TeX-3.0 default pdf generation no good

[billyoc@…]

TeX needs additional fonts for proper pdf generation.

[Randy McMurchy]

Closing this old bug as TeX has been updated to 3.0 and the new version contains many new font sets.

[alexander@…]

There are still issues with generated PDF files. Testcase: cat >test.tex <<"EOF" \documentclass{article} \usepackage[koi8-r]{inputenc} \usepackage[russian]{babel} \begin{document} \CYRA \end{document} EOF pdflatex test.tex The resulting PDF looks correctly but text extraction is broken. pdftotext -enc UTF-8 test.pdf - | xxd The expected output is: 0000000: d090 0a0a 310a 0a0c 0a ....1.... The actual output is: 0000000: c380 0a0a 310a 0a0c 0a ....1.... The issue is fixed for xpdf by installing the "cm-super" font package from CTAN. This package also makes the PDF output look better in Adobe Reader for Windows. For text extraction and searching to work in Adobe Reader for Windows, also the tiny "cmap" package from CTAN is required in addition to "cm-super".

[alexander@…]

Forgot to change tetex version number in bug title

[bdubbs@…]

Alexander,

In reviewing this bug, the instructions for installing the cm-super fonts are

unclear.

1. I don't see an easy way to get the fonts. That is, I cannot find a

prepackaged .gz or bz2 file. Do you know where a tarball is located?

2. It looks like the steps are:
   1. Copy the */pfb files to /usr/share/texmf/fonts/type1/public/cm-super/
   2. Edit the file /usr/share/texmf/dvips/config/updmap
   3. Run updmap

The problem is that my system does not have the file in 2b and I don't know if any parameters are needed for 2c.

Can you give more specific instructions?

[alexander@…]

The installation instructions that come with cm-super are indeed outdated and don't work with tetex-3.0.

1. In fact a tar.gz file can be autogenerated on-the-fly by some ftp mirrors,

e.g: ​ftp://tug.ctan.org/tex-archive/fonts/ps-type1/cm-super.tar.gz Note that it works only with some, not all, ftp clients (wget and Konqueror are OK), and doesn't work at all if you are behind certain versions of Microsoft proxy server.

2. a. *.pfb files go into /usr/share/texmf/fonts/type1/public/cm-super/
   2. *.afm files (uncompressed) optionally go into /usr/share/texmf/fonts/afm/public/cm-super/
   3. *.enc files go to .../texmf/fonts/enc/dvips/cm-super/
   4. *.map files for dvips go to .../texmf/fonts/map/dvips/cm-super/
   5. looks like *.map files for dvipdfm could go to .../texmf/fonts/map/dvips/cm-super/, but they don't seem to be used here at all.
   6. the file to be edited is: .../texmf/web2c/updmap.cfg. Add lines:

MixedMap cm-super-t1.map MixedMap cm-super-t2a.map MixedMap cm-super-t2b.map MixedMap cm-super-t2c.map MixedMap cm-super-ts1.map MixedMap cm-super-x2.map

7. updmap doesn't need any parameters

[alexander@…]

oops,

5. looks like *.map files for dvipdfm could go to .../texmf/fonts/map/dvipdfm/cm-super/, but they don't seem to be used here at all.

[bdubbs@…]

Output from pdflatex test.tex > test.out

[bdubbs@…]

Output from pdflatex test.tex > test.out

[bdubbs@…]

output from pdflatex test.tex > test.out

[bdubbs@…]

Trying to fix this bug...

Alex,

I installed the fonts. The result of updmap includes:

updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-t1.map' updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-t2a.map' updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-t2b.map' updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-t2c.map' updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-ts1.map' updmap: using map file `/usr/share/texmf/fonts/map/dvipdfm/cm-super/cm-super-x2.map'

When running the test in Comment #2, I get: # pdftotext -enc UTF-8 test.pdf - | xxd 0000000: 310a 0a0c 0a 1....

Attaching working files.

I need to get this fixed asap if we want to get it into 6.1.

[alexander@…]

The PDF file you attached doesn't even contain the cyrillic leter A when viewed. Something went wrong, possibly because:

1) you disabled font embedding (the Adobe PDF specification _requires_ embedding of fonts with non-ISO-8859-1 glyphs). 2) Not sure why you have /root/.texmf-var/ mentioned in the output.

[alexander@…]

In your output, it looks like updmap found only dvipdfm map files, not dvips (also used by pdflatex)

[bdubbs@…]

Changing target milestone to 6.2 so we have time to validate the fix.

[alexander@…]

Found the possible problem. 1) Looks like dvipdfm map files are harmful. Try removing .../texmf/fonts/map/dvipdfm/cm-super/ at all, rerunning updmap and retesting the PDF output. Even without them, dvipdfm uses Type1 fonts. 2) I don't like the fact that the configuration is saved in /root/.texmf-var: this way, all changes in font configuration apply to root only. Fix: TEXMFVAR=/usr/share/texmfvar updmap Not sure if that also applies to "texconfig dvips paper letter" in the book.

[alexander@…]

Reposting using Mozilla Firefox instead of Links -- hopefully, this will fix the formatting.

Found the possible problem.

1) Looks like dvipdfm map files are harmful. Try removing .../texmf/fonts/map/dvipdfm/cm-super/ at all, rerunning updmap and retesting the PDF output. Even without them, dvipdfm uses Type1 fonts.

2) I don't like the fact that the configuration is saved in /root/.texmf-var: this way, all changes in font configuration apply to root only.

Fix: TEXMFVAR=/usr/share/texmf-var updmap

Not sure if that also applies to "texconfig dvips paper letter" command in the book.

[bdubbs@…]

Alex, look at:

​http://anduin.linuxfromscratch.org/~bdubbs/blfs-book-xsl/pst/typesetting.html#tex

to see if the changes I made fixes this bug.

[bdubbs@…]

Added installation instructions for cm-super fonts.