added CCITTFaxFilter support and a very crude image extraction.

pdfminer/ccitt.py

@@ -329,6 +329,8 @@ class CCITTG4Parser(BitParser):
             except self.ByteSkip:
                 self._accept = self._parse_mode
                 self._state = self.MODE
+            except self.EOFB:
+                break
         return
 
     def _parse_mode(self, mode):
@@ -394,7 +396,7 @@ class CCITTG4Parser(BitParser):
 
     def _get_bits(self):
         return ''.join( str(b) for b in self._curline[:self._curpos] )
-    
+
     def _get_refline(self, i):
         if i < 0:
             return '[]'+''.join( str(b) for b in self._refline )
@@ -667,6 +669,43 @@ class TestCCITTG4Parser(unittest.TestCase):
         return
 
 
+##  CCITTFaxDecoder
+##
+class CCITTFaxDecoder(CCITTG4Parser):
+    
+    def __init__(self, width, bytealign=False, reversed=False):
+        CCITTG4Parser.__init__(self, width, bytealign=bytealign)
+        self.reversed = reversed
+        self._buf = ''
+        return
+    
+    def close(self):
+        return self._buf
+    
+    def output_line(self, y, bits):
+        bytes = array.array('B', [0]*((len(bits)+7)/8))
+        if self.reversed:
+            bits = [ not b for b in bits ]
+        for (i,b) in enumerate(bits):
+            if b:
+                bytes[i/8] += (128,64,32,16,8,4,2,1)[i%8]
+        self._buf += bytes.tostring()
+        return
+
+
+def ccittfaxdecode(data, params):
+    K = params.get('K')
+    cols = params.get('Columns')
+    bytealign = params.get('EncodedByteAlign')
+    reversed = params.get('BlackIs1')
+    if K == -1:
+        parser = CCITTFaxDecoder(cols, bytealign=bytealign, reversed=reversed)
+    else:
+        raise ValueError(K)
+    parser.feedbytes(data)
+    return parser.close()
+    
+    
 # test
 def main(argv):
     import pygame
@@ -691,10 +730,7 @@ def main(argv):
         fp = file(path,'rb')
         (_,_,k,w,h,_) = path.split('.')
         parser = Parser(int(w))
-        try:
+        parser.feedbytes(fp.read())
-            parser.feedbytes(fp.read())
-        except parser.EOFB:
-            pass
         parser.close()
         fp.close()
     return

pdfminer/converter.py

@@ -1,14 +1,12 @@
 #!/usr/bin/env python2
-import sys, os.path
+import sys
 from pdfdevice import PDFDevice, PDFTextDevice
 from pdffont import PDFUnicodeNotDefined
-from pdftypes import LITERALS_DCT_DECODE
-from pdfcolor import LITERAL_DEVICE_GRAY, LITERAL_DEVICE_RGB
 from layout import LTContainer, LTPage, LTText, LTLine, LTRect, LTCurve
 from layout import LTFigure, LTImage, LTChar, LTTextLine
 from layout import LTTextBox, LTTextBoxVertical, LTTextGroup
 from utils import apply_matrix_pt, mult_matrix
-from utils import enc, bbox2str, create_bmp
+from utils import enc, bbox2str
 
 
 ##  PDFLayoutAnalyzer
@@ -139,28 +137,6 @@ class PDFConverter(PDFLayoutAnalyzer):
         self.outfp = outfp
         self.codec = codec
         return
-
-    def write_image(self, image):
-        stream = image.stream
-        filters = stream.get_filters()
-        if len(filters) == 1 and filters[0] in LITERALS_DCT_DECODE:
-            ext = '.jpg'
-            data = stream.get_rawdata()
-        elif image.colorspace is LITERAL_DEVICE_RGB:
-            ext = '.bmp'
-            data = create_bmp(stream.get_data(), stream.bits*3, image.width, image.height)
-        elif image.colorspace is LITERAL_DEVICE_GRAY:
-            ext = '.bmp'
-            data = create_bmp(stream.get_data(), stream.bits, image.width, image.height)
-        else:
-            ext = '.img'
-            data = stream.get_data()
-        name = image.name+ext
-        path = os.path.join(self.outdir, name)
-        fp = file(path, 'wb')
-        fp.write(data)
-        fp.close()
-        return name
     
 
 ##  TextConverter
@@ -222,7 +198,7 @@ class HTMLConverter(PDFConverter):
 
     def __init__(self, rsrcmgr, outfp, codec='utf-8', pageno=1, laparams=None, 
                  scale=1, fontscale=0.7, layoutmode='normal', showpageno=True,
-                 pagemargin=50, outdir=None,
+                 pagemargin=50, imagewriter=None,
                  rect_colors={'curve':'black', 'page':'gray'},
                  text_colors={'char':'black'}):
         PDFConverter.__init__(self, rsrcmgr, outfp, codec=codec, pageno=pageno, laparams=laparams)
@@ -231,7 +207,7 @@ class HTMLConverter(PDFConverter):
         self.layoutmode = layoutmode
         self.showpageno = showpageno
         self.pagemargin = pagemargin
-        self.outdir = outdir
+        self.imagewriter = imagewriter
         self.rect_colors = rect_colors
         self.text_colors = text_colors
         if self.debug:
@@ -278,8 +254,8 @@ class HTMLConverter(PDFConverter):
         return
 
     def place_image(self, item, borderwidth, x, y, w, h):
-        if self.outdir is not None:
+        if self.imagewriter is not None:
-            name = self.write_image(item)
+            name = self.imagewriter.export_image(item)
             self.write('<img src="%s" border="%d" style="position:absolute; left:%dpx; top:%dpx;" '
                        'width="%d" height="%d" />\n' %
                        (enc(name), borderwidth,
@@ -400,9 +376,10 @@ class HTMLConverter(PDFConverter):
 ##
 class XMLConverter(PDFConverter):
 
