@myorm/graphql
v0.0.5
Published
Plugin that converts existing MyORM contexts into GraphQL endpoints.
Downloads
3
Maintainers
Readme
MyORM
GraphQL
The MyORM
GraphQL plugin serves as a fast and easy way to integrate your database to a GraphQL endpoint by providing various MyORMContext
objects connected to tables, which then creates your root query and mutation types.
Getting Started
Run the following commands.
npm i @myorm/graphql
npm i @myorm/mysql-adapter # or whichever adapter you prefer to use.
Create your database:
CREATE DATABASE auth;
USE auth;
CREATE TABLE User (
Id INT AUTO_INCREMENT,
FirstName VARCHAR(32) NOT NULL,
LastName VARCHAR(32) NOT NULL,
Username VARCHAR(20) NOT NULL,
PassChecksum VARCHAR(36) NOT NULL,
PRIMARY KEY (Id)
);
CREATE TABLE UserRoleXref (
UserId INT NOT NULL,
RoleId INT NOT NULL,
PRIMARY KEY (UserId, RoleId),
FOREIGN KEY (UserId) REFERENCES User (Id)
);
CREATE TABLE Role (
Id INT AUTO_INCREMENT,
Title VARCHAR(20) NOT NULL,
Description VARCHAR(64),
PRIMARY KEY (Id)
);
Construct your TypeScript types:
interface User {
Id?: number;
FirstName: string;
LastName: string;
Username: string;
PassChecksum: string;
UserRoles?: UserRoleXref[];
};
interface UserRoleXref {
UserId?: number;
RoleId?: number;
User?: User;
Role?: Role;
}
interface Role {
Id?: number;
Title: string;
Description?: string;
UserRoles?: UserRoleXref[];
}
Import @myorm/myorm
, @myorm/mysql-adapter
, and @myorm/graphql
.
import { MyORMContext } from '@myorm/myorm';
import { adapter, createMySql2Pool } from '@myorm/mysql-adapter';
import { MyORMGraphQL } from '@myorm/graphql';
Configure your connection to your database.
const pool = createMySql2Pool({
user: 'root',
password: 'root',
host: 'localhost',
port: 3306,
database:
});
const connection = adapter(pool);
Construct your MyORMContext
objects.
const users = new MyORMContext<User>(connection, "User");
const userRoles = new MyORMContext<UserRoleXref>(connection, "UserRoleXref");
const roles = new MyORMContext<Role>(connection, "Role");
Configure relationships (if any exist)
users.hasMany(m => m.UserRoles.fromTable("UserRoleXref").withKeys("Id", "UserId")
.andThatHasOne(m => m.Role.withKeys("Id", "RoleId")));
Create your MyORMGraphQL
object.
const gql = new MyORMGraphQL("auth_connection");
Add your contexts to your MyORMGraphQL
object using .addContext()
.
gql.addContext(users);
gql.addContext(userRoles);
gql.addContext(roles);
Finally, create your root query and root mutation types!
const rootQueryObject = gql.createRootQueryObject();
const rootMutationObject = gql.createRootMutationObject();
Using the express framework and the express-graphql library, you can pass your objects in and start your server!
import express from 'express';
import { graphqlHTTP } from 'express-graphql';
import { GraphQLSchema } from "graphql";
app.use('/graphql', graphqlHTTP({
schema: new GraphQLSchema({
query: rootQueryObject,
mutation: rootMutationObject
}),
graphiql: true
}));
app.listen(5555, () => {
console.log(`Serving on 5555`);
});
If you navigate to the graphiql url, http://localhost:5555/graphql, you will see something like this in the documentation. (on the right side of the screen)
As you navigate throughout the various custom object types that were defined by MyORM
, you will see various other documentation like so:
With this new endpoint, you can query your database using GraphQL syntax like so:
query {
User {
Id,
FirstName
}
}
The above query would generate the following MySQL command:
SELECT `User`.`Id` AS `Id`
,`User`.`FirstName` AS `FirstName`
FROM User AS User
As you can see, the command will be tailored based on the GraphQL query command provided.
