feat(14-02): add TenderRssFeedSource CRUD, tick poll gate, and wire RssAdapter

Global admin-managed RSS feed list (TenderRssFeedSource, D-08/D-14) with
a save-time hostname/SSRF guard (TenderRssFeedSourceService) — RSS feed
URLs are runtime admin input, so the code-level SourceRegistry denylist
gate does not cover them; a separate check rejects DENYLISTED_PORTALS
hostnames, non-http(s) schemes, and private/loopback hosts.

Adds TenderSourcePollConfig.pollGranularity ('day' | 'tick', D-15):
pollDueSources() branches per source — 'day' sources keep the existing
lastIngestedDay gate byte-unchanged, 'tick' sources (rss) fetch on every
active scheduler tick regardless of lastIngestedDay, since the day-cursor
gate was built for a genuine daily batch-export API and would otherwise
silently cap RSS to one fetch per calendar day.

Wires RssAdapter.fetchTenders() to fan out over active feed rows (native
fetch + AbortController 15s + response-size ceiling, catch-per-feed),
registers it in tenders.module.ts, and seeds the 'rss' poll config
active with pollGranularity='tick' plus a default-active service.bund.de
feed row (subreport-elvis has no single canonical URL — zero rows seeded,
admin adds relevant municipality feeds).

Migration applied locally per project convention (host -> container IP).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

apps/api/prisma/migrations/20260723111946_add_rss_feed_source_and_poll_granularity/migration.sql

@@ -0,0 +1,20 @@
+-- AlterTable
+ALTER TABLE "CalendarSource" ADD COLUMN     "domain" TEXT;
+
+-- AlterTable
+ALTER TABLE "TenderSourcePollConfig" ADD COLUMN     "pollGranularity" TEXT NOT NULL DEFAULT 'day';
+
+-- CreateTable
+CREATE TABLE "TenderRssFeedSource" (
+    "id" TEXT NOT NULL,
+    "url" TEXT NOT NULL,
+    "label" TEXT NOT NULL,
+    "isActive" BOOLEAN NOT NULL DEFAULT true,
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updatedAt" TIMESTAMP(3) NOT NULL,
+
+    CONSTRAINT "TenderRssFeedSource_pkey" PRIMARY KEY ("id")
+);
+
+-- CreateIndex
+CREATE UNIQUE INDEX "TenderRssFeedSource_url_key" ON "TenderRssFeedSource"("url");

apps/api/prisma/schema.prisma

@@ -419,6 +419,31 @@ model TenderSourcePollConfig {
   pollIntervalMin Int       @default(60) // D-04 default hourly
   isActive        Boolean   @default(false)
   lastIngestedDay DateTime? // day-cursor, NOT a timestamp (RESEARCH Pattern 1)
+  // Phase 14, Plan 02 (D-15/Pitfall 1): 'day' | 'tick'. 'day' sources
+  // (doe-opendata/ai-netserver/cosinex-dtvp) keep the lastIngestedDay-gated
+  // once-per-calendar-day fetch, unchanged. 'tick' sources (rss) are
+  // fetched on every active scheduler tick, ignoring lastIngestedDay
+  // entirely — the day-cursor gate was built for a genuine daily
+  // batch-export API and would otherwise silently cap RSS to one fetch
+  // per day regardless of pollIntervalMin.
+  pollGranularity String    @default("day")
+  createdAt       DateTime  @default(now())
+  updatedAt       DateTime  @updatedAt
+}
+
+// Phase 14, Plan 02 (INGEST-04, D-14) — admin-managed GLOBAL RSS feed list.
+// Deliberately NO tenantId (mirrors TenderSourcePollConfig's global/
+// RLS-exempt stance, D-08: RSS feeds are public and identical for every
+// tenant). Feed URLs are RUNTIME admin input — unlike the hardcoded
+// NETSERVER_PORTALS/COSINEX_BASE_URL constants, the code-level
+// SourceRegistry denylist gate does NOT cover this data (RESEARCH.md
+// Pitfall 3); TenderRssFeedSourceService enforces a SEPARATE save-time
+// hostname/SSRF guard (T-14-02-01) on create/update.
+model TenderRssFeedSource {
+  id        String   @id @default(uuid())
+  url       String   @unique
+  label     String
+  isActive  Boolean  @default(true)
   createdAt DateTime @default(now())
   updatedAt DateTime @updatedAt
 }

apps/api/src/tenders/adapters/rss.adapter.spec.ts

@@ -1,6 +1,6 @@
 import { readFileSync } from 'fs';
 import { join } from 'path';
-import { describe, expect, it } from 'vitest';
+import { afterEach, describe, expect, it, vi } from 'vitest';
 import { RssAdapter } from './rss.adapter';
 
