@agentrhq/webcmd 0.1.1 → 0.2.0

package/clis/ctrip/flight.js

@@ -1,136 +0,0 @@
-/**
- * 携程机票 oneway search — domestic + international flight search by route + date.
- *
- * 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).
- *
- * 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 '@agentrhq/webcmd/errors';
-import { cli, Strategy } from '@agentrhq/webcmd/registry';
-import { buildFlightExtractJs, buildScrollUntilJs, parseIataCode, parseIsoDate } from './utils.js';
-
-const MIN_LIMIT = 1;
-const MAX_LIMIT = 50;
-const DEFAULT_LIMIT = 20;
-
-function parseFlightLimit(raw) {
-    if (raw === undefined || raw === null || raw === '') return DEFAULT_LIMIT;
-    const parsed = Number(raw);
-    if (!Number.isFinite(parsed) || !Number.isInteger(parsed)) {
-        throw new ArgumentError(`--limit must be an integer between ${MIN_LIMIT} and ${MAX_LIMIT}, got ${JSON.stringify(raw)}`);
-    }
-    if (parsed < MIN_LIMIT || parsed > MAX_LIMIT) {
-        throw new ArgumentError(`--limit must be between ${MIN_LIMIT} and ${MAX_LIMIT}, got ${parsed}`);
-    }
-    return parsed;
-}
-
-/**
- * Wait for `.flight-list > span > div` to render (the post-load XHR settles
- * 1-3s after navigation), or detect a captcha/login redirect.
- */
-const WAIT_FOR_FLIGHTS_JS = `
-  new Promise((resolve) => {
-    const detect = () => {
-      if (location.pathname.includes('captcha') || /验证码|verify the human/i.test(document.body?.innerText || '')) return 'captcha';
-      if (document.querySelector('.flight-list > span > div')) return 'content';
-      return null;
-    };
-    const found = detect();
-    if (found) return resolve(found);
-    const observer = new MutationObserver(() => {
-      const result = detect();
-      if (result) { observer.disconnect(); resolve(result); }
-    });
-    observer.observe(document.documentElement, { childList: true, subtree: true });
-    setTimeout(() => { observer.disconnect(); resolve('timeout'); }, 8000);
-  })
-`;
-
-cli({
-    site: 'ctrip',
-    name: 'flight',
-    access: 'read',
-    description: '搜索携程一程机票（按出发/到达 IATA 三字码 + 日期）',
-    domain: 'flights.ctrip.com',
-    strategy: Strategy.COOKIE,
-    browser: true,
-    navigateBefore: false,
-    args: [
-        { name: 'from', required: true, positional: true, help: 'Departure IATA code (e.g. BJS / PEK)' },
-        { name: 'to', required: true, positional: true, help: 'Arrival IATA code (e.g. SHA / PVG)' },
-        { name: 'date', required: true, help: 'Departure date (YYYY-MM-DD)' },
-        { name: 'limit', type: 'int', default: DEFAULT_LIMIT, help: `Number of flights (${MIN_LIMIT}-${MAX_LIMIT})` },
-    ],
-    columns: [
-        'rank',
-        'airline', 'flightNo', 'aircraft',
-        'departureTime', 'departureAirport',
-        'arrivalTime', 'arrivalAirport', 'terminal',
-        'price', 'currency', 'cabin',
-        'url',
-    ],
-    func: async (page, kwargs) => {
-        const fromCode = parseIataCode('from', kwargs.from);
-        const toCode = parseIataCode('to', kwargs.to);
-        if (fromCode === toCode) {
-            throw new ArgumentError(`--from and --to must differ (got ${fromCode})`);
-        }
-        const date = parseIsoDate('date', kwargs.date);
-        const limit = parseFlightLimit(kwargs.limit);
-
-        const searchUrl =
-            `https://flights.ctrip.com/online/list/oneway-${fromCode.toLowerCase()}-${toCode.toLowerCase()}` +
-            `?depdate=${date}&cabin=Y_S_C_F&adult=1&child=0&infant=0`;
-        await page.goto(searchUrl);
-        const waitResult = await page.evaluate(WAIT_FOR_FLIGHTS_JS);
-        if (waitResult === 'captcha') {
-            throw new AuthRequiredError('flights.ctrip.com', 'Ctrip is asking for a captcha; complete it in your browser session and retry');
-        }
-        if (waitResult !== 'content') {
-            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 raw = await page.evaluate(buildFlightExtractJs());
-        if (!Array.isArray(raw)) {
-            throw new CommandExecutionError('Ctrip flight DOM extraction returned malformed rows');
-        }
-        const rows = raw;
-        if (rows.length === 0) {
-            if (Number(renderedCardCount) > 0) {
-                throw new CommandExecutionError('Ctrip flight cards rendered but parser did not find required flight anchors');
-            }
-            throw new EmptyResultError('ctrip flight', `No flights for ${fromCode}→${toCode} on ${date}`);
-        }
-        const completeRows = rows
-            .filter((r) => r.departureTime && r.departureAirport && r.arrivalTime && r.arrivalAirport && r.airline && r.flightNo)
-            .slice(0, limit)
-            .map((r, i) => ({
-                rank: i + 1,
-                airline: r.airline,
-                flightNo: r.flightNo,
-                aircraft: r.aircraft,
-                departureTime: r.departureTime,
-                departureAirport: r.departureAirport,
-                arrivalTime: r.arrivalTime,
-                arrivalAirport: r.arrivalAirport,
-                terminal: r.terminal,
-                price: r.price,
-                currency: r.currency,
-                cabin: r.cabin,
-                url: searchUrl,
-            }));
-        if (completeRows.length === 0) {
-            throw new CommandExecutionError('Ctrip flight rows were missing required airline/flight/time/airport anchors');
-        }
-        return completeRows;
-    },
-});
-
-export const __test__ = { parseFlightLimit, WAIT_FOR_FLIGHTS_JS };

