README

High-performance PHP adapter for DuckDB that executes SQL queries directly on massive datasets (CSV, JSON, Parquet) without preprocessing or imports. Features SQLite-compatible interface, FFI integration, and intelligent CSV parsing for zero-ETL analytics on files up to gigabytes in size.

Benchmark:

Tested with a large CSV dataset containing 1 million rows (150 MB) using PHP v8.3.

Note: No optimizations were applied. The benchmark includes database creation from compressed CSV and query execution.

Requirements:

• PHP exec function
• PHP SQLite3 extension
• PHP FFI extension (Used by Loader)

Install:

composer require jakiboy/pducky
vendor/bin/pducky install-binaries

Test:

git clone https://github.com/Jakiboy/pducky.git
cd pducky && composer dump-autoload
composer install-binaries
cd examples && bash generate.sh
bash test.sh

Examples:

Fetch single value:

$price = (new Pducky\Adapter('data.csv'))->import()->single(
	'SELECT `price` FROM `temp` WHERE `ean` = "4567890123456";'
); // 374.08$

Fetch rows:

$rows = (new Pducky\Adapter('data.csv'))->import()->query(
	'SELECT * FROM `temp` LIMIT 100;'
); // []

Create database:

Create database data with table product from a compressed CSV file data.csv.gz.

(new Pducky\Adapter('data.csv.gz'))->import('data', 'product');

Loader query (FFI):

$rows = (new Pducky\Loader())->connect('data.db')
	   ->importCsv('data.csv', 'product')
	   ->query('SELECT * FROM product LIMIT 100;'); // []

References:

• SQL Introduction
• Importing Data

Authors:

• Jakiboy (Initial work)

⭐ Support:

Skip the coffee! If you like the project, a Star would mean a lot.