summaryrefslogtreecommitdiffstats
path: root/module/plugins/ocr
diff options
context:
space:
mode:
authorGravatar Walter Purcaro <vuolter@gmail.com> 2014-09-08 00:29:57 +0200
committerGravatar Walter Purcaro <vuolter@gmail.com> 2014-09-14 11:02:23 +0200
commit68d662e689cd42687341c550fb6ebb74e6968d21 (patch)
tree486cef41bd928b8db704894233b2cef94a6e346f /module/plugins/ocr
parentsave_join -> safe_join & save_path -> safe_filename (diff)
downloadpyload-68d662e689cd42687341c550fb6ebb74e6968d21.tar.xz
module -> pyload
Diffstat (limited to 'module/plugins/ocr')
-rw-r--r--module/plugins/ocr/GigasizeCom.py23
-rw-r--r--module/plugins/ocr/LinksaveIn.py148
-rw-r--r--module/plugins/ocr/NetloadIn.py27
-rw-r--r--module/plugins/ocr/ShareonlineBiz.py38
-rw-r--r--module/plugins/ocr/__init__.py0
5 files changed, 0 insertions, 236 deletions
diff --git a/module/plugins/ocr/GigasizeCom.py b/module/plugins/ocr/GigasizeCom.py
deleted file mode 100644
index ba0b805e6..000000000
--- a/module/plugins/ocr/GigasizeCom.py
+++ /dev/null
@@ -1,23 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from module.plugins.OCR import OCR
-
-
-class GigasizeCom(OCR):
- __name__ = "GigasizeCom"
- __type__ = "ocr"
- __version__ = "0.1"
-
- __description__ = """Gigasize.com ocr plugin"""
- __author_name__ = "pyLoad Team"
- __author_mail__ = "admin@pyload.org"
-
-
- 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
diff --git a/module/plugins/ocr/LinksaveIn.py b/module/plugins/ocr/LinksaveIn.py
deleted file mode 100644
index 3ae139a4e..000000000
--- a/module/plugins/ocr/LinksaveIn.py
+++ /dev/null
@@ -1,148 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from module.plugins.OCR import OCR
-from PIL import Image
-from os import sep
-from os.path import abspath, dirname
-from glob import glob
-
-
-class LinksaveIn(OCR):
- __name__ = "LinksaveIn"
- __type__ = "ocr"
- __version__ = "0.1"
-
- __description__ = """Linksave.in ocr plugin"""
- __author_name__ = "pyLoad Team"
- __author_mail__ = "admin@pyload.org"
-
-
- 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 xrange(frame.size[0]):
- for y in xrange(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 xrange(bg.size[0]):
- for y in xrange(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 xrange(bg.size[0]):
- for y in xrange(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 xrange(new.size[0]):
- for y in xrange(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
diff --git a/module/plugins/ocr/NetloadIn.py b/module/plugins/ocr/NetloadIn.py
deleted file mode 100644
index 0de88302e..000000000
--- a/module/plugins/ocr/NetloadIn.py
+++ /dev/null
@@ -1,27 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from module.plugins.OCR import OCR
-
-class NetloadIn(OCR):
- __name__ = "NetloadIn"
- __type__ = "ocr"
- __version__ = "0.1"
-
- __description__ = """Netload.in ocr plugin"""
- __author_name__ = "pyLoad Team"
- __author_mail__ = "admin@pyload.org"
-
-
- def __init__(self):
- OCR.__init__(self)
-
- def get_captcha(self, image):
- self.load_image(image)
- self.to_greyscale()
- self.clean(3)
- self.clean(3)
- self.run_tesser(True, True, False, False)
-
- self.result_captcha = self.result_captcha.replace(" ", "")[:4] # cut to 4 numbers
-
- return self.result_captcha
diff --git a/module/plugins/ocr/ShareonlineBiz.py b/module/plugins/ocr/ShareonlineBiz.py
deleted file mode 100644
index 0ad018bf9..000000000
--- a/module/plugins/ocr/ShareonlineBiz.py
+++ /dev/null
@@ -1,38 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from module.plugins.OCR import OCR
-
-
-class ShareonlineBiz(OCR):
- __name__ = "ShareonlineBiz"
- __type__ = "ocr"
- __version__ = "0.1"
-
- __description__ = """Shareonline.biz ocr plugin"""
- __author_name__ = "RaNaN"
- __author_mail__ = "RaNaN@pyload.org"
-
-
- def __init__(self):
- OCR.__init__(self)
-
- def get_captcha(self, image):
- self.load_image(image)
- self.to_greyscale()
- self.image = self.image.resize((160, 50))
- self.pixels = self.image.load()
- self.threshold(1.85)
- #self.eval_black_white(240)
- #self.derotate_by_average()
-
- letters = self.split_captcha_letters()
-
- final = ""
- for letter in letters:
- self.image = letter
- self.run_tesser(True, True, False, False)
- final += self.result_captcha
-
- return final
-
- #tesseract at 60%
diff --git a/module/plugins/ocr/__init__.py b/module/plugins/ocr/__init__.py
deleted file mode 100644
index e69de29bb..000000000
--- a/module/plugins/ocr/__init__.py
+++ /dev/null