GraphQL-Tools-Sequelize

Integration of GraphQL-Tools and Sequelize Object-Relational-Mapper (ORM).

About

This Node.js module provides an integration of the GraphQL.js wrapper GraphQL-Tools and the Sequelize Object-Relational-Mapper (ORM) in order to operate on the entities and their relationships of an underlying RDBMS through GraphQL. It provides functions for GraphQL schema definition entries and their corresponding resolver functions for querying and mutating entities and their relationships through GraphQL in a natural Object-Oriented (OO) way. It optionally provides Full-Text-Search (FTS) functionality through ElasticLunr and validation, authorization and tracing hooks. It provides an elaborate CRUD (Create, Read, Update, Delete) functionality for the entities and their relationships.

Installation

1$ npm install \
2  graphql \
3  graphql-tools \
4  graphql-tools-types \
5  graphql-tools-sequelize \
6  sequelize \
7  --save-dev

Usage

Suppose we have a simple domain model, consisting of the two entities OrgUnit and Person and some relationships between them (in UML Class Diagram notation):

1 parentUnit                           supervisor
2 +------+                             +------+
3 |      |                             |      |
4 |      V 0..1                   0..1 V      |
5 |   +-----------+           +-----------+   |
6 +---| OrgUnit   |  belongsTo| Person    |---+
7     |-----------|<----------|-----------|
8     | id        |           | id        |
9     | initials  |--------->*| initials  |
10     | name      |  members  | name      |
11     +-----------+           +-----------+
12               |                 ^
13               |     director    |
14               +-----------------+

With Sequelize ORM this could be defined on the RDBMS level as:

1import Sequelize from "sequelize"
2
3const db = new Sequelize(...)
4const dm = {}
5
6dm.OrgUnit = db.define("OrgUnit", {
7    id:         { type: Sequelize.UUID,        primaryKey: true  },
8    initials:   { type: Sequelize.STRING(3),   allowNull:  false },
9    name:       { type: Sequelize.STRING(100), allowNull:  false }
10})
11dm.Person = db.define("Person", {
12    id:         { type: Sequelize.UUID,        primaryKey: true  },
13    initials:   { type: Sequelize.STRING(3),   allowNull:  false },
14    name:       { type: Sequelize.STRING(100), allowNull:  false }
15})
16dm.OrgUnit.belongsTo(dm.OrgUnit, { as: "parentUnit", foreignKey: "parentUnitId" })
17dm.Person .belongsTo(dm.Person,  { as: "supervisor", foreignKey: "personId"     })
18dm.Person .belongsTo(dm.OrgUnit, { as: "belongsTo",  foreignKey: "orgUnitId"    })
19dm.OrgUnit.hasMany  (dm.Person,  { as: "members",    foreignKey: "orgUnitId"    })
20dm.OrgUnit.hasOne   (dm.Person,  { as: "director",   foreignKey: "directorId"   })

You then establish a GraphQL-to-Sequelize mapping like this:

1import GraphQLToolsSequelize from "graphql-tools-sequelize"
2
3const gts = new GraphQLToolsSequelize(db)
4await gts.boot()

Now you can use this mapping and its factory functions to conveniently create a GraphQL schema definition as the interface for operating on your domain model:

1const definition = `
2    schema {
3        query:    Root
4        mutation: Root
5    }
6    scalar UUID
7    scalar JSON
8    type Root {
9        ${gts.entityQuerySchema("Root", "", "OrgUnit")}
10        ${gts.entityQuerySchema("Root", "", "OrgUnit*")}
11        ${gts.entityQuerySchema("Root", "", "Person")}
12        ${gts.entityQuerySchema("Root", "", "Person*")}
13    }
14    type OrgUnit {
15        ${gts.attrIdSchema("OrgUnit")}
16        ${gts.attrHcSchema("OrgUnit")}
17        initials: String
18        name: String
19        director: Person
20        members: [Person]!
21        parentUnit: OrgUnit
22        ${gts.entityCloneSchema ("OrgUnit")}
23        ${gts.entityCreateSchema("OrgUnit")}
24        ${gts.entityUpdateSchema("OrgUnit")}
25        ${gts.entityDeleteSchema("OrgUnit")}
26    }
27    type Person {
28        ${gts.attrIdSchema("Person")}
29        ${gts.attrHcSchema("Person")}
30        initials: String
31        name: String
32        belongsTo: OrgUnit
33        supervisor: Person
34        ${gts.entityCloneSchema ("Person")}
35        ${gts.entityCreateSchema("Person")}
36        ${gts.entityUpdateSchema("Person")}
37        ${gts.entityDeleteSchema("Person")}
38    }
39`

