover PushNotifications AppHeader-notifications"]'
) # 是否登录成功
login_success = True
except:
pass
try:
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg") # 是否出现英文验证码
except:
english_captcha_element = None
try:
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg") # 是否出现中文验证码
except:
chinese_captcha_element = None
if chinese_captcha_element: # 如果产生中文验证码
time.sleep(1)
ele_position = chinese_captcha_element.location # 获取节点坐标
x_relative = ele_position["x"] # x坐标
y_relative = ele_position["y"] # y坐标
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;'
) # 浏览器上栏高度
browser_navigation_panel_height = 70
time.sleep(3)
base64_text = chinese_captcha_element.get_attribute(
"src") # 提取中文验证码节点的arc属性
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A",
"") # 消除图片bs64编码中的无用符号
fh = open("yzm_cn.jpeg", "wb") # 保存文件
fh.write(base64.b64decode(code))
fh.close()
z = zheye()
positions = z.Recognize('yzm_cn.jpeg') # 使用者也 提取倒立文字坐标
last_position = []
if len(positions) == 2:
if positions[0][0] > positions[1][0]: # 按照顺序排列倒立文字坐标
last_position.append(
[positions[1][0], positions[1][1]])
last_position.append(
[positions[0][0], positions[0][1]])
else:
last_position.append(
[positions[0][0], positions[0][1]])
last_position.append(
[positions[1][0], positions[1][0]])
if len(positions) == 2:
first_position = [
int(last_position[0][1] / 2) + x_relative,
int(last_position[0][0] / 2) + y_relative +
browser_navigation_panel_height
] # 实际页面中 倒立文字图片为正常图片缩放的一倍 所有坐标需要除2取整 来获得可以在页面中使用的坐标
second_position = [
int(last_position[1][1] / 2) + x_relative,
int(last_position[1][0] / 2) + y_relative +
browser_navigation_panel_height
]
move(first_position[0], first_position[1]
) # 坐标 起始点x坐标+倒立文字x坐标 起始点y坐标+浏览器地址栏高度+倒立文字y坐标
click()
move(second_position[0], second_position[1])
click()
else: # 如果只有一个倒立文字
last_position.append([positions[0][1], positions[0][1]])
first_position = [
int(last_position[0][1] / 2) + x_relative,
int(last_position[0][0] / 2) +
browser_navigation_panel_height + y_relative
]
time.sleep(5)
move(first_position[0], first_position[1])
time.sleep(5)
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a") # 全选 然后输入账户密码
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"15292060685")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("qq1362441")
move(954, 619)
click()
if english_captcha_element: # 如果产生英文验证码
time.sleep(1)
base64_text = english_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A",
"") # 消除图片bs64编码中的无用符号
fh = open("yzm_en.jpeg", "wb") # 保存文件
fh.write(base64.b64decode(code))
fh.close()
Yundama = YDMHttp("sph116", "qq1362441", 8730,
"9f94b142759f9fd86bd0e7a912bbc889") # 实例化云打码
code = Yundama.decode("yzm_en.jpeg", 5000, 60) # 识别
while True: # 若识别失败 不停识别 直至成功
if code == "": #
code = Yundama.decode("yzm_en.jpeg", 5000, 60)
time.sleep(0.5)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(Keys.CONTROL + "a") # 找到英文验证码位置
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(code)
move(956, 600)
click()
time.sleep(5)
try:
# notify_element = browser.find_element_by_class_name("Popover PushNotifications AppHeader-notifications") # 查看是否出现 登录成功的节点
# login_success = True
Cookies = browser.get_cookies() # 获取登录成功的cookie
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
f = open('./ArticleSpider/cookies/zhihu/' +
cookie['name'] + '.zhihu', 'wb') # 存储cookie进入本地
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
# browser.close() # 暂时不关闭
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
] # 回调 进入解析
except:
pass
print("======知乎登录成功=========")
def get_captcha_position():
from zheye import zheye
z = zheye()
positions = z.Recognize('captcha.gif')
print(positions)
return positions
def start_requests(self):
chrome_option = Options()
chrome_option.add_argument('--disable-extensions')
chrome_option.add_experimental_option('debuggerAddress', '127.0.0.1:9222/json')
browser = webdriver.Chrome(executable_path='E:/git/chromedriver_win32 (3)/chromedriver.exe', chrome_options=chrome_option)
# browser = webdriver.Chrome(executable_path='E:/git/chromedriver_win32 (3)/chromedriver.exe')
try:
browser.maximize_window() # 将窗口最大化防止定位错误
except:
pass
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_css_selector('.SignFlow-account input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-account input').send_keys('18930059946')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys('Admin@2019')
