@danceiny/gotry 0.0.1-rc.21 → 0.0.1-rc.24

package/ts/src/benchmark-environment-bridge.ts

@@ -260,8 +260,6 @@ interface CollectedText {
 }
 
 interface SubprocessHandle {
-  /** DSH reports -1 when process creation itself failed. */
-  readonly pid: number
   collected: { stdout: CollectedText; stderr: CollectedText }
   done: Promise<{ exitCode: number | null; signal?: string | null }>
   terminate?: () => void
@@ -461,22 +459,44 @@ export function registerBenchmarkEnvironmentBridge(
         }
         env.PYTHONDONTWRITEBYTECODE = '1'
         env.PYTHONNOUSERSITE = '1'
-        handle = subprocess.spawn({
-          argv: [bridge.executable, ...bridge.argv_prefix, 'call', tool, serialized.json!],
-          cwd: bridge.cwd,
-          stdio: {
-            stdin: 'ignore',
-            stdout: { maxBytes: bridge.max_output_bytes },
-            stderr: { maxBytes: bridge.max_output_bytes },
-          },
-          graceMs: Math.min(1_000, bridge.timeout_ms),
-          env,
-          signal: controller.signal,
-        })
-        const status = await handle.done
+        try {
+          handle = subprocess.spawn({
+            argv: [bridge.executable, ...bridge.argv_prefix, 'call', tool, serialized.json!],
+            cwd: bridge.cwd,
+            stdio: {
+              stdin: 'ignore',
+              stdout: { maxBytes: bridge.max_output_bytes },
+              stderr: { maxBytes: bridge.max_output_bytes },
+            },
+            graceMs: Math.min(1_000, bridge.timeout_ms),
+            env,
+            signal: controller.signal,
+          })
+        } catch {
+          if (controller.signal.aborted) return jsonObject({ ok: false, error: 'timed_out' })
+          return jsonObject({ ok: false, error: 'spawn_failed' })
+        }
+        let status: { exitCode: number | null; signal?: string | null }
+        try {
+          status = await handle!.done
+        } catch {
+          if (controller.signal.aborted) return jsonObject({ ok: false, error: 'timed_out' })
+          return jsonObject({ ok: false, error: 'spawn_failed' })
+        }
         if (controller.signal.aborted) return jsonObject({ ok: false, error: 'timed_out' })
-        const stdout = handle.collected.stdout.readFrom(0)
-        const stderr = handle.collected.stderr.readFrom(0)
+        let stdout: { text: string; lossy?: boolean }
+        let stderr: { text: string; lossy?: boolean }
+        try {
+          stdout = handle!.collected.stdout.readFrom(0)
+          stderr = handle!.collected.stderr.readFrom(0)
+        } catch {
+          return jsonObject({
+            ok: false,
+            error: 'runner_failed',
+            exit_code: status.exitCode,
+            signal: status.signal ?? null,
+          })
+        }
         if (stdout.lossy || stderr.lossy) return jsonObject({ ok: false, error: 'output_truncated' })
         if (status.exitCode !== 0) {
           return jsonObject({
@@ -504,9 +524,6 @@ export function registerBenchmarkEnvironmentBridge(
         if (!plainObject(visibleResult) && !Array.isArray(visibleResult)) return jsonObject({ ok: false, error: 'invalid_output' })
         if (!inspectAllowedOutput(visibleResult, descriptor.output_keys)) return jsonObject({ ok: false, error: 'forbidden_output' })
         return jsonObject({ ok: true, result: visibleResult })
-      } catch {
-        if (controller.signal.aborted) return jsonObject({ ok: false, error: 'timed_out' })
-        return jsonObject({ ok: false, error: !handle || handle.pid === -1 ? 'spawn_failed' : 'runner_failed' })
       } finally {
         clearTimeout(timer)
       }

package/ts/src/bookable-facts.ts

@@ -20,6 +20,7 @@
  */
 
 import { createHash } from 'node:crypto'
+import { hasRecognizedAvailableSeat, validateTrainQueryResponseUrl, type SessionTrainOption } from '../capabilities/session/adapters/rail-12306.ts'
 
 export const BOOKABLE_FACT_SCHEMA = 'gotry_bookable_fact.v1' as const
 
@@ -182,6 +183,25 @@ export interface SessionLikeResult {
   options?: SessionLikeOption[]
 }
 
