feat: enhance Telegram import functionality to support delays and cancellations

- Updated telegramCancellationParser to include parsing for delay notices. - Added new test cases for handling delay messages and stop-point cancellations. - Modified telegramImport integration tests to validate new delay handling. - Enhanced documentation to reflect the addition of delay notifications. - Updated TelegramImportPanel component to accommodate new alert types. - Introduced new fixture files for testing delay and cancellation messages.
2026-10-09 02:59:52 +03:00
parent c0d40f9b90
commit 249c338c2a
8 changed files with 390 additions and 69 deletions
--- a/backend/test/telegramCancellationParser.test.js
+++ b/backend/test/telegramCancellationParser.test.js
@@ -2,7 +2,7 @@ const test = require('node:test');
 const assert = require('node:assert/strict');
 const fs = require('node:fs');
 const path = require('node:path');
-const { parseCancellation, resolveCancellationStops, matchCancellationRoutes, buildCancellationAlert, normalizeStopKey, dayBounds } = require('../src/services/telegramCancellationParser');
+const { parseTransitNotice, buildTransitNoticeAlert, parseCancellation, resolveCancellationStops, matchCancellationRoutes, buildCancellationAlert, normalizeStopKey, dayBounds } = require('../src/services/telegramCancellationParser');
 const hospitalText = fs.readFileSync(path.join(__dirname, 'fixtures/telegram-hospital-cancellation.txt'), 'utf8');

 const text = `Уважаемые пассажиры!
@@ -128,8 +128,8 @@ test('dictionary selects the actual stop ID and uses its canonical name in conci
  const matched = matchCancellationRoutes(parsed, routes);
  assert.deepEqual(matched.plans[0].routeIds, [1, 2, 3]);
  const alert = buildCancellationAlert(matched.plans[0], '2026-10-02', 'https://t.me/avp2Polotsk/1166', '2026-10-02T12:52Z', 1);
-  assert.equal(alert.message, '02.10.2026: не выполняются рейсы:\nАвтовокзал — 15:43, 16:43\nЗодчего Иоанна — 15:58, 16:58\nИсточник: https://t.me/avp2Polotsk/1166');
-  assert.doesNotMatch(alert.message, /Уважаемые|С УВАЖЕНИЕМ/);
+  assert.equal(alert.message, '02.10.2026: не выполняются рейсы:\nАвтовокзал — 15:43, 16:43\nЗодчего Иоанна — 15:58, 16:58');
+  assert.doesNotMatch(alert.message, /Уважаемые|С УВАЖЕНИЕМ|Источник:/);
  assert.equal(normalizeStopKey(' « ЗОД-ЧЕГО » '), normalizeStopKey('Зодчего'));
 });