# browser.find_element_by_css_selector('.Button.SignFlow-submitButton').click()
time.sleep(3)
move(902, 599)
click()
time.sleep(10)
login_success = False
while not login_success:
try:
notify_ele = browser.find_element_by_class_name('Popover PushNotifications AppHeader-notifications')
login_success = True
except:
pass
try:
# 查询是否有英文验证码
english_captcha_element = browser.find_element_by_class_name('Captcha-englishImg')
except:
english_captcha_element = None
try:
# 查询是否有中文验证码
chinese_captcha_element = browser.find_element_by_class_name('Captcha-chineseImg')
except:
chinese_captcha_element = None
if chinese_captcha_element:
ele_postion = chinese_captcha_element.location
x_relative = ele_postion['x']
y_relative = ele_postion['y']
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
"""
保存图片 1. 通过保存base64编码 2. 通过crop方法
"""
# 1. 通过保存base64编码
base64_text = chinese_captcha_element.get_attribute("src")
code = base64_text.replace('data:image/jpg;base64,', '').replace('%0A', '')
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
z = zheye()
positions = z.Recognize("yzm_cn.jpeg")
last_postion = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_postion.append([positions[1][1], positions[1][0]])
last_postion.append([positions[0][1], positions[0][0]])
else:
last_postion.append([positions[0][1], positions[0][0]])
last_postion.append([positions[1][1], positions[1][0]])
first_point = [int(last_postion[0][0] / 2), int(last_postion[0][1] / 2)]
second_point = [int(last_postion[1][0] / 2), int(last_postion[1][1] / 2)]
move((x_relative + first_point[0]), y_relative + browser_navigation_panel_height + first_point[1])
click()
move((x_relative + second_point[0]), y_relative + browser_navigation_panel_height + second_point[1])
click()
else:
last_postion.append([positions[0][1], positions[0][0]])
first_point = [int(last_postion[0][0] / 2), int(last_postion[0][1] / 2)]
move((x_relative + first_point[0]), y_relative + browser_navigation_panel_height + first_point[1])
click()
browser.find_element_by_css_selector('.SignFlow-account input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-account input').send_keys('18930059946')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys('Admin@2009')
# browser.find_element_by_css_selector('.Button.SignFlow-submitButton').click()
time.sleep(3)
move(836, 634)
click()
if english_captcha_element:
base64_text = english_captcha_element.get_attribute("src")
code = base64_text.replace('data:image/jpg;base64,', '').replace('%0A', '')
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
yundama = YDMHttp("Patrick_Wang", "A100s200", 7107, "f8ecb1a73b115d41f7a2526fd1d8d5a8")
code = yundama.decode("yzm_en.jpeg", 5000, 60)
while True:
if code == "":
code = yundama.decode("yzm_en.jpeg", 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input').send_keys(code)
browser.find_element_by_css_selector('.SignFlow-account input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-account input').send_keys('18930059946')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys(Keys.CONTROL + 'a')
browser.find_element_by_css_selector('.SignFlow-password input').send_keys('Admin@2009')
# submit_ele = browser.find_element_by_css_selector(".Button.SignFlow-submitButton")
# browser.find_element_by_css_selector(".Button.SignFlow-submitButton").click()
move(836, 634)
click()
time.sleep(10)
try:
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己文件的所在路径
f = open('./ArticleSpider/cookies/' + cookie['name'] + '.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [scrapy.Request(url=self.start_urls[0], dont_filter=True, cookies=cookie_dict)]
except:
pass
def login(self, response):
from selenium import webdriver
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.chrome.options import Options
from mouse import move, click
"""
1.启动chrome（启动之前确保所有的chrome实例己经关闭）
"""
chrome_options = Options()
chrome_options.add_argument("--disable-extensions")
chrome_options.add_experimental_option("excludeSwitches",
['enable-automation'])
chrome_options.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
#
#
browser = webdriver.Chrome(
executable_path="D:/解压文件/chromedriver_win32/chromedriver.exe")
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin?next=%2F")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[1]/div[2]"
).click()
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input'
