summaryrefslogtreecommitdiffstats
path: root/module/plugins
diff options
context:
space:
mode:
Diffstat (limited to 'module/plugins')
-rw-r--r--module/plugins/Base.py118
-rw-r--r--module/plugins/Hoster.py96
-rw-r--r--module/plugins/captcha/GigasizeCom.py19
-rw-r--r--module/plugins/captcha/LinksaveIn.py147
-rw-r--r--module/plugins/captcha/MegauploadCom.py14
-rw-r--r--module/plugins/captcha/__init__.py0
-rw-r--r--module/plugins/internal/NetloadInOCR.py (renamed from module/plugins/captcha/NetloadIn.py)11
-rw-r--r--module/plugins/internal/OCR.py (renamed from module/plugins/captcha/captcha.py)3
-rw-r--r--module/plugins/internal/ShareonlineBizOCR.py (renamed from module/plugins/captcha/ShareonlineBiz.py)8
9 files changed, 125 insertions, 291 deletions
diff --git a/module/plugins/Base.py b/module/plugins/Base.py
index b846bbd60..61fa211f4 100644
--- a/module/plugins/Base.py
+++ b/module/plugins/Base.py
@@ -18,8 +18,11 @@
"""
import sys
+from time import time, sleep
+from random import randint
+
from module.utils import decode
-from module.utils.fs import exists, makedirs, join
+from module.utils.fs import exists, makedirs, join, remove
# TODO
# more attributes if needed
@@ -32,6 +35,9 @@ class Fail(Exception):
class Retry(Exception):
""" raised when start again from beginning """
+class Abort(Exception):
+ """ raised when aborted """
+
class Base(object):
"""
The Base plugin class with all shared methods and every possible attribute for plugin definition.
@@ -83,6 +89,9 @@ class Base(object):
#: :class:`InteractionManager`
self.im = core.interactionManager
+ #: last interaction task
+ self.task = None
+
def logInfo(self, *args, **kwargs):
""" Print args to log at specific level
@@ -118,11 +127,7 @@ class Base(object):
getattr(self.log, level)("%s: %s" % (self.__name__, sep.join(strings)))
def setConfig(self, option, value):
- """ Set config value for current plugin
-
- :param option:
- :param value:
- """
+ """ Set config value for current plugin """
self.core.config.set(self.__name__, option, value)
def getConf(self, option):
@@ -130,11 +135,7 @@ class Base(object):
return self.core.config.get(self.__name__, option)
def getConfig(self, option):
- """ Returns config value for current plugin
-
- :param option:
- :return:
- """
+ """ Returns config value for current plugin """
return self.getConf(option)
def setStorage(self, key, value):
@@ -167,6 +168,14 @@ class Base(object):
sys.stdout = sys._stdout
embed()
+ def abort(self):
+ """ Check if plugin is in an abort state, is overwritten by subtypes"""
+ return False
+
+ def checkAbort(self):
+ """ Will be overwriten to determine if control flow should be aborted """
+ if self.abort: raise Abort()
+
def load(self, url, get={}, post={}, ref=True, cookies=True, just_header=False, decode=False):
"""Load content at url and returns it
@@ -180,6 +189,7 @@ class Base(object):
:return: Loaded content
"""
if not hasattr(self, "req"): raise Exception("Plugin type does not have Request attribute.")
