fo3nix / php-graphql-oqm
GraphQL Object-to-Query Mapper (OQM) which generates objects AND typed data models from an API schema.
Installs: 99
Dependents: 1
Suggesters: 0
Security: 0
Stars: 1
Watchers: 0
Forks: 27
pkg:composer/fo3nix/php-graphql-oqm
Requires
- php: ^7.1 || ^8.0
- nesbot/carbon: ^2.67 | ^3.10
- pavel-paliy/php-graphql-client: ^1.0
Requires (Dev)
- ext-json: *
- phpunit/phpunit: ^7.5|^9
This package is auto-updated.
Last update: 2025-10-16 11:35:26 UTC
README
Forked from mghoneimy/php-graphql-oqm with added Result Object functionality.
This package utilizes the introspection feature of GraphQL APIs to generate a set of classes that map to the structure of the API schema. The generated classes can then be used in a very simple and intuitive way to query the API server.
Interacting with GraphQL API's using PHP has never been easier!
New Features in This Fork
This fork adds Result Object functionality that allows you to:
- Generate strongly-typed result objects from GraphQL schema
- Hydrate these objects from API response arrays using
fromArray()
method - Hydrate objects from JSON strings using
fromJson()
method - Convert objects back to arrays using
asArray()
method - Automatic handling of date/time fields with Carbon
- Recursive handling of nested objects
Installation
Run the following command to install the package using composer:
composer require fo3nix/php-graphql-oqm
Generating The Schema Objects
After installing the package, the first step is to generate the schema objects. This can be easily achieved by executing the following command:
php vendor/bin/generate_schema_objects
This script will retrieve the API schema types using the introspection feature in GraphQL, then generate the schema
objects from the types, and save them in the schema_object
directory in the root directory of the package. You can
override the default write directory by providing the "Custom classes writing dir" value when running the command.
You can also specify all options via command line options:
php vendor/bin/generate_schema_objects \
-u "https://graphql-pokemon.vercel.app/" \
-h "Authorization" \
-v "Bearer 123" \
-d "customClassesWritingDirectory" \
-n "Vendor\Custom\Namespace"
or if you prefer long arguments
php vendor/bin/generate_schema_objects \
--url "https://graphql-pokemon.vercel.app/" \
--authorization-header-name "Authorization" \
--authorization-header-value "Bearer 123" \
--directory "customClassesWritingDirectory" \
--namespace "Vendor\Custom\Namespace"
Usage
Building and Running Queries
In all the examples below I'm going to use the super cool public Pokemon GraphQL API as an illustration.
Check out the API at: https://graphql-pokemon.now.sh/
And Github Repo: https://github.com/lucasbento/graphql-pokemon
After generating the schema objects for the public Pokemon API, we can easily query the API by using the
RootQueryObject
. Here's an example:
$rootObject = new RootQueryObject(); $rootObject ->selectPokemons((new RootPokemonsArgumentsObject())->setFirst(5)) ->selectName() ->selectId() ->selectFleeRate() ->selectAttacks() ->selectFast() ->selectName();
What this query does is that it selects the first 5 pokemons returning their names, ids, flee rates, fast attacks with their names. Easy right!?
All what remains is that we actually run the query to obtain results:
$results = $client->runQuery($rootObject->getQuery());
For more on how to use the client class refer to:
- https://github.com/mghoneimy/php-graphql-client#constructing-the-client
- https://github.com/mghoneimy/php-graphql-client#running-queries
Using Result Objects (New Feature)
The new Result Object functionality allows you to convert raw GraphQL query results into strongly-typed PHP objects:
// Step 1: Build the query (same as before) $queryRoot = new RootQueryObject(); $pokemonQuery = $queryRoot ->selectPokemons((new RootPokemonsArgumentsObject())->setFirst(5)) ->selectName() ->selectId() ->selectFleeRate(); // Step 2: Run the query and get results as array $results = $client->runQuery($queryRoot->getQuery(), true); // true = get as array $pokemonsData = $results->getData()['pokemons']; // Step 3: Hydrate the raw array into Result Objects $pokemons = array_map(function ($pokemonData) { return Pokemon::fromArray($pokemonData); }, $pokemonsData); // Step 4: Use the type-hinted objects! foreach ($pokemons as $pokemon) { echo "Name: " . $pokemon->getName() . "\n"; echo "ID: " . $pokemon->getId() . "\n"; echo "Flee Rate: " . $pokemon->getFleeRate() . "\n"; // Convert back to array if needed $pokemonArray = $pokemon->asArray(); } // You can also hydrate from JSON $pokemon = Pokemon::fromJson('{"name":"Pikachu","id":"025","fleeRate":0.1}');
Benefits of using Result Objects:
- Full IDE autocompletion for all properties
- Type safety with proper return type hints
- Automatic handling of date/time fields with Carbon
- Recursive handling of nested objects
- Easy conversion between objects and arrays/JSON
Notes
A couple of notes about schema objects to make your life easier when using the generating classes:
Dealing With Object Selectors
Whilst scalar field setters return an instance of the current query object, object field selectors return objects of
the nested query object. This means that setting the $rootObject
reference to the result returned by an object
selector means that the root query object reference is gone.
