Merge branch 'master' of github.com:euske/pdfminer

pdfminer/image.py

@@ -1,9 +1,13 @@
 #!/usr/bin/env python2
+import cStringIO
+import logging
 import sys
 import struct
 import os, os.path
+from PIL import Image
+from PIL import ImageChops
 from pdftypes import LITERALS_DCT_DECODE
-from pdfcolor import LITERAL_DEVICE_GRAY, LITERAL_DEVICE_RGB
+from pdfcolor import LITERAL_DEVICE_GRAY, LITERAL_DEVICE_RGB, LITERAL_DEVICE_CMYK
 
 def align32(x):
     return ((x+3)/4)*4
@@ -77,7 +81,15 @@ class ImageWriter(object):
         path = os.path.join(self.outdir, name)
         fp = file(path, 'wb')
         if ext == '.jpg':
-            fp.write(stream.get_rawdata())
+            raw_data = stream.get_rawdata()
+            if LITERAL_DEVICE_CMYK in image.colorspace:
+                ifp = cStringIO.StringIO(raw_data)
+                i = Image.open(ifp) 
+                i = ImageChops.invert(i)
+                i = i.convert('RGB')
+                i.save(fp, 'JPEG')
+            else:
+                fp.write(raw_data)
         elif image.bits == 1:
             bmp = BMPWriter(fp, 1, width, height)
             data = stream.get_data()

pdfminer/layout.py

@@ -609,8 +609,8 @@ class LTLayoutContainer(LTContainer):
                 group = LTTextGroupLRTB([obj1,obj2])
             plane.remove(obj1)
             plane.remove(obj2)
-            dists = [ (c,d,o1,o2) for (c,d,o1,o2) in dists
-                      if o1 in plane and o2 in plane ]
+            # this line is optimized -- don't change without profiling
+            dists = [ n for n in dists if n[2] in plane._objs and n[3] in plane._objs ]
             for other in plane:
                 dists.append((0, dist(group,other), group, other))
             dists.sort()

pdfminer/pdfparser.py

@@ -553,10 +553,10 @@ class PDFDocument(object):
             if 'Limits' in d:
                 (k1,k2) = list_value(d['Limits'])
                 if key < k1 or k2 < key: return None
-                if 'Names' in d:
-                    objs = list_value(d['Names'])
-                    names = dict(choplist(2, objs))
-                    return names[key]
+            if 'Names' in d:
+                objs = list_value(d['Names'])
+                names = dict(choplist(2, objs))
+                return names[key]
             if 'Kids' in d:
                 for c in list_value(d['Kids']):
                     v = lookup(dict_value(c))