Andrew Lentvorski wrote:
> David Looney wrote:
>> So something about your scanned PDF (way image is stored ?) must be
>> different.
>
> Yes, it is. And you're going to laugh.
>
> The stupid scanner took a 600 dpi gray scan...
>
> And then dithered it to black and white. ARGGGHHHH!
That doesn't sound like a very user friendly thing to do ....
> All the conversions actually work *flawlessly*. The problem is in the
> final rendering for display or print.
I've found a bug in convert/gs with large enough images in PDF files,
however, that have a cropbox specified. Convert something.jpg to
something.pdf works, but then running convert something.pdf
somethingelse.jpg does not give back the original image, but rather a
background the original size with only a little image in the left lower
corner the size specified in the cropbox field of the PDF.
There is of course the direct approach (snarfed from the net):
#!/usr/bin/env python
import Image, re, zlib, sys
def stripImages(fn):
buf= open(fn,'rb').read()
fnS= fn.split(".")[0]
s =
re.findall("(?s)/XObject\s+/Subtype\s+/Image(.*?)stream\s*\012(.*?)endstream",
buf)
print len(s)
for i in s:
try:
name = re.findall("(?i)/name\s+/(\w+)",i[0])[0]
width= re.findall("(?i)/Width\s+(\d+)",i[0])[0]
height= re.findall("(?i)/Height\s+(\d+)",i[0])[0]
filter = re.findall("(?i)/filter\s+/(\w+)",i[0])[0]
colorSpace = re.findall("(?i)/ColorSpace\s+/(\w+)",i[0])[0]
except IndexError:
print "Skip:", i[0]
continue
print "Found:", name, width, height, filter, colorSpace
if filter=="FlateDecode":
im = zlib.decompress(i[1])
im = Image.fromstring("RGB", (int(width),int(height)), im)
im.save("%s_%s.jpg"%(fnS,name))
elif filter == "DCTDecode":
open("%s_%s.jpg"%(fnS,name),'wb').write(i[1])
stripImages(sys.argv[1])
Which will yank that jpeg right out of that pdf.
David Looney
--
And I sincerely believe, with you, that banking establishments are more
dangerous than standing armies.... Thomas Jefferson (letter to John
Taylor, 5/28/1816)
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
