django-psycopg-infinity 0.1.1

Installation

pip install django-psycopg-infinity

Configuration

Set the custom database backend in your Django settings:

DATABASES = {
    "default": {
        "ENGINE": "django_psycopg_infinity.backends.postgresql",
        # ... other settings
    },
}

Then use DateTimeInfinityField in your models:

from django.db import models
from django_psycopg_infinity.fields import DateTimeInfinityField

class MyModel(models.Model):
    valid_until = DateTimeInfinityField(null=True, blank=True)

Usage

from datetime import datetime

# Store positive infinity
obj.valid_until = datetime.max
obj.save()

# Or use the string representation
obj.valid_until = "infinity"
obj.save()

# Retrieve - the returned datetime matches the column type:
# aware (UTC) for `timestamp with time zone` columns,
# naive for `timestamp without time zone` columns.
obj.refresh_from_db()
assert obj.valid_until == datetime.max

Convenience constants

The package avoids calling timezone.get_default_timezone() at import time so it can be imported before Django is fully configured. If you need timezone-aware infinity sentinels (e.g. for model field defaults), derive them in your own module:

from django_psycopg_infinity.utils import INFINITY, NEGATIVE_INFINITY, make_aware

LOCAL_INFINITY = make_aware(INFINITY)
LOCAL_NEGATIVE_INFINITY = make_aware(NEGATIVE_INFINITY)

Then use them as defaults:

class MyModel(models.Model):
    end_time = DateTimeInfinityField(default=LOCAL_INFINITY)

For fields backed by timestamp without time zone columns, use the bare INFINITY / NEGATIVE_INFINITY constants instead — a tz-aware default would mismatch the naive values that the loader returns from those columns.

Development

Tests require a running PostgreSQL instance:

cp .env.example .env   # edit as needed
uv run runtests.py

Releasing

Releases are published to PyPI automatically when a version tag is pushed. The CI runs the full test suite before publishing.

Create a new release on GitHub with a v-prefixed tag (e.g. v0.1.0) and the workflow handles the rest.

Versioning is derived from git tags via hatch-vcs, so there are no version strings to update manually.

License

Apache-2.0