gql-types-generator
Package to generate TypeScript types depending on GraphQL scheme, mutations and
queries.
Currently supported GraphQL entities: type
, interface
, input
,
type and interface fields arguments
, operations
(of any kind and selections).
GraphQL entities to be supported in near future: fragments
, inline fragments
,
directives
.
TLDR;
If you dont want to read a lot of text and documentation, try to compile
your scheme and look what happens. You can find an example
here.
Just use command in command
file.
Install
npm install --save gql-types-generator
yarn add gql-types-generator
Usage
Command line interface
After installation of package is done, gql-types-generator
command
becomes available. Type to get help:
$ gql-types-generator --help
Usage: gql-types-generator [options] <schema-globs>
Options:
--operations <globs> globs to find queries and mutations
--operations-file <filename> operations file name. If passed, all operations will be placed into a single file
--operations-wrap wraps operations with graphql-tag, making exports from operations not strings, but graphql's DocumentNode (default: false)
--operations-selection-separate creates separated types for each selection set (default: false)
--schema-file <filename> schema file name
--remove-description states if description should be removed (default: false)
--display <sort> how to display compiled types. Valid values are "as-is" and "default". By default, generator compiles scalars first, then enums, interfaces, inputs, unions and then types. "as-is" places types as they are placed in schema
(default: "default")
--scalars <scalars> defines scalars types. Must be a JSON, where key is scalar name and value is its type
--output-directory <path> path to directory where typings will be saved
-h, --help display help for command
When using CLI, each glob will be formatted as process.cwd()
+ glob. You can
pass an array of globs using comma between them like
src/schema1.graphql,src/schema2.graphql
.
If an error occurring during types compilation, library returns code 1.
Programmatic control
Library provides such functions as compile
, compileSchema
and
compileOperations
to generate types. Type definitions for
these functions.
compile(options: CompileOptions)
Name | Type | Description |
---|
options.outputDirectory | string | Full path to output directory |
options.removeDescription | boolean? | Should library remove descriptions |
options.display | DisplayType? | How to display compiled types. Valid values are "as-is" and "default". By default, generator compiles scalars first, then enums, interfaces, inputs, unions and then types. "as-is" places types as they are placed in schema |
options.schemaPath | PathType | Defines paths to schema |
options.operationsPath | PathType? | Defines paths to operations |
options.schemaFileName | string? | Defines schema file name. For example - schema.ts |
options.operationsFileName | string? | Defines operations file name. For example - operation.ts . If passed, all operations will be placed into a single file |
options.operationsWrap | boolean? | States of compiled types should be graphql s DocumentNode and not string |
options.scalars | ScalarsMap? | Defines types of scalars |
Example
import {compile} from 'gql-types-generator';
import * as path from 'path';
compile({
outputDirectory: path.resolve(__dirname, 'compiled'),
removeDescription: false,
display: 'as-is',
operationsPath: {
glob: {
cwd: process.cwd(),
glob: 'gql/operations/*.graphql'
}
},
schemaPath: {
path: [
path.resolve(__dirname, 'gql/schema/part1.graphql'),
path.resolve(__dirname, 'gql/schema/part2.graphql'),
]
},
schemaPath: {
glob: {
cwd: process.cwd(),
glob: 'gql/schema/*.graphql'
}
},
schemaPath: {
definition: 'type Query { ... }'
},
operationsPath: {
path: path.resolve(__dirname, 'gql/getUsers.graphql'),
},
schemaFileName: 'my-compiled-schema.ts',
operationsFileName: 'my-compiled-operations.ts',
operationsWrap: true,
scalars: {
MyCustomScalar: 'Date',
AnotherScalar: 'number | string | Record<string, string>',
AndAnotherOneScalar: '"string literal"',
}
});
compileSchema(options: CompileSchemaOptions)
Name | Type | Description |
---|
options.schema | string | Schema definition |
options.outputDirectory | string | Full path to output directory |
options.fileName | string? | Output schema file name |
options.display | DisplayType? | How to display compiled types. Valid values are "as-is" and "default". By default, generator compiles scalars first, then enums, interfaces, inputs, unions and then types. "as-is" places types as they are placed in schema |
options.removeDescription | boolean? | Should library remove descriptions |
options.scalars | ScalarsMap? | Defines types of scalars |
Example
import {compileSchema} from 'gql-types-generator';
import * as path from 'path';
compileSchema({
schema: 'type Query { ... }',
outputDirectory: path.resolve(__dirname, 'gql/compiled'),
fileName: 'my-compiled-schema.ts',
display: 'default',
removeDescription: true,
});
compileOperations(options: CompileOperationsOptions)
Name | Type | Description |
---|
options.operations | string | Operations definition |
options.outputDirectory | string | Full path to output directory |
options.schema | GraphQLSchema | Built GQL schema |
options.schemaFileName | string | Schema file name. Used to pass in relative imports if they are required |
options.removeDescription | boolean? | Should library remove descriptions |
options.fileName | string? | Output operations file name. If passed, all operations will be placed into a single file |
options.wrapWithTag | boolean? | States of compiled types should be graphql s DocumentNode and not string |
Example
import {compileOperations} from 'gql-types-generator';
import * as path from 'path';
compileOperations({
operations: 'query getUser() { ... } mutation register() { ... }',
outputDirectory: path.resolve(__dirname, 'gql/compiled'),
schema: gqlSchema,
schemaFileName: 'my-compiled-schema.ts',
removeDescription: true,
fileName: 'my-compiled-operations.ts',
wrapWithTag: false,
});
Compilation result
Schema
Command creates a directory on passed --output-directory
path, generates
d.ts
definition file and compiled js
code. Name of these files are taken
from --schema-file
argument (schema.ts
by default).
d.ts
contains all schema types and by default exports constant schema
which is a text representation of schema.
Each schema type definition consists of interface
and namespace
with the
same name. All interface fields refers to namespace fields. So, if you want
to get some Query
field type you could use Query['someField']
or Query.someField
. They return the same thing. It is recommended
to use Query.*
-like syntax for better experience.
If --scalars
passed, compiled type of scalar will be taken from this map.
If scalar not found, it will be any
. Must be a JSON with values of type
string or number.
Operations
To compile operations, it is required to use --operations
argument. This
values must be a glob which refers to files where operations are defined.
Library creates single file with name --operations-file
if it is passed
or 2 separate files d.ts
and js
for each command with name computed as
operationName + toCamelCase(operationType)
in directory on passed
--output-directory
. So, if operation was query getUsers { ... }
,
created files will be getUsersQuery.d.ts
and getUsersQuery.js
.
d.ts
exports selection and namespace with Arguments
if they exist.
Additionally namespace contains subselections represented as other
namespacesjs
exports representation of operation
If --operations-wrap
passed, wraps each operation string with graphql-tag
package making each operation not string, but graphql
s Document Node
.
Useful when you use these operations on frontend with Apollo client.