Merge pull request #1896 from MITx/feature/straz/i18n

Feature/straz/i18n - rake tasks to generate strings

conf/locale/config

-{"locales" : ["en", "fr", "de"]}
+{"locales" : ["en"]}

conf/locale/en/LC_MESSAGES/messages.po

+# empty

i18n/dummy.py

@@ -86,7 +86,7 @@ class Dummy (Converter):
     def init_msgs(self, msgs):
         """
         Make sure the first msg in msgs has a plural property.
-        msgs is list of instances of pofile.Msg
+        msgs is list of instances of polib.POEntry
         """
         if len(msgs)==0:
             return
@@ -100,8 +100,8 @@ class Dummy (Converter):
 
     def convert_msg(self, msg):
         """
-        Takes one Msg object and converts it (adds a dummy translation to it)
-        msg is an instance of pofile.Msg
+        Takes one POEntry object and converts it (adds a dummy translation to it)
+        msg is an instance of polib.POEntry
         """
         source = msg.msgid
         if len(source)==0:

i18n/execute.py

+import os, subprocess, logging, json
+
+def init_module():
+    """
+    Initializes module parameters
+    """
+    global BASE_DIR, LOCALE_DIR, CONFIG_FILENAME, SOURCE_MSGS_DIR, SOURCE_LOCALE, LOG
+
+    # BASE_DIR is the working directory to execute django-admin commands from.
+    # Typically this should be the 'mitx' directory.
+    BASE_DIR = os.path.normpath(os.path.dirname(os.path.abspath(__file__))+'/..')
+
+    # Source language is English
+    SOURCE_LOCALE = 'en'
+
+    # LOCALE_DIR contains the locale files.
+    # Typically this should be 'mitx/conf/locale'
+    LOCALE_DIR = BASE_DIR + '/conf/locale'
+
+    # CONFIG_FILENAME contains localization configuration in json format
+    CONFIG_FILENAME = LOCALE_DIR + '/config'
+
+    # SOURCE_MSGS_DIR contains the English po files.
+    SOURCE_MSGS_DIR = messages_dir(SOURCE_LOCALE)
+
+    # Default logger.
+    LOG = get_logger()
+    
+
+def messages_dir(locale):
+    """
+    Returns the name of the directory holding the po files for locale.
+    Example: mitx/conf/locale/en/LC_MESSAGES
+    """
+    return os.path.join(LOCALE_DIR, locale, 'LC_MESSAGES')
+
+def get_logger():
+    """Returns a default logger"""
+    log = logging.getLogger(__name__)
+    log.setLevel(logging.INFO)
+    log_handler = logging.StreamHandler()
+    log_handler.setFormatter(logging.Formatter('%(asctime)s [%(levelname)s] %(message)s'))
+    log.addHandler(log_handler)
+    return log
+
+# Run this after defining messages_dir and get_logger, because it depends on these.
+init_module()    
+
+def execute (command, working_directory=BASE_DIR, log=LOG):
+    """
+    Executes shell command in a given working_directory.
+    Command is a string to pass to the shell.
+    Output is logged to log.
+    """
+    log.info(command)
+    subprocess.call(command.split(' '), cwd=working_directory)
+    
+def get_config():
+    """Returns data found in config file, or returns None if file not found"""
+    config_path = os.path.abspath(CONFIG_FILENAME)
+    if not os.path.exists(config_path):
+        log.warn("Configuration file cannot be found: %s" % \
+                 os.path.relpath(config_path, BASE_DIR))
+        return None
+    with open(config_path) as stream:
+        return json.load(stream)
+    
+def create_dir_if_necessary(pathname):
+    dirname = os.path.dirname(pathname)
+    if not os.path.exists(dirname):
+        os.makedirs(dirname)
+
+
+def remove_file(filename, log=LOG, verbose=True):
+    """
+    Attempt to delete filename.
+    Log a warning if file does not exist.
+    Logging filenames are releative to BASE_DIR to cut down on noise in output.
+    """
+    if verbose:
+        log.info('Deleting file %s' % os.path.relpath(filename, BASE_DIR))
+    if not os.path.exists(filename):
+        log.warn("File does not exist: %s" % os.path.relpath(filename, BASE_DIR))
+    else:
+        os.remove(filename)
+

