Improve Captcha

7 files changed, 53 insertions, 504 deletions

diff --git a/module/plugins/internal/AdYouLike.py b/module/plugins/internal/AdYouLike.py
deleted file mode 100644
index d14babb51..000000000
--- a/module/plugins/internal/AdYouLike.py
+++ /dev/null
@@ -1,92 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import re
-
-from module.common.json_layer import json_loads
-from module.plugins.internal.CaptchaService import CaptchaService
-
-
-class AdYouLike(CaptchaService):
-    __name__    = "AdYouLike"
-    __type__    = "captcha"
-    __version__ = "0.07"
-    __status__  = "stable"
-
-    __description__ = """AdYouLike captcha service plugin"""
-    __license__     = "GPLv3"
-    __authors__     = [("Walter Purcaro", "vuolter@gmail.com")]
-
-
-    AYL_PATTERN      = r'Adyoulike\.create\s*\((.+?)\)'
-    CALLBACK_PATTERN = r'(Adyoulike\.g\._jsonp_\d+)'
-
-
-    def detect_key(self, data=None):
-        html = data or self.retrieve_data()
-
-        m = re.search(self.AYL_PATTERN, html)
-        n = re.search(self.CALLBACK_PATTERN, html)
-        if m and n:
-            self.key = (m.group(1).strip(), n.group(1).strip())
-            self.log_debug("Ayl: %s | Callback: %s" % self.key)
-            return self.key   #: Key is the tuple(ayl, callback)
-        else:
-            self.log_warning(_("Ayl or callback pattern not found"))
-            return None
-
-
-    def challenge(self, key=None, data=None):
-        ayl, callback = key or self.retrieve_key(data)
-
-        #: {'adyoulike':{'key':"P~zQ~O0zV0WTiAzC-iw0navWQpCLoYEP"},
-        #: 'all':{'element_id':"ayl_private_cap_92300",'lang':"fr",'env':"prod"}}
-        ayl = json_loads(ayl)
-
-        html = self.plugin.load("http://api-ayl.appspot.com/challenge",
-                                    get={'key'     : ayl['adyoulike']['key'],
-                                         'env'     : ayl['all']['env'],
-                                         'callback': callback})
-        try:
-            challenge = json_loads(re.search(callback + r'\s*\((.+?)\)', html).group(1))
-
-        except AttributeError:
-            self.fail(_("AdYouLike challenge pattern not found"))
-
-        self.log_debug("Challenge: %s" % challenge)
-
-        return self.result(ayl, challenge), challenge
-
-
-    def result(self, server, challenge):
-        #: Adyoulike.g._jsonp_5579316662423138
-        #: ({'translations':{'fr':{'instructions_visual':"Recopiez « Soonnight » ci-dessous :"}},
-        #: 'site_under':true,'clickable':true,'pixels':{'VIDEO_050':[],'DISPLAY':[],'VIDEO_000':[],'VIDEO_100':[],
-        #: 'VIDEO_025':[],'VIDEO_075':[]},'medium_type':"image/adyoulike",
-        #: 'iframes':{'big':"<iframe src=\"http://www.soonnight.com/campagn.html\" scrolling=\"no\"
-        #: height=\"250\" width=\"300\" frameborder=\"0\"></iframe>"},'shares':{},'id':256,
-        #: 'token':"e6QuI4aRSnbIZJg02IsV6cp4JQ9~MjA1",'formats':{'small':{'y':300,'x':0,'w':300,'h':60},
-        #: 'big':{'y':0,'x':0,'w':300,'h':250},'hover':{'y':440,'x':0,'w':300,'h':60}},
-        #: 'tid':"SqwuAdxT1EZoi4B5q0T63LN2AkiCJBg5"})
-
-        if isinstance(server, basestring):
-            server = json_loads(server)
-
-        if isinstance(challenge, basestring):
-            challenge = json_loads(challenge)
-
-        try:
