Object database for Python

Newt DB, the amphibious database

Work with your data as ordinary Python objects in memory. PostgreSQL keeps it safe, indexes it, and answers search queries from inside or outside your application.

Read the getting started guide See how the pieces fit

pip install newt.db

Python objects PostgreSQL storage JSON search ZODB transactions

Why Newt DB

The ease of objects, the reach of SQL

Newt DB sits between your Python code and PostgreSQL. Object access stays natural, while every saved object also becomes queryable data.

Objects in memory

Read and change ordinary Python objects. Data move in and out of memory as needed, so a database can grow well beyond the size of a single Python process.

Indexed in PostgreSQL

Every object state is stored as JSON beside the native pickle. PostgreSQL indexes that JSON, so search runs close to the data and stays fast.

Objects or raw rows

Ask for application objects, or ask for plain data. Newt DB returns whichever form suits the job, including counts, joins and reports from other tools.

Mature foundations

ZODB supplies transparent persistence. RelStorage moves it into PostgreSQL. Newt DB adds the JSON layer and the search API without hiding either project.

How it works

Two mature projects, one database

Newt DB does not replace ZODB or RelStorage. It adds a searchable JSON view of the same records, then gives you helpers for querying it.

Your Python code Ordinary objects, containers and transactions

ZODB and RelStorage Object persistence with a PostgreSQL back end

PostgreSQL JSON data, indexes, full text search and raw SQL

Because the JSON view is lossy, Newt DB keeps the native pickle as the source of truth for loading objects. The JSON copy is read only, and that is what makes it safe to index and query from anywhere.

Read the architecture page How JSON is produced

Quick start

From pip to a committed object

You need a PostgreSQL server, version 9.5 or later. A local install is enough to begin; a hosted server is a sensible choice for production.

Install the package

Newt DB is published on PyPI and installs with pip.

pip install newt.db
Open a connection

The default database is fine for a first run. A connection string works when you need a dedicated database and user.
```
import newt.db

connection = newt.db.connection('')
root = connection.root
```
Add and commit an object

Objects are added to the root, or to a container below it. Nothing is written to PostgreSQL until you commit.
```
root.first = newt.db.Object(name='My first object')
root.first.child = newt.db.Object(name='First child')
connection.commit()
```

Continue with the full guide

Documentation

Explore the manual

Start with the guide, then move into a topic or jump straight to the API.

See the full contents

01 Getting startedConnections, objects, collections, search and raw queries. 02 ArchitectureZODB, RelStorage and the JSON layer in plain terms. 03 Fine printPersistence limits, object model notes and transactions. 04 TopicsConfiguration, administration, serialization and operations. 05 Asynchronous updaterBuild JSON data in the background with newt updater. 06 Following changesStream committed changes into other systems. 07 API referenceEvery function, method and class with its arguments. 08 IndexFind a name quickly across the whole manual.

Build with the project

Read the source, ask a question on the mailing list, or report a bug. The community page has every route into the project.

Open community resources