commit 4dc5bc79f2ae965a9ec9b6330bcf644c3063d34d
parent 86dc5318b583376277f455a1468b7f2508a6dbf1
Author: Chidam Gopal <cgopal@mozilla.com>
Date: Tue, 9 Dec 2025 21:40:20 +0000
Bug 2005046 - Insight updation for Day 0 and incremental updates r=cdipersio,ai-models-reviewers
Differential Revision: https://phabricator.services.mozilla.com/D275669
Diffstat:
3 files changed, 328 insertions(+), 7 deletions(-)
diff --git a/browser/components/aiwindow/models/InsightsConstants.sys.mjs b/browser/components/aiwindow/models/InsightsConstants.sys.mjs
@@ -3,6 +3,7 @@
* file, You can obtain one at https://mozilla.org/MPL/2.0/. */
export const HISTORY = "history";
+export const CONVERSATION = "conversation";
/**
* Insight categories
diff --git a/browser/components/aiwindow/models/InsightsManager.sys.mjs b/browser/components/aiwindow/models/InsightsManager.sys.mjs
@@ -17,10 +17,13 @@ import { InsightStore } from "moz-src:///browser/components/aiwindow/services/In
import {
CATEGORIES,
INTENTS,
+ HISTORY as SOURCE_HISTORY,
+ CONVERSATION as SOURCE_CONVERSATION,
} from "moz-src:///browser/components/aiwindow/models/InsightsConstants.sys.mjs";
import {
getFormattedInsightAttributeList,
parseAndExtractJSON,
+ generateInsights,
} from "moz-src:///browser/components/aiwindow/models/Insights.sys.mjs";
import {
messageInsightClassificationSystemPrompt,
@@ -28,10 +31,19 @@ import {
} from "moz-src:///browser/components/aiwindow/models/prompts/insightsPrompts.sys.mjs";
import { INSIGHTS_MESSAGE_CLASSIFY_SCHEMA } from "moz-src:///browser/components/aiwindow/models/InsightsSchemas.sys.mjs";
-const K_DOMAINS = 30;
-const K_TITLES = 60;
-const K_SEARCHES = 10;
+const K_DOMAINS_FULL = 100;
+const K_TITLES_FULL = 60;
+const K_SEARCHES_FULL = 10;
+const K_DOMAINS_DELTA = 30;
+const K_TITLES_DELTA = 60;
+const K_SEARCHES_DELTA = 10;
+const DEFAULT_HISTORY_FULL_LOOKUP_DAYS = 60;
+const DEFAULT_HISTORY_FULL_MAX_RESULTS = 3000;
+const DEFAULT_HISTORY_DELTA_MAX_RESULTS = 500;
+
+const LAST_HISTORY_INSIGHTS_TS_ATTRIBUTE = "last_history_insight_ts";
+const LAST_CONVERSATION_INSIGHTS_TS_ATTRIBUTE = "last_chat_insight_ts";
/**
* InsightsManager class
*/
@@ -52,6 +64,89 @@ export class InsightsManager {
}
/**
+ * Generates and persists insights derived from the user's recent browsing history.
+ *
+ * This method:
+ * 1. Reads {@link last_history_insight_ts} via {@link getLastHistoryInsightsTimestamp}.
+ * 2. Decides between:
+ * - Full processing (first run, no prior timestamp):
+ * * Uses a days-based cutoff (DEFAULT_HISTORY_FULL_LOOKUP_DAYS).
+ * * Uses max-results cap (DEFAULT_HISTORY_FULL_MAX_RESULTS).
+ * * Uses full top-k settings (K_DOMAINS_FULL, K_TITLES_FULL, K_SEARCHES_FULL).
+ * - Delta processing (subsequent runs, prior timestamp present):
+ * * Uses an absolute cutoff via `sinceMicros = lastTsMs * 1000`.
+ * * Uses a smaller max-results cap (DEFAULT_HISTORY_DELTA_MAX_RESULTS).
