commit d0f27a2fe758a1098430a3e670e709e13d15a55c
parent 8d845f554d7c762d506142b182eefb0333f214f5
Author: Marco Bonardo <mbonardo@mozilla.com>
Date: Tue, 2 Dec 2025 13:10:54 +0000
Bug 1992920 - Limit semantic history search to a list of supported locales. r=jteow,urlbar-reviewers,cgopal
Differential Revision: https://phabricator.services.mozilla.com/D272336
Diffstat:
7 files changed, 244 insertions(+), 20 deletions(-)
diff --git a/browser/app/profile/firefox.js b/browser/app/profile/firefox.js
@@ -688,6 +688,7 @@ pref("places.semanticHistory.featureGate", true);
#else
pref("places.semanticHistory.featureGate", false);
#endif
+pref("places.semanticHistory.supportedRegions", "[[\"AU\",[\"en-*\"]],[\"CA\",[\"en-*\"]],[\"GB\",[\"en-*\"]],[\"IE\",[\"en-*\"]],[\"NZ\",[\"en-*\"]],[\"PH\",[\"en-*\"]],[\"US\",[\"en-*\"]]]");
// Minimum length threshold for semantic history search
pref("browser.urlbar.suggest.semanticHistory.minLength", 5);
diff --git a/browser/components/urlbar/UrlbarProvidersManager.sys.mjs b/browser/components/urlbar/UrlbarProvidersManager.sys.mjs
@@ -18,6 +18,7 @@ const lazy = {};
ChromeUtils.defineESModuleGetters(lazy, {
ObjectUtils: "resource://gre/modules/ObjectUtils.sys.mjs",
PlacesUtils: "resource://gre/modules/PlacesUtils.sys.mjs",
+ Region: "resource://gre/modules/Region.sys.mjs",
SkippableTimer: "moz-src:///browser/components/urlbar/UrlbarUtils.sys.mjs",
UrlbarMuxer: "moz-src:///browser/components/urlbar/UrlbarUtils.sys.mjs",
UrlbarPrefs: "moz-src:///browser/components/urlbar/UrlbarPrefs.sys.mjs",
@@ -463,6 +464,15 @@ export class ProvidersManager {
// history and bookmarks even if search engines are not available.
}
+ // Some providers depend on Region/Locale info and must access Region.home
+ // synchronously, so we ensure Region is initialized.
+ try {
+ await lazy.Region.init();
+ } catch (ex) {
+ // We continue anyway, region will be null and providers should handle
+ // that gracefully.
+ }
+
if (query.canceled) {
return;
}
diff --git a/browser/components/urlbar/tests/unit/test_UrlbarProviderSemanticHistorySearch.js b/browser/components/urlbar/tests/unit/test_UrlbarProviderSemanticHistorySearch.js
@@ -17,6 +17,13 @@ const { UrlbarProviderSemanticHistorySearch } = ChromeUtils.importESModule(
const { getPlacesSemanticHistoryManager } = ChromeUtils.importESModule(
"resource://gre/modules/PlacesSemanticHistoryManager.sys.mjs"
);
+ChromeUtils.defineLazyGetter(this, "QuickSuggestTestUtils", () => {
+ const { QuickSuggestTestUtils: module } = ChromeUtils.importESModule(
+ "resource://testing-common/QuickSuggestTestUtils.sys.mjs"
+ );
+ module.init(this);
+ return module;
+});
let semanticManager = getPlacesSemanticHistoryManager();
let hasSufficientEntriesStub = sinon
@@ -46,6 +53,12 @@ add_task(async function setup() {
Services.prefs
.getDefaultBranch("")
.setIntPref("browser.urlbar.suggest.semanticHistory.minLength", 5);
+
+ let cleanup = await QuickSuggestTestUtils.setRegionAndLocale({
+ region: "US",
+ locale: "en-US",
+ });
+ registerCleanupFunction(cleanup);
});
add_task(async function test_startQuery_adds_results() {
diff --git a/toolkit/components/ml/tests/browser/browser_ml_semantic_history_search_perf.js b/toolkit/components/ml/tests/browser/browser_ml_semantic_history_search_perf.js
@@ -2,6 +2,10 @@
* http://creativecommons.org/publicdomain/zero/1.0/ */
"use strict";
+const { sinon } = ChromeUtils.importESModule(
+ "resource://testing-common/Sinon.sys.mjs"
+);
+
const UPDATE_TASK_LATENCY = "update-task-latency";
const SEARCH_LATENCY = "search-latency";
const INFERENCE_LATENCY = "inference-latency";
@@ -247,6 +251,13 @@ async function prepareSemanticSearchTest({
return { skip: true };
}
+ // Skip featureGate, Region and other non critical checks.
