custom exceptions addtions / fixes.

README.md

@@ -40,7 +40,6 @@ Alternatively, clone this repository and run `python setup.py install`.
 
 - Python 2.7 - 3.x
 - **[Requests](https://github.com/kennethreitz/requests)** >= 2.9.2
-- **[Brotli](https://pypi.org/project/Brotli/)** >= 1.0.7
 - **[requests_toolbelt](https://pypi.org/project/requests-toolbelt/)** >= 0.9.1
 
 `python setup.py install` will install the Python dependencies automatically. The javascript interpreters and/or engines you decide to use are the only things you need to install yourself, excluding js2py which is part of the requirements as the default.

cloudscraper/__init__.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import re
 import sys
 import ssl
@@ -43,6 +44,10 @@
 except ImportError:
     from urllib.parse import urlparse
 
+# Add exceptions path
+sys.path.append(os.path.join(os.path.dirname(__file__), 'exceptions'))
+import cloudflare_exceptions  # noqa: E402
+
 # ------------------------------------------------------------------------------- #
 
 __version__ = '1.2.21'
@@ -198,7 +203,7 @@ def request(self, method, url, *args, **kwargs):
                 sys.tracebacklimit = 0
                 _ = self._solveDepthCnt
                 self._solveDepthCnt = 0
-                raise RuntimeError(
+                raise cloudflare_exceptions.Cloudflare_Loop_Protection(
                     "!!Loop Protection!! We have tried to solve {} time(s) in a row.".format(_)
                 )
 
@@ -281,7 +286,7 @@ def is_Firewall_Blocked(resp):
     def is_Challenge_Request(self, resp):
         if self.is_Firewall_Blocked(resp):
             sys.tracebacklimit = 0
-            raise RuntimeError('Cloudflare has blocked this request (Code 1020 Detected).')
+            raise cloudflare_exceptions.Cloudflare_Block('Cloudflare has blocked this request (Code 1020 Detected).')
 
         if self.is_reCaptcha_Challenge(resp) or self.is_IUAM_Challenge(resp):
             return True
@@ -319,7 +324,7 @@ def unescape(html_text):
 
         except AttributeError:
             sys.tracebacklimit = 0
-            raise RuntimeError(
+            raise cloudflare_exceptions.Cloudflare_Error_IUAM(
                 "Cloudflare IUAM detected, unfortunately we can't extract the parameters correctly."
             )
 
@@ -330,7 +335,7 @@ def unescape(html_text):
                 interpreter
             ).solveChallenge(body, hostParsed.netloc)
         except Exception as e:
-            raise RuntimeError(
+            raise cloudflare_exceptions.Cloudflare_Error_IUAM(
                 'Unable to parse Cloudflare anti-bots page: {}'.format(
                     getattr(e, 'message', e)
                 )
@@ -359,7 +364,7 @@ def reCaptcha_Challenge_Response(provider, provider_params, body, url):
             ).groupdict()
         except (AttributeError):
             sys.tracebacklimit = 0
-            raise RuntimeError(
+            raise cloudflare_exceptions.Cloudflare_Error_reCaptcha(
                 "Cloudflare reCaptcha detected, unfortunately we can't extract the parameters correctly."
             )
 
@@ -406,7 +411,7 @@ def Challenge_Response(self, resp, **kwargs):
 
             if not self.recaptcha or not isinstance(self.recaptcha, dict) or not self.recaptcha.get('provider'):
                 sys.tracebacklimit = 0
-                raise RuntimeError(
+                raise cloudflare_exceptions.Cloudflare_reCaptcha_Provider(
                     "Cloudflare reCaptcha detected, unfortunately you haven't loaded an anti reCaptcha provider "
                     "correctly via the 'recaptcha' parameter."
                 )
@@ -442,7 +447,7 @@ def Challenge_Response(self, resp, **kwargs):
                         self.delay = delay
                 except (AttributeError, ValueError):
                     sys.tracebacklimit = 0
-                    raise RuntimeError("Cloudflare IUAM possibility malformed, issue extracing delay value.")
+                    raise cloudflare_exceptions.Cloudflare_Error_IUAM("Cloudflare IUAM possibility malformed, issue extracing delay value.")
 
