Non-embedded standard fonts in PDF files are not displayed/rendered correctly

[corax]

Binary package hint: evince

1)
Natty:
lsb_release -rd
Description: Ubuntu Natty (development branch)
Release: 11.04

Maverick:
lsb_release -rd
Description: Ubuntu 10.10
Release: 10.10

Lucid:
Ubuntu 10.04 Lucid Lynx (amd64)

2)
Natty:
apt-cache policy acroread
acroread:
  Installed: 9.4.2-0natty1
  Candidate: 9.4.2-0natty1
  Version table:
 *** 9.4.2-0natty1 0
        500 http://archive.canonical.com/ubuntu/ natty/partner i386 Packages 100 /var/lib/dpkg/status

apt-cache policy evince
evince:
  Installed: 2.32.0-0ubuntu12.1
  Candidate: 2.32.0-0ubuntu12.1
  Version table:
 *** 2.32.0-0ubuntu12.1 0
        500 http://us.archive.ubuntu.com/ubuntu/ natty-proposed/main i386
Packages
        100 /var/lib/dpkg/status
     2.32.0-0ubuntu12 0
        500 http://us.archive.ubuntu.com/ubuntu/ natty/main i386 Packages

apt-cache policy libcairo2
libcairo2:
  Installed: 1.10.2-2ubuntu2
  Candidate: 1.10.2-2ubuntu2
  Version table:
 *** 1.10.2-2ubuntu2 0
        500 http://us.archive.ubuntu.com/ubuntu/ natty/main i386 Packages 100 /var/lib/dpkg/status

Maverick:
apt-cache policy evince
evince:
  Installed: 2.32.0-0ubuntu1
  Candidate: 2.32.0-0ubuntu1
  Version table:
 *** 2.32.0-0ubuntu1 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy libcairo2
libcairo2:
  Installed: 1.10.0-1ubuntu3
  Candidate: 1.10.0-1ubuntu3
  Version table:
 *** 1.10.0-1ubuntu3 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick-updates/main i386 Packages
        100 /var/lib/dpkg/status
     1.10.0-1ubuntu2 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages

apt-cache policy acroread
acroread:
  Installed: 9.4-1maverick1
  Candidate: 9.4-1maverick1
  Version table:
 *** 9.4-1maverick1 0
        500 http://archive.canonical.com/ubuntu/ maverick/partner i386 Packages
        100 /var/lib/dpkg/status

3) When viewing PDF attachment "PDF sample for Evince", which has font "Times Roman" or "Helvetica", the text looks different in Evince in comparison to Adobe Reader.

4) What is expected is that it looks the same in both.

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: evince 2.30.3-0ubuntu1.1
ProcVersionSignature: Ubuntu 2.6.32-25.45-generic 2.6.32.21+drm33.7
Uname: Linux 2.6.32-25-generic x86_64
Architecture: amd64
Date: Thu Oct 28 13:51:21 2010
InstallationMedia: Ubuntu 10.04 "Lucid Lynx" - Release Candidate amd64 (20100419.1)
ProcEnviron:
 LANG=en_DK.utf8
 SHELL=/bin/bash
SourcePackage: evince

[corax (cpp)]

Here is an example1 of how a portion of a PDF-file is displayed in Adobe Reader. Showing it correctly.

[corax (cpp)]

Here is an example1 of how a portion of a PDF-file is displayed in Evince. Showing it incorrectly.

[rusivi2 (rusivi2-deactivatedaccount)]

corax, thank you for reporting this bug and helping to make Ubuntu better. In Maverick, opened a new OOo Writer document, created text with Helvecta & Times New Roman fonts, exported as PDF, looks good in both Adobe Reader & Evince (please see attached PDF file & screenshot). As per our IRC discussion, please attach a PDF file that demonstrates the reported problem and toggle the status to new. Marking incomplete.

lsb_release -rd
Description: Ubuntu 10.10
Release: 10.10

apt-cache policy openoffice.org-writer
openoffice.org-writer:
  Installed: 1:3.2.1-7ubuntu1
  Candidate: 1:3.2.1-7ubuntu1
  Version table:
 *** 1:3.2.1-7ubuntu1 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy evince
evince:
  Installed: 2.32.0-0ubuntu1
  Candidate: 2.32.0-0ubuntu1
  Version table:
 *** 2.32.0-0ubuntu1 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy acroread
acroread:
  Installed: 9.4-1maverick1
  Candidate: 9.4-1maverick1
  Version table:
 *** 9.4-1maverick1 0
        500 http://archive.canonical.com/ubuntu/ maverick/partner i386 Packages
        100 /var/lib/dpkg/status

[corax (cpp)]

The best I can provide for now is info on the issue as I haven't been able to find another online generated PDF file that doesn't use Embedded fonts.

