jackwener · dxxbb · Jul 16, 2026
diff --git a/clis/ctrip/ctrip.test.js b/clis/ctrip/ctrip.test.js
@@ -682,6 +682,39 @@ describe('ctrip buildFlightExtractJs (JSDOM)', () => {
         }]);
     });
 
+    it('extracts current data-testid cards via stable field ids', () => {
+        const html = `
+          <div data-testid="flight-item-1">
+            <span id="airlineName3U8282_1784385000000-0">四川航空</span>
+            <div id="comfort-3U8282_1784385000000-0">
+              <span>3U8282<span>空客320(中)</span></span>
+            </div>
+            <div>22:30</div>
+            <span id="departureFlightTrain3U8282_1784385000000-0">首都国际机场</span>
+            <span id="departureTerminal3U8282_1784385000000-0">T2</span>
+            <div>01:55<span id="crossDays3U8282_1784385000000-0">+1天</span></div>
+            <span id="arrivalFlightTrain3U8282_1784385000000-0">长水国际机场</span>
+            <span id="arrivalTerminal3U8282_1784385000000-0">T1</span>
+            <span id="travelPackage_price_undefined">900<dfn>¥</dfn></span>
+            <div>经济舱 3.3折</div>
+          </div>
+        `;
+        const rows = runExtract(html);
+        expect(rows).toEqual([{
+            airline: '四川航空',
+            flightNo: '3U8282',
+            aircraft: '空客320(中)',
+            departureTime: '22:30',
+            departureAirport: '首都国际机场',
+            arrivalTime: '01:55',
+            arrivalAirport: '长水国际机场',
+            terminal: 'T1',
+            price: 900,
+            currency: '¥',
+            cabin: '经济舱',
+        }]);
+    });
+
     it('omits terminal when not present after arrAirport', () => {
         const html = `
           <div class="flight-list"><span>

diff --git a/clis/ctrip/flight.js b/clis/ctrip/flight.js
@@ -4,15 +4,15 @@
  * Unlike `hotel-search`, the flight rows are NOT in `__NEXT_DATA__` — they
  * arrive via a post-load XHR that the daemon network buffer currently can't
  * capture (see MEMORY `daemon_capture_pipeline_bug_2026_05_07`). We instead
- * extract from the rendered `.flight-list > span > div` cards using a
- * position-anchored innerText parser (see `buildFlightExtractJs` in utils).
+ * extract from the rendered flight cards using stable visible-UI anchors
+ * (see `FLIGHT_CARD_SELECTOR` and `buildFlightExtractJs` in utils).
  *
  * Round-trip + advanced filters (airline whitelist, cabin selection beyond
  * 全舱位) are out of scope for v1 — track in #1481 follow-up if requested.
  */
 import { ArgumentError, AuthRequiredError, CommandExecutionError, EmptyResultError } from '@jackwener/opencli/errors';
 import { cli, Strategy } from '@jackwener/opencli/registry';
-import { buildFlightExtractJs, buildScrollUntilJs, parseIataCode, parseIsoDate } from './utils.js';
+import { FLIGHT_CARD_SELECTOR, buildFlightExtractJs, buildScrollUntilJs, parseIataCode, parseIsoDate } from './utils.js';
 
 const MIN_LIMIT = 1;
 const MAX_LIMIT = 50;
@@ -31,14 +31,14 @@ function parseFlightLimit(raw) {
 }
 
 /**
- * Wait for `.flight-list > span > div` to render (the post-load XHR settles
+ * Wait for a flight card to render (the post-load XHR settles
  * 1-3s after navigation), or detect a captcha/login redirect.
  */
 const WAIT_FOR_FLIGHTS_JS = `
   new Promise((resolve) => {
     const detect = () => {
       if (location.pathname.includes('captcha') || /验证码|verify the human/i.test(document.body?.innerText || '')) return 'captcha';
-      if (document.querySelector('.flight-list > span > div')) return 'content';
+      if (document.querySelector(${JSON.stringify(FLIGHT_CARD_SELECTOR)})) return 'content';
       return null;
     };
     const found = detect();
@@ -96,7 +96,7 @@ cli({
             throw new CommandExecutionError(`Ctrip flight page did not render flight cards (state=${String(waitResult)})`);
         }
         // Scroll until enough flight cards rendered (Ctrip lazy-loads beyond ~8).
-        const renderedCardCount = await page.evaluate(buildScrollUntilJs('.flight-list > span > div', limit));
+        const renderedCardCount = await page.evaluate(buildScrollUntilJs(FLIGHT_CARD_SELECTOR, limit));
         const raw = await page.evaluate(buildFlightExtractJs());
         if (!Array.isArray(raw)) {
             throw new CommandExecutionError('Ctrip flight DOM extraction returned malformed rows');

diff --git a/clis/ctrip/utils.js b/clis/ctrip/utils.js
@@ -314,11 +314,13 @@ export function mapHotelRow(entry, index) {
     };
 }
 
