TupleTool Search Functionality

Using a small language model to compute similarity between users input and the documentation for the TupleTools, users will be able to search for variables and TupleTools.

src/components/semantic.jsx

+/*****************************************************************************\
+* (c) Copyright 2024 CERN for the benefit of the LHCb Collaboration           *
+*                                                                             *
+* This software is distributed under the terms of the GNU General Public      *
+* Licence version 3 (GPL Version 3), copied verbatim in the file "COPYING".   *
+*                                                                             *
+* In applying this licence, CERN does not waive the privileges and immunities *
+* granted to it by virtue of its status as an Intergovernmental Organization  *
+* or submit itself to any jurisdiction.                                       *
+\*****************************************************************************/
+import {env} from "@xenova/transformers";
+
+// To always download the model from huggingface.com
+env.allowLocalModels = false;
+env.useBrowserCache = false;
+// Create a new worker
+const worker = new Worker(new URL("./worker.jsx", import.meta.url), {
+    type: "module",
+});
+/**
+ * @type {function}
+ * Promise resolve function for loading the model
+ */
+let loadResolve;
+/**
+ * @type {function}
+ */
+let queryResolve;
+
+worker.onmessage = function (event) {
+    const message = event.data;
+    switch (message.type) {
+        case "progress":
+            if (message.progress.status === "ready") {
+                loadResolve();
+            }
+            break;
+        case "corpus":
+            // the corpus is embedded
+            queryResolve();
+            break;
+        case "result":
+            queryResolve(message.result);
+            break;
+    }
+};
+/**
+ *
+ * @param {string} modelname
+ * Load the model with the provided model name
+ * @returns
+ */
+export async function loadModel(modelname) {
+    worker.postMessage({type: "init", model: modelname});
+    return new Promise((resolve) => {
+        loadResolve = resolve;
+    });
+}
+/**
+ * Passes corpus embedding from backend to the worker
+ * @returns
+ */
+export async function embedCorpus(metadata) {
+    worker.postMessage({type: "corpus", kgdoc: metadata.metadata.kgdoc, emb: metadata.metadata.embedding});
+    return new Promise((resolve) => {
+        queryResolve = resolve;
+    });
+}
+/**
+ *
+ * @param {string} query#
+ * Calculate the similarity between the query and the corpus
+ * @returns
+ */
+export async function getSimilarity(query) {
+    worker.postMessage({type: "similarity", query: query});
+    return new Promise((resolve) => {
+        queryResolve = resolve;
+    });
+}