The "Example.pdf" you've attached to this issue is using Embedded-fonts ... AND ... the embedded fonts are not Times Roman or Helvetica (only their linux substitutes). But if you can generate or create a PDF-document with the following three non-embedded fonts - "Times Roman"(type 1), "Helvetica"(type 1) and "Helvetica-Bold"(type 1) ... and show that all three texts inside the document are displayed the same in both Adobe Reader and Evince ... then the bug is something else.
However I'm not able to use OpenOffice Writer to create examples as it will embed the fonts and/or insert the substitute fonts.
If you have a method for generating PDFs, that can allow me to select the non-embedded fonts, then plz let me know and I'll make an example available in this bug-report.

[Marcel Stimberg (marcelstimberg)]

Hi,
I can't reproduce your issue here (Ubuntu 10.10). Attached is a (latex-generated) file with Times Roman and Helvetica but not using font embedding. The file renders fine.

[corax (cpp)]

Thank you very much Marcel for you sample and your help. Can you generate one more PDF that also includes the Helvetica-Bold font (type 1) ?. So all three mentioned fonts are there. Just to be sure.
(and only if possible can you upload a PDF1.6-version as well... not important)

[Marcel Stimberg (marcelstimberg)]

Ok, the attached document contains all three fonts:
$ pdffonts times_helvetica_bold.pdf
name type emb sub uni object ID
------------------------------------ ----------------- --- --- --- ---------
Helvetica Type 1 no no no 9 0
Times-Roman Type 1 no no no 8 0
Helvetica-Bold Type 1 no no no 10 0

I don't know how to use latex and dvipdf to create a PDF 1.6 doc, though.

[corax (cpp)]

I think I'm a bit closer in understanding the issue. Thanx to Marcel's PDF creations (thank you Marcel, very appreciated), I've had a file for comparison. As I suspected, there is an "encoding"-problem or maybe rather an "encoding-interpretation"-problem. When creating a PDF through PostScript, the issue becomes "ironed" out, so this issue only arises when creating PDFs directly (like on-demand, streaming PDF's on the internet). My guess is that when Evince encounters an encoding that hasn't been implemented or can't be read, it assumes the encoding should be "Identity-H" (see the following links)

http://indesignsecrets.com/cid-identity-h-fonts-are-back.php

http://www.talkgraphics.com/showthread.php?33648-What-is-identity-H

Where as xPDF and Adobe Reader, either has the encoding implemented or assumes the encoding is "Custom", but display it as "Ansi". Evince displays it as "Identity-H" an tries to draw the text/letters as some weird font or attached vector-graphics image. So there is probably two issues ...one is the PDF creation, which is not always done "beautifully" ...two is the rendering, which doesn't always account for PDF-format discrepancies.
But either way, I would still recommend that the interpretation of discrepancies in Evince should be the same as Adobe Reader and xpdf. (even though I'm a believer of creating data-files according to their definition as agreed)

For further testing, you would proberly need to create a PDF that uses the non-embedded fonts AND is created directly, without using any form of PostScript. (I'll would suggest PDFLib - www.pdflib.com)

[corax (cpp)]

PDFLib had a java-library, so I took the liberty of creating a sample. See for yourself an compare in Adobe Reader, xPDF and Evince.

[rusivi2 (rusivi2-deactivatedaccount)]

In Maverick, Attachment "PDF sample for Evince" shows a discrepancy between evince & acroread. Marking confirmed.

lsb_release -rd
Description: Ubuntu 10.10
Release: 10.10

apt-cache policy acroread
acroread:
  Installed: 9.4-1maverick1
  Candidate: 9.4-1maverick1
  Version table:
 *** 9.4-1maverick1 0
        500 http://archive.canonical.com/ubuntu/ maverick/partner i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy evince
evince:
  Installed: 2.32.0-0ubuntu1
  Candidate: 2.32.0-0ubuntu1
  Version table:
 *** 2.32.0-0ubuntu1 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages
        100 /var/lib/dpkg/status

[steffmeister (steffed)]

Sorry for making some noise if this does not fit here, but I have a PDF where fonts seem also to be missing, I attach it, maybe it can be helpful.

[corax (cpp)]

The problem/issue still exists .. both in Lucid and in Maverick.
Are there any news, development or timetable in regards to this bug?

[rusivi2 (rusivi2-deactivatedaccount)]

Created attachment 46013
PDF sample for Evince

Evince bug may be found at:
https://bugzilla.gnome.org/show_bug.cgi?id=633843

Downstream bug may be found at:
https://bugs.launchpad.net/ubuntu/+source/evince/+bug/667752

In Maverick, Attachment "PDF sample for Evince" shows a discrepancy between how
it is displayed in evince versus acroread. Advised by Evince developer post bug here.

