Call recorder: parse conference links from calendar event text (#22555)

2026-07-05 21:14:13 +05:30
parent 3cd4498bb2
commit fe8ec0463a
6 changed files with 225 additions and 7 deletions
@@ -7,7 +7,7 @@ import {
  fetchAllNodes,
  type ConnectionPage,
 } from 'src/logic-functions/data/fetch-all-nodes.util';
-import { isNonEmptyString } from 'src/logic-functions/utils/is-non-empty-string.util';
+import { resolveConferenceLinkUrl } from 'src/logic-functions/domain/resolve-conference-link-url.util';
 import { stripRestrictedFieldValue } from 'src/logic-functions/data/strip-restricted-field-value.util';

 type CalendarEventNode = {
@@ -18,6 +18,8 @@ type CalendarEventNode = {
  endsAt?: string | null;
  iCalUid?: string | null;
  conferenceLink?: { primaryLinkUrl?: string | null } | null;
+  location?: string | null;
+  description?: string | null;
  callRecorderPreference?: string | null;
 };

@@ -49,6 +51,8 @@ export const fetchCalendarEventsByFilter = async (
              conferenceLink: {
                primaryLinkUrl: true,
              },
+              location: true,
+              description: true,
              callRecorderPreference: true,
            },
          },
@@ -68,11 +72,13 @@ export const fetchCalendarEventsByFilter = async (
    startsAt: calendarEvent.startsAt ?? undefined,
    endsAt: calendarEvent.endsAt ?? undefined,
    iCalUid: calendarEvent.iCalUid ?? undefined,
-    conferenceLinkUrl: isNonEmptyString(
-      calendarEvent.conferenceLink?.primaryLinkUrl,
-    )
-      ? calendarEvent.conferenceLink.primaryLinkUrl
-      : undefined,
+    conferenceLinkUrl: resolveConferenceLinkUrl({
+      conferenceLinkUrl: calendarEvent.conferenceLink?.primaryLinkUrl,
+      location: stripRestrictedFieldValue(calendarEvent.location ?? undefined),
+      description: stripRestrictedFieldValue(
+        calendarEvent.description ?? undefined,
+      ),
+    }),
    callRecorderPreference: isString(calendarEvent.callRecorderPreference)
      ? calendarEvent.callRecorderPreference
      : undefined,
@@ -0,0 +1,80 @@
+import { describe, expect, it } from 'vitest';
+
+import { extractConferenceLinkUrlFromText } from 'src/logic-functions/domain/extract-conference-link-url-from-text.util';
+
+describe('extractConferenceLinkUrlFromText', () => {
+  it.each([
+    [
+      'extracts a Zoom link from a plain text invitation',
+      'Felix is inviting you to a scheduled Zoom meeting.\n\nJoin Zoom Meeting\nhttps://us02web.zoom.us/j/81234567890?pwd=aBcDeF123\n\nMeeting ID: 812 3456 7890',
+      'https://us02web.zoom.us/j/81234567890?pwd=aBcDeF123',
+    ],
+    [
+      'extracts a bare Zoom link (location field)',
+      'https://zoom.us/j/81234567890',
+      'https://zoom.us/j/81234567890',
+    ],
+    [
+      'extracts a Zoom personal room link',
+      'Join me at https://company.zoom.us/my/felix today',
+      'https://company.zoom.us/my/felix',
+    ],
+    [
+      'extracts a Zoom Government link',
+      'Join: https://example.zoomgov.com/j/1609618851',
+      'https://example.zoomgov.com/j/1609618851',
+    ],
+    [
+      'extracts a Zoom link from an HTML body and decodes &amp;',
+      '<div><a href="https://us02web.zoom.us/j/81234567890?pwd=aBcDeF123&amp;uname=Felix">Join Zoom Meeting</a></div>',
+      'https://us02web.zoom.us/j/81234567890?pwd=aBcDeF123&uname=Felix',
+    ],
+    [
+      'strips trailing sentence punctuation',
+      'Join here: https://zoom.us/j/81234567890.',
+      'https://zoom.us/j/81234567890',
+    ],
+    [
+      'extracts a Google Meet link',
+      'Join the call on https://meet.google.com/abc-defg-hij',
+      'https://meet.google.com/abc-defg-hij',
+    ],
+    [
+      'extracts a Microsoft Teams link',
+      'Click here to join: https://teams.microsoft.com/l/meetup-join/19%3ameeting_ABC123%40thread.v2/0?context=%7b%22Tid%22%3a%22111%22%7d',
+      'https://teams.microsoft.com/l/meetup-join/19%3ameeting_ABC123%40thread.v2/0?context=%7b%22Tid%22%3a%22111%22%7d',
+    ],
+    [
+      'extracts a Webex link',
+      'Join at https://company.webex.com/company/j.php?MTID=m1234abcd',
+      'https://company.webex.com/company/j.php?MTID=m1234abcd',
+    ],
+    [
+      'extracts a GoTo Meeting link',
+      'https://global.gotomeeting.com/join/123456789',
+      'https://global.gotomeeting.com/join/123456789',
+    ],
+    [
+      'prefers a Zoom link over a Google Meet link when both are present',
+      'Video call: https://meet.google.com/abc-defg-hij\nJoin Zoom Meeting\nhttps://zoom.us/j/81234567890',
+      'https://zoom.us/j/81234567890',
+    ],
+  ])('%s', (_label, text, expectedUrl) => {
+    expect(extractConferenceLinkUrlFromText(text)).toBe(expectedUrl);
+  });
+
+  it.each([
+    ['a non-conferencing URL', 'Agenda: https://docs.google.com/document/d/abc'],
+    ['a Zoom marketing page URL', 'Learn more at https://zoom.us/pricing'],
+    ['plain text without links', 'Conference Room A'],
+    ['an empty string', ''],
+    ['a blank string', '   '],
+  ])('returns undefined for %s', (_label, text) => {
+    expect(extractConferenceLinkUrlFromText(text)).toBeUndefined();
