guide42/plan

Fast and simple validation library

0.5.0 2020-06-21 15:58 UTC

Requires

• php: >=7.2

Requires (Dev)

• kahlan/kahlan: @stable

ISC

• Juan M

• security
• validator
• validation
• filter

README

Plan is a data validation library for PHP. It's planed to be used for validating data from external sources.

It has two core design goals:

1. Simple: use language own features: construct schema from literals and function composition, errors are exceptions;
2. Lightweight: lots of validations included without 3rd party libraries;

Usage

Simplest way would be requiring guide42/plan with composer.

use plan\{Schema, MultipleInvalid, assert as v, filter as f};

$userSchema = new Schema(array(
    'id' => v\int(),
    'type' => v\any('user', 'admin'),
    'name' => v\all(
        v\length(4, 20),
        f\intl\alnum()
    ),
));

$_POST = ['id' => 3004, 'type' => 'user', 'name' => 'John'];

try {
    $user = $userSchema($_POST);
} catch (MultipleInvalid $errors) {
    $messages = $errors->getMessages();
}

Concepts

Defining schemas is the first step to validate an input data. An schema is a tree of validators that can be manually defined or created based on literals and arrays.

With the schema defined, a plan must be make to validate data. This is made through the \plan\Schema class. This object, when called like a function, will validate the data and return the modified (or not) data. If any error occurs and exception will be thrown.

Schema information will be always be trusted, therefore will not be validate. Contrary input data will be never be trusted.

Literals

Scalars are treated as literals that are matched using the identity operator.

$plan = new Schema('Hello World');

assert('Hello World' === $plan('Hello World'));

As any plan validator it throws an \plan\Invalid when fails.

$plan = new Schema(42);
$plan(42);

try {
    $plan(10);
} catch (MultipleInvalid $errors) {
    assert('[ 10 is not 42 ]' === $errors->getMessage());
}

Arrays

Plan will distinguish between indexed and associative arrays. If an array has all indexes numeric and sequential will be considerer a sequence. If not will be considerer a dictionary.

Sequences

A sequence will be treated as a list of possible valid values. Will require that the input data is sequence that contains one or more elements of the schema. Elements can be repeated.

An array of possible values:

$plan = new Schema([1, 'one']);
$plan([1]);
$plan([1, 'one', 1, 1, 'one', 1]);

An empty array will be a sequence that accept any value.

$plan = new Schema([]);
$plan(['anything', 123, true]);

Dictionaries

A dictionary will be used to validate structures. Each key in data will be checked with the validator of the same key in the schema.

Validators

All core validators live in \plan\assert namespace.

literal

See Literals.

type

Will validate the type of data. The data type will be not casted.

$plan = new Schema(v\type('integer'));
$plan(123);

try {
    $plan('123');
} catch (MultipleInvalid $errors) {
    assert('[ "123" is not integer ]' === $errors->getMessage());
}

Aliases of this validator are: bool, int, float, str.

scalar

Wrapper around is_scalar function.

instance

Wrapper around instanceof type operator.

iterable

Given data must be an array or implement Iterable interface.

required

Rejected values are null and ''.

optional

Accepts null and empty string without calling the given schema.

$plan = new Schema(v\dict(array(
    'id' => v\optional(v\int()),
    'name' => v\optional(v\str()),
)));
$plan(array('id' => 3004));
$plan(array('name' => 'John'));
$plan(array());

seq

See Sequences.

This is normally accepted as "a list of something (or something else)".

• A list of email? new Schema([v\email()]).

• A list of people, but some of them are in text and some as a dictionary?

  $plan = new Schema([v\str(), array(
      'name'  => v\str(),
      'email' => v\email(),
  )]);
  $plan([
      array('name' => 'Kevin', 'email' => '[e-mail]'),
      array('name' => 'Jane', 'email' => '[e-mail]'),
      'John Doe <[e-mail]>',
  ]);

dict

See Dictionaries.

Elements of the dictionary not found in data will be called with null; any additional data key will throw an exception.

