can resolve captcha with 2captcha

workers/SolveCaptch.py

@@ -1,82 +1,53 @@
-import random
 # for recaptcha
-import urllib
-from speech_recognition import Recognizer, AudioFile
+import logging
 import random
-import os
+import re
+import time
+
+import requests
+
+CAPCHA_NOT_READY = "CAPCHA_NOT_READY"
+REGEX_DATA_SITE_KEY = "data-sitekey=[\"a-z0-9A-Z]+"
 
 
 class SolveCaptcha:
     def __init__(self, page):
         self.page = page
+        self.logger = logging.getLogger("SolveCaptcha")
         self.main_frame = None
         self.recaptcha = None
 
     def delay(self):
         self.page.wait_for_timeout(random.randint(1, 3) * 1000)
 
-    def presetup(self):
-        name = self.page.locator(
-            "//iframe[@title='reCAPTCHA']").get_attribute("name")
-        self.recaptcha = self.page.frame(name=name)
+    def start(self, handle_solution_received):
+        self.logger.info("start to resolve captcha")
+        content = self.page.content()
+        data_sitekey = re.findall(REGEX_DATA_SITE_KEY, content)
+        self.logger.info(data_sitekey)
+        if len(data_sitekey) == 1:
+            key_with_comma = data_sitekey[0].split("=")[-1]
+            key = key_with_comma.replace("\"", '')
+            print("key is : " + key)
+            self.solve_captcha(key, handle_solution_received)
 
-        self.recaptcha.click("//div[@class='recaptcha-checkbox-border']")
-        self.delay()
-        s = self.recaptcha.locator("//span[@id='recaptcha-anchor']")
-        if s.get_attribute("aria-checked") != "false":  # solved already
-            return
-
-        # self.main_frame = self.page.frame(name=self.page.locator(
-        #     "//iframe[contains(@src,'https://www.google.com/recaptcha/api2/bframe?')]").get_attribute("name"))
-        # self.main_frame.click("id=recaptcha-audio-button")
-
-    def start(self):
-        print("start to resolve captcha")
-        self.presetup()
-        tries = 0
-        while (tries <= 5):
-            self.delay()
-            try:
-                self.solve_captcha()
-            except Exception as e:
-                print("exception:")
-                print(e)
-                self.main_frame.click("id=recaptcha-reload-button")
-            else:
-                s = self.recaptcha.locator("//span[@id='recaptcha-anchor']")
-                if s.get_attribute("aria-checked") != "false":
-                    self.page.click("id=recaptcha-demo-submit")
-                    self.delay()
-                    break
-            tries += 1
-
-    def solve_captcha(self):
-        print("solve_captcha()")
-        # self.main_frame.click(
-        #     "//button[@aria-labelledby='audio-instructions rc-response-label']")
-        # href = self.main_frame.locator(
-        #     "//a[@class='rc-audiochallenge-tdownload-link']").get_attribute("href")
-        #
-        # print("retrieve mp3 file")
-        # urllib.request.urlretrieve(href, "audio.mp3")
-        #
-        # sound = pydub.AudioSegment.from_mp3(
-        #     "audio.mp3").export("audio.wav", format="wav")
-        #
-        # recognizer = Recognizer()
-        #
-        # recaptcha_audio = AudioFile("audio.wav")
-        # with recaptcha_audio as source:
-        #     audio = recognizer.record(source)
-        # print("recognizer mp3 file")
-        #
-        # text = recognizer.recognize_google(audio)
-        # print("recognized text: " + text)
-        # self.main_frame.fill("id=audio-response", text)
-        # self.main_frame.click("id=recaptcha-verify-button")
-        # self.delay()
-
-    def __del__(self):
-        pass
-        # os.remove("audio.mp3")
-        # os.remove("audio.wav")
+    def solve_captcha(self, google_key: str, handle_solution_received):
+        self.logger.info("solve_captcha()")
+        url_get = "http://2captcha.com/in.php?key=e7e3cd0977aba2dab49e0ea052ca58b1&method=userrecaptcha&googlekey={}&pageurl={}".format(
+            google_key, self.page.url)
+        res = requests.get(url_get)
+        print(res.text)
+        results = res.text.split("|")
+        self.captcha_id = results[-1]
+        #       wait for 15 seconds
+        time.sleep(15)
+        #         get result of the captcha
+        url_response = "http://2captcha.com/res.php?key=e7e3cd0977aba2dab49e0ea052ca58b1&action=get&id={}".format(
+            self.captcha_id)
+        solution = CAPCHA_NOT_READY
+        while solution == CAPCHA_NOT_READY:
+            solution_res = requests.get(url_response)
+            time.sleep(5)
+            solution = solution_res.text
+        print(solution)
+        handle_solution_received(solution.split("|")[-1])