+export const FLIGHT_CARD_SELECTOR = '.flight-list > span > div, [data-testid^="flight-item-"]';
+
 /**
- * Build the browser-context IIFE that extracts flight rows from `.flight-list`.
+ * Build the browser-context IIFE that extracts flight rows from Ctrip's flight cards.
  *
- * Flights are rendered as `.flight-list > span > div` cards. Each card's innerText
- * has a stable ordering (verified 2026-05-12 on bjs→sha route):
+ * Legacy cards use `.flight-list > span > div` and have a stable innerText ordering
+ * (verified 2026-05-12 on bjs→sha route):
  *
  *   [airline, flightNo, aircraft, lowPriceTag?, depTime, depAirport,
  *    arrTime, arrAirport, terminal?, savings?, promo?, currency, price,
@@ -327,7 +329,8 @@ export function mapHotelRow(entry, index) {
  * `lowPriceTag` (e.g. "当日低价") + `terminal` (e.g. "T2") + `savings` + `promo`
  * are optional — we use position-of-first-time-match to anchor and parse around it.
  *
- * The host is baked in so `normalizeUrl` for booking links resolves on the calling site.
+ * Current cards use `data-testid="flight-item-*"` plus stable field-id prefixes.
+ * Keep both paths so the adapter remains compatible while Ctrip rolls the DOM out.
  */
 export function buildFlightExtractJs() {
     return `
@@ -339,7 +342,7 @@ export function buildFlightExtractJs() {
         const isFlightNo = (s) => /^[A-Z0-9]{2}\\d{3,4}[A-Z]?$/.test(s);
 
         const rows = [];
-        document.querySelectorAll('.flight-list > span > div').forEach((card) => {
+        document.querySelectorAll(${JSON.stringify(FLIGHT_CARD_SELECTOR)}).forEach((card) => {
           // Collect ordered text chunks (text nodes only, skip whitespace-only).
           const chunks = [];
           const walk = (node) => {
@@ -353,6 +356,46 @@ export function buildFlightExtractJs() {
             }
           };
           walk(card);
+
+          if (card.matches('[data-testid^="flight-item-"]')) {
+            const airlineNode = card.querySelector('[id^="airlineName"]');
+            const airline = cleanText(airlineNode?.textContent);
+            const idFlightNo = /^airlineName([A-Z0-9]{2}\\d{3,4}[A-Z]?)(?:_|$)/.exec(airlineNode?.id || '')?.[1] || null;
+            const flightNoIdx = chunks.findIndex(isFlightNo);
+            const flightNo = idFlightNo || (flightNoIdx >= 0 ? chunks[flightNoIdx] : null);
+            const aircraftCandidate = flightNoIdx >= 0 ? chunks[flightNoIdx + 1] : null;
+            const aircraft = aircraftCandidate && !isTime(aircraftCandidate) && !/^(共享|经停|直飞)$/.test(aircraftCandidate)
+              ? aircraftCandidate
+              : null;
+            const times = chunks.filter(isTime);
+            const departureAirport = cleanText(card.querySelector('[id^="departureFlightTrain"]')?.textContent) || null;
+            const arrivalAirport = cleanText(card.querySelector('[id^="arrivalFlightTrain"]')?.textContent) || null;
+
+            if (!airline || !isFlightNo(flightNo) || times.length < 2 || !departureAirport || !arrivalAirport) return;
+
+            const priceNode = card.querySelector('[id^="travelPackage_price_"]');
+            const priceText = cleanText(priceNode?.textContent);
+            const priceMatch = /(\\d[\\d,]*(?:\\.\\d+)?)/.exec(priceText);
+            const currency = cleanText(priceNode?.querySelector('dfn')?.textContent) ||
+              ((priceText.match(/[¥$€£]/) || [])[0] ?? null);
+            const cabinMatch = cleanText(card.textContent).match(/(超级经济舱|经济舱|公务\\/头等舱|公务舱|商务舱|头等舱)/);
+
+            rows.push({
+              airline,
+              flightNo,
+              aircraft,
+              departureTime: times[0],
+              departureAirport,
+              arrivalTime: times[1],
+              arrivalAirport,
+              terminal: cleanText(card.querySelector('[id^="arrivalTerminal"]')?.textContent) || null,
+              price: priceMatch ? Number(priceMatch[1].replace(/,/g, '')) : null,
+              currency,
+              cabin: cabinMatch?.[1] || null,
+            });
+            return;
+          }
+
           if (chunks.length < 8) return;
 
           // Anchor on first HH:MM — that's depTime; depAirport is immediately after.