cmlenz@160: # -*- coding: utf-8 -*-
cmlenz@160: #
jruigrok@530: # Copyright (C) 2007-2011 Edgewall Software
cmlenz@160: # All rights reserved.
cmlenz@160: #
cmlenz@160: # This software is licensed as described in the file COPYING, which
cmlenz@160: # you should have received as part of this distribution. The terms
cmlenz@160: # are also available at http://babel.edgewall.org/wiki/License.
cmlenz@160: #
cmlenz@160: # This software consists of voluntary contributions made by many
cmlenz@160: # individuals. For the exact contribution history, see the revision
cmlenz@160: # history and logs, available at http://babel.edgewall.org/log/.
cmlenz@160: 
cmlenz@160: """Writing of files in the ``gettext`` MO (machine object) format.
cmlenz@160: 
cmlenz@234: :since: version 0.9
cmlenz@160: :see: `The Format of MO Files
cmlenz@160:        <http://www.gnu.org/software/gettext/manual/gettext.html#MO-Files>`_
cmlenz@160: """
cmlenz@160: 
cmlenz@160: import array
cmlenz@160: import struct
cmlenz@160: 
cmlenz@334: from babel.messages.catalog import Catalog, Message
cmlenz@334: 
cmlenz@334: __all__ = ['read_mo', 'write_mo']
cmlenz@161: __docformat__ = 'restructuredtext en'
cmlenz@161: 
cmlenz@334: 
cmlenz@334: LE_MAGIC = 0x950412deL
cmlenz@334: BE_MAGIC = 0xde120495L
cmlenz@334: 
cmlenz@334: def read_mo(fileobj):
cmlenz@334:     """Read a binary MO file from the given file-like object and return a
cmlenz@334:     corresponding `Catalog` object.
cmlenz@334:     
cmlenz@334:     :param fileobj: the file-like object to read the MO file from
cmlenz@334:     :return: a catalog object representing the parsed MO file
cmlenz@334:     :rtype: `Catalog`
cmlenz@334:     
cmlenz@334:     :note: The implementation of this function is heavily based on the
cmlenz@334:            ``GNUTranslations._parse`` method of the ``gettext`` module in the
cmlenz@334:            standard library.
cmlenz@334:     """
cmlenz@334:     catalog = Catalog()
cmlenz@334:     headers = {}
cmlenz@334: 
cmlenz@334:     filename = getattr(fileobj, 'name', '')
cmlenz@334: 
cmlenz@334:     buf = fileobj.read()
cmlenz@334:     buflen = len(buf)
cmlenz@335:     unpack = struct.unpack
cmlenz@334: 
cmlenz@334:     # Parse the .mo file header, which consists of 5 little endian 32
cmlenz@334:     # bit words.
cmlenz@334:     magic = unpack('<I', buf[:4])[0] # Are we big endian or little endian?
