django-linguist is a Django application for flexible model translations.
Here a few principles that define this application in comparaison to others applications:
- Translations are stored in single one table and you can also use a different one per model
- No "one i18n table per model", say "goodbye" to nightmares :)
- No more painful migrations
- Not tied to model class names, you are free to use your own identifiers
- No ORM query hacks, it does not patch anything and it will be easier for you to upgrade your Django
- No magic, it uses metaclasses and mixins and everything is explicit
- Dead simple to plug in an existing project
- Django admin ready
If you are looking for a "one-i18n-table-per-model" way, django-parler is an awesome alternative.
$ pip install django-linguist
In your settings.py
, add linguist
to INSTALLED_APPS
:
INSTALLED_APPS = (
# Your other apps here
'linguist',
)
Then synchronize database:
# >= Django 1.7
$ python manage.py migrate linguist
# < Django 1.7
$ python manage.py syncdb
That's all.
In three steps:
- Add
linguist.metaclasses.ModelMeta
to your model as metaclass - Add
linguist.mixins.ManagerMixin
to your model manager - Add
linguist
settings in your model's Meta
Don't worry, it's fairly simple:
from django.db import models
from django.utils.six import with_metaclass
from django.utils.translation import ugettext_lazy as _
from linguist.metaclasses import ModelMeta as LinguistMeta
from linguist.mixins import ManagerMixin as LinguistManagerMixin
class PostManager(LinguistManagerMixin, models.Manager):
pass
class Post(with_metaclass(LinguistMeta, models.Model)):
title = models.CharField(max_length=255)
body = models.TextField()
created_at = models.DateTimeField(auto_now_add=True)
objects = PostManager()
class Meta:
verbose_name = _('post')
verbose_name_plural = _('posts')
linguist = {
'identifier': 'can-be-anything-you-want',
'fields': ('title', 'body'),
'default_language': 'fr',
}
The linguist
meta requires:
identifier
: a unique identifier for your model (can be anything you want)fields
: list or tuple of model fields to translate
And optionally requires:
default_language
: the default language to usedefault_language_field
: the field that contains the default language to use (see below)decider
: the translation model to use instead of the default one (see below)
That's all. You're ready.
Sometimes, you need to define default language at instance level. Linguist
supports this feature via the default_language_field
option. Add a field
in your model that will store the default language then simply give the field
name to Linguist.
Let's take an example:
from django.db import models
from django.utils.six import with_metaclass
from django.utils.translation import ugettext_lazy as _
from linguist.metaclasses import ModelMeta as LinguistMeta
from linguist.mixins import ManagerMixin as LinguistManagerMixin
class PostManager(LinguistManagerMixin, models.Manager):
pass
class Post(with_metaclass(LinguistMeta, models.Model)):
title = models.CharField(max_length=255)
body = models.TextField()
created_at = models.DateTimeField(auto_now_add=True)
lang = models.CharField(max_length=5, default='en')
objects = PostManager()
class Meta:
verbose_name = _('post')
verbose_name_plural = _('posts')
linguist = {
'identifier': 'can-be-anything-you-want',
'fields': ('title', 'body'),
'default_language': 'en',
'default_language_field': 'lang',
}
By default, Linguist stores translations into linguist.models.Translation
table. So in a single one table. If you need to use another table for a specific
model, Linguist provides a way to override this behavior: use deciders.
That's really easy to implement.
You can do it in three steps:
- Create a model that inherits from
linguist.models.base.Translation
- Don't forget to define it as concrete (
abstract = False
in Meta) - Give this model to Linguist meta
decider
option
This example will show you the light:
from django.db import models
from django.utils.six import with_metaclass
from django.utils.translation import ugettext_lazy as _
from linguist.metaclasses import ModelMeta as LinguistMeta
from linguist.mixins import ManagerMixin as LinguistManagerMixin
from linguist.models.base import Translation
# Our Post model decider
class PostTranslation(Translation):
class Meta:
abstract = False
class PostManager(LinguistManagerMixin, models.Manager):
pass
class Post(with_metaclass(LinguistMeta, models.Model)):
title = models.CharField(max_length=255)
body = models.TextField()
created_at = models.DateTimeField(auto_now_add=True)
objects = PostManager()
class Meta:
verbose_name = _('post')
verbose_name_plural = _('posts')
linguist = {
'identifier': 'can-be-anything-you-want',
'fields': ('title', 'body'),
'default_language': 'fr',
'decider': PostTranslation,
}
Simply use linguist.admin.TranslatableModelAdmin
class:
from django.contrib import admin
from linguist.admin import TranslatableModelAdmin
from .models import Post
class PostAdmin(TranslatableModelAdmin):
list_display = ('title', 'body', 'created_at')
admin.site.register(Post, PostAdmin)
Bonus! You can display instance's languages in list_display
via the
languages_column
property provided by the admin class:
from django.contrib import admin
from linguist.admin import TranslatableModelAdmin
from .models import Post
class PostAdmin(TranslatableModelAdmin):
list_display = ('title', 'body', 'languages_column', 'created_at')
admin.site.register(Post, PostAdmin)
Linguist adds virtual language fields to your models. For the example above, if
we have en
, fr
and it
in settings.LANGUAGES
, it
dynamically adds the following fields in Post
model:
Post.title_en
Post.title_fr
Post.title_it
Post.body_en
Post.body_fr
Post.body_it
These fields are virtuals. They don't exist in Post
table. There are
wrappers around linguist.Translation
model. All translations will be stored
in this table.