You also use it and its factory functions to define the corresponding GraphQL resolver functions:

1import GraphQLToolsTypes from "graphql-tools-types"
2
3const resolvers = {
4    UUID: GraphQLToolsTypes.UUID({ name: "UUID", storage: "string" }),
5    JSON: GraphQLToolsTypes.JSON({ name: "JSON" }),
6    Root: {
7        OrgUnit:    gts.entityQueryResolver ("Root", "", "OrgUnit"),
8        OrgUnits:   gts.entityQueryResolver ("Root", "", "OrgUnit*"),
9        Person:     gts.entityQueryResolver ("Root", "", "Person"),
10        Persons:    gts.entityQueryResolver ("Root", "", "Person*"),
11    },
12    OrgUnit: {
13        id:         gts.attrIdResolver      ("OrgUnit"),
14        hc:         gts.attrHcResolver      ("OrgUnit"),
15        director:   gts.entityQueryResolver ("OrgUnit", "director",   "Person"),
16        members:    gts.entityQueryResolver ("OrgUnit", "members",    "Person*"),
17        parentUnit: gts.entityQueryResolver ("OrgUnit", "parentUnit", "OrgUnit"),
18        clone:      gts.entityCloneResolver ("OrgUnit"),
19        create:     gts.entityCreateResolver("OrgUnit"),
20        update:     gts.entityUpdateResolver("OrgUnit"),
21        delete:     gts.entityDeleteResolver("OrgUnit")
22    },
23    Person: {
24        id:         gts.attrIdResolver      ("Person"),
25        hc:         gts.attrHcResolver      ("Person"),
26        belongsTo:  gts.entityQueryResolver ("Person", "belongsTo",  "OrgUnit"),
27        supervisor: gts.entityQueryResolver ("Person", "supervisor", "Person"),
28        clone:      gts.entityCloneResolver ("Person"),
29        create:     gts.entityCreateResolver("Person"),
30        update:     gts.entityUpdateResolver("Person"),
31        delete:     gts.entityDeleteResolver("Person")
32    }
33}

Then you use the established schema definition and resolver functions to generate an executable GraphQL schema with the help of GraphQL-Tools:

1import * as GraphQLTools from "graphql-tools"
2
3const schema = GraphQLTools.makeExecutableSchema({
4    typeDefs: [ definition ],
5    resolvers: resolvers
6})

Finally, you now can execute GraphQL queries against your RDBMS:

1const query = `query { OrgUnits { name } }`
2const variables = {}
3
4GraphQL.graphql(schema, query, null, null, variables).then((result) => {
5    console.log("OK", util.inspect(result, { depth: null, colors: true }))
6}).catch((result) => {
7    console.log("ERROR", result)
8})

The following GraphQL mutation is a more elaborate example of how CRUD operations look like and what is possible:

1mutation {
2    m1: Person {
3        c1: create(id: "c9965340-a6c8-11e6-ac95-080027e303e4", with: {
4            initials:   "BB",
5            name:       "Big Boss"
6        }) { id }
7        c2: create(id: "ca1ace2c-a6c8-11e6-8ef0-080027e303e4", with: {
8            initials:   "JD",
9            name:       "John Doe",
10            supervisor: "c9965340-a6c8-11e6-ac95-080027e303e4"
11        }) { id }
12    }
13    m2: OrgUnit {
14        c1: create(id: "ca8c588a-a6c8-11e6-8f19-080027e303e4", with: {
15            initials: "EH",
16            name:     "Example Holding",
17            director: "c9965340-a6c8-11e6-ac95-080027e303e4"
18        }) { id }
19        c2: create(id: "cabaa4ce-a6c8-11e6-9d6d-080027e303e4", with: {
20            initials:   "EC",
21            name:       "Example Corporation",
22            parentUnit: "ca8c588a-a6c8-11e6-8f19-080027e303e4",
23            director:   "ca1ace2c-a6c8-11e6-8ef0-080027e303e4",
24            members: { set: [
25                "c9965340-a6c8-11e6-ac95-080027e303e4",
26                "ca1ace2c-a6c8-11e6-8ef0-080027e303e4"
27            ] }
28        }) { id }
29    }
30    q1: OrgUnits(where: {
31        initials: "EC"
32    }) {
33        name
34        director   { initials name }
35        members    { initials name }
36        parentUnit {
37            name
38            director   { initials name }
39            members    { initials name }
40        }
41    }
42    m3: Person(id: "c9965340-a6c8-11e6-ac95-080027e303e4") {
43        update(with: { initials: "XXX" }) {
44            id initials name
45        }
46    }
47    c1: Person(id: "c9965340-a6c8-11e6-ac95-080027e303e4") {
48        clone {
49            id initials name
50        }
51    }
52    m4: Person(id: "c9965340-a6c8-11e6-ac95-080027e303e4") {
53        delete
54    }
55    q2: Persons {
56        id initials name
57    }
58}

For more details, see the all-in-one sample, which even provides a network interface through HAPI and the GraphiQL web interface on top of it (with the help of its HAPI integration HAPI-Plugin-GraphiQL).

