[CaptchaService] New service AdYouLike + AdsCaptcha detect_key fixup

2 files changed, 164 insertions, 124 deletions

diff --git a/module/plugins/hoster/DlFreeFr.py b/module/plugins/hoster/DlFreeFr.py
index 14b74c779..793c81b1c 100644
--- a/module/plugins/hoster/DlFreeFr.py
+++ b/module/plugins/hoster/DlFreeFr.py
@@ -3,9 +3,9 @@
 import pycurl
 import re
 
-from module.common.json_layer import json_loads
 from module.network.Browser import Browser
 from module.network.CookieJar import CookieJar
+from module.plugins.internal.CaptchaService import AdYouLike
 from module.plugins.internal.SimpleHoster import SimpleHoster, create_getInfo, replace_patterns
 
 
@@ -33,87 +33,10 @@ class CustomBrowser(Browser):
         return Browser.load(self, *args, **kwargs)
 
 
-class AdYouLike:
-    """
-    Class to support adyoulike captcha service
-    """
-    ADYOULIKE_INPUT_PATTERN = r'Adyoulike\.create\((.*?)\);'
-    ADYOULIKE_CALLBACK = r'Adyoulike\.g\._jsonp_5579316662423138'
-    ADYOULIKE_CHALLENGE_PATTERN = ADYOULIKE_CALLBACK + r'\((.*?)\)'
-
-
-    def __init__(self, plugin, engine="adyoulike"):
-        self.plugin = plugin
-        self.engine = engine
-
-
-    def challenge(self, html):
-        adyoulike_data_string = None
-        m = re.search(self.ADYOULIKE_INPUT_PATTERN, html)
-        if m:
-            adyoulike_data_string = m.group(1)
-        else:
-            self.plugin.fail("Can't read AdYouLike input data")
-
-        # {"adyoulike":{"key":"P~zQ~O0zV0WTiAzC-iw0navWQpCLoYEP"},
-        # "all":{"element_id":"ayl_private_cap_92300","lang":"fr","env":"prod"}}
-        ayl_data = json_loads(adyoulike_data_string)
-
-        res = self.plugin.load("http://api-ayl.appspot.com/challenge",
-                               get={'key'     : ayl_data[self.engine]['key'],
-                                    'env'     : ayl_data['all']['env'],
-                                    'callback': self.ADYOULIKE_CALLBACK})
-
-        m = re.search(self.ADYOULIKE_CHALLENGE_PATTERN, res)
-        challenge_string = None
-        if m:
-            challenge_string = m.group(1)
-        else:
-            self.plugin.fail("Invalid AdYouLike challenge")
-        challenge_data = json_loads(challenge_string)
-
-        return ayl_data, challenge_data
-
-
-    def result(self, ayl, 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"})
-        """
-        response = None
-        try:
-            instructions_visual = challenge['translations'][ayl['all']['lang']]['instructions_visual']
-            m = re.search(u".*«(.*)».*", instructions_visual)
-            if m:
-                response = m.group(1).strip()
-            else:
-                self.plugin.fail("Can't parse instructions visual")
-        except KeyError:
-            self.plugin.fail("No instructions visual")
-
-        #TODO: Supports captcha
-
-        if not response:
-            self.plugin.fail("AdYouLike result failed")
-
-        return {"_ayl_captcha_engine" : self.engine,
-                "_ayl_env"            : ayl['all']['env'],
-                "_ayl_tid"            : challenge['tid'],
-                "_ayl_token_challenge": challenge['token'],
-                "_ayl_response"       : response}
-
-
 class DlFreeFr(SimpleHoster):
     __name__    = "DlFreeFr"
     __type__    = "hoster"
-    __version__ = "0.25"
+    __version__ = "0.26"
 
