Fix looping challenges and invalid cookies. resolves #1006 (#1010)

Co-authored-by: ilike2burnthing <59480337+ilike2burnthing@users.noreply.github.com>

CHANGELOG.md

@@ -1,5 +1,51 @@
 # Changelog
 
+## v3.3.12 (2023/12/15)
+
+* Fix looping challenges and invalid cookies
+
+## v3.3.11 (2023/12/11)
+
+* Update UC 3.5.4 & Selenium 4.15.2. Thanks @txtsd
+
+## v3.3.10 (2023/11/14)
+
+* Add LANG ENV - resolves issues with YGGtorrent
+
+## v3.3.9 (2023/11/13)
+
+* Fix for Docker build, capture TypeError
+
+## v3.3.8 (2023/11/13)
+
+* Fix headless=true for Chrome 117+. Thanks @NabiKAZ
+* Support running Chrome 119 from source. Thanks @koleg and @Chris7X
+* Fix "OSError: [WinError 6] The handle is invalid" on exit. Thanks @enesgorkemgenc
+
+## v3.3.7 (2023/11/05)
+
+* Bump to rebuild. Thanks @JoachimDorchies
+
+## v3.3.6 (2023/09/15)
+
+* Update checkbox selector, again
+
+## v3.3.5 (2023/09/13)
+
+* Change checkbox selector, support languages other than English
+
+## v3.3.4 (2023/09/02)
+
+* Update checkbox selector
+
+## v3.3.3 (2023/08/31)
+
+* Update undetected_chromedriver to v3.5.3
+
+## v3.3.2 (2023/08/03)
+
+* Fix URL domain in Prometheus exporter
+
 ## v3.3.1 (2023/08/03)
 
 * Fix for Cloudflare verify checkbox

Dockerfile

@@ -62,17 +62,17 @@ ENTRYPOINT ["/usr/bin/dumb-init", "--"]
 CMD ["/usr/local/bin/python", "-u", "/app/flaresolverr.py"]
 
 # Local build
-# docker build -t ngosang/flaresolverr:3.3.1 .
-# docker run -p 8191:8191 ngosang/flaresolverr:3.3.1
+# docker build -t ngosang/flaresolverr:3.3.12 .
+# docker run -p 8191:8191 ngosang/flaresolverr:3.3.12
 
 # Multi-arch build
 # docker run --rm --privileged multiarch/qemu-user-static --reset -p yes
 # docker buildx create --use
-# docker buildx build -t ngosang/flaresolverr:3.3.1 --platform linux/386,linux/amd64,linux/arm/v7,linux/arm64/v8 .
+# docker buildx build -t ngosang/flaresolverr:3.3.12 --platform linux/386,linux/amd64,linux/arm/v7,linux/arm64/v8 .
 #   add --push to publish in DockerHub
 
 # Test multi-arch build
 # docker run --rm --privileged multiarch/qemu-user-static --reset -p yes
 # docker buildx create --use
-# docker buildx build -t ngosang/flaresolverr:3.3.1 --platform linux/arm/v7 --load .
-# docker run -p 8191:8191 --platform linux/arm/v7 ngosang/flaresolverr:3.3.1
+# docker buildx build -t ngosang/flaresolverr:3.3.12 --platform linux/arm/v7 --load .
+# docker run -p 8191:8191 --platform linux/arm/v7 ngosang/flaresolverr:3.3.12

README.md

@@ -113,7 +113,7 @@ This also speeds up the requests since it won't have to launch a new browser ins
 | Parameter | Notes                                                                                                                                                                                                                                                                                                            |
 |-----------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
 | session   | Optional. The session ID that you want to be assigned to the instance. If isn't set a random UUID will be assigned.                                                                                                                                                                                              |
-| proxy     | Optional, default disabled. Eg: `"proxy": {"url": "http://127.0.0.1:8888"}`. You must include the proxy schema in the URL: `http://`, `socks4://` or `socks5://`. Authorization (username/password) is supported. Eg: `"proxy": {"url": "http://127.0.0.1:8888", username": "testuser", "password": "testpass"}` |
+| proxy     | Optional, default disabled. Eg: `"proxy": {"url": "http://127.0.0.1:8888"}`. You must include the proxy schema in the URL: `http://`, `socks4://` or `socks5://`. Authorization (username/password) is supported. Eg: `"proxy": {"url": "http://127.0.0.1:8888", "username": "testuser", "password": "testpass"}` |
 