+ self.checkAbort()
res = self.req.load(url, get, post, ref, cookies, just_header, decode=decode)
@@ -225,6 +235,92 @@ class Base(object):
return res
+ def invalidTask(self):
+ if self.task:
+ self.task.invalid()
+
+ def invalidCaptcha(self):
+ self.logDebug("Deprecated method .invalidCaptcha, use .invalidTask")
+ self.invalidTask()
+
+ def correctTask(self):
+ if self.task:
+ self.task.correct()
+
+ def correctCaptcha(self):
+ self.logDebug("Deprecated method .correctCaptcha, use .correctTask")
+ self.correctTask()
+
+ def decryptCaptcha(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())[-6:].replace(".", "")
+ temp_file = open(join("tmp", "tmpCaptcha_%s_%s.%s" % (self.__name__, id, imgtype)), "wb")
+ temp_file.write(img)
+ temp_file.close()
+
+ name = "%sOCR" % self.__name__
+ has_plugin = name in self.core.pluginManager.getPlugins("internal")
+
+ if self.core.captcha:
+ OCR = self.core.pluginManager.loadClass("internal", name)
+ else:
+ OCR = None
+
+ if OCR and not forceUser:
+ sleep(randint(3000, 5000) / 1000.0)
+ self.checkAbort()
+
+ ocr = OCR()
+ result = ocr.get_captcha(temp_file.name)
+ else:
+ task = self.im.newCaptchaTask(img, imgtype, temp_file.name, result_type)
+ self.task = task
+ self.im.handleTask(task)
+
+ while task.isWaiting():
+ if self.abort():
+ self.im.removeTask(task)
+ raise Abort()
+ sleep(1)
+
+ #TODO
+ self.im.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" % str(result))
+
+ if not self.core.debug:
+ try:
+ remove(temp_file.name)
+ except:
+ pass
+
+ return result
+
def fail(self, reason):
""" fail and give reason """
raise Fail(reason) \ No newline at end of file
diff --git a/module/plugins/Hoster.py b/module/plugins/Hoster.py
index 32c587aa5..b330743e6 100644
--- a/module/plugins/Hoster.py
+++ b/module/plugins/Hoster.py
@@ -17,10 +17,8 @@
@author: RaNaN, spoob, mkaay
"""
-from time import time, sleep
-from random import randint
-
import os
+from time import time
if os.name != "nt":
from module.utils.fs import chown
@@ -35,9 +33,6 @@ from module.utils.fs import save_join, save_filename, fs_encode, fs_decode,\
# Import for Hoster Plugins
chunks = _chunks
-class Abort(Exception):
- """ raised when aborted """
-
class Reconnect(Exception):
""" raised when reconnected """
@@ -170,6 +165,9 @@ class Hoster(Base):
"""the 'main' method of every plugin, you **have to** overwrite it"""
raise NotImplementedError
+ def abort(self):
+ return self.pyfile.abort
+
def resetAccount(self):
""" dont use account and retry download """
self.account = None
@@ -208,7 +206,7 @@ class Hoster(Base):
while self.pyfile.waitUntil > time():
self.thread.m.reconnecting.wait(2)
- if self.pyfile.abort: raise Abort
+ self.checkAbort()
if self.thread.m.reconnecting.isSet():
self.waiting = False
self.wantReconnect = False
@@ -243,88 +241,6 @@ class Hoster(Base):
self.retries += 1
raise Retry(reason)
- def invalidCaptcha(self):
- if self.cTask:
- self.cTask.invalid()
-
- def correctCaptcha(self):
- if self.cTask:
- self.cTask.correct()
-
- def decryptCaptcha(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())[-6:].replace(".", "")
- temp_file = open(join("tmp", "tmpCaptcha_%s_%s.%s" % (self.__name__, id, imgtype)), "wb")
- temp_file.write(img)
- temp_file.close()
-
- has_plugin = self.__name__ in self.core.pluginManager.getPlugins("captcha")
-
- if self.core.captcha:
- Ocr = self.core.pluginManager.loadClass("captcha", self.__name__)
- else:
- Ocr = None
-
- if Ocr and not forceUser:
- sleep(randint(3000, 5000) / 1000.0)
- if self.pyfile.abort: raise Abort
-
- ocr = Ocr()
- result = ocr.get_captcha(temp_file.name)
- else:
- captchaManager = self.core.captchaManager
- task = captchaManager.newTask(img, imgtype, temp_file.name, result_type)
- self.cTask = task
- captchaManager.handleCaptcha(task)
-
- while task.isWaiting():
- if self.pyfile.abort:
- captchaManager.removeTask(task)
- raise Abort