-            instructions_visual = challenge['translations'][server['all']['lang']]['instructions_visual']
-            result = re.search(u'«(.+?)»', instructions_visual).group(1).strip()
-
-        except AttributeError:
-            self.fail(_("AdYouLike result not found"))
-
-        result = {'_ayl_captcha_engine' : "adyoulike",
-                  '_ayl_env'            : server['all']['env'],
-                  '_ayl_tid'            : challenge['tid'],
-                  '_ayl_token_challenge': challenge['token'],
-                  '_ayl_response'       : response}
-
-        self.log_debug("Result: %s" % result)
-
-        return result
diff --git a/module/plugins/internal/AdsCaptcha.py b/module/plugins/internal/AdsCaptcha.py
deleted file mode 100644
index f487042e2..000000000
--- a/module/plugins/internal/AdsCaptcha.py
+++ /dev/null
@@ -1,64 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import random
-import re
-
-from module.plugins.internal.CaptchaService import CaptchaService
-
-
-class AdsCaptcha(CaptchaService):
-    __name__    = "AdsCaptcha"
-    __type__    = "captcha"
-    __version__ = "0.10"
-    __status__  = "stable"
-
-    __description__ = """AdsCaptcha captcha service plugin"""
-    __license__     = "GPLv3"
-    __authors__     = [("pyLoad Team", "admin@pyload.org")]
-
-
-    CAPTCHAID_PATTERN  = r'api\.adscaptcha\.com/Get\.aspx\?.*?CaptchaId=(\d+)'
-    PUBLICKEY_PATTERN = r'api\.adscaptcha\.com/Get\.aspx\?.*?PublicKey=([\w-]+)'
-
-
-    def detect_key(self, data=None):
-        html = data or self.retrieve_data()
-
-        m = re.search(self.PUBLICKEY_PATTERN, html)
-        n = re.search(self.CAPTCHAID_PATTERN, html)
-        if m and n:
-            self.key = (m.group(1).strip(), n.group(1).strip())  #: Key is the tuple(PublicKey, CaptchaId)
-            self.log_debug("Key: %s | ID: %s" % self.key)
-            return self.key
-        else:
-            self.log_warning(_("Key or id pattern not found"))
-            return None
-
-
-    def challenge(self, key=None, data=None):
-        PublicKey, CaptchaId = key or self.retrieve_key(data)
-
-        html = self.plugin.load("http://api.adscaptcha.com/Get.aspx",
-                                    get={'CaptchaId': CaptchaId,
-                                         'PublicKey': PublicKey})
-        try:
-            challenge = re.search("challenge: '(.+?)',", html).group(1)
-            server    = re.search("server: '(.+?)',", html).group(1)
-
-        except AttributeError:
-            self.fail(_("AdsCaptcha challenge pattern not found"))
-
-        self.log_debug("Challenge: %s" % challenge)
-
-        return self.result(server, challenge), challenge
-
-
-    def result(self, server, challenge):
-        result = self.decrypt_image("%sChallenge.aspx" % server,
-                                    get={'cid': challenge, 'dummy': random.random()},
-                                    cookies=True,
-                                    input_type="jpg")
-
-        self.log_debug("Result: %s" % result)
-
-        return result
diff --git a/module/plugins/internal/Captcha.py b/module/plugins/internal/Captcha.py
index af7f66ed5..942021f26 100644
--- a/module/plugins/internal/Captcha.py
+++ b/module/plugins/internal/Captcha.py
@@ -1,12 +1,14 @@
 # -*- coding: utf-8 -*-
 
