DefPixel

Define your pixels. Design your voxels.
Define your connections, Design your models, and Dive into your data in Python.

DefPixel is a Python library for defining data source connections, models, and access. Use a provided data access implementation or build your own, and work with data directly or through standard dataclasses.

Connect · model · use

• Define your connections: storage-specific configuration and data access capabilities.
• Design your models: storage specifications and validation rules attached to standard dataclasses.
• Dive into your data: direct access without models, or storage and retrieval through models.

Configuration and Connection provide connection foundations. Database, KeyValue, Document, and Index Protocols describe capabilities. Modeling approaches and support vary by implementation. Validators also provide independent value validation and conversion.

Requirements and installation

Python 3.10+ is required. The base pip install DefPixel installs only the core, without implementation code.

Sqlite

Select the Sqlite implementation with the Sqlite extra. It uses Python's standard-library sqlite3. Model-based usage requires Sqlite 3.35+ as used by Python.

pip install "DefPixel[Sqlite]"

PostgreSQL

Install the PostgreSQL extra to include the implementation and PostgreSQL driver. A running PostgreSQL 15–18 server is required. This installation excludes Sqlite.

pip install "DefPixel[PostgreSQL]"

Use pip install "DefPixel[Sqlite,PostgreSQL]" for both implementations. Each extra selects a separate, same-version distribution; existing import paths are unchanged. Before migrating an older integrated installation, uninstall DefPixel and any implementation distributions, then reinstall with the required extras. Sqlite is no longer included in the base installation. Omitting an extra later does not remove an already installed implementation.

MariaDB development status

The current source provides MariaDB connections, direct SQL, sessions, native pools, dataclass mapping, types, DDL, model CRUD, SELECT-backed views, query composition, scalar/aggregate/date-time/JSON/spatial/vector functions and JSON_TABLE sources through the MariaDB implementation selection. VECTOR types/indexes and vector functions require 11.7+; MBRCoveredBy requires 12.0+; JSON_TABLE requires 10.6+. Insert returns the first RETURNING row; Update/Delete return None. Final integration and regression validation remain in progress; YEAR=0 support is deferred. No release containing this feature has been published. Source development uses python -m pip install -e . --config-settings EXTRAS=MariaDB; include other implementation names in EXTRAS to keep them selected.

Usage at a glance

Database

Sqlite

Execute SQL directly for dictionary results or map standard dataclasses to tables and retrieve objects. This complete example requires Sqlite 3.35+ as used by Python.

from dataclasses import dataclass
from DefPixel.DataAccessObject.Implements.Sqlite import (
  Configuration, Connection, column, table,
)
from DefPixel.DataAccessObject.Implements.Sqlite.Constraints import PrimaryKey
from DefPixel.DataAccessObject.Implements.Sqlite.Executors import Create, Insert, Select
from DefPixel.DataAccessObject.Implements.Sqlite.Types import Integer, Text


@table(name='users', constraints=(PrimaryKey('id'),))
@dataclass
class User:
  id: int = column(Integer())
  name: str = column(Text())


connection = Connection(Configuration(':memory:'))
connection.connect()
try:
  with connection.session() as session:
    session.run(Create(User))
    session.run(Insert(User).values(id=1, name='Alice'))

    rows = session.execute(
      'SELECT id, name FROM users WHERE id = ?', 1,
    ).rows()
    print(rows)

    users = session.run(Select(User), fetch=User)
    print(users)
finally:
  connection.close()

Output:

[{'id': 1, 'name': 'Alice'}]
[User(id=1, name='Alice')]

Sessions commit on normal completion and roll back on exceptions. Direct SQL does not require models. Models remain ordinary Python objects; attribute assignment does not automatically persist changes. No custom Model base class or Service/Repository/DTO structure is required.

PostgreSQL

PostgreSQL also supports direct SQL alongside dataclass mappings. Install the PostgreSQL extra above and adjust the connection settings for your server. Run this example in a development database: it creates and drops the defpixel_readme_users table.

from dataclasses import dataclass
from DefPixel.DataAccessObject.Implements.PostgreSQL import (
  Configuration, Connection, Create, Drop, Insert, PrimaryKey, Select, column, table,
)
from DefPixel.DataAccessObject.Implements.PostgreSQL.Types import Integer, Text


@table('defpixel_readme_users', constraints=PrimaryKey('defpixel_readme_users_pk', 'id'))
@dataclass
class User:
  id: int = column(Integer())
  name: str = column(Text())


connection = Connection(Configuration(
  host='127.0.0.1', port=5432, database='defpixel',
  username='defpixel', password='change-me-local-only',
))
connection.connect()
try:
  with connection.session() as session:
    session.run(Create(User))
    session.run(Insert(User).values(id=1, name='Alice'))

    rows = session.execute(
      'SELECT id, name FROM defpixel_readme_users WHERE id = %s', 1,
    ).rows()
    print(rows)

    users = session.run(Select(User), fetch=User)
    print(users)
    session.run(Drop(User))
finally:
  connection.close()

Output:

[{'id': 1, 'name': 'Alice'}]
[User(id=1, name='Alice')]

Direct SQL binds positional arguments with %s and keyword arguments with %(name)s. PostgreSQL constraints take a name as their first argument.

Supported implementations

• Sqlite: direct SQL, dataclass mappings, CRUD, tables, views, constraints, indexes, joins, aggregates, and transactions. Sqlite modeling requires Sqlite 3.35+.
• PostgreSQL: direct SQL, dataclass mappings, CRUD, schemas, sequences, tables, views, constraints, indexes, joins, aggregate/window/JSON functions, nested transactions, and connection pools. Requires PostgreSQL 15–18 and the PostgreSQL extra.

Build your own

Implement the capability Protocols for the store you need, without inheriting a provided implementation or adapting every store to SQL.

The source repository includes step-by-step implementation development guides and Sqlite and PostgreSQL user manuals.