cmlenz@56: # -*- coding: utf-8 -*-
cmlenz@56: #
cmlenz@56: # Copyright (C) 2007 Edgewall Software
cmlenz@56: # All rights reserved.
cmlenz@56: #
cmlenz@56: # This software is licensed as described in the file COPYING, which
cmlenz@56: # you should have received as part of this distribution. The terms
cmlenz@56: # are also available at http://babel.edgewall.org/wiki/License.
cmlenz@56: #
cmlenz@56: # This software consists of voluntary contributions made by many
cmlenz@56: # individuals. For the exact contribution history, see the revision
cmlenz@56: # history and logs, available at http://babel.edgewall.org/log/.
cmlenz@56: 
cmlenz@56: """Data structures for message catalogs."""
cmlenz@56: 
cmlenz@67: from datetime import datetime
cmlenz@56: import re
cmlenz@56: try:
cmlenz@56:     set
cmlenz@56: except NameError:
cmlenz@56:     from sets import Set as set
cmlenz@67: import time
cmlenz@56: 
cmlenz@67: from babel import __version__ as VERSION
cmlenz@64: from babel.core import Locale
cmlenz@67: from babel.messages.plurals import PLURALS
cmlenz@67: from babel.util import odict, UTC
cmlenz@56: 
cmlenz@56: __all__ = ['Message', 'Catalog']
cmlenz@56: __docformat__ = 'restructuredtext en'
cmlenz@56: 
cmlenz@56: PYTHON_FORMAT = re.compile(r'\%(\([\w]+\))?[diouxXeEfFgGcrs]').search
cmlenz@56: 
cmlenz@56: 
cmlenz@56: class Message(object):
cmlenz@56:     """Representation of a single message in a catalog."""
cmlenz@56: 
cmlenz@84:     def __init__(self, id, string='', locations=(), flags=(), comments=()):
cmlenz@56:         """Create the message object.
cmlenz@56:         
cmlenz@56:         :param id: the message ID, or a ``(singular, plural)`` tuple for
cmlenz@56:                    pluralizable messages
cmlenz@56:         :param string: the translated message string, or a
cmlenz@56:                        ``(singular, plural)`` tuple for pluralizable messages
cmlenz@56:         :param locations: a sequence of ``(filenname, lineno)`` tuples
cmlenz@56:         :param flags: a set or sequence of flags
cmlenz@84:         :param comments: a sequence of translator comments for the message
cmlenz@56:         """
cmlenz@56:         self.id = id
cmlenz@68:         if not string and self.pluralizable:
cmlenz@68:             string = (u'', u'')
cmlenz@56:         self.string = string
cmlenz@70:         self.locations = list(locations)
cmlenz@56:         self.flags = set(flags)
cmlenz@67:         if id and self.python_format:
cmlenz@56:             self.flags.add('python-format')
cmlenz@56:         else:
cmlenz@56:             self.flags.discard('python-format')
cmlenz@84:         self.comments = list(comments)
cmlenz@56: 
cmlenz@56:     def __repr__(self):
cmlenz@56:         return '<%s %r>' % (type(self).__name__, self.id)
cmlenz@56: 
cmlenz@67:     def fuzzy(self):
cmlenz@67:         return 'fuzzy' in self.flags
cmlenz@67:     fuzzy = property(fuzzy, doc="""\
cmlenz@67:         Whether the translation is fuzzy.
cmlenz@67:         
cmlenz@67:         >>> Message('foo').fuzzy
cmlenz@67:         False
cmlenz@67:         >>> Message('foo', 'foo', flags=['fuzzy']).fuzzy
cmlenz@67:         True
cmlenz@67:         
cmlenz@67:         :type:  `bool`
cmlenz@67:         """)
cmlenz@67: 
cmlenz@56:     def pluralizable(self):
cmlenz@56:         return isinstance(self.id, (list, tuple))
cmlenz@56:     pluralizable = property(pluralizable, doc="""\
cmlenz@56:         Whether the message is plurizable.
