feat(13-04): NetServer-Adapter fuer tender24/lhs-vpbw/vergabe.landbw (INGEST-02)

- Package-Legitimacy-Checkpoint bestaetigt: cheerio (sauberer Gate-Pass, 27M Downloads/Woche) statt node-html-parser installiert - EIN config-getriebener NetServerAdapter bedient alle 3 AI-AG-Portale via PublicationSearchControllerServlet-Trefferliste, parst <table> mit cheerio zu RawTenderRecord[], sourcePortal je Zeile korrekt - sourceNoticeId = data-oid (stabil, per Zeile eindeutig); sourceUrl = Such-URL als dokumentierter Best-Effort-Fallback (kein Deep-Link ohne JS-Ausfuehrung ermittelbar, Open Question 2) - Fehlertoleranz: try/catch pro Zeile + pro Portal, []-Fallback bei Totalausfall, AbortController 15s-Timeout, hardcodierte Portal-URLs (SSRF-Guard T-13-04-01) - Live-gecapturte Fixture (tender24.de, 2026-07-23) mit Umlaut-/&- Buyer, gefuellter und leerer Deadline; 14 Spec-Tests gruen Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-23 09:13:01 +02:00
parent 2c28605397
commit 043ada9438
5 changed files with 651 additions and 21 deletions
@@ -26,6 +26,7 @@
    "@tessera/shared": "workspace:*",
    "adm-zip": "^0.6.0",
    "argon2": "^0.44.0",
+    "cheerio": "^1.2.0",
    "class-transformer": "^0.5.1",
    "class-validator": "^0.15.1",
    "cookie-parser": "^1.4.7",
@@ -0,0 +1,140 @@
+<!DOCTYPE html>
+<html lang="de">
+<head><meta charset="utf-8"><title>Vergabe@Net - Suchergebnisse</title></head>
+<body>
+<!-- Live-gecapturte NetServer-Trefferliste (tender24.de, 2026-07-23), gekuerzt auf 6 Zeilen fuer die Test-Fixture. -->
+<table class="table table-responsive table-striped table-hover tableHorizontalHeader">
+<caption>Suchergebnisse</caption>
+<thead>
+  <tr>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=Publishing&amp;Order=desc" title="Nach dem Erscheinungsdatum sortieren" class=" tableActive">Erschienen am</a>
+    </th>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a class="activeChange" href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=TenderName&amp;Order=desc" title="Nach dem Namen der Veröffentlichung sortieren">Ausschreibung</a>
+    </th>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a class="activeChange" href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=AuthorityName&amp;Order=desc" title="Nach dem Namen der ausschreibenden Stelle sortieren">Vergabestelle</a>
+    </th>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a class="activeChange" href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=TenderKind&amp;Order=desc" title="Nach der Verfahrensart sortieren">Verfahrensart</a>
+    </th>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a class="activeChange" href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=TenderLawCaption&amp;Order=desc" title="Nach dem Rechtsrahmen sortieren">Rechtsrahmen</a>
+    </th>
+    
+    <th class="tabelleFirstLine background-blue">
+      <a class="activeChange" id="tableLastHeader" href="PublicationSearchControllerServlet?function=SearchPublications&amp;OrderBy=EndTime&amp;Order=desc" title="Nach der Abgabefrist sortieren">Abgabefrist</a>
+    </th>
+  </tr>
+</thead>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19f800ce3b1-36b275763103203a" data-category="InvitationToTender">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    Wärmenetz Wunsiedel Planungs- und Koordinierungsleistungen (STKP-BY-2026-11)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Sterr-Kölln &amp; Partner</td>
+  
+  <td class="tenderType">Verhandlungsverfahren mit Teilnahmewettbewerb</td>
+  
+  <td class="tenderType">VOL</td>
+  
+  <td class="tenderDeadline">20.08.2026 24:00</td>
+</tr>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19e21c643b8-4268f33140e63c36" data-category="ContractAward">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    POLY_BAU_2026_0003_000023 (Sanierung BT4) Stahlbauarbeiten (POLY_BAU_2026_0003)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Max-Planck-Institut für Polymerforschung (IPF)</td>
+  
+  <td class="tenderType">Freihändige Vergabe</td>
+  
+  <td class="tenderType">VOB</td>
+  
+  <td class="tenderDeadline"></td>
