rf(json-schema): Factor JSON schema validation out of context.dataset

bids-validator/src/schema/applyRules.ts

@@ -4,6 +4,7 @@ import { BIDSContext } from './context.ts'
 import { expressionFunctions } from './expressionLanguage.ts'
 import { logger } from '../utils/logger.ts'
 import { memoize } from '../utils/memoize.ts'
+import { compile } from '../validators/json.ts'
 
 /**
  * Given a schema and context, evaluate which rules match and test them.
@@ -467,7 +468,7 @@ function evalJsonCheck(
       return
     }
 
-    const validate = context.dataset.ajv.compile(metadataDef)
+    const validate = compile(metadataDef)
     const result = validate(context.sidecar[keyName])
     if (result === false) {
       const evidenceBase = `Failed for this file.key: ${originFileKey} Schema path: ${schemaPath}`

bids-validator/src/schema/context.ts

@@ -16,9 +16,6 @@ import { loadHeader } from '../files/nifti.ts'
 import { buildAssociations } from './associations.ts'
 import { ValidatorOptions } from '../setup/options.ts'
 import { logger } from '../utils/logger.ts'
-import { Ajv, JSONSchemaType, ValidateFunction } from '../deps/ajv.ts'
-import { memoize } from '../utils/memoize.ts'
-import { Schema } from '../types/schema.ts'
 
 export class BIDSContextDataset implements ContextDataset {
   dataset_description: Record<string, unknown>
@@ -28,22 +25,15 @@ export class BIDSContextDataset implements ContextDataset {
   ignored: any[]
   modalities: any[]
   subjects?: ContextDatasetSubjects
-  ajv: Ajv
   sidecarKeyValidated: Set<string>
 
-  constructor(options?: ValidatorOptions, schema?: Schema, description = {}) {
+  constructor(options?: ValidatorOptions, description = {}) {
     this.dataset_description = description
     this.files = []
     this.tree = {}
     this.ignored = []
     this.modalities = []
-    this.ajv = new Ajv({ strictSchema: false })
-    // @ts-expect-error
-    this.ajv.compile = memoize(this.ajv.compile)
     this.sidecarKeyValidated = new Set<string>()
-    if (schema) {
-      this.setCustomAjvFormats(schema)
-    }
     if (options) {
       this.options = options
     }
@@ -56,27 +46,6 @@ export class BIDSContextDataset implements ContextDataset {
       this.dataset_description.DatasetType = 'raw'
     }
   }
-
-  setCustomAjvFormats(schema: Schema): void {
-    if (typeof schema.objects.formats !== 'object') {
-      // logger.warning(
-      console.log(
-        `schema.objects.formats missing from schema, format validation disabled.`,
-      )
-      return
-    }
-    const schemaFormats = schema.objects.formats
-    for (let key of Object.keys(schemaFormats)) {
-      if (typeof schemaFormats[key]['pattern'] !== 'string') {
-        // logger.warning(
-        console.log(
-          `schema.objects.formats.${key} pattern missing or invalid. Skipping this format for addition to context json validator`,
-        )
-        continue
-      }
-      this.ajv.addFormat(key, schemaFormats[key]['pattern'])
-    }
-  }
 }
 
 export class BIDSContextDatasetSubjects implements ContextDatasetSubjects {

bids-validator/src/setup/loadSchema.ts

@@ -3,6 +3,7 @@ import { objectPathHandler } from '../utils/objectPathHandler.ts'
 import * as schemaDefault from 'https://bids-specification.readthedocs.io/en/latest/schema.json' assert {
   type: 'json',
 }
+import { setCustomMetadataFormats } from '../validators/json.ts'
 
 /**
  * Load the schema from the specification
@@ -18,11 +19,12 @@ export async function loadSchema(version = 'latest'): Promise<Schema> {
   } else if (version === 'latest' || versionRegex.test(version)) {
     schemaUrl = `https://bids-specification.readthedocs.io/en/${version}/schema.json`
   }
+  let schema: Schema | undefined = undefined
   try {
     const schemaModule = await import(/* @vite-ignore */ schemaUrl, {
       assert: { type: 'json' },
     })
-    return new Proxy(
+    schema = new Proxy(
       schemaModule.default as object,
       objectPathHandler,
     ) as Schema
@@ -32,9 +34,11 @@ export async function loadSchema(version = 'latest'): Promise<Schema> {
     console.error(
       `Warning, could not load schema from ${schemaUrl}, falling back to internal version`,
     )
-    return new Proxy(
+    schema = new Proxy(
       schemaDefault.default as object,
       objectPathHandler,
     ) as Schema
   }
+  setCustomMetadataFormats(schema)
+  return schema
 }

bids-validator/src/types/context.ts

@@ -1,5 +1,4 @@
 import { ValidatorOptions } from '../setup/options.ts'
-import { Ajv } from '../deps/ajv.ts'
 
 export interface ContextDatasetSubjects {
   sub_dirs: string[]
@@ -14,7 +13,6 @@ export interface ContextDataset {
   modalities: any[]
   subjects?: ContextDatasetSubjects
   options?: ValidatorOptions
-  ajv: Ajv
   sidecarKeyValidated: Set<string>
 }
 export interface ContextSubjectSessions {

bids-validator/src/validators/bids.ts

@@ -48,9 +48,9 @@ export async function validate(
   if (ddFile) {
     const description = await ddFile.text().then((text) => JSON.parse(text))
     summary.dataProcessed = description.DatasetType === 'derivative'
-    dsContext = new BIDSContextDataset(options, schema, description)
+    dsContext = new BIDSContextDataset(options, description)
   } else {
-    dsContext = new BIDSContextDataset(options, schema)
+    dsContext = new BIDSContextDataset(options)
     issues.addNonSchemaIssue('MISSING_DATASET_DESCRIPTION', [] as IssueFile[])
   }
 

bids-validator/src/validators/json.ts

@@ -0,0 +1,29 @@
+import { Ajv, JSONSchemaType, ValidateFunction } from '../deps/ajv.ts'
+import { Schema } from '../types/schema.ts'
+import { memoize } from '../utils/memoize.ts'
+import { logger } from '../utils/logger.ts'
+
+const metadataValidator = new Ajv({ strictSchema: false })
+
+// Bind the method to the instance before memoizing to avoid losing the context
+export const compile = memoize(metadataValidator.compile.bind(metadataValidator))
+
+export function setCustomMetadataFormats(schema: Schema): void {
+  if (typeof schema.objects.formats !== 'object') {
+    logger.warning(
+      `schema.objects.formats missing from schema, format validation disabled.`,
+    )
+    return
+  }
+  const schemaFormats = schema.objects.formats
+  for (let key of Object.keys(schemaFormats)) {
+    const pattern = schemaFormats[key]['pattern']
+    if (typeof pattern !== 'string') {
+      logger.warning(
+        `schema.objects.formats.${key} pattern missing or invalid. Skipping this format for addition to context json validator`,
+      )
+      continue
+    }
+    metadataValidator.addFormat(key, pattern)
+  }
+}