+ * * Uses delta top-k settings (K_DOMAINS_DELTA, K_TITLES_DELTA, K_SEARCHES_DELTA).
+ * 3. Calls {@link getAggregatedBrowserHistory} with the computed options to obtain
+ * domain, title, and search aggregates.
+ * 4. Fetches existing insights via {@link getAllInsights}.
+ * 5. Ensures a shared OpenAI engine via {@link ensureOpenAIEngine} and calls
+ * {@link generateInsights} to produce new/updated insights.
+ * 6. Persists those insights via {@link saveInsights}, which also updates
+ * `last_history_insight_ts` in {@link InsightStore.updateMeta}.
+ *
+ * @returns {Promise<Insight[]>}
+ * A promise that resolves to the list of persisted history insights
+ * (newly created or updated), sorted and shaped as returned by
+ * {@link InsightStore.addInsight}.
+ */
+ static async generateInsightsFromBrowsingHistory() {
+ const now = Date.now();
+ // get last history insight timestamp in ms
+ const lastTsMs = await this.getLastHistoryInsightsTimestamp();
+ const isDelta = typeof lastTsMs === "number" && lastTsMs > 0;
+ // set up the options based on delta or full (first) run
+ let recentHistoryOpts = {};
+ let topkAggregatesOpts;
+ if (isDelta) {
+ recentHistoryOpts = {
+ sinceMicros: lastTsMs * 1000,
+ maxResults: DEFAULT_HISTORY_DELTA_MAX_RESULTS,
+ };
+ topkAggregatesOpts = {
+ k_domains: K_DOMAINS_DELTA,
+ k_titles: K_TITLES_DELTA,
+ k_searches: K_SEARCHES_DELTA,
+ now,
+ };
+ } else {
+ recentHistoryOpts = {
+ days: DEFAULT_HISTORY_FULL_LOOKUP_DAYS,
+ maxResults: DEFAULT_HISTORY_FULL_MAX_RESULTS,
+ };
+ topkAggregatesOpts = {
+ k_domains: K_DOMAINS_FULL,
+ k_titles: K_TITLES_FULL,
+ k_searches: K_SEARCHES_FULL,
+ now,
+ };
+ }
+
+ const [domainItems, titleItems, searchItems] =
+ await this.getAggregatedBrowserHistory(
+ recentHistoryOpts,
+ topkAggregatesOpts
+ );
+ const sources = { history: [domainItems, titleItems, searchItems] };
+ const existingInsights = await this.getAllInsights();
+ const existingInsightsSummaries = existingInsights.map(
+ i => i.insight_summary
+ );
+ const engine = await this.ensureOpenAIEngine();
+ const insights = await generateInsights(
+ engine,
+ sources,
+ existingInsightsSummaries
+ );
+ const { persistedInsights } = await this.saveInsights(
+ insights,
+ SOURCE_HISTORY,
+ now
+ );
+ return persistedInsights;
+ }
+
+ /**
* Retrieves and aggregates recent browser history into top-k domain, title, and search aggregates.
*
* @param {object} [recentHistoryOpts={}]
@@ -85,9 +180,9 @@ export class InsightsManager {
static async getAggregatedBrowserHistory(
recentHistoryOpts = {},
topkAggregatesOpts = {
- k_domains: K_DOMAINS,
- k_titles: K_TITLES,
- k_searches: K_SEARCHES,
+ k_domains: K_DOMAINS_DELTA,
+ k_titles: K_TITLES_DELTA,
+ k_searches: K_SEARCHES_DELTA,
now: undefined,
}
) {
@@ -122,6 +217,77 @@ export class InsightsManager {
}
/**
+ * Returns the last timestamp (in ms since Unix epoch) when a history-based
+ * insight was generated, as persisted in InsightStore.meta.
+ *
+ * If the store has never been updated, this returns 0.