+import time
+
 from module.plugins.internal.Plugin import Plugin
 
 
 class Captcha(Plugin):
     __name__    = "Captcha"
     __type__    = "captcha"
-    __version__ = "0.01"
+    __version__ = "0.02"
     __status__  = "stable"
 
     __description__ = """Base anti-captcha plugin"""
@@ -35,13 +37,21 @@ class Captcha(Plugin):
         pass
 
 
-    def decrypt_image(self, url, get={}, post={}, ref=False, cookies=False, decode=False,
-                      input_type='png', output_type='textual', try_ocr=True):
-        image = self.load(url, get=get, post=post, ref=ref, cookies=cookies, decode=decode)
-        return self.decrypt(image, input_type, output_type, try_ocr)
+    def recognize(self, image):
+        """
+        Extend to build your custom anti-captcha ocr
+        """
+        pass
+
+
+    def decrypt(self, url, get={}, post={}, ref=False, cookies=False, decode=False,
+                input_type='png', output_type='textual', ocr=True):
+        img = self.load(url, get=get, post=post, ref=ref, cookies=cookies, decode=decode)
+        return self._decrypt(img, input_type, output_type, ocr)
 
 
-    def decrypt(self, data, input_type='png', output_type='textual', try_ocr=True):
+    #@TODO: Definitely dhoose a better name for this method!
+    def _decrypt(self, raw, input_type='png', output_type='textual', ocr=None):
         """
         Loads a captcha and decrypts it with ocr, plugin, user input
 
@@ -53,56 +63,53 @@ class Captcha(Plugin):
         :param output_type: 'textual' if text is written on the captcha\
         or 'positional' for captcha where the user have to click\
         on a specific region on the captcha
-        :param try_ocr: if True, ocr is not used
+        :param ocr: if True, ocr is not used
 
         :return: result of decrypting
         """
-        id = ("%.2f" % time.time())[-6:].replace(".", "")
+        time_ref = ("%.2f" % time.time())[-6:].replace(".", "")
+
+        with open(os.path.join("tmp", "captcha_image_%s_%s.%s" % (self.plugin.__name__, time_ref, input_type)), "wb") as tmp_img:
+            tmp_img.write(raw)
+
+            if ocr is not False:
+                if isinstance(ocr, basestring):
+                    OCR = self.pyload.pluginManager.loadClass("captcha", ocr)  #: Rename `captcha` to `ocr` in 0.4.10
 
-        with open(os.path.join("tmp", "tmpCaptcha_%s_%s.%s" % (self.plugin.__name__, id, input_type)), "wb") as tmpCaptcha:
-            tmpCaptcha.write(img)
+                    if self.plugin.pyfile.abort:
+                        self.abort()
 
-        has_plugin = self.plugin.__name__ in self.pyload.pluginManager.ocrPlugins
+                    result = OCR(self.plugin.pyfile).recognize(tmp_img.name)
 
-        if self.pyload.captcha:
-            Ocr = self.pyload.pluginManager.loadClass("ocr", self.plugin.__name__)
-        else:
-            Ocr = None
+                else:
+                    result = self.recognize(tmp_img.name)
 
-        if Ocr and try_ocr:
-            time.sleep(random.randint(3000, 5000) / 1000.0)
-            if self.pyfile.abort:
-                self.abort()
+            else:
+                captchaManager = self.pyload.captchaManager
 
-            ocr = Ocr(self.pyfile)
-            result = ocr.get_captcha(tmpCaptcha.name)
-        else:
-            captchaManager = self.pyload.captchaManager
-            task = captchaManager.newTask(img, input_type, tmpCaptcha.name, output_type)
-            self.task = task
-            captchaManager.handleCaptcha(task)
+                try:
+                    self.task = captchaManager.newTask(img, input_type, tmp_img.name, output_type)
+                    captchaManager.handleCaptcha(self.task)
 
-            while task.isWaiting():
-                if self.pyfile.abort:
-                    captchaManager.removeTask(task)
-                    self.abort()
-                time.sleep(1)
+                    while self.task.isWaiting():
+                        if self.plugin.pyfile.abort:
+                            self.abort()
+                        time.sleep(1)
+                finally:
+                    captchaManager.removeTask(self.task)
 
-            captchaManager.removeTask(task)
+                if self.task.error:
+                    self.fail(task.error)
 
-            if task.error and has_plugin:  #: Ignore default error message since the user could use try_ocr
-                self.fail(_("Pil and tesseract not installed and no Client connected for captcha decrypting"))
-            elif task.error:
-                self.fail(task.error)
-            elif not task.result:
-                self.fail(_("No captcha result obtained in appropiate time by any of the plugins"))
+                elif not self.task.result:
+                    self.fail(_("No captcha result obtained in appropiate time by any of the plugins"))
 
-            result = task.result
-            self.log_debug("Received captcha result: %s" % result)
+                result = task.result
+                self.log_debug("Received captcha result: %s" % result)  #@TODO: Remove from here?
 