lsb_release -rd
Description: Ubuntu 10.10
Release: 10.10

apt-cache policy acroread
acroread:
  Installed: 9.4-1maverick1
  Candidate: 9.4-1maverick1
  Version table:
 *** 9.4-1maverick1 0
        500 http://archive.canonical.com/ubuntu/ maverick/partner i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy evince
evince:
  Installed: 2.32.0-0ubuntu1
  Candidate: 2.32.0-0ubuntu1
  Version table:
 *** 2.32.0-0ubuntu1 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages
        100 /var/lib/dpkg/status

apt-cache policy libcairo2
libcairo2:
  Installed: 1.10.0-1ubuntu3
  Candidate: 1.10.0-1ubuntu3
  Version table:
 *** 1.10.0-1ubuntu3 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick-updates/main i386 Packages
        100 /var/lib/dpkg/status
     1.10.0-1ubuntu2 0
        500 http://us.archive.ubuntu.com/ubuntu/ maverick/main i386 Packages

[corax (cpp)]

There are still issues in Evince in the upcoming release 11.04 (beta2)

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Adrian Johnson (ajohnson-redneon)]

Created attachment 57075
adjust substitute font width using first character in each string

There are two problems here:
 1) The "hello world" text is too narrow
 2) The "www.pdflib.com" text has spike on each path segment.

The first problem is caused by the code that adjusts the size of text drawn with a substituted font to match the width in the pdf file. This code matches the width of the letter 'm' in the font with the width in the pdf file. The problem in this case is the pdf file has bogus widths for characters not used included the letter 'm'. The attached patch fixes this by using the first letter of each string to set the font size.

The second problem is a cairo bug. In this case the pdf file is stroking the outline of a Type 3 font. Each glyph in the Type 3 font strokes a path. Cairo needs to implement stroke to path to make this work correctly.

[Germán Poo-Caamaño (gpoo)]

FWIW, it complains with hundreds of messages like:

Syntax Error (6326): No font in show/space
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6337): No font in show
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6347): No font in show
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6365): No font in show/space
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6371): No font in show/space
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6379): No font in show/space
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6384): No font in show/space
Syntax Error: Unknown font tag 'F1.1'
Syntax Error (6389): No font in show/space

[Albert Astals Cid (aacid)]

Adrian do you think this needs to go in? What about the comments by Germán?

[Adrian Johnson (ajohnson-redneon)]

Carlos, could you review the patch.

I don't see the Syntax Error messages. Maybe it depends on the font. What does pdffonts -subst show?

[Carlos Garcia Campos (carlosgc)]

The problem with the patch is that in some cases it's clear that it improves the output, but in some others I have doubts whether the output is better or worse, I'll attach some examples found while running my tests.

[Carlos Garcia Campos (carlosgc)]

Created attachment 70366
Test results examples

Let me know if you want me to send the PDFs to you

[corax (cpp)]

This problem is still here in Ubuntu 14.04 LTS ... and it has become a serious matter.
More and more of my own customers are switching to Ubuntu, which is a positive thing IMO. But with the increasing use of online web-services to handle billing, document exchanges and online-shopping ... more and more streaming-PDF files are being created on-the-fly as everyone is expected to be able to read that particular format. But when people can't read the amount due on an invoice or the delivery-date on a webshopping receipt, because the font-rendering of the pre-installed document viewer in Ubuntu won't display them correctly (or sometimes not at all), it really becomes highly inconvenient for both regular users as well as more experienced user. The Adobe Reader installation package is no longer available in ubuntu 14.04's standard repositories, which was a go-to option for the more experienced user.
SO BUCKLE UP PEOPLE, let's see if we can't fix this issue. And hopefully before we use a couple of gigabytes on the web in forum- and chatlogs about the subject of creating temporary fixes and work-arounds of such a basic thing.

best of luck to all

[madbiologist (me-again)]

Posted in the downstream bug report by the original bug reporter on 17th September 2014:

This problem is still here in Ubuntu 14.04 LTS ... and it has become a serious matter.
More and more of my own customers are switching to Ubuntu, which is a positive thing IMO. But with the increasing use of online web-services to handle billing, document exchanges and online-shopping ... more and more streaming-PDF files are being created on-the-fly as everyone is expected to be able to read that particular format. But when people can't read the amount due on an invoice or the delivery-date on a webshopping receipt, because the font-rendering of the pre-installed document viewer in Ubuntu won't display them correctly (or sometimes not at all), it really becomes highly inconvenient for both regular users as well as more experienced user. The Adobe Reader installation package is no longer available in ubuntu 14.04's standard repositories, which was a go-to option for the more experienced user.
SO BUCKLE UP PEOPLE, let's see if we can't fix this issue. And hopefully before we use a couple of gigabytes on the web in forum- and chatlogs about the subject of creating temporary fixes and work-arounds of such a basic thing.

best of luck to all

[dino99 (9d9)]

# 633843 RESOLVED NOTGNOME

[Jason Crain (jcrain)]

The "narrow text" part of this should be fixed with bug 94054.

[madbiologist (me-again)]

The fix mentioned in comment #26 is included in poppler 0.43.0. Ubuntu 16.10 "Yakkety Yak" is scheduled for release on 20th October 2016 and will contain poppler 0.44.0-3ubuntu2.

[Gitlab-migration (gitlab-migration)]

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/poppler/poppler/issues/9.