 #### + `sessions.list`
 
@@ -232,6 +232,7 @@ This is the same as `request.get` but it takes one more param:
 | LOG_HTML           | false                  | Only for debugging. If `true` all HTML that passes through the proxy will be logged to the console in `debug` level.                                          |
 | CAPTCHA_SOLVER     | none                   | Captcha solving method. It is used when a captcha is encountered. See the Captcha Solvers section.                                                            |
 | TZ                 | UTC                    | Timezone used in the logs and the web browser. Example: `TZ=Europe/London`.                                                                                   |
+| LANG               | none                   | Language used in the web browser. Example: `LANG=en_GB`.                                                                                   |
 | HEADLESS           | true                   | Only for debugging. To run the web browser in headless mode or visible.                                                                                       |
 | BROWSER_TIMEOUT    | 40000                  | If you are experiencing errors/timeouts because your system is slow, you can try to increase this value. Remember to increase the `maxTimeout` parameter too. |
 | TEST_URL           | https://www.google.com | FlareSolverr makes a request on start to make sure the web browser is working. You can change that URL if it is blocked in your country.                      |

package.json

@@ -1,6 +1,6 @@
 {
   "name": "flaresolverr",
-  "version": "3.3.1",
+  "version": "3.3.12",
   "description": "Proxy server to bypass Cloudflare protection",
   "author": "Diego Heras (ngosang / ngosang@hotmail.es)",
   "license": "MIT"

requirements.txt

@@ -1,6 +1,6 @@
 bottle==0.12.25
 waitress==2.1.2
-selenium==4.11.2
+selenium==4.15.2
 func-timeout==4.3.5
 prometheus-client==0.17.1
 # required by undetected_chromedriver

src/bottle_plugins/prometheus_plugin.py

@@ -2,7 +2,8 @@ import logging
 import os
 import urllib.parse
 
-from dtos import V1ResponseBase
+from bottle import request
+from dtos import V1RequestBase, V1ResponseBase
 from metrics import start_metrics_http_server, REQUEST_COUNTER, REQUEST_DURATION
 
 PROMETHEUS_ENABLED = os.environ.get('PROMETHEUS_ENABLED', 'false').lower() == 'true'
@@ -39,8 +40,12 @@ def prometheus_plugin(callback):
 
         domain = "unknown"
         if res.solution and res.solution.url:
-            parsed_url = urllib.parse.urlparse(res.solution.url)
-            domain = parsed_url.hostname
+            domain = parse_domain_url(res.solution.url)
+        else:
+            # timeout error
+            req = V1RequestBase(request.json)
+            if req.url:
+                domain = parse_domain_url(req.url)
 
         run_time = (res.endTimestamp - res.startTimestamp) / 1000
         REQUEST_DURATION.labels(domain=domain).observe(run_time)
@@ -54,4 +59,8 @@ def prometheus_plugin(callback):
             result = "error"
         REQUEST_COUNTER.labels(domain=domain, result=result).inc()
 
+    def parse_domain_url(url):
+        parsed_url = urllib.parse.urlparse(url)
+        return parsed_url.hostname
+
     return wrapper

src/flaresolverr_service.py

@@ -252,11 +252,11 @@ def _resolve_challenge(req: V1RequestBase, method: str) -> ChallengeResolutionT:
 def click_verify(driver: WebDriver):
     try:
         logging.debug("Try to find the Cloudflare verify checkbox...")
-        iframe = driver.find_element(By.XPATH, "//iframe[@title='Widget containing a Cloudflare security challenge']")
+        iframe = driver.find_element(By.XPATH, "//iframe[starts-with(@id, 'cf-chl-widget-')]")
         driver.switch_to.frame(iframe)
         checkbox = driver.find_element(
             by=By.XPATH,
-            value='//*[@id="cf-stage"]//label[@class="ctp-checkbox-label"]/input',
+            value='//*[@id="challenge-stage"]/div/label/input',
         )
         if checkbox:
             actions = ActionChains(driver)
@@ -294,6 +294,9 @@ def _evil_logic(req: V1RequestBase, driver: WebDriver, method: str) -> Challenge
 
     # navigate to the page
     logging.debug(f'Navigating to... {req.url}')
+    # Workaround for "challenge not detected" caused by the devtools window
+    driver.get(req.url)
+    driver.start_session() # required to bypass Cloudflare
     if method == 'POST':
         _post_request(req, driver)
     else:

src/undetected_chromedriver/__init__.py

@@ -17,11 +17,12 @@ by UltrafunkAmsterdam (https://github.com/ultrafunkamsterdam)
 from __future__ import annotations
 