+</tr>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19efec148b6-1a2bbf5ec4dade45" data-category="ContractAward">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    Zellanalyse- und Zellsortiersystem mit spektraler und bildbasierter Echtzeitanalyse (PFOR_2026_02)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Max-Planck-Institut für Herz- und Lungenforschung (IHL)</td>
+  
+  <td class="tenderType">Verhandlungsverfahren ohne Teilnahmewettbewerb</td>
+  
+  <td class="tenderType">UVgO/VgV</td>
+  
+  <td class="tenderDeadline"></td>
+</tr>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19db46612b3-1a04aea066286c1" data-category="ContractAward">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    VE117_1 Malerarbeiten (VE117_1)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Drees &amp; Sommer SE Smart Infrastructure</td>
+  
+  <td class="tenderType">Offenes Verfahren</td>
+  
+  <td class="tenderType">VOB</td>
+  
+  <td class="tenderDeadline"></td>
+</tr>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19eacb06976-633a2f5db089eb5c" data-category="ContractAward">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    ZV - Lautertal - Geschlossene Kanalsanierung (Schmutzwasserhauptkanal) Tiefenlauter (1200-0452-2026/000656)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Stadt Coburg - Beschaffungsamt</td>
+  
+  <td class="tenderType">Freihändige Vergabe</td>
+  
+  <td class="tenderType">VOB</td>
+  
+  <td class="tenderDeadline"></td>
+</tr>
+<tr class="tableRow clickable-row publicationDetail" tabindex="0" style="cursor: pointer;" data-oid="54321-NetTender-19cb85baaa1-34fafe93210c7594" data-category="ContractAward">
+  
+  <td >22.07.2026</td>
+  
+  <td class="tender">
+    Planungsleistungen Wärmenetz Obertürkheim (SWS-2026-0001)
+    
+  </td>
+  
+  <td class="tenderAuthority" >Stadtwerke Stuttgart GmbH</td>
+  
+  <td class="tenderType">Verhandlungsverfahren mit Teilnahmewettbewerb</td>
+  
+  <td class="tenderType">VOL</td>
+  
+  <td class="tenderDeadline"></td>
+</tr>
+</table>
+</body>
+</html>
@@ -0,0 +1,255 @@
+import { readFileSync } from 'fs';
+import { join } from 'path';
+import { afterEach, describe, expect, it, vi } from 'vitest';
+import { NetServerAdapter } from './netserver.adapter';
+
+/**
+ * Real, live-captured NetServer results table (6 rows, trimmed from a
+ * 161 KB live GET against
+ * `https://www.tender24.de/NetServer/PublicationSearchControllerServlet?function=SearchPublications`,
+ * 2026-07-23) — covers an umlaut buyer name ("Sterr-Kölln & Partner",
+ * exercising `&amp;`-entity decoding), a populated deadline row, and
+ * several empty-deadline rows (already-awarded / no open deadline).
+ */
+const FIXTURE_PATH = join(
+  __dirname,
+  '..',
+  '__fixtures__',
+  'netserver-search.html',
+);
+const FIXTURE_HTML = readFileSync(FIXTURE_PATH, 'utf8');
+const FIXTURE_ROW_COUNT = 6;
+
+function stubFetchWithHtml(html: string): void {
+  vi.stubGlobal(
+    'fetch',
+    vi.fn(async () => {
+      return {
+        ok: true,
+        status: 200,
+        text: async () => html,
+      } as Response;
+    }),
+  );
+}
+
+function stubFetchThrowing(): void {
+  vi.stubGlobal(
+    'fetch',
+    vi.fn(async () => {
+      throw new Error('network unreachable');
+    }),
+  );
+}
+
+describe('NetServerAdapter', () => {
+  afterEach(() => {
+    vi.unstubAllGlobals();
+  });
+
+  it('declares sourceType ai-netserver and the 3 AI-AG NetServer portals', () => {
+    const adapter = new NetServerAdapter();
+    expect(adapter.sourceType).toBe('ai-netserver');
+    expect(adapter.portals).toEqual(['tender24', 'lhs-vpbw', 'vergabe.landbw']);
+  });
+
+  describe('parseSearchResults (pure, fixture-driven)', () => {
+    it('parses all rows of the fixture into RawTenderRecord[], sourcePortal set per row', () => {
+      const adapter = new NetServerAdapter();
+      const fetchedAt = new Date();
+
+      const records = adapter.parseSearchResults(
+        FIXTURE_HTML,
+        'tender24',
+        fetchedAt,
+      );
+
+      expect(records.length).toBeGreaterThanOrEqual(1);
+      expect(records).toHaveLength(FIXTURE_ROW_COUNT);
+      for (const record of records) {
+        expect(record.sourceType).toBe('ai-netserver');
+        expect(record.sourcePortal).toBe('tender24');