- 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" % str(result))
-
- if not self.core.debug:
- try:
- remove(temp_file.name)
- except:
- pass
-
- return result
-
-
- def load(self, *args, **kwargs):
- """ See 'Base' load method for more info """
- if self.pyfile.abort: raise Abort
- return Base.load(self, *args, **kwargs)
def download(self, url, get={}, post={}, ref=True, cookies=True, disposition=False):
"""Downloads the content at url to download folder
@@ -338,8 +254,8 @@ class Hoster(Base):
the filename will be changed if needed
:return: The location where the file was saved
"""
-
self.checkForSameFiles()
+ self.checkAbort()
self.pyfile.setStatus("downloading")
diff --git a/module/plugins/captcha/GigasizeCom.py b/module/plugins/captcha/GigasizeCom.py
deleted file mode 100644
index d31742eb5..000000000
--- a/module/plugins/captcha/GigasizeCom.py
+++ /dev/null
@@ -1,19 +0,0 @@
-# -*- coding: utf-8 -*-
-from captcha import OCR
-
-class GigasizeCom(OCR):
- def __init__(self):
- OCR.__init__(self)
-
- def get_captcha(self, image):
- self.load_image(image)
- self.threshold(2.8)
- self.run_tesser(True, False, False, True)
- return self.result_captcha
-
-if __name__ == '__main__':
- ocr = GigasizeCom()
- import urllib
- urllib.urlretrieve('http://www.gigasize.com/randomImage.php', "gigasize_tmp.jpg")
-
- print ocr.get_captcha('gigasize_tmp.jpg')
diff --git a/module/plugins/captcha/LinksaveIn.py b/module/plugins/captcha/LinksaveIn.py
deleted file mode 100644
index 3ad7b265a..000000000
--- a/module/plugins/captcha/LinksaveIn.py
+++ /dev/null
@@ -1,147 +0,0 @@
-from captcha import OCR
-import Image
-from os import sep
-from os.path import dirname
-from os.path import abspath
-from glob import glob
-
-
-class LinksaveIn(OCR):
- __name__ = "LinksaveIn"
- def __init__(self):
- OCR.__init__(self)
- self.data_dir = dirname(abspath(__file__)) + sep + "LinksaveIn" + sep
-
- def load_image(self, image):
- im = Image.open(image)
- frame_nr = 0
-
- lut = im.resize((256, 1))
- lut.putdata(range(256))
- lut = list(lut.convert("RGB").getdata())
-
- new = Image.new("RGB", im.size)
- npix = new.load()
- while True:
- try:
- im.seek(frame_nr)
- except EOFError:
- break
- frame = im.copy()
- pix = frame.load()
- for x in range(frame.size[0]):
- for y in range(frame.size[1]):
- if lut[pix[x, y]] != (0,0,0):
- npix[x, y] = lut[pix[x, y]]
- frame_nr += 1
- new.save(self.data_dir+"unblacked.png")
- self.image = new.copy()
- self.pixels = self.image.load()
- self.result_captcha = ''
-
- def get_bg(self):
- stat = {}
- cstat = {}
- img = self.image.convert("P")
- for bgpath in glob(self.data_dir+"bg/*.gif"):
- stat[bgpath] = 0
- bg = Image.open(bgpath)
-
- bglut = bg.resize((256, 1))
- bglut.putdata(range(256))
- bglut = list(bglut.convert("RGB").getdata())
-
- lut = img.resize((256, 1))
- lut.putdata(range(256))
- lut = list(lut.convert("RGB").getdata())
-
- bgpix = bg.load()
- pix = img.load()
- for x in range(bg.size[0]):
- for y in range(bg.size[1]):
- rgb_bg = bglut[bgpix[x, y]]
- rgb_c = lut[pix[x, y]]
- try:
- cstat[rgb_c] += 1
- except:
- cstat[rgb_c] = 1
- if rgb_bg == rgb_c:
- stat[bgpath] += 1
- max_p = 0
- bg = ""
- for bgpath, value in stat.items():
- if max_p < value:
- bg = bgpath
- max_p = value
- return bg
-
- def substract_bg(self, bgpath):
- bg = Image.open(bgpath)
- img = self.image.convert("P")
-
- bglut = bg.resize((256, 1))
- bglut.putdata(range(256))
- bglut = list(bglut.convert("RGB").getdata())
-
- lut = img.resize((256, 1))
- lut.putdata(range(256))
- lut = list(lut.convert("RGB").getdata())
-
- bgpix = bg.load()
- pix = img.load()
- orgpix = self.image.load()
- for x in range(bg.size[0]):
- for y in range(bg.size[1]):
- rgb_bg = bglut[bgpix[x, y]]
- rgb_c = lut[pix[x, y]]
- if rgb_c == rgb_bg:
- orgpix[x, y] = (255,255,255)
-
- def eval_black_white(self):
- new = Image.new("RGB", (140, 75))