Don't:
$rootObject = (new RootQueryObject())->selectAttacks()->selectSpecial()->selectName();
This way you end up with reference to the PokemonAttackQueryObject
, and the reference to the RootQueryObject
is gone.
Do:
$rootObjet = new RootQueryObject(); $rootObject->selectAttacks()->selectSpecial()->selectName();
This way you can keep track of the RootQueryObject
reference and develop your query safely.
Dealing With Multiple Object Selectors
Suppose we want to get the pokemon "Charmander", retrieve his evolutions, evolution requirements, and evolution requirements of his evolutions, how can we do that?
We can't do this:
$rootObject = new RootQueryObject(); $rootObject->selectPokemon( (new RootPokemonArgumentsObject())->setName('charmander') ) ->selectEvolutions() ->selectName() ->selectNumber() ->selectEvolutionRequirements() ->selectName() ->selectAmount() ->selectEvolutionRequirements() ->selectName() ->selectAmount();
This is because the reference is now pointing to the evolution requirements of the evolutions of charmander and not charmander himself.
The best way to do this is by structuring the query like this:
$rootObject = new RootQueryObject(); $charmander = $rootObject->selectPokemon( (new RootPokemonArgumentsObject())->setName('charmander') ); $charmander->selectEvolutions() ->selectName() ->selectNumber() ->selectEvolutionRequirements() ->selectName() ->selectAmount(); $charmander->selectEvolutionRequirements() ->selectName() ->selectAmount();
This way we have kept the reference to charmander safe and constructed our query in an intuitive way.
Generally, whenever there's a branch off (just like in the case of getting evolutions and evolution requirements of the same object) the best way to do it is to structure the query like a tree, where the root of the tree becomes the reference to the object being branch off from. In this case, charmander is the root and evolutions and evolution requirements are 2 sub-trees branched off it.
Improving Query Objects Readability
A couple of hints on how to keep your query objects more readable:
- Store nodes that will be used as roots in branch offs in meaningful variables, just like the case with charmander.
- Write each selector on a separate line.
- Every time you use an object selector, add an extra indentation to the next selectors.
- Move construction of a new object in the middle of a query (such as an ArgumentsObject construction) to a new line.
Schema Objects Generation
After running the generation script, the SchemaInspector will run queries on the GraphQL server to retrieve the API schema. After that, the SchemaClassGenerator will traverse the schema from the root queryType recursively, creating a class for every object in the schema spec.
The SchemaClassGenerator will generate a different schema object depending on the type of object being scanned using the following mapping from GraphQL types to SchemaObject types:
- OBJECT:
QueryObject
- INPUT_OBJECT:
InputObject
- ENUM:
EnumObject
Additionally, an ArgumentsObject
will be generated for the arguments on each field in every object. The arguments
object naming convention is:
{CURRENT_OBJECT}{FIELD_NAME}ArgumentsObject
The QueryObject
The object generator will start traversing the schema from the root queryType
, creating a class for each query object
it encounters according to the following rules:
- The
RootQueryObject
is generated for the type corresponding to thequeryType
in the schema declaration, this object is the start of all GraphQL queries. - For a query object of name {OBJECT_NAME}, a class with name
{OBJECT_NAME}QueryObject
will be created. - For each selection field in the selection set of the query object, a corresponding selector method will be created,
according to the following rules:
- Scalar fields will have a simple selector created for them, which will add the field name to the selection set. The simple selector will return a reference to the query object being created (this).
- Object fields will have an object selector created for them, which will create a new query object internally and nest it inside the current query. The object selector will return instance of the new query object created.
- For every list of arguments tied to an object field an
ArgumentsObject
will be created with a setter corresponding to every argument value according to the following rules:- Scalar arguments: will have a simple setter created for them to set the scalar argument value.
- List arguments: will have a list setter created for them to set the argument value with an
array
- Input object arguments: will have an input object setter created for them to set the argument value with an object
of type
InputObject
The InputObject
For every input object the object generator encounters while traversing the schema, it will create a corresponding class according to the following rules:
- For an input object of name {OBJECT_NAME}, a class with name
{OBJECT_NAME}InputObject
will be created - For each field in the input object declaration, a setter will be created according to the following rules:
- Scalar fields: will have a simple setter created for them to set the scalar value.
- List fields: will have a list setter created for them to set the value with an
array
- Input object arguments: will have an input object setter created for them to set the value with an object of type
InputObject
The EnumObject
For every enum the object generator encounters while traversing the schema, it will create a corresponding ENUM class according to the following rules:
- For an enum object of name {OBJECT_NAME}, a class with name
{OBJECT_NAME}EnumObject
will be created - For each EnumValue in the ENUM declaration, a const will be created to hold its value in the class
The ResultObject (New Feature)
For every object in the schema, a corresponding Result Object class will be generated with:
- Properties for each field in the object
- Getter methods for each property with proper type hints
fromArray()
method to hydrate the object from an arrayfromJson()
method to hydrate the object from a JSON stringasArray()
method to convert the object back to an array- Special handling for date/time fields using Carbon
- Recursive handling of nested objects
License
MIT