1

I read sometimes pdf's written in Finnish. I was unable to search text containing letters ä and ö by Xpdf and Okular. Is there a PDF-reader which finds those letters correctly? I had a problem with the file elisanet.fi/matti.t.lehtinen/Geom2011.pdf.

1

I am using Evince and have to read German every once in a while. Evince correctly finds ö and ä (and other German special characters like ß).

I do have special keyboard shortcuts for those characters (using the right meta key), but you can also copy and paste them in the search field if your keyboard cannot produce them.

One thing to keep in mind though is that it is possible to create a PDF where the output looks like ä, however the graphic being composed of separate 'ä' and '¨' (a separate diaeresis/umlaut). In that case you will not find it by searching the accented character. So try out finding just 'a' and 'o' on the actual PDFs before switching your viewer.


If you look at the file with pdfedit then for the highlighted segment (sorry it is a bit small, it is on page 2 of the document):

enter image description here

the encoding is in an array like this:

enter image description here

As you can see there is little connection between the diaeresis and the 'o' (from the first field), the offsets place it there and that is not a "simple" UTF16 to UTF-8 or other character encoding conversion that you can automate.

If the text didn't have so much math in it you could try rendering to image and then do OCR, but in this case I think you better try and contact the author and get the original (probably LaTeX) source if you need to search in the text.

  • Hmm. I tried to use Evince to find word "myös" from the pdf elisanet.fi/matti.t.lehtinen/Geom2011.pdf but failed to find any. – FinnLinuxUser Sep 29 '14 at 12:25
  • @FinnLinuxUser I you copy that text out from the PDF (on page 2) and paste in a terminal you will see that that is not a normal character sequence. I guess LaTeX -> PDF has been doing something special there, and not finding that is a problem within the PDF. – Anthon Sep 29 '14 at 12:28
  • Okay. Are there some program which can convert those strange letters to normal character sequences? – FinnLinuxUser Sep 29 '14 at 12:33
  • @FinnLinuxUser If the image of the character is actually two characters you are out of luck, they don't have to be next to each other in the data stream within the PDF. They don't even have to be in the same stream. This might just be a special encoding. I will have a closer look at what is actually there. – Anthon Sep 29 '14 at 12:41
  • @FinnLinuxUser I updated my answer – Anthon Sep 29 '14 at 12:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.