datafusion-dft

dft - Batteries included DataFusion

Overview

dft is a batteries-included suite of DataFusion applications that provides:

• Data Source Integration: Query files from S3, local filesystems, or HuggingFace datasets
• Table Format Support: Native support for Delta Lake and Vortex
• Extensibility: UDFs defined in WASM
• Helper Functions: Built-in functions for JSON and Parquet data processing

The project offers four complementary interfaces:

1. Text User Interface (TUI): An interactive SQL IDE with real-time query analysis, benchmarking, and catalog exploration (requires tui feature)
2. Command Line Interface (CLI): A scriptable engine for executing queries from files or command line
3. FlightSQL Server: A standards-compliant SQL interface for programmatic access (requires flightsql feature)
4. HTTP Server: A REST API for SQL queries and catalog exploration (requires http feature)

All interfaces share the same execution engine, allowing you to develop locally with the TUI and then seamlessly deploy with the server implementations.

dft builds upon datafusion-cli with enhanced interactivity, additional integrations, and ready-to-use server implementations.

User Guide

Installation

From crates.io (Recommended)

# Core CLI and server interfaces
cargo install datafusion-dft

# With TUI interface
cargo install datafusion-dft --features=tui

# For full functionality with all features (including TUI)
cargo install datafusion-dft --all-features

If you don't have Rust installed, follow the installation instructions.

Feature Flags

Common feature combinations:

…

See the Features documentation for all available features.

Note: The TUI (Text User Interface) is optional and requires the tui feature flag. The CLI, FlightSQL server, and HTTP server are always available.

Running the apps

…

Benchmarking

dft includes built-in benchmarking to measure query performance with detailed timing breakdowns:

# Serial benchmark (default) - measures query performance in isolation
dft -c "SELECT * FROM my_table" --bench

# Concurrent benchmark - measures throughput under load
dft -c "SELECT * FROM my_table" --bench --concurrent

# Custom iteration count
dft -c "SELECT * FROM my_table" --bench -n 100

# Save results to CSV for analysis
dft -c "SELECT * FROM my_table" --bench --save results.csv

# Compare serial vs concurrent performance
dft -c "SELECT * FROM my_table" --bench --save results.csv
dft -c "SELECT * FROM my_table" --bench --concurrent --save results.csv --append

Benchmark Output:

• Timing breakdown by phase: logical planning, physical planning, execution
• Statistics: min, max, mean, median for each phase
• Row counts validation across all runs
• CSV export with concurrency_mode column for result comparison

Serial vs Concurrent:

• Serial: Pure query execution time without contention (baseline performance)
• Concurrent: Throughput measurement with parallel execution (reveals bottlenecks and contention)
• Concurrent mode uses adaptive concurrency: min(iterations, CPU cores)

Setting Up Tables with DDL

dft can automatically load table definitions at startup, giving you a persistent "database-like" experience.

Using DDL Files

1. Create a DDL file (default: ~/.config/dft/ddl.sql)
2. Add your table and view definitions:

-- S3 data source (requires s3 feature)
CREATE EXTERNAL TABLE users 
STORED AS NDJSON 
LOCATION 's3://bucket/users';

-- Parquet files
CREATE EXTERNAL TABLE transactions 
STORED AS PARQUET 
LOCATION 's3://bucket/transactions';

-- Local files
CREATE EXTERNAL TABLE listings 
STORED AS PARQUET 
LOCATION 'file://folder/listings';

-- Create views from tables
CREATE VIEW users_listings AS 
SELECT * FROM users 
LEFT JOIN listings USING (user_id);

-- Delta Lake table (requires deltalake feature)
CREATE EXTERNAL TABLE delta_table 
STORED AS DELTATABLE 
LOCATION 's3://bucket/delta_table';

Loading DDL

• TUI (requires tui feature): DDL is automatically loaded at startup
• CLI: Add --run-ddl flag to execute DDL before your query
• Custom Path: Configure a custom DDL path in your config file
  toml复制

  [execution]
  ddl_path = "/path/to/my/ddl.sql"

Quick Reference