@mazjq/zhiro 0.1.2 → 0.2.1

package/dist/{chunk-YJNBYSVM.js → chunk-2XJ45GUM.js}

@@ -8,25 +8,25 @@ import {
   nodeArgs,
   noteModelRefused,
   upsertFact
-} from "./chunk-UMCKC7XP.js";
+} from "./chunk-VOUILWMI.js";
 import {
   callingAgent,
   delegateRun,
   isExternal
-} from "./chunk-F6DLONN7.js";
+} from "./chunk-7IV2NG2D.js";
 import {
   DomainError,
   NotFound
-} from "./chunk-LH25QWMP.js";
+} from "./chunk-ZKD2BKN3.js";
 import {
   parseBaseUrl,
   parseCsv,
   readBase
-} from "./chunk-YYNUPNBH.js";
+} from "./chunk-4UWMDKMH.js";
 import {
   now,
   tx
-} from "./chunk-5KCN32HQ.js";
+} from "./chunk-4EPNDXRS.js";
 import {
   __name
 } from "./chunk-2SSYR7VW.js";
@@ -1256,7 +1256,12 @@ function brief(kind, d) {
   }
 }
 __name(brief, "brief");
-var toRow = /* @__PURE__ */ __name((r) => ({ kind: r.kind, data: JSON.parse(r.data_json), source: r.source, fetchedAt: r.fetched_at, ageDays: Math.floor((Date.now() - Date.parse(r.fetched_at)) / 864e5) }), "toRow");
+var EVERY_DAYS = { listing: 180, basic: 90, sectors: 90, finance: 30, dividends: 30, holders: 30, forecast: 7, reports: 7, irm: 7, announcements: 1, news: 1, ipo: 3, sme: 365 };
+var everyDays = /* @__PURE__ */ __name((kind, data) => kind === "listing" && data?.none ? 30 : kind === "ipo" && !data?.szse ? 30 : EVERY_DAYS[kind] ?? 30, "everyDays");
+var toRow = /* @__PURE__ */ __name((r, now3 = Date.now()) => {
+  const data = JSON.parse(r.data_json), age = (now3 - Date.parse(r.fetched_at)) / 864e5, every = everyDays(r.kind, data);
+  return { kind: r.kind, data, source: r.source, fetchedAt: r.fetched_at, ageDays: Math.floor(age), everyDays: every, stale: age > every };
+}, "toRow");
 function getIntelDetail(s, companyId, kind) {
   const r = s.domain.db.prepare("select * from company_intel where company_id = ? and kind = ?").get(companyId, kind);
   if (!r) throw new NotFound(`企业情况 ${companyId} ${kind}`);
@@ -1271,97 +1276,110 @@ function getIntel(s, companyId) {
     return JSON.stringify(b) === JSON.stringify(row2.data) ? row2 : { ...row2, data: b, more: true };
   });
   const sme = smeMatch(s.domain.db, co.name);
-  const listing = rows.find((r) => r.kind === "listing")?.data ?? null;
-  return { companyId, companyName: co.name, listing, rows: rows.filter((r) => r.kind !== "listing"), sme, stale: rows.length === 0 || rows.some((r) => r.ageDays > 30) };
+  const listingRow = rows.find((r) => r.kind === "listing");
+  const staleKinds = rows.length === 0 ? ["listing"] : rows.filter((r) => r.stale).map((r) => r.kind);
+  return { companyId, companyName: co.name, listing: listingRow?.data ?? null, rows: rows.filter((r) => r.kind !== "listing"), sme, stale: staleKinds.length > 0, staleKinds, refreshing: running2.has(companyId) };
 }
 __name(getIntel, "getIntel");
+function readIntel(s, companyId) {
+  const card = getIntel(s, companyId);
+  const kept = card.listing !== null;
+  if (kept && card.stale && !card.refreshing) void refreshIntel(s, companyId).catch(() => {
+  });
+  return { ...card, refreshing: card.refreshing || kept && card.stale };
+}
+__name(readIntel, "readIntel");
 var put = /* @__PURE__ */ __name((s, companyId, kind, data, source) => s.domain.db.prepare("insert or replace into company_intel(company_id, kind, data_json, source, fetched_at) values (?,?,?,?,?)").run(companyId, kind, JSON.stringify(data), source, (/* @__PURE__ */ new Date()).toISOString()), "put");
-async function refreshIntel(s, companyId) {
+var running2 = /* @__PURE__ */ new Map();
+var failed = /* @__PURE__ */ new Map();
+var mayTry = /* @__PURE__ */ __name((key, now3) => (failed.get(key)?.until ?? 0) <= now3, "mayTry");
+var noteFailure = /* @__PURE__ */ __name((key, now3) => {
+  const n = (failed.get(key)?.n ?? 0) + 1;
+  failed.set(key, { n, until: now3 + Math.min(24, 2 ** (n - 1)) * 36e5 });
+}, "noteFailure");
+function refreshIntel(s, companyId, opts = {}) {
+  const on = running2.get(companyId);
+  if (on) return on;
+  const job = refreshKinds(s, companyId, !!opts.force).finally(() => running2.delete(companyId));
+  running2.set(companyId, job);
+  return job;
+}
+__name(refreshIntel, "refreshIntel");
+async function refreshKinds(s, companyId, force) {
   const co = s.domain.db.prepare("select name from company where id = ?").get(companyId);
   if (!co) throw new NotFound(`company ${companyId}`);
+  const now3 = Date.now();
+  const have = new Map(s.domain.db.prepare("select * from company_intel where company_id = ?").all(companyId).map((r) => [r.kind, toRow(r, now3)]));
   const errors = {};
-  let listing = await resolveListing(co.name).catch((e) => {
-    errors.listing = String(e?.message ?? e);
-    return null;
-  });
-  let via = "cninfo:topSearch";
-  if (!listing) {
-    await sleep(300);
-    listing = await resolveNeeq(co.name, normName(co.name)).catch((e) => {
-      errors.listing = String(e?.message ?? e);
-      return null;
-    });
-    via = "eastmoney:codetable";
-  }
-  if (!listing) {
+  const due = /* @__PURE__ */ __name((kind) => force || !have.has(kind) || have.get(kind).stale, "due");
+  const attempt = /* @__PURE__ */ __name(async (kind, fn, source) => {
+    const key = `${companyId}:${kind}`;
+    if (!force && !mayTry(key, now3)) return;
+    try {
+      put(s, companyId, kind, await fn(), source);
+      failed.delete(key);
+    } catch (e) {
+      errors[kind] = String(e?.message ?? e).slice(0, 200);
+      noteFailure(key, now3);
+    }
     await sleep(300);
-    listing = await resolveUsListing(co.name).catch((e) => {