@@ -163,3 +163,117 @@ test('a shared explicit departure list may name several route numbers', () => {
    assert.ok(parsed.groups.every(g => g.departures[0].times.join(',') === '18:00,19:00'));
  }
 });
+
+const delayText = fs.readFileSync(path.join(__dirname, 'fixtures/telegram-delay-route-4.txt'), 'utf8');
+const stopPointText = fs.readFileSync(path.join(__dirname, 'fixtures/telegram-stop-point-cancellation.txt'), 'utf8');
+const route4Stops = [{ id: 41, name: 'Мариненко' }, { id: 42, name: 'Боровуха' }];
+const route4 = [{ id: 4, route_number: '4' }, { id: 40, route_number: '4' }];
+
+function route4Alert(parsed, publishedAt) {
+  const resolved = resolveCancellationStops(parsed, route4Stops, []);
+  const matched = matchCancellationRoutes(resolved, route4);
+  assert.deepEqual(matched.reasons, []);
+  return buildTransitNoticeAlert(matched.plans, parsed.serviceDate, 'https://t.me/avp2Polotsk/1700', publishedAt, 1, parsed);
+}
+
+test('screenshot delay is a delay notice with both departures and Minsk publication-day expiry', () => {
+  const publishedAt = '2026-10-08T16:14:01Z';
+  const parsed = parseTransitNotice({ text: delayText, publishedAt });
+  assert.equal(parsed.status, 'ready');
+  assert.equal(parsed.alertType, 'delay');
+  assert.equal(parsed.serviceDate, '2026-10-08');
+  assert.deepEqual(parsed.groups[0].departures.map(d => d.times), [['19:05'], ['20:00']]);
+  const alert = route4Alert(parsed, publishedAt);
+  assert.equal(alert.alert_type, 'delay');
+  assert.deepEqual(alert.route_numbers, ['4']);
+  assert.equal(alert.title, 'Задержка отдельных рейсов №4');
+  assert.equal(alert.message, '08.10.2026: с опозданием выполняются рейсы:\nМариненко — 19:05\nБоровуха — 20:00');
+  assert.equal(alert.start_time, '2026-10-08T16:14:01.000Z');
+  assert.equal(alert.end_time, '2026-10-08T21:00:00.000Z');
+});
+
+test('screenshot stop-point heading and quoted continuation preserve four cancelled departures', () => {
+  const publishedAt = '2026-10-08T02:03:13Z';
+  const parsed = parseTransitNotice({ text: stopPointText, publishedAt, editedAt: '2026-10-09T01:00Z' }, { preserveSourceStops: true });
+  assert.equal(parsed.status, 'ready');
+  assert.equal(parsed.alertType, 'cancellation');
+  assert.equal(parsed.serviceDate, '2026-10-08');
+  assert.deepEqual(parsed.groups, [{ routeNumber: '4', departures: [
+    { stop: 'Мариненко', sourceStop: 'Мариненко', times: ['06:00', '07:30'] },
+    { stop: 'Боровуха', sourceStop: 'Боровуха', times: ['06:40', '08:15'] }
+  ] }]);
+  const alert = route4Alert(parsed, publishedAt);
+  assert.equal(alert.alert_type, 'cancellation');
+  assert.equal(alert.message, '08.10.2026: не выполняются рейсы:\nМариненко — 06:00, 07:30\nБоровуха — 06:40, 08:15');
+  assert.equal(alert.end_time, '2026-10-08T21:00:00.000Z');
+});
+
+test('stop-point heading supports quote variants, mixed prefixes and real dates independently from times', () => {
+  for (const [opening, closing] of [['"', '"'], ['«', '»'], ['“', '”']]) {
+    const text = stopPointText.replace(/"(Мариненко|Боровуха)"/g, (_, name) => opening + name + closing).replace('6.00', '09.10');
+    const parsed = parseTransitNotice({ text: 'На 09.10\n' + text, publishedAt: '2026-10-08T02:03:13Z' });
+    assert.equal(parsed.status, 'ready');
+    assert.equal(parsed.serviceDate, '2026-10-09');
+    assert.deepEqual(parsed.groups[0].departures[0].times, ['09:10', '07:30']);
+  }
+  const mixed = stopPointText.replace('"Боровуха"', 'ОТ ОСТАНОВКИ «Боровуха»');
+  assert.equal(parseTransitNotice({ text: mixed, publishedAt: '2026-10-08T02:03:13Z' }).status, 'ready');
+  for (const text of [stopPointText.replace('6.00', '25.00'), 'На 31.02.2026\n' + stopPointText,
+    stopPointText.replace('с остановочного пункта', ''), stopPointText + '\nДополнительно в 09:00']) {
+    assert.equal(parseTransitNotice({ text, publishedAt: '2026-10-08T02:03:13Z' }).status, 'review');
+  }
+});
+
+test('possible delay stays possible and unspecified duration is never invented or silently dropped', () => {
+  const publishedAt = '2026-10-08T16:14:01Z';
+  for (const wording of ['могут выполняться рейсы с опозданием', 'рейсы могут прибыть с опозданием', 'рейсы могут прийти с опозданием']) {
+    const parsed = parseTransitNotice({ text: delayText.replace('будут выполняться рейсы с опозданием', wording), publishedAt });
+    assert.equal(parsed.status, 'ready');
+    assert.equal(parsed.delayPossible, true);
+    const alert = route4Alert(parsed, publishedAt);
+    assert.match(alert.title, /Возможное опоздание/);
+    assert.match(alert.message, /возможны опоздания рейсов/);
+    assert.doesNotMatch(alert.message, /минут|час/);
+  }
+  for (const duration of [' на 20 минут', ' на 20 мин.', ' на 1 час']) {
+    const parsed = parseTransitNotice({ text: delayText.replace('с опозданием:', 'с опозданием' + duration + ':'), publishedAt });
+    assert.equal(parsed.status, 'review');
+    assert.equal(parsed.alertType, 'delay');
+    assert.match(parsed.reasons[0], /величина опоздания/);
+  }
+});
+
+test('mixed types, negated delay, conditions and incomplete departures stay on review; unrelated execution is ignored', () => {
+  const publishedAt = '2026-10-08T16:14:01Z';
+  for (const text of [delayText + '\n' + stopPointText,
+    delayText.replace('будут выполняться', 'не могут выполняться'),
+    delayText + '\nДвижение восстановлено.',
+    delayText.replace(' НА 20.00', ''),
+    delayText + '\nОстальные рейсы будут выполняться.']) {
+    assert.equal(parseTransitNotice({ text, publishedAt }).status, 'review', text);
+  }
+  assert.equal(parseTransitNotice({ text: 'Будут выполняться рейсы по маршруту №4.', publishedAt }).status, 'ignored');
+  assert.equal(parseTransitNotice({ text: 'С опозданием поздравляем с праздником!', publishedAt }).status, 'ignored');
+  const withMedia = parseTransitNotice({ text: delayText, publishedAt, hasMedia: true });
+  assert.equal(withMedia.status, 'review');
+  assert.equal(withMedia.alertType, 'delay');
+});
+
+test('multiple delayed routes are combined while every stop and departure remains visible', () => {
+  const publishedAt = '2026-10-08T16:14:01Z';
+  const parsed = parseTransitNotice({ text: delayText + '\nПО МАРШРУТУ №27 С ОСТАНОВКИ АВ/ВОКЗАЛ НА 20.15,21.00.', publishedAt });
+  const resolved = resolveCancellationStops(parsed, [...route4Stops, { id: 1, name: 'Автовокзал' }], []);
+  const matched = matchCancellationRoutes(resolved, [...route4, { id: 27, route_number: '27' }]);
+  const alert = buildTransitNoticeAlert(matched.plans, parsed.serviceDate, '', publishedAt, 1, parsed);
+  assert.equal(alert.alert_type, 'delay');
+  assert.deepEqual(alert.route_numbers, ['4', '27']);
+  assert.match(alert.message, /Маршрут №4:\nМариненко — 19:05\nБоровуха — 20:00/);
+  assert.match(alert.message, /Маршрут №27:\nАвтовокзал — 20:15, 21:00/);
+});
+
+test('calendar words in a stop name are not mistaken for a service date', () => {
+  const parsed = parseTransitNotice({ text: stopPointText.replace('Мариненко', '1 Мая'), publishedAt: '2026-10-08T02:03:13Z' });
+  assert.equal(parsed.status, 'ready');
+  assert.equal(parsed.serviceDate, '2026-10-08');
+  assert.equal(parsed.groups[0].departures[0].stop, '1 Мая');
+});