i18n/extract.py

+#!/usr/bin/python
+  
+"""
+See https://edx-wiki.atlassian.net/wiki/display/ENG/PO+File+workflow
+
+ This task extracts all English strings from all source code
+ and produces three human-readable files:
+   conf/locale/en/LC_MESSAGES/django-partial.po
+   conf/locale/en/LC_MESSAGES/djangojs.po
+   conf/locale/en/LC_MESSAGES/mako.po
+
+ This task will clobber any existing django.po file.
+ This is because django-admin.py makemessages hardcodes this filename
+ and it cannot be overridden.
+ 
+"""
+
+import os
+from datetime import datetime
+from polib import pofile
+from execute import execute, create_dir_if_necessary, remove_file, \
+     BASE_DIR, LOCALE_DIR, SOURCE_MSGS_DIR, LOG
+
+
+# BABEL_CONFIG contains declarations for Babel to extract strings from mako template files
+# Use relpath to reduce noise in logs
+BABEL_CONFIG = os.path.relpath(LOCALE_DIR + '/babel.cfg', BASE_DIR)
+
+# Strings from mako template files are written to BABEL_OUT
+# Use relpath to reduce noise in logs
+BABEL_OUT = os.path.relpath(SOURCE_MSGS_DIR + '/mako.po', BASE_DIR)
+
+
+def main ():
+    create_dir_if_necessary(LOCALE_DIR)
+    generated_files = ('django-partial.po', 'djangojs.po', 'mako.po')
+
+    for filename in generated_files:
+        remove_file(os.path.join(SOURCE_MSGS_DIR, filename))
+
+    # Extract strings from mako templates
+    babel_mako_cmd = 'pybabel extract -F %s -c "TRANSLATORS:" . -o %s' % (BABEL_CONFIG, BABEL_OUT)
+
+    # Extract strings from django source files
+    make_django_cmd = 'django-admin.py makemessages -l en --ignore=src/* --ignore=i18n/* ' \
+                      + '--extension html'
+
+    # Extract strings from javascript source files
+    make_djangojs_cmd = 'django-admin.py makemessages -l en -d djangojs --ignore=src/* ' \
+                        + '--ignore=i18n/* --extension js'
+    execute(babel_mako_cmd, working_directory=BASE_DIR)
+    execute(make_django_cmd, working_directory=BASE_DIR)
+    # makemessages creates 'django.po'. This filename is hardcoded.
+    # Rename it to django-partial.po to enable merging into django.po later.
+    os.rename(os.path.join(SOURCE_MSGS_DIR, 'django.po'), 
+              os.path.join(SOURCE_MSGS_DIR, 'django-partial.po'))
+    execute(make_djangojs_cmd, working_directory=BASE_DIR)
+
+    for filename in generated_files:
+        LOG.info('Cleaning %s' % filename)
+        po = pofile(os.path.join(SOURCE_MSGS_DIR, filename))
+        # replace default headers with edX headers
+        fix_header(po)
+        # replace default metadata with edX metadata
+        fix_metadata(po)
+        # remove key strings which belong in messages.po
+        strip_key_strings(po)
+        po.save()
+
+# By default, django-admin.py makemessages creates this header:
+"""
+SOME DESCRIPTIVE TITLE.
+Copyright (C) YEAR THE PACKAGE'S COPYRIGHT HOLDER
+This file is distributed under the same license as the PACKAGE package.
+FIRST AUTHOR <EMAIL@ADDRESS>, YEAR.
+"""
+
+def fix_header(po):
+    """
+    Replace default headers with edX headers
+    """
+    header = po.header
+    fixes = (
+        ('SOME DESCRIPTIVE TITLE', 'edX translation file'),
+        ('Translations template for PROJECT.', 'edX translation file'),
+        ('YEAR', '%s' % datetime.utcnow().year),
+        ('ORGANIZATION', 'edX'),
+        ("THE PACKAGE'S COPYRIGHT HOLDER", "EdX"),
+        ('This file is distributed under the same license as the PROJECT project.',
+         'This file is distributed under the GNU AFFERO GENERAL PUBLIC LICENSE.'),
+        ('This file is distributed under the same license as the PACKAGE package.',
+         'This file is distributed under the GNU AFFERO GENERAL PUBLIC LICENSE.'),
+        ('FIRST AUTHOR <EMAIL@ADDRESS>',
+         'EdX Team <info@edx.org>')
+        )
+    for (src, dest) in fixes:
+        header = header.replace(src, dest)
+    po.header = header
+
+# By default, django-admin.py makemessages creates this metadata:
+"""
+{u'PO-Revision-Date': u'YEAR-MO-DA HO:MI+ZONE',
+ u'Language': u'',
+ u'Content-Transfer-Encoding': u'8bit',
+ u'Project-Id-Version': u'PACKAGE VERSION',
+ u'Report-Msgid-Bugs-To': u'',
+ u'Last-Translator': u'FULL NAME <EMAIL@ADDRESS>', 
+ u'Language-Team': u'LANGUAGE <LL@li.org>',
+ u'POT-Creation-Date': u'2013-04-25 14:14-0400',
+ u'Content-Type': u'text/plain; charset=UTF-8',
+ u'MIME-Version': u'1.0'}
+""" 
+
+def fix_metadata(po):
+    """
+    Replace default metadata with edX metadata
+    """
+    fixes = {'PO-Revision-Date': datetime.utcnow(),
+             'Report-Msgid-Bugs-To': 'translation_team@edx.org',
+             'Project-Id-Version': '0.1a',
+             'Language' : 'en',
+             'Language-Team': 'translation team <translation_team@edx.org>',
+             }
+    if po.metadata.has_key('Last-Translator'):
+        del po.metadata['Last-Translator']
+    po.metadata.update(fixes)
+
+def strip_key_strings(po):
+    """
+    Removes all entries in PO which are key strings.
+    These entries should appear only in messages.po, not in any other po files.
+    """
+    newlist = [entry for entry in po if not is_key_string(entry.msgid)]
+    del po[:]
+    po += newlist
+
+def is_key_string(string):
+    """
+    returns True if string is a key string.
+    Key strings begin with underscore.
+    """
+    return len(string)>1 and string[0]=='_'
+
+if __name__ == '__main__':
+    main()