+  }, "attempt");
+  let listing = have.get("listing")?.data?.none ? null : have.get("listing")?.data ?? null;
+  let changed = false;
+  if (due("listing")) {
+    const was = have.get("listing")?.data;
+    let found = await resolveListing(co.name).catch((e) => {
       errors.listing = String(e?.message ?? e);
       return null;
     });
-    via = "eastmoney:search";
-  }
-  put(s, companyId, "listing", listing ?? { none: true }, listing ? via : "cninfo:topSearch+eastmoney:codetable+eastmoney:search");
-  if (listing && listing.market === "美股") {
-    for (const [kind, fn, source] of [["basic", () => fetchUsBasic(listing.code), "eastmoney:USF10"], ["finance", () => fetchUsFinance(listing.code), "eastmoney:USF10"]]) {
-      try {
-        put(s, companyId, kind, await fn(), source);
-      } catch (e) {
-        errors[kind] = String(e?.message ?? e).slice(0, 200);
-      }
+    let via = "cninfo:topSearch";
+    if (!found) {
       await sleep(300);
+      found = await resolveNeeq(co.name, normName(co.name)).catch((e) => {
+        errors.listing = String(e?.message ?? e);
+        return null;
+      });
+      via = "eastmoney:codetable";
     }
-  }
-  if (listing && listing.market === "新三板") {
-    for (const [kind, fn, source] of [["basic", () => fetchNeeqBasic(listing.code), "neeq:detailCompany+年报"], ["finance", () => fetchNeeqFinance(listing.code), "eastmoney:xinsanban"]]) {
-      try {
-        put(s, companyId, kind, await fn(), source);
-      } catch (e) {
-        errors[kind] = String(e?.message ?? e).slice(0, 200);
-      }
+    if (!found) {
       await sleep(300);
+      found = await resolveUsListing(co.name).catch((e) => {
+        errors.listing = String(e?.message ?? e);
+        return null;
+      });
+      via = "eastmoney:search";
     }
-  }
-  if (listing && listing.market === "港股") {
-    try {
-      put(s, companyId, "finance", await fetchHkFinance(listing.code), "eastmoney:HKF10");
-    } catch (e) {
-      errors.finance = String(e?.message ?? e).slice(0, 200);
+    if (found || !errors.listing) put(s, companyId, "listing", found ?? { none: true }, found ? via : "cninfo:topSearch+eastmoney:codetable+eastmoney:search");
+    if (found || !errors.listing) {
+      changed = !was || (was.none ? !!found : !found || was.code !== found.code || was.market !== found.market);
+      listing = found;
     }
-    await sleep(300);
   }
-  if (listing && listing.market === "A股") {
-    const steps = [
-      ["basic", () => fetchBasic(listing.code), "eastmoney:F10"],
-      ["finance", () => fetchFinance(listing.code), "eastmoney:ZYZB+sina:lrb"],
-      ["dividends", () => fetchDividends(listing.code), "eastmoney:datacenter"],
-      ["announcements", () => fetchAnnouncements(listing.code, listing.orgId), "cninfo:hisAnnouncement"],
-      ["holders", () => fetchHolders(listing.code), "eastmoney:F10-shareholders"],
-      ["irm", () => fetchIrm(listing.code), "cninfo:irm"],
-      ["sectors", () => fetchSectors(listing.code), "eastmoney:slist"],
-      ["reports", () => fetchReports(listing.code), "eastmoney:reportapi"],
-      ["forecast", () => fetchForecast(listing.code), "eastmoney:datacenter"]
-    ];
-    for (const [kind, fn, source] of steps) {
-      try {
-        put(s, companyId, kind, await fn(), source);
-      } catch (e) {
-        errors[kind] = String(e?.message ?? e).slice(0, 200);
-      }
-      await sleep(300);
-    }
-  }
-  for (const [kind, fn, source] of [["news", () => fetchNews(co.name), "wallstreetcn:search"], ["ipo", () => fetchIpo(co.name), "szse:projectrends"]]) {
-    try {
-      put(s, companyId, kind, await fn(), source);
-    } catch (e) {
-      errors[kind] = String(e?.message ?? e).slice(0, 200);
-    }
-    await sleep(300);
+  const go = /* @__PURE__ */ __name((kind) => changed || due(kind), "go");
+  const steps = [];
+  if (listing?.market === "美股") steps.push(["basic", () => fetchUsBasic(listing.code), "eastmoney:USF10"], ["finance", () => fetchUsFinance(listing.code), "eastmoney:USF10"]);
+  if (listing?.market === "新三板") steps.push(["basic", () => fetchNeeqBasic(listing.code), "neeq:detailCompany+年报"], ["finance", () => fetchNeeqFinance(listing.code), "eastmoney:xinsanban"]);
+  if (listing?.market === "港股") steps.push(["finance", () => fetchHkFinance(listing.code), "eastmoney:HKF10"]);
+  if (listing?.market === "A股") {
+    const l = listing;
+    steps.push(
+      ["basic", () => fetchBasic(l.code), "eastmoney:F10"],
+      ["finance", () => fetchFinance(l.code), "eastmoney:ZYZB+sina:lrb"],
+      ["dividends", () => fetchDividends(l.code), "eastmoney:datacenter"],
+      ["announcements", () => fetchAnnouncements(l.code, l.orgId), "cninfo:hisAnnouncement"],
+      ["holders", () => fetchHolders(l.code), "eastmoney:F10-shareholders"],
+      ["irm", () => fetchIrm(l.code), "cninfo:irm"],
+      ["sectors", () => fetchSectors(l.code), "eastmoney:slist"],
+      ["reports", () => fetchReports(l.code), "eastmoney:reportapi"],
+      ["forecast", () => fetchForecast(l.code), "eastmoney:datacenter"]
+    );
   }
+  steps.push(["news", () => fetchNews(co.name), "wallstreetcn:search"]);
+  if (listing?.market !== "A股") steps.push(["ipo", () => fetchIpo(co.name), "szse:projectrends"]);
+  for (const [kind, fn, source] of steps) if (go(kind)) await attempt(kind, fn, source);
   return { ...getIntel(s, companyId), errors };
 }
-__name(refreshIntel, "refreshIntel");
+__name(refreshKinds, "refreshKinds");
 