).send_keys("13232732408")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("abc713912")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/button"
).click()
time.sleep(3)
while True:
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/button'
).click()
has_en = False
has_cn = False
try:
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div[2]/img'
)
has_cn = True # 中文
except:
pass
try:
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/span/div/img'
)
has_en = True # 英文
except:
pass
if has_cn or has_en:
break
time.sleep(3)
# 未登录前为false
login_succes = False
while not login_succes:
try:
notify_element = browser.find_element_by_class_name(
"Popover AppHeader-menu")
login_succes = True
except:
pass
try:
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg") # 如果是英文验证码
except:
english_captcha_element = None
try:
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg") # 如果是中文验证码
except:
chinese_captcha_element = None
# 这个是中文验证码的登录
if chinese_captcha_element:
# 精准获取xy坐标
ele_postion = chinese_captcha_element.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
# 但是他的计算方法并不是从窗口那里开始获取的，所以下面是固定的代码，可以去掉窗口位置,
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;'
) # browser_navigation_panel_height就是地址栏的高度
# 做一个图片保存转换
base64_text = chinese_captcha_element.get_attribute("src")
import base64
code = base64_text.replace(
"data:image/jpg;base64,", ''
).replace(
"%0A", ""
) # 获取文本后的前边的文本替换成空的，但是这个base64和一般的base64不太一样，还多了一段%0A，所以也要替换掉
fh = open("yzm_cn.jpeg", "wb") # 保存这个图片叫yzm_cn.jpg
fh.write(base64.b64decode(code))
fh.close()
from zheye import zheye
z = zheye()
positions = z.Recognize('yzm_cn.jpeg')
pos_arr = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
pos_arr.append([positions[1][1], positions[1][0]])
pos_arr.append([positions[0][1], positions[0][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
pos_arr.append([positions[1][1], positions[1][0]])
else:
pos_arr.append([positions[0][1], positions[0][0]])
if len(positions) == 2:
# 有两个倒立文字
first_position = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
] # 原始图片在zheye项目的相比小一半，所以除以2，这是第一个元素
second_position = [
int(pos_arr[1][0] / 2),
int(pos_arr[1][1] / 2)
] # 原始图片在zheye项目的相比小一半，所以除以2，这是第二个元素
move((x_relative + first_position[0]),
y_relative + browser_navigation_panel_height +
second_position[1]) # 这是点击第一个元素
click()
move((x_relative + second_position[0]),
y_relative + browser_navigation_panel_height +
second_position[1]) # 这是点击第二个元素
click()
else:
# 这是有一个倒立文字
first_position = [
int(pos_arr[0][0] / 2),
int(pos_arr[0][1] / 2)
] # 原始图片在zheye项目的相比小一半，所以除以2，这是第一个元素
move((x_relative + first_position[0]),
y_relative + browser_navigation_panel_height +
second_position[1]) # 这是点击第一个元素
click()
# 再做一次登录
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[1]/div[2]"
).click()
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input"
).send_keys("13232732408")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("abc713912")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/button'
).click()
# 英文验证码
if english_captcha_element:
# 做一个图片保存转换
base64_text = english_captcha_element.get_attribute("src")
import base64
code = base64_text.replace(
"data:image/jpg;base64,", ''
).replace(
"%0A", ""
) # 获取文本后的前边的文本替换成空的，但是这个base64和一般的base64不太一样，还多了一段%0A，所以也要替换掉
fh = open("yzm_en.jpeg", "wb") # 保存这个图片叫yzm_cn.jpg
fh.write(base64.b64decode(code))
print(code)
fh.close()
from ArticleSpider.tools.chaojiying import Chaojiying_Client
Chaojiying = Chaojiying_Client("1171242903", "130796abc",
"905526")
im = open('D:\BaiduNetdiskDownload\ArticleSpider\yzm_en.jpeg',
'rb').read()
code = Chaojiying.PostPic(im, 1902)
print("英文验证码:")
print(code)
# 做一个while循环，怕一次不成功识别，循环到成功，再做一个break
while True:
if code == "":
chaojiying = Chaojiying_Client('1171242903',
'130796abc', '905526')
im = open(
'D:\BaiduNetdiskDownload\ArticleSpider\yzm_en.jpeg',
'rb').read()
code = chaojiying.PostPic(im, 1902)
# print("chaojiyingshibie结果:")
# print(code)
else:
break
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[1]/div[2]"
).click()
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label'
).send_keys(code["pic_str"])
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[2]/div/label/input"
).send_keys("13232732408")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("abc713912")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/button'
