django-loadjson
===============
|Codeship Status for onebit0fme/django-loadjson|
Django management command to load json data of any shape by defining
"manifest" that describes mapping instructions.
Requirements
------------
Python 2.7, 3.4, 3.5 Django >= 1.7
Installation
------------
``pip install django-loadjson``
Quick setup
-----------
Include loadjson in settings.py
::
INSTALLED_APPS = [
... ,
'loadjson',
...
]
Define loadjson data directories - a place loadjson will look for data
(.json) files.
::
LOAD_JSON = {
'DATA_DIRS': [os.path.join(BASE_DIR, 'dumpdata')],
}
Dump all your .json files inside the specified directory(ies).
Each ``*.json`` file must also have corresponding ``*.manifest.json``
that describes how the data should be handled. For "manifest" reference,
see ``Manifest`` section.
Once the data and manifest is in place, run
``python manage.py loadjson <data_name>``, where <data\_name>
corresponds to the filename of the data (with or without the .json
part).
Note, loadjson will look in all specified directories for the requested
<data\_name> and will use the first file it will find. Same goes for the
manifest file. Data file and manifest do not have to live in the same
directory, but both must be in a path of defined "DATA\_DIRS".
It is also possible to handle data and manifest in other ways by
providing custom "finder\_classes". See ``Advanced Usage`` for
instructions.
Manifest
--------
- model (required) - a string in format "."
- mapping (required) - an object representing the model - data mapping.
- parsers (optional) - an object the describes how to parse the data.
Supported types:
- ``string`` - convert to string
- ``integer`` - convert to integer
- ``boolean`` - convert to boolean. Optional, define
``"invert": true`` to invert.
- ``datetime`` - parse datetime string
- ``relative_key`` - lookup the relative key by the field value in
another dataset. Required:
- ``data_name`` - is a data name where related object should be
looked up.
- ``rk_lookup`` - the field for the lookup in the related dataset
using one key. Ex., "email".
- ``lookup`` - overwrite the related data lookup field(s). By
default will use the lookup fields that are defined in data
manifest.
- ``many`` (optional) - for many-to-many relationship.
- ``relative_object`` - get/create/update a related object.
``data_name`` or ``manifest`` required
- ``data_name`` - for regular manifest lookup
- ``manifest`` - define manifest right on the spot
- ``many`` - one related object or many-to-many relationship.
- lookup (required for updates are relative lookups) - a string or a
list of fields to use when looking up an object. Ex., ``id``,
``email``, ``["username", "email"]``. Note, lookup fields are used to
lookup an object. The result of a lookup must be one object, so
choose accordingly.
- nullable (optional) - a list of fields that are nullable.
- m2m\_fields (optional) - a list of many-to-many fields. Note, if data
contains many-to-many field, this field should include it, or
alternatively use custom adaptors to handle it, otherwise Django will
throw an error when saving.
Example:
~~~~~~~~
users.json
::
[
{
"username": "PinkRabbit",
"email": "pink.rabbit@example.com"
"member_since": "",
"active": true,
"not_superuser": true,
"preferences": {
"email_notifications: true,
"number_of_friends": "2"
},
"friends": ["blue.hippo@example.com", "funny.tiger@example.com"],
"badges": [
{
"name": "For being pink",
"date_awarded": ""
},
{
"name": "For being there for friends",
"date_awarded": ""
}
]
},
...
]
users.manifest.json
::
{
"model": "account.User",
"lookup": "pk",
"mapping": {
"username": "username",
"email": "email",
"date_joined": "member_since",
"active": "active",
"is_superuser": "not_superuser",
"email_notifications": "preferences.email_notifications",
"friends_number": "preferences.number_of_friends",
"friends": "friends",
"badges": "badges"
},
"parsers": {
"is_superuser: {
"type": "boolean",
"invert": true
},
"date_joined": {
"type": "datetime"
},
"friends": {
"type": "relative_key",
"data_name": "users",
"rk_lookup": "email",
"lookup": "username",
"many": true
},
"badges": {
"type": "relative_object",
"many": true,
"manifest" {
"model": "account.Badge"
"mapping": {
"name": "name",
"awarded": "date_awarded"
},
"parsers": {
"awarded": {
"type": "datetime"
}
},
"lookup": "name"
}
}
}
}
LOAD\_JSON settings
-------------------
All loadjson-related settings should go into ``LOAD_JSON`` var inside
project's settings.py.
Possible LOAD\_JSON settings:
- ``DATA_DIRS`` (required) - a list of absolute paths to use by the
default finder to find data and manifest.
- ``ADAPTOR_CLASSES`` (optional) - a list of classes that are used to
customize the import. Extend ``loadjson.adaptors.BaseAdaptor`` to
define your custom adaptors. Defaults to None.
- ``MODEL_HANDLER`` (optional) - a string that references a class that
handles model lookups. Defaults to ``loadjson.adaptors.ModelHandler``
- ``FINDER_CLASSES`` (optional) - a list of classes that are used to
find data. By default loadjson uses
``loadjson.finders.DefaultDataFinder`` that uses defined
``DATA_DIRS`` to find data and manifest.
- ``MANIFEST_DEFAULTS`` (optional) - a dictionary of default manifest
values to use.
Defining ADAPTOR\_CLASSES
~~~~~~~~~~~~~~~~~~~~~~~~~
``ADAPTOR_CLASSES`` are used to further massage the data before saving.
To define Adaptor Class, extend ``loadjson.adaptors.BaseAdaptor`` and
overwrite ``adapt`` and/or ``adapt_post_save`` methods like so:
::
from loadjson.adaptors import BaseAdaptor
class MyCusomAdaptor(BaseAdaptor):
"""
Available attributes:
- model - model class
- app_model - a string in format <app_label>.<model_name>
- manifest - a dictionary with defined manifest values
"""
def adapt(self, data):
"""
Method provides a base hook to provide additional data, set defaults,
or modify the data before saving.
Usage: what returned gets saved
"""
return data
def adapt_post_save(self, obj, data, m2m_data):
"""
In some cases (like saving many-to-many relations) data might require
some additional tweaks. That is done here.
Note: Many-to-Many objects are attached by default, however in case if many-to-many relationship
is done through a custom model, this method provides a hook to process such customization.
"""
pass
Don't forget to include your custom adaptors in
LOAD\_JSON.ADAPTOR\_CLASSES.
Advanced usage
--------------
MODEL\_HANDLER
~~~~~~~~~~~~~~
Define custom ModelHandle if you want to use custom model manager, or
your same method has any customization. ``get``, ``create``,
``get_or_create``, ``update_or_create`` methods are available for
overwrite.
FINDER\_CLASSES
~~~~~~~~~~~~~~~
It is also possible to overwrite or extend the default finder classes.
The purpose of the finder class is to lookup the data by data\_name.
``DefaultDataFinder`` is a default finder, but can be extended by
defining ``FINDER_CLASSES`` in LOAD\_JSON settings. Note, if you define
this setting and not include ``DefaultDataFinder``, then it will not be
used, unless you add it explicitly as well.
It is possible to define as many finders as you will, but be aware that
only the first occurrence found will be used.
.. |Codeship Status for onebit0fme/django-loadjson| image:: https://codeship.com/projects/6e3309b0-d07e-0133-fd7c-1e2b23aa142a/status?branch=master
:target: https://codeship.com/projects/141380