+        expect(record.sourceNoticeId).toMatch(/^54321-NetTender-/);
+        expect(record.sourceUrl).toBe(
+          'https://www.tender24.de/NetServer/PublicationSearchControllerServlet?function=SearchPublications',
+        );
+        expect(record.fetchedAt).toBe(fetchedAt);
+      }
+    });
+
+    it('decodes an umlaut + &amp;-entity buyer name correctly (Sterr-Kölln & Partner)', () => {
+      const adapter = new NetServerAdapter();
+      const records = adapter.parseSearchResults(
+        FIXTURE_HTML,
+        'tender24',
+        new Date(),
+      );
+
+      const row = records.find(
+        (r) => r.sourceNoticeId === '54321-NetTender-19f800ce3b1-36b275763103203a',
+      );
+      expect(row).toBeDefined();
+      const payload = row!.ocdsPayload as { buyerName: string | null };
+      expect(payload.buyerName).toBe('Sterr-Kölln & Partner');
+    });
+
+    it('uses the row data-oid attribute as sourceNoticeId', () => {
+      const adapter = new NetServerAdapter();
+      const records = adapter.parseSearchResults(
+        FIXTURE_HTML,
+        'lhs-vpbw',
+        new Date(),
+      );
+
+      const ids = records.map((r) => r.sourceNoticeId);
+      expect(ids).toContain('54321-NetTender-19f800ce3b1-36b275763103203a');
+      expect(ids).toContain('54321-NetTender-19e21c643b8-4268f33140e63c36');
+      // every id is unique — no accidental row-collapsing
+      expect(new Set(ids).size).toBe(ids.length);
+    });
+
+    it('maps a populated deadline row to a Date and an empty-deadline row to null (never throws)', () => {
+      const adapter = new NetServerAdapter();
+      const records = adapter.parseSearchResults(
+        FIXTURE_HTML,
+        'tender24',
+        new Date(),
+      );
+
+      const withDeadline = records.find(
+        (r) => r.sourceNoticeId === '54321-NetTender-19f800ce3b1-36b275763103203a',
+      );
+      const withoutDeadline = records.find(
+        (r) => r.sourceNoticeId === '54321-NetTender-19e21c643b8-4268f33140e63c36',
+      );
+      expect(withDeadline).toBeDefined();
+      expect(withoutDeadline).toBeDefined();
+
+      const withDeadlinePayload = withDeadline!.ocdsPayload as {
+        deadlineAt: string | null;
+      };
+      const withoutDeadlinePayload = withoutDeadline!.ocdsPayload as {
+        deadlineAt: string | null;
+      };
+      // "20.08.2026 24:00" -> normalized to 2026-08-21T00:00Z (NetServer's 24:00 = next-day midnight)
+      expect(withDeadlinePayload.deadlineAt).toBe('2026-08-21T00:00:00.000Z');
+      expect(withoutDeadlinePayload.deadlineAt).toBeNull();
+    });
+
+    it('returns [] for empty HTML instead of throwing', () => {
+      const adapter = new NetServerAdapter();
+      expect(adapter.parseSearchResults('', 'tender24', new Date())).toEqual(
+        [],
+      );
+    });
+
+    it('returns [] for HTML with no matching result rows instead of throwing', () => {
+      const adapter = new NetServerAdapter();
+      const broken = '<html><body><p>Keine Treffer</p></body></html>';
+      expect(
+        adapter.parseSearchResults(broken, 'tender24', new Date()),
+      ).toEqual([]);
+    });
+
+    it('skips a row missing data-oid without throwing or affecting other rows', () => {
+      const adapter = new NetServerAdapter();
+      const html = `
+        <table>
+          <tr class="tableRow publicationDetail">
+            <td>22.07.2026</td>
+            <td class="tender">Ohne OID</td>
+            <td class="tenderAuthority">Test-Vergabestelle</td>
+            <td class="tenderType">Offenes Verfahren</td>
+            <td class="tenderType">VOB</td>
+            <td class="tenderDeadline"></td>
+          </tr>
+          <tr class="tableRow publicationDetail" data-oid="test-oid-1">
+            <td>22.07.2026</td>
+            <td class="tender">Mit OID</td>
+            <td class="tenderAuthority">Test-Vergabestelle 2</td>
+            <td class="tenderType">Offenes Verfahren</td>
+            <td class="tenderType">VOB</td>
+            <td class="tenderDeadline"></td>
+          </tr>
+        </table>
+      `;
+      const records = adapter.parseSearchResults(html, 'tender24', new Date());
+      expect(records).toHaveLength(1);