$dict = array('name' => 'John', 'age' => 42);
$plan = new Schema($dict);

try {
    $plan(array('age' => 42));
} catch (MultipleInvalid $errors) {
    assert('{ [name]: null is not string }' === $errors->getMessage());
}

try {
    $plan(array('name' => 'John', 'age' => 42, 'sex' => 'male'));
} catch (MultipleInvalid $errors) {
    assert('{ Extra key sex not allowed }' === $errors->getMessage());
}

The validator dict accept two more parameters to change this behavior.

$required = false; // Accept any keys
$extra    = true;  // Accept extra keys

$dict = array('name' => 'John', 'age' => 42);
$plan = new Schema(v\dict($dict, $required, $extra));
$plan(array(
    'name' => 'John',
    'sex'  => 'male', // This could be whatever
                      // as it would not be validated
));

Both parameters (required and extra) could be arrays, so only the given keys will be taken in account.

$dict = array('name' => 'John', 'age' => 42);
$plan = new Schema(v\dict($dict, ['age'], ['sex']));
$plan(array('name' => 'John', 'age' => 42, 'sex' => 'male'));

try {
    $plan(array('name' => 'John', 'hobby' => 'sailing'));
} catch (MultipleInvalid $errors) {
    assert('{ Extra key hobby not allowed, [age]: Required age not provided }' === $errors->getMessage());
}

If the extra parameter is a dictionary it will be compiled and treat it as a validator for each extra key.

$extra = array('dob' => v\instance('\\DateTime'));

$dict = array('name' => 'John', 'age' => 42);
$plan = new Schema(v\dict($dict, true, $extra));
$plan(array('name' => 'John', 'age' => 42, 'dob' => new \DateTime));

try {
    $plan(array('name' => 'John', 'age' => 42, 'dob' => '1970-01-01'));
} catch (MultipleInvalid $errors) {
    assert('{ Extra key dob is not valid: Expected \DateTime (is not an object) }' === $errors->getMessage());
}

There is no way of treat all items with the same validator. Nor having a default validator for extra keys.

keys

Is also possible to validate and/or filter the list of keys of a dictionary.

$schema = v\keys(function(array $keys) {
    return array_filter($keys, function($key) {
        return $key === 'two';
    });
});

$result = $schema([
    'one' => 1,
    'two' => 2,
]);

assert($result === [ 'two' => 2 ]);

object

The structure of an object can also be validated.

$structure = array('name' => v\str());
$class     = 'stdClass';
$byref     = true;

$plan = new Schema(v\object($structure, $class, $byref));
$plan((object) array('name' => 'John'));

try {
    $plan((object) array('name' => false));
} catch (MultipleInvalid $errors) {
    assert('{ [name]: false is not string }' === $errors->getMessage());
}

any

Accept any of the given list of validators, as a valid value.

$plan = new Schema(array(
    'Connection' => v\any('ethernet', 'wireless'),
));
$plan(array('Connection' => 'ethernet'));
$plan(array('Connection' => 'wireless'));

try {
    $plan(array('Connection' => 'any'));
} catch (MultipleInvalid $errors) {
    assert('{ [Connection]: No valid value found }' === $errors->getMessage());
}

This is useful when you only need one choice a of set of values. If you need any quantity of choices use a sequence instead.

all

Require all validators to be valid.

$plan = new Schema(v\all(v\str(), v\length(3, 17)));
$plan('Hello World');

try {
    $plan('No');
} catch (MultipleInvalid $errors) {
    assert('[ Value must be at least 3 ]' === $errors->getMessage());
}

not

Negative the given validator.

$plan = new Schema(v\not(v\str()));
$plan(true);
$plan(123);

try {
    $plan('fail');
} catch (MultipleInvalid $errors) {
    assert('[ Validator passed ]' === $errors->getMessage());
}

iif

Simple conditional.

$class = 'stdClass';
$plan = new Schema(v\iif(null !== $class,
    v\instance($class),
    v\type('object')
));

$plan(new stdClass);

try {
    $plan(new Exception('Arr..'));
} catch (MultipleInvalid $errors) {
    assert('[ Expected stdClass (is Exception) ]' === $errors->getMessage());
}

length

The given data length is between some minimum and maximum value. This works with strings using strlen or count for everything else.

