summaryrefslogtreecommitdiffstats
path: root/module/plugins/internal
diff options
context:
space:
mode:
Diffstat (limited to 'module/plugins/internal')
-rw-r--r--module/plugins/internal/AdYouLike.py12
-rw-r--r--module/plugins/internal/AdsCaptcha.py20
-rw-r--r--module/plugins/internal/Captcha.py102
-rw-r--r--module/plugins/internal/CaptchaService.py46
-rw-r--r--module/plugins/internal/Hoster.py92
-rw-r--r--module/plugins/internal/ReCaptcha.py32
-rw-r--r--module/plugins/internal/SimpleHoster.py6
-rw-r--r--module/plugins/internal/SolveMedia.py20
-rw-r--r--module/plugins/internal/XFSHoster.py2
9 files changed, 179 insertions, 153 deletions
diff --git a/module/plugins/internal/AdYouLike.py b/module/plugins/internal/AdYouLike.py
index fcb9b8372..d14babb51 100644
--- a/module/plugins/internal/AdYouLike.py
+++ b/module/plugins/internal/AdYouLike.py
@@ -3,10 +3,10 @@
import re
from module.common.json_layer import json_loads
-from module.plugins.internal.Captcha import Captcha
+from module.plugins.internal.CaptchaService import CaptchaService
-class AdYouLike(Captcha):
+class AdYouLike(CaptchaService):
__name__ = "AdYouLike"
__type__ = "captcha"
__version__ = "0.07"
@@ -21,8 +21,8 @@ class AdYouLike(Captcha):
CALLBACK_PATTERN = r'(Adyoulike\.g\._jsonp_\d+)'
- def detect_key(self, html=None):
- html = html or self.retrieve_html()
+ 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)
@@ -35,8 +35,8 @@ class AdYouLike(Captcha):
return None
- def challenge(self, key=None, html=None):
- ayl, callback = key or self.retrieve_key(html)
+ 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"}}
diff --git a/module/plugins/internal/AdsCaptcha.py b/module/plugins/internal/AdsCaptcha.py
index b99697a84..f487042e2 100644
--- a/module/plugins/internal/AdsCaptcha.py
+++ b/module/plugins/internal/AdsCaptcha.py
@@ -3,10 +3,10 @@
import random
import re
-from module.plugins.internal.Captcha import Captcha
+from module.plugins.internal.CaptchaService import CaptchaService
-class AdsCaptcha(Captcha):
+class AdsCaptcha(CaptchaService):
__name__ = "AdsCaptcha"
__type__ = "captcha"
__version__ = "0.10"
@@ -21,8 +21,8 @@ class AdsCaptcha(Captcha):
PUBLICKEY_PATTERN = r'api\.adscaptcha\.com/Get\.aspx\?.*?PublicKey=([\w-]+)'
- def detect_key(self, html=None):
- html = html or self.retrieve_html()
+ 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)
@@ -35,8 +35,8 @@ class AdsCaptcha(Captcha):
return None
- def challenge(self, key=None, html=None):
- PublicKey, CaptchaId = key or self.retrieve_key(html)
+ 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,
@@ -54,10 +54,10 @@ class AdsCaptcha(Captcha):
def result(self, server, challenge):
- result = self.plugin.decryptCaptcha("%sChallenge.aspx" % server,
- get={'cid': challenge, 'dummy': random.random()},
- cookies=True,
- imgtype="jpg")
+ result = self.decrypt_image("%sChallenge.aspx" % server,
+ get={'cid': challenge, 'dummy': random.random()},
+ cookies=True,
+ input_type="jpg")
self.log_debug("Result: %s" % result)
diff --git a/module/plugins/internal/Captcha.py b/module/plugins/internal/Captcha.py
index 814c36756..af7f66ed5 100644
--- a/module/plugins/internal/Captcha.py
+++ b/module/plugins/internal/Captcha.py
@@ -6,24 +6,28 @@ from module.plugins.internal.Plugin import Plugin
class Captcha(Plugin):
__name__ = "Captcha"
__type__ = "captcha"
- __version__ = "0.31"
+ __version__ = "0.01"
__status__ = "stable"
- __description__ = """Base captcha service plugin"""
+ __description__ = """Base anti-captcha plugin"""
__license__ = "GPLv3"
__authors__ = [("Walter Purcaro", "vuolter@gmail.com")]
- def __init__(self, plugin):
+ def __init__(self, plugin): #@TODO: pass pyfile instead plugin, so store plugin's html in its associated pyfile as data
self.pyload = plugin.core
self.info = {} #: Provide information in dict here
self.plugin = plugin
- self.key = None #: Last key detected
+ self.task = None #: captchaManager task
self.init()
+ def _log(self, type, args):
