updated old class to convert words to new format

package.json

@@ -140,6 +140,7 @@
     "@types/node": "^22.5.5",
     "@types/node-fetch": "^2.5.6",
     "@types/sinon": "^17.0.3",
+    "@types/uuid": "^10.0.0",
     "@types/wikidata-sdk": "^5.15.0",
     "eslint": "^8.18.0",
     "eslint-config-standard": "^17.0.0",

src/services/embeddings/embeddings-v1.class.js

@@ -17,22 +17,22 @@ class Service {
   }
 
   async find(params) {
-    const namespace = `embeddings_${params.query.language}`
+    const namespace = `embeddings_${params.query.language_code}`
     // use en to get embedding vector for the queried word
     //
     // https:// solrdev.dhlab.epfl.ch/solr/impresso_embeddings_de/select?q=word_s:amour&fl=embedding_bv
     debug('[find] with params', params.query)
 
     const bvRequest = {
-      q: `word_s:(${escapeValue(params.query.q)})`,
+      q: `word_s:(${escapeValue(params.query.term)})`,
       fl: 'embedding_bv',
       namespace,
     }
     const bv = await measureTime(
       () =>
         asFindAll(this.solr, namespace, bvRequest).then(res => {
           if (!res.response.docs.length) {
-            throw new NotFound(`word "${params.query.q}" not found in available embeddings`)
+            throw new NotFound(`word "${params.query.term}" not found in available embeddings`)
           }
           return res.response.docs[0].embedding_bv
         }),
@@ -60,8 +60,8 @@ class Service {
       limit: params.query.limit,
       offset: params.query.offset,
       info: {
-        q: params.query.q,
-        language: params.query.language,
+        q: params.query.term,
+        language: params.query.language_code,
       },
     }
   }

src/services/embeddings/embeddings-v1.hooks.ts

@@ -0,0 +1,72 @@
+import { HookContext } from '@feathersjs/feathers'
+import { ImpressoApplication } from '../../types'
+import { v4 } from 'uuid'
+import { WordMatch } from '../../models/generated/schemas'
+
+const { queryWithCommonParams, validate } = require('../../hooks/params')
+
+export default {
+  before: {
+    all: [],
+    find: [
+      validate(
+        {
+          language_code: {
+            choices: ['fr', 'de', 'lb'],
+          },
+          term: {
+            required: true,
+            regex: /^[A-zÀ-ÿ'()\s]+$/,
+            max_length: 500,
+            transform: (d: string) =>
+              d
+                .replace(/[^A-zÀ-ÿ]/g, ' ')
+                .toLowerCase()
+                .split(/\s+/)
+                .sort((a: string, b: string) => a.length - b.length)
+                .pop(),
+          },
+        },
+        'GET'
+      ),
+      queryWithCommonParams(),
+    ],
+    get: [],
+    create: [],
+    update: [],
+    patch: [],
+    remove: [],
+  },
+
+  after: {
+    all: [],
+    find: [
+      (context: HookContext<ImpressoApplication>) => {
+        if (Array.isArray(context.result.data)) {
+          context.result.data = context.result.data.map((word: string) => {
+            return {
+              word,
+              id: v4(),
+              languageCode: context.params.query.language ?? 'fr',
+            } satisfies WordMatch
+          })
+        }
+      },
+    ],
+    get: [],
+    create: [],
+    update: [],
+    patch: [],
+    remove: [],
+  },
+
+  error: {
+    all: [],
+    find: [],
+    get: [],
+    create: [],
+    update: [],
+    patch: [],
+    remove: [],
+  },
+}