django-model-import 0.8.0

A Django library for importing CSVs and other structured data quickly using Django's ModelForm for validation and deserialisation into an instance.

django-model-import

Django Model Import is a light weight CSV importer built for speed.

It uses a standard Django ModelForm to parse each row, giving you a familiar API to work with for data validation and model instantiation. In most cases, if you already have a ModelForm for the ContentType you are importing you do not need to create an import specific form.

To present feedback to the end-user running the import you can easily generate a preview of the imported data by toggling the commit parameter.

It also provides some import optimized fields for ForeignKey's, allowing preloading all possible values, or caching each lookup as it occurs, or looking up a model where multiple fields are needed to uniquely identify a resource.

Installation

poetry add django-model-import

Quickstart

import djangomodelimport

class BookImporter(djangomodelimport.ImporterModelForm):
    name = forms.CharField()
    author = CachedChoiceField(queryset=Author.objects.all(), to_field='name')

    class Meta:
        model = Book
        fields = (
            'name',
            'author',
        )

with default_storage.open('books.csv', 'rb') as fh:
    data = fh.read().decode("utf-8")

# Use tablib
parser = djangomodelimport.TablibCSVImportParser(BookImporter)
headers, rows = parser.parse(data)

# Process
importer = djangomodelimport.ModelImporter(BookImporter)
preview = importer.process(headers, rows, commit=False)
errors = preview.get_errors()

if errors:
    print(errors)

importresult = importer.process(headers, rows, commit=True)
for result in importresult.get_results():
    print(result.instance)

Composite key lookups

Often a relationship cannot be referenced via a single unique string. For this we can use a CachedChoiceField with a CompositeLookupWidget. The widget looks for the values under the type and variant columns in the source CSV, and does a unique lookup with the field names specified in to_field, e.g. queryset.get(type__name=type, name=variant).

The results of each get are cached internally for the remainder of the import minimising any database access.

class AssetImporter(ImporterModelForm):
    site = djangomodelimport.CachedChoiceField(queryset=Site.objects.active(), to_field='ref')
    type = djangomodelimport.CachedChoiceField(queryset=AssetType.objects.filter(is_active=True), to_field='name')
    type_variant = djangomodelimport.CachedChoiceField(
        queryset=InspectionItemTypeVariant.objects.filter(is_active=True),
        required=False,
        widget=djangomodelimport.CompositeLookupWidget(source=('type', 'variant')),
        to_field=('type__name', 'name'),
    )
    contractor = djangomodelimport.CachedChoiceField(queryset=Contractor.objects.active(), to_field='name')

Flat related fields

Often you'll have a OneToOneField or just a ForeignKey to another model, but you want to be able to create/update that other model via this one. You can flatten all of the related model's fields onto this importer using FlatRelatedField.

class ClientImporter(ImporterModelForm):
    primary_contact = FlatRelatedField(
        queryset=ContactDetails.objects.all(),
        fields={
            'contact_name': {'to_field': 'name', 'required': True},
            'email': {'to_field': 'email'},
            'email_cc': {'to_field': 'email_cc'},
            'mobile': {'to_field': 'mobile'},
            'phone_bh': {'to_field': 'phone_bh'},
            'phone_ah': {'to_field': 'phone_ah'},
            'fax': {'to_field': 'fax'},
        },
    )

    class Meta:
        model = Client
        fields = (
            'name',
            'ref',
            'is_active',
            'account',

            'primary_contact',
        )

Tests

Run tests with python example/manage.py test testapp