Merge pull request #1977 from effigies/feat/subjects

feat(schema): Load subject data

bids-validator/src/schema/context.test.ts

@@ -88,7 +88,7 @@ rootFileTree.files = [
 ]
 rootFileTree.directories = [subjectFileTree]
 
-const context = new BIDSContext(anatFileTree, dataFile, issues)
+let context = new BIDSContext(anatFileTree, dataFile, issues)
 
 Deno.test('test context LoadSidecar', async (t) => {
   await context.loadSidecar(rootFileTree)
@@ -106,3 +106,16 @@ Deno.test('test context LoadSidecar', async (t) => {
     assert(anatValue, 'anat')
   })
 })
+
+context = new BIDSContext(rootFileTree, dataFile, issues)
+
+Deno.test('test context loadSubjects', async (t) => {
+  await context.loadSubjects()
+  await t.step('context produces correct subjects object', () => {
+    assert(context.dataset.subjects, "subjects object exists")
+    assert(context.dataset.subjects.sub_dirs.length == 1, "there is only one sub dir found")
+    assert(context.dataset.subjects.sub_dirs[0] == 'sub-01', "that sub dir is sub-01")
+    // no participants.tsv so this should be empty
+    assert(context.dataset.subjects.participant_id == undefined, "no participant_id is populated")
+  })
+})

bids-validator/src/schema/context.ts

@@ -24,15 +24,14 @@ export class BIDSContextDataset implements ContextDataset {
   tree: object
   ignored: any[]
   modalities: any[]
-  subjects: ContextDatasetSubjects[]
+  subjects?: ContextDatasetSubjects
 
   constructor(options?: ValidatorOptions, description = {}) {
     this.dataset_description = description
     this.files = []
     this.tree = {}
     this.ignored = []
     this.modalities = []
-    this.subjects = [] as ContextDatasetSubjects[]
     if (options) {
       this.options = options
     }
@@ -47,6 +46,22 @@ export class BIDSContextDataset implements ContextDataset {
   }
 }
 
+export class BIDSContextDatasetSubjects implements ContextDatasetSubjects {
+  sub_dirs: string[]
+  participant_id?: string[]
+  phenotype?: string[]
+
+  constructor(
+    sub_dirs?: string[],
+    participant_id?: string[],
+    phenotype?: string[],
+  ) {
+    this.sub_dirs = sub_dirs ? sub_dirs : []
+    this.participant_id = participant_id
+    this.phenotype = phenotype
+  }
+}
+
 const defaultDsContext = new BIDSContextDataset()
 
 export class BIDSContext implements Context {
@@ -200,8 +215,54 @@ export class BIDSContext implements Context {
       .catch((error) => {})
   }
 
+  // This is currently done for every file. It should be done once for the dataset.
+  async loadSubjects(): Promise<void> {
+    if (this.dataset.subjects != null) {
+      return
+    }
+    this.dataset.subjects = new BIDSContextDatasetSubjects()
+    // Load subject dirs from the file tree
+    this.dataset.subjects.sub_dirs = this.fileTree.directories
+      .filter((dir) => dir.name.startsWith('sub-'))
+      .map((dir) => dir.name)
+
+    // Load participants from participants.tsv
+    const participants_tsv = this.fileTree.files.find(
+      (file) => file.name === 'participants.tsv',
+    )
+    if (participants_tsv) {
+      const participantsText = await participants_tsv.text()
+      const participantsData = parseTSV(participantsText)
+      this.dataset.subjects.participant_id = participantsData[
+        'participant_id'
+      ] as string[]
+    }
+
+    // Load phenotype from phenotype/*.tsv
+    const phenotype_dir = this.fileTree.directories.find(
+      (dir) => dir.name === 'phenotype',
+    )
+    if (phenotype_dir) {
+      const phenotypeFiles = phenotype_dir.files.filter((file) =>
+        file.name.endsWith('.tsv'),
+      )
+      // Collect observed participant_ids
+      const seen = new Set() as Set<string>
+      for (const file of phenotypeFiles) {
+        const phenotypeText = await file.text()
+        const phenotypeData = parseTSV(phenotypeText)
+        const participant_id = phenotypeData['participant_id'] as string[]
+        if (participant_id) {
+          participant_id.forEach((id) => seen.add(id))
+        }
+      }
+      this.dataset.subjects.phenotype = Array.from(seen)
+    }
+  }
+
   async asyncLoads() {
     await Promise.allSettled([
+      this.loadSubjects(),
       this.loadSidecar(),
       this.loadColumns(),
       this.loadAssociations(),

bids-validator/src/schema/expressionLanguage.ts

@@ -88,7 +88,10 @@ export const expressionFunctions = {
     return arg.substr(start, end - start)
   },
   sorted: <T>(list: T[]): T[] => {
-    list.sort()
-    return list
+    // Copy, sort, return
+    return list.slice().sort()
+  },
+  allequal: <T>(a: T[], b: T[]): boolean => {
+    return (a != null && b != null) && a.length === b.length && a.every((v, i) => v === b[i])
   },
 }

bids-validator/src/types/context.ts

@@ -2,16 +2,16 @@ import { ValidatorOptions } from '../setup/options.ts'
 
 export interface ContextDatasetSubjects {
   sub_dirs: string[]
-  participant_id: string[]
-  phenotype: string[]
+  participant_id?: string[]
+  phenotype?: string[]
 }
 export interface ContextDataset {
   dataset_description: Record<string, unknown>
   files: any[]
   tree: object
   ignored: any[]
   modalities: any[]
-  subjects: ContextDatasetSubjects[]
+  subjects?: ContextDatasetSubjects
   options?: ValidatorOptions
 }
 export interface ContextSubjectSessions {