+/** 12306 exact-date fact freshness rule: capture must be non-future and <=15 minutes old. */
+export const TRAIN_FACT_FRESHNESS_RULE = 'captured_at_non_future_and_max_age_15m' as const
+export const TRAIN_FACT_MAX_AGE_MS = 15 * 60 * 1000
+
+export interface SessionTrainFactResult {
+  outcome: {
+    kind: 'recognized-nonempty' | 'recognized-empty' | 'malformed' | 'transport-runtime-error'
+    trains?: SessionTrainOption[]
+  }
+  collection?: {
+    requested: { from: string; to: string; date: string }
+    request: { fromStationTelecode: string; toStationTelecode: string; date: string }
+    response: { url: string; fromStationTelecode: string; toStationTelecode: string; date: string }
+    batchId: string
+    queryId: string
+    fetchedAt: string
+  }
+}
+
 const YMD = /^\d{4}-\d{2}-\d{2}$/
 
 function hmOf(dateTime: string | undefined): string | undefined {
@@ -289,6 +309,67 @@ export function factsFromSession(q: { origin: string; destination: string; date:
   return facts
 }
 
+function trainCollectionIsFresh(
+  q: { from: string; to: string; date: string },
+  r: SessionTrainFactResult,
+  now: Date,
+): r is SessionTrainFactResult & { collection: NonNullable<SessionTrainFactResult['collection']> } {
+  const c = r.collection
+  if (!c || !c.batchId.trim() || !c.queryId.trim() || c.queryId !== `session:12306-train:${c.batchId}`) return false
+  if (c.requested.from !== q.from || c.requested.to !== q.to || c.requested.date !== q.date) return false
+  if (!c.request || !c.response
+    || c.request.date !== c.requested.date
+    || c.response.fromStationTelecode !== c.request.fromStationTelecode
+    || c.response.toStationTelecode !== c.request.toStationTelecode
+    || c.response.date !== c.request.date
+    || !validateTrainQueryResponseUrl(c.response.url, c.request).ok) return false
+  const fetchedMs = Date.parse(c.fetchedAt)
+  const nowMs = now.getTime()
+  if (!Number.isFinite(fetchedMs) || !Number.isFinite(nowMs)) return false
+  return fetchedMs <= nowMs && nowMs - fetchedMs <= TRAIN_FACT_MAX_AGE_MS
+}
+
+/** 12306 typed outcome → exact-date train facts.
+ * The converter never invents collection identity or timestamps: both must come
+ * from the session invocation binding. Only recognized-empty can create a
+ * negative fact. A recognized row needs Y plus a closed-set available seat token.
+ */
+export function factsFromSessionTrain(
+  q: { from: string; to: string; date: string },
+  r: SessionTrainFactResult,
+  now: Date = new Date(),
+): FlightFact[] {
+  if (!trainCollectionIsFresh(q, r, now)) return []
+  const c = r.collection
+  const source = 'session:12306-train'
+  const fetchedAt = c.fetchedAt
+  if (r.outcome.kind === 'recognized-empty') {
+    return [negativeFact(c.queryId, 'train', q.from, q.to, q.date, source, fetchedAt)]
+  }
+  if (r.outcome.kind !== 'recognized-nonempty') return []
+  const facts: FlightFact[] = []
+  for (const o of r.outcome.trains ?? []) {
+    if (o.canWebBuy !== 'Y' || !hasRecognizedAvailableSeat(o)) continue
+    facts.push({
+      schema: BOOKABLE_FACT_SCHEMA,
+      fact_id: makeFactId(['train', q.from, q.to, q.date, o.trainCode, c.queryId]),
+      kind: 'train',
+      route: { origin: q.from, destination: q.to },
+      date: q.date,
+      flight_no: o.trainCode,
+      dep_local: o.depTime,
+      arr_local: o.arrTime,
+      tier: 'live_inventory',
+      bookability: 'bookable_exact_date',
+      source,
+      query_id: c.queryId,
+      fetched_at: fetchedAt,
+      as_of: fetchedAt.slice(0, 10),
+    })
+  }
+  return facts
+}
+
 // ---------------------------------------------------------------------------
 // 注册表(纯函数视图;持久化在 capabilities/fact-log.ts)
 // ---------------------------------------------------------------------------
@@ -304,9 +385,13 @@ export function dedupeFacts<T extends BookableFact>(facts: T[]): T[] {
 }
 
