Bug#687230: lintian: check for duplicated translation data in binary packages

[Paul Wise <pabs@debian.org>]

Package: lintian
Severity: wishlist

Please add a check for duplicated translation data. By duplicated
translation data I mean the same translation data in multiple formats.
This happens when both the source and binary forms of translation data
are shipped in binary packages. The source forms of translation data are
generally not needed in binary packages but here may be some cases where
the source form is needed. However, when both are shipped, we can be
fairly sure the source form is not needed. The test should compare the
lists of source and binary translation files in a package after strip
any language codes from the file-names. 

There are two standard types that I am aware of:

      * gettext:
              * source form is *.po (special plain text format)
              * binary form is *.mo
              * files are named domain.ext and are located in a path
                containing the language code
                      * some may be called domain.langcode.ext instead
      * Qt:
              * source form is *.ts (an XML based format)
              * binary form is *.qm
              * files are named domain_langcode.ext or
                domain.langcode.ext

There are also various custom formats but it will not be possible to
check those automatically.

apt-file indicates quite a number of packages that this lintian check
could apply to, some semi-useful commands:

apt-file search -x \\.[pm]o$ | grep -C2 \\.po$ | grep -C2 \\.mo$
apt-file search -x '\.(ts|qm)$' | grep -C2 \\.ts$ | grep -C2 \\.qm$

-- 
bye,
pabs

http://wiki.debian.org/PaulWise