Here is another example, including the User's Roles
query {
User {
Id,
FirstName,
LastName,
UserRoles {
UserId,
RoleId,
Role {
Id,
Title,
Description
}
}
}
}
The above query would generate the following MySQL command:
SELECT `User`.`Id` AS `Id`
,`User`.`FirstName` AS `FirstName`
,`User`.`LastName` AS `LastName`
,`__User_UserRoles__`.`UserId` AS `UserRoles<|UserId`
,`__User_UserRoles__`.`RoleId` AS `UserRoles<|RoleId`
,`__User_UserRoles_Role__`.`Id` AS `UserRoles<|Role<|Id`
,`__User_UserRoles_Role__`.`Title` AS `UserRoles<|Role<|Title`
,`__User_UserRoles_Role__`.`Description` AS `UserRoles<|Role<|Description`
FROM User AS User
LEFT JOIN `UserRoleXref` AS `__User_UserRoles__`
ON `User`.`Id` = `__User_UserRoles__`.`UserId`
LEFT JOIN `Role` AS `__User_UserRoles_Role__`
ON `__User_UserRoles__`.`RoleId` = `__User_UserRoles_Role__`.`Id`
Here is an example using default arguments:
query {
User(FirstName: "John", LastName: "Doe"){
Id,
FirstName,
LastName,
UserRoles {
UserId,
RoleId,
Role {
Id,
Title,
Description
}
}
}
}
The above query would generate the following MySQL command:
SELECT `User`.`Id` AS `Id`
,`User`.`FirstName` AS `FirstName`
,`User`.`LastName` AS `LastName`
,`__User_UserRoles__`.`UserId` AS `UserRoles<|UserId`
,`__User_UserRoles__`.`RoleId` AS `UserRoles<|RoleId`
,`__User_UserRoles_Role__`.`Id` AS `UserRoles<|Role<|Id`
,`__User_UserRoles_Role__`.`Title` AS `UserRoles<|Role<|Title`
,`__User_UserRoles_Role__`.`Description` AS `UserRoles<|Role<|Description`
FROM User AS User
LEFT JOIN `UserRoleXref` AS `__User_UserRoles__`
ON `User`.`Id` = `__User_UserRoles__`.`UserId`
LEFT JOIN `Role` AS `__User_UserRoles_Role__`
ON `__User_UserRoles__`.`RoleId` = `__User_UserRoles_Role__`.`Id`
WHERE `User`.`FirstName` = "John"
AND `User`.LastName = "Doe"
You can also edit your database with the mutation object, containing respective insert
, update
, and delete
functionalities.
Each context will receive an edit root type, prepended respectively with update
, insert
, and delete
.
If, for some reason, you would like to ignore certain functionalities of your mutation, you can pass into the .addContext
function's fourth (4th) parameter, mutationOptions
the respective disable property to disable such behavior.
gql.addContext(someContext, undefined, undefined, { disableInserts: true });
The mutation object acts the same as the query object, with the exception of the return values for the update
and delete
query functions, where the return values will only ever be numRowsAffected
which would be the total number of rows affected from the execution.
Additionally, the update
query function will automatically have double the parameters as the query
query function, where one set of arguments are prepended with filterBy_
and the other set being the arguments you use to choose what column to set the new value to.
Finally, in the insert
and update
query functions, any argument that is deemed to be a virtually generated column or an identity
column (automatically increments within the database) will be omitted.