).click()
time.sleep(10)
try:
notify_element = browser.find_element_by_xpath(
'//*[@id="Popover17-toggle"]/img')
login_succes = True
Cookies = browser.get_cookies()
print(Cookies)
cookie_dict = {}
import pickle
for cookie in Cookies:
# 写入文件
# 此处大家修改一下自己的文件所在路径
f = open(
'./ArticleSpider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
yield [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict,
callback=check_login)
]
except:
pass
def start_requests(self):
# 本地启动chrome
chrome_option = Options()
chrome_option.add_argument("--disable-extensions")
chrome_option.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(
executable_path=
"C:/Users/Administrator/PycharmProjects/Envs/Scripts/chromedriver.exe",
chrome_options=chrome_option)
try:
browser.maximize_window()
except:
pass
try:
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[1]/div[2]"
).click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"16601052213")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("ZHUHAIOO")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(1)
login_success = False
while not login_success:
try:
notify_ele = browser.find_element_by_xpath(
"//*[@id='root']/div/div[2]/header/div[1]/a/svg")
login_success = True
except:
pass
try:
english_captcha_element = browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/span/div/img'
)
except:
english_captcha_element = None
try:
chinese_captcha_element = browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[4]/div/div[2]/img"
)
except:
chinese_captcha_element = None
if chinese_captcha_element:
# 图片位置坐标x和y
ele_position = chinese_captcha_element.location
x_relative = ele_position["x"]
y_relative = ele_position["y"]
# 浏览器上部的y的坐标
browser_navigation_panel_height = 70
# 图片保存
base64_text = chinese_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
fh = open("yzm_cn.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
z = zheye()
positions = z.Recognize('yzm_cn.jpeg')
last_position = []
if len(positions) == 2:
# 如果第一个元素的x坐标大于第二个元素的x坐标，则第二个元素是第一个倒立文字
if positions[0][1] > positions[1][1]:
# 所以列表里放文字的时候返过来放，先放第二个元素的xy，再放第一个元素的xy,后面的是x，前面的是y
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
# 浏览器中的图片实际要比
first_position = [
int(last_position[0][0]) / 2,
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0]) / 2,
int(last_position[1][1] / 2)
]
mouse.move(
x_relative + first_position[0],
y_relative + browser_navigation_panel_height +
first_position[1])
mouse.click()
time.sleep(3)
mouse.move(
x_relative + second_position[0],
y_relative + browser_navigation_panel_height +
second_position[1])
mouse.click()
else:
last_position.append(
[positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0]) / 2,
int(last_position[0][1] / 2)
]
mouse.move(
x_relative + first_position[0],
y_relative + browser_navigation_panel_height +
first_position[1])
mouse.click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input"
).send_keys("16601052213")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("1qaz@4321")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
if english_captcha_element:
# 图片保存
base64_text = english_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
chaojiying = Chaojiying_Client('16601052213', 'ZHUHAIOO00',
'905609')
im = open('yzm_en.jpeg', 'rb').read()
json_data = chaojiying.PostPic(im, 1902)
code = json_data["pic_str"]
while json_data["err_no"] != 0:
if code == "":
json_data = chaojiying.PostPic(im, 1902)
code = json_data["pic_str"]
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[1]/div/form/div[4]/div/div/label/input'
).send_keys(code)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input"
).send_keys("16601052213")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
"//*[@id='root']/div/main/div/div/div/div[1]/div/form/div[3]/div/label/input"
).send_keys("1qaz@4321")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
except:
cookies = browser.get_cookies()
pickle.dump(
cookies,
open(
"C:/Users/Administrator/PycharmProjects/Envs/Scripts/ArticleSpider/cookies/zhihu.cookie",
"wb"))
cookie_dict = {}
for cookie in cookies:
cookie_dict[cookie["name"]] = cookie["value"]
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
# -*- coding: utf-8 -*-
__author__ = 'dongnanzhy'
# 参考：https://github.com/muchrooms/zheye
import sys
sys.path.insert(0, "../")
from zheye import zheye
z = zheye()
positions = z.Recognize('captcha_cn.gif')