+      expect(records[0]?.sourceNoticeId).toBe('test-oid-1');
+    });
+  });
+
+  describe('fetchTenders (portal fan-out, fetch mocked)', () => {
+    it('fetches all 3 portals and aggregates their parsed records', async () => {
+      stubFetchWithHtml(FIXTURE_HTML);
+      const adapter = new NetServerAdapter();
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records.length).toBe(FIXTURE_ROW_COUNT * 3);
+      const portals = new Set(records.map((r) => r.sourcePortal));
+      expect(portals).toEqual(new Set(['tender24', 'lhs-vpbw', 'vergabe.landbw']));
+    });
+
+    it('returns [] for a portal whose fetch throws, without aborting the other portals (D-01)', async () => {
+      const adapter = new NetServerAdapter();
+      let callCount = 0;
+      vi.stubGlobal(
+        'fetch',
+        vi.fn(async () => {
+          callCount += 1;
+          if (callCount === 1) throw new Error('portal 1 unreachable');
+          return { ok: true, status: 200, text: async () => FIXTURE_HTML } as Response;
+        }),
+      );
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      // first portal fails -> contributes 0 records; the other two succeed.
+      expect(records.length).toBe(FIXTURE_ROW_COUNT * 2);
+    });
+
+    it('returns [] for every portal when fetch always throws, without propagating (D-01 total-failure fallback)', async () => {
+      stubFetchThrowing();
+      const adapter = new NetServerAdapter();
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records).toEqual([]);
+    });
+
+    it('returns [] without throwing when a portal responds non-2xx', async () => {
+      vi.stubGlobal(
+        'fetch',
+        vi.fn(async () => {
+          return { ok: false, status: 503 } as Response;
+        }),
+      );
+      const adapter = new NetServerAdapter();
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records).toEqual([]);
+    });
+  });
+
+  it('never imports or uses axios (native fetch is the sole HTTP client convention)', () => {
+    const source = readFileSync(
+      join(__dirname, 'netserver.adapter.ts'),
+      'utf8',
+    );
+    expect(source).not.toMatch(/from ['"]axios['"]/);
+  });
+
+  it('never interpolates a portal base URL from a variable/parameter (SSRF guard, T-13-04-01)', () => {
+    const source = readFileSync(
+      join(__dirname, 'netserver.adapter.ts'),
+      'utf8',
+    );
+    // NETSERVER_PORTALS values must be string literals, not template
+    // expressions referencing an external input.
+    expect(source).toMatch(/baseUrl: 'https:\/\/www\.tender24\.de'/);
+    expect(source).toMatch(/baseUrl: 'https:\/\/lhs-vpbw\.vmstart\.de'/);
+    expect(source).toMatch(/baseUrl: 'https:\/\/vergabe\.landbw\.de'/);
+  });
+});
@@ -0,0 +1,229 @@
+import { Injectable, Logger } from '@nestjs/common';
+import * as cheerio from 'cheerio';
+import type { RawTenderRecord, SourceType } from '../tender.types';
+import type { TenderSourceAdapter } from './tender-source-adapter.interface';
+
+/**
+ * NetServerAdapter — ONE config-driven adapter (INGEST-02) serving the
+ * three AI-AG "Vergabe@Net" portals (tender24, lhs-vpbw, vergabe.landbw).
+ * Each portal exposes the same public, auth-free HTML results table at
+ * `/NetServer/PublicationSearchControllerServlet?function=SearchPublications`
+ * (live verified: tender24 -> HTTP 200, 161 KB `<table>`, 2026-07-23).
+ *
+ * Security notes (threat model T-13-04-01/02/03):
+ * - Portal base URLs are hardcoded constants (NETSERVER_PORTALS), NEVER
+ *   interpolated from user/admin input (SSRF guard, T-10-06 pattern).
+ * - Native fetch + AbortController 15s timeout (DoeOpenDataAdapter idiom)
+ *   — no axios anywhere in this codebase.
+ * - Per-row try/catch (Pitfall 2): a single malformed row is skipped
+ *   (logger.warn), never aborts the whole portal. A totally unparsable
+ *   page, or a portal whose fetch throws, resolves to [] for that portal
+ *   instead of throwing — pollDueSources' catch-per-source (Plan 13-03) is
+ *   a second, coarser safety net, but the adapter itself must never take
+ *   down the other two portals in the same tick (D-01).
+ * - Extracted title/buyerName are plain text (cheerio `.text()`, not
+ *   `.html()`) — no raw HTML is ever carried into RawTenderRecord/Tender,