i18n/generate.py

+#!/usr/bin/python
+
+"""
+ See https://edx-wiki.atlassian.net/wiki/display/ENG/PO+File+workflow
+
+
+ This task merges and compiles the human-readable .pofiles on the 
+ local filesystem into machine-readable .mofiles. This is typically
+ necessary as part of the build process since these .mofiles are
+ needed by Django when serving the web app.
+
+ The configuration file (in mitx/conf/locale/config) specifies which
+ languages to generate.
+"""
+
+import os
+from execute import execute, get_config, messages_dir, remove_file, \
+     BASE_DIR, LOG, SOURCE_LOCALE
+
+def merge(locale, target='django.po'):
+    """
+    For the given locale, merge django-partial.po, messages.po, mako.po -> django.po
+    """
+    LOG.info('Merging locale={0}'.format(locale))
+    locale_directory = messages_dir(locale)
+    files_to_merge = ('django-partial.po', 'messages.po', 'mako.po')
+    validate_files(locale_directory, files_to_merge)
+
+    # merged file is merged.po
+    merge_cmd = 'msgcat -o merged.po ' + ' '.join(files_to_merge)
+    execute(merge_cmd, working_directory=locale_directory)
+
+    # rename merged.po -> django.po (default)
+    merged_filename = os.path.join(locale_directory, 'merged.po')
+    django_filename = os.path.join(locale_directory, target)
+    os.rename(merged_filename, django_filename) # can't overwrite file on Windows
+
+def validate_files(dir, files_to_merge):
+    """
+    Asserts that the given files exist.
+    files_to_merge is a list of file names (no directories).
+    dir is the directory in which the files should appear.
+    raises an Exception if any of the files are not in dir.
+    """
+    for path in files_to_merge:
+        pathname = os.path.join(dir, path)
+        if not os.path.exists(pathname):
+            raise Exception("File not found: {0}".format(pathname))
+
+def main ():
+    configuration = get_config()
+    if configuration == None:
+        LOG.warn('Configuration file not found, using only English.')
+        locales = (SOURCE_LOCALE,)
+    else:
+        locales = configuration['locales']
+    for locale in locales:
+        merge(locale)
+
+    compile_cmd = 'django-admin.py compilemessages'
+    execute(compile_cmd, working_directory=BASE_DIR)
+
+if __name__ == '__main__':
+    main()

