Dynamically register data/tables schemas (#57905)

src/data-directory/lib/data-schemas/index.ts

@@ -1,8 +1,37 @@
+import fs from 'fs'
+import path from 'path'
+import { fileURLToPath } from 'url'
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url))
+
 interface DataSchemas {
   [key: string]: string
 }
 
-const dataSchemas: DataSchemas = {
+// Auto-discover table schemas from data/tables/ directory
+function loadTableSchemas(): DataSchemas {
+  const tablesDir = path.join(process.cwd(), 'data/tables')
+  const schemasDir = path.join(__dirname, 'tables')
+  const tableSchemas: DataSchemas = {}
+
+  if (fs.existsSync(tablesDir)) {
+    const yamlFiles = fs.readdirSync(tablesDir).filter((file) => file.endsWith('.yml'))
+
+    for (const yamlFile of yamlFiles) {
+      const name = path.basename(yamlFile, '.yml')
+      const schemaPath = path.join(schemasDir, `${name}.ts`)
+
+      if (fs.existsSync(schemaPath)) {
+        tableSchemas[`data/tables/${yamlFile}`] = `@/data-directory/lib/data-schemas/tables/${name}`
+      }
+    }
+  }
+
+  return tableSchemas
+}
+
+// Manual schema registrations for non-table data
+const manualSchemas: DataSchemas = {
   'data/features': '@/data-directory/lib/data-schemas/features.js',
   'data/variables': '@/data-directory/lib/data-schemas/variables',
   'data/learning-tracks': '@/data-directory/lib/data-schemas/learning-tracks.js',
@@ -10,8 +39,12 @@ const dataSchemas: DataSchemas = {
   'data/code-languages.yml': '@/data-directory/lib/data-schemas/code-languages',
   'data/glossaries/candidates.yml': '@/data-directory/lib/data-schemas/glossaries-candidates.js',
   'data/glossaries/external.yml': '@/data-directory/lib/data-schemas/glossaries-external.js',
-  'data/tables/supported-code-languages.yml':
-    '@/data-directory/lib/data-schemas/supported-code-languages.js',
+}
+
+// Combine manual registrations with auto-discovered table schemas
+const dataSchemas: DataSchemas = {
+  ...manualSchemas,
+  ...loadTableSchemas(),
 }
 
 export default dataSchemas

src/data-directory/tests/data-schemas.ts

@@ -1,6 +1,6 @@
 import yaml from 'js-yaml'
-import { readFileSync } from 'fs'
-import { extname, basename } from 'path'
+import { readFileSync, existsSync, readdirSync } from 'fs'
+import { extname, basename, join } from 'path'
 
 import walk from 'walk-sync'
 import { beforeAll, describe, expect, test } from 'vitest'
@@ -49,3 +49,24 @@ describe('single data files', () => {
     expect(isValid, formattedErrors).toBe(true)
   })
 })
+
+describe('YAML-powered tables', () => {
+  test('all table files have corresponding schemas', () => {
+    const tablesDir = join(process.cwd(), 'data/tables')
+    const schemasDir = join(__dirname, '../lib/data-schemas/tables')
+
+    if (existsSync(tablesDir)) {
+      const yamlFiles = readdirSync(tablesDir).filter((file) => file.endsWith('.yml'))
+
+      for (const yamlFile of yamlFiles) {
+        const name = basename(yamlFile, '.yml')
+        const schemaPath = join(schemasDir, `${name}.ts`)
+        expect(existsSync(schemaPath)).toBe(true)
+
+        // Also verify it's registered in the dataSchemas
+        const dataKey = `data/tables/${yamlFile}`
+        expect(dataSchemas[dataKey]).toBeDefined()
+      }
+    }
+  })
+})