From ce1c2b6b05c08b669357947e61ae40efce7fc50f Mon Sep 17 00:00:00 2001
From: Walter Purcaro <vuolter@gmail.com>
Date: Mon, 16 Feb 2015 10:46:28 +0100
Subject: module temp
---
module/plugins/ocr/GigasizeCom.py | 24 ++++++
module/plugins/ocr/LinksaveIn.py | 158 +++++++++++++++++++++++++++++++++++
module/plugins/ocr/NetloadIn.py | 29 +++++++
module/plugins/ocr/ShareonlineBiz.py | 39 +++++++++
module/plugins/ocr/__init__.py | 1 +
5 files changed, 251 insertions(+)
create mode 100644 module/plugins/ocr/GigasizeCom.py
create mode 100644 module/plugins/ocr/LinksaveIn.py
create mode 100644 module/plugins/ocr/NetloadIn.py
create mode 100644 module/plugins/ocr/ShareonlineBiz.py
create mode 100644 module/plugins/ocr/__init__.py
(limited to 'module/plugins/ocr')
diff --git a/module/plugins/ocr/GigasizeCom.py b/module/plugins/ocr/GigasizeCom.py
new file mode 100644
index 000000000..6982e6ca9
--- /dev/null
+++ b/module/plugins/ocr/GigasizeCom.py
@@ -0,0 +1,24 @@
+# -*- coding: utf-8 -*-
+
+from pyload.plugin.OCR import OCR
+
+
+class GigasizeCom(OCR):
+ __name__ = "GigasizeCom"
+ __type__ = "ocr"
+ __version__ = "0.10"
+
+ __description__ = """Gigasize.com ocr plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("pyLoad Team", "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
new file mode 100644
index 000000000..7ced74f4b
--- /dev/null
+++ b/module/plugins/ocr/LinksaveIn.py
@@ -0,0 +1,158 @@
+# -*- coding: utf-8 -*-
+
+try:
+ from PIL import Image
+except ImportError:
+ import Image
+
+from glob import glob
+from os import sep
+from os.path import abspath, dirname
+
+from pyload.plugin.OCR import OCR
+
+
+class LinksaveIn(OCR):
+ __name__ = "LinksaveIn"
+ __type__ = "ocr"
+ __version__ = "0.10"
+
+ __description__ = """Linksave.in ocr plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("pyLoad Team", "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 Exception:
+ cstat[rgb_c] = 1
+ if rgb_bg == rgb_c:
+ stat[bgpath] += 1
+ max_p = 0
+ bg = ""
+ for bgpath, value in stat.iteritems():
+ 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
new file mode 100644
index 000000000..8939b5318
--- /dev/null
+++ b/module/plugins/ocr/NetloadIn.py
@@ -0,0 +1,29 @@
+# -*- coding: utf-8 -*-
+
+from pyload.plugin.OCR import OCR
+
+
+class NetloadIn(OCR):
+ __name__ = "NetloadIn"
+ __type__ = "ocr"
+ __version__ = "0.10"
+
+ __description__ = """Netload.in ocr plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("pyLoad Team", "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
new file mode 100644
index 000000000..bbc3d1762
--- /dev/null
+++ b/module/plugins/ocr/ShareonlineBiz.py
@@ -0,0 +1,39 @@
+# -*- coding: utf-8 -*-
+
+from pyload.plugin.OCR import OCR
+
+
+class ShareonlineBiz(OCR):
+ __name__ = "ShareonlineBiz"
+ __type__ = "ocr"
+ __version__ = "0.10"
+
+ __description__ = """Shareonline.biz ocr plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("RaNaN", "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
new file mode 100644
index 000000000..40a96afc6
--- /dev/null
+++ b/module/plugins/ocr/__init__.py
@@ -0,0 +1 @@
+# -*- coding: utf-8 -*-
--
cgit v1.2.3