i18n/make_dummy.py

@@ -18,20 +18,12 @@
 import os, sys
 import polib
 from dummy import Dummy
+from execute import create_dir_if_necessary
 
-# Dummy language 
-# two letter language codes reference:
-# see http://www.loc.gov/standards/iso639-2/php/code_list.php
-#
-# Django will not localize in languages that django itself has not been
-# localized for. So we are using a well-known language: 'fr'.
-
-OUT_LANG = 'fr'
-
-def main(file):
+def main(file, locale):
     """
     Takes a source po file, reads it, and writes out a new po file
-    containing a dummy translation.
+    in :param locale: containing a dummy translation.
     """
     if not os.path.exists(file):
         raise IOError('File does not exist: %s' % file)
@@ -40,29 +32,36 @@ def main(file):
     converter.init_msgs(pofile.translated_entries())
     for msg in pofile:
         converter.convert_msg(msg)
-    new_file = new_filename(file, OUT_LANG)
+    new_file = new_filename(file, locale)
     create_dir_if_necessary(new_file)
     pofile.save(new_file)
-    
-
 
-def new_filename(original_filename, new_lang):
-    """Returns a filename derived from original_filename, using new_lang as the locale"""
+def new_filename(original_filename, new_locale):
+    """Returns a filename derived from original_filename, using new_locale as the locale"""
     orig_dir = os.path.dirname(original_filename)
     msgs_dir = os.path.basename(orig_dir)
     orig_file = os.path.basename(original_filename)
-    return '%s/%s/%s/%s' % (os.path.abspath(orig_dir + '/../..'),
-                            new_lang,
-                            msgs_dir,
-                            orig_file)
+    return os.path.join(orig_dir,
+                        '/../..',
+                        new_locale,
+                        msgs_dir,
+                        orig_file)
 
 
-def create_dir_if_necessary(pathname):
-    dirname = os.path.dirname(pathname)
-    if not os.path.exists(dirname):
-        os.makedirs(dirname)
+# Dummy language 
+# two letter language codes reference:
+# see http://www.loc.gov/standards/iso639-2/php/code_list.php
+#
+# Django will not localize in languages that django itself has not been
+# localized for. So we are using a well-known language: 'fr'.
+
+DEFAULT_LOCALE = 'fr'
 
 if __name__ == '__main__':
     if len(sys.argv)<2:
         raise Exception("missing file argument")
-    main(sys.argv[1])
+    if len(sys.argv)<2:
+        locale = DEFAULT_LOCALE
+    else:
+        locale = sys.argv[2]
+    main(sys.argv[1], locale)

i18n/tests/__init__.py

+from test_extract import TestExtract
+from test_generate import TestGenerate
+from test_converter import TestConverter
+from test_dummy import TestDummy