$plan = new Schema(v\length(2, 4));
$plan('abc');
$plan(['a', 'b', 'c']);

try {
    $plan('hello');
} catch (MultipleInvalid $errors) {
    assert('[ Value must be at most 4 ]' === $errors->getMessage());
}

validate

A wrapper for validate filters using filter_var.

$plan = new Schema(v\validate('email'));
$plan('[e-mail]');

try {
    $plan('john(@)example.org');
} catch (MultipleInvalid $errors) {
    assert('[ Expected email ]' === $errors->getMessage());
}

It accepts the name of the filter.

Aliases are: url, email, ip. And the "like-type": boolval, intval, floatval. Note that this will check that a string resemble to a boolean/int/float; for checking if the input data IS a boolean/int/float use the type validator. None of this will modify the input data.

datetime

Validates if given datetime in string can be parsed by given format.

match

Value must be a string that matches the regular expression.

Filters

The input data can also be filtered, and the validation will return the modified data. By convention is called validator when it will not modified the input data; and filter when modification to the data are performed.

Core filters will be found in the \plan\filter namespace.

type

Will cast the data into the given type.

$plan = new Schema(f\type('int'));
$data = $plan('123 users');

assert(123 === $data);

Note that boolval, strval, intval, floatval are not aliases of this filter but wrappers of the homonymous functions.

sanitize

Sanitization filters.

$plan = new Schema(f\sanitize('email'));
$data = $plan('(john)@example.org');

assert('[e-mail]' === $data);

Aliases are: url, email.

datetime

Will parse a datetime formated string into a \DateTimeImmutable object.

$plan = new Schema(f\datetime('Y-m-d H:i:s'));
$data = $plan('2009-02-23 23:59:59')->format('m-d');

assert('02-23' === $data);

String Filters

All string transformations fall under the \plan\filter\str namespace.

nullempty

Change empty strings into nulls.

$plan = new Schema(f\str\nullempty());
$data = $plan('');

assert(null === $data);

strip

Wrapper of trim function.

$plan = new Schema(f\str\strip("\t"));
$data = $plan('\t  Hello World\t');

assert('  Hello World' === $data);

Internationalization

This library supports some filters to be language dependant. Before using any of them make sure that the correct locale is set (ex. by using setlocale).

chars

Will keep only characters in the current language and numbers. Optionally white-space could be keeped too.

$lower      = true; // all lower-case characters
$upper      = true; // all upper-case characters
$number     = true; // all numbers
$whitespace = true; // the only one not language dependant

$plan = new Schema(f\intl\chars($lower, $upper, $number, $whitespace));
$data = $plan('Hello World ☃!!1');

assert('Hello World 1' === $data);

Aliases are: alpha, alnum.

Writing Validators

A simple callable can be a validator.

Any validation error is thrown with the Invalid exception. If several errors must be reported, MultipleInvalid is an exception that could contain multiple exceptions. All other exceptions are considerer as errors in the validator.

$passwordStrength = function($data, $path = null)
{
    $type = v\str(); // Use another validator to check that `$data` is
    $data = $type($data); // an string, if not will throw an exception.

    // Because we are going to throw more than one error, we will
    // accumulate in this variable.
    $errors = [];

    if (strlen($data) < 8) {
        $errors[] = new Invalid('Must be at least 8 characters');
    }

    if (!preg_match('/[A-Z]/', $data)) {
        $errors[] = new Invalid('Must have at least one uppercase letter');
    }

    if (!preg_match('/[a-z]/', $data)) {
        $errors[] = new Invalid('Must have at least one lowercase letter');
    }

    if (!preg_match('/\d/', $data)) {
        $errors[] = new Invalid('Must have at least one digit');
    }

    if (count($errors) > 0) {
        throw new MultipleInvalid($errors);
    }

    // If everything went OK, we return the data so it can continue to be
    // checked by the chain.
    return $data;
};

$validator = new Schema(v\all(v\str(), $passwordStrength, v\not('hunter2')));
$validated = $validator('heLloW0rld');

Acknowledgments

This library is heavily inspired in Voluptuous by Alec Thomas.

Badges