TuringDB

TuringDB is a columnar graph database with git-like versioning. Its query language is a Cypher dialect — largely compatible with standard Cypher but with TuringDB-specific extensions. The Python SDK query() method returns results as pandas DataFrames.

Setup

Default: connect to a running TuringDB server over HTTP. The server listens on http://localhost:6666 by default; start one with the turingdb CLI if it isn't already running (see startup.md).

from turingdb import TuringDB

client = TuringDB(host="http://localhost:6666")   # HTTP/JSON client (the default backend)
client.create_graph("my_graph")  # create new graph
# or
client.load_graph("my_graph")    # load an existing on-disk graph by name
client.set_graph("my_graph")     # make it active

TuringDB(...) supports three backends via type: "json" (HTTP, the default — pass host="http://localhost:6666"), "native" (binary protocol over HTTP — host=/port=), and "embedded" (in-process, no server — pass data_dir=). Use the server (json) by default; it's required for the browser visualizer (-ui), list_available_graphs() to discover on-disk graphs, concurrent multi-process access, and S3 transfers. Reach for "embedded" only when you want a self-contained, in-process engine with no server to manage. See startup.md for connecting/starting the server and introspection.md for the full backend reference.