+ return super(Captcha, self)._log(type, (self.plugin.__name__,) + args)
+
+
def init(self):
"""
Initialize additional data structures
@@ -31,29 +35,87 @@ class Captcha(Plugin):
pass
- #@TODO: Recheck in 0.4.10
- def retrieve_key(self, html):
- if self.detect_key(html):
- return self.key
+ 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 decrypt(self, data, input_type='png', output_type='textual', try_ocr=True):
+ """
+ Loads a captcha and decrypts it with ocr, plugin, user input
+
+ :param url: url of captcha image
+ :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 try_ocr: if True, ocr is not used
+
+ :return: result of decrypting
+ """
+ id = ("%.2f" % time.time())[-6:].replace(".", "")
+
+ with open(os.path.join("tmp", "tmpCaptcha_%s_%s.%s" % (self.plugin.__name__, id, input_type)), "wb") as tmpCaptcha:
+ tmpCaptcha.write(img)
+
+ has_plugin = self.plugin.__name__ in self.pyload.pluginManager.ocrPlugins
+
+ if self.pyload.captcha:
+ Ocr = self.pyload.pluginManager.loadClass("ocr", self.plugin.__name__)
else:
- self.fail(_("%s key not found") % self.__name__)
+ Ocr = None
+ if Ocr and try_ocr:
+ time.sleep(random.randint(3000, 5000) / 1000.0)
+ if self.pyfile.abort:
+ self.abort()
- #@TODO: Recheck in 0.4.10
- def retrieve_html(self):
- if hasattr(self.plugin, "html") and self.plugin.html:
- return self.plugin.html
+ ocr = Ocr(self.pyfile)
+ result = ocr.get_captcha(tmpCaptcha.name)
else:
- self.fail(_("%s html not found") % self.__name__)
+ captchaManager = self.pyload.captchaManager
+ task = captchaManager.newTask(img, input_type, tmpCaptcha.name, output_type)
+ self.task = task
+ captchaManager.handleCaptcha(task)
+
+ while task.isWaiting():
+ if self.pyfile.abort:
+ captchaManager.removeTask(task)
+ self.abort()
+ time.sleep(1)
+
+ captchaManager.removeTask(task)
+
+ 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"))
+
+ result = task.result
+ self.log_debug("Received captcha result: %s" % result)
+ if not self.pyload.debug:
+ try:
+ os.remove(tmpCaptcha.name)
+ except Exception:
+ pass
- def detect_key(self, html=None):
- raise NotImplementedError
+ return result
- def challenge(self, key=None, html=None):
- raise NotImplementedError
+ def invalid(self):
+ self.log_error(_("Invalid captcha"))
+ if self.task:
+ self.task.invalid()
- def result(self, server, challenge):
- raise NotImplementedError
+ def correct(self):
+ self.log_info(_("Correct captcha"))
+ if self.task:
+ self.task.correct()
diff --git a/module/plugins/internal/CaptchaService.py b/module/plugins/internal/CaptchaService.py
new file mode 100644
index 000000000..05af8ccec
--- /dev/null
+++ b/module/plugins/internal/CaptchaService.py
@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+
+from module.plugins.internal.Plugin import Plugin
+
+
+class Captcha(Plugin):
+ __name__ = "Captcha"
+ __type__ = "captcha"
+ __version__ = "0.31"
+ __status__ = "stable"
+
+ __description__ = """Base anti-captcha service plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("Walter Purcaro", "vuolter@gmail.com")]
+
+
+ def init(self):
+ self.key = None #: Last key detected
+
+
+ #@TODO: Recheck in 0.4.10
+ def retrieve_key(self, data):
+ if self.detect_key(data):
+ return self.key
+ else:
+ self.fail(_("%s key not found") % self.__name__)
+
+
+ #@TODO: Recheck in 0.4.10, html is now pyfile.data
+ def retrieve_data(self):
+ if hasattr(self.plugin, "html") and self.plugin.html:
+ return self.plugin.html
+ else:
+ self.fail(_("%s data not found") % self.__name__)
+
+
+ def detect_key(self, data=None):
+ raise NotImplementedError
+
+
+ def challenge(self, key=None, data=None):
+ raise NotImplementedError
+
+
+ def result(self, server, challenge):
+ raise NotImplementedError
diff --git a/module/plugins/internal/Hoster.py b/module/plugins/internal/Hoster.py
index 35e6ef23e..af3e80acf 100644
--- a/module/plugins/internal/Hoster.py
+++ b/module/plugins/internal/Hoster.py
@@ -12,6 +12,7 @@ if os.name != "nt":
import grp