     __pattern__ = r'http://(?:www\.)?dl\.free\.fr/(\w+|getfile\.pl\?file=/\w+)'
 
@@ -169,9 +92,7 @@ class DlFreeFr(SimpleHoster):
         action, inputs = self.parseHtmlForm('action="getfile.pl"')
 
         adyoulike = AdYouLike(self)
-        ayl, challenge = adyoulike.challenge(self.html)
-        result = adyoulike.result(ayl, challenge)
-        inputs.update(result)
+        inputs.update(adyoulike.challenge())
 
         self.load("http://dl.free.fr/getfile.pl", post=inputs)
         headers = self.getLastHeaders()
diff --git a/module/plugins/internal/CaptchaService.py b/module/plugins/internal/CaptchaService.py
index 7009e6986..965799e8e 100644
--- a/module/plugins/internal/CaptchaService.py
+++ b/module/plugins/internal/CaptchaService.py
@@ -4,10 +4,12 @@ import re
 
 from random import random
 
+from module.common.json_layer import json_loads
+
 
 class CaptchaService:
     __name__    = "CaptchaService"
-    __version__ = "0.15"
+    __version__ = "0.16"
 
     __description__ = """Base captcha service plugin"""
     __license__     = "GPLv3"
@@ -34,7 +36,7 @@ class CaptchaService:
 
         m = re.search(self.KEY_PATTERN, html)
         if m:
-            self.key = m.group("KEY")
+            self.key = m.group(1).strip()
             self.plugin.logDebug("%s key: %s" % (self.__name__, self.key))
             return self.key
         else:
@@ -59,8 +61,8 @@ class ReCaptcha(CaptchaService):
     __authors__     = [("pyLoad Team", "admin@pyload.org")]
 
 
-    KEY_PATTERN = r'recaptcha(/api|\.net)/(challenge|noscript)\?k=(?P<KEY>[\w-]+)'
-    KEY_AJAX_PATTERN = r'Recaptcha\.create\s*\(\s*["\'](?P<KEY>[\w-]+)'
+    KEY_PATTERN      = r'recaptcha(?:/api|\.net)/(?:challenge|noscript)\?k=([\w-]+)'
+    KEY_AJAX_PATTERN = r'Recaptcha\.create\s*\(\s*["\']([\w-]+)'
 
 
     def detect_key(self, html=None):
@@ -74,7 +76,7 @@ class ReCaptcha(CaptchaService):
 
         m = re.search(self.KEY_PATTERN, html) or re.search(self.KEY_AJAX_PATTERN, html)
         if m:
-            self.key = m.group("KEY")
+            self.key = m.group(1).strip()
             self.plugin.logDebug("ReCaptcha key: %s" % self.key)
             return self.key
         else:
@@ -91,36 +93,43 @@ class ReCaptcha(CaptchaService):
                 self.plugin.fail(errmsg)
                 raise TypeError(errmsg)
 
-        js = self.plugin.req.load("http://www.google.com/recaptcha/api/challenge", get={'k': key})
+        html = self.plugin.req.load("http://www.google.com/recaptcha/api/challenge", get={'k': key})
         try:
-            challenge = re.search("challenge : '(.+?)',", js).group(1)
-            server = re.search("server : '(.+?)',", js).group(1)
+            challenge = re.search("challenge : '(.+?)',", html).group(1)
+            server    = re.search("server : '(.+?)',", html).group(1)
         except:
-            self.plugin.error("ReCaptcha challenge pattern not found")
-
-        result = self.result(server, challenge)
+            errmsg = _("ReCaptcha challenge pattern not found")
+            self.plugin.fail(errmsg)
+            raise ValueError(errmsg)
 
-        self.plugin.logDebug("ReCaptcha result: %s" % result, "challenge: %s" % challenge)
+        self.plugin.logDebug("ReCaptcha challenge: %s" % challenge)
 
-        return challenge, result
+        return challenge, self.result(server, challenge)
 
 
     def result(self, server, challenge):
-        return self.plugin.decryptCaptcha("%simage" % server, get={'c': challenge},
-                                          cookies=True, forceUser=True, imgtype="jpg")
+        result = self.plugin.decryptCaptcha("%simage" % server,
+                                            get={'c': challenge},
+                                            cookies=True,
+                                            forceUser=True,
+                                            imgtype="jpg")
+
+        self.plugin.logDebug("ReCaptcha result: %s" % result)
+
+        return result
 
 
 class AdsCaptcha(CaptchaService):
     __name__    = "AdsCaptcha"
-    __version__ = "0.05"
+    __version__ = "0.06"
 
