fix(queries): match Dia/Arc chrome forks and surface empty Browser view

src/components/SelectableVisualization.vue

@@ -38,6 +38,9 @@ div(v-if="editable || !activityStore.buckets.loaded || has_prerequisites || !set
                  :namefunc="e => e.data.classname",
                  :colorfunc="e => e.data.app",
                  with_limit)
+    b-alert.small.px-2.py-1(v-if="isBrowserVis && browserAllowlistMiss" show variant="info")
+      | {{ $t('activity.browserAllowlistMiss') }}
+      |  (#[a(href="https://github.com/ActivityWatch/aw-webui/issues/927") #927])
     div(v-if="type == 'top_domains'")
       aw-summary(:fields="activityStore.browser.top_domains",
                  :namefunc="e => e.data.$domain",
@@ -132,6 +135,7 @@ import { useCategoryStore } from '~/stores/categories';
 import { useBucketsStore } from '~/stores/buckets';
 import { useViewsStore } from '~/stores/views';
 import { useSettingsStore } from '~/stores/settings';
+import { isBrowserAllowlistMiss } from '~/util/browserAllowlist';
 
 import moment from 'moment';
 
@@ -282,6 +286,12 @@ export default {
     has_prerequisites() {
       return this.visualizations[this.type].available;
     },
+    isBrowserVis() {
+      return ['top_domains', 'top_urls', 'top_browser_titles'].includes(this.type);
+    },
+    browserAllowlistMiss() {
+      return isBrowserAllowlistMiss(this.activityStore.browser);
+    },
     supports_period: function () {
       if (this.type == 'sunburst_clock' || this.type == 'vis_timeline') {
         return this.isSingleDay;

src/i18n/locales/de.ts

@@ -268,6 +268,8 @@ export default {
   },
   activity: {
     title: 'Aktivität',
+    browserAllowlistMiss:
+      'Für diesen Zeitraum wurde kein passendes Browserfenster gefunden. Falls Sie in einem Chromium-/Firefox-Derivat gesurft haben, wird dessen App-Name möglicherweise noch nicht erkannt.',
     for: 'für',
     host: 'Host:',
     timeActive: 'Aktive Zeit:',

src/i18n/locales/en.ts

@@ -269,6 +269,8 @@ export default {
   },
   activity: {
     title: 'Activity',
+    browserAllowlistMiss:
+      'No matching browser window for this period. If you were browsing in a Chromium/Firefox fork, its app name may not be recognized yet.',
     for: 'for',
     host: 'Host:',
     timeActive: 'Time active:',

src/i18n/locales/ru.ts

@@ -265,6 +265,8 @@ export default {
   },
   activity: {
     title: 'Активность',
+    browserAllowlistMiss:
+      'Для этого периода не найдено подходящее окно браузера. Если вы использовали форк Chromium/Firefox, имя его приложения может пока не распознаваться.',
     for: 'для',
     host: 'Хост:',
     timeActive: 'Активное время:',

src/i18n/locales/sv.ts

@@ -273,6 +273,8 @@ export default {
   },
   activity: {
     title: 'Aktivitet',
+    browserAllowlistMiss:
+      'Inget matchande webbläsarfönster för den här perioden. Om du surfade i en Chromium-/Firefox-fork kanske dess appnamn inte känns igen ännu.',
     for: 'för',
     host: 'Värd:',
     timeActive: 'Aktiv tid:',

src/i18n/locales/uk.ts

@@ -265,6 +265,8 @@ export default {
   },
   activity: {
     title: 'Активність',
+    browserAllowlistMiss:
+      'Для цього періоду не знайдено відповідного вікна браузера. Якщо ви користувалися форком Chromium/Firefox, назва його застосунку може ще не розпізнаватися.',
     for: 'для',
     host: 'Хост:',
     timeActive: 'Активний час:',

src/i18n/locales/zh-CN.ts

@@ -260,6 +260,8 @@ export default {
   },
   activity: {
     title: '活动',
+    browserAllowlistMiss:
+      '此时间段内未找到匹配的浏览器窗口。如果您使用的是 Chromium/Firefox 衍生浏览器，其应用名称可能尚未被识别。',
     for: '查看',
     host: '主机：',
     timeActive: '活跃时间：',