+ *
+ * @returns {Promise<number>} Milliseconds since Unix epoch
+ */
+ static async getLastHistoryInsightsTimestamp() {
+ const meta = await InsightStore.getMeta();
+ return meta.last_history_insights_ts || 0;
+ }
+
+ /**
+ * Persist a list of generated insights and update the appropriate meta timestamp.
+ *
+ * @param {Array<object>|null|undefined} generatedInsights
+ * Array of InsightPartial-like objects to persist.
+ * @param {"history"|"conversation"} source
+ * Source of these insights; controls which meta timestamp to update.
+ * @param {number} [nowMs=Date.now()]
+ * Optional "now" timestamp in ms, for meta update fallback.
+ *
+ * @returns {Promise<{ persistedInsights: Array<object>, newTimestampMs: number | null }>}
+ */
+ static async saveInsights(generatedInsights, source, nowMs = Date.now()) {
+ const persistedInsights = [];
+
+ if (Array.isArray(generatedInsights)) {
+ for (const insightPartial of generatedInsights) {
+ const stored = await InsightStore.addInsight(insightPartial);
+ persistedInsights.push(stored);
+ }
+ }
+
+ // Decide which meta field to update
+ let metaKey;
+ if (source === SOURCE_HISTORY) {
+ metaKey = LAST_HISTORY_INSIGHTS_TS_ATTRIBUTE;
+ } else if (source === SOURCE_CONVERSATION) {
+ metaKey = LAST_CONVERSATION_INSIGHTS_TS_ATTRIBUTE;
+ } else {
+ // Unknown source: don't update meta, just return persisted results.
+ return {
+ persistedInsights,
+ newTimestampMs: null,
+ };
+ }
+
+ // Compute new timestamp: prefer max(updated_at) if present, otherwise fall back to nowMs.
+ let newTsMs = nowMs;
+ if (persistedInsights.length) {
+ const maxUpdated = persistedInsights.reduce(
+ (max, i) => Math.max(max, i.updated_at ?? 0),
+ 0
+ );
+ if (maxUpdated > 0) {
+ newTsMs = maxUpdated;
+ }
+ }
+
+ await InsightStore.updateMeta({
+ [metaKey]: newTsMs,
+ });
+
+ return {
+ persistedInsights,
+ newTimestampMs: newTsMs,
+ };
+ }
+
+ /**
* Builds the prompt to classify a user message into insight categories and intents.
*
* @param {string} message User message to classify
diff --git a/browser/components/aiwindow/models/tests/xpcshell/test_InsightsManager.js b/browser/components/aiwindow/models/tests/xpcshell/test_InsightsManager.js
@@ -12,7 +12,12 @@ const { sinon } = ChromeUtils.importESModule(
const { InsightsManager } = ChromeUtils.importESModule(
"moz-src:///browser/components/aiwindow/models/InsightsManager.sys.mjs"
);
-const { CATEGORIES, INTENTS } = ChromeUtils.importESModule(
+const {
+ CATEGORIES,
+ INTENTS,
+ HISTORY: SOURCE_HISTORY,
+ CONVERSATION: SOURCE_CONVERSATION,
+} = ChromeUtils.importESModule(
"moz-src:///browser/components/aiwindow/models/InsightsConstants.sys.mjs"
);
const { getFormattedInsightAttributeList } = ChromeUtils.importESModule(
@@ -504,3 +509,152 @@ add_task(
}
}
);
+
+/**
+ * Tests saveInsights correctly persists history insights and updates last_history_insight_ts.