cmlenz@56:         
cmlenz@56:         >>> Message('foo').pluralizable
cmlenz@56:         False
cmlenz@56:         >>> Message(('foo', 'bar')).pluralizable
cmlenz@56:         True
cmlenz@56:         
cmlenz@61:         :type:  `bool`
cmlenz@56:         """)
cmlenz@56: 
cmlenz@56:     def python_format(self):
cmlenz@56:         ids = self.id
cmlenz@56:         if not isinstance(ids, (list, tuple)):
cmlenz@56:             ids = [ids]
cmlenz@56:         return bool(filter(None, [PYTHON_FORMAT(id) for id in ids]))
cmlenz@56:     python_format = property(python_format, doc="""\
cmlenz@56:         Whether the message contains Python-style parameters.
cmlenz@56:         
cmlenz@56:         >>> Message('foo %(name)s bar').python_format
cmlenz@56:         True
cmlenz@56:         >>> Message(('foo %(name)s', 'foo %(name)s')).python_format
cmlenz@56:         True
cmlenz@56:         
cmlenz@61:         :type:  `bool`
cmlenz@56:         """)
cmlenz@56: 
cmlenz@56: 
cmlenz@56: class Catalog(object):
palgarvio@78:     """Representation of a message catalog."""
cmlenz@56: 
cmlenz@67:     def __init__(self, locale=None, domain=None, project=None, version=None,
palgarvio@78:                  msgid_bugs_address=None, creation_date=None,
palgarvio@78:                  revision_date=None, last_translator=None, charset='utf-8'):
cmlenz@64:         """Initialize the catalog object.
cmlenz@64:         
cmlenz@64:         :param locale: the locale identifier or `Locale` object, or `None`
cmlenz@64:                        if the catalog is not bound to a locale (which basically
cmlenz@64:                        means it's a template)
palgarvio@78:         :param domain: the message domain
palgarvio@78:         :param project: the project's name
palgarvio@78:         :param version: the project's version
palgarvio@78:         :param msgid_bugs_address: the address to report bugs about the catalog
palgarvio@78:         :param creation_date: the date the catalog was created
palgarvio@78:         :param revision_date: the date the catalog was revised
palgarvio@78:         :param last_translator: the name and email of the last translator
cmlenz@64:         """
cmlenz@64:         self.domain = domain #: the message domain
cmlenz@64:         if locale:
cmlenz@64:             locale = Locale.parse(locale)
cmlenz@64:         self.locale = locale #: the locale or `None`
cmlenz@67:         self._messages = odict()
cmlenz@67: 
cmlenz@67:         self.project = project or 'PROJECT' #: the project name
cmlenz@84:         self.version = version or 'VERSION' #: the project version
palgarvio@78:         self.msgid_bugs_address = msgid_bugs_address or 'EMAIL@ADDRESS'
cmlenz@84: 
cmlenz@67:         if creation_date is None:
cmlenz@67:             creation_date = time.localtime()
cmlenz@67:         elif isinstance(creation_date, datetime):
cmlenz@67:             if creation_date.tzinfo is None:
cmlenz@67:                 creation_date = creation_date.replace(tzinfo=UTC)
cmlenz@67:             creation_date = creation_date.timetuple()
cmlenz@67:         self.creation_date = creation_date #: creation date of the template
cmlenz@67:         if revision_date is None:
cmlenz@67:             revision_date = time.localtime()
cmlenz@67:         elif isinstance(revision_date, datetime):
cmlenz@67:             if revision_date.tzinfo is None:
cmlenz@67:                 revision_date = revision_date.replace(tzinfo=UTC)
cmlenz@67:             revision_date = revision_date.timetuple()
cmlenz@67:         self.revision_date = revision_date #: last revision date of the catalog
cmlenz@67:         self.last_translator = last_translator #: last translator name + email
cmlenz@68:         self.charset = charset or 'utf-8'
cmlenz@67: 
cmlenz@67:     def headers(self):
cmlenz@67:         headers = []