src/queries.ts

@@ -255,8 +255,16 @@ export function appQuery(
 // variants (upper/lowercase, spacing, .exe suffix) are handled by
 // browser_appname_regex using (?i) flag. See test/unit/queries.test.node.ts for
 // the complete list of known app names these patterns cover.
-const browser_appnames: Record<string, string[]> = {
-  chrome: ['com.google.Chrome', 'com.google.ChromeDev', 'org.chromium.Chromium'],
+export const browser_appnames: Record<string, string[]> = {
+  // Chromium forks (Dia, Arc) run the chrome extension by default, so their
+  // web events land in the chrome bucket. Reverse-domain identifiers don't
+  // match the process-name regex below and have to live here (#927).
+  chrome: [
+    'com.google.Chrome',
+    'com.google.ChromeDev',
+    'org.chromium.Chromium',
+    'company.thebrowser.dia',
+  ],
   firefox: ['org.mozilla.firefox', 'io.gitlab.librewolf-community', 'net.waterfox.waterfox'],
   opera: ['com.opera.Opera'],
   brave: ['com.brave.Browser'],
@@ -294,6 +302,7 @@ export const browser_appname_regex: Record<string, string> = {
   // default their events land in the chrome bucket and their app names have to be matched
   // here (#927, ActivityWatch/activitywatch#1094). The standalone arc key below only covers
   // setups where Arc was picked explicitly in the settings, which changes the bucket name.
+  // Fork alternatives are $-anchored so names like "archive" / "Dialog" don't match.
   chrome: '(?i)^(google[-_ ]?chrome|chrome|chromium|arc(\\.exe)?$|dia(\\.exe)?$)',
   firefox: '(?i)(firefox|librewolf|waterfox|nightly)',
   opera: '(?i)(opera)',
@@ -310,29 +319,50 @@ export const browser_appname_regex: Record<string, string> = {
 
 // Returns a list of active browser events (where the browser was the active window) from all browser buckets
 function browserEvents(params: DesktopQueryParams): string {
+  const browsers = browsersWithBuckets(params.bid_browsers);
+  // Chrome regex also matches Arc, and a settings-override Arc bucket can
+  // coexist with the default chrome bucket. Those two streams can duplicate
+  // the same Arc activity; union_no_overlap is only for that pair. Distinct
+  // browsers (Chrome + Firefox, etc.) may overlap in time and must concat.
+  const mixChromeArc =
+    browsers.some(([browserName]) => browserName === 'chrome') &&
+    browsers.some(([browserName]) => browserName === 'arc');
+
   let code = `
     browser_events = [];
   `;
 
-  _.each(browsersWithBuckets(params.bid_browsers), ([browserName, bucketId]) => {
+  _.each(browsers, ([browserName, bucketId]) => {
     const browser_appnames_str = JSON.stringify(browser_appnames[browserName]);
     code += `events_${browserName} = flood(query_bucket("${bucketId}"));
        window_${browserName} = filter_keyvals(events, "app", ${browser_appnames_str});`;
 
-    // Add regex-based matching to cover case/spacing/versioning variants (e.g., Firefox.exe, firefox-esr-esr140)
+    // Add regex-based matching to cover case/spacing/versioning variants (e.g., Firefox.exe, firefox-esr-esr140).
     const pattern = browser_appname_regex[browserName];
     if (pattern) {
       code += `
        window_${browserName}_re = filter_keyvals_regex(events, "app", ${JSON.stringify(pattern)});
        window_${browserName} = sort_by_timestamp(concat(window_${browserName}, window_${browserName}_re));`;
     }
 