         if not self.pyload.debug:
             try:
-                os.remove(tmpCaptcha.name)
+                os.remove(tmp_img.name)
             except Exception:
                 pass
 
diff --git a/module/plugins/internal/OCR.py b/module/plugins/internal/OCR.py
index ee5571f77..5c22866c8 100644
--- a/module/plugins/internal/OCR.py
+++ b/module/plugins/internal/OCR.py
@@ -20,7 +20,7 @@ from module.utils import save_join as fs_join
 class OCR(Plugin):
     __name__    = "OCR"
     __type__    = "ocr"
-    __version__ = "0.12"
+    __version__ = "0.13"
     __status__  = "stable"
 
     __description__ = """OCR base plugin"""
@@ -133,7 +133,7 @@ class OCR(Plugin):
             pass
 
 
-    def get_captcha(self, name):
+    def recognize(self, name):
         raise NotImplementedError
 
 
diff --git a/module/plugins/internal/ReCaptcha.py b/module/plugins/internal/ReCaptcha.py
deleted file mode 100644
index b4f9ef1eb..000000000
--- a/module/plugins/internal/ReCaptcha.py
+++ /dev/null
@@ -1,197 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import random
-import re
-import time
-import urlparse
-
-from base64 import b64encode
-
-from module.plugins.internal.CaptchaService import CaptchaService
-
-
-class ReCaptcha(CaptchaService):
-    __name__    = "ReCaptcha"
-    __type__    = "captcha"
-    __version__ = "0.18"
-    __status__  = "stable"
-
-    __description__ = """ReCaptcha captcha service plugin"""
-    __license__     = "GPLv3"
-    __authors__     = [("pyLoad Team", "admin@pyload.org"),
-                       ("Walter Purcaro", "vuolter@gmail.com"),
-                       ("zapp-brannigan", "fuerst.reinje@web.de")]
-
-
-    KEY_V1_PATTERN = r'(?:recaptcha(?:/api|\.net)/(?:challenge|noscript)\?k=|Recaptcha\.create\s*\(\s*["\'])([\w-]+)'
-    KEY_V2_PATTERN = r'(?:data-sitekey=["\']|["\']sitekey["\']:\s*["\'])([\w-]+)'
-
-
-    def detect_key(self, data=None):
-        html = data or self.retrieve_data()
-
-        m = re.search(self.KEY_V2_PATTERN, html) or re.search(self.KEY_V1_PATTERN, html)
-        if m:
-            self.key = m.group(1).strip()
-            self.log_debug("Key: %s" % self.key)
-            return self.key
-        else:
-            self.log_warning(_("Key pattern not found"))
-            return None
-
-
-    def challenge(self, key=None, data=None, version=None):
-        key = key or self.retrieve_key(data)
-
-        if version in (1, 2):
-            return getattr(self, "_challenge_v%s" % version)(key)
-
-        else:
-            return self.challenge(key,
-                                  version=2 if re.search(self.KEY_V2_PATTERN, html or self.retrieve_data()) else 1)
-
-
-    def _challenge_v1(self, key):
-        html = self.plugin.load("http://www.google.com/recaptcha/api/challenge",
-                                    get={'k': key})
-        try:
-            challenge = re.search("challenge : '(.+?)',", html).group(1)
-            server    = re.search("server : '(.+?)',", html).group(1)
-
-        except AttributeError:
-            self.fail(_("ReCaptcha challenge pattern not found"))
-
-        self.log_debug("Challenge: %s" % challenge)
-
-        return self.result(server, challenge, key)
-
-
-    def result(self, server, challenge, key):
-        self.plugin.load("http://www.google.com/recaptcha/api/js/recaptcha.js")
-        html = self.plugin.load("http://www.google.com/recaptcha/api/reload",
-                                    get={'c'     : challenge,
-                                         'k'     : key,
-                                         'reason': "i",
-                                         'type'  : "image"})
-
-        try:
-            challenge = re.search('\(\'(.+?)\',',html).group(1)
-
-        except AttributeError:
-            self.fail(_("ReCaptcha second challenge pattern not found"))
-
-        self.log_debug("Second challenge: %s" % challenge)
-        result = self.decrypt("%simage" % server,
-                              get={'c': challenge},
-                              cookies=True,
-                              input_type="jpg",
-                              try_ocr=False)
-
-        self.log_debug("Result: %s" % result)
-
-        return result, challenge
-
-
-    def _collect_api_info(self):
-        html = self.plugin.load("http://www.google.com/recaptcha/api.js")
-        a    = re.search(r'po.src = \'(.*?)\';', html).group(1)
-        vers = a.split("/")[5]
-
-        self.log_debug("API version: %s" % vers)
-
