Extracting ASCII text from a PDF Document

Extracting ASCII text from a PDF Document

[Martin McCormick]

I have a PDF document that does have embedded ASCII text in it.
It plays fine on a Macintosh that has no OCR software on it but
uses Voiceover. Voiceover just runs on ASCII so the ASCII is
there.

	I need to use the file on a Debian system so I hope I am
just using a2ps and pstotext wrong.

	if one uses pstotext on this document, it immediately
errors out. If I use a2ps and give it -o outfilename.ps, a2ps
runs but I may be producing an image file as there is no text
from the document, talk about sound and fury signifying nothing.

	If one runs pstotext on that output file, one gets a
single form feed for each page and nothing else.

	The PDF document is not protected.

	Any suggestions as to how to extract the text are
welcome. Thanks.

Martin McCormick

Re: Extracting ASCII text from a PDF Document

[Kirk Reiser]

What happens when you run pdftotext on the file?

On Thu, 12 Aug 2010, Martin McCormick wrote:

> I have a PDF document that does have embedded ASCII text in it.
> It plays fine on a Macintosh that has no OCR software on it but
> uses Voiceover. Voiceover just runs on ASCII so the ASCII is
> there.
>
> 	I need to use the file on a Debian system so I hope I am
> just using a2ps and pstotext wrong.
>
> 	if one uses pstotext on this document, it immediately
> errors out. If I use a2ps and give it -o outfilename.ps, a2ps
> runs but I may be producing an image file as there is no text
> from the document, talk about sound and fury signifying nothing.
>
> 	If one runs pstotext on that output file, one gets a
> single form feed for each page and nothing else.
>
> 	The PDF document is not protected.
>
> 	Any suggestions as to how to extract the text are
> welcome. Thanks.
>
> Martin McCormick
>
> _______________________________________________
> Blinux-list mailing list
> Blinux-list@redhat.com
> https://www.redhat.com/mailman/listinfo/blinux-list
>

--
Kirk Reiser				The Computer Braille Facility
e-mail: kirk@braille.uwo.ca		University of Western Ontario
phone: (519) 661-3061

Re: Extracting ASCII text from a PDF Document

[Chris Brannon]

Martin McCormick wrote:
> I have a PDF document that does have embedded ASCII text in it.
> 	I need to use the file on a Debian system so I hope I am
> just using a2ps and pstotext wrong.

Don't do that!  Use pdftotext instead.
On my distribution, ArchLinux, pdftotext is provided by the "poppler"
package.  I don't know which package you need for Debian.
Perhaps it's in xpdf.

One thing you'll notice when converting PDF to plain text is that certain
two-letter combinations are replaced with UTF-8-encoded Unicode characters.
Only the gods know why.
Common examples are fi, fl, and ff.
Of course, most screenreaders won't render those correctly.

-- Chris

Re: Extracting ASCII text from a PDF Document

[Geoff Shang]

On Thu, 12 Aug 2010, Martin McCormick wrote:

> I have a PDF document that does have embedded ASCII text in it.

Iuse pdftotext from the poppler-utils package.

Note that this expects an output filename rather than sending to stdout.

Geoff.

Re: Extracting ASCII text from a PDF Document

[Martin McCormick]

Kirk Reiser writes:
> What happens when you run pdftotext on the file?

$ pstotext  BCD996XT_v1.04.00_Protocol.pdf

< BCD996XT Operation Specification >
200
GPL Ghostscript GPL Ghostscript 8.628.62: : Unrecoverable error, exit code 1
Unrecoverable error, exit code 1
7.13. REMOTE

I think that 7.13.remote is some stray text that got outputted
from the file before pstotext exploded.

	The output would have gone to standard output had it
worked.

Martin

Re: Extracting ASCII text from a PDF Document

[Kirk Reiser]

pdftotext is a different program, mine with the -v argument returns:

pdftotext version 3.02
Copyright 1996-2007 Glyph & Cog, LLC


It also outputs to a file with the basename but containing a .txt
extension.  I believe it is part of the xpdf utilities.


On Thu, 12 Aug 2010, Martin McCormick wrote:

> Kirk Reiser writes:
>> What happens when you run pdftotext on the file?
>
> $ pstotext  BCD996XT_v1.04.00_Protocol.pdf
>
> < BCD996XT Operation Specification >
> 200
> GPL Ghostscript GPL Ghostscript 8.628.62: : Unrecoverable error, exit code 1
> Unrecoverable error, exit code 1
> 7.13. REMOTE
>
> I think that 7.13.remote is some stray text that got outputted
> from the file before pstotext exploded.
>
> 	The output would have gone to standard output had it
> worked.
>
> Martin
>
> _______________________________________________
> Blinux-list mailing list
> Blinux-list@redhat.com
> https://www.redhat.com/mailman/listinfo/blinux-list
>

--
Kirk Reiser				The Computer Braille Facility
e-mail: kirk@braille.uwo.ca		University of Western Ontario
phone: (519) 661-3061

Re: Extracting ASCII text from a PDF Document

[Hart Larry]

Well, certainly a majority of pdf files never read well, seemingly better 
results with pdftohtml, but if I have that wrong try pdf2html, since I am not 
at home, I cannot check.  Anyway the results are somewhat smoother.
Hart

Re: Extracting ASCII text from a PDF Document

[Lee Maschmeyer]

Speaking of smoothness, the -layout switch is often helpful. The table of 
contents actually comes out looking like a table of contents!

I also ran pdfinfo for the first time just now. It's pretty interesting. 
Next time I have a file I can't read I wonder if it'll help.

-- 
Lee Maschmeyer
Wayne State University Computing Center
5925 Woodward, #281
Detroit MI 48202
USA 

Re: Extracting ASCII text from a PDF Document

[Martin McCormick]

Kirk Reiser writes:
> pdftotext is a different program, mine with the -v argument returns:
> 
> pdftotext version 3.02
> Copyright 1996-2007 Glyph & Cog, LLC
> 
> 
> It also outputs to a file with the basename but containing a .txt
> extension.  I believe it is part of the xpdf utilities.

Thank you very much. I do have pdftotext and I probably need to
upgrade it as mine is 3.00 but it read the document just fine.

	I got confused and thought pstotext was what I needed as
the man page says it will convert a postscript or pdf document
to ASCII text.

	Anyway, it looks like the problem is solved by calling
the right application.

Martin McCormick