             sleep(self.delay)
 
@@ -590,7 +595,7 @@ def get_tokens(cls, url, **kwargs):
                 break
         else:
             sys.tracebacklimit = 0
-            raise RuntimeError(
+            raise cloudflare_exceptions.Cloudflare_Error_IUAM(
                 "Unable to find Cloudflare cookies. Does the site actually "
                 "have Cloudflare IUAM (I'm Under Attack Mode) enabled?"
             )

cloudscraper/exceptions/cloudflare_exceptions.py

@@ -0,0 +1,31 @@
+# ------------------------------------------------------------------------------- #
+
+
+class Cloudflare_Loop_Protection(Exception):
+    """
+    Raise error for recursive depth protection
+    """
+
+
+class Cloudflare_Block(Exception):
+    """
+    Raise error for Cloudflare 1020 block
+    """
+
+
+class Cloudflare_Error_IUAM(Exception):
+    """
+    Raise error for problem extracting IUAM paramters from Cloudflare payload
+    """
+
+
+class Cloudflare_Error_reCaptcha(Exception):
+    """
+    Raise error for problem extracting reCaptcha paramters from Cloudflare payload
+    """
+
+
+class Cloudflare_reCaptcha_Provider(Exception):
+    """
+    Raise error for reCaptcha from Cloudflare, no provider loaded.
+    """

cloudscraper/reCaptcha/exceptions.py → cloudscraper/exceptions/reCaptcha_exceptions.py

@@ -1,49 +1,49 @@
 # ------------------------------------------------------------------------------- #
 
 
-class reCaptchaServiceUnavailable(Exception):
+class reCaptcha_Service_Unavailable(Exception):
     """
     Raise error for external services that cannot be reached
     """
 
 
-class reCaptchaErrorFromAPI(Exception):
+class reCaptcha_Error_From_API(Exception):
     """
     Raise error for error from API response.
     """
 
 
-class reCaptchaAccountError(Exception):
+class reCaptcha_Account_Error(Exception):
     """
     Raise error for reCaptcha provider account problem.
     """
 
 
-class reCaptchaTimeout(Exception):
+class reCaptcha_Timeout(Exception):
     """
     Raise error for reCaptcha provider taking too long.
     """
 
 
-class reCaptchaBadParameter(Exception):
+class reCaptcha_Bad_Parameter(NotImplementedError):
     """
     Raise error for bad or missing Parameter.
     """
 
 
-class reCaptchaBadJobID(Exception):
+class reCaptcha_Bad_Job_ID(Exception):
     """
     Raise error for invalid job id.
     """
 
 
-class reCaptchaReportError(Exception):
+class reCaptcha_Report_Error(Exception):
     """
     Raise error for reCaptcha provider unable to report bad solve.
     """
 
 
-class reCaptchaImportError(Exception):
+class reCaptcha_Import_Error(Exception):
     """
     Raise error for reCaptcha, cannot import a module.
     """

cloudscraper/reCaptcha/2captcha.py

@@ -1,14 +1,14 @@
 from __future__ import absolute_import
 
 import requests
-from . import exceptions
+import reCaptcha_exceptions
 
 try:
     import polling
 except ImportError:
     import sys
     sys.tracebacklimit = 0
-    raise exceptions.reCaptchaImportError(
+    raise reCaptcha_exceptions.reCaptcha_Import_Error(
         "Please install the python module 'polling' via pip or download it from "
         "https://github.com/justiniso/polling/"
     )
@@ -28,7 +28,7 @@ def __init__(self):
     @staticmethod
     def checkErrorStatus(response, request_type):
         if response.status_code in [500, 502]:
-            raise exceptions.reCaptchaServiceUnavailable('2Captcha: Server Side Error {}'.format(response.status_code))
+            raise reCaptcha_exceptions.reCaptcha_Service_Unavailable('2Captcha: Server Side Error {}'.format(response.status_code))
 
