@danceiny/gotry 0.0.1-rc.18 → 0.0.1-rc.20

package/dist/capabilities/session-search.js

@@ -3,6 +3,8 @@ import { extensionCookieNames, extensionSearchJob, classifyBridgeFailure } from
 import { appendFileSync, mkdirSync } from 'node:fs';
 import { dirname } from 'node:path';
 import { buildEntryUrl, NETWORK_HINTS, parseBatchSearch, LOGIN_COOKIE_NAMES, SITE_DOMAIN } from './session/adapters/ctrip-flight.js';
+import { buildHotelEntryUrl, parseCtripHotelList, HOTEL_SITE_HOST } from './session/adapters/ctrip-hotel.js';
+import { buildTrainEntryUrl, parseLeftTicketQuery, TRAIN_SITE_HOST } from './session/adapters/rail-12306.js';
 import { EXTENSION_STORE_URL } from './session/extension-bridge.js';
 const MIN_INTERVAL_MS = 30_000;
 const lastCallAt = new Map();
@@ -48,6 +50,315 @@ export function appendExtensionAudit(auditPath, entry) {
         appendFileSync(auditPath, JSON.stringify(record) + '\n');
     } catch  {}
 }
+export function hotelCityUnresolvedHint(cities) {
+    return `城市 ${cities.join('/')} 不在携程酒店城市码表——先 web 搜「hotels.ctrip.com ${cities[0] ?? ''} 酒店」拿到 list 页 URL 里的 city= 数字,带 cityId 重试`;
+}
+export async function sessionHotelSearch(q) {
+    const started = Date.now();
+    const ts = new Date().toISOString();
+    const site = 'ctrip-hotel';
+    const err = (verdict, error)=>({
+            ok: false,
+            via: 'session-ctrip-hotel-error',
+            evidence: `[会话:${site}@error@${ts}] ${error}`,
+            latencyMs: Date.now() - started,
+            verdict,
+            error
+        });
+    if ((q.checkIn ? 1 : 0) !== (q.checkOut ? 1 : 0) || q.checkIn && !/^\d{4}-\d{2}-\d{2}$/.test(q.checkIn)) {
+        return err('error', 'checkIn/checkOut 须成对且为 YYYY-MM-DD(未定档期可不带日期)');
+    }
+    if (q.checkIn) {
+        const now = new Date();
+        const today = `${now.getFullYear()}-${String(now.getMonth() + 1).padStart(2, '0')}-${String(now.getDate()).padStart(2, '0')}`;
+        if (q.checkIn < today || (q.checkOut ?? '') < q.checkIn) {
+            return err('error', `入住 ${q.checkIn}/退房 ${q.checkOut ?? ''} 不是未来合法区间(今天 ${today})——向用户确认日期后再查`);
+        }
+    }
+    const last = lastCallAt.get(site) ?? 0;
+    if (Date.now() - last < MIN_INTERVAL_MS) {
+        return err('cooldown', `rate limit: last call ${Date.now() - last}ms ago, min ${MIN_INTERVAL_MS}ms`);
+    }
+    lastCallAt.set(site, Date.now());
+    const entry = buildHotelEntryUrl({
+        to: q.to,
+        cityId: q.cityId,
+        checkIn: q.checkIn,
+        checkOut: q.checkOut,
+        adults: q.adults
+    });
+    if (!entry.ok || !entry.url) {
+        return err('error', hotelCityUnresolvedHint(entry.unresolved ?? [
+            q.to
+        ]));
+    }
+    const mode = resolveTransportMode(q.profileDir);
+    if (mode === 'extension') {
+        const login = await extensionCookieNames({
+            site,
+            domain: SITE_DOMAIN.replace(/^\./, ''),
+            ticketNames: LOGIN_COOKIE_NAMES
+        });
+        if (!login.ok) {
+            const verdict = classifyBridgeFailure(login.kind);
+            if (verdict === 'needs-extension') {
+                return {
+                    ok: false,
+                    via: 'session-ctrip-hotel-error',
+                    evidence: '[会话:ctrip-hotel-needs-extension@ts]',
+                    latencyMs: Date.now() - started,
+                    verdict,
+                    error: login.summary,
+                    installUrl: EXTENSION_STORE_URL,
+                    installAction: 'add-to-chrome'
+                };
+            }
+            return err(verdict, login.summary);
+        }
+        if (login.tickets.length === 0 && !q.allowAnonymous) {
+            return err('needs-login', '未检出你本人登录态——调用 gotry_session_login 为用户打开携程登录入口(登录在携程官网完成;gotry 永不经手密码/验证码/cookie 值)');
+        }
+        const r = await extensionSearchJob({
+            site,
+            url: entry.url,
+            timeoutMs: q.timeoutMs
+        });
+        appendExtensionAudit(q.auditPath, {
+            kind: 'extension-session-job',
+            site,
+            url: entry.url,
+            jobId: 'search',
+            result: r.ok ? r.timedOut ? 'timeout' : `body ${r.body.length}B title="${r.title.slice(0, 60)}"` : `${r.kind}:${r.summary.slice(0, 120)}`
+        });
+        if (!r.ok) {
+            const verdict = classifyBridgeFailure(r.kind);
+            if (verdict === 'needs-extension') {
+                return {
+                    ok: false,
+                    via: 'session-ctrip-hotel-error',
+                    evidence: '[会话:ctrip-hotel-needs-extension@ts]',
+                    latencyMs: Date.now() - started,
+                    verdict,
+                    error: r.summary,
+                    installUrl: EXTENSION_STORE_URL,
+                    installAction: 'add-to-chrome'
+                };
+            }
+            return err(verdict, r.summary);
+        }
+        const title = r.title;
+        const head = r.body.slice(0, 5000);
+        if (CHALLENGE_RE.test(title + head)) {
+            return err('challenged', `风控/验证码命中(title=${title.slice(0, 60)});按红线不重试不绕过,交还用户`);
+        }
+        const hotels = parseCtripHotelList(r.body);
+        const verdict = hotels.length > 0 ? 'hit' : 'miss';
+        return {
+            ok: true,
+            via: 'session-ctrip-hotel',
+            evidence: `[会话:${site}@${ts}] ${hotels.length} hotels;transport=extension(被动嗅探,零系统弹窗;扩展零写行为=物理只读)${q.allowAnonymous ? ';anonymous=自检态' : ''}`,
+            latencyMs: Date.now() - started,
+            verdict,
+            hotels
+        };
+    }
+    const t = await openSession({
+        profileDir: q.profileDir,
+        headless: q.headless,
+        auditPath: q.auditPath,
+        mode: mode === 'persistent' ? 'persistent' : 'cdp',
+        newPage: true
+    });
+    if (!t.ok) {
+        return err(classifyTransportFailure(t.summary, q.profileDir === undefined), t.summary);
+    }
+    try {
+        const loggedIn = async ()=>{
+            const cookies = await t.browser.cookies().catch(()=>[]);
+            return cookies.some((c)=>c.domain.includes(SITE_DOMAIN.replace(/^\./, '')) && LOGIN_COOKIE_NAMES.includes(c.name));
+        };
+        if (!await loggedIn() && !q.allowAnonymous) {
+            return err('needs-login', '未检出你本人登录态——调用 gotry_session_login 为用户打开携程登录入口(登录在携程官网完成;gotry 永不经手密码/验证码/cookie 值)');
+        }
+        let settled = false;
+        let body = '';
+        const heard = new Promise((resolve)=>{
+            t.page.on('response', async (res)=>{
+                if (settled) return;
+                const u = res.url();
+                const hostHit = u.includes(HOTEL_SITE_HOST);
+                if (!hostHit) return;
+                try {
+                    const text = await res.text();
+                    if (text && (hostHit && /hotels\.ctrip\.com\/(hotels\/api|domestic\/pc\/api)|GetHotelListBySOA|GetHotelListByCity|HotelSearch|hotelsearch/i.test(u) || text.length <= 2_000_000 && /"hotelList"|"hotelMatchInfos"|"hotelName"/.test(text))) {
+                        body = text;
+                        settled = true;
+                        resolve();
+                    }
+                } catch  {}
+            });
+            setTimeout(()=>resolve(), q.timeoutMs ?? 30_000);
+        });
+        await t.page.goto(entry.url, {
+            waitUntil: 'domcontentloaded',
+            timeout: 30_000
+        });
+        await heard;
+        const title = await t.page.title().catch(()=>'');
+        const headHtml = (await t.page.content().catch(()=>'')).slice(0, 5000);
+        if (CHALLENGE_RE.test(title + headHtml)) {
+            return err('challenged', `风控/验证码命中(title=${title.slice(0, 60)});按红线不重试不绕过,交还用户`);
+        }
+        const hotels = parseCtripHotelList(body);
+        const verdict = hotels.length > 0 ? 'hit' : 'miss';
+        return {
+            ok: true,
+            via: 'session-ctrip-hotel',
+            evidence: `[会话:${site}@${ts}] ${hotels.length} hotels;guard blocked=${t.guard.blockedCount()}/${t.guard.requestCount()}${q.allowAnonymous ? ';anonymous=自检态' : ''}`,
+            latencyMs: Date.now() - started,
+            verdict,
+            hotels
+        };
+    } catch (e) {
+        return err('error', e instanceof Error ? e.message.slice(0, 200) : String(e));
+    } finally{
+        await t.close();
+    }
+}
+export function trainStationUnresolvedHint(cities) {
+    return `城市 ${cities.join('/')} 不在 12306 城市电报码表——web 搜「12306 ${cities[0] ?? ''} 余票」拿到 kyfw 查询页 URL 里的 fs=城市,XXX 三位码,带 fromStationTelecode/toStationTelecode 重试`;
+}
+export async function sessionTrainSearch(q) {
+    const started = Date.now();
+    const ts = new Date().toISOString();
+    const site = 'train-12306';
+    const err = (verdict, error)=>({
+            ok: false,
+            via: 'session-train-12306-error',
+            evidence: `[会话:${site}@error@${ts}] ${error}`,
+            latencyMs: Date.now() - started,
+            verdict,
+            error
+        });
+    const last = lastCallAt.get(site) ?? 0;
+    if (Date.now() - last < MIN_INTERVAL_MS) {
+        return err('cooldown', `rate limit: last call ${Date.now() - last}ms ago, min ${MIN_INTERVAL_MS}ms`);
+    }
+    lastCallAt.set(site, Date.now());
+    const entry = buildTrainEntryUrl({
+        from: q.from,
+        to: q.to,
+        date: q.date,
+        fromStationTelecode: q.fromStationTelecode,
+        toStationTelecode: q.toStationTelecode
+    });
+    if (!entry.ok || !entry.url) {
+        return err('error', trainStationUnresolvedHint(entry.unresolved ?? [
+            q.from,
+            q.to
+        ]));
+    }
+    const mode = resolveTransportMode(q.profileDir);
+    if (mode === 'extension') {
+        const r = await extensionSearchJob({
+            site,
+            url: entry.url,
+            timeoutMs: q.timeoutMs
+        });
+        appendExtensionAudit(q.auditPath, {
+            kind: 'extension-session-job',
+            site,
+            url: entry.url,
+            jobId: 'search',
+            result: r.ok ? r.timedOut ? 'timeout' : `body ${r.body.length}B title="${r.title.slice(0, 60)}"` : `${r.kind}:${r.summary.slice(0, 120)}`
+        });
+        if (!r.ok) {
+            const verdict = classifyBridgeFailure(r.kind);
+            if (verdict === 'needs-extension') {
+                return {
+                    ok: false,
+                    via: 'session-train-12306-error',
+                    evidence: '[会话:train-12306-needs-extension@ts]',
+                    latencyMs: Date.now() - started,
+                    verdict,
+                    error: r.summary,
+                    installUrl: EXTENSION_STORE_URL,
+                    installAction: 'add-to-chrome'
+                };
+            }
+            return err(verdict, r.summary);
+        }
+        const title = r.title;
+        const head = r.body.slice(0, 5000);
+        if (CHALLENGE_RE.test(title + head)) {
+            return err('challenged', `风控/验证码命中(title=${title.slice(0, 60)});按红线不重试不绕过,交还用户`);
+        }
+        const trains = parseLeftTicketQuery(r.body, entry.url);
+        const verdict = trains.length > 0 ? 'hit' : 'miss';
+        return {
+            ok: true,
+            via: 'session-train-12306',
+            evidence: `[会话:${site}@${ts}] ${trains.length} trains;transport=extension(公开查询面,被动嗅探,零系统弹窗;扩展零写行为=物理只读)`,
+            latencyMs: Date.now() - started,
+            verdict,
+            trains
+        };
+    }
+    const t = await openSession({
+        profileDir: q.profileDir,
+        headless: q.headless,
+        auditPath: q.auditPath,
+        mode: mode === 'persistent' ? 'persistent' : 'cdp',
+        newPage: true
+    });
+    if (!t.ok) {
+        return err(classifyTransportFailure(t.summary, q.profileDir === undefined), t.summary);
+    }
+    try {
+        let settled = false;
+        let body = '';
+        const heard = new Promise((resolve)=>{
+            t.page.on('response', async (res)=>{
+                if (settled) return;
+                const u = res.url();
+                if (!/leftTicket\/query/i.test(u)) return;
+                try {
+                    const text = await res.text();
+                    if (text) {
+                        body = text;
+                        settled = true;
+                        resolve();
+                    }
+                } catch  {}
+            });
+            setTimeout(()=>resolve(), q.timeoutMs ?? 25_000);
+        });
+        await t.page.goto(entry.url, {
+            waitUntil: 'domcontentloaded',
+            timeout: 30_000
+        });
+        await heard;
+        const title = await t.page.title().catch(()=>'');
+        const headHtml = (await t.page.content().catch(()=>'')).slice(0, 5000);
+        if (CHALLENGE_RE.test(title + headHtml)) {
+            return err('challenged', `风控/验证码命中(title=${title.slice(0, 60)});按红线不重试不绕过,交还用户`);
+        }
+        const trains = parseLeftTicketQuery(body, entry.url);
+        const verdict = trains.length > 0 ? 'hit' : 'miss';
+        return {
+            ok: true,
+            via: 'session-train-12306',
+            evidence: `[会话:${site}@${ts}] ${trains.length} trains(公开查询面);guard blocked=${t.guard.blockedCount()}/${t.guard.requestCount()}`,
+            latencyMs: Date.now() - started,
+            verdict,
+            trains
+        };
+    } catch (e) {
+        return err('error', e instanceof Error ? e.message.slice(0, 200) : String(e));
+    } finally{
+        await t.close();
+    }
+}
 export async function sessionFlightSearch(q) {
     const started = Date.now();
     const ts = new Date().toISOString();

package/dist/capabilities/visa-policy.js

@@ -0,0 +1,126 @@
+import { readFileSync } from 'node:fs';
+import { BOOKABLE_FACT_SCHEMA, makeFactId } from '../src/bookable-facts.js';
+const SOURCE_HOST = 'https://cs.mfa.gov.cn';
+const SOURCE_ID = 'cs-mfa';
+const USER_AGENT = 'GoTry/0.0.1 (+https://github.com/Danceiny/gotry; polite policy reader)';
+const CONTINENTS = [
+    'yz_645708',
+    'feizhou_645714',
+    'ouzhou_645711',
+    'bz_645718',
+    'dyz_645717'
+];
+export function countryPageUrl(continent, country) {
+    if (!/^[a-z0-9_]+$/.test(continent) || !/^[a-z0-9_]+$/.test(country)) return null;
+    return `${SOURCE_HOST}/zggmcg/ljmdd/${continent}/${country}/`;
+}
+function stripTags(html) {
+    return html.replace(/<(script|style)[\s\S]*?<\/\1>/gi, ' ').replace(/<[^>]+>/g, ' ').replace(/&nbsp;/g, ' ').replace(/&amp;/g, '&').replace(/\s+/g, ' ');
+}
+export function extractVisaSection(pageText) {
+    const text = stripTags(pageText);
+    const start = text.indexOf('签证入境');
+    if (start < 0) return null;
+    const rest = text.slice(start + '签证入境'.length);
+    const sectionEnds = [
+        '海关检查',
+        '海关申报',
+        '医疗条件',
+        '交通出行',
+        '实用信息',
+        '领事保护',
+        '国家安全',
+        '特别提醒'
+    ].map((kw)=>rest.indexOf(kw)).filter((i)=>i > 0);
+    const numHeaders = Array.from(rest.matchAll(/[一二三四五六七八九十]+、(?!免签|落地签|签证|入境规定)/g)).map((m)=>m.index).filter((i)=>i > 0);
+    const allEnds = [
+        ...sectionEnds,
+        ...numHeaders
+    ].sort((a, b)=>a - b);
+    const end = allEnds.length > 0 ? allEnds[0] : Math.min(rest.length, 4_000);
+    const section = rest.slice(0, end).trim();
+    return section.length > 20 ? section : null;
+}
+export function splitPolicyEntries(section) {
+    const parts = section.split(/(?:[一二三四五六七八九十]+、|[（(]\s*[一二三四五六七八九十]+\s*[）)])\s*/).map((p)=>p.trim());
+    return parts.filter((p)=>p.length > 30);
+}
+export function policyFactsFromMfa(input) {
+    const entries = splitPolicyEntries(input.section);
+    const asOf = input.fetchedAt.slice(0, 10);
+    const queryId = `visa-policy:${SOURCE_ID}:${input.countryPath}:${asOf}`;
+    return entries.map((entry, i)=>({
+            schema: BOOKABLE_FACT_SCHEMA,
+            fact_id: makeFactId([
+                input.countryPath,
+                String(i),
+                entry.slice(0, 64),
+                queryId
+            ]),
+            kind: 'policy',
+            subject: `${input.countryLabel} 签证入境(条款 ${i + 1})`,
+            statement: entry,
+            source: `${SOURCE_ID}@${input.fetchedAt}`,
+            query_id: queryId,
+            fetched_at: input.fetchedAt,
+            as_of: asOf,
+            review_by: defaultReviewBy(input.fetchedAt)
+        }));
+}
+export function defaultReviewBy(fetchedAt) {
+    const d = new Date(fetchedAt);
+    d.setDate(d.getDate() + 30);
+    return d.toISOString().slice(0, 10);
+}
+export function listCountryPaths(continentListHtml, continent) {
+    const hits = continentListHtml.matchAll(new RegExp(`href="(\\./([a-z0-9_]+)/)"[^>]*>([^<]{2,20})<`, 'g'));
+    const out = [];
+    for (const m of hits){
+        if (!m[2] || !m[3]) continue;
+        out.push({
+            path: m[2],
+            label: m[3].trim()
+        });
+    }
+    return out.map((x)=>x.path).filter((p)=>/^[a-z0-9_]+$/.test(p));
+}
+export function continentListUrl(continent) {
+    if (!/^[a-z0-9_]+$/.test(continent)) return null;
+    return `${SOURCE_HOST}/zggmcg/ljmdd/${continent}/`;
+}
+export function readPageFixture(path) {
+    return readFileSync(path, 'utf-8');
+}
+export async function fetchVisaPolicyPage(url, timeoutMs = 20_000) {
+    const controller = new AbortController();
+    const timer = setTimeout(()=>controller.abort(), timeoutMs);
+    try {
+        const res = await fetch(url, {
+            headers: {
+                'user-agent': USER_AGENT,
+                'accept-language': 'zh-CN,zh'
+            },
+            signal: controller.signal,
+            redirect: 'follow'
+        });
+        if (!res.ok) return {
+            ok: false,
+            error: `HTTP ${res.status}`
+        };
+        return {
+            ok: true,
+            html: await res.text()
+        };
+    } catch (e) {
+        const msg = controller.signal.aborted ? `timeout after ${timeoutMs}ms` : String(e.message ?? e);
+        return {
+            ok: false,
+            error: msg
+        };
+    } finally{
+        clearTimeout(timer);
+    }
+}
+
+
+//# sourceURL=ts/capabilities/visa-policy.ts