feat(scanner): Ausschluss-Unternehmen je Suchagent

Neues Feld exclude_companies: steht ein Eintrag im Firmennamen der
Anzeige (z. B. "Hays"), verwirft der Agent sie. Beim Speichern werden
vorhandene Treffer dieser Firmen gelöscht (außer "in Bearbeitung").

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

server/src/db.js

@@ -191,6 +191,12 @@ async function migrateColumns() {
       'alarm_keywords',
       'ALTER TABLE icon_scanner_agents ADD COLUMN alarm_keywords TEXT NULL AFTER excludes',
     ],
+    // Scanner: Ausschluss-Unternehmen je Agent ("Hays"), 20.10.2026.
+    [
+      'icon_scanner_agents',
+      'exclude_companies',
+      'ALTER TABLE icon_scanner_agents ADD COLUMN exclude_companies TEXT NULL AFTER alarm_keywords',
+    ],
     [
       'icon_scanner_matches',
       'alarm',

server/src/routes/scanner.js

@@ -4,7 +4,7 @@ import { requireAdmin, requireAuth } from '../middleware/auth.js';
 import { logHistory } from '../history.js';
 import { SOURCES, SOURCE_IDS, sourceReady } from '../scanner/sources.js';
 import { parseList, runScan, scanStatus } from '../scanner/pipeline.js';
-import { stripQuotes } from '../scanner/matching.js';
+import { isExcludedCompany, stripQuotes } from '../scanner/matching.js';
 import {
   getLastScanAt,
   loadScannerSecrets,
@@ -37,6 +37,7 @@ const agentOut = (row) => ({
   // Nur Quellen, die es noch gibt (Jooble ist am 29.09.2026 entfallen).
   sources: parseList(row.sources).filter((id) => SOURCE_IDS.includes(id)),
   excludes: parseList(row.excludes),
+  exclude_companies: parseList(row.exclude_companies),
   alarm_keywords: parseList(row.alarm_keywords),
   is_active: Boolean(row.is_active),
   matches: Number(row.matches ?? 0),
@@ -121,32 +122,54 @@ function readAgent(body) {
   const keywords = cleanList(body?.keywords);
   const sources = cleanList(body?.sources).filter((id) => SOURCE_IDS.includes(id));
   const excludes = cleanList(body?.excludes);
+  const excludeCompanies = cleanList(body?.exclude_companies);
   const alarmKeywords = cleanList(body?.alarm_keywords);
   if (!name) return { error: 'Bitte einen Namen angeben.' };
   if (keywords.length === 0 && alarmKeywords.length === 0) {
     return { error: 'Mindestens ein Schlagwort oder Alarm-Schlagwort angeben.' };
   }
   if (sources.length === 0) return { error: 'Mindestens eine Börse auswählen.' };
-  return { name, keywords, alarmKeywords, sources, excludes, is_active: body?.is_active !== false };
+  return { name, keywords, alarmKeywords, sources, excludes, excludeCompanies, is_active: body?.is_active !== false };
 }
 
-/** Spaltenwerte in der Reihenfolge name, keywords, alarm_keywords, sources, excludes, is_active. */
+/**
+ * Spaltenwerte in der Reihenfolge name, keywords, alarm_keywords, sources,
+ * excludes, exclude_companies, is_active.
+ */
 const agentValues = (agent) => [
   agent.name,
   JSON.stringify(agent.keywords),
   JSON.stringify(agent.alarmKeywords),
   JSON.stringify(agent.sources),
   JSON.stringify(agent.excludes),
+  JSON.stringify(agent.excludeCompanies),
   agent.is_active ? 1 : 0,
 ];
 