 
-__version__ = "3.5.0"
+__version__ = "3.5.4"
 
 import json
 import logging
 import os
+import pathlib
 import re
 import shutil
 import subprocess
@@ -373,6 +374,18 @@ class Chrome(selenium.webdriver.chrome.webdriver.WebDriver):
                 browser_executable_path or find_chrome_executable()
             )
 
+        if not options.binary_location or not \
+                pathlib.Path(options.binary_location).exists():
+                raise FileNotFoundError(
+                    "\n---------------------\n"
+                    "Could not determine browser executable."
+                    "\n---------------------\n"
+                    "Make sure your browser is installed in the default location (path).\n"
+                    "If you are sure about the browser executable, you can specify it using\n"
+                    "the `browser_executable_path='{}` parameter.\n\n"
+                    .format("/path/to/browser/executable" if IS_POSIX else "c:/path/to/your/browser.exe")
+                )
+
         self._delay = 3
 
         self.user_data_dir = user_data_dir
@@ -383,7 +396,7 @@ class Chrome(selenium.webdriver.chrome.webdriver.WebDriver):
         if no_sandbox:
             options.arguments.extend(["--no-sandbox", "--test-type"])
 
-        if headless or options.headless:
+        if headless or getattr(options, 'headless', None):
             #workaround until a better checking is found
             try:
                 v_main = int(self.patcher.version_main) if self.patcher.version_main else 108
@@ -451,11 +464,9 @@ class Chrome(selenium.webdriver.chrome.webdriver.WebDriver):
             )
             self.browser_pid = browser.pid
 
-        # Fix for Chrome 115
-        # https://github.com/seleniumbase/SeleniumBase/pull/1967
+
         service = selenium.webdriver.chromium.service.ChromiumService(
-            executable_path=self.patcher.executable_path,
-            service_args=["--disable-build-check"]
+            self.patcher.executable_path
         )
 
         super(Chrome, self).__init__(
@@ -480,7 +491,7 @@ class Chrome(selenium.webdriver.chrome.webdriver.WebDriver):
         else:
             self._web_element_cls = WebElement
 
-        if options.headless:
+        if headless or getattr(options, 'headless', None):
             self._configure_headless()
 
     def _configure_headless(self):
@@ -800,7 +811,11 @@ class Chrome(selenium.webdriver.chrome.webdriver.WebDriver):
                 else:
                     logger.debug("successfully removed %s" % self.user_data_dir)
                     break
+
+                try:
                     time.sleep(0.1)
+                except OSError:
+                    pass
 
         # dereference patcher, so patcher can start cleaning up as well.
         # this must come last, otherwise it will throw 'in use' errors
@@ -895,8 +910,6 @@ def find_chrome_executable():
             if item is not None:
                 for subitem in (
                     "Google/Chrome/Application",
-                    "Google/Chrome Beta/Application",
-                    "Google/Chrome Canary/Application",
                 ):
                     candidates.add(os.sep.join((item, subitem, "chrome.exe")))
     for candidate in candidates:

src/undetected_chromedriver/patcher.py

@@ -3,9 +3,11 @@
 
 from distutils.version import LooseVersion
 import io
+import json
 import logging
 import os
 import pathlib
+import platform
 import random
 import re
 import shutil
@@ -24,21 +26,9 @@ IS_POSIX = sys.platform.startswith(("darwin", "cygwin", "linux", "linux2"))
 
 class Patcher(object):
     lock = Lock()
-    url_repo = "https://chromedriver.storage.googleapis.com"
-    zip_name = "chromedriver_%s.zip"
     exe_name = "chromedriver%s"
 
     platform = sys.platform
-    if platform.endswith("win32"):
-        zip_name %= "win32"
-        exe_name %= ".exe"
-    if platform.endswith(("linux", "linux2")):
-        zip_name %= "linux64"
-        exe_name %= ""
-    if platform.endswith("darwin"):
-        zip_name %= "mac64"
-        exe_name %= ""
-
     if platform.endswith("win32"):
         d = "~/appdata/roaming/undetected_chromedriver"
     elif "LAMBDA_TASK_ROOT" in os.environ:
@@ -72,6 +62,20 @@ class Patcher(object):
         prefix = "undetected"
         self.user_multi_procs = user_multi_procs
 
+        try:
+            # Try to convert version_main into an integer
+            version_main_int = int(version_main)
+            # check if version_main_int is less than or equal to e.g 114
+            self.is_old_chromedriver = version_main and version_main_int <= 114
+        except (ValueError,TypeError):
+            # If the conversion fails, print an error message
+            print("version_main cannot be converted to an integer")
+            # Set self.is_old_chromedriver to False if the conversion fails
+            self.is_old_chromedriver = False
+
+        # Needs to be called before self.exe_name is accessed
+        self._set_platform_name()
+
         if not os.path.exists(self.data_path):
             os.makedirs(self.data_path, exist_ok=True)
 
