PO-to-JSON Distutils/Setuptools Command Integration

po2json.py

""" :mod:`po2json`
    ~~~~~~~~~~~~~~

Convert PO files to JSON-serialized files.

By Hyunjun Kim <[email protected]> for StyleShare.

Skeleton source code from `babel.messages.frontend.compile_catalog` that does compiling PO to MO files.

"""
try:
    import omnijson as json
except ImportError:
    import json
import os

from babel.messages.pofile import read_po
from distutils import log
from distutils.cmd import Command
from distutils.errors import DistutilsOptionError


class po2json(Command):
    """Catalog to-json conversion command for use in ``setup.py`` scripts. """

    description = 'convert message catalogs to JSON-serialized format files'
    user_options = [
       ('domain=', 'D',
         "domain of PO file (default 'messages')"),
        ('input-dir=', None,
         'path to base directory containing the catalogs'),
        ('input-file=', 'i',
         'name of the input file'),
        ('output-dir=', None,
         'path to base directory for creating output files'),
        ('output-file=', 'o',
         "name of the output file (default "
         "'<output_dir>/<locale>/LC_MESSAGES/<domain>.json')"),
        ('locale=', 'l',
         'locale of the catalog to compile'),
        ('statistics', None,
         'print statistics about translations') 
    ]
    boolean_options = ['statistics']

    def initialize_options(self):
        self.domain = 'messages'
        self.input_dir = None
        self.input_file = None
        self.output_dir = None
        self.output_file = None
        self.locale = None
        self.statistics = False 

    def finalize_options(self):
        if not self.input_file and not self.input_dir:
            raise DistutilsOptionError('you must specify either the input file '
                                       'or the base directory')
        if not self.output_file and not self.output_dir:
            raise DistutilsOptionError('you must specify either the output file '
                                       'or the base directory') 

    def run(self):
        po_files = []
        json_files = []
        if not self.input_file:
            if self.locale:
                po_files.append((self.locale,
                                 os.path.join(self.input_dir, 'LC_MESSAGES',
                                              self.domain + '.po')))
                json_files.append((self.locale,
                                   os.path.join(self.output_dir,
                                                self.locale,
                                                'LC_MESSAGES',
                                                self.domain + '.json')))
            else:
                for locale in os.listdir(self.input_dir):
                    po_file = os.path.join(self.input_dir, locale,
                                           'LC_MESSAGES',
                                           self.domain + '.po')
                    if os.path.exists(po_file):
                        po_files.append((locale, po_file))
                        json_files.append(os.path.join(self.output_dir,
                                                       locale,
                                                       'LC_MESSAGES',
                                                       self.domain + '.json'))
        else:
            po_files.append((self.locale, self.input_file))
            if self.output_file:
                json_files.append(self.output_file)
            else:
                json_files.append(os.path.join(self.output_dir,
                                               self.locale,
                                               'LC_MESSAGES',
                                               self.domain + '.json'))
        if not po_files:
            raise DistutilsOptionError('no message catalogs found')

        for idx, (locale, po_file) in enumerate(po_files):
            json_file = json_files[idx]
            with open(po_file, 'r') as infile:
                catalog = read_po(infile, locale)

            if self.statistics:
                translated = 0
                for message in list(catalog)[1:]:
                    if message.string:
                        translated += 1
                percentage = 0
                if len(catalog):
                    percentage = translated * 100 // len(catalog)
                log.info("%d of %d messages (%d%%) translated in %r",
                         translated, len(catalog), percentage, po_file)

            for message, errors in catalog.check():
                for error in errors:
                    log.error('error: %s:%d: %s', po_file, message.lineno,
                              error)

            log.info('compiling catalog %r to %r', po_file, json_file)

            messages = {}
            for message in catalog:
                msgid = message.id
                if isinstance(msgid, (list, tuple)):
                    msgid = msgid[0]
                messages[msgid] = message.string

            json_compiled = json.dumps({
                'messages': messages,
                'plural_expr': catalog.plural_expr,
                'locale': str(locale),
                'domain': self.domain,
            })

            if not os.path.exists(os.path.dirname(json_file)):
                os.makedirs(os.path.dirname(json_file))
            with open(json_file, 'wb') as outfile:
                outfile.write(json_compiled)