+/**
+ * Bereits gespeicherte Treffer von Ausschluss-Unternehmen entfernen - sonst
+ * blieben sie bis zum Aufraeumen sichtbar. Treffer "in Bearbeitung" bleiben,
+ * an denen arbeitet schon jemand.
+ */
+async function dropExcludedMatches(agentId, agent) {
+  if (agent.excludeCompanies.length === 0) return;
+  const [rows] = await pool.query(
+    `SELECT m.id, j.company FROM icon_scanner_matches m JOIN icon_scanner_jobs j ON j.id = m.job_id
+      WHERE m.agent_id = ? AND m.status <> 'in_bearbeitung'`,
+    [agentId]
+  );
+  const ids = rows.filter((row) => isExcludedCompany(agent, row.company)).map((row) => row.id);
+  if (ids.length > 0) await pool.query('DELETE FROM icon_scanner_matches WHERE id IN (?)', [ids]);
+}
+
 router.post('/agents', async (req, res, next) => {
   const agent = readAgent(req.body);
   if (agent.error) return res.status(400).json({ error: agent.error });
   try {
     const [result] = await pool.query(
-      `INSERT INTO icon_scanner_agents (name, keywords, alarm_keywords, sources, excludes, is_active, created_by)
-       VALUES (?, ?, ?, ?, ?, ?, ?)`,
+      `INSERT INTO icon_scanner_agents
+         (name, keywords, alarm_keywords, sources, excludes, exclude_companies, is_active, created_by)
+       VALUES (?, ?, ?, ?, ?, ?, ?, ?)`,
       [...agentValues(agent), req.user.id]
     );
     await logHistory({
@@ -169,11 +192,13 @@ router.put('/agents/:id', async (req, res, next) => {
   try {
     const [result] = await pool.query(
       `UPDATE icon_scanner_agents
-          SET name = ?, keywords = ?, alarm_keywords = ?, sources = ?, excludes = ?, is_active = ?
+          SET name = ?, keywords = ?, alarm_keywords = ?, sources = ?, excludes = ?, exclude_companies = ?,
+              is_active = ?
         WHERE id = ?`,
       [...agentValues(agent), Number(req.params.id)]
     );
     if (result.affectedRows === 0) return res.status(404).json({ error: 'Suchagent nicht gefunden.' });
+    await dropExcludedMatches(Number(req.params.id), agent);
     const [[row]] = await pool.query('SELECT * FROM icon_scanner_agents WHERE id = ?', [Number(req.params.id)]);
     res.json({ agent: agentOut(row) });
   } catch (err) {

server/src/scanner/matching.js

@@ -54,6 +54,16 @@ export function agentTerms(agent) {
   });
 }
 
+/**
+ * Ausschluss-Unternehmen: geprueft wird nur der Firmenname der Anzeige, nicht
+ * der Text - "Hays" soll Anzeigen von Hays ausblenden, nicht jede, die Hays
+ * irgendwo erwaehnt.
+ */
+export function isExcludedCompany(agent, company) {
+  if (!company) return false;
+  return (agent.excludeCompanies ?? []).some((term) => containsTerm(company, term));
+}
+
 /**
  * @returns {{ score: number, hits: string[], alarm: boolean } | null}
  *          null = ausgeschlossen oder kein einziges Schlagwort gefunden.
@@ -64,6 +74,7 @@ export function scoreJob(agent, job) {
   const all = `${title}\n${job.company ?? ''}\n${job.description ?? ''}`;
 
   if ((agent.excludes ?? []).some((term) => containsTerm(all, term))) return null;
+  if (isExcludedCompany(agent, job.company)) return null;
 
   const keywords = agentTerms(agent);
   if (keywords.length === 0) return null;

server/src/scanner/pipeline.js

@@ -153,6 +153,7 @@ async function loadActiveAgents() {
     alarmKeywords: parseList(row.alarm_keywords),
     sources: parseList(row.sources),
     excludes: parseList(row.excludes),
+    excludeCompanies: parseList(row.exclude_companies),
   }));
 }
 

server/src/schema.sql

@@ -477,6 +477,8 @@ CREATE TABLE IF NOT EXISTS icon_scanner_agents (
   -- Schlagwoerter mit Alarm: werden wie `keywords` gesucht und gezaehlt,
   -- ein Fund meldet sich zusaetzlich mit Ton und Pop-up.
   alarm_keywords TEXT NULL,
+  -- Unternehmen, deren Anzeigen dieser Agent nie zeigt (z. B. ["Hays"]).
+  exclude_companies TEXT NULL,
   is_active   TINYINT(1) NOT NULL DEFAULT 1,
   created_by  INT UNSIGNED NULL,
   created_at  TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,