+  });
+
+  it('returns undefined for null and undefined', () => {
+    expect(extractConferenceLinkUrlFromText(null)).toBeUndefined();
+    expect(extractConferenceLinkUrlFromText(undefined)).toBeUndefined();
+  });
+});
@@ -0,0 +1,56 @@
+import { describe, expect, it } from 'vitest';
+
+import { resolveConferenceLinkUrl } from 'src/logic-functions/domain/resolve-conference-link-url.util';
+
+describe('resolveConferenceLinkUrl', () => {
+  it('returns the structured conference link when present', () => {
+    expect(
+      resolveConferenceLinkUrl({
+        conferenceLinkUrl: 'https://meet.google.com/abc-defg-hij',
+        location: 'https://zoom.us/j/81234567890',
+        description: 'Join Zoom Meeting https://zoom.us/j/99999999999',
+      }),
+    ).toBe('https://meet.google.com/abc-defg-hij');
+  });
+
+  it('parses the location when the structured link is empty', () => {
+    expect(
+      resolveConferenceLinkUrl({
+        conferenceLinkUrl: undefined,
+        location: 'https://zoom.us/j/81234567890',
+        description: 'Weekly sync',
+      }),
+    ).toBe('https://zoom.us/j/81234567890');
+  });
+
+  it('parses the description when the structured link and location have none', () => {
+    expect(
+      resolveConferenceLinkUrl({
+        conferenceLinkUrl: '',
+        location: 'Conference Room A',
+        description:
+          'Felix is inviting you to a scheduled Zoom meeting.\nJoin Zoom Meeting\nhttps://us02web.zoom.us/j/81234567890?pwd=aBcDeF123',
+      }),
+    ).toBe('https://us02web.zoom.us/j/81234567890?pwd=aBcDeF123');
+  });
+
+  it('prefers the location link over the description link', () => {
+    expect(
+      resolveConferenceLinkUrl({
+        conferenceLinkUrl: null,
+        location: 'https://zoom.us/j/11111111111',
+        description: 'Join Zoom Meeting https://zoom.us/j/22222222222',
+      }),
+    ).toBe('https://zoom.us/j/11111111111');
+  });
+
+  it('returns undefined when no conference link exists anywhere', () => {
+    expect(
+      resolveConferenceLinkUrl({
+        conferenceLinkUrl: undefined,
+        location: 'Conference Room A',
+        description: 'Quarterly review agenda',
+      }),
+    ).toBeUndefined();
+  });
+});
@@ -0,0 +1,33 @@
+import { isNonEmptyString } from 'src/logic-functions/utils/is-non-empty-string.util';
+
+// Ordered by provider precedence: the first matching provider wins (e.g. a
+// pasted Zoom invitation beats an auto-added Meet link).
+const CONFERENCE_LINK_URL_PATTERNS: RegExp[] = [
+  /https:\/\/(?:[\w-]+\.)*(?:zoom\.us|zoomgov\.com)\/(?:j|my|s|w|wc\/join)\/[^\s"'<>\\]+/i,
+  /https:\/\/meet\.google\.com\/[^\s"'<>\\]+/i,
+  /https:\/\/teams\.(?:microsoft|live)\.com\/(?:l\/meetup-join|meet)\/[^\s"'<>\\]+/i,
+  /https:\/\/(?:[\w-]+\.)*webex\.com\/(?:meet\/|join\/|[\w-]+\/j\.php\?)[^\s"'<>\\]+/i,
+  /https:\/\/(?:[\w-]+\.)*(?:gotomeeting\.com\/join|gotomeet\.me)\/[^\s"'<>\\]+/i,
+];
+
+// Outlook HTML bodies encode ampersands; plain-text links can end mid-sentence.
+const cleanExtractedConferenceLinkUrl = (url: string): string =>
+  url.replace(/&amp;/gi, '&').replace(/[.,;:!?)\]]+$/, '');
+
+export const extractConferenceLinkUrlFromText = (
+  text: string | null | undefined,
+): string | undefined => {
+  if (!isNonEmptyString(text)) {
+    return undefined;
+  }
+
+  for (const pattern of CONFERENCE_LINK_URL_PATTERNS) {
+    const match = text.match(pattern);
+
+    if (match !== null) {
+      return cleanExtractedConferenceLinkUrl(match[0]);
+    }
+  }
+
+  return undefined;
+};
@@ -0,0 +1,25 @@
+import { extractConferenceLinkUrlFromText } from 'src/logic-functions/domain/extract-conference-link-url-from-text.util';
+import { isNonEmptyString } from 'src/logic-functions/utils/is-non-empty-string.util';
+
+type ResolveConferenceLinkUrlInput = {
+  conferenceLinkUrl: string | null | undefined;
+  location: string | null | undefined;
+  description: string | null | undefined;
+};
+
+// The server only fills conferenceLink from structured provider data; Zoom and
+// other third-party links often exist only as text in location/description.
+export const resolveConferenceLinkUrl = ({
+  conferenceLinkUrl,
+  location,
+  description,
+}: ResolveConferenceLinkUrlInput): string | undefined => {
+  if (isNonEmptyString(conferenceLinkUrl)) {
+    return conferenceLinkUrl;
+  }
+
+  return (
+    extractConferenceLinkUrlFromText(location) ??
+    extractConferenceLinkUrlFromText(description)
+  );
+};
@@ -11,6 +11,8 @@ import { type RemovedCallRecorderOccurrence } from 'src/logic-functions/types/re
 import { computeRealMeetingKey } from 'src/logic-functions/domain/compute-real-meeting-key.util';
 import { getUniqueSortedIds } from 'src/logic-functions/utils/get-unique-sorted-ids.util';
 import { reconcileCallRecorderForCalendarEventIds } from 'src/logic-functions/flows/reconcile-call-recorder.util';
+import { resolveConferenceLinkUrl } from 'src/logic-functions/domain/resolve-conference-link-url.util';
+import { stripRestrictedFieldValue } from 'src/logic-functions/data/strip-restricted-field-value.util';