+ let canUseSemanticStub = sinon.stub(semanticManager, "canUseSemanticSearch");
+ canUseSemanticStub.get(() => true);
+ registerCleanupFunction(() => {
+ canUseSemanticStub.restore();
+ });
+
semanticManager.embedder.options = CUSTOM_EMBEDDER_OPTIONS;
await semanticManager.embedder.ensureEngine();
diff --git a/toolkit/components/nimbus/FeatureManifest.yaml b/toolkit/components/nimbus/FeatureManifest.yaml
@@ -598,6 +598,19 @@ urlbar:
description: >-
semantic similarity distance threshold (0.0 - 1.0)
defaulted at 0.6.
+ semanticHistorySupportedRegions:
+ type: string
+ setPref:
+ branch: user
+ pref: places.semanticHistory.supportedRegions
+ description: >-
+ map-like JSON of regions and locales for which the feature is supported.
+ When `semanticHistoryEnable` is false, this list is ignored and the
+ feature is disabled everywhere. When true, the feature is enabled only
+ in the regions/locales listed here.
+ The format is like: '[["AU",["en-*"]],["CA",["en-*"]], …]' where `-*`
+ can be used as a wildcard to match any locale that starts with the given
+ prefix.
showDebuggingIcons:
type: boolean
setPref:
diff --git a/toolkit/components/places/PlacesSemanticHistoryManager.sys.mjs b/toolkit/components/places/PlacesSemanticHistoryManager.sys.mjs
@@ -11,15 +11,17 @@
* @import {OpenedConnection} from "resource://gre/modules/Sqlite.sys.mjs"
*/
-const lazy = {};
+import { XPCOMUtils } from "resource://gre/modules/XPCOMUtils.sys.mjs";
+const lazy = {};
ChromeUtils.defineESModuleGetters(lazy, {
- DeferredTask: "resource://gre/modules/DeferredTask.sys.mjs",
AsyncShutdown: "resource://gre/modules/AsyncShutdown.sys.mjs",
- PlacesUtils: "resource://gre/modules/PlacesUtils.sys.mjs",
+ DeferredTask: "resource://gre/modules/DeferredTask.sys.mjs",
+ EmbeddingsGenerator: "chrome://global/content/ml/EmbeddingsGenerator.sys.mjs",
PlacesSemanticHistoryDatabase:
"resource://gre/modules/PlacesSemanticHistoryDatabase.sys.mjs",
- EmbeddingsGenerator: "chrome://global/content/ml/EmbeddingsGenerator.sys.mjs",
+ PlacesUtils: "resource://gre/modules/PlacesUtils.sys.mjs",
+ Region: "resource://gre/modules/Region.sys.mjs",
});
ChromeUtils.defineLazyGetter(lazy, "logger", function () {
@@ -33,6 +35,44 @@ ChromeUtils.defineLazyGetter(lazy, "PAGES_FRECENCY_FIELD", () => {
: "frecency";
});
+// This list is based on the current model capabilities. It is a Map-like list
+// of regions where English is predominant, and a common language is latin-based.
+// Each country code is assigned to an array of supported BCP 47 language tags,
+// a tag can end with "-*" to match any variants (match at the start).
+// The list of supported region and locales is loaded from the
+// places.semanticHistory.supportedRegions string pref, and this is used as a
+// fallback if we fail to parse the pref.
+/** @type {[string, string[]][]} */
+const ENABLED_REGIONS_DEFAULT = [
+ ["AU", ["en-*"]],
+ ["CA", ["en-*"]],
+ ["GB", ["en-*"]],
+ ["IE", ["en-*"]],
+ ["NZ", ["en-*"]],
+ ["PH", ["en-*"]],
+ ["US", ["en-*"]],
+];
+XPCOMUtils.defineLazyPreferenceGetter(
+ lazy,
+ "supportedRegions",
+ "places.semanticHistory.supportedRegions",
+ JSON.stringify(ENABLED_REGIONS_DEFAULT),
+ null,
+ val => {
+ try {
+ return new Map(JSON.parse(val));
+ } catch (ex) {
+ // Supposing a user may empty the pref to disable the feature, as they
+ // don't know it should be a JSON string, we'll treat that as an empty
+ // Map, so the feature is disabled.
+ if (val === "") {
+ return new Map();
+ }
+ return new Map(ENABLED_REGIONS_DEFAULT);
+ }
+ }
+);
+
// Time between deferred task executions.
const DEFERRED_TASK_INTERVAL_MS = 3000;
// Maximum time to wait for an idle before the task is executed anyway.