cmlenz@334:     if magic == LE_MAGIC:
cmlenz@335:         version, msgcount, origidx, transidx = unpack('<4I', buf[4:20])
cmlenz@334:         ii = '<II'
cmlenz@334:     elif magic == BE_MAGIC:
cmlenz@335:         version, msgcount, origidx, transidx = unpack('>4I', buf[4:20])
cmlenz@334:         ii = '>II'
cmlenz@334:     else:
cmlenz@334:         raise IOError(0, 'Bad magic number', filename)
cmlenz@334: 
cmlenz@334:     # Now put all messages from the .mo file buffer into the catalog
cmlenz@334:     # dictionary
cmlenz@334:     for i in xrange(0, msgcount):
cmlenz@335:         mlen, moff = unpack(ii, buf[origidx:origidx + 8])
cmlenz@334:         mend = moff + mlen
cmlenz@334:         tlen, toff = unpack(ii, buf[transidx:transidx + 8])
cmlenz@334:         tend = toff + tlen
cmlenz@334:         if mend < buflen and tend < buflen:
cmlenz@334:             msg = buf[moff:mend]
cmlenz@334:             tmsg = buf[toff:tend]
cmlenz@334:         else:
cmlenz@334:             raise IOError(0, 'File is corrupt', filename)
cmlenz@334: 
cmlenz@334:         # See if we're looking at GNU .mo conventions for metadata
cmlenz@334:         if mlen == 0:
cmlenz@334:             # Catalog description
cmlenz@334:             lastkey = key = None
cmlenz@334:             for item in tmsg.splitlines():
cmlenz@334:                 item = item.strip()
cmlenz@334:                 if not item:
cmlenz@334:                     continue
cmlenz@334:                 if ':' in item:
cmlenz@334:                     key, value = item.split(':', 1)
cmlenz@334:                     lastkey = key = key.strip().lower()
cmlenz@335:                     headers[key] = value.strip()
cmlenz@334:                 elif lastkey:
cmlenz@335:                     headers[lastkey] += '\n' + item
cmlenz@334: 
cmlenz@335:         if '\x04' in msg: # context
cmlenz@335:             ctxt, msg = msg.split('\x04')
cmlenz@335:         else:
cmlenz@335:             ctxt = None
cmlenz@335: 
cmlenz@335:         if '\x00' in msg: # plural forms
cmlenz@334:             msg = msg.split('\x00')
cmlenz@334:             tmsg = tmsg.split('\x00')
cmlenz@335:             if catalog.charset:
cmlenz@335:                 msg = [x.decode(catalog.charset) for x in msg]
cmlenz@335:                 tmsg = [x.decode(catalog.charset) for x in tmsg]
cmlenz@334:         else:
cmlenz@335:             if catalog.charset:
cmlenz@335:                 msg = msg.decode(catalog.charset)
cmlenz@335:                 tmsg = tmsg.decode(catalog.charset)
cmlenz@335:         catalog[msg] = Message(msg, tmsg, context=ctxt)
cmlenz@334: 
cmlenz@334:         # advance to next entry in the seek tables
cmlenz@335:         origidx += 8
cmlenz@334:         transidx += 8
cmlenz@334: 
cmlenz@334:     catalog.mime_headers = headers.items()
cmlenz@334:     return catalog
cmlenz@334: 
cmlenz@160: def write_mo(fileobj, catalog, use_fuzzy=False):
cmlenz@160:     """Write a catalog to the specified file-like object using the GNU MO file
cmlenz@160:     format.
cmlenz@160:     
cmlenz@160:     >>> from babel.messages import Catalog
cmlenz@160:     >>> from gettext import GNUTranslations
cmlenz@160:     >>> from StringIO import StringIO
cmlenz@160:     
cmlenz@160:     >>> catalog = Catalog(locale='en_US')
cmlenz@160:     >>> catalog.add('foo', 'Voh')
fschwarz@544:     <Message ...>
cmlenz@160:     >>> catalog.add((u'bar', u'baz'), (u'Bahr', u'Batz'))
fschwarz@544:     <Message ...>
cmlenz@160:     >>> catalog.add('fuz', 'Futz', flags=['fuzzy'])
fschwarz@544:     <Message ...>
palgarvio@172:     >>> catalog.add('Fizz', '')
fschwarz@544:     <Message ...>
palgarvio@174:     >>> catalog.add(('Fuzz', 'Fuzzes'), ('', ''))
fschwarz@544:     <Message ...>
cmlenz@160:     >>> buf = StringIO()
cmlenz@160:     
cmlenz@160:     >>> write_mo(buf, catalog)
cmlenz@160:     >>> buf.seek(0)
cmlenz@160:     >>> translations = GNUTranslations(fp=buf)
cmlenz@160:     >>> translations.ugettext('foo')
