Description : Library for json schemas validation
This library is mainly composed of 3 interfaces and 3 classes implementing them.
Allows to encapsulate the data into an object and format it for DataSchema validation
JsonData::format()
uses json_decode()
to format json string into php array.
Allows to validate data unit by checking if the data is null, optional and his length, pattern and type.
JsonRule can validate :
- Array types: list, array
- Integer types: int, integer, long
- Float types: float, double
- Numeric types: numeric, number
- Boolean types: boolean, bool
- Character types: char, character
- Typed Array types: numeric list, string list, boolean list, integer list, float list, character list
- Object types: entity, object
- Strings
Only strings and typed arrays can be length checked.
Only strings can be date format checked.
Only integers, floats, numbers and typed arrays can be min/max checked.
Only strings, characters, string typed arrays and character typed array can be pattern checked.
Only strings, integers, floats, numbers, typed list can be date format checked.
An object is a "child" schema and an array typed value is an object list.
DataSchema is the library main class. It allows to validate DataType based on "child" schema or Rule
DataSchema::validate()
method allows this validation. If the DataType does not validate the DataSchema, DataSchema::validate()
will return false and DataSchema::getLastError()
will return the validation error.
JsonSchema implements DataSchema, validates JsonData and uses JsonRule for validation.
See tests for examples
A JsonSchema has a type : object
or list
.
Objects are composed of rules and "child" schemas if needed.
This is a schema definition :
use hunomina\Validator\Json\Schema\JsonSchema;
$schema = (new JsonSchema())->setSchema([
'success' => ['type' => 'bool'],
'error' => ['type' => 'string', 'null' => true],
'user' => ['type' => 'object', 'null' => true, 'optional' => true, 'schema' => [
'name' => ['type' => 'string'],
'age' => ['type' => 'int']
]]
]);
Schemas are just php arrays passe to JsonSchema::setSchema()
method.
This schema is composed of 3 elements :
-
a rule
success
which :- is a boolean
- can not be null
- is not optional
-
a rule
error
which :- is a string
- can be null
- is not optional
-
a "child" schema
user
which :- is an object and therefor is represented by a schema which contains 2 elements : a
name
(string) and anage
(integer) - can be null
- is optional
- is an object and therefor is represented by a schema which contains 2 elements : a
When a data unit is being validated using this schema by calling the JsonSchema::validate()
method, the schema will check recursively if the data respects the rules and the "child" schemas.
If the data has :
- a boolean element
success
- a null or string element
error
- an optionally, null or object element
user
which must have :- a string element
name
- an integer element
age
- a string element
This data is valid :
use hunomina\Validator\Json\Data\JsonData;
$data = (new JsonData())->setDataFromArray([
'success' => true,
'error' => null,
'user' => [
'name' => 'test',
'age' => 10
]
]);
This one is not :
use hunomina\Validator\Json\Data\JsonData;
$data = (new JsonData())->setDataFromArray([
'success' => true,
'error' => null,
'user' => 'test'
]);
As said earlier, rules can be used to validate length or data pattern.
This schema uses the pattern validation on the name
element and the length validation on the geolocation
element :
use hunomina\Validator\Json\Schema\JsonSchema;
$schema = (new JsonSchema())->setSchema([
'name' => ['type' => 'string', 'pattern' => '/^[a-z]+$/'],
'geolocation' => ['type' => 'integer-list', 'length' => 2]
]);
When calling the JsonSchema::validate()
method, the schema will recursively check all the rule set and "child" schemas. If one rule or one "child" schema is invalid, JsonSchema::validate()
returns false
.
The "first level" schema is an object
typed schema. It could be changed but is not meant to.
Finally, if a "child" schema is typed as an object
, the schema will validate it as described above. If it's typed as a list
, the schema will simply check each element of the data as an object
type using the given "child" schema.