package/clis/ctrip/hotel-search.js

@@ -1,132 +0,0 @@
-/**
- * 携程酒店 list — search hotels by city + date range.
- *
- * Reads `window.__NEXT_DATA__.props.pageProps.initListData.hotelList` directly
- * from the SSR-rendered hotel listing page. Ctrip serves first 13 hotels
- * (10 organic + ~3 promoted) inline; `&pageSize=N` URL params are ignored
- * server-side so we cap default limit accordingly (see
- * `~/.webcmd/sites/ctrip/notes.md`).
- *
- * Reuses the existing `mapHotelRow` + `pickHotelMapCoords` helpers from utils.js
- * so the column shape stays consistent if future variants (hotel-detail) also
- * project from the same `hotelInfo` shape.
- *
- * Anti-bot: not detected on first-page navigation (PR #1481 recon 2026-05-12).
- */
-import { ArgumentError, AuthRequiredError, CommandExecutionError, EmptyResultError } from '@agentrhq/webcmd/errors';
-import { cli, Strategy } from '@agentrhq/webcmd/registry';
-import { mapHotelRow, parseCityId, parseIsoDate } from './utils.js';
-
-const MIN_LIMIT = 1;
-const MAX_LIMIT = 30;
-const DEFAULT_LIMIT = 10;
-
-function parseHotelLimit(raw) {
-    if (raw === undefined || raw === null || raw === '') return DEFAULT_LIMIT;
-    const parsed = Number(raw);
-    if (!Number.isFinite(parsed) || !Number.isInteger(parsed)) {
-        throw new ArgumentError(`--limit must be an integer between ${MIN_LIMIT} and ${MAX_LIMIT}, got ${JSON.stringify(raw)}`);
-    }
-    if (parsed < MIN_LIMIT || parsed > MAX_LIMIT) {
-        throw new ArgumentError(`--limit must be between ${MIN_LIMIT} and ${MAX_LIMIT}, got ${parsed}`);
-    }
-    return parsed;
-}
-
-/**
- * Wait for SSR state to be populated, or detect a login/captcha gate.
- *
- * Ctrip occasionally serves a captcha redirect (`/captcha`) when traffic
- * looks bot-like; we catch that as AuthRequired so the agent can pop a
- * human session instead of looping on an empty extract.
- */
-const WAIT_FOR_SSR_JS = `
-  new Promise((resolve) => {
-    const detect = () => {
-      if (location.pathname.includes('captcha') || /验证码|verify the human/i.test(document.body?.innerText || '')) return 'captcha';
-      const hotels = window.__NEXT_DATA__?.props?.pageProps?.initListData?.hotelList;
-      if (Array.isArray(hotels)) return 'content';
-      return null;
-    };
-    const found = detect();
-    if (found) return resolve(found);
-    const observer = new MutationObserver(() => {
-      const result = detect();
-      if (result) { observer.disconnect(); resolve(result); }
-    });
-    observer.observe(document.documentElement, { childList: true, subtree: true });
-    setTimeout(() => { observer.disconnect(); resolve('timeout'); }, 5000);
-  })
-`;
-
-const EXTRACT_HOTELS_JS = `
-  (() => {
-    const list = window.__NEXT_DATA__?.props?.pageProps?.initListData?.hotelList;
-    if (!Array.isArray(list)) return null;
-    return list;
-  })()
-`;
-
-function assertCheckinBeforeCheckout(checkin, checkout) {
