linksave captcha 2nd try

14 files changed, 113 insertions, 1 deletions

diff --git a/module/captcha/LinksaveIn.py b/module/captcha/LinksaveIn.py
index cd4e97f87..4219f03b5 100644
--- a/module/captcha/LinksaveIn.py
+++ b/module/captcha/LinksaveIn.py
@@ -1,9 +1,17 @@
 from captcha import OCR
 import Image
+from os import sep
+from os.path import dirname
+from os.path import abspath
+from glob import glob
+import tempfile
+
+from pprint import pprint
 
 class LinksaveIn(OCR):
     def __init__(self):
         OCR.__init__(self)
+        self.data_dir = dirname(abspath(__file__)) + sep + "LinksaveIn" + sep
     
     def load_image(self, image):
         im = Image.open(image)
@@ -27,12 +35,115 @@ class LinksaveIn(OCR):
                     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 run_tesser(self):
+        self.logger.debug("create tmp tif")
+        tmp = tempfile.NamedTemporaryFile(suffix=".tif")
+        self.logger.debug("create tmp txt")
+        tmpTxt = tempfile.NamedTemporaryFile(suffix=".txt")
+        self.logger.debug("save tiff")
+        self.image.save(tmp.name, 'TIFF')
+        self.logger.debug("run tesseract")
+        self.run(['tesseract', tmp.name, tmpTxt.name.replace(".txt", ""), "nobatch", self.data_dir+"tesser_conf"])
+        self.logger.debug("read txt")
+
+        with open(tmpTxt.name, 'r') as f:
+            self.result_captcha = f.read().replace("\n", "")
 
     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(6)
+        self.image.save(self.data_dir+"cleaned_pass2.png")
+        letters = self.split_captcha_letters()
+        
         self.run_tesser()
 
         return self.result_captcha
@@ -42,5 +153,5 @@ if __name__ == '__main__':
     ocr = LinksaveIn()
     testurl = "http://linksave.in/captcha/cap.php?hsh=2229185&code=ZzHdhl3UffV3lXTH5U4b7nShXj%2Bwma1vyoNBcbc6lcc%3D"
     urllib.urlretrieve(testurl, "captcha.gif")
-
+    
     print ocr.get_captcha('captcha.gif')
diff --git a/module/captcha/LinksaveIn/bg/flecken_1.gif b/module/captcha/LinksaveIn/bg/flecken_1.gif
new file mode 100644
index 000000000..df2f51217
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/flecken_1.gif
diff --git a/module/captcha/LinksaveIn/bg/flecken_2.gif b/module/captcha/LinksaveIn/bg/flecken_2.gif
new file mode 100644
index 000000000..838276188
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/flecken_2.gif
diff --git a/module/captcha/LinksaveIn/bg/gewebe_fein.gif b/module/captcha/LinksaveIn/bg/gewebe_fein.gif
new file mode 100644
index 000000000..502f18cc4
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/gewebe_fein.gif
diff --git a/module/captcha/LinksaveIn/bg/gewebe_grob.gif b/module/captcha/LinksaveIn/bg/gewebe_grob.gif
new file mode 100644
index 000000000..e66a365ad
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/gewebe_grob.gif
diff --git a/module/captcha/LinksaveIn/bg/gitter.gif b/module/captcha/LinksaveIn/bg/gitter.gif
new file mode 100644
index 000000000..ec52ef68d
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/gitter.gif
diff --git a/module/captcha/LinksaveIn/bg/mauer_horizontal.gif b/module/captcha/LinksaveIn/bg/mauer_horizontal.gif
new file mode 100644
index 000000000..3d75fafa8
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/mauer_horizontal.gif
diff --git a/module/captcha/LinksaveIn/bg/mauer_vertikal.gif b/module/captcha/LinksaveIn/bg/mauer_vertikal.gif
new file mode 100644
index 000000000..2ada6fdae
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/mauer_vertikal.gif
diff --git a/module/captcha/LinksaveIn/bg/scheckig.gif b/module/captcha/LinksaveIn/bg/scheckig.gif
new file mode 100644
index 000000000..8bfb45c56
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/scheckig.gif
diff --git a/module/captcha/LinksaveIn/bg/wellen.gif b/module/captcha/LinksaveIn/bg/wellen.gif
new file mode 100644
index 000000000..a181ebe74
--- /dev/null
+++ b/module/captcha/LinksaveIn/bg/wellen.gif
diff --git a/module/captcha/LinksaveIn/cleaned_pass1.png b/module/captcha/LinksaveIn/cleaned_pass1.png
new file mode 100644
index 000000000..f0cc0ae47
--- /dev/null
+++ b/module/captcha/LinksaveIn/cleaned_pass1.png
diff --git a/module/captcha/LinksaveIn/cleaned_pass2.png b/module/captcha/LinksaveIn/cleaned_pass2.png
new file mode 100644
index 000000000..fb2ef44cd
--- /dev/null
+++ b/module/captcha/LinksaveIn/cleaned_pass2.png
diff --git a/module/captcha/LinksaveIn/tesser_conf b/module/captcha/LinksaveIn/tesser_conf
new file mode 100644
index 000000000..34ca8fa02
--- /dev/null
+++ b/module/captcha/LinksaveIn/tesser_conf
@@ -0,0 +1 @@
+tessedit_char_whitelist 0123456789
diff --git a/module/captcha/LinksaveIn/unblacked.png b/module/captcha/LinksaveIn/unblacked.png
new file mode 100644
index 000000000..4fa056d3d
--- /dev/null
+++ b/module/captcha/LinksaveIn/unblacked.png