cmlenz@67:         headers.append(('Project-Id-Version',
cmlenz@67:                         '%s %s' % (self.project, self.version)))
palgarvio@78:         headers.append(('Report-Msgid-Bugs-To', self.msgid_bugs_address))
cmlenz@67:         headers.append(('POT-Creation-Date',
cmlenz@67:                         time.strftime('%Y-%m-%d %H:%M%z', self.creation_date)))
cmlenz@67:         if self.locale is None:
cmlenz@67:             headers.append(('PO-Revision-Date', 'YEAR-MO-DA HO:MI+ZONE'))
cmlenz@67:             headers.append(('Last-Translator', 'FULL NAME <EMAIL@ADDRESS>'))
cmlenz@67:             headers.append(('Language-Team', 'LANGUAGE <LL@li.org>'))
cmlenz@67:         else:
cmlenz@67:             headers.append(('PO-Revision-Date',
cmlenz@67:                             time.strftime('%Y-%m-%d %H:%M%z', self.revision_date)))
cmlenz@67:             headers.append(('Last-Translator', self.last_translator))
cmlenz@67:             headers.append(('Language-Team', '%s <LL@li.org>' % self.locale))
cmlenz@84:             headers.append(('Plural-Forms', self.plural_forms))
cmlenz@67:         headers.append(('MIME-Version', '1.0'))
cmlenz@68:         headers.append(('Content-Type',
cmlenz@68:                         'text/plain; charset=%s' % self.charset))
cmlenz@67:         headers.append(('Content-Transfer-Encoding', '8bit'))
cmlenz@67:         headers.append(('Generated-By', 'Babel %s' % VERSION))
cmlenz@67:         return headers
cmlenz@67:     headers = property(headers, doc="""\
cmlenz@67:     The MIME headers of the catalog, used for the special ``msgid ""`` entry.
cmlenz@67:     
cmlenz@67:     The behavior of this property changes slightly depending on whether a locale
cmlenz@67:     is set or not, the latter indicating that the catalog is actually a template
cmlenz@67:     for actual translations.
cmlenz@67:     
cmlenz@67:     Here's an example of the output for such a catalog template:
cmlenz@67:     
cmlenz@67:     >>> catalog = Catalog(project='Foobar', version='1.0',
cmlenz@67:     ...                   creation_date=datetime(1990, 4, 1, 15, 30))
cmlenz@67:     >>> for name, value in catalog.headers:
cmlenz@67:     ...     print '%s: %s' % (name, value)
cmlenz@67:     Project-Id-Version: Foobar 1.0
palgarvio@78:     Report-Msgid-Bugs-To: EMAIL@ADDRESS
cmlenz@67:     POT-Creation-Date: 1990-04-01 15:30+0000
cmlenz@67:     PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE
cmlenz@67:     Last-Translator: FULL NAME <EMAIL@ADDRESS>
cmlenz@67:     Language-Team: LANGUAGE <LL@li.org>
cmlenz@67:     MIME-Version: 1.0
cmlenz@67:     Content-Type: text/plain; charset=utf-8
cmlenz@67:     Content-Transfer-Encoding: 8bit
cmlenz@67:     Generated-By: Babel ...
cmlenz@67:     
cmlenz@67:     And here's an example of the output when the locale is set:
cmlenz@67:     
cmlenz@67:     >>> catalog = Catalog(locale='de_DE', project='Foobar', version='1.0',
cmlenz@67:     ...                   creation_date=datetime(1990, 4, 1, 15, 30),
cmlenz@67:     ...                   revision_date=datetime(1990, 8, 3, 12, 0),
cmlenz@67:     ...                   last_translator='John Doe <jd@example.com>')
cmlenz@67:     >>> for name, value in catalog.headers:
cmlenz@67:     ...     print '%s: %s' % (name, value)
cmlenz@67:     Project-Id-Version: Foobar 1.0
palgarvio@78:     Report-Msgid-Bugs-To: EMAIL@ADDRESS
cmlenz@67:     POT-Creation-Date: 1990-04-01 15:30+0000
cmlenz@67:     PO-Revision-Date: 1990-08-03 12:00+0000
cmlenz@67:     Last-Translator: John Doe <jd@example.com>
cmlenz@67:     Language-Team: de_DE <LL@li.org>
cmlenz@84:     Plural-Forms: nplurals=2; plural=(n != 1)
cmlenz@67:     MIME-Version: 1.0
cmlenz@67:     Content-Type: text/plain; charset=utf-8
cmlenz@67:     Content-Transfer-Encoding: 8bit
cmlenz@67:     Generated-By: Babel ...
