#sql-database #data-fusion #table #data-source #duck-db #provider #sqlite

datafusion-table-providers

Extend the capabilities of DataFusion to support additional data sources via implementations of the TableProvider trait

5 releases

0.2.3 Oct 23, 2024
0.2.2 Oct 11, 2024
0.2.1 Oct 11, 2024
0.2.0 Oct 10, 2024
0.1.0 Sep 6, 2024

#224 in Database interfaces

Download history 77/week @ 2024-08-31 44/week @ 2024-09-07 9/week @ 2024-09-14 22/week @ 2024-09-21 12/week @ 2024-09-28 222/week @ 2024-10-05 537/week @ 2024-10-12 1044/week @ 2024-10-19 1382/week @ 2024-10-26 1231/week @ 2024-11-02

4,346 downloads per month

Apache-2.0

500KB
11K SLoC

DataFusion Table Providers

Note: This is not an official Apache Software Foundation project.

The goal of this repo is to extend the capabilities of DataFusion to support additional data sources via implementations of the TableProvider trait.

Many of the table providers in this repo are for querying data from other database systems. Those providers also integrate with the datafusion-federation crate to allow for more efficient query execution, such as pushing down joins between multiple tables from the same database system, or efficiently implementing TopK style queries (SELECT * FROM table ORDER BY foo LIMIT 10).

To use these table providers with efficient federation push-down, add the datafusion-federation crate and create a DataFusion SessionContext using the Federation optimizer rule and query planner with:

use datafusion::prelude::SessionContext;

let state = datafusion_federation::default_session_state();
let ctx = SessionContext::with_state(state);

// Register the specific table providers into ctx
// queries will now automatically be federated

Table Providers

  • PostgreSQL
  • MySQL
  • SQLite
  • DuckDB
  • Flight SQL

Examples

Run the included examples to see how to use the table providers:

DuckDB

# Read from a table in a DuckDB file
cargo run --example duckdb --features duckdb
# Create an external table backed by DuckDB directly in DataFusion
cargo run --example duckdb_external_table --features duckdb
# Use the result of a DuckDB function call as the source of a table
cargo run --example duckdb_function --features duckdb

SQLite

cargo run --example sqlite --features sqlite

Postgres

In order to run the Postgres example, you need to have a Postgres server running. You can use the following command to start a Postgres server in a Docker container the example can use:

docker run --name postgres -e POSTGRES_PASSWORD=password -e POSTGRES_DB=postgres_db -p 5432:5432 -d postgres:16-alpine
# Wait for the Postgres server to start
sleep 30

# Create a table in the Postgres server and insert some data
docker exec -i postgres psql -U postgres -d postgres_db <<EOF
CREATE TABLE companies (
   id INT PRIMARY KEY,
  name VARCHAR(100)
);

INSERT INTO companies (id, name) VALUES (1, 'Acme Corporation');
EOF
cargo run --example postgres --features postgres

MySQL

In order to run the MySQL example, you need to have a MySQL server running. You can use the following command to start a MySQL server in a Docker container the example can use:

docker run --name mysql -e MYSQL_ROOT_PASSWORD=password -e MYSQL_DATABASE=mysql_db -p 3306:3306 -d mysql:9.0
# Wait for the MySQL server to start
sleep 30

# Create a table in the MySQL server and insert some data
docker exec -i mysql mysql -uroot -ppassword mysql_db <<EOF
CREATE TABLE companies (
   id INT PRIMARY KEY,
  name VARCHAR(100)
);

INSERT INTO companies (id, name) VALUES (1, 'Acme Corporation');
EOF
cargo run --example mysql --features mysql

Flight SQL

brew install roapi
# or
#cargo install --locked --git https://github.com/roapi/roapi --branch main --bins roapi
roapi -t taxi=https://d37ci6vzurychx.cloudfront.net/trip-data/yellow_tripdata_2024-01.parquet &

cargo run --example flight-sql --features flight

Dependencies

~58–82MB
~1.5M SLoC