cmlenz@160:     u'Voh'
cmlenz@160:     >>> translations.ungettext('bar', 'baz', 1)
cmlenz@160:     u'Bahr'
cmlenz@160:     >>> translations.ungettext('bar', 'baz', 2)
cmlenz@160:     u'Batz'
cmlenz@160:     >>> translations.ugettext('fuz')
cmlenz@160:     u'fuz'
palgarvio@172:     >>> translations.ugettext('Fizz')
palgarvio@172:     u'Fizz'
palgarvio@174:     >>> translations.ugettext('Fuzz')
palgarvio@174:     u'Fuzz'
palgarvio@174:     >>> translations.ugettext('Fuzzes')
palgarvio@174:     u'Fuzzes'
cmlenz@160:     
cmlenz@160:     :param fileobj: the file-like object to write to
cmlenz@160:     :param catalog: the `Catalog` instance
cmlenz@160:     :param use_fuzzy: whether translations marked as "fuzzy" should be included
cmlenz@160:                       in the output
cmlenz@160:     """
cmlenz@160:     messages = list(catalog)
cmlenz@160:     if not use_fuzzy:
cmlenz@160:         messages[1:] = [m for m in messages[1:] if not m.fuzzy]
pjenvey@248:     messages.sort()
cmlenz@160: 
cmlenz@160:     ids = strs = ''
cmlenz@160:     offsets = []
cmlenz@160: 
cmlenz@160:     for message in messages:
cmlenz@160:         # For each string, we need size and file offset.  Each string is NUL
cmlenz@160:         # terminated; the NUL does not count into the size.
cmlenz@160:         if message.pluralizable:
cmlenz@160:             msgid = '\x00'.join([
cmlenz@160:                 msgid.encode(catalog.charset) for msgid in message.id
cmlenz@160:             ])
palgarvio@173:             msgstrs = []
palgarvio@173:             for idx, string in enumerate(message.string):
palgarvio@173:                 if not string:
cmlenz@330:                     msgstrs.append(message.id[min(int(idx), 1)])
palgarvio@173:                 else:
palgarvio@173:                     msgstrs.append(string)
cmlenz@160:             msgstr = '\x00'.join([
palgarvio@173:                 msgstr.encode(catalog.charset) for msgstr in msgstrs
cmlenz@160:             ])
cmlenz@160:         else:
cmlenz@160:             msgid = message.id.encode(catalog.charset)
palgarvio@172:             if not message.string:
palgarvio@172:                 msgstr = message.id.encode(catalog.charset)
palgarvio@172:             else:
palgarvio@172:                 msgstr = message.string.encode(catalog.charset)
cmlenz@335:         if message.context:
palgarvio@420:             msgid = '\x04'.join([message.context.encode(catalog.charset),
palgarvio@420:                                  msgid])
cmlenz@160:         offsets.append((len(ids), len(msgid), len(strs), len(msgstr)))
cmlenz@160:         ids += msgid + '\x00'
cmlenz@160:         strs += msgstr + '\x00'
cmlenz@160: 
cmlenz@160:     # The header is 7 32-bit unsigned integers.  We don't use hash tables, so
cmlenz@160:     # the keys start right after the index tables.
cmlenz@160:     keystart = 7 * 4 + 16 * len(messages)
cmlenz@160:     valuestart = keystart + len(ids)
cmlenz@160: 
cmlenz@160:     # The string table first has the list of keys, then the list of values.
cmlenz@160:     # Each entry has first the size of the string, then the file offset.
cmlenz@160:     koffsets = []
cmlenz@160:     voffsets = []
cmlenz@160:     for o1, l1, o2, l2 in offsets:
cmlenz@160:         koffsets += [l1, o1 + keystart]
cmlenz@160:         voffsets += [l2, o2 + valuestart]
cmlenz@160:     offsets = koffsets + voffsets
cmlenz@160: 
cmlenz@160:     fileobj.write(struct.pack('Iiiiiii',
cmlenz@334:         LE_MAGIC,                   # magic
cmlenz@160:         0,                          # version
cmlenz@160:         len(messages),              # number of entries
cmlenz@160:         7 * 4,                      # start of key index
cmlenz@160:         7 * 4 + len(messages) * 8,  # start of value index
cmlenz@160:         0, 0                        # size and offset of hash table
cmlenz@160:     ) + array.array("i", offsets).tostring() + ids + strs)