cmlenz@67:     
cmlenz@67:     :type: `list`
cmlenz@67:     """)
cmlenz@67: 
cmlenz@68:     def num_plurals(self):
cmlenz@68:         num = 2
cmlenz@68:         if self.locale:
cmlenz@68:             if str(self.locale) in PLURALS:
cmlenz@68:                 num = PLURALS[str(self.locale)][0]
cmlenz@68:             elif self.locale.language in PLURALS:
cmlenz@68:                 num = PLURALS[self.locale.language][0]
cmlenz@68:         return num
cmlenz@84:     num_plurals = property(num_plurals, doc="""\
cmlenz@84:     The number of plurals used by the locale.
cmlenz@84:     """)
cmlenz@68: 
cmlenz@67:     def plural_forms(self):
cmlenz@67:         num, expr = ('INTEGER', 'EXPRESSION')
cmlenz@67:         if self.locale:
cmlenz@67:             if str(self.locale) in PLURALS:
cmlenz@67:                 num, expr = PLURALS[str(self.locale)]
cmlenz@67:             elif self.locale.language in PLURALS:
cmlenz@67:                 num, expr = PLURALS[self.locale.language]
cmlenz@67:         return 'nplurals=%s; plural=%s' % (num, expr)
cmlenz@67:     plural_forms = property(plural_forms, doc="""\
cmlenz@67:     Return the plural forms declaration for the locale.
cmlenz@67:     
cmlenz@67:     >>> Catalog(locale='en_US').plural_forms
cmlenz@67:     'nplurals=2; plural=(n != 1)'
cmlenz@67:     >>> Catalog(locale='pt_BR').plural_forms
cmlenz@67:     'nplurals=2; plural=(n > 1)'
cmlenz@67:     
cmlenz@67:     :type: `str`
cmlenz@67:     """)
cmlenz@67: 
cmlenz@67:     def __contains__(self, id):
cmlenz@67:         """Return whether the catalog has a message with the specified ID."""
cmlenz@69:         return self._key_for(id) in self._messages
cmlenz@69: 
cmlenz@69:     def __len__(self):
cmlenz@84:         """The number of messages in the catalog.
cmlenz@84:         
cmlenz@84:         This does not include the special ``msgid ""`` entry.
cmlenz@84:         """
cmlenz@69:         return len(self._messages)
cmlenz@56: 
cmlenz@56:     def __iter__(self):
cmlenz@64:         """Iterates through all the entries in the catalog, in the order they
cmlenz@64:         were added, yielding a `Message` object for every entry.
cmlenz@64:         
cmlenz@64:         :rtype: ``iterator``
cmlenz@64:         """
cmlenz@67:         buf = []
cmlenz@67:         for name, value in self.headers:
cmlenz@67:             buf.append('%s: %s' % (name, value))
cmlenz@67:         yield Message('', '\n'.join(buf), flags=set(['fuzzy']))
cmlenz@69:         for key in self._messages:
cmlenz@69:             yield self._messages[key]
cmlenz@56: 
cmlenz@56:     def __repr__(self):
cmlenz@64:         locale = ''
cmlenz@64:         if self.locale:
cmlenz@64:             locale = ' %s' % self.locale
cmlenz@64:         return '<%s %r%s>' % (type(self).__name__, self.domain, locale)
cmlenz@56: 
cmlenz@56:     def __delitem__(self, id):
cmlenz@64:         """Delete the message with the specified ID."""
cmlenz@69:         key = self._key_for(id)
cmlenz@69:         if key in self._messages:
cmlenz@69:             del self._messages[key]
cmlenz@56: 
cmlenz@56:     def __getitem__(self, id):
cmlenz@64:         """Return the message with the specified ID.
