Django QBE 2

A reusable Django app for building authorized reports without writing SQL. Choose columns, add typed conditions, follow model relationships, preview paginated results and export a CSV. Supports Python 3.13–3.14 and Django 6.0–6.1.

Experimental alpha: this is the initial implementation, not a production release or a drop-in replacement for the original package. The current release candidate is 0.1.0a3; check PyPI for its availability. Inspired by versae/qbe, rebuilt around Django's ORM, explicit permissions and typed query definitions.

Try the demo

From a checkout containing this implementation:

uv sync --python 3.14 --frozen --group dev
uv run python demo/manage.py migrate
uv run python demo/manage.py seed_demo
uv run python demo/manage.py changepassword qbe2_demo_alice
uv run python demo/manage.py runserver 127.0.0.1:8000

Open the report builder and log in as qbe2_demo_alice with the password you chose. The seed creates six synthetic customers and 24 orders for two users. Each user can view only their own records. It never sets a shared password or overwrites an existing non-demo account. An optional superuser created with createsuperuser can inspect all demo records.

Install in another project

In your project's virtual environment, install the explicit alpha version after checking its availability on PyPI:

python -m pip install django-qbe2==0.1.0a3

To try the checkout before publication, build and install its wheel:

uv build
python -m pip install /absolute/path/to/dist/django_qbe2-0.1.0a3-py3-none-any.whl

Add the app and mount its namespaced URLs alongside Django Admin:

# Keep your existing Django apps, including admin, auth, sessions and messages.
INSTALLED_APPS += ["django_qbe2"]

from django.urls import include, path

# Keep your existing admin URL and other routes.
urlpatterns += [path("qbe/", include("django_qbe2.urls"))]

Keep standard authentication, sessions and CSRF middleware, APP_DIRS template discovery and the admin context processors. Serve the packaged assets with your usual Django static-file setup. The reporting app itself has no models to migrate. The catalog is empty until you register reports.

Register reports and their policies

Create qbe.py in your own installed app. Modules are discovered at startup, without fetching report data. Register related targets as well as root models:

from django_qbe2.registry import Report, register
from .models import Customer, Order


@register
class CustomerReport(Report):
    model = Customer
    fields = ("name",)

    def get_queryset(self, request):
        return Customer.objects.filter(owner=request.user)


@register
class OrderReport(Report):
    model = Order
    fields = ("number", "customer__name", "total", "status")
    field_labels = {"customer__name": "Customer", "total": "Total amount"}

    def get_queryset(self, request):
        return Order.objects.filter(owner=request.user)

Users need active staff status and Django view permissions for the root and any related model used. Staff status alone does not grant report access. Every report must provide an explicit row policy; Django model permissions do not imply row authorization. The same policies apply to columns, filters, ordering, HTML and CSV.

Django model choices require no QBE-specific configuration. Register the field, and equality filters offer its options automatically. The complete documentation example defines an Order status using TextChoices. Adapt the example's fields and ownership policy to your own models; owner is not a required field name.

What is included

• Text, integer, decimal, boolean, date, datetime and UUID fields.
• Generic model-choice selectors and readable HTML labels; CSV keeps stored codes.
• AND conditions, ascending/descending order and stable pagination.
• One forward ForeignKey/OneToOne hop with related-row authorization.
• Bounded CSV streaming, exact decimals, Unicode and safe text formula handling.
• Django Admin styling, light/dark modes and responsive, keyboard-accessible controls.

Saved queries, aggregates, OR/NOT groups, ManyToMany, inverse relationships, arbitrary joins, multi-database selection, Excel and the visual diagram are later work. CSV uses a synchronous streaming iterator intended for WSGI deployment. See the public documentation for installation, row policies, settings, API reference and troubleshooting. Sources are plain English reStructuredText under docs/source, readable by developers and agents. Contributor instructions live in AGENTS.md.

Build the documentation

uv sync --python 3.14 --frozen --group dev --group docs
uv run --group docs sphinx-build -n -W -b html docs/source docs/_build/html
uv run --group docs sphinx-build -n -W -b text docs/source docs/_build/text
python -m http.server 8001 --bind 127.0.0.1 --directory docs/_build/html

Open http://127.0.0.1:8001/. Sphinx is a documentation dependency only. CI checks HTML and plain-text builds with warnings treated as errors. A manual GitHub Pages workflow is prepared; the documentation site is not yet published.

License and attribution

MIT, copyright Carlos A. Rocha. The project acknowledges Javier de la Rosa and the contributors to versae/qbe as its inspiration. The current code is a new implementation; preserve upstream notices if you reuse upstream code in a future change.