I have a pdf file that has text with Hungarian accented characters, and is:
- correctly displayed in xpdf and
- in kpdf,
- but AdobeReader 7.0 throws "Cannot find or generate TimesNewRoman, Bold font" error, and replaces all non-accented characters with a dot on the screen. (MS True Type fonts are installed in KDE)

I want to extract the text from this pdf.

- pdftotext version 3.01:
Extracts the text, but very often inserts spaces inside a word, deviding it into two or more words. Sometimes it mixes up the order of syllables, too, so the resulting text is unacceptable.
- pdftohtml version 0.36:
The same error
- AdobeReader: cannot even display the pdf corretly, it is of no use

Is there an other - even manual - way to extract the text correctly?

What about just copying the whole text from, say, kpdf and then pasting into an editor? Just go to the "Tools" menu in kpdf and choose the "Select Tool" and copy the whole text.
Cheers,
jdk

The text is 300+ pages, and i have many such files, whilst kpdf only makes it possible to copy one page at a time. It would take days till I process them all.

An other issue: i also have some files that are copy-protected, but printing-allowed. So I print them into pdf (from kpdf and acroread), and try to copy text out of the newly generated pdf.
Here is what I get:
"De miért itt? $ Np] JpSLHVHQ EHFV~V]LN D ]DNy DOi D IHJ\YHUWRNEDQ WDUWRWW %HUHWWához. Hirtelen feltámad a szél, és végigsodor a szokatlanul kihalt utcán néhány IDOHYHOHW PHJ YDODPL HOGRERWW SDStUIHFQLW 0LQWKD HJ\ NLFVLW V|WpWHEE LV OHQQH GH KiW SHUV]H HVWH YDQ PiU KRYi WHWWH D] HV]pW $ V]pO YLV]RQW téliesen jeges – IXUFVDtJ\PiMXVN|]HSpQ0HJERU]RQJ"

The text of the newly-generated pdf is correct in the viewer, but incorrect when copied out. Pdftotext gives the same result.

What the heck is this?

P.S:
PSRESOURCEPATH is set for acroread, and the same path is given on the command line (-sFONTPATH) to the gs backend working as a pdf printer.

It looks like incorrectly rendered unicode characters to me. It is possible to drag the rectangle in kpdf over more than one page. I wouldn't fancy doing it for 300 pages though. I'm out of ideas.
good luck.
jdk

This thread http://www.linuxquestions.org/questi...erters-702769/ may help you.

The ocr solutions would be cool, but gocr s*cks, whilst tesseract has no Hungarian language support yet. (I plan to do it one day, though).
So, I will try the OpenOffice plugin now. Thx.

Doh. That extension is for OO 3.0. There is no 3.0 version for SuSE 10.1.

So it is the good old "Please upgrade your operating system and buy new hardware to view this file" case again.
I did it once, when I badly needed a firefox plugin, but no mood to do it again.

I have access to another system, running SuSE 11.0. I installed OO 3.0 there, and also the plugin. Of course the plugin install was not smooth, a little googling here, a little source downloading there, searching for and loading libraries here and there, and the pdfimport plugin was installed in not more than 2 hours.

Now it failed to import one of the pdf's in question, because it was encrypted. At least it was fast to fail with that file. But then it started to import an other pdf a half of an hour ago, and the progress bar is at 20% on a PIV 2.4G CPU now.

These "tools" are only good for wasting time...

Edit:
The pdfimport plugin finished finally. Of course the result is good for nothing, as it invented the best way by which neither text nor layout is preserved from the pdf file: a hundred thousand text boxes are scattered over the pages of the converted file. These text boxes often contain just a single character. And neither the resulting file can be saved as text, nor selecting all and copying on the clipboard is possible.