     __description__ = """AdsCaptcha captcha service plugin"""
     __license__     = "GPLv3"
     __authors__     = [("pyLoad Team", "admin@pyload.org")]
 
 
-    ID_PATTERN = r'api\.adscaptcha\.com/Get\.aspx\?[^"\']*CaptchaId=(?P<ID>\d+)'
-    KEY_PATTERN = r'api\.adscaptcha\.com/Get\.aspx\?[^"\']*PublicKey=(?P<KEY>[\w-]+)'
+    CAPTCHAID_PATTERN  = r'api\.adscaptcha\.com/Get\.aspx\?[^"\']*CaptchaId=(\d+)'
+    PUBLICKEY_PATTERN = r'api\.adscaptcha\.com/Get\.aspx\?[^"\']*PublicKey=([\w-]+)'
 
 
     def detect_key(self, html=None):
@@ -132,18 +141,18 @@ class AdsCaptcha(CaptchaService):
                 self.plugin.fail(errmsg)
                 raise TypeError(errmsg)
 
-        m = re.search(self.ID_PATTERN, html)
-        n = re.search(self.KEY_PATTERN, html)
+        m = re.search(self.PUBLICKEY_PATTERN, html)
+        n = re.search(self.CAPTCHAID_PATTERN, html)
         if m and n:
-            self.key = (m.group("ID"), m.group("KEY"))
-            self.plugin.logDebug("AdsCaptcha id|key: %s | %s" % self.key)
+            self.key = (m.group(1).strip(), n.group(1).strip())  #: key is the tuple(PublicKey, CaptchaId)
+            self.plugin.logDebug("AdsCaptcha key|id: %s | %s" % self.key)
             return self.key
         else:
-            self.plugin.logDebug("AdsCaptcha id or key not found")
+            self.plugin.logDebug("AdsCaptcha key or id not found")
             return None
 
 
-    def challenge(self, key=None):  #: key is a tuple(CaptchaId, PublicKey)
+    def challenge(self, key=None):
         if not key:
             if self.detect_key():
                 key = self.key
@@ -152,25 +161,31 @@ class AdsCaptcha(CaptchaService):
                 self.plugin.fail(errmsg)
                 raise TypeError(errmsg)
 
-        CaptchaId, PublicKey = key
+        PublicKey, CaptchaId = key
 
-        js = self.plugin.req.load("http://api.adscaptcha.com/Get.aspx", get={'CaptchaId': CaptchaId, 'PublicKey': PublicKey})
+        html = self.plugin.req.load("http://api.adscaptcha.com/Get.aspx", get={'CaptchaId': CaptchaId, 'PublicKey': PublicKey})
         try:
-            challenge = re.search("challenge: '(.+?)',", js).group(1)
-            server = re.search("server: '(.+?)',", js).group(1)
+            challenge = re.search("challenge: '(.+?)',", html).group(1)
+            server    = re.search("server: '(.+?)',", html).group(1)
         except:
-            self.plugin.error("AdsCaptcha challenge pattern not found")
-
-        result = self.result(server, challenge)
+            errmsg = _("AdsCaptcha challenge pattern not found")
+            self.plugin.fail(errmsg)
+            raise ValueError(errmsg)
 
-        self.plugin.logDebug("AdsCaptcha result: %s" % result, "challenge: %s" % challenge)
+        self.plugin.logDebug("AdsCaptcha challenge: %s" % challenge)
 
-        return challenge, result
+        return challenge, self.result(server, challenge)
 
 
     def result(self, server, challenge):
-        return self.plugin.decryptCaptcha("%sChallenge.aspx" % server, get={'cid': challenge, 'dummy': random()},
-                                          cookies=True, imgtype="jpg")
+        result = self.plugin.decryptCaptcha("%sChallenge.aspx" % server,
+                                            get={'cid': challenge, 'dummy': random()},
+                                            cookies=True,
+                                            imgtype="jpg")
+
+        self.plugin.logDebug("AdsCaptcha result: %s" % result)
+
+        return result
 
 
 class SolveMedia(CaptchaService):
@@ -182,7 +197,7 @@ class SolveMedia(CaptchaService):
     __authors__     = [("pyLoad Team", "admin@pyload.org")]
 