+ *   preventing stored-XSS via portal markup (V5).
+ *
+ * Open Question 2 (13-RESEARCH.md) resolved by live inspection: NetServer
+ * result rows carry NO per-row `<a href>` — navigation to the detail page
+ * is client-side JS only (`data-oid` + a click handler in an external
+ * script), with no statically-derivable deep-link URL in the search HTML.
+ * Documented, best-effort choice (D-01): `sourceNoticeId` is the row's
+ * `data-oid` attribute (stable, unique per notice); `sourceUrl` falls back
+ * to the portal's search-results URL itself (NOT a per-notice deep link) —
+ * this is a known limitation, not a bug; see 13-04-SUMMARY.md.
+ */
+const NETSERVER_FETCH_TIMEOUT_MS = 15_000;
+const SEARCH_PATH =
+  '/NetServer/PublicationSearchControllerServlet?function=SearchPublications';
+
+interface NetServerPortalConfig {
+  readonly baseUrl: string;
+}
+
+/**
+ * Hardcoded per-portal base URLs (SSRF guard — never built from input).
+ * Each base URL points directly at the host that resolves `/NetServer`
+ * without relying on a redirect/meta-refresh stub (Pitfall 4: a bare
+ * `vergabe.landbw.de/` root only serves a 641-byte meta-refresh page; the
+ * base URL below targets the resolved host directly).
+ */
+const NETSERVER_PORTALS: Record<string, NetServerPortalConfig> = {
+  'tender24': { baseUrl: 'https://www.tender24.de' },
+  'lhs-vpbw': { baseUrl: 'https://lhs-vpbw.vmstart.de' },
+  'vergabe.landbw': { baseUrl: 'https://vergabe.landbw.de' },
+};
+
+@Injectable()
+export class NetServerAdapter implements TenderSourceAdapter {
+  readonly sourceType: SourceType = 'ai-netserver';
+  /** Config-driven multi-portal adapter (INGEST-02) — one class, 3 portals. */
+  readonly portals = ['tender24', 'lhs-vpbw', 'vergabe.landbw'] as const;
+
+  private readonly logger = new Logger(NetServerAdapter.name);
+
+  /**
+   * dayCursor is accepted for interface conformance but NOT used as a
+   * query filter — the NetServer search servlet has no documented
+   * incremental/date-range parameter (unlike DÖE's day-batch export);
+   * pagination/date-filtering is best-effort out of scope for this plan
+   * (13-RESEARCH.md Open Question 2 — first page is sufficient for the
+   * MVP proof). Every configured portal is fetched independently and a
+   * failure on one never prevents the others from being polled (D-01).
+   */
+  async fetchTenders(_dayCursor: string): Promise<RawTenderRecord[]> {
+    const fetchedAt = new Date();
+    const records: RawTenderRecord[] = [];
+
+    for (const portal of this.portals) {
+      const config = NETSERVER_PORTALS[portal];
+      if (!config) continue; // defensive — portals[] and the config map are kept in sync
+
+      try {
+        const html = await this.fetchPortalHtml(config.baseUrl);
+        records.push(...this.parseSearchResults(html, portal, fetchedAt));
+      } catch (error) {
+        this.logger.warn(
+          `NetServer portal '${portal}' fetch failed, skipping this portal for this tick: ${(error as Error).message}`,
+        );
+      }
+    }
+
+    return records;
+  }
+
+  private async fetchPortalHtml(baseUrl: string): Promise<string> {
+    const controller = new AbortController();
+    const timeout = setTimeout(
+      () => controller.abort(),
+      NETSERVER_FETCH_TIMEOUT_MS,
+    );
+
+    try {
+      const res = await fetch(`${baseUrl}${SEARCH_PATH}`, {
+        signal: controller.signal,
+        redirect: 'follow',
+      });
+      if (!res.ok) {
+        throw new Error(`NetServer fetch failed: HTTP ${res.status}`);
+      }
+      return await res.text();
+    } finally {
+      clearTimeout(timeout);
+    }
+  }
+
+  /**
+   * Parses the `PublicationSearchControllerServlet` results `<table>` into
+   * RawTenderRecord[]. Tolerant by construction (Pitfall 2, D-01): a single
+   * row that throws during extraction is skipped (logger.warn) rather than
+   * aborting the whole portal; if the page itself is fundamentally
+   * unparsable (cheerio.load throws, or simply no matching rows exist),
+   * this returns `[]` instead of propagating.
+   */
+  parseSearchResults(
+    html: string,
+    portal: string,
+    fetchedAt: Date,
+  ): RawTenderRecord[] {