@@ -97,9 +101,33 @@ class Patcher(object):
             self._custom_exe_path = True
             self.executable_path = executable_path
 
+        # Set the correct repository to download the Chromedriver from
+        if self.is_old_chromedriver:
+            self.url_repo = "https://chromedriver.storage.googleapis.com"
+        else:
+            self.url_repo = "https://googlechromelabs.github.io/chrome-for-testing"
+
         self.version_main = version_main
         self.version_full = None
 
+    def _set_platform_name(self):
+        """
+        Set the platform and exe name based on the platform undetected_chromedriver is running on
+        in order to download the correct chromedriver.
+        """
+        if self.platform.endswith("win32"):
+            self.platform_name = "win32"
+            self.exe_name %= ".exe"
+        if self.platform.endswith(("linux", "linux2")):
+            self.platform_name = "linux64"
+            self.exe_name %= ""
+        if self.platform.endswith("darwin"):
+            if self.is_old_chromedriver:
+                self.platform_name = "mac64"
+            else:
+                self.platform_name = "mac-x64"
+            self.exe_name %= ""
+
     def auto(self, executable_path=None, force=False, version_main=None, _=None):
         """
 
@@ -111,16 +139,15 @@ class Patcher(object):
         Returns:
 
         """
-        # if self.user_multi_procs and \
-        #         self.user_multi_procs != -1:
-        #     # -1 being a skip value used later in this block
-        #
         p = pathlib.Path(self.data_path)
+        if self.user_multi_procs:
             with Lock():
-            files = list(p.rglob("*chromedriver*?"))
-            for file in files:
-                if self.is_binary_patched(file):
-                    self.executable_path = str(file)
+                files = list(p.rglob("*chromedriver*"))
+                most_recent = max(files, key=lambda f: f.stat().st_mtime)
+                files.remove(most_recent)
+                list(map(lambda f: f.unlink(), files))
+                if self.is_binary_patched(most_recent):
+                    self.executable_path = str(most_recent)
                     return True
 
         if executable_path:
@@ -202,7 +229,11 @@ class Patcher(object):
     def cleanup_unused_files(self):
         p = pathlib.Path(self.data_path)
         items = list(p.glob("*undetected*"))
-        print(items)
+        for item in items:
+            try:
+                item.unlink()
+            except:
+                pass
 
     def patch(self):
         self.patch_exe()
@@ -214,13 +245,33 @@ class Patcher(object):
         :return: version string
         :rtype: LooseVersion
         """
-        path = "/latest_release"
-        if self.version_main:
-            path += f"_{self.version_main}"
+        # Endpoint for old versions of Chromedriver (114 and below)
+        if self.is_old_chromedriver:
+            path = f"/latest_release_{self.version_main}"
             path = path.upper()
             logger.debug("getting release number from %s" % path)
             return LooseVersion(urlopen(self.url_repo + path).read().decode())
 
+        # Endpoint for new versions of Chromedriver (115+)
+        if not self.version_main:
+            # Fetch the latest version
+            path = "/last-known-good-versions-with-downloads.json"
+            logger.debug("getting release number from %s" % path)
+            with urlopen(self.url_repo + path) as conn:
+                response = conn.read().decode()
+
+            last_versions = json.loads(response)
+            return LooseVersion(last_versions["channels"]["Stable"]["version"])
+
+        # Fetch the latest minor version of the major version provided
+        path = "/latest-versions-per-milestone-with-downloads.json"
+        logger.debug("getting release number from %s" % path)
+        with urlopen(self.url_repo + path) as conn:
+            response = conn.read().decode()
+
+        major_versions = json.loads(response)
+        return LooseVersion(major_versions["milestones"][str(self.version_main)]["version"])
+
     def parse_exe_version(self):
         with io.open(self.executable_path, "rb") as f:
             for line in iter(lambda: f.readline(), b""):
@@ -234,10 +285,16 @@ class Patcher(object):
 
