Lets one import grampslocale directly in a test script:
>>> from gramps.gen.utils.grampslocale import GrampsLocale
>>> import os
>>> gl = GrampsLocale(lang='fr_FR.UTF-8', localedir=os.path.join('build', 'mo'))
>>> tr = gl.get_translation().gettext
>>> tr("List of known family trees in your database path\n")
u'Liste des arbres familiaux connus dans votre chemin de base de donn\xe9es\n'
svn: r21640
As noted in the previous change, importing const into setup.py tried to initialize GrampsLocale and ResourcePath, which won't work. Since all we want is the VERSION string, move that to a new file, gramps/version.py
svn: r21614
Uses GRAMPS_RESOURCES to override the root location of the Gramps data files (which would normally be $(prefix)/share).
In setup.py install, writes $(prefix)/share to gen/utils/resource-path, installs that file, then deletes it again from the source tree. The presence or absence determines whether Gramps is running from the source directory or from an installation.
Const.py is now a static file; const.py.in is no longer used. Note that because importing const into setup tried to initialize things that we don't want initialized, VERSION is removed, and imported from gramps.version -- a file which will be added in the next change. Consequently, this commit will not run.
svn: r21613
Make sure that lang overrides the language list only if $LANG is set. Don't add encodings to the language list (e.g., en_US, not en_US.UTF-8). Start collation with None, because `if collation:` is True for collation = "". defaults may have whitespace at the end of "root", so use "startswith".
svn: r21596
Replace "en" with "C" when setting $LANGUAGE ("en" confuses GtkBuilder), and explicitly iterate over languages when finding
the translator, because left to itself gettext will ignore English.
svn: r21595
0002933: Problems in Narrative Web Report with surnames beginning with V or W. Use collation primary difference to define index groupings. Special case for 'V' and 'W' in Swedish removed because the default CLDR has a primary difference between them.
0003434: NarrativeWeb new alphabet index sorted incorrectly. Fix Cyrillic sort order, and contractions for Slovak and Czech (among other languages).
0003933: References inside pages in NarWeb report and non-ascii characters. Sorting issues for Polish characters.
0004423: Web report does not handle Czech "CH" character properly. Dz contraction is not present in the CLDR. Slovak is similar.
0005088: Narrated Web Site Report sort order different Windows vs Linux. Resolved by using PyICU (if available).
0005645: can't sort greek names
0005767: Sorting in Narrative Webb does not work correctly. Fix sorting in Individuals, Surnames, Families, Events and Places to use PyICU (if available). This bug covers sorting in the index header (as well as in the body of the index pages).
svn: r21509