 
-    KEY_PATTERN = r'api\.solvemedia\.com/papi/challenge\.(no)?script\?k=(?P<KEY>.+?)["\']'
+    KEY_PATTERN = r'api\.solvemedia\.com/papi/challenge\.(?:no)?script\?k=(.+?)["\']'
 
 
     def challenge(self, key=None):
@@ -198,16 +213,120 @@ class SolveMedia(CaptchaService):
         try:
             challenge = re.search(r'<input type=hidden name="adcopy_challenge" id="adcopy_challenge" value="([^"]+)">',
                                   html).group(1)
-            server = "http://api.solvemedia.com/papi/media"
+            server    = "http://api.solvemedia.com/papi/media"
         except:
-            self.plugin.error("SolveMedia challenge pattern not found")
+            errmsg = _("SolveMedia challenge pattern not found")
+            self.plugin.fail(errmsg)
+            raise ValueError(errmsg)
+
+        self.plugin.logDebug("SolveMedia challenge: %s" % challenge)
+
+        return challenge, self.result(server, challenge)
+
+
+    def result(self, server, challenge):
+        result = self.plugin.decryptCaptcha(server, get={'c': challenge}, imgtype="gif")
+
+        self.plugin.logDebug("SolveMedia result: %s" % result)
+
+        return result
+
+
+class AdYouLike(CaptchaService):
+    __name__    = "AdYouLike"
+    __version__ = "0.02"
+
+    __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, html=None):
+        if not html:
+            if hasattr(self.plugin, "html") and self.plugin.html:
+                html = self.plugin.html
+            else:
+                errmsg = _("AdYouLike html not found")
+                self.plugin.fail(errmsg)
+                raise TypeError(errmsg)
 
-        result = self.result(server, challenge)
+        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.plugin.logDebug("AdYouLike ayl|callback: %s | %s" % self.key)
+            return self.key   #: key is the tuple(ayl, callback)
+        else:
+            self.plugin.logDebug("AdYouLike ayl or callback not found")
+            return None
 
-        self.plugin.logDebug("SolveMedia result: %s" % result, "challenge: %s" % challenge)
 
-        return challenge, result
+    def challenge(self, key=None):
+        if not key:
+            if self.detect_key():
+                key = self.key
+            else:
+                errmsg = _("AdYouLike key not found")
+                self.plugin.fail(errmsg)
+                raise TypeError(errmsg)
+
+        ayl, callback = key
+
+        # {"adyoulike":{"key":"P~zQ~O0zV0WTiAzC-iw0navWQpCLoYEP"},
+        # "all":{"element_id":"ayl_private_cap_92300","lang":"fr","env":"prod"}}
+        ayl = json_loads(ayl)
+
+        html = self.plugin.req.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:
+            errmsg = _("AdYouLike challenge pattern not found")
+            self.plugin.fail(errmsg)
+            raise ValueError(errmsg)
+
+        self.plugin.logDebug("AdYouLike challenge: %s" % challenge)
+
+        return self.result(ayl, challenge)
 
 
     def result(self, server, challenge):
-        return self.plugin.decryptCaptcha(server, get={'c': challenge}, imgtype="gif")
+        # 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:
+            errmsg = _("AdYouLike result not found")
+            self.plugin.fail(errmsg)
+            raise ValueError(errmsg)
+
+        result = {'_ayl_captcha_engine' : "adyoulike",
+                  '_ayl_env'            : server['all']['env'],
+                  '_ayl_tid'            : challenge['tid'],
+                  '_ayl_token_challenge': challenge['token'],
+                  '_ayl_response'       : response}
+
+        self.plugin.logDebug("AdYouLike result: %s" % result)
+
+        return result