marsha 0.0.2

No project description provided

# Marsha

[](https://gitlab.com/deckar01/marsha/commits/master)
[](https://gitlab.com/deckar01/marsha/commits/master)

A serialization and validation library for python 3 based on the typing module.

## Status

### :microscope: Experimental

Don't put this code on servers. You can, but you probably shouldn't.

## Documentation

### :book: [https://deckar01.gitlab.io/marsha](https://deckar01.gitlab.io/marsha)

## Features

### :guardsman: Strictness

When using python's builtin types for schema annotations, the data **must** be
instances of the specified types or the load operation will fail with an exception.
Unexpected and missing data will error by default.

### :spy: Clarity

Validation errors mirror the structure of the data, provide insight into the values
that caused the error, and present meaningful information to help resolve the problem.

### :runner: Speed

The default functionality is fast and lean. The performance degrades gracefully as
complex functionality is explicitly opted into by the developer.

### :cartwheel: Flexibility

Custom data types are created by defining formatter classes, then transfored into
types compatible with the builtin typing module.

### :dancer: Expressiveness

Python's builtin typing generics can be leveraged to declare complex type unions that
know how to automatically handle multiple types using a syntax that is explicit and concise.

## Usage

_Need to create database model instances from user input and expose them in serialized form?_

Creating a schema that is separate from the model is a useful way to declare a "view" and
allows defining multiple schemas to control how data is exposed in different contexts.

```py
from my_app.database_models import Artist, Album

import marsha
from typing import List

@marsha.schema(Artist)
class ArtistSchema:
name: str

@marsha.schema(Album)
class AlbumSchema:
title: str
artists: List[Artist]

# dict -> Album
album = marsha.load(album_data, AlbumSchema)
# Do stuff
album.save()
# Album -> dict
response_data = marsha.dump(album)
```

_Just need to validate some data and convert it to the correct type?_

You can automatically register the type annotations of any model as a schema by
adding the `schema` decorator.

```py
import marsha
from typing import List

@marsha.schema()
class Artist(dict):
name: str

@marsha.schema()
class Album(dict):
title: str
artists: List[Artist]

# dict -> dict
album = marsha.load(album_data, AlbumSchema)
# Do stuff
first_artist = album['artists'][0]
```

## Development

### Dependencies

```sh
# python >= 3.6
pip install -r dev-requirements.txt
```

### Testing

```sh
# Quick testing.
flake8
pytest
# Ensure all statements and branches are covered with tests.
pytest --cov=marsha --cov-branch --cov-report html --cov-report term
# Ensure documentations changes render correctly.
make html
```