 /**
@@ -20,16 +20,46 @@ const SUBREPORT_ELVIS_FIXTURE = readFileSync(
   'utf8',
 );
 
+/** Minimal prisma-shaped fake — only `tenderRssFeedSource.findMany` is used by RssAdapter. */
+function makeFakePrisma(feeds: any[] = []) {
+  return {
+    tenderRssFeedSource: {
+      findMany: vi.fn(async () => feeds),
+    },
+  } as any;
+}
+
+/** Stubs global `fetch` to resolve with `xml` for every call (fan-out tests). */
+function stubFetchResolvingXml(xmlByUrl: Record<string, string>): void {
+  vi.stubGlobal(
+    'fetch',
+    vi.fn(async (url: string) => {
+      const xml = xmlByUrl[url];
+      if (xml === undefined) {
+        return { ok: false, status: 404 } as Response;
+      }
+      const bytes = Buffer.from(xml, 'utf8');
+      return {
+        ok: true,
+        status: 200,
+        headers: new Headers({ 'content-length': String(bytes.byteLength) }),
+        arrayBuffer: async () =>
+          bytes.buffer.slice(bytes.byteOffset, bytes.byteOffset + bytes.byteLength),
+      } as unknown as Response;
+    }),
+  );
+}
+
 describe('RssAdapter', () => {
   it('declares sourceType rss and the symbolic rss portal', () => {
-    const adapter = new RssAdapter();
+    const adapter = new RssAdapter(makeFakePrisma());
     expect(adapter.sourceType).toBe('rss');
     expect(adapter.portals).toEqual(['rss']);
   });
 
   describe('parseFeed (pure, fixture-driven)', () => {
     it('parses service.bund.de items: sourceType/sourcePortal set, pubDate present, numeric HTML entities decoded in title', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
 
       const records = adapter.parseFeed(SERVICE_BUND_FIXTURE, 'service-bund');
 
@@ -54,7 +84,7 @@ describe('RssAdapter', () => {
     });
 
     it('uses the item guid as sourceNoticeId for service.bund.de (plain-text guid, no isPermaLink attribute)', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const records = adapter.parseFeed(SERVICE_BUND_FIXTURE, 'service-bund');
 
       const ids = records.map((r) => r.sourceNoticeId);
@@ -63,7 +93,7 @@ describe('RssAdapter', () => {
     });
 
     it('parses subreport-elvis items: publishedAt null (no item pubDate), title from CDATA, guid isPermaLink=false extracted as plain text', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
 
       const records = adapter.parseFeed(
         SUBREPORT_ELVIS_FIXTURE,
@@ -92,7 +122,7 @@ describe('RssAdapter', () => {
     });
 
     it('never carries description HTML into the record (V5 — no stored-XSS surface)', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const records = adapter.parseFeed(
         SUBREPORT_ELVIS_FIXTURE,
         'subreport-neuss',
@@ -106,7 +136,7 @@ describe('RssAdapter', () => {
     });
 
     it('bare-minimum bag: buyerName/procedureType/deadlineAt always null (baseline mapping only, D-04/D-05)', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const records = adapter.parseFeed(SERVICE_BUND_FIXTURE, 'service-bund');
 
       for (const record of records) {
@@ -124,26 +154,26 @@ describe('RssAdapter', () => {
     });
 
     it('returns [] for empty XML instead of throwing', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       expect(adapter.parseFeed('', 'empty-feed')).toEqual([]);
     });
 
     it('returns [] for malformed XML instead of throwing', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       expect(
         adapter.parseFeed('<rss><channel><item><title>', 'broken-feed'),
       ).toEqual([]);
     });
 
     it('returns [] for well-formed XML with no <item> at all', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const xml =
         '<?xml version="1.0"?><rss version="2.0"><channel><title>Empty</title></channel></rss>';
       expect(adapter.parseFeed(xml, 'no-items-feed')).toEqual([]);
     });
 
     it('skips a single item missing <link> without aborting the rest', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const xml = `<?xml version="1.0"?>
         <rss version="2.0"><channel>
           <item><title>Ohne Link</title><guid>no-link-guid</guid></item>
@@ -157,7 +187,7 @@ describe('RssAdapter', () => {
     });
 
     it('falls back to sha256(link) for sourceNoticeId when guid is absent', () => {
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       const xml = `<?xml version="1.0"?>
         <rss version="2.0"><channel>
           <item><title>No Guid</title><link>https://example.invalid/no-guid</link></item>
@@ -177,11 +207,107 @@ describe('RssAdapter', () => {
       // only proves RssAdapter's OWN output shape (parseFeed), not the
       // normalizer dispatch itself (kept in the normalizer's own spec file
       // to avoid duplicating TenderNormalizerService test infrastructure).
-      const adapter = new RssAdapter();
+      const adapter = new RssAdapter(makeFakePrisma());
       expect(adapter.sourceType).toBe('rss');
     });
   });
 