- pix = new.load()
- orgpix = self.image.load()
- thresh = 4
- for x in range(new.size[0]):
- for y in range(new.size[1]):
- rgb = orgpix[x, y]
- r, g, b = rgb
- pix[x, y] = (255,255,255)
- if r > max(b, g)+thresh:
- pix[x, y] = (0,0,0)
- if g < min(r, b):
- pix[x, y] = (0,0,0)
- if g > max(r, b)+thresh:
- pix[x, y] = (0,0,0)
- if b > max(r, g)+thresh:
- pix[x, y] = (0,0,0)
- self.image = new
- self.pixels = self.image.load()
-
- def get_captcha(self, image):
- self.load_image(image)
- bg = self.get_bg()
- self.substract_bg(bg)
- self.eval_black_white()
- self.to_greyscale()
- self.image.save(self.data_dir+"cleaned_pass1.png")
- self.clean(4)
- self.clean(4)
- self.image.save(self.data_dir+"cleaned_pass2.png")
- letters = self.split_captcha_letters()
- final = ""
- for n, letter in enumerate(letters):
- self.image = letter
- self.image.save(ocr.data_dir+"letter%d.png" % n)
- self.run_tesser(True, True, False, False)
- final += self.result_captcha
-
- return final
-
-if __name__ == '__main__':
- import urllib
- ocr = LinksaveIn()
- testurl = "http://linksave.in/captcha/cap.php?hsh=2229185&code=ZzHdhl3UffV3lXTH5U4b7nShXj%2Bwma1vyoNBcbc6lcc%3D"
- urllib.urlretrieve(testurl, ocr.data_dir+"captcha.gif")
-
- print ocr.get_captcha(ocr.data_dir+'captcha.gif')
diff --git a/module/plugins/captcha/MegauploadCom.py b/module/plugins/captcha/MegauploadCom.py
deleted file mode 100644
index 469ee4094..000000000
--- a/module/plugins/captcha/MegauploadCom.py
+++ /dev/null
@@ -1,14 +0,0 @@
-from captcha import OCR
-
-class MegauploadCom(OCR):
- __name__ = "MegauploadCom"
- def __init__(self):
- OCR.__init__(self)
-
- def get_captcha(self, image):
- self.load_image(image)
- self.run_tesser(True, True, False, True)
- return self.result_captcha
-
-if __name__ == '__main__':
- ocr = MegauploadCom()
diff --git a/module/plugins/captcha/__init__.py b/module/plugins/captcha/__init__.py
deleted file mode 100644
index e69de29bb..000000000
--- a/module/plugins/captcha/__init__.py
+++ /dev/null
diff --git a/module/plugins/captcha/NetloadIn.py b/module/plugins/internal/NetloadInOCR.py
index 7f2e6a8d1..e50978701 100644
--- a/module/plugins/captcha/NetloadIn.py
+++ b/module/plugins/internal/NetloadInOCR.py
@@ -1,7 +1,10 @@
-from captcha import OCR
+# -*- coding: utf-8 -*-
+
+from OCR import OCR
+
+class NetloadInOCR(OCR):
+ __version__ = 0.1
-class NetloadIn(OCR):
- __name__ = "NetloadIn"
def __init__(self):
OCR.__init__(self)
@@ -18,7 +21,7 @@ class NetloadIn(OCR):
if __name__ == '__main__':
import urllib
- ocr = NetloadIn()
+ ocr = NetloadInOCR()
urllib.urlretrieve("http://netload.in/share/includes/captcha.php", "captcha.png")
print ocr.get_captcha('captcha.png')
diff --git a/module/plugins/captcha/captcha.py b/module/plugins/internal/OCR.py
index 4cbb736c1..9f8b7ef8c 100644
--- a/module/plugins/captcha/captcha.py
+++ b/module/plugins/internal/OCR.py
@@ -33,8 +33,7 @@ import JpegImagePlugin
class OCR(object):
-
- __name__ = "OCR"
+ __version__ = 0.1
def __init__(self):
self.logger = logging.getLogger("log")
diff --git a/module/plugins/captcha/ShareonlineBiz.py b/module/plugins/internal/ShareonlineBizOCR.py
index b07fb9b0f..c5c2e92e8 100644
--- a/module/plugins/captcha/ShareonlineBiz.py
+++ b/module/plugins/internal/ShareonlineBizOCR.py
@@ -17,10 +17,10 @@
# along with this program; if not, see <http://www.gnu.org/licenses/>.
#
###
-from captcha import OCR
+from OCR import OCR
-class ShareonlineBiz(OCR):
- __name__ = "ShareonlineBiz"
+class ShareonlineBizOCR(OCR):
+ __version__ = 0.1
def __init__(self):
OCR.__init__(self)
@@ -48,6 +48,6 @@ class ShareonlineBiz(OCR):
if __name__ == '__main__':
import urllib
- ocr = ShareonlineBiz()
+ ocr = ShareonlineBizOCR()
urllib.urlretrieve("http://www.share-online.biz/captcha.php", "captcha.jpeg")
print ocr.get_captcha('captcha.jpeg')
generated by cgit v1.2.3 (git 2.25.1) at 2025-01-31 02:46:00 +0100