chore: update search scripts, extra sources and post drafts

Adds the one-off scripts accumulated across recent sessions, removes temp scripts whose searches now live in config.ts as role-family sweeps, and updates fetch-other.ts plus the LinkedIn post drafts.
2026-08-27 09:03:05 +02:00
parent 3f5360f44c
commit b4957025b1
31 changed files with 2008 additions and 283 deletions
@@ -0,0 +1,83 @@
+// jobs.ch — public search API sweep, Switzerland-wide.
+// Endpoint: /api/v1/public/search?query=…&rows=…&page=…
+import { NON_ENGLISH_RE, JUNIOR_TITLE_RE, JAVA_TITLE_RE, SOFTWARE_TITLE_RE } from "../fetch.ts";
+import { readFileSync } from "fs";
+
+const csv = readFileSync("./results/jobs.csv", "utf8");
+const existing = new Set([...csv.matchAll(/"(https?:\/\/[^"]+)"/g)].map((m) => m[1]));
+
+const QUERIES = [
+  "software engineer", "full stack engineer", "backend engineer", "platform engineer",
+  "devops engineer", "data engineer", "AI engineer", "machine learning engineer",
+  "python developer", "typescript developer", "ruby on rails", "node.js developer",
+  "tech lead", "engineering manager", "staff engineer", "principal engineer",
+  "solution architect", "data scientist",
+];
+
+const MS_RE = /\b(\.net|c#|azure|sharepoint|dynamics|power\s?(bi|platform|apps)|sccm|intune|windows|sap|abap)\b/i;
+// Entry-level roles whose *title* looks neutral — caught only in the description.
+const JUNIOR_TEXT_RE = /\b(0\s*[-–to]+\s*[1-5]\s*years|entry.level|junior|graduate program|no experience required)\b/i;
+const DE_FR_WORDS_RE =
+  /\b(und|oder|wir|Sie|Ihre|Ihrer|Kenntnisse|Erfahrung|Aufgaben|Anforderungen|Stelle|Unternehmen|Mitarbeiter|Bereich|für|mit|nous|vous|votre|pour|avec|des|les|une|dans)\b/g;
+
+interface Doc {
+  job_id: string;
+  title: string;
+  preview?: string;
+  company_name?: string;
+  place?: string;
+  publication_date?: string;
+  language_skills?: Array<{ language: string; level: number }>;
+  _links?: { detail_en?: { href: string } };
+}
+
+const seen = new Map<string, Doc>();
+
+for (const q of QUERIES) {
+  // rows is capped at 20 by the API (422 above that); sort=date puts fresh postings first.
+  for (let page = 1; page <= 6; page++) {
+    const url = `https://www.jobs.ch/api/v1/public/search?query=${encodeURIComponent(q)}&rows=20&page=${page}&sort=date`;
+    const res = await fetch(url, {
+      headers: { "User-Agent": "Mozilla/5.0", Accept: "application/json" },
+    });
+    if (!res.ok) break;
+    const data = (await res.json()) as { documents: Doc[]; num_pages: number };
+    for (const d of data.documents ?? []) if (!seen.has(d.job_id)) seen.set(d.job_id, d);
+    if (page >= (data.num_pages ?? 1)) break;
+    await Bun.sleep(250);
+  }
+  console.log(`${q}: ${seen.size} unique so far`);
+}
+
+console.log(`\n${seen.size} raw postings. Filtering…\n`);
+
+const out: Array<Doc & { link: string }> = [];
+let dropped = { title: 0, ms: 0, lang: 0, junior: 0, dup: 0 };
+
+for (const d of seen.values()) {
+  const t = d.title;
+  if (NON_ENGLISH_RE.test(t) || JUNIOR_TITLE_RE.test(t) || JAVA_TITLE_RE.test(t) || !SOFTWARE_TITLE_RE.test(t)) { dropped.title++; continue; }
+  if (MS_RE.test(t)) { dropped.ms++; continue; }
+
+  // German/French required at a working level → drop
+  const needsDeFr = (d.language_skills ?? []).some(
+    (l) => (l.language === "de" || l.language === "fr") && l.level >= 3
+  );
+  const preview = (d.preview ?? "").replace(/<[^>]+>/g, " ");
+  const deWords = (preview.match(DE_FR_WORDS_RE) ?? []).length;
+  if (needsDeFr || deWords >= 4) { dropped.lang++; continue; }
+  if (JUNIOR_TEXT_RE.test(preview)) { dropped.junior++; continue; }
+
+  const link = d._links?.detail_en?.href ?? `https://www.jobs.ch/en/vacancies/detail/${d.job_id}/`;
+  if (existing.has(link)) { dropped.dup++; continue; }
+  out.push({ ...d, link });
+}
+
+out.sort((a, b) => (b.publication_date ?? "").localeCompare(a.publication_date ?? ""));
+
+console.log(`dropped: ${dropped.title} title, ${dropped.ms} MS-stack, ${dropped.lang} DE/FR, ${dropped.junior} entry-level, ${dropped.dup} already in CSV`);
+console.log(`\n=== ${out.length} jobs.ch CANDIDATES ===\n`);
+for (const j of out) {
+  console.log(`${j.title} | ${j.company_name} | ${j.place} | ${(j.publication_date ?? "").slice(0, 10)}`);
+  console.log(`  ${j.link}`);
+}