 // apps/daemon/src/service/industry.ts
 var UA3 = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/128.0 Safari/537.36";
@@ -2300,6 +2318,7 @@ var tools = {
       degree: Type.Optional(Type.String({ description: "学历要求含任一词" })),
       city: Type.Optional(Type.String()),
       watchedOnly: Type.Optional(Type.Boolean()),
+      internship: Type.Optional(Type.Boolean({ description: "只列实习岗：岗位名含「实习」/ intern，或来源把这批标成实习" })),
       includeExpired: Type.Optional(Type.Boolean({ description: "默认只列截止日未过的岗位；true 连已截止的也列" })),
       limit: Type.Optional(Type.Number({ default: 100 })),
       offset: Type.Optional(Type.Number({ default: 0, description: "跳过前几条（翻页）；简版结果的 hasMore 说还有没有" })),
@@ -2596,7 +2615,7 @@ var tools = {
   "research.company": defineTool({
     name: "research.company",
     description: "派发一次公司调研：由系统侧 Claude Code worker（有联网搜索）在隔离 run 目录中检索官网/招聘系统/就业网/公众号/牛客，产出带证据等级的中文报告，落 vault/research/。异步，返回 runId；用 research.get 查看。",
-    input: Type.Object({ companyName: Type.String(), question: Type.Optional(Type.String({ description: "具体想知道什么；默认 2027 届校招全貌" })) }),
+    input: Type.Object({ companyName: Type.String(), question: Type.Optional(Type.String({ description: "具体想知道什么；默认 2027 届校招全貌。「企业情况」= 经营情况；「就业评估」= 就业评估报告（D-29：职路先放同行百分位和历年财务，再由调研员写利润为什么变、钱去了哪、激励是否覆盖技术人员、上市进展、和用户的匹配，每条结论带证据、反面说法和限制，不打分）" })) }),
     output: Type.Object({ runId: Id, workflowId: Id }),
     effects: EGRESS,
     execution: "job",
@@ -2671,6 +2690,31 @@ var tools = {
     execution: "sync",
     errors: ["MISSING_FIELD", "BAD_RUN"]
   }),
+  "employer.compare": defineTool({
+    name: "employer.compare",
+    description: "比较几家上市公司作为雇主的位置（D-29，不打总分，不花模型额度）。每家在自己行业的全部 A 股公司里（A 股按申万一级，新三板按证监会行业，港股美股按 peerIndustry 或一起比的 A 股公司最多的那个申万一级；港美股不披露职工现金，待遇信号缺；yardstick 说明怎么量的，要转述给用户），按四项各算百分位：成长（营收三年复合增速）、研发（研发费用占营收）、待遇信号（每名员工对应的职工现金支出，粗指标）、盈利（净利率）；再在三种偏好（成长研发优先 / 均衡 / 稳定回报优先，可另给 weights）下排这几家的名次。under 里每种偏好给：place（按该权重的名次）、usually（权重上下浮动时名次的通常范围）、top3Share（进前三的权重占比，是偏好情景的比例，不是现实概率）；缺指标的不估值，给 byKnown / best / worst。outdoneBy 列出在四项上都不差且至少一项更好的公司（不叫梯队）。movedTogether 只说同期一起发生了什么，不是原因，原因要读年报或公告核实。labels 带着规则。未上市公司在 notFound 里，不和上市公司混排。第一次查一个行业约 20–30 秒，之后 7 天内用同一份数据（data 里有取数时间）。",
+    input: Type.Object({
+      companies: Type.Array(Type.String({ description: "公司名、简称或 6 位股票代码" }), { minItems: 1, maxItems: 15 }),
+      weights: Type.Optional(Type.Object({ growth: Type.Optional(Type.Number()), rd: Type.Optional(Type.Number()), pay: Type.Optional(Type.Number()), profit: Type.Optional(Type.Number()) }, { description: "用户自己的偏好权重（相对大小即可，会归一）" })),
+      spread: Type.Optional(Type.Number({ description: "权重上下浮动多少（0.02–1，默认 0.1）" })),
+      year: Type.Optional(Type.Number({ description: "用哪一年的年报，默认最近一个已全部披露的年份" })),
+      peerIndustry: Type.Optional(Type.String({ description: "港股、美股公司拿哪个申万一级行业的 A 股公司当同行（如 机械设备、电子、汽车）；不给就用一起比的 A 股公司最多的那个行业" }))
+    }),
+    output: Type.Object({
+      year: Type.Number(),
+      peerGroup: Type.String(),
+      data: Type.Unknown(),
+      measures: Type.Array(Type.Unknown()),
+      companies: Type.Array(Type.Unknown()),
+      scenarios: Type.Unknown(),
+      spread: Type.Number(),
+      notFound: Type.Array(Type.Unknown()),
+      method: Type.String()
+    }),
+    effects: READ,
+    execution: "sync",
+    errors: ["MISSING_FIELD"]
+  }),
   "labor.us": defineTool({
     name: "labor.us",
     description: "美国某些职业的就业人数和年薪（均值、中位数）：美国劳工统计局 BLS 的 OEWS 官方年度统计，全国、各行业合计（免费）。按 SOC 职业代码查，一次最多 8 个，如 17-2141 机械工程师、17-2071 电气工程师、17-2072 电子工程师、17-2199 其他工程师、15-1252 软件开发、15-2051 数据科学家、15-1221 计算机与信息研究科学家。返回里的 title 是 BLS 对这个代码的叫法，和你要查的职业不一样就是代码给错了。没有密钥，每天大约只能问 25 次，把要查的职业放进一次调用。",
@@ -3116,9 +3160,9 @@ var tools = {
   }),
   "company.intel": defineTool({