When you set/get post.title
, Linguist will use the current active language
and will set/get the correct field for this language. For example, if your
default language is English (en
), then Post.title
will refer to post.title_en
.
The ModelMixin
enhance your model with the following properties and methods:
instance.linguist_identifier
(read-only property)- Your model identifier defined in the related translation class.
Shortcut pointing on
instance._linguist.identifier
. instance.default_language
(read-write property)- The default language to use.
Shortcut pointing on
instance._linguist.default_language
. instance.translatable_fields
(read-only property)- Translatable fields defined in the related translation class.
Shorcut pointing on
instance._linguist.fields
. instance.available_languages
(read-only property)- Available languages for this instance (content translated in these languages).
instance.cached_translations_count
(read-only property)- Returns the number of cached translations. Each time you set a new language
and set content on translatable fields, a cache is created for each language
and field. It will be used to create
Translation
objets at instance saving. instance.active_language()
- Set the current active language for the instance.
instance.clear_translations_cache()
- Remove all cached translations. Be aware, any content you set will be dropped. So no translation will be created/updated at saving.
# Let's create a new Post
>>> post = Post()
# Set English content
>>> post.activate_language('en')
>>> post.title = 'Hello'
# Now set French content
>>> post.activate_language('fr')
>>> post.title = 'Bonjour'
# Be sure everything works as expected for English
>>> post.activate_language('en')
>>> post.title
Hello
# And now for French
>>> post.activate_language('fr')
>>> post.title
Bonjour
# Sweet! Save translations!
>>> post.save()
To improve performances, you can preload/prefetch translations.
For a queryset (your queryset must inherit from Linguist manager/queryset):
>>> Post.objects.with_translations()
For a list of objects (all your objects must inherit from Linguist model):
>>> from linguist.helpers import prefetch_translations
>>> posts = list(Post.objects.all())
>>> prefetch_translations(posts)
For an instance (it must inherit from Linguist model):
>>> post = Post.objects.first()
>>> post.prefetch_translations()
All translations will be cached in instances. Database won't be hit anymore.
This preloading system takes three parameters:
field_names
: list of translatable field names to filter onlanguages
: list of languages to filter onpopulate_missing
: boolean if you want to populate cache for missing translations (defaults toTrue
)chunks_length
: chunk limit for SELECT IN ids for translations
For example, we only want to prefetch post titles in English without populating missing translations with an empty string:
>>> Post.objects.with_translations(field_names=['title'], languages=['en'], populate_missing=False)
It works the same for:
- QuerySet
with_translations()
- Helper
prefetch_translations()
- Instance method
prefetch_translations()
What does "populating missing translations" mean?
Simple. By default, when you prefetch translations, instances cache will be populated
with empty strings for all supported languages (see settings
). For example, if
you have en
, fr
and it
as supported languages and only have English
translations, if you try to access other languages, an empty string will be returned
without any database hit:
>>> Post.objects.with_translations()
>>> post.title_fr # no database hit here because
''
Now, if you explicitly set populate_missing
to False
, if a translation
is not found, it will be fetched from database.
>>> Post.objects.with_translations(populate_missing=False)
>>> post.title_fr # database hit here
''
# Don't have pip?
$ sudo easy_install pip
# Don't already have virtualenv?
$ sudo pip install virtualenv
# Clone and install dependencies
$ git clone https://github.com/ulule/django-linguist.git
$ cd django-linguist
$ make devenv
# Enable virtual environment.
$ source .venv/bin/activate
# Launch tests
$ make test
# Launch example project
$ make serve
- python 2.7: Django 1.8, 1.9
- Python 3.4: Django 1.8, 1.9