+    const records: RawTenderRecord[] = [];
+
+    let $: cheerio.CheerioAPI;
+    try {
+      $ = cheerio.load(html);
+    } catch (error) {
+      this.logger.warn(
+        `NetServer HTML totally unparsable (${portal}), returning [] (D-01): ${(error as Error).message}`,
+      );
+      return [];
+    }
+
+    $('tr.tableRow.publicationDetail').each((_, el) => {
+      try {
+        const row = $(el);
+        const oid = row.attr('data-oid');
+        if (!oid) return; // no stable id -> unusable row, skip (Pitfall 2)
+
+        const cells = row.find('td');
+        const publishedRaw = $(cells.get(0)).text().trim();
+        const title = $(cells.get(1)).text().trim() || 'Unbenannte Ausschreibung';
+        const buyerName = $(cells.get(2)).text().trim() || null;
+        const procedureType = $(cells.get(3)).text().trim() || null;
+        const legalFramework = $(cells.get(4)).text().trim() || null;
+        const deadlineRaw = $(cells.get(5)).text().trim();
+
+        const publishedAt = parseNetServerDate(publishedRaw);
+        const deadlineAt = parseNetServerDeadline(deadlineRaw);
+
+        records.push({
+          sourceType: this.sourceType,
+          sourcePortal: portal,
+          sourceNoticeId: oid,
+          // Best-effort fallback (Open Question 2 / D-01): the search
+          // results URL, NOT a per-notice deep link — see class doc.
+          sourceUrl: `${NETSERVER_PORTALS[portal]?.baseUrl ?? ''}${SEARCH_PATH}`,
+          fetchedAt,
+          publishedAt,
+          eformsPayload: null,
+          // NetServer has no eForms/OCDS structure; the extracted table
+          // fields are carried through this generic bag so a future
+          // normalizer extension (out of scope for this plan) can map
+          // them without re-parsing HTML.
+          ocdsPayload: {
+            title,
+            buyerName,
+            procedureType,
+            legalFramework,
+            deadlineAt: deadlineAt ? deadlineAt.toISOString() : null,
+          },
+        });
+      } catch (error) {
+        this.logger.warn(
+          `Skipping unparsable NetServer row (${portal}): ${(error as Error).message}`,
+        );
+      }
+    });
+
+    return records;
+  }
+}
+
+/** Parses NetServer's `dd.MM.yyyy` publish-date column; null if absent/malformed. */
+function parseNetServerDate(raw: string): Date | null {
+  const match = raw.match(/^(\d{2})\.(\d{2})\.(\d{4})$/);
+  if (!match) return null;
+  const [, dd, mm, yyyy] = match;
+  const date = new Date(
+    Date.UTC(Number(yyyy), Number(mm) - 1, Number(dd)),
+  );
+  return Number.isNaN(date.getTime()) ? null : date;
+}
+
+/**
+ * Parses NetServer's `dd.MM.yyyy HH:mm` deadline column. Frequently empty
+ * (many rows have no deadline, e.g. already-awarded ContractAward entries)
+ * — empty/malformed input returns null, never throws (RawTenderRecord's
+ * deadline is only exposed via the raw `ocdsPayload` bag in this plan;
+ * nullability here matches the project-wide "deadline is often null"
+ * convention, RESEARCH.md Pattern 4). NetServer uses `24:00` to mean
+ * midnight at the end of the given day — normalized to `00:00` the next day.
+ */
+function parseNetServerDeadline(raw: string): Date | null {
+  if (!raw) return null;
+  const match = raw.match(/^(\d{2})\.(\d{2})\.(\d{4})\s+(\d{2}):(\d{2})$/);
+  if (!match) return null;
+  const [, dd, mm, yyyy, hh, min] = match;
+  const day = Number(dd);
+  const month = Number(mm) - 1;
+  const year = Number(yyyy);
+  const hour = Number(hh);
+  const minute = Number(min);
+
+  const date =
+    hour === 24
+      ? new Date(Date.UTC(year, month, day + 1, 0, minute))
+      : new Date(Date.UTC(year, month, day, hour, minute));
+
+  return Number.isNaN(date.getTime()) ? null : date;
+}
@@ -62,6 +62,9 @@ importers:
      argon2:
        specifier: ^0.44.0
        version: 0.44.0
+      cheerio:
+        specifier: ^1.2.0
+        version: 1.2.0
      class-transformer:
        specifier: ^0.5.1
        version: 0.5.1
@@ -2569,6 +2572,10 @@ packages:
    resolution: {integrity: sha512-quS9HgjQpdaXOvsZz82Oz7uxtXiy6UIsIQcpBj7HRw2M63Skasm9qlDocAM7jNuaxdhpPU7c4kJN+gA5MCu4ww==}
    engines: {node: '>=18.17'}