         errors = {
             'in.php': {
@@ -75,7 +75,7 @@ def checkErrorStatus(response, request_type):
         }
 
         if response.json().get('status') is False and response.json().get('request') in errors.get(request_type):
-            raise exceptions.reCaptchaErrorFromAPI(
+            raise reCaptcha_exceptions.reCaptcha_Error_From_API(
                 '{} {}'.format(
                     response.json().get('request'),
                     errors.get(request_type).get(response.json().get('request'))
@@ -86,7 +86,7 @@ def checkErrorStatus(response, request_type):
 
     def reportJob(self, jobID):
         if not jobID:
-            raise exceptions.reCaptchaBadJobID(
+            raise reCaptcha_exceptions.reCaptcha_Bad_Job_ID(
                 "2Captcha: Error bad job id to request reCaptcha."
             )
 
@@ -116,7 +116,7 @@ def _checkRequest(response):
         if response:
             return True
         else:
-            raise exceptions.reCaptchaReportError(
+            raise reCaptcha_exceptions.reCaptcha_Report_Error(
                 "2Captcha: Error - Failed to report bad reCaptcha solve."
             )
 
@@ -152,7 +152,7 @@ def _checkRequest(response):
         if response:
             return response.json().get('request')
         else:
-            raise exceptions.reCaptchaTimeout(
+            raise reCaptcha_exceptions.reCaptcha_Timeout(
                 "2Captcha: Error failed to solve reCaptcha."
             )
 
@@ -188,7 +188,7 @@ def _checkRequest(response):
         if response:
             return response.json().get('request')
         else:
-            raise exceptions.reCaptchaBadJobID(
+            raise reCaptcha_exceptions.reCaptcha_Bad_Job_ID(
                 '2Captcha: Error no job id was returned.'
             )
 
@@ -198,7 +198,7 @@ def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
         jobID = None
 
         if not reCaptchaParams.get('api_key'):
-            raise exceptions.reCaptchaBadParameter(
+            raise reCaptcha_exceptions.reCaptcha_Bad_Parameter(
                 "2Captcha: Missing api_key parameter."
             )
 
@@ -215,11 +215,11 @@ def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
                 if jobID:
                     self.reportJob(jobID)
             except polling.TimeoutException:
-                raise exceptions.reCaptchaTimeout(
+                raise reCaptcha_exceptions.reCaptcha_Timeout(
                     "2Captcha: reCaptcha solve took to long and also failed reporting the job the job id {}.".format(jobID)
                 )
 
-            raise exceptions.reCaptchaTimeout(
+            raise reCaptcha_exceptions.reCaptcha_Timeout(
                 "2Captcha: reCaptcha solve took to long to execute job id {}, aborting.".format(jobID)
             )
 

cloudscraper/reCaptcha/9kw.py

@@ -2,14 +2,14 @@
 
 import re
 import requests
-from . import exceptions
+import reCaptcha_exceptions
 
 try:
     import polling
 except ImportError:
     import sys
     sys.tracebacklimit = 0
-    raise exceptions.reCaptchaImportError(
+    raise reCaptcha_exceptions.reCaptcha_Import_Error(
         "Please install the python module 'polling' via pip or download it from "
         "https://github.com/justiniso/polling/"
     )
@@ -30,7 +30,7 @@ def __init__(self):
     @staticmethod
     def checkErrorStatus(response):
         if response.status_code in [500, 502]:
-            raise exceptions.reCaptchaServiceUnavailable(
+            raise reCaptcha_exceptions.reCaptcha_Service_Unavailable(
                 '9kw: Server Side Error {}'.format(response.status_code)
             )
 
