Workflow

Summary

Chevere Workflow is a PHP library for building and executing multi-step procedures with automatic dependency resolution. Define independent jobs that can run synchronously or asynchronously, pass data between them using typed responses, and let the engine handle execution order automatically.

Key features:

• Declarative job definitions - Define what to do, not how to orchestrate it
• Automatic dependency graph - Jobs execute in optimal order based on their dependencies
• Sync and async execution - Mix blocking and non-blocking jobs freely
• Type-safe responses - Access job outputs with full type safety
• Conditional execution - Run jobs based on variables or previous responses
• Built-in retry policies - Handle transient failures automatically
• Testable components - Each job is independently testable

Installing

Workflow is available through Packagist and the repository source is at chevere/workflow.

composer require chevere/workflow

Quick Start

Here's a minimal example to get you started:

use function Chevere\Workflow\{workflow, sync, variable, run};

// 1. Define a workflow with jobs
$workflow = workflow(
    greet: sync(
        fn(string $name): string => "Hello, {$name}!",
        name: variable('username')
    )
);

// 2. Run with variables
$result = run($workflow, username: 'World');

// 3. Get typed responses
echo $result->response('greet')->string();
// Output: Hello, World!

Core Concepts

Workflow is built around four main concepts:

Concept	Description
Job	A unit of work (Action class or Closure) that produces a response
Variable	External input provided when running the workflow
Response	Reference to output from a previous job
Graph	Automatic execution order based on job dependencies

How It Works

1. You define jobs using sync() or async() functions
2. Jobs declare their inputs: literal values, variable() references, or response() from other jobs
3. The engine builds a dependency graph automatically
4. Jobs execute in optimal order (parallel when possible)
5. Access typed responses after execution

Functions Reference

Function	Purpose
workflow()	Create a workflow from named jobs
sync()	Create a synchronous (blocking) job
async()	Create an asynchronous (non-blocking) job
variable()	Declare a runtime variable
response()	Reference another job's output
run()	Execute a workflow with variables

---

Jobs

Jobs are the building blocks of a workflow. Each job wraps an executable unit (Action or Closure) and declares its input arguments.

Creating Jobs with Closures

Use closures for simple, inline operations:

use function Chevere\Workflow\{workflow, sync, async, variable, response, run};

$workflow = workflow(
    // Simple calculation
    add: sync(
        fn(int $a, int $b): int => $a + $b,
        a: 10,
        b: variable('number')
    ),
    // Format the result
    format: sync(
        fn(int $sum): string => "Sum: {$sum}",
        sum: response('add')
    )
);

$result = run($workflow, number: 5);
echo $result->response('format')->string(); // Sum: 15

Creating Jobs with Action Classes

For complex or reusable logic, use Action classes:

use Chevere\Action\Action;

class FetchUser extends Action
{
    public function __invoke(int $userId): array
    {
        // Fetch user from database
        return ['id' => $userId, 'name' => 'John', 'email' => 'john@example.com'];
    }
}

class SendEmail extends Action
{
    public function __invoke(string $email, string $subject): bool
    {
        // Send email logic
        return true;
    }
}

$workflow = workflow(
    user: sync(
        FetchUser::class,
        userId: variable('id')
    ),
    notify: sync(
        SendEmail::class,
        email: response('user', 'email'),
        subject: 'Welcome!'
    )
);

$result = run($workflow, id: 123);

Sync vs Async Jobs

Synchronous jobs (sync) block execution until complete. Use for operations that must run in sequence:

workflow(
    first: sync(ActionA::class),  // Runs first
    second: sync(ActionB::class), // Waits for first
    third: sync(ActionC::class),  // Waits for second
);
// Graph: first → second → third

Asynchronous jobs (async) run concurrently when they have no dependencies:

workflow(
    resize1: async(ResizeImage::class, size: 'thumb'),
    resize2: async(ResizeImage::class, size: 'medium'),
    resize3: async(ResizeImage::class, size: 'large'),
    store: sync(StoreFiles::class, files: response('resize1'), ...)
);
// Graph: [resize1, resize2, resize3] → store

Job Arguments

Jobs accept three types of arguments:

workflow(
    example: sync(
        MyAction::class,
        literal: 'fixed value',           // Literal value
        dynamic: variable('userInput'),   // Runtime variable
        chained: response('otherJob'),    // Previous job output
    )
);

Jobs can define I/O rules via chevere/parameter. Workflow derives parameter and return definitions from the closure signature or Action reflection, and validates inputs and responses at runtime.

---

Variables

Variables are placeholders for values provided at runtime. Declare them with variable():

$workflow = workflow(
    job1: sync(
        SomeAction::class,
        name: variable('userName'),
        age: variable('userAge')
    )
);

// Provide values when running
$result = run($workflow, userName: 'Alice', userAge: 30);

All declared variables must be provided when running the workflow.