+  cheerio@1.2.0:
+    resolution: {integrity: sha512-WDrybc/gKFpTYQutKIK6UvfcuxijIZfMfXaYm8NMsPQxSYvf+13fXUJ4rztGGbJcBQ/GF55gvrZ0Bc0bj/mqvg==}
+    engines: {node: '>=20.18.1'}
+
  chokidar@4.0.3:
    resolution: {integrity: sha512-Qgzu8kfBvo+cA4962jnP1KkS6Dop5NS6g7R5LFYJr4b8Ub94PPQXUksCw9PvXoeXPRRddRNC5C1JQUR2SMGtnA==}
    engines: {node: '>= 14.16.0'}
@@ -7905,7 +7912,6 @@ snapshots:
      domelementtype: 2.3.0
      domhandler: 5.0.3
      domutils: 3.2.2
-    optional: true

  cheerio@1.0.0:
    dependencies:
@@ -7922,6 +7928,20 @@ snapshots:
      whatwg-mimetype: 4.0.0
    optional: true

+  cheerio@1.2.0:
+    dependencies:
+      cheerio-select: 2.1.0
+      dom-serializer: 2.0.0
+      domhandler: 5.0.3
+      domutils: 3.2.2
+      encoding-sniffer: 0.2.1
+      htmlparser2: 10.1.0
+      parse5: 7.3.0
+      parse5-htmlparser2-tree-adapter: 7.1.0
+      parse5-parser-stream: 7.1.2
+      undici: 7.28.0
+      whatwg-mimetype: 4.0.0
+
  chokidar@4.0.3:
    dependencies:
      readdirp: 4.1.2
@@ -8108,7 +8128,6 @@ snapshots:
      domhandler: 5.0.3
      domutils: 3.2.2
      nth-check: 2.1.1
-    optional: true