@@ -93,17 +93,17 @@ def checkErrorStatus(response):
 
         if response.text.startswith('{'):
             if response.json().get('error'):
-                raise exceptions.reCaptchaErrorFromAPI(error_codes.get(int(response.json().get('error'))))
+                raise reCaptcha_exceptions.reCaptcha_Error_From_API(error_codes.get(int(response.json().get('error'))))
         else:
             error_code = int(re.search(r'^00(?P<error_code>\d+)', response.text).groupdict().get('error_code', 0))
             if error_code:
-                raise exceptions.reCaptchaErrorFromAPI(error_codes.get(error_code))
+                raise reCaptcha_exceptions.reCaptcha_Error_From_API(error_codes.get(error_code))
 
     # ------------------------------------------------------------------------------- #
 
     def requestJob(self, jobID):
         if not jobID:
-            raise exceptions.reCaptchaBadJobID(
+            raise reCaptcha_exceptions.reCaptcha_Bad_Job_ID(
                 "9kw: Error bad job id to request reCaptcha against."
             )
 
@@ -134,7 +134,7 @@ def _checkRequest(response):
         if response:
             return response.json().get('answer')
         else:
-            raise exceptions.reCaptchaTimeout("9kw: Error failed to solve reCaptcha.")
+            raise reCaptcha_exceptions.reCaptcha_Timeout("9kw: Error failed to solve reCaptcha.")
 
     # ------------------------------------------------------------------------------- #
 
@@ -170,15 +170,15 @@ def _checkRequest(response):
         if response:
             return response.json().get('captchaid')
         else:
-            raise exceptions.reCaptchaBadJobID('9kw: Error no valid job id was returned.')
+            raise reCaptcha_exceptions.reCaptcha_Bad_Job_ID('9kw: Error no valid job id was returned.')
 
     # ------------------------------------------------------------------------------- #
 
     def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
         jobID = None
 
         if not reCaptchaParams.get('api_key'):
-            raise exceptions.reCaptchaBadParameter("9kw: Missing api_key parameter.")
+            raise reCaptcha_exceptions.reCaptcha_Bad_Parameter("9kw: Missing api_key parameter.")
 
         self.api_key = reCaptchaParams.get('api_key')
 
@@ -192,7 +192,7 @@ def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
             jobID = self.requestSolve(site_url, site_key)
             return self.requestJob(jobID)
         except polling.TimeoutException:
-            raise exceptions.reCaptchaTimeout(
+            raise reCaptcha_exceptions.reCaptcha_Timeout(
                 "9kw: reCaptcha solve took to long to execute 'captchaid' {}, aborting.".format(jobID)
             )
 

cloudscraper/reCaptcha/anticaptcha.py

@@ -1,13 +1,13 @@
 from __future__ import absolute_import
 
 import sys
-from . import exceptions
+import reCaptcha_exceptions
 
 try:
     from python_anticaptcha import AnticaptchaClient, NoCaptchaTaskProxylessTask
 except ImportError:
     sys.tracebacklimit = 0
-    raise exceptions.reCaptchaImportError(
+    raise reCaptcha_exceptions.reCaptcha_Import_Error(
         "Please install the python module 'python_anticaptcha' via pip or download it from "
         "https://github.com/ad-m/python-anticaptcha"
     )
@@ -24,7 +24,7 @@ def __init__(self):
 
     def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
         if not reCaptchaParams.get('api_key'):
-            raise exceptions.reCaptchaBadParameter("anticaptcha: Missing api_key parameter.")
+            raise reCaptcha_exceptions.reCaptcha_Bad_Parameter("anticaptcha: Missing api_key parameter.")
 
         client = AnticaptchaClient(reCaptchaParams.get('api_key'))
 
@@ -35,7 +35,7 @@ def getCaptchaAnswer(self, site_url, site_key, reCaptchaParams):
 
         if not hasattr(client, 'createTaskSmee'):
             sys.tracebacklimit = 0
-            raise exceptions.reCaptchaImportError(
+            raise reCaptcha_exceptions.reCaptcha_Import_Error(
                 "Please upgrade 'python_anticaptcha' via pip or download it from https://github.com/ad-m/python-anticaptcha"
             )