python编辑器中文字体倒立的_scrapy爬虫识别知乎倒立文字

参考github-zheye编写好的程序

程序环境：python3.6+tensorflow1.5

python3.7+tensorflow2.2如果用于老式机器，会一直失败，请注意，降级到3.6+1.5

知乎会识别chromedriver里面的关键字，从而禁止我们访问，我们可以加入以下代码规避检测

chrome_option = Options()

chrome_option.add_argument("--disable-extensions")

chrome_option.add_experimental_option("excludeSwitches",['enable-automation'])

chrome_option.add_experimental_option("debuggerAddress", "127.0.0.1:9222")

知乎识别汉字倒立文字

def deal_with_chinese_captcha(self, Captcha_chineseImg):

"""

deal with chinese captcha

:param chinese_captcha_element:

:return:

"""

ele_position = Captcha_chineseImg.location

x_relative = ele_position["x"]

y_relative = ele_position["y"]

browser_navigation_panel_height = self.browser.execute_script(

'return window.outerHeight - window.innerHeight;')

base64_text = Captcha_chineseImg.get_attribute("src")

# 这里需要注意这个OA

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()

# 识别汉字坐标，关键代码

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]])

# 获取坐标，这里因为实际显示的文字比生成的文字小，是1:2的关系，所以还原

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)]

# 点击第一个汉字，这里x_relative是图片位置，后面是汉字坐标

# y_relative是图片但是要加上浏览器地址栏坐标+汉字纵坐标

move(x_relative + first_position[0],

y_relative + browser_navigation_panel_height + first_position[1])

click()

# 点击第二个汉字

time.sleep(1)

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()

# 输入验证码后继续登录

self.browser.find_element_by_name("username").send_keys(Keys.CONTROL + "a")

self.browser.find_element_by_name("username").send_keys(ZHIHU_USER)

time.sleep(2)

self.browser.find_element_by_name("password").send_keys(Keys.CONTROL + "a")

self.browser.find_element_by_name("password").send_keys(ZHIHU_PWD)

time.sleep(2)

self.browser.find_element_by_xpath('//*[@id="root"]/div/main/div/div/div/div[1]/div/form/button').click()

if self.check_login():

login_success = True