import pwd
+from module.plugins.internal import Captcha
from module.plugins.internal.Plugin import (Plugin, Abort, Fail, Reconnect, Retry, Skip
chunks, fixurl as _fixurl, replace_patterns, seconds_to_midnight,
set_cookies, parse_html_form, parse_html_tag_attr_value,
@@ -122,8 +123,8 @@ class Hoster(Plugin):
#: Js engine, see `JsEngine`
self.js = self.pyload.js
- #: Captcha task
- self.c_task = None
+ #: Captcha stuff
+ self.captcha = Captcha(self)
#: Some plugins store html code here
self.html = None
@@ -257,7 +258,7 @@ class Hoster(Plugin):
time.sleep(1)
else:
while pyfile.waitUntil > time.time():
- self.thread.m.reconnecting.wait(2)
+ self.thread.m.reconnecting.wait(1)
if pyfile.abort:
self.abort()
@@ -329,89 +330,6 @@ class Hoster(Plugin):
raise Retry(reason)
- def invalid_captcha(self):
- self.log_error(_("Invalid captcha"))
- if self.c_task:
- self.c_task.invalid()
-
-
- def correct_captcha(self):
- self.log_info(_("Correct captcha"))
- if self.c_task:
- self.c_task.correct()
-
-
- def decrypt_captcha(self, url, get={}, post={}, cookies=False, forceUser=False,
- imgtype='jpg', result_type='textual'):
- """
- Loads a captcha and decrypts it with ocr, plugin, user input
-
- :param url: url of captcha image
- :param get: get part for request
- :param post: post part for request
- :param cookies: True if cookies should be enabled
- :param forceUser: if True, ocr is not used
- :param imgtype: Type of the Image
- :param result_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
-
- :return: result of decrypting
- """
- img = self.load(url, get=get, post=post, cookies=cookies)
-
- id = ("%.2f" % time.time())[-6:].replace(".", "")
-
- with open(os.path.join("tmp", "tmpCaptcha_%s_%s.%s" % (self.__name__, id, imgtype)), "wb") as tmpCaptcha:
- tmpCaptcha.write(img)
-
- has_plugin = self.__name__ in self.pyload.pluginManager.ocrPlugins
-
- if self.pyload.captcha:
- Ocr = self.pyload.pluginManager.loadClass("ocr", self.__name__)
- else:
- Ocr = None
-
- if Ocr and not forceUser:
- time.sleep(random.randint(3000, 5000) / 1000.0)
- if self.pyfile.abort:
- self.abort()
-
- ocr = Ocr(self.pyfile)
- result = ocr.get_captcha(tmpCaptcha.name)
- else:
- captchaManager = self.pyload.captchaManager
- task = captchaManager.newTask(img, imgtype, tmpCaptcha.name, result_type)
- self.c_task = task
- captchaManager.handleCaptcha(task)
-
- while task.isWaiting():
- if self.pyfile.abort:
- captchaManager.removeTask(task)
- self.abort()
- time.sleep(1)
-
- captchaManager.removeTask(task)
-
- if task.error and has_plugin: #: Ignore default error message since the user could use 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"))
-
- result = task.result
- self.log_debug("Received captcha result: %s" % result)
-
- if not self.pyload.debug:
- try:
- os.remove(tmpCaptcha.name)
- except Exception:
- pass
-
- return result
-
-
def fixurl(self, url):
url = _fixurl(url)
@@ -447,7 +365,7 @@ class Hoster(Plugin):
if self.pyload.debug:
self.log_debug("DOWNLOAD URL " + url, *["%s=%s" % (key, val) for key, val in locals().iteritems() if key not in ("self", "url")])
- self.correct_captcha()
+ self.captcha.correct()
self.check_for_same_files()
self.pyfile.setStatus("downloading")
diff --git a/module/plugins/internal/ReCaptcha.py b/module/plugins/internal/ReCaptcha.py
index 4d04c07db..b4f9ef1eb 100644
--- a/module/plugins/internal/ReCaptcha.py
+++ b/module/plugins/internal/ReCaptcha.py
@@ -7,10 +7,10 @@ import urlparse
from base64 import b64encode
-from module.plugins.internal.Captcha import Captcha
+from module.plugins.internal.CaptchaService import CaptchaService
-class ReCaptcha(Captcha):
+class ReCaptcha(CaptchaService):
__name__ = "ReCaptcha"
__type__ = "captcha"
__version__ = "0.18"
@@ -27,8 +27,8 @@ class ReCaptcha(Captcha):
KEY_V2_PATTERN = r'(?:data-sitekey=["\']|["\']sitekey["\']:\s*["\'])([\w-]+)'
- def detect_key(self, html=None):
- html = html or self.retrieve_html()
+ 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:
@@ -40,15 +40,15 @@ class ReCaptcha(Captcha):
return None
- def challenge(self, key=None, html=None, version=None):
- key = key or self.retrieve_key(html)
+ 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_html()) else 1)
+ version=2 if re.search(self.KEY_V2_PATTERN, html or self.retrieve_data()) else 1)
def _challenge_v1(self, key):
@@ -81,11 +81,11 @@ class ReCaptcha(Captcha):
self.fail(_("ReCaptcha second challenge pattern not found"))
self.log_debug("Second challenge: %s" % challenge)
- result = self.plugin.decryptCaptcha("%simage" % server,
- get={'c': challenge},
- cookies=True,
- forceUser=True,
- imgtype="jpg")
+ result = self.decrypt("%simage" % server,
+ get={'c': challenge},
+ cookies=True,
+ input_type="jpg",
+ try_ocr=False)
self.log_debug("Result: %s" % result)
@@ -170,10 +170,10 @@ class ReCaptcha(Captcha):
self.log_debug("Token #3: %s" % token3.group(1))
millis_captcha_loading = int(round(time.time() * 1000))
- captcha_response = self.plugin.decryptCaptcha("https://www.google.com/recaptcha/api2/payload",
- get={'c':token3.group(1), 'k':key},
- cookies=True,
- forceUser=True)
+ 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)
diff --git a/module/plugins/internal/SimpleHoster.py b/module/plugins/internal/SimpleHoster.py
index 9d1ebc0bf..5960794cc 100644
--- a/module/plugins/internal/SimpleHoster.py
+++ b/module/plugins/internal/SimpleHoster.py
@@ -301,8 +301,8 @@ class SimpleHoster(Hoster):
def check_file(self):
lastDownload = fs_encode(self.last_download)
- if self.c_task and not self.last_download:
- self.invalid_captcha()
+ if self.captcha.task and not self.last_download:
+ self.captcha.invalid()
self.retry(10, reason=_("Wrong captcha"))
elif self.check_download({'Empty file': re.compile(r'\A((.|)(\2|\s)*)\Z')},
@@ -396,7 +396,7 @@ class SimpleHoster(Hoster):
self.fail(_("Connection from your current IP address is not allowed"))
elif re.search('captcha|code', errmsg, re.I):
- self.invalid_captcha()
+ self.captcha.invalid()
self.retry(10, reason=_("Wrong captcha"))
elif re.search('countdown|expired', errmsg, re.I):
diff --git a/module/plugins/internal/SolveMedia.py b/module/plugins/internal/SolveMedia.py
index 5d701a5a2..ce4ebb007 100644
--- a/module/plugins/internal/SolveMedia.py
+++ b/module/plugins/internal/SolveMedia.py
@@ -3,10 +3,10 @@
import re
from module.plugins.internal.Plugin import Fail
-from module.plugins.internal.Captcha import Captcha
+from module.plugins.internal.CaptchaService import CaptchaService
-class SolveMedia(Captcha):
+class SolveMedia(CaptchaService):
__name__ = "SolveMedia"
__type__ = "captcha"
__version__ = "0.15"
@@ -20,8 +20,8 @@ class SolveMedia(Captcha):
KEY_PATTERN = r'api\.solvemedia\.com/papi/challenge\.(?:no)?script\?k=(.+?)["\']'
- def detect_key(self, html=None):
- html = html or self.retrieve_html()
+ def detect_key(self, data=None):
+ html = data or self.retrieve_data()
m = re.search(self.KEY_PATTERN, html)
if m:
@@ -33,8 +33,8 @@ class SolveMedia(Captcha):
return None
- def challenge(self, key=None, html=None):
- key = key or self.retrieve_key(html)
+ 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})
@@ -95,10 +95,10 @@ class SolveMedia(Captcha):
def result(self, server, challenge):
- result = self.plugin.decryptCaptcha(server,
- get={'c': challenge},
- cookies=True,
- imgtype="gif")
+ result = self.decrypt_image(server,
+ get={'c': challenge},
+ cookies=True,
+ input_type="gif")
self.log_debug("Result: %s" % result)
diff --git a/module/plugins/internal/XFSHoster.py b/module/plugins/internal/XFSHoster.py
index 09799aa28..ec9a18a48 100644
--- a/module/plugins/internal/XFSHoster.py
+++ b/module/plugins/internal/XFSHoster.py
@@ -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.decrypt_captcha(captcha_url)
+ inputs['code'] = self.captcha.decrypt_image(captcha_url)
return
m = re.search(self.CAPTCHA_BLOCK_PATTERN, self.html, re.S)
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-09 18:27:06 +0200