Here is an example of inserting a User.
mutation {
insertUser(FirstName: "John", LastName: "Doe") {
Id,
John,
Doe
}
}
Here is an example of updating the User, "John Doe" to "Jane Doe". (assuming Id is 1
)
mutation {
updateUser(filterBy_Id: 1, FirstName: "Jane") {
numRowsAffected
}
}
Here is an example of deleting the User, "Jane Doe" (assuming Id is 1
)
mutation {
deleteUser(Id: 1) {
numRowsAffected
}
}
Generated behavior
@myorm/graphql
generates all of the GraphQL object types, as well as the root query and root mutation types. With that being the case, it is important to know what you would all have access to.
If there are any requests to have abilities to modify what can be generated, please open an issue here specifically stating what should be done.
Querying
The behavior that is generated using the @myorm/graphql
plugin consists of allowing querying to the table using various optional filtering arguments that are added by default by each column.
Each column as a filter checks for equality
and expects the exact type as it appears in the database. Meaning, if you have a column, named Foo
, as a VARCHAR
column, then the GraphQL API will only accept an argument of type string
, and it will ONLY check for equality.
There may be columns you want different behavior, or you just don't want the behavior overall. In those cases, you can read the Features section to learn more on how to work with your various filtering arguments.
Additionally, with the default columns as filtering arguments, there will always be two arguments for all contexts, skip
, and take
. These filtering arguments allow you to automatically skip some records and/or limit the number of records you would like to retrieve.
Inserting
The behavior that is generated using the @myorm/graphql
plugin consists of each context mutation to expect arguments that belong to the main schema for that table, while maintaining what columns are required and what are not.
NOTE: Columns marked as an identity column (one that auto increments) or is virtually generated is omitted from the arguments that can be accepted.
Updating
The behavior that is generated using the @myorm/graphql
plugin consists of each context mutation to expect arguments that belong to the main schema for that table to be updated, but each argument is optional. Additionally, there will be an equal number of default-generated arguments that are prepended with filterBy_
, which are what you would use to help determine which rows should get updated.
NOTE: If, for some reason, the API call results in the update occurring to all records, then MyORM
should throw an exception, unless the allowUpdateOnAll
option is enabled.
NOTE: Columns marked as an identity column (one that auto increments) or is virtually generated is omitted from the arguments that can be accepted.
Deleting
The behavior that is generated using the @myorm/graphql
plugin consists of each context mutation to expect arguments that belong to the main schema for that table to be used to determine what records should be deleted. This is similar to the filterBy_
arguments in Updating, so each argument will be optional.
NOTE: If, for some reason, the API call results in the delete occurring to all records, then MyORM
should throw an exception. There is no generated behavior that allows for truncation.
Arguments
Various features exist within the @myorm/graphql
plugin, such as adding custom arguments, removing default arguments, or altering default arguments.
Adding custom arguments
You can add custom arguments in the case that it would make more sense.
NOTE: This functionality is not available for Insert
. For Update
and Delete
, the arguments will act as filter arguments.
For example, say you have a table called "Track" in a database that stores music, and you would like your GraphQL endpoint to be queried to receive all tracks where the duration of the track is between two numbers (in milliseconds). In this case, you may want two separate arguments called DurationUpperBound
and DurationLowerBound
.
To add an argument, you just provide a callback within the .addContext()
function in the second argument, where the callback you provide has access to a single parameter which has 4 properties, Query
, Insert
, Update
, and Delete
. Each of these properties have the function .addArgument
, which you can use to add an argument to its respective root GraphQLObjectType
.
addArgument
takes in 3 arguments:
argName
:{name: string, description?: string}
- Name and description of the argument as it should appear in GraphQLdefinitionCallback
:function
- A callback that appears exactly as it does inMyORM
's.where()
function, but this function has an additional paramter calledargVal
, which would be the argument value that was provided by the API caller.graphqlType
:GraphQLScalarType
- A GraphQLScalarType that specifies what type the expected argument should be in the API.