 /** route+date 的最新查询结论(同日多次查询:最新一次为准,避免「昨天 miss 今天 hit」陈旧否定) */
-export function latestFactsForRouteDate(facts: BookableFact[], origin: string, destination: string, date: string): FlightFact[] {
+export function latestFactsForRouteDate(
+  facts: BookableFact[], origin: string, destination: string, date: string,
+  kind?: 'flight' | 'train',
+): FlightFact[] {
   const hits = facts.filter((f): f is FlightFact =>
     (f.kind === 'flight' || f.kind === 'train')
+    && (!kind || f.kind === kind)
     && f.route.origin === origin && f.route.destination === destination && f.date === date)
   const latestQueryTs = new Map<string, string>()
   for (const f of hits) {
@@ -319,7 +404,7 @@ export function latestFactsForRouteDate(facts: BookableFact[], origin: string, d
 }
 
 // ---------------------------------------------------------------------------
-// 判定原语①:航班 claim 可述性(fail closed)
+// 判定原语①:航班 claim 可述性(fail closed,flight-only)
 // ---------------------------------------------------------------------------
 
 export type FlightClaimVerdict =
@@ -335,10 +420,16 @@ export interface FlightClaim {
   date?: string
 }
 
+/**
+ * 航班 claim 裁决(flight-only):注册表只接受 `kind:'flight'` 的同号事实
+ * ——train 事实走 `railClaimVerdict` 独立裁决,不得混入航班 not_in_source
+ * 路径(issue #299)。中文高铁/动车/城际/直达代码(G/D/C/Z + 3–4 位)在
+ * 闸侧 `extractClaims` 已分流到 `trains` claim,本函数无需再做扩展。
+ */
 export function flightClaimVerdict(facts: BookableFact[], claim: FlightClaim): { verdict: FlightClaimVerdict; fact?: FlightFact; reason: string } {
   const no = claim.flight_no.toUpperCase().replace(/\s+/g, '')
   const sameNo = facts.filter((f): f is FlightFact =>
-    (f.kind === 'flight' || f.kind === 'train') && f.flight_no.toUpperCase() === no && f.bookability === 'bookable_exact_date')
+    f.kind === 'flight' && f.flight_no.toUpperCase() === no && f.bookability === 'bookable_exact_date')
   const routeMatched = sameNo.filter(f =>
     (!claim.origin || f.route.origin === claim.origin || f.route.origin_airport === claim.origin)
     && (!claim.destination || f.route.destination === claim.destination || f.route.dest_airport === claim.destination)
@@ -346,7 +437,7 @@ export function flightClaimVerdict(facts: BookableFact[], claim: FlightClaim): {
   if (routeMatched.length > 0) return { verdict: 'traceable', fact: routeMatched[0], reason: `回溯 ${routeMatched[0]!.query_id}` }
   // 无同号可下单事实:该 route+date 是否查过?
   if (claim.origin && claim.destination && claim.date) {
-    const scoped = latestFactsForRouteDate(facts, claim.origin, claim.destination, claim.date)
+    const scoped = latestFactsForRouteDate(facts, claim.origin, claim.destination, claim.date, 'flight')
     if (scoped.length > 0) {
       return {
         verdict: 'not_in_source',
@@ -360,6 +451,63 @@ export function flightClaimVerdict(facts: BookableFact[], claim: FlightClaim): {
   return { verdict: 'route_unqueried', reason: `航班 ${no} 无可回溯的 exact-date 事实(缺 route/date 上下文)` }
 }
 
