Browse documentation

#Flow PHP

Flow is a strongly typed, memory-efficient data processing framework for PHP. It gives you one fluent API to read, transform, and write data in CSV, JSON, XML, Parquet, REST, RDBMS, SEAL, and other formats. You do not write boilerplate for each format.

This framework is for people who build pipelines, ETL jobs, exports, imports, or reports in PHP, and for whom one-off scripts are no longer sufficient.

#Hello, Flow

Install the core and the adapters you need:

composer require flow-php/etl flow-php/etl-adapter-csv flow-php/etl-adapter-json

Then write your first pipeline:

<?php

use function Flow\ETL\DSL\{data_frame, ref};
use function Flow\ETL\Adapter\CSV\from_csv;
use function Flow\ETL\Adapter\JSON\to_json;

data_frame()
    ->read(from_csv('orders.csv'))
    ->filter(ref('amount')->isNotNull())
    ->write(to_json('orders.json'))
    ->run();

That is a complete program. The pipeline reads from top to bottom. It streams rows in constant memory, for each input size.

#Mental model

Every Flow pipeline has three stages:

• Extract: ->read(...) pulls rows from a source (file, API, database).
• Transform: ->filter(), ->withColumn(), ->join(), ->groupBy(), ->window(), and friends shape the rows.
• Load: ->write(...) streams the result into a loader. ->run() executes the full pipeline.

Data moves through the pipeline as DataFrame → Rows batches of columns. Each Rows batch carries the schema that types its columns.

#Where to go from here

New to Flow

• Installation: composer setup and runtime tips.
• Quick Start: from empty project to working pipeline in five minutes.

Learn the API

• Data Frame: the core: filter, join, group by, window, partition, sort, limit.
• DSL Reference: every function in the DSL, with signatures and examples.

Working with a specific source or destination

• Browse Adapters: CSV, JSON, XML, Parquet, Excel, HTTP, PostgreSQL, Doctrine, SEAL, ChartJS.

See it running

• Examples: runnable snippets grouped by topic and format. Open any example in the Playground.

Going to production

• Schema, Save Mode, Caching, Telemetry.

#What Flow isn't

Flow is not a database engine, a query planner, or a distributed compute framework. It runs inside each PHP runtime (CLI, FPM, queue worker) and processes rows in constant memory. For cluster-scale joins or SQL planning, use a distributed engine. For easy, type-safe data pipelines inside a PHP application, use Flow.

#Get help

• Open an issue on GitHub.
• Drop in on Discord.
• Sponsor to support development.

Found a typo or an outdated section? Edit this page on GitHub ↗