i18n/tests/test_converter.py

+import os
+from unittest import TestCase
+
+import converter
+
+class UpcaseConverter (converter.Converter):
+    """
+    Converts a string to uppercase. Just used for testing.
+    """
+    def inner_convert_string(self, string):
+        return string.upper()
+
+
+class TestConverter(TestCase):
+    """
+    Tests functionality of i18n/converter.py
+    """
+
+    def test_converter(self):
+        """
+        Tests with a simple converter (converts strings to uppercase).
+        Assert that embedded HTML and python tags are not converted.
+        """
+        c = UpcaseConverter()
+        test_cases = (
+            # no tags
+            ('big bad wolf', 'BIG BAD WOLF'),
+            # one html tag
+            ('big <strong>bad</strong> wolf', 'BIG <strong>BAD</strong> WOLF'),
+            # two html tags
+            ('big <b>bad</b> <i>wolf</i>', 'BIG <b>BAD</b> <i>WOLF</i>'),
+            # one python tag
+            ('big %(adjective)s wolf', 'BIG %(adjective)s WOLF'),
+            # two python tags
+            ('big %(adjective)s %(noun)s', 'BIG %(adjective)s %(noun)s'),
+            # both kinds of tags
+            ('<strong>big</strong> %(adjective)s %(noun)s',
+             '<strong>BIG</strong> %(adjective)s %(noun)s'),
+            )
+        for (source, expected) in test_cases:
+            result = c.convert(source)
+            self.assertEquals(result, expected)

i18n/tests/test_dummy.py

+import os, string, random
+from unittest import TestCase
+from polib import POEntry
+
+import dummy
+
+
+class TestDummy(TestCase):
+    """
+    Tests functionality of i18n/dummy.py
+    """
+
+    def setUp(self):
+        self.converter = dummy.Dummy()
+
+    def test_dummy(self):
+        """
+        Tests with a dummy converter (adds spurious accents to strings).
+        Assert that embedded HTML and python tags are not converted.
+        """
+        test_cases = (("hello my name is Bond, James Bond",
+                       u'h\xe9ll\xf6 my n\xe4m\xe9 \xefs B\xf6nd, J\xe4m\xe9s B\xf6nd Lorem i#'),
+
+                      ('don\'t convert <a href="href">tag ids</a>',
+                        u'd\xf6n\'t \xe7\xf6nv\xe9rt <a href="href">t\xe4g \xefds</a> Lorem ipsu#'),
+                      
+                      ('don\'t convert %(name)s tags on %(date)s',
+                        u"d\xf6n't \xe7\xf6nv\xe9rt %(name)s t\xe4gs \xf6n %(date)s Lorem ips#")
+                      )
+        for (source, expected) in test_cases:
+            result = self.converter.convert(source)
+            self.assertEquals(result, expected)
+
+    def test_singular(self):
+        entry = POEntry()
+        entry.msgid = 'A lovely day for a cup of tea.'
+        expected = u'\xc0 l\xf6v\xe9ly d\xe4y f\xf6r \xe4 \xe7\xfcp \xf6f t\xe9\xe4. Lorem i#'
+        self.converter.convert_msg(entry)
+        self.assertEquals(entry.msgstr, expected)
+
+    def test_plural(self):
+        entry = POEntry()
+        entry.msgid = 'A lovely day for a cup of tea.'
+        entry.msgid_plural = 'A lovely day for some cups of tea.'
+        expected_s = u'\xc0 l\xf6v\xe9ly d\xe4y f\xf6r \xe4 \xe7\xfcp \xf6f t\xe9\xe4. Lorem i#'
+        expected_p = u'\xc0 l\xf6v\xe9ly d\xe4y f\xf6r s\xf6m\xe9 \xe7\xfcps \xf6f t\xe9\xe4. Lorem ip#'
+        self.converter.convert_msg(entry)
+        result = entry.msgstr_plural
+        self.assertEquals(result['0'], expected_s)
+        self.assertEquals(result['1'], expected_p)