---

Responses

Use response() to pass output from one job to another. This automatically establishes a dependency.

$workflow = workflow(
    fetch: sync(
        FetchData::class,
        url: variable('endpoint')
    ),
    process: sync(
        ProcessData::class,
        data: response('fetch')  // Gets entire response from 'fetch'
    ),
    extract: sync(
        ExtractField::class,
        value: response('fetch', 'items')  // Gets 'items' key from response
    )
);

Accessing Nested Response Keys

When a job returns an array, access specific keys:

response('user')           // Entire response
response('user', 'id')     // $response['id']
response('user', 'profile') // $response['profile']

---

Execution Graph

The workflow engine automatically builds an execution graph based on job dependencies. Jobs without dependencies run in parallel (when using async), while dependent jobs wait for their dependencies.

$workflow = workflow(
    // Independent async jobs run in parallel
    thumb: async(ImageResize::class, size: 'thumb', file: variable('image')),
    medium: async(ImageResize::class, size: 'medium', file: variable('image')),
    large: async(ImageResize::class, size: 'large', file: variable('image')),
    // Sync job waits for all above
    store: sync(
        StoreFiles::class,
        thumb: response('thumb'),
        medium: response('medium'),
        large: response('large')
    )
);

// View the execution graph
$graph = $workflow->jobs()->graph()->toArray();
// [
//     ['thumb', 'medium', 'large'],  // Level 0: parallel
//     ['store']                       // Level 1: after dependencies
// ]

graph TD
    thumb --> store
    medium --> store
    large --> store

Loading

---

Running Workflows

Execute a workflow with the run() function:

use function Chevere\Workflow\run;

// Basic execution
$result = run($workflow, var1: 'value1', var2: 'value2');

// With dependency injection container
$result = run($workflow, $container, var1: 'value1');

Accessing Responses

The run result provides type-safe access to job responses:

$result = run($workflow, ...);

// Get typed responses
$result->response('jobName')->string();     // string
$result->response('jobName')->int();        // int
$result->response('jobName')->float();      // float
$result->response('jobName')->bool();       // bool
$result->response('jobName')->array();      // array

// Access array keys directly
$result->response('jobName', 'key')->string();
$result->response('jobName', 'nested', 'key')->int();

Check Skipped Jobs

When using conditional execution, check which jobs were skipped:

if ($result->skip()->contains('optionalJob')) {
    // Job was skipped
}

---

Conditional Execution

Control whether a job runs using withRunIf() (run when conditions are met) or withRunIfNot() (skip when conditions are met). Both methods accept the same kinds of conditions and are evaluated at run-time.

Accepted condition types

• boolean literal — evaluated directly
• variable('name') — runtime argument coerced to boolean
• response('job') or response('job', 'key') — uses another job's output
• Closure — invokes a closure passing the current RunInterface context

---

use function Chevere\Workflow\{workflow, sync, variable, run, response};

$workflow = workflow(
    isTooBig: sync(
        fn(string $path, int $maxBytes): bool => filesize($path) > $maxBytes,
        path: variable('file'),
        maxBytes: variable('maxBytes')
    ),
    compress: sync(
        CompressImage::class,
        file: variable('file')
    )->withRunIf(
        true,                       // literal
        variable('shouldCompress'), // workflow variable
        response('isTooBig'),       // job response value
        fn(RunInterface $run) => $run->variable('shouldCompress')->bool(), // closure condition variable
        fn(RunInterface $run) => $run->response('isTooBig')->bool(), // closure condition using response
    )
);
$result = run($workflow,
    file: '/path/to/image.jpg',
    shouldCompress: true,
    maxBytes: 1_000_000
);

---

Explicit Dependencies

While response() creates implicit dependencies, use withDepends() for explicit control:

$workflow = workflow(
    setup: async(SetupAction::class),
    process: async(
        ProcessAction::class,
        data: variable('input')
    )->withDepends('setup')  // Wait for setup even without using its response
);

---

Retry Policy

Configure automatic retries for jobs that may fail transiently:

$workflow = workflow(
    fetch: sync(
        FetchFromApi::class,
        url: variable('endpoint')
    )->withRetry(
        timeout: 300,     // Max 300 seconds total
        maxAttempts: 5,   // Try up to 5 times
        delay: 10         // Wait 10 seconds between attempts
    )
);

Parameter	Type	Default	Description
timeout	int<0, max>	0	Max execution time in seconds (0 = unlimited)
maxAttempts	int<1, max>	1	Total attempts including initial
delay	int<0, max>	0	Seconds between retries (0 = immediate)

Retry delays use non-blocking sleep, making them safe for async runtimes.

