firehed/input

PHP Input Handling Tools

2.3.1 2021-10-08 21:50 UTC

README

An input validation framework with a boring name

Test Static analysis Lint codecov

Changelog

Concept

Input validation is an important task in any web application, but remains an extremely tedious task. This replaces intermingled checks with a proper data structure that clearly defines the required and optional inputs.

The design reolves around the idea of API-driven design, where each API endpoint is its own object, however does not explicitly require this format

  • it is capable of validating for any object that defines the input requirements. What it cannot easily handle is the common pattern of large controlelrs that are respnsible for many URLs, since each URL has its own validation requirements. It's certainly possible to structure your code in a way to make this work, but that is liable to become more complicated than the benefit it provides.

Data handling steps

Raw input is transformed into safe data in two primary steps:

  • Parsing
  • Validation

Parsing is responsible for transforming the raw input string into an associative array. If your application is structured to do so, this step can be skipped entirely.

Validation is the most useful part of the library - taking a defined set of optional and required parameters and their types, and comparing the input values to the spec. The implementation prevents invalid data from being propagated entirely; it is not possible to create a SafeInput object (which your application will use) from invalid data!

Upon completion of this process, a SafeInput object is returned that contains data in accordance with the spec defined by the object implementing ValidationInterface (missing optional values are null).

Because this library exists to provide trustable data, it will actively prevent you from second-guessing it; for example, using isset or empty on the data structure will throw an exception. It is the author's experience that acting unable to trust your validated data is an anti-pattern and a code smell; if you insist on doing so, this is not the right tool for you. Forcing trust like this tends to prevent documentation from driting apart from reality.

Example

A basic example follows:

some_controller_file.php

<?php
// This would be in its own file
use Firehed\Input\Interfaces\ValidationInterface;

use Firehed\Input\Containers\SafeInput;
use Firehed\Input\Objects as O;
class Endpoint
    implements ValidationInterface {

    public function getOptionalInputs() {
        return [
            'bar' => new O\Text(),
            'baz' => (new O\Text())->setDefaultValue('my baz'),
        ];
    }

    public function getRequiredInputs() {
        return [
            'foo' => new O\Text(),
        ];
    }

    public function execute(SafeInput $i) {
        // ... do some magic
        // $i['foo'] will be a string
        // $i['bar'] will be a string or null, since it was optional
        // $i['baz'] will be a string or 'my baz', since it was an optional with a default value
    }
}

index.php

<?php
// This is the core of your Front Controller

use Firehed\Input\Containers\RawInput;
use Firehed\Input\Parsers\URLEncoded;

// The endpoint should be detrmined by your router
$endpoint = new Endpoint();

// The parser should be determined by the Content-Type header
$parser = new URLEncoded();


try {
    $input = (new RawInput("foo=world"))
        ->parse($parser)
        ->validate($endpoint);
    $endpoint->execute($input);
} catch (Firehed\Input\Exceptions\InputException $e) {
    // The input contained invalid data
} catch (Exception $e) {
    // Do any logging, error responses, etc.
    echo $e;
}