i18n/tests/test_extract.py

+import os, polib
+from unittest import TestCase
+from nose.plugins.skip import SkipTest
+from datetime import datetime, timedelta
+
+import extract
+from execute import SOURCE_MSGS_DIR
+
+# Make sure setup runs only once
+SETUP_HAS_RUN = False
+
+class TestExtract(TestCase):
+    """
+    Tests functionality of i18n/extract.py
+    """
+    generated_files = ('django-partial.po', 'djangojs.po', 'mako.po')
+
+    def setUp(self):
+        # Skip this test because it takes too long (>1 minute)
+        # TODO: figure out how to declare a "long-running" test suite
+        # and add this test to it.
+        raise SkipTest() 
+
+        global SETUP_HAS_RUN
+        
+        # Subtract 1 second to help comparisons with file-modify time succeed,
+        # since os.path.getmtime() is not millisecond-accurate
+        self.start_time = datetime.now() - timedelta(seconds=1)
+        super(TestExtract, self).setUp()
+        if not SETUP_HAS_RUN:
+            # Run extraction script. Warning, this takes 1 minute or more
+            extract.main()
+            SETUP_HAS_RUN = True
+
+    def get_files (self):
+        """
+        This is a generator.
+        Returns the fully expanded filenames for all extracted files
+        Fails assertion if one of the files doesn't exist.
+        """
+        for filename in self.generated_files:
+            path = os.path.join(SOURCE_MSGS_DIR, filename)
+            exists = os.path.exists(path)
+            self.assertTrue(exists, msg='Missing file: %s' % filename)
+            if exists:
+                yield path
+
+    def test_files(self):
+        """
+        Asserts that each auto-generated file has been modified since 'extract' was launched.
+        Intended to show that the file has been touched by 'extract'.
+        """
+
+        for path in self.get_files():
+            self.assertTrue(datetime.fromtimestamp(os.path.getmtime(path)) > self.start_time,
+                            msg='File not recently modified: %s' % os.path.basename(path))
+
+    def test_is_keystring(self):
+        """
+        Verifies is_keystring predicate
+        """
+        entry1 = polib.POEntry()
+        entry2 = polib.POEntry()
+        entry1.msgid = "_.lms.admin.warning.keystring"
+        entry2.msgid = "This is not a keystring"
+        self.assertTrue(extract.is_key_string(entry1.msgid))
+        self.assertFalse(extract.is_key_string(entry2.msgid))
+    
+    def test_headers(self):
+        """Verify all headers have been modified"""
+        for path in self.get_files():
+            po = polib.pofile(path)
+            header = po.header
+            self.assertEqual(header.find('edX translation file'), 0,
+                             msg='Missing header in %s:\n"%s"' % \
+                             (os.path.basename(path), header))
+
+    def test_metadata(self):
+        """Verify all metadata has been modified"""
+        for path in self.get_files():    
+            po = polib.pofile(path)
+            metadata = po.metadata
+            value = metadata['Report-Msgid-Bugs-To']
+            expected = 'translation_team@edx.org'
+            self.assertEquals(expected, value)