---

Exception Handling

When a job fails, a WorkflowException wraps the original exception:

use Chevere\Workflow\Exceptions\WorkflowException;

try {
    $result = run($workflow, ...);
} catch (WorkflowException $e) {
    echo $e->name;        // Name of the failed job
    echo $e->job;         // Job instance
    echo $e->throwable;   // Original exception
}

---

Using WorkflowTrait

For class-based workflow management, use WorkflowTrait:

use Chevere\Workflow\Traits\WorkflowTrait;
use function Chevere\Workflow\{workflow, sync, variable};

class OrderProcessor
{
    use WorkflowTrait;

    public function process(int $orderId): void
    {
        $workflow = workflow(
            validate: sync(ValidateOrder::class, id: variable('orderId')),
            charge: sync(ChargePayment::class, order: response('validate')),
            fulfill: sync(FulfillOrder::class, order: response('charge'))
        );

        $this->execute($workflow, orderId: $orderId);
    }

    public function getResult(): string
    {
        return $this->run()->response('fulfill')->string();
    }
}

---

Testing

Testing Actions

Test your Action classes independently:

use PHPUnit\Framework\TestCase;

class FetchUserTest extends TestCase
{
    public function testFetchUser(): void
    {
        $action = new FetchUser();
        $result = $action(userId: 123);

        $this->assertSame(123, $result['id']);
        $this->assertArrayHasKey('name', $result);
    }
}

Testing Workflow Graph

Verify execution order:

public function testWorkflowGraph(): void
{
    $workflow = workflow(
        a: async(ActionA::class),
        b: async(ActionB::class),
        c: sync(ActionC::class, x: response('a'), y: response('b'))
    );

    $graph = $workflow->jobs()->graph()->toArray();

    $this->assertSame([['a', 'b'], ['c']], $graph);
}

Testing Responses

Test complete workflow execution:

public function testWorkflowResponses(): void
{
    $result = run($workflow, input: 'test');

    $this->assertSame('expected', $result->response('job1')->string());
    $this->assertSame(42, $result->response('job2', 'count')->int());
}

Testing Exceptions

Use ExpectWorkflowExceptionTrait for error scenarios:

use Chevere\Workflow\Traits\ExpectWorkflowExceptionTrait;

class WorkflowExceptionTest extends TestCase
{
    use ExpectWorkflowExceptionTrait;

    public function testJobFailure(): void
    {
        $this->expectWorkflowException(
            closure: fn() => run($workflow, input: 'invalid'),
            exception: InvalidArgumentException::class,
            job: 'validate',
            message: 'Invalid input provided'
        );
    }
}

---

Real-World Examples

Image Processing Pipeline

$workflow = workflow(
    // Parallel image resizing
    thumb: async(
        ImageResize::class,
        file: variable('image'),
        width: 150,
        height: 150
    ),
    medium: async(
        ImageResize::class,
        file: variable('image'),
        width: 800
    ),
    // Store after all resizing completes
    store: sync(
        StoreFiles::class,
        thumb: response('thumb'),
        medium: response('medium'),
        directory: variable('outputDir')
    )
);

$result = run($workflow,
    image: '/uploads/photo.jpg',
    outputDir: '/processed/'
);

User Registration Flow

$workflow = workflow(
    validate: sync(
        ValidateRegistration::class,
        email: variable('email'),
        password: variable('password')
    ),
    createUser: sync(
        CreateUser::class,
        data: response('validate')
    ),
    sendWelcome: async(
        SendWelcomeEmail::class,
        user: response('createUser')
    ),
    logEvent: async(
        LogRegistration::class,
        userId: response('createUser', 'id')
    )
);

Conditional Processing

$workflow = workflow(
    analyze: sync(
        AnalyzeContent::class,
        content: variable('text')
    ),
    translate: sync(
        TranslateContent::class,
        text: variable('text'),
        targetLang: variable('lang')
    )->withRunIf(
        variable('needsTranslation')
    ),
    publish: sync(
        PublishContent::class,
        content: response('analyze'),
        translated: response('translate')
    )
);

$result = run($workflow,
    text: 'Hello world',
    lang: 'es',
    needsTranslation: true
);

---

Demo

Run the included examples:

php demo/hello-world.php          # Basic workflow
php demo/chevere.php              # Chained jobs
php demo/closure.php              # Using closures
php demo/sync-vs-async.php        # Performance comparison
php demo/image-resize.php         # Parallel processing
php demo/run-if.php               # Conditional execution

See the demo directory for all examples.

Documentation

Documentation is available at chevere.org/packages/workflow.

For a comprehensive introduction, read Workflow for PHP on Rodolfo's blog.

License

Copyright Rodolfo Berrios A.

This software is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.