+// ---------------------------------------------------------------------------
+// 判定原语①b:车次 claim 可述性(rail-only,fail closed;issue #299)
+//   闸侧 `extractClaims` 已把 G/D/C/Z + 3–4 位 token 收集为 train claim;
+//   本裁决只在 train 事实里查同号 bookable_exact_date,缺事实/缺上下文/
+// 路线未查均 fail-closed。支持的事实生产端是 `factsFromFlyai kind:'train'`
+// 与 `factsFromSessionTrain`,两者均经真实工具路径追加到事实日志; session
+// 结果必须先通过 typed parser outcome + seat availability/freshness contract。
+// 不得用历史班期或 static-schedule 凑合格回溯。
+// ---------------------------------------------------------------------------
+
+export type RailClaimVerdict =
+  | 'traceable'          // 注册表内有 bookable_exact_date 的该 train 事实(同号+同 route+同 date)
+  | 'not_in_source'      // 该 route+date 查过 exact-date,结果里没有它——最强违例
+  | 'route_unqueried'    // 该 route+date 从未查过 exact-date——无证据断言
+  | 'rail_claim_unverified' // 缺 route/date 上下文,或缺同号事实且 route/date 无 exact-date 证据
+
+export interface RailClaim {
+  flight_no: string
+  origin?: string
+  destination?: string
+  date?: string
+}
+
+const TRAIN_FACT_SOURCES = new Set(['flyai', 'session:12306-train'])
+
+function isStructuredTrainFact(f: BookableFact): f is FlightFact {
+  return f.kind === 'train'
+    && f.schema === BOOKABLE_FACT_SCHEMA
+    && TRAIN_FACT_SOURCES.has(f.source)
+    && f.query_id.trim().length > 0
+    && f.tier === 'live_inventory'
+}
+
+export function railClaimVerdict(facts: BookableFact[], claim: RailClaim): { verdict: RailClaimVerdict; fact?: FlightFact; reason: string } {
+  const no = claim.flight_no.toUpperCase().replace(/\s+/g, '')
+  if (!claim.origin || !claim.destination || !claim.date) {
+    return { verdict: 'rail_claim_unverified', reason: `车次 ${no} 无可回溯的 exact-date 事实(缺完整 route/date 上下文)` }
+  }
+  // 先按 train kind 选择该 route+date 的最新查询批次,再在该批次匹配车次。
+  // 这样旧 hit 不会越过更新的 miss/unavailable,flight 批次也不能遮蔽 train 批次。
+  const scoped = latestFactsForRouteDate(facts, claim.origin, claim.destination, claim.date, 'train')
+    .filter(isStructuredTrainFact)
+  const sameNo = scoped.filter(f =>
+    f.flight_no.toUpperCase() === no
+      && f.bookability === 'bookable_exact_date')
+  if (sameNo.length > 0) return { verdict: 'traceable', fact: sameNo[0], reason: `回溯 ${sameNo[0]!.query_id}` }
+  if (scoped.length > 0) {
+    return {
+      verdict: 'not_in_source',
+      reason: scoped.some(f => f.bookability === 'unavailable_exact_date')
+        ? `${claim.origin}→${claim.destination} ${claim.date} train exact-date 源返回 0 条(${scoped[0]!.query_id})——不得用历史班期/相邻日期填充`
+        : `${claim.origin}→${claim.destination} ${claim.date} train exact-date 源在架 ${scoped.filter(f => f.bookability === 'bookable_exact_date').map(f => f.flight_no).join('/')}——无 ${no}`,
+    }
+  }
+  return { verdict: 'route_unqueried', reason: `${claim.origin}→${claim.destination} ${claim.date} train 路线从未经 exact-date 源核验` }
+}
+
 // ---------------------------------------------------------------------------
 // 判定原语②:同日衔接硬约束(办事结束 → 门到机场 → 值机截止 → 起飞)
 // ---------------------------------------------------------------------------
@@ -405,7 +553,7 @@ export function checkSameDayDeparture(
   c: SameDayDepartureConstraint,
 ): SameDayDepartureVerdict {
   const earliest = hmToMin(c.errand_end_local) + c.airport_transit_min + c.checkin_deadline_min
-  const scoped = latestFactsForRouteDate(facts, q.origin, q.destination, q.date)
+  const scoped = latestFactsForRouteDate(facts, q.origin, q.destination, q.date, 'flight')
   const nonstopBookable = scoped.filter(f => f.bookability === 'bookable_exact_date' && f.nonstop !== false)
   const feasible: FlightFact[] = []
   const excluded: Array<{ flight_no: string; reason: string }> = []
@@ -526,8 +674,9 @@ export function defaultReviewBy(tripDate: string): string {
 }
 