@@ -94,6 +134,7 @@ class PlacesSemanticHistoryManager {
* @param {number} [options.changeThresholdCount=3] - Threshold of changed rows to trigger updates.
* @param {number} [options.distanceThreshold=0.6] - Cosine distance threshold to determine similarity.
* @param {boolean} [options.testFlag=false] - Flag for test behavior.
+ * @param {number} [options.deferredTaskInterval=DEFERRED_TASK_INTERVAL_MS] - Interval for deferred task execution.
*/
constructor({
backend = "static-embeddings",
@@ -103,6 +144,7 @@ class PlacesSemanticHistoryManager {
changeThresholdCount = 3,
distanceThreshold = 0.6,
testFlag = false,
+ deferredTaskInterval = DEFERRED_TASK_INTERVAL_MS,
} = {}) {
this.QueryInterface = ChromeUtils.generateQI([
"nsIObserver",
@@ -149,6 +191,7 @@ class PlacesSemanticHistoryManager {
this.#changeThresholdCount = changeThresholdCount;
this.#distanceThreshold = distanceThreshold;
this.testFlag = testFlag;
+ this.#deferredTaskInterval = deferredTaskInterval;
this.#updateTaskLatency = [];
lazy.logger.trace("PlaceSemanticManager constructor");
@@ -289,13 +332,41 @@ class PlacesSemanticHistoryManager {
* else false
*/
get canUseSemanticSearch() {
+ // This requires Region to have been initialized somewhere else
+ // asynchronously, so consumer is responsible for that, otherwise it may
+ // be null.
return (
this.qualifiedForSemanticSearch &&
Services.prefs.getBoolPref("browser.ml.enable", true) &&
- Services.prefs.getBoolPref("places.semanticHistory.featureGate", false)
+ Services.prefs.getBoolPref("places.semanticHistory.featureGate", false) &&
+ this.#isSupportedLocale(Services.locale.appLocaleAsBCP47)
);
}
+ /**
+ * Check if the given locale is supported for Semantic History Search.
+ *
+ * @param {string} appLocale BCP 47 language tag.
+ * @returns {boolean} Whether the locale is supported.
+ */
+ #isSupportedLocale(appLocale) {
+ // Per BCP-47 comparisons must be performend in a case-insensitive manner.
+ appLocale = appLocale.toLowerCase();
+ let supportedLocales = lazy.supportedRegions.get(lazy.Region.home) ?? [];
+ for (let localePattern of supportedLocales) {
+ localePattern = localePattern.toLowerCase();
+ if (
+ localePattern.endsWith("*") &&
+ appLocale.startsWith(localePattern.slice(0, -1))
+ ) {
+ return true;
+ } else if (localePattern == appLocale) {
+ return true;
+ }
+ }
+ return false;
+ }
+
handlePlacesEvents(events) {
for (const { type } of events) {
switch (type) {
@@ -330,15 +401,6 @@ class PlacesSemanticHistoryManager {
}
/**
- * Sets the DeferredTask interval for testing purposes.
- *
- * @param {number} val minimum milliseconds between deferred task executions.
- */
- setDeferredTaskIntervalForTests(val) {
- this.#deferredTaskInterval = val;
- }
-
- /**
* Creates or updates the DeferredTask for managing updates to the semantic DB.
*/
#createOrUpdateTask() {
diff --git a/toolkit/components/places/tests/unit/test_PlacesSemanticHistoryManager.js b/toolkit/components/places/tests/unit/test_PlacesSemanticHistoryManager.js
@@ -10,6 +10,14 @@ ChromeUtils.defineESModuleGetters(this, {
"resource://gre/modules/PlacesSemanticHistoryManager.sys.mjs",
});
+ChromeUtils.defineLazyGetter(this, "QuickSuggestTestUtils", () => {
+ const { QuickSuggestTestUtils: module } = ChromeUtils.importESModule(
+ "resource://testing-common/QuickSuggestTestUtils.sys.mjs"
+ );
+ module.init(this);
+ return module;
+});
+
// Must be supported, and a multiple of 8. see EmbeddingsGenerator.sys.mjs for
// a list of supported values.