     name: "company.intel",
-    description: "企业情况卡（ADR-029）：是否上市（A股/港股/新三板/美股）、基本盘（行业/省份/员工数/成立与上市日期）、近三年营收净利与研发费用、分红、近一年公告里的关键词（专精特新/政府补助/IPO/扩产/对外投资/减持等）、专精特新名录匹配；每类各带 fetchedAt 与 ageDays。stale=true 表示没有数据或有一类超过 30 天没更新，此时先 company.intelRefresh。非上市公司只有名录匹配，其余要 research.company 联网调研。列表是摘要：more=true 的类别（新闻、互动易问答全文、研报、股东、分红、公告命中、行业概念）用 company.intelDetail 取全部。",
+    description: "企业情况卡（ADR-029）：是否上市（A股/港股/新三板/美股）、基本盘（行业/省份/员工数/成立与上市日期）、近三年营收净利与研发费用、分红、近一年公告里的关键词（专精特新/政府补助/IPO/扩产/对外投资/减持等）、专精特新名录匹配；每类各带 fetchedAt、ageDays 和 everyDays（这一类多久更新一次：新闻公告 1 天、问答研报业绩预告 7 天、在审 IPO 3 天、财报分红股东 30 天、行业员工 90 天、上市信息 180 天）。读的时候立即返回已有的；过期的类（staleKinds）会在后台自动更新（refreshing=true），稍后再读就是新的，一般不用调 company.intelRefresh。非上市公司只有名录匹配，其余要 research.company 联网调研。列表是摘要：more=true 的类别（新闻、互动易问答全文、研报、股东、分红、公告命中、行业概念）用 company.intelDetail 取全部。",
     input: Type.Object({ companyId: Id }),
-    output: Type.Object({ companyId: Id, companyName: Type.String(), listing: Type.Unknown(), rows: Type.Array(Type.Object({ kind: Type.String(), data: Type.Unknown(), source: Type.String(), fetchedAt: Type.String(), ageDays: Type.Number(), more: Type.Optional(Type.Boolean()) })), sme: Type.Unknown(), stale: Type.Boolean() }),
+    output: Type.Object({ companyId: Id, companyName: Type.String(), listing: Type.Unknown(), rows: Type.Array(Type.Object({ kind: Type.String(), data: Type.Unknown(), source: Type.String(), fetchedAt: Type.String(), ageDays: Type.Number(), everyDays: Type.Number(), stale: Type.Boolean(), more: Type.Optional(Type.Boolean()) })), sme: Type.Unknown(), stale: Type.Boolean(), staleKinds: Type.Array(Type.String()), refreshing: Type.Boolean() }),
     effects: READ,
     execution: "sync",
     errors: ["NOT_FOUND"]
@@ -3134,8 +3178,8 @@ var tools = {
   }),
   "company.intelRefresh": defineTool({
     name: "company.intelRefresh",
-    description: "重新抓取一家公司的企业情况（巨潮、东方财富 F10、新浪财报、巨潮公告；新三板走股转年报 PDF + 东财三板 F10；匿名只读，约 2–30 秒）；抓不到的类别保留旧值并在 errors 里说明。不花模型额度。",
-    input: Type.Object({ companyId: Id }),
+    description: "立即重新抓取一家公司的企业情况并等它完成（巨潮、东方财富 F10、新浪财报、巨潮公告；新三板走股转年报 PDF + 东财三板 F10；匿名只读，约 2–30 秒）。默认全部类别重抓；force=false 只抓过期的类。抓不到的类别保留旧值并在 errors 里说明。不花模型额度。",
+    input: Type.Object({ companyId: Id, force: Type.Optional(Type.Boolean({ description: "默认 true：全部重抓；false：只抓过期的类" })) }),
     output: Type.Object({ companyId: Id, companyName: Type.String(), listing: Type.Unknown(), rows: Type.Array(Type.Unknown()), sme: Type.Unknown(), stale: Type.Boolean(), errors: Type.Record(Type.String(), Type.String()) }),
     effects: NETWORK_READ,
     execution: "sync",
@@ -3613,6 +3657,15 @@ var tools = {
     execution: "sync",
     errors: ["NOT_AVAILABLE"]
   }),
+  "agent.model": defineTool({
+    name: "agent.model",
+    description: "Zhiro 里用 Claude Code / Codex 时用哪个模型：model 为空 = 跟随它自己的设置；否则用这里选的（只影响 Zhiro 里的用法，不改它自己的设置）。可选的模型见 config.get 的 chat.models.choices。只能由本人。",
+    input: Type.Object({ backend: Type.Union([Type.Literal("claude"), Type.Literal("codex")]), model: Type.String() }),
+    output: Type.Object({ backend: Type.String(), model: Type.String(), using: Type.String() }),
+    effects: WRITE,
+    execution: "sync",
+    errors: []
+  }),
   "agent.harnessConfirm": defineTool({
     name: "agent.harnessConfirm",
     description: "在你自己的 agent（Claude Code / Codex / WorkBuddy 经 MCP）里用 Zhiro 时，确认和撤销在哪里完成：harness = 在 agent 里（agent 复述、你同意后它来确认）；window = 只在 Zhiro 窗口。只能由本人。",
@@ -3718,12 +3771,16 @@ export {
   loadSources,
   startSmeResearch,
   smeRun,
+  resolveNeeq,
   INTEL_QUESTION,
   PLACE,
+  normName,
+  smeMatch,
   importSme,
   getIntelDetail,
-  getIntel,
+  readIntel,
   refreshIntel,
+  companyFinancials,
   DATA_TOOLS,
   dataFile,
   keepData,