print(positions)
def start_requests(self):
chrome_option = Options()
"""
手动启动Chrome,避免被反爬识别
Chrome安装目录下，命令行执行：chrome.exe --remote-debugging-port=9222
验证启动是否成功：http://127.0.0.1:9222/json
"""
chrome_option.add_argument("--disable-extensions")
chrome_option.add_experimental_option("debuggerAddress",
"127.0.0.1:9222")
browser = webdriver.Chrome(chrome_options=chrome_option)
try:
browser.maximize_window()
except:
pass
browser.get("https://www.zhihu.com/signin")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys("xxx")
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("xxx")
browser.find_element_by_css_selector(
".Button.SignFlow-submitButton").click()
time.sleep(3)
login_success = False
if login_success:
cookies_list = browser.get_cookies()
print(cookies_list)
cookie_dict = {}
import pickle
for cookie in cookies_list:
# 写入文件
f = open(
'./Article_Spider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
while not login_success:
try:
browser.find_element_by_class_name(
"Popover PushNotifications AppHeader-notifications")
login_success = True
cookies_list = browser.get_cookies()
print(cookies_list)
cookie_dict = {}
for cookie in cookies_list:
# 写入文件
f = open(
'./Article_Spider/cookies/zhihu/' + cookie['name'] +
'.zhihu', 'wb')
import pickle
pickle.dump(cookie, f)
f.close()
cookie_dict[cookie['name']] = cookie['value']
browser.close()
return [
scrapy.Request(url=self.start_urls[0],
dont_filter=True,
cookies=cookie_dict)
]
except:
pass
try:
english_captcha_element = browser.find_element_by_class_name(
"Captcha-englishImg")
except:
english_captcha_element = None
try:
chinese_captcha_element = browser.find_element_by_class_name(
"Captcha-chineseImg")
except:
chinese_captcha_element = None
# 识别中文倒立汉字
if chinese_captcha_element:
ele_postion = chinese_captcha_element.location
x_relative = ele_postion["x"]
y_relative = ele_postion["y"]
browser_navigation_panel_height = browser.execute_script(
'return window.outerHeight - window.innerHeight;')
base64_text = chinese_captcha_element.get_attribute("src")
code = base64_text.replace("data:image/jpg;base64,",
"").replace("%0A", "")
fh = open("verify_code.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
z = zheye()
positions = z.Recognize('verify_code.jpeg')
last_position = []
if len(positions) == 2:
if positions[0][1] > positions[1][1]:
last_position.append(
[positions[1][1], positions[1][0]])
last_position.append(
[positions[0][1], positions[0][0]])
else:
last_position.append(
[positions[0][1], positions[0][0]])
last_position.append(
[positions[1][1], positions[1][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
second_position = [
int(last_position[1][0] / 2),
int(last_position[1][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
browser_navigation_panel_height + first_position[1])
click()
time.sleep(3)
move(
x_relative + second_position[0], y_relative +
browser_navigation_panel_height + second_position[1])
click()
else:
last_position.append([positions[0][1], positions[0][0]])
first_position = [
int(last_position[0][0] / 2),
int(last_position[0][1] / 2)
]
move(
x_relative + first_position[0], y_relative +
browser_navigation_panel_height + first_position[1])
click()
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"13662241324")
time.sleep(2)
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("root0503")
move(674, 527)
click()
# 识别英语字母验证码
if english_captcha_element:
base64_text = english_captcha_element.get_attribute("src")
code = base64_text.replace('data:image/jpg;base64,',
'').replace("%0A", "")
# print code
fh = open("yzm_en.jpeg", "wb")
fh.write(base64.b64decode(code))
fh.close()
from tools.yundama_requests import YDMHttp
yundama = YDMHttp("xxx", "xxx", 3129, "xxx")
code = yundama.decode("yzm_en.jpeg", 5000, 60)
while True:
if code == "":
code = yundama.decode("yzm_en.jpeg", 5000, 60)
else:
break
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(Keys.CONTROL + "a")
browser.find_element_by_xpath(
'//*[@id="root"]/div/main/div/div/div/div[2]/div[1]/form/div[3]/div/div/div[1]/input'
).send_keys(code)
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-accountInput.Input-wrapper input").send_keys(
"13662241324")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys(Keys.CONTROL + "a")
browser.find_element_by_css_selector(
".SignFlow-password input").send_keys("root0503")
move(674, 527)
click()
