django-transmeta 0.6.1

Creating translatable models

Look at this model:

class Book(models.Model):
    title = models.CharField(max_length=200)
    description = models.TextField()
    body = models.TextField(default='')
    price = models.FloatField()

Suppose you want to make description and body translatable. The resulting model after using transmeta is:

from transmeta import TransMeta

class Book(models.Model):
    __metaclass__ = TransMeta

    title = models.CharField(max_length=200)
    description = models.TextField()
    body = models.TextField(default='')
    price = models.FloatField()

    class Meta:
        translate = ('description', 'body', )

Make sure you have set the default and available languages in your settings.py:

LANGUAGE_CODE = 'es'

ugettext = lambda s: s # dummy ugettext function, as django's docs say

LANGUAGES = (
    ('es', ugettext('Spanish')),
    ('en', ugettext('English')),
)

This is the SQL generated with the ./manage.py sqlall command:

BEGIN;
CREATE TABLE "fooapp_book" (
    "id" serial NOT NULL PRIMARY KEY,
    "title" varchar(200) NOT NULL,
    "description_en" text,
    "description_es" text NOT NULL,
    "body_es" text NOT NULL,
    "body_en" text NOT NULL,
    "price" double precision NOT NULL
)
;
COMMIT;

Notes: * transmeta creates one column for each language. Don’t worry about needing new languages in the future, transmeta solves this problem for you. * If one field is null=False and doesn’t have a default value, transmeta will create only one NOT NULL field, for the default language. Fields for other secondary languages will be nullable. Also, the primary language will be required in the admin app, while the other fields will be optional (with blank=True). This was done so because the normal approach for content translation is first add content in the main language and later have translators translate into other languages. * You can use ./manage.py syncdb to create database schema.

Playing in the python shell

transmeta creates one field for every available language for every translatable field defined in a model. Field names are suffixed with language short codes, e.g.: description_es, description_en, and so on. In addition it creates a field_name getter to retrieve the field value in the active language.

Let’s play a bit in a python shell to best understand how this works:

>>> from fooapp.models import Book
>>> b = Book.objects.create(description_es=u'mi descripcion', description_en=u'my description')
>>> b.description
u'my description'
>>> from django.utils.translation import activate
>>> activate('es')
>>> b.description
u'mi descripcion'
>>> b.description_en
u'my description'

Adding new languages

If you need to add new languages to the existing ones you only need to change your settings.py and ask transmeta to sync the DB again. For example, to add French to our project, you need to add it to LANGUAGES in settings.py:

LANGUAGES = (
    ('es', ugettext('Spanish')),
    ('en', ugettext('English')),
    ('fr', ugettext('French')),
)

And execute a special sync_transmeta_db command:

$ ./manage.py sync_transmeta_db

Missing languages in "description" field from "fooapp.book" model: fr

SQL to synchronize "fooapp.book" schema:
   ALTER TABLE "fooapp_book" ADD COLUMN "description_fr" text

Are you sure that you want to execute the previous SQL: (y/n) [n]: y
Executing SQL... Done

Missing languages in "body" field from "fooapp.book" model: fr

SQL to synchronize "fooapp.book" schema:
   ALTER TABLE "fooapp_book" ADD COLUMN "body_fr" text

Are you sure that you want to execute the previous SQL: (y/n) [n]: y
Executing SQL... Done

And done!

Adding new translatable fields

Now imagine that, after several months using this web app (with many books created), you need to make book price translatable (for example because book price depends on currency).

To achieve this, first add price to the model’s translatable fields list:

class Book(models.Model):
    ...
    price = models.FloatField()

    class Meta:
        translate = ('description', 'body', 'price', )

All that’s left now is calling the sync_transmeta_db command to update the DB schema:

$ ./manage.py sync_transmeta_db

Available languages:
    1. Spanish
    2. English
Choose a language in which to put current untranslated data.
What's the language of current data? (1-2): 1

Missing languages in "price" field from "fooapp.book" model: es, en

SQL to synchronize "fooapp.book" schema:
    ALTER TABLE "fooapp_book" ADD COLUMN "price_es" double precision
    UPDATE "fooapp_book" SET "price_es" = "price"
    ALTER TABLE "fooapp_book" ALTER COLUMN "price_es" SET NOT NULL
    ALTER TABLE "fooapp_book" ADD COLUMN "price_en" double precision
    ALTER TABLE "fooapp_book" DROP COLUMN "price"

Are you sure that you want to execute the previous SQL: (y/n) [n]: y
Executing SQL...Done

What the hell this command does?

sync_transmeta_db command not only creates new database columns for new translatable field… it copy data from old price field into one of languages, and that is why command ask you for destination language field for actual data.

Admin integration

transmeta transparently displays all translatable fields into the admin interface. This is easy because models have in fact many fields (one for each language).

Changing form fields in the admin is quite a common task, and transmeta includes the canonical_fieldname utility function to apply these changes for all language fields at once. It’s better explained with an example:

from transmeta import canonical_fieldname

class BookAdmin(admin.ModelAdmin):
    def formfield_for_dbfield(self, db_field, **kwargs):
        field = super(BookAdmin, self).formfield_for_dbfield(db_field, **kwargs)
        db_fieldname = canonical_fieldname(db_field)
        if db_fieldname == 'description':
            # this applies to all description_* fields
            field.widget = MyCustomWidget()
        elif field.name == 'body_es':
            # this applies only to body_es field
            field.widget = MyCustomWidget()
        return field

0.6.1 (2011-03-17)

• get_all_translatable_fields does not returned the correct tuple. Problems with inheritance.

0.6.0 (2011-02-24)

• Make compatible with Django 1.2 and 1.3 when using ugettext_lazy in models verbose_name, fixing a hidden bug also for Django 1.1