Merge pull request #272 from effigies/fix/datatype

feat: Load datatype and modality when building file context

Author: rwblair

src/schema/context.ts

@@ -15,6 +15,7 @@ import type { BIDSFile } from '../types/filetree.ts'
 import { FileTree } from '../types/filetree.ts'
 import { ColumnsMap } from '../types/columns.ts'
 import { readEntities } from './entities.ts'
+import { findDatatype } from './datatypes.ts'
 import { DatasetIssues } from '../issues/datasetIssues.ts'
 import { walkBack } from '../files/inheritance.ts'
 import { parseGzip } from '../files/gzip.ts'
@@ -147,16 +148,20 @@ export class BIDSContext implements Context {
     dsContext?: BIDSContextDataset,
     fileTree?: FileTree,
   ) {
+    this.dataset = dsContext ? dsContext : new BIDSContextDataset({ tree: fileTree })
+
     this.filenameRules = []
     this.file = file
-    const bidsEntities = readEntities(file.name)
-    this.suffix = bidsEntities.suffix
-    this.extension = bidsEntities.extension
-    this.entities = bidsEntities.entities
-    this.dataset = dsContext ? dsContext : new BIDSContextDataset({ tree: fileTree })
+
+    const { entities, suffix, extension } = readEntities(file.name)
+    const { datatype, modality } = findDatatype(file, this.dataset.schema)
+    this.entities = entities
+    this.suffix = suffix
+    this.extension = extension
+    this.datatype = datatype
+    this.modality = modality
+
     this.subject = {} as Subject
-    this.datatype = ''
-    this.modality = ''
     this.sidecar = {}
     this.sidecarKeyOrigin = {}
     this.columns = new ColumnsMap() as Record<string, string[]>

src/schema/datatypes.test.ts

@@ -0,0 +1,57 @@
+import { assert, assertObjectMatch } from '@std/assert'
+import { loadSchema } from '../setup/loadSchema.ts'
+import { BIDSContext, BIDSContextDataset } from '../schema/context.ts'
+import { pathsToTree } from '../files/filetree.ts'
+import type { BIDSFile } from '../types/filetree.ts'
+
+import { findDatatype, modalityTable } from './datatypes.ts'
+
+const schema = await loadSchema()
+
+// Creates a file object as part of a minimal file tree
+const makeFile = (path: string): BIDSFile => pathsToTree([path]).get(path) as BIDSFile
+
+Deno.test('test modalityTable', async (t) => {
+  await t.step('empty schema', () => {
+    assertObjectMatch(modalityTable({}), {})
+  })
+
+  await t.step('real schema', async () => {
+    const table = modalityTable(schema)
+    // Memoization check
+    assert(modalityTable(schema) == table)
+    // spot check
+    assertObjectMatch(table, {
+      'anat': 'mri',
+      'perf': 'mri',
+      'eeg': 'eeg',
+    })
+  })
+})
+
+Deno.test('test findDatatype', async (t) => {
+  await t.step('root files', async () => {
+    const path = makeFile('/participants.tsv')
+    assertObjectMatch(findDatatype(path, schema), { datatype: '', modality: '' })
+  })
+  await t.step('non-datatype parent', async () => {
+    const path = makeFile('/stimuli/image.png')
+    assertObjectMatch(findDatatype(path, schema), { datatype: '', modality: '' })
+  })
+  await t.step('phenotype file', async () => {
+    const path = makeFile('/phenotype/survey.tsv')
+    assertObjectMatch(findDatatype(path, schema), { datatype: 'phenotype', modality: '' })
+  })
+  await t.step('data files', async () => {
+    for (
+      const [filename, datatype, modality] of [
+        ['/sub-01/anat/sub-01_T1w.nii.gz', 'anat', 'mri'],
+        ['/sub-01/eeg/sub-01_task-rest_eeg.edf', 'eeg', 'eeg'],
+        ['/sub-01/perf/sub-01_task-rest_bold.nii.gz', 'perf', 'mri'],
+      ]
+    ) {
+      const path = makeFile(filename)
+      assertObjectMatch(findDatatype(path, schema), { datatype, modality })
+    }
+  })
+})

src/schema/datatypes.ts

@@ -0,0 +1,30 @@
+import type { BIDSFile } from '../types/filetree.ts'
+import type { Schema } from '../types/schema.ts'
+import { memoize } from '../utils/memoize.ts'
+
+function _modalityTable(schema: Schema): Record<string, string> {
+  const modalities: Record<string, string> = {}
+  const rules = (schema.rules?.modalities ?? {}) as Record<string, { datatypes: string[] }>
+  for (const [modality, { datatypes }] of Object.entries(rules)) {
+    for (const datatype of datatypes) {
+      modalities[datatype] = modality
+    }
+  }
+  return modalities
+}
+
+// Construct once per schema; should only be multiple in tests
+export const modalityTable = memoize(_modalityTable)
+
+export function findDatatype(
+  file: BIDSFile,
+  schema: Schema,
+): { datatype: string; modality: string } {
+  const lookup = modalityTable(schema)
+  const datatype = file.parent?.name
+  if (!schema?.objects?.datatypes[datatype]) {
+    return { datatype: '', modality: '' }
+  }
+  const modality = lookup[datatype] ?? ''
+  return { datatype, modality }
+}

src/schema/modalities.ts

@@ -1,12 +1,7 @@
 import { assertEquals } from '@std/assert'
 import { SEPARATOR_PATTERN } from '@std/path'
 import { BIDSContext } from '../schema/context.ts'
-import {
-  _findRuleMatches,
-  datatypeFromDirectory,
-  findDirRuleMatches,
-  hasMatch,
-} from './filenameIdentify.ts'
+import { _findRuleMatches, findDirRuleMatches, hasMatch } from './filenameIdentify.ts'
 import { BIDSFileDeno } from '../files/deno.ts'
 import { FileIgnoreRules } from '../files/ignore.ts'
 import { loadSchema } from '../setup/loadSchema.ts'
@@ -50,19 +45,6 @@ Deno.test('test _findRuleMatches', async (t) => {
   )
 })
 