 const CALENDAR_EVENT_OBJECT_NAME = 'calendarEvent';

@@ -18,14 +20,20 @@ const CALL_RECORDER_RELEVANT_CALENDAR_EVENT_FIELDS = [
  'title',
  'callRecorderPreference',
  'conferenceLink',
+  'location',
+  'description',
  'startsAt',
  'endsAt',
  'isCanceled',
  'iCalUid',
 ];

+// location and description are key fields because the conference link is
+// parsed out of them when the structured conferenceLink is empty.
 const CALL_RECORDER_KEY_CALENDAR_EVENT_FIELDS = [
  'conferenceLink',
+  'location',
+  'description',
  'startsAt',
  'iCalUid',
 ];
@@ -33,6 +41,8 @@ const CALL_RECORDER_KEY_CALENDAR_EVENT_FIELDS = [
 type CalendarEventForDatabaseEvent = {
  id: string;
  conferenceLink?: { primaryLinkUrl?: string | null } | null;
+  location?: string | null;
+  description?: string | null;
  iCalUid?: string | null;
  startsAt?: string | null;
 };
@@ -156,7 +166,15 @@ const buildRemovedOccurrence = (
    calendarEventId: calendarEvent.id,
    realMeetingKey: computeRealMeetingKey({
      calendarEventId: calendarEvent.id,
-      conferenceLinkUrl: calendarEvent.conferenceLink?.primaryLinkUrl,
+      conferenceLinkUrl: resolveConferenceLinkUrl({
+        conferenceLinkUrl: calendarEvent.conferenceLink?.primaryLinkUrl,
+        location: stripRestrictedFieldValue(
+          calendarEvent.location ?? undefined,
+        ),
+        description: stripRestrictedFieldValue(
+          calendarEvent.description ?? undefined,
+        ),
+      }),
      iCalUid: calendarEvent.iCalUid ?? undefined,
      startsAt: calendarEvent.startsAt ?? undefined,
    }),