k2pdfopt: optimizes PDFs for viewing on e-readers

[willus]

Originally Posted by RefUser

It compiles well on ARM btw (OCR switched off for now).
I tested it on a 25 page document, which took about 10.5 minutes.

Wow--30 seconds per page is a little painful, but I guess not terrible. I don't think you'll want to add Tesseract to that. It could get really ugly! But I'm impressed you got it to run already. Nice work. What platform? If you want to send me the ARM binary for a given platform, I'd be happy to host it on willus.com. Did you comment out the function that pulls in libgsl? Also, if you made mods to the code for ARM, you can send them to me and I'll incorporate them into future releases.

[RefUser]

Originally Posted by willus

Wow--30 seconds per page is a little painful, but I guess not terrible. I don't think you'll want to add Tesseract to that. It could get really ugly! But I'm impressed you got it to run already. Nice work. What platform? If you want to send me the ARM binary for a given platform, I'd be happy to host it on willus.com. Did you comment out the function that pulls in libgsl? Also, if you made mods to the code for ARM, you can send them to me and I'll incorporate them into future releases.

The platform is Maemo 5 which runs on Nokia's N900http://maemo.org/. It has many packages from debian already ported, so I just had to compile libpng and mupdf, including its thirdparty libraries.
I didn't manage to link it completely statically, so while I can send you the binary, everyone who wants to use it, needs the correct versions of libjasper and libdjavulibre which in turn need libjpeg62.
If I don't figure it out in the near future, I may send you a zip file with the dynamic libraries included.
For Android a copy of libc (and libm) may be needed in either case since their version is stripped down to Android's needs.

[willus]

Can somebody with a Kindle 3 (or newer) let me know if they can highlight the text from the OCR output of k2pdfopt? I cannot do this on my Kindle 2, but I think that's because Kindle 2's cannot highlight PDF files. Anybody know for certain?

[RefUser]

I managed to link k2pdfopt statically for ARM (armv5, still without OCR) and uploaded it to (Link outdated, see below) for the moment. Even a test on my PRS-T1 passed succesfully (about the same speed as mentioned before), so it should run on most android devices.
If any one tries another port: don't rely on the thirdparty libraries of mupdf (especially libjpeg).

Originally Posted by willus

Can somebody with a Kindle 3 (or newer) let me know if they can highlight the text from the OCR output of k2pdfopt? I cannot do this on my Kindle 2, but I think that's because Kindle 2's cannot highlight PDF files. Anybody know for certain?

I have no kindle but on the PRS-T1 I can highlight the text.

[willus]

Originally Posted by RefUser

I managed to link k2pdfopt statically for ARM (armv5, still without OCR) ... it should run on most android devices.

Available on the k2pdfopt download page now. I ran UPX on it to shrink it down to ~6 MB.

[geekmaster]

I updated the dead link in the tools index wiki with this new merged thread. I wonder what other broken links the thread merge created.

[RefUser]

Originally Posted by willus

Available on the k2pdfopt download page now. I ran UPX on it to shrink it down to ~6 MB.

The upx'ed version segfaults. I've uploaded a stripped version (uncompressed 11MB) to Dropbox (which is now available from willus.com, see below).

[willus]

Originally Posted by RefUser

The upx'ed version segfaults. I've uploaded a stripped version (uncompressed 11MB) to http://dl.dropbox.com/u/2299259/k2pdfopt/k2pdfopt.xz

Sorry about that. I UPX'd it on x64 Linux but I thought it would work. I've put up your new version. Do they make UPX for ARM? Maybe you can try UPX-ing it on your system? No big deal for now. I'm guessing I won't get a ton of ARM downloads right away, but I could be wrong.

[willus]

Originally Posted by geekmaster

I updated the dead link in the tools index wiki with this new merged thread. I wonder what other broken links the thread merge created.

Thank you. I asked the moderator specifically about re-directing the dead thread link to this thread, but was told it couldn't be done. I don't know why you couldn't replace the dead thread with a single post that points to this thread, though. I would think that would be a very common need whenever threads are merged, so I'm a bit puzzled it can't be done.

[pdurrant]

Originally Posted by willus

Thank you. I asked the moderator specifically about re-directing the dead thread link to this thread, but was told it couldn't be done. I don't know why you couldn't replace the dead thread with a single post that points to this thread, though. I would think that would be a very common need whenever threads are merged, so I'm a bit puzzled it can't be done.

We don't merge threads very often. Perhaps it could have been done a different way, but we do the best we can at the time.

[geekmaster]

Originally Posted by pdurrant

We don't merge threads very often. Perhaps it could have been done a different way, but we do the best we can at the time.

Can you search for mobileread links to the dead thread (including links to posts in that thread)?

Unfortunately, only a site mod (or creator of the post containing a now-dead link) can replace those dead links to point at the new merged location.

Any descriptive references such as "see post #234 above" could also be incorrect now if the post numbers changed.

Otherwise, we need to wait and report dead links when we stumble across them.

[willus]

Originally Posted by willus

Can somebody with a Kindle 3 (or newer) let me know if they can highlight the text from the OCR output of k2pdfopt? I cannot do this on my Kindle 2, but I think that's because Kindle 2's cannot highlight PDF files. Anybody know for certain?

I've been told that the OCR text can be highlighted on a Kindle 3 and Kindle Touch (see this thread), and presumably all newer versions. I'm not very pleased that Amazon has left the Kindle 2 (and DX, apparently) folks in the dust in this regard. And yes, I know I can get a 3.x update for the k2, but I prefer not to hack it.

What's also odd is that the kindle software for the iPad can't do PDF text highlighting. Or is that just because my iPad software is tied to my purchase of a Kindle 2? Anybody know?

[willus]

Originally Posted by willus

Sorry about that. I UPX'd it on x64 Linux but I thought it would work. I've put up your new version. Do they make UPX for ARM? Maybe you can try UPX-ing it on your system? No big deal for now. I'm guessing I won't get a ton of ARM downloads right away, but I could be wrong.

After about a week of binary downloads, I've had roughly 1% of them in the ARM binary format (18 out of ~1800).

[dianner]

Hi - I tried to convert a PDF file with this tool using the default options and it doesn't seem to work. This PDF file has 2 columns on some pages and 3 columns on others. I expected the pages to be 1 column, but they continued to be 2 and 3. I also tried specifying "co" and "1" on the interactive display, with the same result. I believe this PDF file is created from a hard copy document being scanned in. Could this be a problem? Or, is there some other option to specify? I've looked at the help for the options, but I can't see any that make sense. Thanks for any help.

[DaleDe]

If a page is really just an image of the text then there is no way to change the number of columns. Often scanned pages are just images.