-        language = a.split("__")[1].split(".")[0]
-
-        self.log_debug("API language: %s" % language)
-
-        html = self.plugin.load("https://apis.google.com/js/api.js")
-        b    = re.search(r'"h":"(.*?)","', html).group(1)
-        jsh  = b.decode('unicode-escape')
-
-        self.log_debug("API jsh-string: %s" % jsh)
-
-        return vers, language, jsh
-
-
-    def _prepare_time_and_rpc(self):
-        self.plugin.load("http://www.google.com/recaptcha/api2/demo")
-
-        millis = int(round(time.time() * 1000))
-
-        self.log_debug("Time: %s" % millis)
-
-        rand = random.randint(1, 99999999)
-        a    = "0.%s" % str(rand * 2147483647)
-        rpc  = int(100000000 * float(a))
-
-        self.log_debug("Rpc-token: %s" % rpc)
-
-        return millis, rpc
-
-
-    def _challenge_v2(self, key, parent=None):
-        if parent is None:
-            try:
-                parent = urlparse.urljoin("http://", urlparse.urlparse(self.plugin.pyfile.url).netloc)
-
-            except Exception:
-                parent = ""
-
-        botguardstring      = "!A"
-        vers, language, jsh = self._collect_api_info()
-        millis, rpc         = self._prepare_time_and_rpc()
-
-        html = self.plugin.load("https://www.google.com/recaptcha/api2/anchor",
-                                    get={'k'       : key,
-                                         'hl'      : language,
-                                         'v'       : vers,
-                                         'usegapi' : "1",
-                                         'jsh'     : "%s#id=IO_%s" % (jsh, millis),
-                                         'parent'  : parent,
-                                         'pfname'  : "",
-                                         'rpctoken': rpc})
-
-        token1 = re.search(r'id="recaptcha-token" value="(.*?)">', html)
-        self.log_debug("Token #1: %s" % token1.group(1))
-
-        html = self.plugin.load("https://www.google.com/recaptcha/api2/frame",
-                                get={'c'      : token1.group(1),
-                                     'hl'     : language,
-                                     'v'      : vers,
-                                     'bg'     : botguardstring,
-                                     'k'      : key,
-                                     'usegapi': "1",
-                                     'jsh'    : jsh},
-                                decode="unicode-escape")
-
-        token2 = re.search(r'"finput","(.*?)",', html)
-        self.log_debug("Token #2: %s" % token2.group(1))
-
-        token3 = re.search(r'"rresp","(.*?)",', html)
-        self.log_debug("Token #3: %s" % token3.group(1))
-
-        millis_captcha_loading = int(round(time.time() * 1000))
-        captcha_response = self.decrypt_image("https://www.google.com/recaptcha/api2/payload",
-                                              get={'c':token3.group(1), 'k':key},
-                                              cookies=True,
-                                              try_ocr=False)
-        response = b64encode('{"response":"%s"}' % captcha_response)
-
-        self.log_debug("Result: %s" % response)
-
-        timeToSolve     = int(round(time.time() * 1000)) - millis_captcha_loading
-        timeToSolveMore = timeToSolve + int(float("0." + str(random.randint(1, 99999999))) * 500)
-
-        html = self.plugin.load("https://www.google.com/recaptcha/api2/userverify",
-                                    post={'k'       : key,
-                                          'c'       : token3.group(1),
-                                          'response': response,
-                                          't'       : timeToSolve,
-                                          'ct'      : timeToSolveMore,
-                                          'bg'      : botguardstring})
-
-        token4 = re.search(r'"uvresp","(.*?)",', html)
-        self.log_debug("Token #4: %s" % token4.group(1))
-
-        result = token4.group(1)
-
-        return result, None
diff --git a/module/plugins/internal/SolveMedia.py b/module/plugins/internal/SolveMedia.py
deleted file mode 100644
index ce4ebb007..000000000
--- a/module/plugins/internal/SolveMedia.py
+++ /dev/null
@@ -1,105 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import re
-
-from module.plugins.internal.Plugin import Fail
-from module.plugins.internal.CaptchaService import CaptchaService
-
-
-class SolveMedia(CaptchaService):
-    __name__    = "SolveMedia"
