fix(qajobsboard): drop expired LinkedIn reposts and resolve hiring location

Probe application links for closed listings and feed expires_at; enrich vague Remote/Worldwide rows with real country before blocked-countries filtering.

Co-authored-by: Cursor <cursoragent@cursor.com>

extractors/qajobsboard/manifest.ts

@@ -9,6 +9,11 @@ import type {
   ExtractorRunResult,
 } from "@shared/types/extractors";
 import type { CreateJobInput } from "@shared/types/jobs";
+import {
+  isLinkedInJobUrl,
+  isPostingExpiredByDate,
+  probeApplicationLink,
+} from "./src/application-link.js";
 import {
   extractJobLocationFromText,
   fetchQaJobDetailEnrichment,
@@ -37,6 +42,7 @@ interface QaJobBoardlyJob {
   location?: string;
   location_limits?: string[];
   published_at?: string;
+  expires_at?: string;
   application_link?: string;
   description?: DescriptionBlock;
   company?: { name?: string; logo?: string };
@@ -156,8 +162,6 @@ function mapJob(raw: QaJobBoardlyJob): CreateJobInput | null {
   };
 }
 
-const DETAIL_ENRICH_CONCURRENCY = 4;
-
 function needsDetailEnrichment(location: string | undefined): boolean {
   if (!location?.trim()) return true;
   return isVagueLocationLabel(location);
@@ -167,31 +171,43 @@ async function enrichJobsFromDetailPages(
   jobs: CreateJobInput[],
   shouldCancel?: () => boolean,
 ): Promise<CreateJobInput[]> {
-  const enriched = jobs.map((job) => ({ ...job }));
-  const targets = enriched
-    .map((job, index) => ({ job, index }))
-    .filter(({ job }) => needsDetailEnrichment(job.location));
+  const enriched: CreateJobInput[] = [];
 
-  for (let offset = 0; offset < targets.length; offset += DETAIL_ENRICH_CONCURRENCY) {
+  for (const job of jobs) {
     if (shouldCancel?.()) break;
-    const batch = targets.slice(offset, offset + DETAIL_ENRICH_CONCURRENCY);
-    await Promise.all(
-      batch.map(async ({ job, index }) => {
-        try {
-          const detail = await fetchQaJobDetailEnrichment(job.jobUrl);
-          if (!detail) return;
-          enriched[index] = {
-            ...job,
-            ...(detail.location ? { location: detail.location } : {}),
-            ...(detail.jobDescription
-              ? { jobDescription: detail.jobDescription }
-              : {}),
-          };
-        } catch {
-          // keep feed row when detail page fetch fails
+
+    let current = { ...job };
+
+    const applicationLink = job.applicationLink?.trim();
+    if (applicationLink && isLinkedInJobUrl(applicationLink)) {
+      try {
+        const probe = await probeApplicationLink(applicationLink);
+        if (probe?.expired) continue;
+        if (probe?.location && needsDetailEnrichment(current.location)) {
+          current = { ...current, location: probe.location };
         }
-      }),
-    );
+      } catch {
+        // keep row when LinkedIn probe fails
+      }
+    }
+
+    if (needsDetailEnrichment(current.location)) {
+      try {
+        const detail = await fetchQaJobDetailEnrichment(current.jobUrl);
+        if (detail?.expired) continue;
+        current = {
+          ...current,
+          ...(detail?.location ? { location: detail.location } : {}),
+          ...(detail?.jobDescription
+            ? { jobDescription: detail.jobDescription }
+            : {}),
+        };
+      } catch {
+        // keep feed row when detail page fetch fails
+      }
+    }
+
+    enriched.push(current);
   }
 
   return enriched;
@@ -238,6 +254,7 @@ export const manifest: ExtractorManifest = {
 
       for (const row of rows as QaJobBoardlyJob[]) {
         if (out.length >= cap) break;
+        if (isPostingExpiredByDate(asString(row.expires_at))) continue;
         if (terms.length > 0 && !terms.some((t) => matchesTerm(row, t)))
           continue;
         const mapped = mapJob(row);