Here is an example using what was discussed above:
import { GraphQLInt } from 'graphql';
// ... initialization
gql.addContext(tracks, ({ Query }) => {
Query.addArgument({
name: "DurationUpperBound"
},
(m,argVal) => m.DurationInMilliseconds.lessThanOrEqualTo(argVal),
GraphQLInt
);
Query.addArgument({
name: "DurationLowerBound"
},
(m,argVal) => m.DurationInMilliseconds.greaterThanOrEqualTo(argVal),
GraphQLInt
);
});
Removing default arguments
You can remove default arguments that you would not like to see in the GraphQL API.
For example, say you have a table called "Track" in a database that stores music, and you would like your GraphQL endpoint to not be queried based on the duration at all. In this case, you would want to remove the ability to query by Duration
.
To remove an argument, you just provide a callback within the .addContext()
function in the second argument, where the callback you provide has access to a single parameter which has 4 properties, Query
, Insert
, Update
, and Delete
. Each of these properties have the function .removeArgument
, which you can use to remove an argument to its respective root GraphQLObjectType
.
NOTE: When using this function with Insert
, the only arguments you are allowed to remove are optional arguments. This is to avoid any errors with database transactions.
Here is an example where we remove the original Duration
field:
// ... initialization
gql.addContext(tracks, ({ Query }) => {
Query.removeArgument(m => m.Duration);
// removing multiple arguments (the same arguments we added from the previous .addArgument example.)
Query.removeArgument(m => [m.FileSize, m.Name]);
});
Altering default arguments
You can alter default arguments on Query arguments that you would like to see different behavior in the GraphQL API.
For example, say you have a table called "Track" in a database that stores music, and you would like your GraphQL endpoint to have the Bytes
column to be queried from a string, rather than an integer, where the string would be expected to be two numbers separated by a dash (-), implying a range of numbers.
To change an argument, you just provide a callback within the .addContext()
function in the second argument, where the callback you provide has access to a single parameter which has 4 properties, Query
, Insert
, Update
, and Delete
. Each of these properties have the function .changeArgument
, which you can use to change an argument to its respective root GraphQLObjectType
.
With the .changeArgument
function, it will also remove the old argument that you reference in the callback you provide.
The dereferenced variable returns an object containing four (4) functions:
.definedAs
: Accepts a callback function that has access to two parameters,model
andargVal
, wheremodel
is the model that the context represents. This is to be treated like a regular.where()
function inMyORM
, except the value intended to be used isargVal
..describedAs
: Accepts a string which would be the new definition for the argument..namedAs
: Accepts a string which would be the new name for the argument..typedAs
: Accepts aGraphQLScalarType
which would be the new type for the argument.
NOTE: Any edit query type does not have access to the .definedAs
and .typedAs
functions, as the original behavior for these arguments is required for the plugin to work as intended.
Each of these functions are optional and can be used in any order. However, if .definedAs()
is used with a different expected argument value, then .typedAs()
should also be used, otherwise, GraphQL will throw an error about an unexpected argument type.
TypeScript will dynamically change the functions that are available based on what has and has not been used.
import { GraphQLString } from 'graphql';
// ... initialization
gql.addContext(tracks, ({ Query }) => {
Query.changeArgument(m => m.Duration
// optional: this is the behavior of how the argument should be handled.
// (NOTE: If the type of `argVal` is different than the intended argument type, then `.typedAs` must be used,
// otherwise GraphQL will throw an error if the argument is used.)
.definedAs((m,argVal) => m.Duration.between(parseInt(argVal.split('-')[0]), parseInt(argVal.split('-')[1])))
// optional: type for GraphQL to expect in the API.
.typedAs(GraphQLString)
// optional: you can change the name of the argument here.
.namedAs("DurationRange")
// optional: you can change the description from the normal description.
.describedAs("some description")
);
});
Other possibilites
Mixing adapters
Since @myorm/graphql
works on each context individually, it is allowed to connect multiple different contexts (as in, connected to different adapters) and therefore, different databases on different servers, and it will not affect the performance of any one context nor the GraphQL endpoint itself.