django-modelish

Because programmers are lazy...

Modelish is a quick attempt to save a few keystrokes. Many data modeling tools are a GUI nightmare, with dropdowns and lots of buttons. But sometimes, especially when you are sketching something out, writing out full Django models can be a bit tedious.

Modelish will parse a file with a simplified model declaration in a flavor of YAML and generate Python code for model definition.

Normally I would say source code generation is a "Bad Idea", but really even in Python, Django models are a mostly declarative syntax

Quickstart

Install django-modelish:

pip install django-modelish

You can then start banging out models like this

Poll:
    doc: This is the poll model
    question:
        type: char
        max_length: 200
    pub_date:
        type: datetime
        args: date published

Choice:
    poll:
        type: fk
        args: [Poll]
    choice_text-char:
        max_length: 200
    votes:
        type: int

You then just use the CLI tool:

modelish path-to-file.yml

This will generate code like this:

class Poll(models.Model):
    """This is the poll model"""

    pub_date = models.DateTimeField(
        'date published')
    question = models.CharField(
        max_length=200,
        blank=True)

class Choice(models.Model):
    choice_text = models.CharField(
        max_length=200,
        blank=True)
    poll = models.ForeignKey(
        'Poll')
    votes = models.IntegerField()

Syntax

The syntax is pretty simple:

<ModelName>:
    [doc: docstring]
    <fieldname>[-type-alias]:
        [type: type-alias],
        [args: arglist],
        kwarg: value

The field type is represented by a shortened 'type' alias so 'char' becomes 'models.CharField' etc (see grammarish below).

The type can be specified one of two ways, either following the field name with a hyphen, or explicitly declared as a type in the field definition.

Positional args to the model specified as an explicit list in square brackets, or as a comma delimited string.

The remaining block of the field definition consists of kwarg/value pairs.

Working with grammarish

Modelish works with a grammar composed of type aliases and defaults. The standard types are:

auto: AutoField
bigint: BigIntegerField
bool: BooleanField
char: CharField
date: DateField
datetime: DateTimeField
decimal: DecimalField
email: EmailField
file: FileField
float: FloatField
image: ImageField
int: IntegerField
ip: IPAddressField
gip: GenericIPAddressField
nbool: NullBooleanField
pint: PositiveIntegerField
psint: PositiveSmallIntegerField
slug: SlugField
sint: SmallIntegerField
text: TextField
time: TimeField
url: URLField
fk: ForeignKey
m2m: ManyToManyField
timestamp: DateTimeField

For each type - a set of default kwargs is defined in the grammar as defaults:

bool:
    default: true
char:
    max_length: 100
    blank: true
nbool:
    null: true
timestamp:
    auto-now: true

This default grammar can be replace, or enhanced by passing your own yaml files to the command. Use --grammar to replace the default grammer, and use --extra-grammar to merge in and update the default grammer with your own additions or changes.

That's it - this isn't meant to be any sort of full featured model builder or data modeler, it is just a simple DSLish bootstrap tool to give you a models.py starting point with a little less typing.

What about editor completion snippets? Yes - that is an option, but it tends to only lets you set what kwargs you want in a all or nothing way. I'm open to learning about better in editor techniques for this use however.