cmlenz@64:         
cmlenz@64:         :param id: the message ID
cmlenz@64:         :return: the message with the specified ID, or `None` if no such message
cmlenz@64:                  is in the catalog
cmlenz@67:         :rtype: `Message`
cmlenz@64:         """
cmlenz@69:         return self._messages.get(self._key_for(id))
cmlenz@56: 
cmlenz@56:     def __setitem__(self, id, message):
cmlenz@64:         """Add or update the message with the specified ID.
cmlenz@64:         
cmlenz@64:         >>> catalog = Catalog()
cmlenz@64:         >>> catalog[u'foo'] = Message(u'foo')
cmlenz@64:         >>> catalog[u'foo']
cmlenz@64:         <Message u'foo'>
cmlenz@64:         
cmlenz@64:         If a message with that ID is already in the catalog, it is updated
cmlenz@64:         to include the locations and flags of the new message.
cmlenz@64:         
cmlenz@64:         >>> catalog = Catalog()
cmlenz@64:         >>> catalog[u'foo'] = Message(u'foo', locations=[('main.py', 1)])
cmlenz@64:         >>> catalog[u'foo'].locations
cmlenz@64:         [('main.py', 1)]
cmlenz@64:         >>> catalog[u'foo'] = Message(u'foo', locations=[('utils.py', 5)])
cmlenz@64:         >>> catalog[u'foo'].locations
cmlenz@64:         [('main.py', 1), ('utils.py', 5)]
cmlenz@64:         
cmlenz@64:         :param id: the message ID
cmlenz@64:         :param message: the `Message` object
cmlenz@64:         """
cmlenz@56:         assert isinstance(message, Message), 'expected a Message object'
cmlenz@69:         key = self._key_for(id)
cmlenz@69:         current = self._messages.get(key)
cmlenz@56:         if current:
cmlenz@69:             if message.pluralizable and not current.pluralizable:
cmlenz@69:                 # The new message adds pluralization
cmlenz@69:                 current.id = message.id
cmlenz@70:                 current.string = message.string
cmlenz@56:             current.locations.extend(message.locations)
cmlenz@56:             current.flags |= message.flags
cmlenz@56:             message = current
cmlenz@56:         else:
cmlenz@56:             if isinstance(id, (list, tuple)):
cmlenz@68:                 assert isinstance(message.string, (list, tuple))
cmlenz@69:             self._messages[key] = message
cmlenz@56: 
cmlenz@84:     def add(self, id, string=None, locations=(), flags=(), comments=()):
cmlenz@64:         """Add or update the message with the specified ID.
cmlenz@64:         
cmlenz@64:         >>> catalog = Catalog()
cmlenz@64:         >>> catalog.add(u'foo')
cmlenz@64:         >>> catalog[u'foo']
cmlenz@64:         <Message u'foo'>
cmlenz@64:         
cmlenz@64:         This method simply constructs a `Message` object with the given
cmlenz@64:         arguments and invokes `__setitem__` with that object.
cmlenz@64:         
cmlenz@64:         :param id: the message ID, or a ``(singular, plural)`` tuple for
cmlenz@64:                    pluralizable messages
cmlenz@64:         :param string: the translated message string, or a
cmlenz@64:                        ``(singular, plural)`` tuple for pluralizable messages
cmlenz@64:         :param locations: a sequence of ``(filenname, lineno)`` tuples
cmlenz@64:         :param flags: a set or sequence of flags
cmlenz@84:         :param comments: a list of translator comments
cmlenz@64:         """
palgarvio@80:         self[id] = Message(id, string, list(locations), flags, comments)
cmlenz@69: 
cmlenz@69:     def _key_for(self, id):
cmlenz@69:         """The key for a message is just the singular ID even for pluralizable
cmlenz@69:         messages.
cmlenz@69:         """
cmlenz@69:         key = id
cmlenz@69:         if isinstance(key, (list, tuple)):
cmlenz@69:             key = id[0]
cmlenz@69:         return key