1 files changed, 117 insertions, 32 deletions
diff --git a/module/plugins/internal/Captcha.py b/module/plugins/internal/Captcha.py
index b4af46493..c08050ee8 100644
--- a/module/plugins/internal/Captcha.py
+++ b/module/plugins/internal/Captcha.py
@@ -1,56 +1,141 @@
 # -*- coding: utf-8 -*-
 
-from module.plugins.Plugin import Base
+from __future__ import with_statement
 
+import os
+import time
+import traceback
 
-#@TODO: Extend (new) Plugin class; remove all `html` args
-class Captcha(Base):
+from module.plugins.internal.Plugin import Plugin
+
+
+class Captcha(Plugin):
     __name__    = "Captcha"
     __type__    = "captcha"
-    __version__ = "0.29"
+    __version__ = "0.42"
+    __status__  = "testing"
 
-    __description__ = """Base captcha service plugin"""
+    __description__ = """Base anti-captcha plugin"""
     __license__     = "GPLv3"
-    __authors__     = [("pyLoad Team", "admin@pyload.org")]
-
+    __authors__     = [("Walter Purcaro", "vuolter@gmail.com")]
 
-    key = None  #: last key detected
 
+    def __init__(self, plugin):  #@TODO: Pass pyfile instead plugin, so store plugin's html in its associated pyfile as data
+        self._init(plugin.pyload)
 
-    def __init__(self, plugin):
         self.plugin = plugin
-        super(Captcha, self).__init__(plugin.core)
+        self.task   = None  #: captchaManager task
+
+        self.init()
+
+
+    def init(self):
+        """
+        Initialize additional data structures
+        """
+        pass
+
+
+    def _log(self, level, plugintype, pluginname, messages):
+        return self.plugin._log(level,
+                                plugintype,
+                                self.plugin.__name__,
+                                (self.__name__,) + messages)
+
+
+    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='jpg', output_type='textual', ocr=True, timeout=120):
+        img = self.load(url, get=get, post=post, ref=ref, cookies=cookies, decode=decode)
+        return self._decrypt(img, input_type, output_type, ocr, timeout)
+
+
+    #@TODO: Definitely choose a better name for this method!
+    def _decrypt(self, raw, input_type='jpg', output_type='textual', ocr=False, timeout=120):
+        """
+        Loads a captcha and decrypts it with ocr, plugin, user input
+
+        :param raw: image raw data
+        :param get: get part for request
+        :param post: post part for request
+        :param cookies: True if cookies should be enabled
+        :param input_type: Type of the Image
+        :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 ocr: if True, ocr is not used
+
+        :return: result of decrypting
+        """
+        result   = ""
+        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:
+            if isinstance(ocr, basestring):
+                OCR = self.pyload.pluginManager.loadClass("captcha", ocr)  #: Rename `captcha` to `ocr` in 0.4.10
+                result = OCR(self.plugin).recognize(tmp_img.name)
+            else:
+                result = self.recognize(tmp_img.name)
+
+        if not result:
+            captchaManager = self.pyload.captchaManager
+
+            try:
+                self.task = captchaManager.newTask(raw, input_type, tmp_img.name, output_type)
+
+                captchaManager.handleCaptcha(self.task)
+
+                self.task.setWaiting(max(timeout, 50))  #@TODO: Move to `CaptchaManager` in 0.4.10
+                while self.task.isWaiting():
+                    if self.plugin.pyfile.abort:
+                        self.plugin.abort()
+                    time.sleep(1)
+
+            finally:
+                captchaManager.removeTask(self.task)
+
+            if self.task.error:
+                self.fail(self.task.error)
 
+            elif not self.task.result:
+                self.invalid()
+                self.plugin.retry(reason=_("No captcha result obtained in appropiate time"))
 
-    #@TODO: Recheck in 0.4.10
-    def fail(self, reason):
-        self.plugin.fail(reason)
-        raise AttributeError(reason)
+            result = self.task.result
 
+        if not self.pyload.debug:
+            try:
+                os.remove(tmp_img.name)
 
-    #@TODO: Recheck in 0.4.10
-    def retrieve_key(self, html):
-        if self.detect_key(html):
-            return self.key
-        else:
-            self.fail(_("%s key not found") % self.__name__)
+            except OSError, e:
+                self.log_warning(_("Error removing: %s") % tmp_img.name, e)
+                traceback.print_exc()
 
+        self.log_info(_("Captcha result: ") + result)  #@TODO: Remove from here?
 
-    #@TODO: Recheck in 0.4.10
-    def retrieve_html(self):
-        if hasattr(self.plugin, "html") and self.plugin.html:
-            return self.plugin.html
-        else:
-            self.fail(_("%s html not found") % self.__name__)
+        return result
 
 
-    def detect_key(self, html=None):
-        raise NotImplementedError
+    def invalid(self):
+        if not self.task:
+            return
 
+        self.log_error(_("Invalid captcha"))
+        self.task.invalid()
 
-    def challenge(self, key=None, html=None):
-        raise NotImplementedError
 
+    def correct(self):
+        if not self.task:
+            return
 
-    def result(self, server, challenge):
-        raise NotImplementedError
+        self.log_info(_("Correct captcha"))
+        self.task.correct()