i18n/tests/test_generate.py

+import os, string, random
+from unittest import TestCase
+from datetime import datetime, timedelta
+
+import generate
+from execute import get_config, messages_dir, SOURCE_MSGS_DIR, SOURCE_LOCALE
+
+class TestGenerate(TestCase):
+    """
+    Tests functionality of i18n/generate.py
+    """
+    generated_files = ('django-partial.po', 'djangojs.po', 'mako.po')
+
+    def setUp(self):
+        self.configuration = get_config()
+
+        # Subtract 1 second to help comparisons with file-modify time succeed,
+        # since os.path.getmtime() is not millisecond-accurate
+        self.start_time = datetime.now() - timedelta(seconds=1)
+
+    def test_configuration(self):
+        """
+        Make sure we have a valid configuration file,
+        and that it contains an 'en' locale.
+        """
+        self.assertIsNotNone(self.configuration)
+        locales = self.configuration['locales']
+        self.assertIsNotNone(locales)
+        self.assertIsInstance(locales, list)
+        self.assertIn('en', locales)
+
+    def test_merge(self):
+        """
+        Tests merge script on English source files.
+        """
+        filename = os.path.join(SOURCE_MSGS_DIR, random_name())
+        generate.merge(SOURCE_LOCALE, target=filename)
+        self.assertTrue(os.path.exists(filename))
+        os.remove(filename)
+
+    def test_main(self):
+        """
+        Runs generate.main() which should merge source files,
+        then compile all sources in all configured languages.
+        Validates output by checking all .mo files in all configured languages.
+        .mo files should exist, and be recently created (modified
+        after start of test suite)
+        """
+        generate.main()
+        for locale in self.configuration['locales']:
+            for filename in ('django.mo', 'djangojs.mo'):
+                path = os.path.join(messages_dir(locale), filename)
+                exists = os.path.exists(path)
+                self.assertTrue(exists, msg='Missing file in locale %s: %s' % (locale, filename))
+                self.assertTrue(datetime.fromtimestamp(os.path.getmtime(path)) >= self.start_time,
+                                msg='File not recently modified: %s' % path)
+
+def random_name(size=6):
+    """Returns random filename as string, like test-4BZ81W"""
+    chars = string.ascii_uppercase + string.digits
+    return 'test-' + ''.join(random.choice(chars) for x in range(size))

i18n/update.py

-#!/usr/bin/python
-
-import os, subprocess, logging, json
-from make_dummy import create_dir_if_necessary, main as dummy_main
-
-'''
-Generate or update all translation files
- Usage:
-    $ update.py
-
-
- 1. extracts files from mako templates
- 2. extracts files from django templates and python source files
- 3. extracts files from django javascript files
- 4. generates dummy text translations
- 5. compiles po files to mo files
-
- Configuration (e.g. known languages) declared in mitx/conf/locale/config
-'''
-
-# -----------------------------------
-# BASE_DIR is the working directory to execute django-admin commands from.
-# Typically this should be the 'mitx' directory.
-BASE_DIR = os.path.abspath(os.path.dirname(os.path.abspath(__file__))+'/..')
-
-# LOCALE_DIR contains the locale files.
-# Typically this should be 'mitx/conf/locale'
-LOCALE_DIR = BASE_DIR + '/conf/locale'
-
-# MSGS_DIR contains the English po files
-MSGS_DIR = LOCALE_DIR + '/en/LC_MESSAGES'
-
-# CONFIG_FILENAME contains localization configuration in json format
-CONFIG_FILENAME = LOCALE_DIR + '/config'
-
-# BABEL_CONFIG contains declarations for Babel to extract strings from mako template files
-BABEL_CONFIG = LOCALE_DIR + '/babel.cfg'
-
-# Strings from mako template files are written to BABEL_OUT
-BABEL_OUT = MSGS_DIR + '/mako.po'
-
-# These are the shell commands invoked by main()
-COMMANDS = {
-    'babel_mako': 'pybabel extract -F %s -c "TRANSLATORS:" . -o %s' % (BABEL_CONFIG, BABEL_OUT),
-    'make_django': 'django-admin.py makemessages --all --ignore=src/* --extension html -l en',
-    'make_djangojs': 'django-admin.py makemessages --all -d djangojs --ignore=src/* --extension js -l en',
-    'msgcat' : 'msgcat -o merged.po django.po %s' % BABEL_OUT,
-    'rename_django' : 'mv django.po django_old.po',
-    'rename_merged' : 'mv merged.po django.po',
-    'compile': 'django-admin.py compilemessages'
-    
-    }
-
-def execute (command_kwd, log, working_directory=BASE_DIR):
-    '''
-    Executes command_kwd, which references a shell command in COMMANDS.
-    '''
-    full_cmd = COMMANDS[command_kwd]
-    log.info('%s' % full_cmd)
-    subprocess.call(full_cmd.split(' '), cwd=working_directory)
-    
-def make_log ():
-    '''returns a logger'''
-    log = logging.getLogger(__name__)
-    log.setLevel(logging.INFO)
-    log_handler = logging.StreamHandler()
-    log_handler.setFormatter(logging.Formatter('%(asctime)s [%(levelname)s] %(message)s'))
-    log.addHandler(log_handler)
-    return log
-
-def get_config ():
-    '''Returns data found in config file, or returns None if file not found'''
-    config_path = os.path.abspath(CONFIG_FILENAME)
-    if not os.path.exists(config_path):
-        return None
-    with open(config_path) as stream:
-        return json.load(stream)
-
-def main ():
-    log = make_log()
-    create_dir_if_necessary(LOCALE_DIR)
-    log.info('Executing all commands from %s' % BASE_DIR)
-
-    remove_files = ['django.po', 'djangojs.po', 'nonesuch']
-    for filename in remove_files:
-        path = MSGS_DIR + '/' + filename
-        log.info('Deleting file %s' % path)
-        if not os.path.exists(path):
-            log.warn("File does not exist: %s" % path)
-        else:
-            os.remove(path)
-
-    # Generate or update human-readable .po files from all source code.
-    execute('babel_mako', log=log)
-    execute('make_django', log=log)
-    execute('make_djangojs', log=log)
-    execute('msgcat', log=log, working_directory=MSGS_DIR)
-    execute('rename_django', log=log, working_directory=MSGS_DIR)
-    execute('rename_merged', log=log, working_directory=MSGS_DIR)
-
-    # Generate dummy text files from the English .po files
-    log.info('Generating dummy text.')
-    dummy_main(LOCALE_DIR + '/en/LC_MESSAGES/django.po')
-    dummy_main(LOCALE_DIR + '/en/LC_MESSAGES/djangojs.po')
-
-    # Generate machine-readable .mo files
-    execute('compile', log)
-
-if __name__ == '__main__':
-    main()

