[Bug 997891] Re: sometimes cannot build pdfs for de, sl, pt, es, nl, pl, or it locales

[Vibhav Pant]

** Also affects: sphinx (Debian) via
   http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=672586
   Importance: Unknown
       Status: Unknown

-- 
You received this bug notification because you are a member of Ubuntu
Sponsors Team, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/997891

Title:
  sometimes cannot build pdfs for de, sl, pt, es, nl, pl, or it locales

Status in “sphinx” package in Ubuntu:
  Confirmed
Status in “sphinx” source package in Precise:
  Confirmed
Status in “sphinx” source package in Quantal:
  Confirmed
Status in “sphinx” package in Debian:
  Unknown

Bug description:
  Newer versions of docutils (>=0.8) report full language name rather
  than shortcodes and hence breaking pdf compilation in de, sl, pt, es,
  nl, pl and it locales as the \shorthandoff directive is not added and
  triggering latex is generated.

  I submitted this patch upstream last night and it was accepted.
  https://bitbucket.org/birkenfeld/sphinx/pull-request/64/added-
  language-names-to-the-shorthandoff

  SRU Info:
  [Impact] - On systems where python-docutils is version 0.8 or above, this bug prevents pdf generation from sphinx generated latex source for the de, sl, pt, es, nl, pl and it locales. I expect that my upstream patch will be released with the next version of python-sphinx.

  [Development Fix] - A debdiff for quantal has been attached. It is
  functionally identical to the precise-proposed debdiff as the sphinx
  versions are still identical.

  [Stable Fix] - A debdiff for precise-proposed has been attached. It
  adds full language names to the pool of shortcodes which the tex file
  writer uses to evaluate the necessity of the \shorthandoff directive.

  [Test Case] - Note: The debian maintainer offers a better minimal test
  here: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=672586#31

  I encountered this issue while trying to build the Mahara manual (
  git://gitorious.org/mahara/manual.git ) on precise.

  * Install git-core and python-sphinx along with its recommends.
  * git clone git://gitorious.org/mahara/manual.git
  * cd manual
  * mkdir source/locales/es/LC_MESSAGES/
  * from https://translations.launchpad.net/mahara-manual/1.5/+export get
  the MO version of the translation files. Put all the spanish .mo files
  in that newly created directory
  * add the following to the bottom of source/conf.py:
  language = 'es'
  locale_dirs =  ["locales/"]
  * run 'make clean latexpdf'
  * wait while approximately 258 pages worth of latex are compiled and the
  pdf attempts to compile from those. It'll stop saying this:

  <use groups_csv_add.png> <use groups_csv_add.png> [188
  <./groups_csv_add.png (P
  NG copy)>] <groups_csv.png, id=3006, 692.5875pt x 79.29625pt>
  <use groups_csv.png> <use groups_csv.png>
  ! Bad character code (-1).
  \es at chf ->\char \hyphenchar \font

  l.8537 ...dded","public","hidemembersfrommembers"}

  ?

  * view the build/latex/Mahara.tex file with an editor.
  * You should see the following after the preamble and \def stuff:
  \begin{document}

  \maketitle
  * after applying the changes in the patch the 'make clean latexpdf' command should compile the pdf without stopping. viewing the .tex file again, you should see this:
  \begin{document}
  \shorthandoff{"}
  \maketitle

  [Regression Potential] - I don't believe it should regress, nothing is
  being taken away, rather additional workaround support is added.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/sphinx/+bug/997891/+subscriptions