-    __type__    = "captcha"
-    __version__ = "0.15"
-    __status__  = "stable"
-
-    __description__ = """SolveMedia captcha service plugin"""
-    __license__     = "GPLv3"
-    __authors__     = [("pyLoad Team", "admin@pyload.org")]
-
-
-    KEY_PATTERN = r'api\.solvemedia\.com/papi/challenge\.(?:no)?script\?k=(.+?)["\']'
-
-
-    def detect_key(self, data=None):
-        html = data or self.retrieve_data()
-
-        m = re.search(self.KEY_PATTERN, html)
-        if m:
-            self.key = m.group(1).strip()
-            self.log_debug("Key: %s" % self.key)
-            return self.key
-        else:
-            self.log_warning(_("Key pattern not found")
-            return None
-
-
-    def challenge(self, key=None, data=None):
-        key = key or self.retrieve_key(data)
-
-        html = self.plugin.load("http://api.solvemedia.com/papi/challenge.noscript",
-                                    get={'k': key})
-
-        for i in xrange(1, 11):
-            try:
-                magic = re.search(r'name="magic" value="(.+?)"', html).group(1)
-
-            except AttributeError:
-                self.log_warning(_("Magic pattern not found")
-                magic = None
-
-            try:
-                challenge = re.search(r'<input type=hidden name="adcopy_challenge" id="adcopy_challenge" value="(.+?)">',
-                                      html).group(1)
-
-            except AttributeError:
-                self.fail(_("SolveMedia challenge pattern not found"))
-
-            else:
-                self.log_debug("Challenge: %s" % challenge)
-
-            try:
-                result = self.result("http://api.solvemedia.com/papi/media", challenge)
-
-            except Fail, e:
-                self.log_warning(e)
-                self.plugin.invalidCaptcha()
-                result = None
-
-            html = self.plugin.load("http://api.solvemedia.com/papi/verify.noscript",
-                                        post={'adcopy_response' : result,
-                                              'k'               : key,
-                                              'l'               : "en",
-                                              't'               : "img",
-                                              's'               : "standard",
-                                              'magic'           : magic,
-                                              'adcopy_challenge': challenge,
-                                              'ref'             : self.plugin.pyfile.url})
-            try:
-                redirect = re.search(r'URL=(.+?)">', html).group(1)
-
-            except AttributeError:
-                self.fail(_("SolveMedia verify pattern not found"))
-
-            else:
-                if "error" in html:
-                    self.log_warning(_("Captcha code was invalid"))
-                    self.log_debug("Retry #%d" % i)
-                    html = self.plugin.load(redirect)
-                else:
-                    break
-
-        else:
-            self.fail(_("SolveMedia max retries exceeded"))
-
-        return result, challenge
-
-
-    def result(self, server, challenge):
-        result = self.decrypt_image(server,
-                                    get={'c': challenge},
-                                    cookies=True,
-                                    input_type="gif")
-
-        self.log_debug("Result: %s" % result)
-
-        return result
diff --git a/module/plugins/internal/XFSHoster.py b/module/plugins/internal/XFSHoster.py
index ec9a18a48..18a50a6b0 100644
--- a/module/plugins/internal/XFSHoster.py
+++ b/module/plugins/internal/XFSHoster.py
@@ -4,8 +4,8 @@ import pycurl
 import random
 import re
 
-from module.plugins.internal.ReCaptcha import ReCaptcha
-from module.plugins.internal.SolveMedia import SolveMedia
+from module.plugins.captcha.ReCaptcha import ReCaptcha
+from module.plugins.captcha.SolveMedia import SolveMedia
 from module.plugins.internal.SimpleHoster import SimpleHoster, create_getInfo, seconds_to_midnight
 from module.utils import html_unescape
 
@@ -221,7 +221,7 @@ class XFSHoster(SimpleHoster):
         m = re.search(self.CAPTCHA_PATTERN, self.html)
         if m:
             captcha_url = m.group(1)
-            inputs['code'] = self.captcha.decrypt_image(captcha_url)
+            inputs['code'] = self.captcha.decrypt(captcha_url)
             return
 
         m = re.search(self.CAPTCHA_BLOCK_PATTERN, self.html, re.S)