rakefile

@@ -330,6 +330,12 @@ task :migrate, [:env] do |t, args|
     sh(django_admin(:lms, args.env, 'migrate'))
 end
 
+desc "Run tests for the internationalization library"
+task :test_i18n do
+  test = File.join(REPO_ROOT, "i18n", "tests")
+  sh("nosetests #{test}")
+end
+
 Dir["common/lib/*"].select{|lib| File.directory?(lib)}.each do |lib|
     task_name = "test_#{lib}"
 
@@ -501,6 +507,30 @@ task :autodeploy_properties do
     end
 end
 
+# --- Internationalization tasks
+
+desc "Extract localizable strings from sources"
+task :extract_dev_strings do
+  sh(File.join(REPO_ROOT, "i18n", "extract.py"))
+end
+
+desc "Compile localizable strings from sources. With optional flag 'extract', will extract strings first."
+task :generate_i18n do
+  if ARGV.last.downcase == 'extract'
+    Rake::Task["extract_dev_strings"].execute
+  end
+  sh(File.join(REPO_ROOT, "i18n", "generate.py"))
+end
+
+desc "Simulate international translation by generating dummy strings corresponding to source strings."
+task :dummy_i18n do
+  source_files = Dir["#{REPO_ROOT}/conf/locale/en/LC_MESSAGES/*.po"]
+  dummy_locale = 'fr'
+  cmd = File.join(REPO_ROOT, "i18n", "make_dummy.py")
+  for file in source_files do
+    sh("#{cmd} #{file} #{dummy_locale}")
+  end
+end
 
 # --- Develop and public documentation ---
 desc "Invoke sphinx 'make build' to generate docs."

requirements.txt

 -r repo-requirements.txt
-Babel==0.9.6
 beautifulsoup4==4.1.3
 beautifulsoup==3.2.1
 boto==2.6.0
@@ -62,6 +61,10 @@ newrelic==1.8.0.13
 # Used for documentation gathering
 sphinx==1.1.3
 
+# Used for Internationalization and localization
+Babel==0.9.6
+transifex-client==0.8
+
 # Used for testing
 coverage==3.6
 factory_boy==2.0.2