Application Programming Interface (API)

• import GraphQLToolsSequelize from "graphql-tools-sequelize"
gts = new GraphQLToolsSequelize(sequelize: Sequelize, options?: Object)
  

  Creates a new GraphQL-Tools-Sequelize instance with an existing Sequelize instance sequelize. The options have to given, but can be an empty object. It can contain the following fields:

  • validator(type: String, obj: Object, ctx: Object): Promise<Boolean>:
    Optionally validate entity object obj (of entity type type) just before create or update operations. If the resulting Promise is rejected, the create or update operation fails. The ctx object is just passed through from the GraphQL.graphql() call.

  • authorizer(moment: String, op: String, type: String, obj: Object, ctx: Object): Promise<Boolean>:
    Optionally authorize entity object obj (of entity type type) for operation op (create, read, update or delete) at moment (before or after). Notice that for read there is no before and for delete there is no after, of course. The ctx object is just passed through from the GraphQL.graphql() call. If the resulting Promise is rejected or returns false, the operation fails.

  • tracer(record: Object, ctx: Object): Promise<any>:
    Optionally trace the operation via the action record. The fields of record are: { op: String, arity: String, dstType: String, dstIds: String[], dstAttrs: String[] }. The ctx object is just passed through from the GraphQL.graphql() call.

  • fts: { [String]: String[] }:
    Enables the Full-Text-Search (FTS) mechanism for all configured entity types and their listed attributes.

  • idname: String = "id":
    Configures the GraphQL name of the mandatory unique identifier attribute on each entity. The default name is id.

  • idtype: String = "UUID":
    Configures the GraphQL type of the mandatory unique identifier attribute on each entity. The default type UUID assumes that you define the GraphQL scalar type UUID with the help of GraphQL-Tools-Types.

  • idmake: Function = () => (new UUID(1)).format():
    Configures a function for generating unique identifiers for the mandatory unique identifier attribute on each entity. The default uses pure-uuid to generate UUIDs of version 1.

  • hcname: String = "hc":
    Configures the GraphQL name of the optional hash-code attribute on each entity. The default name is hc. This attribute is NOT persisted and instead calculated on the fly and is intended to be used for optimistic locking purposes.

  • hctype: String = "UUID":
    Configures the GraphQL type of the optional hash-code attribute on each entity. The default type UUID assumes that you define the GraphQL scalar type UUID with the help of GraphQL-Tools-Types.

  • hcmake: Function = (data) => (new UUID(5, "ns:URL", \uri:gts:${data}`)).format()`:
    Configures a function for generating hash-codes for the optional hash-code attribute on each entity. The default uses pure-uuid to generate UUIDs of version 5.

• gts.boot(): Promise:
  

  Bootstrap the GraphQL-Tools-Sequelize instance. It internally mainly initialized the Full-Text-Search (FTS) mechanism.

• gts.attrIdSchema(source: String): String,
  gts.attrIdResolver(source: String): Function:
  

  Generate a GraphQL schema entry and a corresponding GraphQL resolver function for querying the mandatory unique identifier attribute of an entity of type source. By default this generates a schema entry <idname>: <idtype> and a resolver which just returns <object>[<idname>]. This mandatory unique identifier attribute has to be persisted and hence part of the Sequelize schema definition.

• gts.attrHcSchema(source: String): String,
  gts.attrHcResolver(source: String): Function:
  

  Generate a GraphQL schema entry and a corresponding GraphQL resolver function for querying the optional hash-code attribute of an entity of type source. By default this generates a schema entry <hcname>: <hctype> and a resolver which returns something like <hcmake>(data(<object>)), where data() is an internal function which deterministically concatenates the values of all attributes of <object>. This optional hash-code attribute has NOT to be persisted and hence SHOULD NOT BE part of the Sequelize schema definition.