package/dist/{chunk-5KCN32HQ.js → chunk-4EPNDXRS.js}

@@ -399,6 +399,31 @@ var MIGRATIONS = [
   // v36: ADR-050 — every upgrade leaves a line here: when, from / to which schema, by which program, where the original went
   `
   create table upgrade_log (at text not null, from_v integer not null, to_v integer not null, app_version text not null, backup text not null default '');
+  `,
+  // v37: D-27 — BOSS 直聘 visits through the person's agent: one row per one-time permit (issued → used), and the stops
+  `
+  create table boss_visit (
+    id text primary key, session_id text not null, requested_at text not null, not_before text not null, expires_at text not null,
+    consumed_at text, target text not null, kind text not null, outcome text not null default 'issued', detail text not null default ''
+  );
+  create index boss_visit_at on boss_visit(requested_at);
+  create index boss_visit_session on boss_visit(session_id);
+  `,
+  // v38: D-28 — BOSS posts seen on a search page, kept as candidates until their job page is read (then position_id)
+  `
+  create table boss_candidate (
+    job_id text primary key, url text not null, title text not null, salary text not null default '', city text not null default '',
+    district text not null default '', experience text not null default '', degree text not null default '', days text not null default '',
+    months text not null default '', brand text not null default '', scale text not null default '', stage text not null default '',
+    industry text not null default '', skills text not null default '[]', labels text not null default '[]', query text not null default '',
+    city_code text not null default '', first_seen text not null, last_seen text not null, position_id text, detail_at text
+  );
+  create index boss_candidate_seen on boss_candidate(last_seen);
+  `,
+  // v39: D-29 — the A-share list and one industry's statements for a year, kept with the time they were fetched, so a
+  // comparison of employers can say which data it used and be run again on the same data
+  `
+  create table peer_snapshot (key text primary key, fetched_at text not null, data_json text not null);
   `
 ];
 var COUNTS_MAY_CHANGE = {};

package/dist/{chunk-YYNUPNBH.js → chunk-4UWMDKMH.js}

@@ -2,7 +2,7 @@ import {
   json,
   now,
   tx
-} from "./chunk-5KCN32HQ.js";
+} from "./chunk-4EPNDXRS.js";
 import {
   updateConfig
 } from "./chunk-4JJEOOW2.js";
@@ -832,12 +832,12 @@ var CatalogBatch = class {
 };
 
 // apps/daemon/src/data/sheet.ts
-function parseCsv(text2) {
+function parseCsv(text3) {
   const rows = [];
   let row = [];
   let field = "";
   let inQ = false;
-  const s = text2.replace(/^/, "");
+  const s = text3.replace(/^/, "");
   for (let i = 0; i < s.length; i++) {
     const c = s[i];
     if (inQ) {
@@ -866,8 +866,8 @@ function parseCsv(text2) {
   return rows.filter((r) => r.some((f) => f.trim() !== ""));
 }
 __name(parseCsv, "parseCsv");
-function parseDeadline(text2) {
-  const t = (text2 ?? "").trim();
+function parseDeadline(text3) {
+  const t = (text3 ?? "").trim();
   if (!t) return { closesAt: null, kind: "unknown" };
   if (/招满|长期|滚动|持续|尽快|随时/.test(t)) return { closesAt: null, kind: /滚动|持续|尽快|随时/.test(t) ? "rolling" : "until_filled" };
   const m = t.match(/(\d{4})[-/.年](\d{1,2})[-/.月](\d{1,2})/);
@@ -1005,8 +1005,8 @@ function listingsFromTable(table, cfg, publisher) {
   return tableToListings(table, columns, cfg, void 0, publisher);
 }
 __name(listingsFromTable, "listingsFromTable");
-function csvToTable(text2) {
-  const [head = [], ...rows] = parseCsv(text2);
+function csvToTable(text3) {
+  const [head = [], ...rows] = parseCsv(text3);
   const headers = head.map((h) => h.trim());
   const at = new Map(headers.map((h, i) => [h, i]));
   return { headers, rows: rows.map((r) => ({ id: "", get: /* @__PURE__ */ __name((h) => (r[at.get(h) ?? -1] ?? "").trim(), "get") })) };
@@ -1244,7 +1244,7 @@ function zhilianUrl(url) {
 }
 __name(zhilianUrl, "zhilianUrl");
 var ID = /(?:xiaozhaoId|companyXzId|companyOutId|rootCompanyId|orgNumbers?)["']?[:=]\[?["']?(\d{5,7})\b|globalData\.companyId,[\w$]+="?(\d{5,7})\b/g;
-var idsIn = /* @__PURE__ */ __name((text2) => [...new Set([...text2.matchAll(ID)].map((m) => m[1] || m[2]))], "idsIn");
+var idsIn = /* @__PURE__ */ __name((text3) => [...new Set([...text3.matchAll(ID)].map((m) => m[1] || m[2]))], "idsIn");
 async function count(ctx, id) {
   const r = await ctx.fetchJson(API2, { method: "POST", headers: { "content-type": "application/json" }, body: JSON.stringify({ orgNumbers: [id], jobSource: 2, pageIndex: 1, pageSize: 1 }) }).catch(() => null);
   return r?.data?.pageInfo?.totalNum ?? 0;
@@ -1818,9 +1818,9 @@ function literalToJson(s) {
   return o;
 }
 __name(literalToJson, "literalToJson");
-function siteLinks(text2, base) {
+function siteLinks(text3, base) {
   const root = new URL(".", base).href, out = [];
-  for (const m of text2.matchAll(/["']((?:\.\/)?[\w./-]+\.(?:html?|js))(?:\?[^"']*)?["']/g)) {
+  for (const m of text3.matchAll(/["']((?:\.\/)?[\w./-]+\.(?:html?|js))(?:\?[^"']*)?["']/g)) {
     const p = new URL(m[1], base).href;
     if (p.startsWith(root) && !out.includes(p) && !/jquery|swiper|jweixin|\.min\.js/.test(p)) out.push(p);
   }
@@ -2076,9 +2076,140 @@ var csv = {
   }
 };
 
+// apps/daemon/src/data/readers/gaoxiaojob.ts
+var M = "https://m.gaoxiaojob.com";
+var W = "https://www.gaoxiaojob.com";
+var UA3 = { "user-agent": "Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X) AppleWebKit/605.1.15 Mobile/15E148" };
+var DAY = 864e5;
+var LONG = /长期有效|长期招聘|常年|招满为止|招满即止|额满为止/;
+var columnOf = /* @__PURE__ */ __name((url) => /^https?:\/\/(?:www\.|m\.)?gaoxiaojob\.com\/column\/(\d+)\.html/.exec(url.trim())?.[1] ?? null, "columnOf");
+function dayOf(mmdd, today = /* @__PURE__ */ new Date()) {
+  const m = /^(\d{1,2})-(\d{1,2})$/.exec(mmdd.trim());
+  if (!m) return null;
+  const y = today.getUTCFullYear(), mm = m[1].padStart(2, "0"), dd = m[2].padStart(2, "0");
+  const d = `${y}-${mm}-${dd}`;
+  return Date.parse(d) > today.getTime() + DAY ? `${y - 1}-${mm}-${dd}` : d;
+}
+__name(dayOf, "dayOf");
+var text2 = /* @__PURE__ */ __name((html) => html.replace(/<script[\s\S]*?<\/script>|<style[\s\S]*?<\/style>/g, " ").replace(/<br\s*\/?>|<\/p>|<\/div>|<\/li>/g, "\n").replace(/<[^>]+>/g, " ").replace(/&nbsp;/g, " ").replace(/&amp;/g, "&").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/[ \t\r]+/g, " ").replace(/\n\s*/g, "\n"), "text");
+function parseNotice(id, title, html) {
+  const t = text2(html);
+  const flat = t.replace(/\s+/g, " ");
+  const field = /* @__PURE__ */ __name((k) => new RegExp(`${k}[:：]\\s*(.*?)\\s+(?:截止日期|学历要求|所属省份|工作地点|报名方式|栏目分类|需求学科|发布时间)`).exec(flat)?.[1]?.trim() ?? "", "field");
+  const posted = /发布时间[:：]\s*(\d{4}-\d{2}-\d{2})/.exec(flat)?.[1] ?? null;
+  const deadlineText = field("截止日期");
+  const full = /(\d{4})[-/.年](\d{1,2})[-/.月](\d{1,2})/.exec(deadlineText);
+  const closesAt = full ? `${full[1]}-${full[2].padStart(2, "0")}-${full[3].padStart(2, "0")}` : null;
+  const source = /来源链接[:：]\s*(https?:\/\/\S+)/.exec(flat)?.[1] ?? "";
+  const a = t.indexOf("公告详情"), b = t.indexOf("来源于网络", a);
+  const body = (a >= 0 ? t.slice(a + 4, b > a ? b : a + 8e3) : "").trim().slice(0, 8e3);
+  return { id, title, posted, deadlineText, closesAt, place: field("工作地点"), source, body };
+}
+__name(parseNotice, "parseNotice");
+function stillOpen(n, listedOn, today = /* @__PURE__ */ new Date()) {
+  if (n.closesAt) return Date.parse(`${n.closesAt}T23:59:59+08:00`) >= today.getTime();
+  const on = n.posted ?? listedOn;
+  if (!on) return false;
+  const age = (today.getTime() - Date.parse(on)) / DAY;
+  return age <= (LONG.test(n.deadlineText) ? 180 : 90);
+}
+__name(stillOpen, "stillOpen");
+function noticeListings(n, jobs, cfg) {
+  const company = (jobs.find((j) => j.companyName)?.companyName ?? "").replace(/#+$/, "").trim() || unitFrom(n.title);
+  const base = {
+    company,
+    noticeUrl: `${W}/announcement/detail/${n.id}.html`,
+    datePosted: n.posted ?? void 0,
+    validThrough: n.closesAt ?? void 0,
+    deadlineText: n.deadlineText || void 0,
+    cohort: cfg.cohort,
+    batch: n.title.slice(0, 80),
+    // one announcement, one campaign (campaigns group by publisher|company|cohort|batch)
+    publisher: "gaoxiaojob",
+    grade: "聚合",
+    isOriginal: false,
+    confidence: 0.7
+    // a reprint: below the 0.85–0.9 the official readers give
+  };
+  const intern = /* @__PURE__ */ __name((s) => /实习/.test(s) ? "实习" : void 0, "intern");
+  const jd = [n.body, n.source ? `原公告：${n.source}` : ""].filter(Boolean).join("\n\n");
+  if (!jobs.length) return [{ ...base, externalId: `ann:${n.id}`, url: base.noticeUrl, title: n.title, locations: n.place ? [n.place] : [], description: jd, kind: intern(n.title) }];
+  return jobs.map((j) => ({
+    ...base,
+    externalId: `job:${j.jobId}`,
+    url: j.url ? `${W}${j.url}` : base.noticeUrl,
+    title: j.jobName,
+    locations: j.city ? [j.city] : n.place ? [n.place] : [],
+    description: jd,
+    degreeReq: j.education || void 0,
+    majorReq: j.major || void 0,
+    kind: intern(`${j.jobName} ${n.title}`)
+  }));
+}
+__name(noticeListings, "noticeListings");
+var unitFrom = /* @__PURE__ */ __name((title) => /^(.+?)(?:\d{4}|诚聘|招聘|公开|引进|招收)/.exec(title)?.[1]?.trim() || title.slice(0, 30), "unitFrom");
+async function jobsOf(ctx, id) {
+  const out = [];
+  for (let page2 = 1; page2 <= 20; page2++) {
+    const d = await ctx.fetchJson(`${M}/announcement/get-job-list?announcementId=${id}&page=${page2}`, { headers: UA3 });
+    const list = d?.data?.list ?? [];
+    out.push(...list);
+    if (!list.length || out.length >= Number(d?.data?.count ?? 0)) return out;
+    await ctx.sleep(pause());
+  }
+  return out;
+}
+__name(jobsOf, "jobsOf");
+var pause = /* @__PURE__ */ __name(() => 600 + Math.floor(Math.random() * 800), "pause");
+var gaoxiaojob = {
+  id: "gaoxiaojob",
+  name: "高校人才网",
+  input: "www.gaoxiaojob.com/column/<栏目编号>.html（如武汉 48、武汉科研院所 509、武汉企业 511）",
+  mode: "pull",
+  describe: "高校人才网的一个栏目（研究院、实验室、高校团队、事业单位、企业）：读手机版公开列表和每条公告的岗位清单，只收仍有效的；转载来源，记为聚合，零 token",
+  detect: /* @__PURE__ */ __name((url) => !!columnOf(url), "detect"),
+  siteKey: /* @__PURE__ */ __name((url) => `column/${columnOf(url)}`, "siteKey"),
+  async fetch(cfg, ctx) {
+    const col = columnOf(cfg.url ?? "");
+    if (!col) throw new Error("SCHEMA_CHANGED: 不是高校人才网的栏目链接");
+    const maxPages = cfg.maxPages ?? 60;
+    const today = /* @__PURE__ */ new Date();
+    const out = [];
+    let oldPages = 0, page2 = 1;
+    for (; page2 <= maxPages; page2++) {
+      const d = await ctx.fetchJson(`${M}/home/get-announcement-list?columnId=${col}&isEstablishment=0&announcementHeat=0&page=${page2}`, { headers: UA3 });
+      if (d?.code !== 200) throw new Error(`SCHEMA_CHANGED: 列表接口返回 ${d?.code ?? "?"} ${d?.msg ?? ""}`);
+      const items = d?.data ?? [];
+      if (!items.length) break;
+      let fresh = 0;
+      for (const it of items) {
+        const listedOn = dayOf(it.refreshTime ?? "", today);
+        if (listedOn && (today.getTime() - Date.parse(listedOn)) / DAY > 180) continue;
+        fresh++;
+        await ctx.sleep(pause());
+        const n = parseNotice(
+          String(it.id),
+          it.title,
+          await ctx.fetchText(`${W}/announcement/detail/${it.id}.html`)
+          /* the desktop page: its 基本信息 block is what parseNotice reads */
+        );
+        if (!stillOpen(n, listedOn, today)) continue;
+        await ctx.sleep(pause());
+        out.push(...noticeListings(n, await jobsOf(ctx, n.id), cfg));
+      }
+      ctx.log(`gaoxiaojob ${col}: page ${page2} ${items.length} 条，在窗口内 ${fresh}（累计 ${out.length}）`);
+      oldPages = fresh ? 0 : oldPages + 1;
+      if (oldPages >= 2) break;
+      await ctx.sleep(pause());
+    }
+    if (page2 > maxPages) throw new Error(`TIMEOUT: 读到第 ${maxPages} 页还没读完这个栏目，这次不算（读一半会把另一半标成下架）`);
+    return out;
+  }
+};
+
 // apps/daemon/src/data/readers/index.ts
 var capturePush = { id: "capture", name: "任意页面抓取", mode: "push", describe: "任意招聘页/简章：书签一键捕获，或给 URL 由 daemon 抓取（公众号/就业网/官网），模型抽取岗位" };
-var sources = Object.fromEntries([moka, beisen, feishuBase, feishuJobs, zhilian, iguopin, chnenergy, chng, dayee, job51, kuaishou, xiaohongshu, csv, capturePush].map((s) => [s.id, s]));
+var sources = Object.fromEntries([moka, beisen, feishuBase, feishuJobs, zhilian, iguopin, chnenergy, chng, dayee, job51, kuaishou, xiaohongshu, gaoxiaojob, csv, capturePush].map((s) => [s.id, s]));
 function sourceFor(url) {
   return Object.values(sources).find((s) => s.detect?.(url));
 }

