json-schema-cycles
Detect cyclic dependencies in a JSON schema (or more precisely a JSON schema with the definitions
property for the individual type definitions). This package doesn't handle external references, you need to use a ref-parser and squash all types into one schema before using this package.
There are two types of analysis functions, analyzeTypes
and analyzeTypesFast
. The former provides a more in-depth analysis but can be slow on large input. If the input is extremely complex with massive cyclicity, it might even crash the process or run out of memory.
Unless an in-depth analysis is necessary, choose the analyzeTypesFast
function.
Versions
- Since v2 this is a pure ESM package, and requires Node.js >=12.20. It cannot be used from CommonJS.
Example
analyzeTypes
takes a JSON schema object and returns a result object on the same format as FullAnalysisResult
in graph-cycles but with an additional graph
property containing the type dependency graph as an array of [ from: string, to: Array< string > ]
where from
is the type and to
is the dependent types.
analyzeTypesFast
also takes a JSON schema object, but returns an object on the same format as FastAnalysisResult
in graph-cycles, with the same additional graph
property as from analyzeTypes
.
import { analyzeTypes, analyzeTypesFast } from 'json-schema-cycles'
const { cycles, entrypoints, dependencies, dependents, all, graph } = analyzeTypes( schemaObject );
const { cyclic, dependencies, dependents, graph } = analyzeTypesFast( schemaObject );
Check graph-cycles for an understanding of the result object, apart from graph
.
Given the following JSON Schema:
const jsonSchema = {
definitions: {
Link: {},
Subscriber: {
type: 'object',
properties: {
user: { $ref: '#/definitions/User' },
},
},
Message: {
type: 'object',
properties: {
replyTo: { $ref: '#/definitions/Message' },
link: { $ref: '#/definitions/Link' },
subscriber: { $ref: '#/definitions/Subscriber' },
},
},
User: {
type: 'object',
properties: {
parent: { $ref: '#/definitions/User' },
lastMessage: { $ref: '#/definitions/Message' },
},
},
DM: {
type: 'object',
properties: {
lastUser: { $ref: '#/definitions/User' },
},
},
Actions: {
type: 'object',
properties: {
dms: {
type: 'array',
items: { $ref: '#/definitions/DM' },
},
},
},
Product: {},
Cart: {
type: 'array',
items: { $ref: '#/definitions/Product' },
},
}
};
... which can be viewed as:
Cart -> Product
__ ___
| | | |
v | v |
Action -> DM -> User -> Message -> Link
^ |
| v
`--- Subscriber
for a full analysis, the result will be (of the type TypeAnalysisFullResult
):
{
entrypoints: [
[ 'DM' ],
[ 'Actions', 'DM' ],
],
cycles: [
[ 'User' ],
[ 'Message' ],
[ 'User', 'Message', 'Subscriber' ],
],
all: [ 'User', 'Message', 'DM', 'Actions', 'Subscriber' ],
dependencies: [ 'Link' ],
dependents: [ ],
graph: [
[ 'Link', [ ] ],
[ 'Subscriber', [ 'User' ] ],
[ 'Message', [ 'Message', 'Link', 'Subscriber' ] ],
[ 'User', [ 'Message', 'User' ] ],
[ 'DM', [ 'User' ] ],
[ 'Actions', [ 'DM' ] ],
[ 'Product', [ ] ],
[ 'Cart', [ 'Product' ] ],
],
}
for a fast analysis, the result will be (of the type TypeAnalysisFastResult
):
{
cyclic: [ 'User', 'Message', 'DM', 'Actions', 'Subscriber' ],
dependencies: [ 'Link' ],
dependents: [ ],
graph: [
[ 'Link', [ ] ],
[ 'Subscriber', [ 'User' ] ],
[ 'Message', [ 'Message', 'Link', 'Subscriber' ] ],
[ 'User', [ 'Message', 'User' ] ],
[ 'DM', [ 'User' ] ],
[ 'Actions', [ 'DM' ] ],
[ 'Product', [ ] ],
[ 'Cart', [ 'Product' ] ],
],
}
Helpers
Two helper functions are exported; sortTypeAnalysisFullResult
and sortTypeAnalysisFastResult
. They take an analysis result (of type TypeAnalysisFullResult
or TypeAnalysisFastResult
) and return a new object of the same type, with all fields sorted in a deterministic way, which is useful in tests.