Merge pull request #18 from DarwinAwardWinner/master

Avoid leaking file descriptors and convert unrecognized colorspaces to RGB

src/img2pdf.py

@@ -204,7 +204,9 @@ def convert(images, dpi, x, y, title=None, author=None, creator=None, producer=N
     pdf = pdfdoc(3, title, author, creator, producer, creationdate,
                  moddate, subject, keywords)
 
-    for im in images:
+    for imfilename in images:
+        debug_out("Reading %s"%imfilename, verbose)
+        with open(imfilename, "rb") as im:
             rawdata = im.read()
             im.seek(0)
             try:
@@ -263,8 +265,15 @@ def convert(images, dpi, x, y, title=None, author=None, creator=None, producer=N
         else:
             # because we do not support /CCITTFaxDecode
             if color == '1':
+                debug_out("Converting colorspace 1 to L", verbose)
                 imgdata = imgdata.convert('L')
                 color = 'L'
+            elif color in ("RGB", "L"):
+                debug_out("Colorspace is OK: %s"%color, verbose)
+            else:
+                debug_out("Converting colorspace %s to RGB"%color, verbose)
+                imgdata = imgdata.convert('RGB')
+                color = imgdata.mode
             imgdata = zlib.compress(imgdata.tostring())
 
         # pdf units = 1/72 inch
@@ -277,8 +286,6 @@ def convert(images, dpi, x, y, title=None, author=None, creator=None, producer=N
 
         pdf.addimage(color, width, height, imgformat, imgdata, pdf_x, pdf_y)
 
-        im.close()
-
     return pdf.tostring()
 
 
@@ -295,7 +302,7 @@ def valid_date(string):
 parser = argparse.ArgumentParser(
     description='Lossless conversion/embedding of images (in)to pdf')
 parser.add_argument(
-    'images', metavar='infile', type=argparse.FileType('rb'),
+    'images', metavar='infile', type=str,
     nargs='+', help='input file(s)')
 parser.add_argument(
     '-o', '--output', metavar='out', type=argparse.FileType('wb'),