-    def __init__(self, rsrcmgr, outfp, codec='utf-8', pageno=1, laparams=None, outdir=None):
+    def __init__(self, rsrcmgr, outfp, codec='utf-8', pageno=1,
+                 laparams=None, imagewriter=None):
         PDFConverter.__init__(self, rsrcmgr, outfp, codec=codec, pageno=pageno, laparams=laparams)
-        self.outdir = outdir
+        self.imagewriter = imagewriter
         self.write_header()
         return
 
@@ -479,8 +456,8 @@ class XMLConverter(PDFConverter):
             elif isinstance(item, LTText):
                 self.outfp.write('<text>%s</text>\n' % item.get_text())
             elif isinstance(item, LTImage):
-                if self.outdir:
+                if self.imagewriter is not None:
-                    name = self.write_image(item)
+                    name = self.imagewriter.export_image(item)
                     self.outfp.write('<image src="%s" width="%d" height="%d" />\n' %
                                      (enc(name), item.width, item.height))
                 else:

pdfminer/pdftypes.py

@@ -4,6 +4,7 @@ import zlib
 from lzw import lzwdecode
 from ascii85 import ascii85decode, asciihexdecode
 from runlength import rldecode
+from ccitt import ccittfaxdecode
 from psparser import PSException, PSObject
 from psparser import LIT, KWD, STRICT
 from utils import apply_png_predictor
@@ -206,6 +207,7 @@ class PDFStream(PDFObject):
             self.rawdata = None
             return
         for f in filters:
+            params = self.get_any(('DP', 'DecodeParms', 'FDecodeParms'), {})
             if f in LITERALS_FLATE_DECODE:
                 # will get errors if the document is encrypted.
                 try:
@@ -223,15 +225,13 @@ class PDFStream(PDFObject):
             elif f in LITERALS_RUNLENGTH_DECODE:
                 data = rldecode(data)
             elif f in LITERALS_CCITTFAX_DECODE:
-                #data = ccittfaxdecode(data)
+                data = ccittfaxdecode(data, params)
-                raise PDFNotImplementedError('Unsupported filter: %r' % f)
             elif f == LITERAL_CRYPT:
                 # not yet..
                 raise PDFNotImplementedError('/Crypt filter is unsupported')
             else:
                 raise PDFNotImplementedError('Unsupported filter: %r' % f)
             # apply predictors
-            params = self.get_any(('DP', 'DecodeParms', 'FDecodeParms'), {})
             if 'Predictor' in params:
                 pred = int_value(params['Predictor'])
                 if pred == 1:

pdfminer/utils.py

@@ -306,13 +306,3 @@ class Plane(object):
                     obj.y1 <= y0 or y1 <= obj.y0): continue
                 yield obj
         return
-
-
-# create_bmp
-def create_bmp(data, bits, width, height):
-    info = struct.pack('<IiiHHIIIIII', 40, width, height, 1, bits, 0, len(data), 0, 0, 0, 0)
-    assert len(info) == 40, len(info)
-    header = struct.pack('<ccIHHI', 'B', 'M', 14+40+len(data), 0, 0, 14+40)
-    assert len(header) == 14, len(header)
-    # XXX re-rasterize every line
-    return header+info+data

tools/pdf2txt.py

@@ -6,6 +6,7 @@ from pdfminer.pdfdevice import PDFDevice, TagExtractor
 from pdfminer.converter import XMLConverter, HTMLConverter, TextConverter
 from pdfminer.cmapdb import CMapDB
 from pdfminer.layout import LAParams
+from pdfminer.image import ImageWriter
 
 # main
 def main(argv):
@@ -29,7 +30,7 @@ def main(argv):
     # output option
     outfile = None
     outtype = None
-    outdir = None
+    imagewriter = None
     layoutmode = 'normal'
     codec = 'utf-8'
     pageno = 1
@@ -52,7 +53,7 @@ def main(argv):
         elif k == '-W': laparams.word_margin = float(v)
         elif k == '-F': laparams.boxes_flow = float(v)
         elif k == '-Y': layoutmode = v
-        elif k == '-O': outdir = v
+        elif k == '-O': imagewriter = ImageWriter(v)
         elif k == '-t': outtype = v
         elif k == '-c': codec = v
         elif k == '-s': scale = float(v)
@@ -81,10 +82,12 @@ def main(argv):
     if outtype == 'text':
         device = TextConverter(rsrcmgr, outfp, codec=codec, laparams=laparams)
     elif outtype == 'xml':
-        device = XMLConverter(rsrcmgr, outfp, codec=codec, laparams=laparams, outdir=outdir)
+        device = XMLConverter(rsrcmgr, outfp, codec=codec, laparams=laparams,
+                              imagewriter=imagewriter)
     elif outtype == 'html':
         device = HTMLConverter(rsrcmgr, outfp, codec=codec, scale=scale,
-                               layoutmode=layoutmode, laparams=laparams, outdir=outdir)
+                               layoutmode=layoutmode, laparams=laparams,
+                               imagewriter=imagewriter)
     elif outtype == 'tag':
         device = TagExtractor(rsrcmgr, outfp, codec=codec)
     else: