@ha7ch/job-pro 1.0.93

package/dist/unitree.js

@@ -0,0 +1,418 @@
+// Thin client for 宇树科技 (Unitree Robotics) campus recruiting.
+//
+// ============================================================
+// API DISCOVERY (probed 2026-05)
+//
+// Infrastructure:
+//   https://www.unitree.com/position/ (and /cn/position/) →
+//     Nuxt 3 SPA that inlines all job listings in the server-rendered HTML.
+//     The apiBase revealed in window.__NUXT__.config is:
+//       https://api.unitree.com/website
+//     with routes GET_JOB_LIST: "/job/list" and GET_JOB_DETAIL: "/job/info"
+//     (found in /_nuxt/Cd6-Y0rS.js bundle, 2026-05).
+//
+// Dead ends probed:
+//   career.unitree.com        — resolves to 198.18.x.x (IANA reserved / unreachable)
+//   unitree.app.mokahr.com    — same IANA block; no Moka tenant
+//   https://api.unitree.com/website/job/list (GET or POST, any headers) →
+//     HTTP 567 "请求已被站点的安全策略拦截" from Tencent Cloud EdgeOne WAF.
+//     The WAF blocks all non-browser clients regardless of UA/Referer/Origin spoofing.
+//     The endpoint is real (the SPA uses it from a browser context) but is entirely
+//     inaccessible to server-side HTTP clients.
+//
+// WORKING APPROACH — parse SSR HTML from www.unitree.com/position/:
+//   The Nuxt SPA is configured with ssr:false in its __NUXT_DATA__ state
+//   (serverRendered:false), yet the site's CDN pre-renders the page HTML via
+//   a build-time static pass. The full position list (typically ~20-25 jobs)
+//   is embedded verbatim in the returned HTML, including job IDs, titles, city,
+//   category, department, and hot/urgent flags.
+//
+//   HTML job entry format (stripped from tags):
+//     {Title}({JobCode}) 热招 [急招] {City} | {Category} | {Department} {JD text...}
+//   Some newer listings omit the job code:
+//     {Title} 热招 [急招] {City} | {Category} | {Department} {JD text...}
+//
+//   Job detail deep-links use SPA routing at /position/{JobCode} or /cn/position/{JobCode}.
+//   These return 404 from the CDN (SPA-only routes) but are still the canonical apply URLs.
+//
+// ============================================================
+// PositionSummary field mapping (canonical keys — matches all other adapters):
+//   post_id       — job code (e.g. "J10034") or a slug derived from the title
+//   title         — position title (Chinese)
+//   project       — job category (e.g. "技术类" / "销售类")
+//   recruit_label — "热招" / "热招|急招" / "" depending on status flags
+//   bgs           — department (e.g. "研发部" / "销售服务体系")
+//   work_cities   — work location (e.g. "杭州市")
+//   apply_url     — deep link to the SPA position page
+// ============================================================
+import { extractResumeSignals, scoreOverlap, checkResume } from "./tencent.js";
+export { checkResume };
+const SOURCE = "unitree.com";
+const POSITION_PAGE = "https://www.unitree.com/position/";
+const DETAIL_URL = (jobCode) => `https://www.unitree.com/position/${encodeURIComponent(jobCode)}`;
+const DEFAULT_HEADERS = {
+    "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36",
+    Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
+    "Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
+};
+// ---------- HTML parser ----------
+function stripTags(html) {
+    return html
+        .replace(/<[^>]+>/g, " ")
+        .replace(/&nbsp;/g, " ")
+        .replace(/&amp;/g, "&")
+        .replace(/&lt;/g, "<")
+        .replace(/&gt;/g, ">")
+        .replace(/&quot;/g, '"')
+        .replace(/\s+/g, " ");
+}
+function slugify(title) {
+    // Build a stable stub ID for un-coded listings
+    return title
+        .replace(/[^\w一-鿿]/g, "-")
+        .replace(/-+/g, "-")
+        .replace(/^-|-$/g, "")
+        .slice(0, 40);
+}
+function parsePositions(html) {
+    const text = stripTags(html);
+    const positions = [];
+    // ---- Pass 1: jobs with explicit job-code like (J10034) ----
+    // Context before the (Jxxxxx) anchor is the title; after it are the status / city / category / dept.
+    const idPattern = /\(J(\d+)\)/g;
+    let m;
+    const seen = new Set();
+    while ((m = idPattern.exec(text)) !== null) {
+        const jobCode = `J${m[1]}`;
+        if (seen.has(jobCode))
+            continue;
+        seen.add(jobCode);
+        // Title: scan backwards from the match start for a Chinese/ASCII job title.
+        // Titles immediately precede the job code in the stripped text.
+        // Note: some titles include full-width parentheses, e.g. 嵌入式软件工程师（Linux）
+        // so we must not treat （ as a hard word boundary.
+        const beforeSlice = text.slice(Math.max(0, m.index - 140), m.index);
+        // Take the last "word" cluster that looks like a job title.
+        // Allow full-width （）inside the title but stop at half-width ( and common separators.
+        const titleMatch = beforeSlice.match(/([A-Za-z+#（）一-鿿][^\s·|。；：(]{1,50}(?:\/[^\s|·。；：(]{2,20})?)\s*$/);
+        const rawTitle = titleMatch ? titleMatch[1].trim() : "";
+        // Strip any description text that bled in (heuristic: keep last segment after 。 or ；)
+        const title = rawTitle.split(/[。；]\s*/).pop()?.trim() ?? rawTitle;
+        // Status / city / category / dept: scan forward from end of job code
+        const afterSlice = text.slice(m.index + m[0].length, m.index + m[0].length + 200);
+        const isHot = afterSlice.slice(0, 30).includes("热招");
+        const isUrgent = afterSlice.slice(0, 30).includes("急招");
+        const metaMatch = afterSlice.match(/([一-鿿]{2,6}市)\s*\|\s*([^|]{2,20}?)\s*\|\s*([^\s|·。]{2,30})/);
+        const city = metaMatch ? metaMatch[1].trim() : "杭州市";
+        const category = metaMatch ? metaMatch[2].trim() : "";
+        const dept = metaMatch ? metaMatch[3].trim() : "";
+        const recruitParts = [];
+        if (isHot)
+            recruitParts.push("热招");
+        if (isUrgent)
+            recruitParts.push("急招");
+        const recruit_label = recruitParts.join("|");
+        positions.push({
+            post_id: jobCode,
+            title: title || jobCode,
+            project: category,
+            recruit_label,
+            bgs: dept,
+            work_cities: city,
+            apply_url: DETAIL_URL(jobCode),
+        });
+    }
+    // ---- Pass 2: jobs without a (Jxxxxx) code ----
+    // Pattern: ChineseTitle 热招 [急招] City | Category | Dept
+    const noIdPattern = /([^\s·|：；。]{3,30}(?:工程师|设计师|经理|专员|研究员|架构师|科学家|运营|专家|分析师|顾问))\s+热招(?!\s*\()(\s*急招)?\s+([一-鿿]{2,6}市)\s*\|\s*([^|]{2,20}?)\s*\|\s*([^\s|·。]{2,30})/g;
+    let m2;
+    while ((m2 = noIdPattern.exec(text)) !== null) {
+        const title = m2[1].trim();
+        const slug = slugify(title);
+        if (seen.has(slug))
+            continue;
+        seen.add(slug);
+        const isUrgent = Boolean(m2[2]?.trim());
+        const city = m2[3].trim();
+        const category = m2[4].trim();
+        const dept = m2[5].trim();
+        const recruit_label = isUrgent ? "热招|急招" : "热招";
+        positions.push({
+            post_id: slug,
+            title,
+            project: category,
+            recruit_label,
+            bgs: dept,
+            work_cities: city,
+            apply_url: POSITION_PAGE,
+        });
+    }
+    return positions;
+}
+// ---------- fetch helper ----------
+async function fetchPositionHtml() {
+    let response;
+    try {
+        response = await fetch(POSITION_PAGE, { headers: DEFAULT_HEADERS });
+    }
+    catch (err) {
+        return {
+            ok: false,
+            message: `network error: ${err instanceof Error ? err.message : String(err)}`,
+        };
+    }
+    if (!response.ok) {
+        return { ok: false, message: `HTTP ${response.status}: ${response.statusText}` };
+    }
+    let html;
+    try {
+        html = await response.text();
+    }
+    catch (err) {
+        return {
+            ok: false,
+            message: `body read error: ${err instanceof Error ? err.message : String(err)}`,
+        };
+    }
+    return { ok: true, html, message: "ok" };
+}
+// ---------- In-process cache ----------
+// The position list rarely changes; one fetch per Node process is enough.
+let _posCache = null;
+async function getAllPositions() {
+    const now = Date.now();
+    // Cache valid for 5 minutes
+    if (_posCache && now - _posCache.fetchedAt < 5 * 60 * 1000) {
+        return { ok: true, positions: _posCache.positions, message: "ok (cached)", total: _posCache.positions.length };
+    }
+    const result = await fetchPositionHtml();
+    if (!result.ok || !result.html) {
+        return { ok: false, positions: [], message: result.message, total: 0 };
+    }
+    const positions = parsePositions(result.html);
+    _posCache = { positions, fetchedAt: now };
+    return { ok: true, positions, message: "ok", total: positions.length };
+}
+// ---------- searchPositions ----------
+export async function searchPositions(opts = {}) {
+    const pageSize = Math.max(1, Math.min(100, opts.pageSize ?? 20));
+    const page = Math.max(1, opts.page ?? 1);
+    const keyword = (opts.keyword ?? "").trim().toLowerCase();
+    const pool = await getAllPositions();
+    if (!pool.ok) {
+        return {
+            ok: false,
+            source: SOURCE,
+            message: pool.message,
+            apply_url: POSITION_PAGE,
+            positions: [],
+        };
+    }
+    let filtered = pool.positions;
+    if (keyword) {
+        filtered = filtered.filter((p) => {
+            const blob = [p.title, p.project, p.bgs, p.work_cities, p.post_id]
+                .join(" ")
+                .toLowerCase();
+            return blob.includes(keyword);
+        });
+    }
+    const offset = (page - 1) * pageSize;
+    const paginated = filtered.slice(offset, offset + pageSize);
+    return {
+        ok: true,
+        source: SOURCE,
+        page,
+        page_size: pageSize,
+        total: filtered.length,
+        positions: paginated,
+    };
+}
+// ---------- fetchAllPositions ----------
+export async function fetchAllPositions(opts = {}) {
+    const keyword = (opts.keyword ?? "").trim().toLowerCase();
+    const pool = await getAllPositions();
+    if (!pool.ok) {
+        return {
+            ok: false,
+            source: SOURCE,
+            message: pool.message,
+            apply_url: POSITION_PAGE,
+            fetched: 0,
+            positions: [],
+        };
+    }
+    const positions = keyword
+        ? pool.positions.filter((p) => {
+            const blob = [p.title, p.project, p.bgs, p.work_cities, p.post_id]
+                .join(" ")
+                .toLowerCase();
+            return blob.includes(keyword);
+        })
+        : pool.positions;
+    return {
+        ok: true,
+        source: SOURCE,
+        total: positions.length,
+        fetched: positions.length,
+        positions,
+    };
+}
+// ---------- fetchPositionDetail ----------
+// The SSR HTML contains the JD text for each position but does not expose it
+// in a clean structured field. We extract a best-effort description snippet.
+export async function fetchPositionDetail(postId) {
+    const id = (postId ?? "").trim();
+    if (!id) {
+        return { ok: false, source: SOURCE, message: "post_id is required" };
+    }
+    const result = await fetchPositionHtml();
+    if (!result.ok || !result.html) {
+        return { ok: false, source: SOURCE, post_id: id, message: result.message };
+    }
+    const text = stripTags(result.html);
+    // Find the job code anchor or slug and extract surrounding JD text
+    const anchor = id.startsWith("J") ? `(${id})` : id.replace(/-/g, "");
+    const idx = text.indexOf(anchor);
+    if (idx === -1) {
+        return {
+            ok: false,
+            source: SOURCE,
+            post_id: id,
+            message: `post ${id} not found in current page snapshot`,
+        };
+    }
+    // Extract up to 600 chars of JD text following the city|category|dept line
+    const after = text.slice(idx, idx + 800);
+    const descMatch = after.match(/[一-鿿]{2,5}市\s*\|\s*[^|]+\|\s*[^\s|]+\s+(.{50,600})/);
+    const description = descMatch ? descMatch[1].trim() : "";
+    const pool = await getAllPositions();
+    const pos = pool.positions.find((p) => p.post_id === id);
+    return {
+        ok: true,
+        source: SOURCE,
+        post_id: id,
+        title: pos?.title ?? id,
+        project: pos?.project ?? "",
+        bgs: pos?.bgs ?? "",
+        recruit_label: pos?.recruit_label ?? "",
+        description,
+        work_cities: pos?.work_cities ?? "",
+        apply_url: pos?.apply_url ?? DETAIL_URL(id),
+    };
+}
+// ---------- fetchDictionaries ----------
+// Returns the known static taxonomy (Unitree does not expose a filter catalog).
+export async function fetchDictionaries() {
+    const pool = await getAllPositions();
+    return {
+        ok: pool.ok,
+        source: SOURCE,
+        scrape_url: POSITION_PAGE,
+        note: "Unitree's ATS API (api.unitree.com/website) is protected by Tencent Cloud EdgeOne WAF " +
+            "(HTTP 567) and is inaccessible from server-side clients. " +
+            "Job listings are parsed from the SSR HTML of www.unitree.com/position/ instead.",
+        positions_scraped: pool.total,
+        categories: ["技术类", "销售类"],
+        departments: ["研发部", "销售服务体系"],
+        cities: ["杭州市"],
+        message: pool.message,
+    };
+}
+// ---------- notices (no public endpoint) ----------
+const NOTICES_STUB = {
+    ok: false,
+    source: SOURCE,
+    message: "Unitree: no public notices or announcement endpoint available",
+};
+export async function listNotices() {
+    return NOTICES_STUB;
+}
+export async function getNotice(_id) {
+    return NOTICES_STUB;
+}
+export async function findNoticesByQuestion(_question, _opts = {}) {
+    return NOTICES_STUB;
+}
+// ---------- matchResume ----------
+// Extract technical signals from resume text, filter the scraped position list,
+// and return top N by keyword overlap score.
+export async function matchResume(text, opts = {}) {
+    const topN = Math.max(1, opts.topN ?? 5);
+    const candidates = Math.max(topN, opts.candidates ?? 20);
+    const { terms, cities } = extractResumeSignals(text ?? "");
+    if (!terms.length) {
+        return {
+            ok: false,
+            source: SOURCE,
+            message: "could not extract any technical signals from the text",
+            preview: (text ?? "").slice(0, 120),
+        };
+    }
+    const pool = await getAllPositions();
+    if (!pool.ok) {
+        return { ok: false, source: SOURCE, message: pool.message, positions: [] };
+    }
+    const scored = [];
+    for (const p of pool.positions) {
+        const blob = [p.title, p.project, p.bgs, p.work_cities, p.recruit_label].join(" ");
+        const { score, reasons } = scoreOverlap(blob, terms, cities);
+        if (score > 0)
+            scored.push({ score, position: p, reasons });
+    }
+    scored.sort((a, b) => b.score - a.score);
+    let shortlist = scored.slice(0, Math.max(topN, candidates));
+    if (!shortlist.length) {
+        shortlist = pool.positions.slice(0, candidates).map((position) => ({
+            score: 0,
+            position,
+            reasons: [],
+        }));
+    }
+    const matches = shortlist.slice(0, topN).map((s) => {
+        const mr = s.reasons.length > 0
+            ? s.reasons.slice(0, 5)
+            : ["no specific keyword overlap — surfaced from full position list"];
+        return { ...s.position, match_reasons: mr };
+    });
+    return {
+        ok: true,
+        source: SOURCE,
+        extracted_terms: terms,
+        city_preferences: cities,
+        matches,
+        note: "match_reasons surfaces overlapping keywords, not a probability of getting an interview. " +
+            "The only authority on selection is HR.",
+    };
+}
+export { extractResumeSignals, scoreOverlap };
+import { buildBespokeApplySchema as _buildBespokeApplySchema_unitree } from "./apply.js";
+export async function fetchApplicationSchema(postId) {
+    const id = (postId ?? "").trim();
+    if (!id)
+        return { ok: false, source: "unitree.com", message: "post_id is required" };
+    let title = "";
+    let applyUrl = "https://unitree.com";
+    try {
+        const detail = (await fetchPositionDetail(id));
+        if (detail?.ok === false) {
+            return { ok: false, source: "unitree.com", message: detail.message ?? "post not found" };
+        }
+        title = detail?.title ?? "";
+        if (detail?.apply_url)
+            applyUrl = detail.apply_url;
+    }
+    catch { }
+    return {
+        ok: true,
+        schema: _buildBespokeApplySchema_unitree({
+            source: "unitree.com",
+            postId: id,
+            jobTitle: title,
+            applyUrl,
+            submitEndpoint: undefined,
+            submitKind: "external",
+            submitNotes: "Unitree — recruiting funnel runs through a WeChat mini-program; no public submit API. Open apply_url in browser to scan the WeChat QR.",
+        }),
+    };
+}