+ */
+add_task(async function test_saveInsights_history_updates_meta() {
+ const sb = sinon.createSandbox();
+ try {
+ const now = Date.now();
+
+ const generatedInsights = [
+ {
+ insight_summary: "foo",
+ category: "A",
+ intent: "X",
+ score: 1,
+ updated_at: now - 1000,
+ },
+ {
+ insight_summary: "bar",
+ category: "B",
+ intent: "Y",
+ score: 2,
+ updated_at: now + 500,
+ },
+ ];
+
+ const storedInsights = generatedInsights.map((generatedInsight, idx) => ({
+ id: `id-${idx}`,
+ ...generatedInsight,
+ }));
+
+ const addInsightStub = sb
+ .stub(InsightStore, "addInsight")
+ .callsFake(async partial => {
+ // simple mapping: return first / second stored insight based on summary
+ return storedInsights.find(
+ s => s.insight_summary === partial.insight_summary
+ );
+ });
+
+ const updateMetaStub = sb.stub(InsightStore, "updateMeta").resolves();
+
+ const { persistedInsights, newTimestampMs } =
+ await InsightsManager.saveInsights(
+ generatedInsights,
+ SOURCE_HISTORY,
+ now
+ );
+
+ Assert.equal(
+ addInsightStub.callCount,
+ generatedInsights.length,
+ "addInsight should be called once per generated insight"
+ );
+ Assert.deepEqual(
+ persistedInsights.map(i => i.id),
+ storedInsights.map(i => i.id),
+ "Persisted insights should match stored insights"
+ );
+
+ Assert.ok(
+ updateMetaStub.calledOnce,
+ "updateMeta should be called once for history source"
+ );
+ const metaArg = updateMetaStub.firstCall.args[0];
+ Assert.ok(
+ "last_history_insight_ts" in metaArg,
+ "updateMeta should update last_history_insight_ts for history source"
+ );
+ Assert.equal(
+ metaArg.last_history_insight_ts,
+ storedInsights[1].updated_at,
+ "last_history_insight_ts should be set to max(updated_at) among persisted insights"
+ );
+ Assert.equal(
+ newTimestampMs,
+ storedInsights[1].updated_at,
+ "Returned newTimestampMs should match the updated meta timestamp"
+ );
+ } finally {
+ sb.restore();
+ }
+});
+
+/**
+ * Tests saveInsights correctly persists conversation insights and updates last_chat_insight_ts.
+ */
+add_task(async function test_saveInsights_conversation_updates_meta() {
+ const sb = sinon.createSandbox();
+ try {
+ const now = Date.now();
+
+ const generatedInsights = [
+ {
+ insight_summary: "chat-insight",
+ category: "Chat",
+ intent: "Talk",
+ score: 1,
+ updated_at: now,
+ },
+ ];
+ const storedInsight = { id: "chat-1", ...generatedInsights[0] };
+
+ const addInsightStub = sb
+ .stub(InsightStore, "addInsight")
+ .resolves(storedInsight);
+ const updateMetaStub = sb.stub(InsightStore, "updateMeta").resolves();
+
+ const { persistedInsights, newTimestampMs } =
+ await InsightsManager.saveInsights(
+ generatedInsights,
+ SOURCE_CONVERSATION,
+ now
+ );
+
+ Assert.equal(
+ addInsightStub.callCount,
+ 1,
+ "addInsight should be called once for conversation insight"
+ );
+ Assert.equal(
+ persistedInsights[0].id,
+ storedInsight.id,
+ "Persisted insight should match stored insight"
+ );
+
+ Assert.ok(
+ updateMetaStub.calledOnce,
+ "updateMeta should be called once for conversation source"
+ );
+ const metaArg = updateMetaStub.firstCall.args[0];
+ Assert.ok(
+ "last_chat_insight_ts" in metaArg,
+ "updateMeta should update last_chat_insight_ts for conversation source"
+ );
+ Assert.equal(
+ metaArg.last_chat_insight_ts,
+ storedInsight.updated_at,
+ "last_chat_insight_ts should be set to insight.updated_at"
+ );
+ Assert.equal(
+ newTimestampMs,
+ storedInsight.updated_at,
+ "Returned newTimestampMs should match the updated meta timestamp"
+ );
+ } finally {
+ sb.restore();
+ }
+});