 /**
- * 航班事实行。bookable_exact_date 才允许完整班次行;其余一律 fail-closed 措辞。
- * 永不输出无条件 ✓;证据链(source/query_id/fetched_at)随行必达。
+ * 航班/车次事实行。bookable_exact_date 才允许完整班次行;其余一律 fail-closed 措辞。
+ * train 复用此 canonical renderer 但只输出车次语义,不输出直飞或价格;永不输出无条件 ✓;
+ * 证据链(source/query_id/fetched_at)随行必达。
  */
 export function renderFlightFact(f: FlightFact): string {
   return renderFlightFactInner(f) + ` <!-- fact:${f.fact_id} -->`
@@ -538,6 +687,13 @@ function renderFlightFactInner(f: FlightFact): string {
   const anchor = ` <!-- fact:${f.fact_id} -->`
   const route = `${f.route.origin}→${f.route.destination}`
   const evidence = `[${f.source}@${f.fetched_at} #${f.query_id}]`
+  if (f.kind === 'train' && f.bookability === 'unavailable_exact_date') {
+    const review = f.review_by ?? defaultReviewBy(f.date)
+    return `- ${route} ${f.date} 车次${f.flight_no ? ` ${f.flight_no}` : ''}:**未确认/当前不可售**——exact-date 源 ${f.source} 于 ${f.fetched_at.slice(0, 10)} 返回 0 条;到 ${review} 复核,不得用历史班期/相邻日期填充 ${evidence}`
+  }
+  if (f.kind === 'train' && f.bookability !== 'bookable_exact_date') {
+    return `- ${route} ${f.date} 车次${f.flight_no ? ` ${f.flight_no}` : ''}:**未核验**(${f.bookability})——仅作参考,不得作为可下单方案呈现 ${evidence}`
+  }
   if (f.bookability === 'unavailable_exact_date') {
     const review = f.review_by ?? defaultReviewBy(f.date)
     return `- ${route} ${f.date}:**未确认/当前不可售**——exact-date 源 ${f.source} 于 ${f.fetched_at.slice(0, 10)} 返回 0 条;到 ${review} 复核,不得用历史班期/相邻日期填充 ${evidence}`
@@ -545,6 +701,10 @@ function renderFlightFactInner(f: FlightFact): string {
   if (f.bookability !== 'bookable_exact_date') {
     return `- ${route} ${f.date} ${f.flight_no || ''}:**未核验**(${f.bookability})——仅作参考,不得作为可下单方案呈现 ${evidence}`
   }
+  if (f.kind === 'train') {
+    const times = f.dep_local && f.arr_local ? ` ${f.dep_local}→${f.arr_local}` : ''
+    return `- ${route} ${f.date} 车次 ${f.flight_no}${times} ${evidence}`
+  }
   const carrier = f.operating_carrier && f.operating_carrier !== f.marketing_carrier
     ? `营销 ${f.marketing_carrier ?? '?'} / 实际承运 ${f.operating_carrier}`
     : (f.marketing_carrier ?? '')
@@ -574,8 +734,24 @@ export function renderConnection(legA: FlightFact, legB: FlightFact): string {
 /**
  * 政策事实行:恒为「截至 as_of 的现行政策」表述 + 复核 gate,永不用无条件 ✓。
  * 远期(出发日晚于 as_of)必须附 review_by(默认 D-30)。
+ *
+ * 行尾内嵌 `<!-- fact:<fact_id> -->` 锚点(issue #273,ADR-19 typed-anchor 单向生成):
+ * 产物经渲染原语生成即自带溯源锚,闸侧锚点确定性回溯;手改/伪造锚点 = fact_anchor_unknown。
+ * 与 renderFlightFact / renderHotelFact 同源:启发式 POLICY_WORD 仅覆盖手写/历史产物。
  */
 export function renderPolicyFact(p: PolicyFact, tripStart?: string): string {
+  return `${policyCanonicalBody(p, tripStart)} <!-- fact:${p.fact_id} -->`
+}
+
+/**
+ * 政策行 canonical body(不含锚点;issue #359,D-26 残余收口):
+ * 闸侧锚点回溯用同一原文,subject/statement/source/fetched_at/query_id 任意一项
+ * 改动 → 行文本不再等于 canonical body → fact_anchor_unknown fail-closed。
+ * 不走宽松 substring/subset 匹配,避免锚点行被改写 subject/statement 后仍 pass。
+ * 复核提醒(review_by 或 tripStart 派生的 defaultReviewBy)是合法的可选段——
+ * `tripStart` 与 renderPolicyFact 同一传入,产物与闸两侧语义一致。
+ */
+export function policyCanonicalBody(p: PolicyFact, tripStart?: string): string {
   const review = p.review_by ?? (tripStart ? defaultReviewBy(tripStart) : undefined)
   const reviewNote = review ? `;远期政策须复核——到 ${review} 再核验一次` : ''
   return `- ${p.subject}:截至 ${p.as_of} 的现行政策——${p.statement}${reviewNote} [${p.source}@${p.fetched_at} #${p.query_id}]`