package/dist/{chunk-F6DLONN7.js → chunk-7IV2NG2D.js}

@@ -1,10 +1,10 @@
 import {
   DomainError,
   NotFound
-} from "./chunk-LH25QWMP.js";
+} from "./chunk-ZKD2BKN3.js";
 import {
   now
-} from "./chunk-5KCN32HQ.js";
+} from "./chunk-4EPNDXRS.js";
 import {
   __name
 } from "./chunk-2SSYR7VW.js";

package/dist/{chunk-J2D7DYGW.js → chunk-A722HITH.js}

@@ -1,6 +1,6 @@
 import {
   VERSION
-} from "./chunk-NNH4ZG6F.js";
+} from "./chunk-H3EFVYGD.js";
 import {
   COUNTS_MAY_CHANGE,
   MIGRATIONS,
@@ -9,7 +9,7 @@ import {
   migrate,
   openDb,
   schemaOf
-} from "./chunk-5KCN32HQ.js";
+} from "./chunk-4EPNDXRS.js";
 import {
   configFile,
   dbPath,

package/dist/{chunk-LUKCNS5M.js → chunk-GRZ4VMSO.js}

@@ -1,7 +1,7 @@
 import {
   json,
   now
-} from "./chunk-5KCN32HQ.js";
+} from "./chunk-4EPNDXRS.js";
 import {
   __name
 } from "./chunk-2SSYR7VW.js";

package/dist/{chunk-NNH4ZG6F.js → chunk-H3EFVYGD.js}

@@ -1,5 +1,5 @@
 // apps/daemon/src/version.ts
-var VERSION = true ? "0.1.2" : JSON.parse(readFileSync(new URL("../../../package.json", import.meta.url), "utf8")).version;
+var VERSION = true ? "0.2.1" : JSON.parse(readFileSync(new URL("../../../package.json", import.meta.url), "utf8")).version;
 
 export {
   VERSION