         :return: path to downloaded file
         """
-        u = "%s/%s/%s" % (self.url_repo, self.version_full.vstring, self.zip_name)
-        logger.debug("downloading from %s" % u)
-        # return urlretrieve(u, filename=self.data_path)[0]
-        return urlretrieve(u)[0]
+        zip_name = f"chromedriver_{self.platform_name}.zip"
+        if self.is_old_chromedriver:
+            download_url = "%s/%s/%s" % (self.url_repo, self.version_full.vstring, zip_name)
+        else:
+            zip_name = zip_name.replace("_", "-", 1)
+            download_url = "https://edgedl.me.gvt1.com/edgedl/chrome/chrome-for-testing/%s/%s/%s"
+            download_url %= (self.version_full.vstring, self.platform_name, zip_name)
+
+        logger.debug("downloading from %s" % download_url)
+        return urlretrieve(download_url)[0]
 
     def unzip_package(self, fp):
         """
@@ -245,6 +302,12 @@ class Patcher(object):
 
         :return: path to unpacked executable
         """
+        exe_path = self.exe_name
+        if not self.is_old_chromedriver:
+            # The new chromedriver unzips into its own folder
+            zip_name = f"chromedriver-{self.platform_name}"
+            exe_path = os.path.join(zip_name, self.exe_name)
+
         logger.debug("unzipping %s" % fp)
         try:
             os.unlink(self.zip_path)
@@ -253,10 +316,10 @@ class Patcher(object):
 
         os.makedirs(self.zip_path, mode=0o755, exist_ok=True)
         with zipfile.ZipFile(fp, mode="r") as zf:
-            zf.extract(self.exe_name, self.zip_path)
-        os.rename(os.path.join(self.zip_path, self.exe_name), self.executable_path)
+            zf.extractall(self.zip_path)
+        os.rename(os.path.join(self.zip_path, exe_path), self.executable_path)
         os.remove(fp)
-        os.rmdir(self.zip_path)
+        shutil.rmtree
         os.chmod(self.executable_path, 0o755)
         return self.executable_path
 

src/utils.py

@@ -113,7 +113,7 @@ def create_proxy_extension(proxy: dict) -> str:
 
 
 def get_webdriver(proxy: dict = None) -> WebDriver:
-    global PATCHED_DRIVER_PATH
+    global PATCHED_DRIVER_PATH, USER_AGENT
     logging.debug('Launching web browser...')
 
     # undetected_chromedriver
@@ -136,6 +136,14 @@ def get_webdriver(proxy: dict = None) -> WebDriver:
     # https://peter.sh/experiments/chromium-command-line-switches/#use-gl
     options.add_argument('--use-gl=swiftshader')
 
+    language = os.environ.get('LANG', None)
+    if language is not None:
+        options.add_argument('--lang=%s' % language)
+
+    # Fix for Chrome 117 | https://github.com/FlareSolverr/FlareSolverr/issues/910
+    if USER_AGENT is not None:
+        options.add_argument('--user-agent=%s' % USER_AGENT)
+
     proxy_extension_dir = None
     if proxy and all(key in proxy for key in ['url', 'username', 'password']):
         proxy_extension_dir = create_proxy_extension(proxy)
@@ -145,7 +153,7 @@ def get_webdriver(proxy: dict = None) -> WebDriver:
         logging.debug("Using webdriver proxy: %s", proxy_url)
         options.add_argument('--proxy-server=%s' % proxy_url)
 
-    # note: headless mode is detected (options.headless = True)
+    # note: headless mode is detected (headless = True)
     # we launch the browser in head-full mode with the window hidden
     windows_headless = False
     if get_config_headless():
@@ -153,6 +161,8 @@ def get_webdriver(proxy: dict = None) -> WebDriver:
             windows_headless = True
         else:
             start_xvfb_display()
+    # For normal headless mode:
+    # options.add_argument('--headless')
 
     # if we are inside the Docker container, we avoid downloading the driver
     driver_exe_path = None
@@ -162,10 +172,6 @@ def get_webdriver(proxy: dict = None) -> WebDriver:
         driver_exe_path = "/app/chromedriver"
     else:
         version_main = get_chrome_major_version()
-        # Fix for Chrome 115
-        # https://github.com/seleniumbase/SeleniumBase/pull/1967
-        if int(version_main) > 114:
-            version_main = 114
         if PATCHED_DRIVER_PATH is not None:
             driver_exe_path = PATCHED_DRIVER_PATH
 
@@ -295,6 +301,8 @@ def get_user_agent(driver=None) -> str:
         if driver is None:
             driver = get_webdriver()
         USER_AGENT = driver.execute_script("return navigator.userAgent")
+        # Fix for Chrome 117 | https://github.com/FlareSolverr/FlareSolverr/issues/910
+        USER_AGENT = re.sub('HEADLESS', '', USER_AGENT, flags=re.IGNORECASE)
         return USER_AGENT
     except Exception as e:
         raise Exception("Error getting browser User-Agent. " + str(e))