const EMBEDDING_SIZE = 32;
@@ -61,6 +69,11 @@ class MockMLEngine {
add_setup(async function () {
Services.fog.initializeFOG();
+ let cleanup = await QuickSuggestTestUtils.setRegionAndLocale({
+ region: "US",
+ locale: "en-US",
+ });
+ registerCleanupFunction(cleanup);
});
add_task(async function test_tensorToSQLBindable() {
@@ -164,6 +177,105 @@ add_task(async function test_canUseSemanticSearch_not_qualified() {
);
});
+add_task(async function test_canUseSemanticSearch_region_locale() {
+ const semanticManager = createPlacesSemanticHistoryManager();
+
+ Services.prefs.setBoolPref("browser.ml.enable", true);
+ Services.prefs.setBoolPref("places.semanticHistory.featureGate", true);
+
+ semanticManager.qualifiedForSemanticSearch = true;
+ semanticManager.enoughEntries = true;
+
+ let tests = [
+ { region: "US", locale: "en-US", supported: true },
+ { region: "FR", locale: "fr-FR", supported: false },
+ {
+ region: "IT",
+ locale: "it-IT",
+ supported: true,
+ setPref: '[["IT",["it-*"]]]',
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: false,
+ setPref: '[["IT",["it-*"]]]',
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: false,
+ setPref: "[]", // empty list means disable
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: false,
+ setPref: "", // empty string means disable
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: true,
+ setPref: "invalid json", // invalid, should use default.
+ },
+ {
+ region: "US",
+ locale: "en", // wrong locale format
+ supported: false,
+ setPref: '[["US",["en-*"]]]',
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: true,
+ setPref: '[["US",["en-US"]]]',
+ },
+ {
+ region: "US",
+ locale: "es-MX",
+ supported: false,
+ setPref: '[["US",["en-US"]]]',
+ },
+ {
+ region: "US",
+ locale: "es-MX",
+ supported: true,
+ setPref: '[["US",["en-*","es-*"]]]',
+ },
+ {
+ region: "US",
+ locale: "en-US",
+ supported: true,
+ setPref: '[["US",["en-*","es-*"]]]',
+ },
+ ];
+ for (let { region, locale, supported, setPref } of tests) {
+ if (setPref !== undefined) {
+ info("Setting `supportedRegions` pref to " + setPref);
+ Services.prefs.setCharPref(
+ "places.semanticHistory.supportedRegions",
+ setPref
+ );
+ }
+ await QuickSuggestTestUtils.withRegionAndLocale({
+ region,
+ locale,
+ skipSuggestReset: true,
+ callback() {
+ Assert.equal(
+ semanticManager.canUseSemanticSearch,
+ supported,
+ `Check region ${region} and locale ${locale}`
+ );
+ },
+ });
+ if (setPref !== undefined) {
+ Services.prefs.clearUserPref("places.semanticHistory.supportedRegions");
+ }
+ }
+});
+
add_task(async function test_removeDatabaseFilesOnDisable() {
// Ensure Places has been initialized.
Assert.equal(
@@ -269,9 +381,9 @@ add_task(async function test_chunksTelemetry() {
);
Services.prefs.setBoolPref("places.semanticHistory.featureGate", true);
- let semanticManager = createPlacesSemanticHistoryManager();
- // Ensure only one task execution for measuremant purposes.
- semanticManager.setDeferredTaskIntervalForTests(3000);
+ let semanticManager = createPlacesSemanticHistoryManager({
+ deferredTaskInterval: 2000, // lower time to avoid timeouts.
+ });
await semanticManager.getConnection();
semanticManager.embedder.setEngine(new MockMLEngine());
await TestUtils.topicObserved(
@@ -317,9 +429,9 @@ add_task(async function test_duplicate_urlhash() {
);
});
- let semanticManager = createPlacesSemanticHistoryManager();
- // Ensure only one task execution for measuremant purposes.
- semanticManager.setDeferredTaskIntervalForTests(3000);
+ let semanticManager = createPlacesSemanticHistoryManager({
+ deferredTaskInterval: 2000, // lower time to avoid timeouts.
+ });
let conn = await semanticManager.getConnection();
semanticManager.embedder.setEngine(new MockMLEngine());
await TestUtils.topicObserved(
@@ -358,6 +470,7 @@ add_task(async function test_rowid_relations() {
let semanticManager = createPlacesSemanticHistoryManager({
changeThresholdCount: 1,
+ deferredTaskInterval: 2000, // lower time to avoid timeouts.
});
// Ensure we start from an empty database.
await semanticManager.semanticDB.removeDatabaseFiles();
@@ -426,6 +539,7 @@ add_task(async function test_rowid_conflict() {
let semanticManager = createPlacesSemanticHistoryManager({
changeThresholdCount: 1,
+ deferredTaskInterval: 2000, // lower time to avoid timeouts.
});
// Ensure we start from an empty database.
await semanticManager.semanticDB.removeDatabaseFiles();