+    const combineChromeArcDup = mixChromeArc && (browserName === 'chrome' || browserName === 'arc');
     code += `
        events_${browserName} = filter_period_intersect(events_${browserName}, window_${browserName});
-       events_${browserName} = split_url_events(events_${browserName});
+       events_${browserName} = split_url_events(events_${browserName});`;
+    if (!combineChromeArcDup) {
+      code += `
        browser_events = concat(browser_events, events_${browserName});
        browser_events = sort_by_timestamp(browser_events);`;
+    }
   });
+
+  if (mixChromeArc) {
+    // Chrome first so current chrome-bucket events win over a stale Arc bucket.
+    code += `
+       chrome_arc_events = union_no_overlap(events_chrome, events_arc);
+       browser_events = concat(browser_events, chrome_arc_events);
+       browser_events = sort_by_timestamp(browser_events);`;
+  }
   return code;
 }
 

src/util/browserAllowlist.ts

@@ -0,0 +1,21 @@
+/**
+ * Detect the silent-empty Browser view: a browser watcher bucket exists, the
+ * query finished, and the window-event intersection came back empty.
+ *
+ * This is the Chromium-fork failure mode (aw-webui#927): events land in
+ * `aw-watcher-web-chrome_*` but `app` is "Dia"/"Arc"/… and matches nothing.
+ * It is also the honest "didn't browse this period" case, so the UI copy
+ * must cover both.
+ *
+ * `top_domains === null` means the query is still in flight (see
+ * `start_loading` in the activity store) and must not fire the hint.
+ */
+export function isBrowserAllowlistMiss(browser: {
+  available: boolean;
+  duration: number;
+  top_domains: unknown[] | null;
+}): boolean {
+  if (!browser.available) return false;
+  if (browser.top_domains === null) return false;
+  return browser.duration === 0 && browser.top_domains.length === 0;
+}

test/unit/browserAllowlist.test.node.ts

@@ -0,0 +1,23 @@
+import { isBrowserAllowlistMiss } from '~/util/browserAllowlist';
+
+describe('isBrowserAllowlistMiss', () => {
+  const empty = { available: true, duration: 0, top_domains: [] as unknown[] };
+
+  test('true when a browser bucket exists but the window intersection is empty', () => {
+    expect(isBrowserAllowlistMiss(empty)).toBe(true);
+  });
+
+  test('false while the query is still in flight (null fields)', () => {
+    expect(isBrowserAllowlistMiss({ ...empty, top_domains: null })).toBe(false);
+  });
+
+  test('false when no browser watcher bucket is present', () => {
+    expect(isBrowserAllowlistMiss({ ...empty, available: false })).toBe(false);
+  });
+
+  test('false when matched browser events exist', () => {
+    expect(isBrowserAllowlistMiss({ available: true, duration: 12, top_domains: [{}] })).toBe(
+      false
+    );
+  });
+});

test/unit/queries.test.node.ts

@@ -14,6 +14,9 @@
  *             'Google-chrome-beta', 'Google-chrome-unstable'
  *             (Flatpak app IDs retained as exact: 'com.google.Chrome', 'com.google.ChromeDev',
  *              'org.chromium.Chromium')
+ *             Chromium forks that report through the chrome extension bucket (#927):
+ *             'Arc', 'arc.exe', 'Arc.exe', 'Dia', 'Dia.exe'
+ *             (macOS bundle ID retained as exact: 'company.thebrowser.dia')
  *
  *   Firefox:  'Firefox', 'Firefox.exe', 'firefox', 'firefox.exe',
  *             'Firefox Developer Edition', 'firefoxdeveloperedition',
@@ -55,7 +58,12 @@
  *             (Flatpak app ID retained: 'one.ablaze.floorp')
  */
 
-import { browser_appname_regex, querystr_to_array } from '~/queries';
+import {
+  browser_appname_regex,
+  browser_appnames,
+  fullDesktopQuery,
+  querystr_to_array,
+} from '~/queries';
 
 // Convert ActivityWatch (?i) patterns to JS RegExp with i flag for testing.
 // AW server uses Python-style (?i) inline flag; JS uses RegExp 'i' flag instead.
@@ -97,8 +105,9 @@ describe('browser_appname_regex', () => {
 
   test('chrome pattern does not false-positive', () => {
     const re = toRegex(browser_appname_regex.chrome);
-    // Flatpak app IDs are in the exact list, not matched by regex
+    // Flatpak / bundle IDs are in the exact list, not matched by regex
     expect(re.test('com.google.Chrome')).toBe(false);
+    expect(re.test('company.thebrowser.dia')).toBe(false);
     expect(re.test('Slack')).toBe(false);
     expect(re.test('Electron')).toBe(false);
     // The fork alternatives are anchored, so names merely starting with them don't match
@@ -107,6 +116,10 @@ describe('browser_appname_regex', () => {
     expect(re.test('Dialog')).toBe(false);
   });
 
