feat: Retrieve datatype and modality from BIDSFile and schema

Author: effigies

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 }
+}