• gts.entityQuerySchema(source: String, relation: String, target: String): String,
  gts.entityQueryResolver(source: String, relation: String, target: String): Function:
  

  Generate a GraphQL schema entry and a corresponding GraphQL resolver function for querying one, many or all entities of particular entity type target when coming from entity type source -- either directly (in case relation is the empty string) or via relationship relation. The target is either just the name foo of an entity type foo (for relationship of cardinality 0..1) or foo* (for relationship of cardinality 0..N). Based on the combination of relation and the cardinality of target, four distinct GraphQL schema entries (and corresponding GraphQL resolver functions) are generated:

  • empty relation and target cardinality 0..1:
    

    1`# Query one [${target}]() entity by its unique identifier (\`id\`) or condition (\`where\`) or` +
    2`# open an anonymous context for the [${target}]() entity.\n` +
    3`# The [${target}]() entity can be optionally filtered by a condition on some relationships (\`include\`).\n` +
    4`${target}(id: ${idtype}, where: JSON, include: JSON): ${target}\n`

  • empty relation and target cardinality 0..N:
    

    1`# Query one or many [${target}]() entities,\n` +
    2"# by either an (optionally available) full-text-search (`query`)\n" +
    3"# or an (always available) attribute-based condition (`where`),\n" +
    4"# optionally filter them by a condition on some relationships (`include`),\n" +
    5"# optionally sort them (`order`),\n" +
    6"# optionally start the result set at the n-th entity (zero-based `offset`), and\n" +
    7"# optionally reduce the result set to a maximum number of entities (`limit`).\n" +
    8`${target}s(fts: String, where: JSON, include: JSON, order: JSON, offset: Int = 0, limit: Int = 100): [${target}]!\n`

  • non-empty relation and target cardinality 0..1:
    

    1`# Query one [${target}]() entity by following the **${relation}** relation of [${source}]() entity.\n` +
    2`# The [${target}]() entity can be optionally filtered by a condition (\`where\`).\n` +
    3`# The [${target}]() entity can be optionally filtered by a condition on some relationships (\`include\`).\n` +
    4`${relation}(where: JSON, include: JSON): ${target}\n`

  • non-empty relation and target cardinality 0..N:
    

    1`# Query one or many [${target}]() entities\n` +
    2`# by following the **${relation}** relation of [${source}]() entity,\n` +
    3"# optionally filter them by a condition (`where`),\n" +
    4"# optionally filter them by a condition on some relationships (`include`),\n" +
    5"# optionally sort them (`order`),\n" +
    6"# optionally start the result set at the n-th entity (zero-based `offset`), and\n" +
    7"# optionally reduce the result set to a maximum number of entities (`limit`).\n" +
    8`${relation}(where: JSON, include: JSON, order: JSON, offset: Int = 0, limit: Int = 100): [${target}]!\n`

  The comments are intentionally also generated, as they document the entries in the GraphQL schema and are visible through GraphQL schema introspection tools like GraphiQL.

• gts.entity{Create,Clone,Update,Delete}Schema(type: String): String,
  gts.entity{Create,Clone,Update,Delete}Resolver(type: String): Function:
  

  Generate a GraphQL schema entry and a corresponding GraphQL resolver function for mutating one, many or all entities of particular entity type type. The following GraphQL schema entries (and corresponding GraphQL resolver functions) are generated:

  • For entityCreate{Schema,Resolver}(type):
    

    1`# Create new [${type}]() entity, optionally with specified attributes (\`with\`).\n` +
    2`create(id: ${idtype}, with: JSON): ${type}!\n`

  • For entityClone{Schema,Resolver}(type):
    

    1`# Clone one [${type}]() entity by cloning its attributes (but not its relationships).\n` +
    2`clone: ${type}!\n`

  • For entityUpdate{Schema,Resolver}(type):
    

    1`# Update one [${type}]() entity with specified attributes (\`with\`).\n` +
    2`update(with: JSON!): ${type}!\n`

  • For entityDelete{Schema,Resolver}(type):
    

    1`# Delete one [${type}]() entity.\n` +
    2`delete: ${idtype}!\n`

  The comments are intentionally also generated, as they document the entries in the GraphQL schema and are visible through GraphQL schema introspection tools like GraphiQL.

Assumptions

It is assumed that all your Sequelize entities have an attribute id (see also idname configuration option) which is the (technical) primary key of an entity. By default, the type of field id is UUID, but this can be overridden (see idtype configuration option). In case of the type UUID, it is assumed that you define the GraphQL scalar type UUID with the help of GraphQL-Tools-Types.

Notice: all entities are required to have the field id and the type of all id fields have to be the same. But this does not prevent you from having additional domain-specific primary keys per entity of an arbitrary type, of course. GraphQL-Tools-Sequelize just uses the field id for its functionality.

In addition, the scalar type JSON always has to be defined with the help of GraphQL-Tools-Types.

License

Copyright (c) 2016-2023 Dr. Ralf S. Engelschall (http://engelschall.com/)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.