+  test('chrome exact list includes the Dia macOS bundle id', () => {
+    expect(browser_appnames.chrome).toContain('company.thebrowser.dia');
+  });
+
   test('firefox pattern matches all known Firefox/LibreWolf/Waterfox app names', () => {
     const re = toRegex(browser_appname_regex.firefox);
     // Every entry from the old exact-match list
@@ -244,6 +257,76 @@ describe('browser_appname_regex', () => {
   });
 });
 
+describe('chrome fork matching in generated query', () => {
+  const params = {
+    bid_window: 'aw-watcher-window_testhost',
+    bid_afk: 'aw-watcher-afk_testhost',
+    filter_afk: true,
+    include_audible: false,
+    categories: [],
+    filter_categories: [],
+  };
+
+  test('chrome bucket query includes Dia bundle id and process-name regex', () => {
+    const query = fullDesktopQuery({
+      ...params,
+      bid_browsers: ['aw-watcher-web-chrome_testhost'],
+    }).join('\n');
+    expect(query).toContain('company.thebrowser.dia');
+    // JSON.stringify doubles the regex backslash, so the query text has \\.
+    expect(query).toContain('dia(\\\\.exe)?$');
+  });
+
+  test('mixed chrome and Arc buckets preserve both matching paths without overlap', () => {
+    const query = fullDesktopQuery({
+      ...params,
+      bid_browsers: ['aw-watcher-web-chrome_testhost', 'aw-watcher-web-arc_testhost'],
+    }).join('\n');
+    const chromeWindowFilter = query.slice(
+      query.indexOf('window_chrome_re ='),
+      query.indexOf('events_chrome = filter_period_intersect')
+    );
+    expect(chromeWindowFilter).toContain('arc(\\\\.exe)?$');
+    expect(query).toContain('window_arc_re =');
+    expect(query).toContain('arc(\\\\.exe)?$');
+    // Duplicate chrome/Arc streams are unioned with each other, not with every browser.
+    expect(query).toContain('chrome_arc_events = union_no_overlap(events_chrome, events_arc);');
+    expect(query).toContain('browser_events = concat(browser_events, chrome_arc_events);');
+    expect(query).not.toContain(
+      'browser_events = union_no_overlap(browser_events, events_chrome);'
+    );
+    expect(query).not.toContain('browser_events = union_no_overlap(browser_events, events_arc);');
+  });
+
+  test('unrelated browser buckets concat instead of dropping overlaps', () => {
+    const query = fullDesktopQuery({
+      ...params,
+      bid_browsers: ['aw-watcher-web-chrome_testhost', 'aw-watcher-web-firefox_testhost'],
+    }).join('\n');
+    expect(query).toContain('browser_events = concat(browser_events, events_chrome);');
+    expect(query).toContain('browser_events = concat(browser_events, events_firefox);');
+    expect(query).not.toContain('union_no_overlap(browser_events, events_');
+    expect(query).not.toContain('union_no_overlap(events_chrome, events_firefox)');
+    expect(query).not.toContain('chrome_arc_events');
+  });
+
+  test('chrome+arc union does not swallow a third browser', () => {
+    const query = fullDesktopQuery({
+      ...params,
+      bid_browsers: [
+        'aw-watcher-web-chrome_testhost',
+        'aw-watcher-web-arc_testhost',
+        'aw-watcher-web-firefox_testhost',
+      ],
+    }).join('\n');
+    expect(query).toContain('chrome_arc_events = union_no_overlap(events_chrome, events_arc);');
+    expect(query).toContain('browser_events = concat(browser_events, events_firefox);');
+    expect(query).toContain('browser_events = concat(browser_events, chrome_arc_events);');
+    expect(query).not.toContain('union_no_overlap(browser_events, events_firefox)');
+    expect(query).not.toContain('union_no_overlap(browser_events, events_chrome)');
+  });
+});
+
 describe('querystr_to_array', () => {
   test('splits simple multi-statement query correctly', () => {
     const query = 'events = query_bucket("aw-watcher-window_host"); RETURN = {"events": events};';