  css-selector-parser@3.3.0: {}

@@ -8123,8 +8142,7 @@ snapshots:
      mdn-data: 2.27.1
      source-map-js: 1.2.1

-  css-what@6.2.2:
-    optional: true
+  css-what@6.2.2: {}

  css.escape@1.5.1: {}

@@ -8283,10 +8301,8 @@ snapshots:
      domelementtype: 2.3.0
      domhandler: 5.0.3
      entities: 4.5.0
-    optional: true

-  domelementtype@2.3.0:
-    optional: true
+  domelementtype@2.3.0: {}

  domhandler@4.3.1:
    dependencies:
@@ -8296,7 +8312,6 @@ snapshots:
  domhandler@5.0.3:
    dependencies:
      domelementtype: 2.3.0
-    optional: true

  domutils@2.8.0:
    dependencies:
@@ -8310,7 +8325,6 @@ snapshots:
      dom-serializer: 2.0.0
      domelementtype: 2.3.0
      domhandler: 5.0.3
-    optional: true

  dotenv-expand@12.0.3:
    dependencies:
@@ -8368,7 +8382,6 @@ snapshots:
    dependencies:
      iconv-lite: 0.6.3
      whatwg-encoding: 3.1.1
-    optional: true

  enhanced-resolve@5.21.6:
    dependencies:
@@ -8386,13 +8399,11 @@ snapshots:
  entities@3.0.1:
    optional: true

-  entities@4.5.0:
-    optional: true
+  entities@4.5.0: {}

  entities@6.0.1: {}

-  entities@7.0.1:
-    optional: true
+  entities@7.0.1: {}

  entities@8.0.0: {}

@@ -8951,7 +8962,6 @@ snapshots:
      domhandler: 5.0.3
      domutils: 3.2.2
      entities: 7.0.1
-    optional: true

  htmlparser2@7.2.0:
    dependencies:
@@ -9004,7 +9014,6 @@ snapshots:
  iconv-lite@0.6.3:
    dependencies:
      safer-buffer: 2.1.2
-    optional: true

  iconv-lite@0.7.2:
    dependencies:
@@ -10531,12 +10540,10 @@ snapshots:
    dependencies:
      domhandler: 5.0.3
      parse5: 7.3.0
-    optional: true

  parse5-parser-stream@7.1.2:
    dependencies:
      parse5: 7.3.0
-    optional: true

  parse5@7.3.0:
    dependencies:
@@ -12077,10 +12084,8 @@ snapshots:
  whatwg-encoding@3.1.1:
    dependencies:
      iconv-lite: 0.6.3
-    optional: true

-  whatwg-mimetype@4.0.0:
-    optional: true
+  whatwg-mimetype@4.0.0: {}

  whatwg-mimetype@5.0.0: {}