-Deno.test('test datatypeFromDirectory', (t) => {
-  const filesToTest = [
-    ['/sub-01/ses-01/func/sub-01_ses-01_task-nback_run-01_bold.nii', 'func'],
-    ['/sub-01/ses-01/anat/sub-01_ses-01_T1w.nii', 'anat'],
-  ]
-  filesToTest.map((test) => {
-    const file = new BIDSFileDeno(PATH, test[0], ignore)
-    const context = new BIDSContext(file)
-    datatypeFromDirectory(schema, context)
-    assertEquals(context.datatype, test[1])
-  })
-})
-
 Deno.test('test hasMatch', async (t) => {
   await t.step('hasMatch', async () => {
     const fileName = '/sub-01/ses-01/func/sub-01_ses-01_task-nback_run-01_bold.nii'

src/validators/filenameIdentify.test.ts

@@ -15,12 +15,10 @@
 import { globToRegExp, SEPARATOR_PATTERN } from '@std/path'
 import type { GenericSchema, Schema } from '../types/schema.ts'
 import type { BIDSContext } from '../schema/context.ts'
-import type { lookupModality } from '../schema/modalities.ts'
 import type { CheckFunction } from '../types/check.ts'
 import { lookupEntityLiteral } from './filenameValidate.ts'
 
 const CHECKS: CheckFunction[] = [
-  datatypeFromDirectory,
   findRuleMatches,
   hasMatch,
   cleanContext,
@@ -126,29 +124,6 @@ function matchStemRule(node, context): boolean {
   return true
 }
 
-export async function datatypeFromDirectory(schema, context) {
-  const subEntity = schema.objects.entities.subject.name
-  const sesEntity = schema.objects.entities.session.name
-  const parts = context.file.path.split(SEPARATOR_PATTERN)
-  const datatypeIndex = parts.length - 2
-  if (datatypeIndex < 1) {
-    return Promise.resolve()
-  }
-  const dirDatatype = parts[datatypeIndex]
-  if (dirDatatype === 'phenotype') {
-    // Phenotype is a pseudo-datatype for now.
-    context.datatype = dirDatatype
-    return Promise.resolve()
-  }
-  for (const key in schema.rules.modalities) {
-    if (schema.rules.modalities[key].datatypes.includes(dirDatatype)) {
-      context.modality = key
-      context.datatype = dirDatatype
-      return Promise.resolve()
-    }
-  }
-}
-
 export function hasMatch(schema, context) {
   if (
     context.filenameRules.length === 0 &&

src/validators/filenameIdentify.ts

@@ -1,19 +1,19 @@
 import { assert, assertEquals } from '@std/assert'
 import { filenameIdentify } from './filenameIdentify.ts'
 import { filenameValidate } from './filenameValidate.ts'
-import { BIDSContext } from '../schema/context.ts'
+import { BIDSContext, BIDSContextDataset } from '../schema/context.ts'
 import { loadSchema } from '../setup/loadSchema.ts'
 import type { GenericSchema, Schema } from '../types/schema.ts'
 import type { DatasetIssues } from '../issues/datasetIssues.ts'
-import { pathToFile } from '../files/filetree.ts'
+import type { BIDSFile } from '../types/filetree.ts'
+import { pathsToTree } from '../files/filetree.ts'
 
-const schema = await loadSchema() as unknown as GenericSchema
+const schema = await loadSchema()
 
-function validatePath(path: string): DatasetIssues {
-  const context = new BIDSContext(pathToFile(path))
-  filenameIdentify(schema, context)
-  filenameValidate(schema, context)
-  return context.dataset.issues
+function makeContext(path: string): BIDSContext {
+  const tree = pathsToTree([path])
+  const dataset = new BIDSContextDataset({ schema, tree })
+  return new BIDSContext(tree.get(path) as BIDSFile, dataset)
 }
 
 Deno.test('test valid paths', async (t) => {
@@ -54,11 +54,13 @@ Deno.test('test valid paths', async (t) => {
   ]
   for (const filename of validFiles) {
     await t.step(filename, async () => {
-      const issues = validatePath(filename)
+      const context = makeContext(filename)
+      await filenameIdentify(schema, context)
+      await filenameValidate(schema as unknown as GenericSchema, context)
       assertEquals(
-        issues.get({ location: filename }).length,
+        context.dataset.issues.get({ location: filename }).length,
         0,
-        Deno.inspect(issues),
+        Deno.inspect(context.dataset.issues),
       )
     })
   }
@@ -111,9 +113,9 @@ Deno.test('test invalid paths', async (t) => {
   ]
   for (const filename of invalidFiles) {
     await t.step(filename, async () => {
-      const context = new BIDSContext(pathToFile(filename))
+      const context = makeContext(filename)
       await filenameIdentify(schema, context)
-      await filenameValidate(schema, context)
+      await filenameValidate(schema as unknown as GenericSchema, context)
       assert(
         context.dataset.issues.get({
           location: context.file.path,