dj-labeler 1.1.0

A Django library to externalize translation strings from models and forms.

The most annoying thing about Django models is their verbosity when you want to do things right. As soon as you have an international audience, you’ll need to start marking strings for translation. Labeler was created to reduce the noise by externalizing a model’s labels, help texts and error messages. It even provides the same functionality for any Django form.

It’s expected to work with Django 1.8, 1.9 and up.

Installation

Labeler is available on Pypi as dj-labeler:

pip install dj-labeler

Example

Imagine our bookstore models look like this:

from django.db import models

class Author(models.Model):
    name = models.CharField(max_length=200)
    published = models.BooleanField(default=False)
    birthdate = models.DateField(blank=True, null=True)


class Book(models.Model):
    title = models.CharField(max_length=200)
    published_on = models.DateField(blank=True, null=True)
    isbn = models.CharField(max_length=50, unique=True)
    authors = models.ManyToManyField(Author)

Now you want to branch out into a Dutch-speaking market. Instead of relying on Django’s automagical label creation based on the field name, you’ll need to explicitly state your verbose name for each field and mark it as a translatable string. And to avoid any confusion for the people performing the Dutch translation, you want to provide as much context as possible, because an author’s name might not require the same label as the name of a category.

So you end up with this:

from django.db import models
from django.utils.translation import pgettext_lazy

class Author(models.Model):
    name = models.CharField(pgettext_lazy('author', 'name'), max_length=200)
    published = models.BooleanField(pgettext_lazy('author', 'published'), editable=False)
    birthdate = models.DateField(pgettext_lazy('author', 'birthdate'), blank=True, null=True)

    class Meta:
        verbose_name = pgettext_lazy('author model', 'Author')
        verbose_name_plural = pgettext_lazy('author model (plural)', 'Authors')


class Book(models.Model):
    title = models.CharField(pgettext_lazy('book', 'title'), max_length=200)
    published_on = models.DateField(pgettext_lazy('book (date)', 'published'), blank=True, null=True)
    isbn = models.CharField(pgettext_lazy('book', 'isbn'), max_length=50, unique=True)
    authors = models.ManyToManyField(Author, verbose_name=pgettext_lazy('book authors', 'authors'))

    class Meta:
        verbose_name = pgettext_lazy('author model', 'Book')
        verbose_name_plural = pgettext_lazy('author model (plural)', 'Books')

Now add in help text and you’ve got a lot of noise, making it hard to discern the attributes you as a programmer care about most when developing, like the maximum length and whether a field is optional or unique.

Labeler will enable apps to use translatable strings with less noise. Let’s move the strings to a separate file we’ll call i18n.py (but any name will do) and use Labeler’s ModelTranslations:

# i18n.py
from django.utils.translation import pgettext_lazy
from labeler import ModelTranslations

author = ModelTranslations(
    labels=dict(
        name=pgettext_lazy('author', 'name'),
        published=pgettext_lazy('author', 'published'),
        birthdate=pgettext_lazy('author', 'birthdate')
    )
    help_texts=dict(
        birthdate=pgettext_lazy('author', 'When was the author born?')
    ),
    name=pgettext_lazy('author model', 'Author'),
    name_plural=pgettext_lazy('author model (plural)', 'Authors')
)

book = ModelTranslations(
    labels=dict(
        title=pgettext_lazy('book', 'title'),
        published_on=pgettext_lazy('book (date)', 'published'),
        isbn=pgettext_lazy('book', 'isbn'),
        authors=pgettext_lazy('book authors', 'authors')
    ),
    help_texts=dict(
        isbn=pgettext_lazy('book', 'The ISBN will be validated against XYZ database')
    ),
    name=pgettext_lazy('author model', 'Book'),
    name_plural=pgettext_lazy('author model (plural)', 'Books')
)

That’s still a lot of noise, but at least we’ve got it isolated to a single file in our app. Now, because ModelTranslations is simply an extension of dict, you could start doing things like this:

# models.py
from django.db import models
from . import i18n

class Author(models.Model):
    # as above

    class Meta:
        verbose_name = i18n.author['name']
        verbose_name_plural = i18n.author['name_plural']

But that doesn’t cut down on the noise. Instead you should use the inject method/decorator of ModelTranslations (or FormTranslations when dealing with a form). This will make our models lean and mean:

# models.py
from django.db import models
from . import i18n

@i18n.author.inject
class Author(models.Model):
    name = models.CharField(max_length=200)
    published = models.BooleanField(default=False)
    birthdate = models.DateField(blank=True, null=True)


@i18n.book.inject
class Book(models.Model):
    title = models.CharField(max_length=200)
    published_on = models.DateField(blank=True, null=True, unique=True)
    isbn = models.CharField(max_length=50)
    authors = models.ManyToManyField(Author)

Spot the difference with our initial version? This version uses translatable strings simply by decorating our models with our ModelTranslations’ inject.

Usage

Translating models using ModelTranslations

ModelTranslations is a simple dict with some useful methods and properties added on top. Nothing is required, but if you specify labels, help_texts or error_messages, the keys of those dictionaries should refer to existing model fields.

ModelTranslations key	Type	Maps to	Attribute
labels	dict	field	verbose_name
help_texts	dict	field	help_text
error_messages	dict	field	Updates error_messages
name	str	Meta	verbose_name
name_plural	str	Meta	verbose_name_plural

Example:

from django.utils.translation import ugettext_lazy as _
from labeler import ModelTranslations

article = ModelTranslations(
    # verbose_name of the model's fields
    labels=dict(
        title=_('Title'),
        body=_('Body')
    ),
    # help_text of the model's fields
    help_texts=dict(
        title=_('No clickbait titles please!')
    ),
    # update to the listed fields' error_messages
    error_messages=dict(
        title=dict(
            unique=_('Title already exists')
        )
    ),
    # verbose_name of the model
    name=_('article'),
    # verbose_name_plural of the model
    name_plural=_('articles'),
    # Handy dict of error messages for this model, not field-specific
    errors=dict(
        too_clickbaity=_('Please review the article.')
    ),
    # Handy dict for other kinds of messages
    messages=dict(
        first_publication=_('Congratulations! Your first article has been published')
    ),
    # It's just a dict; add whatever you want
    something_else='abc',
    publication_state={
        'published': _('Published'),
        'draft': _('Draft'),
        'trashed': _('Trashed')
    }
)

When everything is good and ready to go, simply inject this into your model:

from . import i18n

@i18n.article.inject
class Article(models.Model):
    # Fields and stuff

The nested labels, error_messages, errors, messages, and help_texts dictionaries are also available as properties. This means custom validation might look like this:

def clean_fields(self, exclude=None):
    super(MyModel, self).clean_fields(exclude)
    if 'title' not in exclude and calculate_clickbait_level(self.title) > 50:
        raise ValidationError({'title': i18n.article.errors['too_clickbaity']})

If you’re dealing with lots of nested dicts, you can use the resolve method:

hard_way = i18n.article.get('errors', {}).get('fieldname', {}).get('invalid', {}).get('state')
easier_way = i18n.article.resolve('errors.fieldname.invalid.state')
easier_way == hard_way

Translating forms using FormTranslations

FormTranslations works exactly like ModelTranslations, but it also supports a nested dictionary empty_labels to override the default empty label on form fields.

FormTranslations key	Type	Maps to	Attribute
labels	dict	field	label
help_texts	dict	field	help_text
empty_labels	dict	field	empty_label
error_messages	dict	field	Updates error_messages

Usage:

# i18n.py
from django.utils.translation import ugettext_lazy as _
from labeler import FormTranslations

article_form = FormTranslations(
    labels=dict(
        title=_('Title'),
        body=_('Body'),
        published=_('When to publish this article'),
        author=_('Author'),
    ),
    help_texts=dict(
        title=_('Limit to 100 characters please'),
        body=_('Formatting is not supported')
    ),
    empty_labels=dict(
        author=_('Please select an author')
    ),
    error_messages=dict(
        title=dict(
            unique=_('That title has already been used. Be more original!')
        )
    )
)

# forms.py
from django import forms
from . import i18n
from .models import Article

@i18n.article_form.inject
class ArticleForm(forms.ModelForm):

    class Meta:
        model = Article
        fields = ('title', 'body', 'published', 'author')

That’s all there is to it.

Changelog

v1.0.1

• Fixes to code in the README and project information