-    if (Date.parse(checkin + 'T00:00:00Z') >= Date.parse(checkout + 'T00:00:00Z')) {
-        throw new ArgumentError(`--checkin must be earlier than --checkout (got ${checkin} >= ${checkout})`);
-    }
-}
-
-cli({
-    site: 'ctrip',
-    name: 'hotel-search',
-    access: 'read',
-    description: '搜索携程酒店列表（按城市 + 入住/离店日期）',
-    domain: 'hotels.ctrip.com',
-    strategy: Strategy.COOKIE,
-    browser: true,
-    navigateBefore: false,
-    args: [
-        { name: 'city', required: true, positional: true, help: 'Numeric Ctrip city ID (use `ctrip search` or `ctrip hotel-suggest` to discover)' },
-        { name: 'checkin', required: true, help: 'Check-in date (YYYY-MM-DD)' },
-        { name: 'checkout', required: true, help: 'Check-out date (YYYY-MM-DD)' },
-        { name: 'limit', type: 'int', default: DEFAULT_LIMIT, help: `Number of hotels (${MIN_LIMIT}-${MAX_LIMIT}); SSR first page returns ~13 entries` },
-    ],
-    columns: [
-        'rank', 'hotelId', 'name', 'enName',
-        'star', 'score', 'scoreLabel', 'reviewCount',
-        'cityName', 'district', 'address',
-        'lat', 'lon',
-        'price', 'currency', 'url',
-    ],
-    func: async (page, kwargs) => {
-        const cityId = parseCityId(kwargs.city);
-        const checkin = parseIsoDate('checkin', kwargs.checkin);
-        const checkout = parseIsoDate('checkout', kwargs.checkout);
-        assertCheckinBeforeCheckout(checkin, checkout);
-        const limit = parseHotelLimit(kwargs.limit);
-
-        const url = `https://hotels.ctrip.com/hotels/list?city=${cityId}&checkin=${checkin}&checkout=${checkout}`;
-        await page.goto(url);
-        const waitResult = await page.evaluate(WAIT_FOR_SSR_JS);
-        if (waitResult === 'captcha') {
-            throw new AuthRequiredError('hotels.ctrip.com', 'Ctrip is asking for a captcha; complete it in your browser session and retry');
-        }
-        if (waitResult !== 'content') {
-            throw new CommandExecutionError(`Ctrip hotel-search page did not expose SSR hotel list (state=${String(waitResult)})`);
-        }
-        const raw = await page.evaluate(EXTRACT_HOTELS_JS);
-        if (!Array.isArray(raw)) {
-            throw new CommandExecutionError('Ctrip hotel-search returned malformed SSR hotel list');
-        }
-        if (raw.length === 0) {
-            throw new EmptyResultError('ctrip hotel-search', `No hotels for city=${cityId} on ${checkin} → ${checkout}`);
-        }
-        const rows = raw
-            .map((entry, i) => mapHotelRow(entry, i))
-            .filter((row) => row.hotelId && row.name)
-            .slice(0, limit);
-        if (rows.length === 0) {
-            throw new CommandExecutionError('Ctrip hotel-search SSR rows were missing required hotelId/name anchors');
-        }
-        return rows;
-    },
-});
-
-export const __test__ = { parseHotelLimit, assertCheckinBeforeCheckout, WAIT_FOR_SSR_JS, EXTRACT_HOTELS_JS };

package/clis/ctrip/hotel-suggest.js

@@ -1,45 +0,0 @@
-/**
- * 携程酒店 hotel-context suggest — public city/business-area/hotel lookup.
- *
- * Distinct from `ctrip/search` (destination): the same backing endpoint is
- * called with `searchType=H`, surfacing Hotel and BusinessArea rows that the
- * destination flavour does not return.
- */
-import { ArgumentError, EmptyResultError } from '@agentrhq/webcmd/errors';
-import { cli, Strategy } from '@agentrhq/webcmd/registry';
-import { fetchSuggest, mapSuggestRow, parseLimit } from './utils.js';
-
-cli({
-    site: 'ctrip',
-    name: 'hotel-suggest',
-    access: 'read',
-    description: '搜索携程酒店上下文联想：城市、商圈、单酒店匹配',
-    strategy: Strategy.PUBLIC,
-    browser: false,
-    args: [
-        { name: 'query', required: true, positional: true, help: 'Search keyword (city, business area, or hotel name)' },
-        { name: 'limit', type: 'int', default: 15, help: 'Number of results (1-50)' },
-    ],
-    columns: [
-        'rank', 'id', 'type', 'displayType', 'name', 'eName',
-        'cityId', 'cityName', 'provinceName', 'countryName',
-        'lat', 'lon', 'score', 'url',
-    ],
-    func: async (kwargs) => {
-        const query = String(kwargs.query || '').trim();
-        if (!query) {
-            throw new ArgumentError('Search keyword cannot be empty');
-        }
-        const limit = parseLimit(kwargs.limit);
-        const raw = await fetchSuggest(query, 'H');
-        const rows = raw
-            .filter((item) => !!item && typeof item === 'object')
-            .slice(0, limit)
-            .map(mapSuggestRow)
-            .filter((row) => row.name);
-        if (!rows.length) {
-            throw new EmptyResultError('ctrip hotel-suggest', 'Try a city, business area, or hotel keyword such as "陆家嘴" or "汉庭酒店"');
-        }
-        return rows;
-    },
-});

package/clis/ctrip/search.js

@@ -1,41 +0,0 @@
-/**
- * 携程旅行 destination suggest — public city/landmark/scenic-spot lookup.
- */
-import { ArgumentError, EmptyResultError } from '@agentrhq/webcmd/errors';
-import { cli, Strategy } from '@agentrhq/webcmd/registry';
-import { fetchSuggest, mapSuggestRow, parseLimit } from './utils.js';
-
-cli({
-    site: 'ctrip',
-    name: 'search',
-    access: 'read',
-    description: '搜索携程目的地、景区、火车站和地标联想结果',
-    strategy: Strategy.PUBLIC,
-    browser: false,
-    args: [
-        { name: 'query', required: true, positional: true, help: 'Search keyword (city, scenic spot, landmark)' },
-        { name: 'limit', type: 'int', default: 15, help: 'Number of results (1-50)' },
-    ],
-    columns: [
-        'rank', 'id', 'type', 'displayType', 'name', 'eName',
-        'cityId', 'cityName', 'provinceName', 'countryName',
-        'lat', 'lon', 'score', 'url',
-    ],
-    func: async (kwargs) => {
-        const query = String(kwargs.query || '').trim();
-        if (!query) {
-            throw new ArgumentError('Search keyword cannot be empty');
-        }
-        const limit = parseLimit(kwargs.limit);
-        const raw = await fetchSuggest(query, 'D');
-        const rows = raw
-            .filter((item) => !!item && typeof item === 'object')
-            .slice(0, limit)
-            .map(mapSuggestRow)
-            .filter((row) => row.name);
-        if (!rows.length) {
-            throw new EmptyResultError('ctrip search', 'Try a destination, scenic spot, or landmark keyword such as "苏州" or "故宫"');
-        }
-        return rows;
-    },
-});