feat: tab suggestions

packages/flutter_mozilla_components/android/src/main/assets/extensions/ml_engine/api/ml.js

@@ -46,9 +46,9 @@ const SMART_TAB_GROUPING_CONFIG = {
  */
 function createModelInput(keywords, documents) {
     if (!keywords || keywords.length === 0) {
-        return `Topic from keywords: titles: \n${documents.join(" \n")}`;
+        return `Topic from keywords: titles: \n${documents.slice(0, 3).join(" \n")}`;
     }
-    return `Topic from keywords: ${keywords.join(", ")}. titles: \n${documents.join(" \n")}`;
+    return `Topic from keywords: ${keywords.join(", ")}. titles: \n${documents.slice(0, 3).join(" \n")}`;
 }
 
 /**
@@ -81,7 +81,6 @@ function cutAtDuplicateWords(phrase) {
     return phrase; // return original phrase
 }
 
-
 /**
    *
    * @param {MLEngine} engine the engine to check
@@ -96,7 +95,29 @@ this.ml = class extends ExtensionAPI {
         return {
             experiments: {
                 ml: {
-                    async containerTopic(keywords, documents) {
+                    async generateEmbeddings(textToEmbedList) {
+                        const inputData = {
+                            inputArgs: textToEmbedList,
+                            runOptions: {
+                                pooling: "mean",
+                                normalize: true,
+                            },
+                        };
+
+                        if (isEngineClosed(this.embeddingEngine)) {
+                            this.embeddingEngine = await createEngine(SMART_TAB_GROUPING_CONFIG.embedding);
+                        }
+
+                        const request = {
+                            args: [inputData.inputArgs],
+                            options: inputData.runOptions,
+                        };
+
+                        const generated = await this.embeddingEngine.run(request);
+
+                        return JSON.stringify(generated);
+                    },
+                    async predictTopic(keywords, documents) {
                         if (isEngineClosed(this.topicEngine)) {
                             const {
                                 featureId,

packages/flutter_mozilla_components/android/src/main/assets/extensions/ml_engine/background.js

@@ -26,15 +26,25 @@ function sendErrorForRequest(id) {
 port.onMessage.addListener(async (message) => {
     let requestId = message["id"]
     switch (message["action"]) {
-        case "getContainerTopic":
+        case "predictDocumentTopic": {
             const documents = message["args"];
             const keywords = (documents.length > 1)
                 ? await browser.experiments.nlp.extractKeywords([documents.slice(0, 3).join(" ")])
                 : [[]];
 
-            browser.experiments.ml.containerTopic(keywords[0], documents)
+            browser.experiments.ml.predictTopic(keywords[0], documents)
                 .then(sendJsonResultForRequest(requestId))
-                .catch(sendErrorForRequest(requestId))
-            break
+                .catch(sendErrorForRequest(requestId));
+
+            break;
+        }
+        case "generateDocumentEmbeddings": {
+            const documents = message["args"];
+            await browser.experiments.ml.generateEmbeddings(documents)
+                .then(sendJsonResultForRequest(requestId))
+                .catch(sendErrorForRequest(requestId));
+
+            break;
+        }
     }
 });

packages/flutter_mozilla_components/android/src/main/assets/extensions/ml_engine/schema.json

@@ -33,7 +33,23 @@
         "description": "Machine Learning utilities",
         "functions": [
             {
-                "name": "containerTopic",
+                "name": "generateEmbeddings",
+                "type": "function",
+                "description": "Generate embeddings for a list of text strings using ML engine",
+                "async": true,
+                "parameters": [
+                    {
+                        "name": "textToEmbedList",
+                        "type": "array",
+                        "items": {
+                            "type": "string"
+                        },
+                        "description": "Array of text strings to generate embeddings for"
+                    }
+                ]
+            },
+            {
+                "name": "predictTopic",
                 "type": "function",
                 "description": "Generate topic from keywords and documents using ML engine",
                 "async": true,