+  describe('fetchTenders (internal fan-out over active TenderRssFeedSource rows, Plan 14-02 Task 2)', () => {
+    afterEach(() => {
+      vi.unstubAllGlobals();
+    });
+
+    it('fetches and parses every active feed, fanning out over findMany({isActive:true})', async () => {
+      const feeds = [
+        { id: 'f1', url: 'https://service-bund.invalid/rss.xml', label: 'service-bund', isActive: true },
+        { id: 'f2', url: 'https://subreport.invalid/rss.xml', label: 'subreport-neuss', isActive: true },
+      ];
+      const prisma = makeFakePrisma(feeds);
+      stubFetchResolvingXml({
+        'https://service-bund.invalid/rss.xml': SERVICE_BUND_FIXTURE,
+        'https://subreport.invalid/rss.xml': SUBREPORT_ELVIS_FIXTURE,
+      });
+      const adapter = new RssAdapter(prisma);
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(prisma.tenderRssFeedSource.findMany).toHaveBeenCalledWith({
+        where: { isActive: true },
+      });
+      const portals = new Set(records.map((r) => r.sourcePortal));
+      expect(portals).toEqual(new Set(['service-bund', 'subreport-neuss']));
+      expect(records.length).toBeGreaterThan(1);
+    });
+
+    it('catch-per-feed: a feed that fails to fetch is skipped, the other feed still resolves (D-01)', async () => {
+      const feeds = [
+        { id: 'f1', url: 'https://broken.invalid/rss.xml', label: 'broken', isActive: true },
+        { id: 'f2', url: 'https://ok.invalid/rss.xml', label: 'ok-feed', isActive: true },
+      ];
+      const prisma = makeFakePrisma(feeds);
+      stubFetchResolvingXml({
+        'https://ok.invalid/rss.xml': SERVICE_BUND_FIXTURE,
+        // 'broken.invalid' deliberately absent -> stub resolves 404 -> throws
+      });
+      const adapter = new RssAdapter(prisma);
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records.every((r) => r.sourcePortal === 'ok-feed')).toBe(true);
+      expect(records.length).toBeGreaterThanOrEqual(1);
+    });
+
+    it('returns [] when there are no active feeds', async () => {
+      const prisma = makeFakePrisma([]);
+      const adapter = new RssAdapter(prisma);
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records).toEqual([]);
+    });
+
+    it('returns [] without throwing when a feed fetch throws (network error)', async () => {
+      const feeds = [
+        { id: 'f1', url: 'https://unreachable.invalid/rss.xml', label: 'unreachable', isActive: true },
+      ];
+      const prisma = makeFakePrisma(feeds);
+      vi.stubGlobal(
+        'fetch',
+        vi.fn(async () => {
+          throw new Error('network unreachable');
+        }),
+      );
+      const adapter = new RssAdapter(prisma);
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records).toEqual([]);
+    });
+
+    it('rejects a feed response exceeding the size ceiling (DoS mitigation, T-14-02-02)', async () => {
+      const feeds = [
+        { id: 'f1', url: 'https://huge.invalid/rss.xml', label: 'huge', isActive: true },
+      ];
+      const prisma = makeFakePrisma(feeds);
+      vi.stubGlobal(
+        'fetch',
+        vi.fn(async () => {
+          return {
+            ok: true,
+            status: 200,
+            headers: new Headers({ 'content-length': String(11 * 1024 * 1024) }),
+            arrayBuffer: async () => new ArrayBuffer(0),
+          } as unknown as Response;
+        }),
+      );
+      const adapter = new RssAdapter(prisma);
+
+      const records = await adapter.fetchTenders('2026-07-23');
+
+      expect(records).toEqual([]); // oversized feed skipped, catch-per-feed (D-01)
+    });
+  });
+
   it('never imports or uses axios (native fetch is the sole HTTP client convention)', () => {
     const source = readFileSync(join(__dirname, 'rss.adapter.ts'), 'utf8');
     expect(source).not.toMatch(/from ['"]axios['"]/);

apps/api/src/tenders/adapters/rss.adapter.ts

@@ -1,6 +1,7 @@
 import { Injectable, Logger } from '@nestjs/common';
 import { createHash } from 'crypto';
 import { XMLParser } from 'fast-xml-parser';
+import { PrismaService } from '../../prisma/prisma.service';
 import type { RawTenderRecord, SourceType } from '../tender.types';
 import type { TenderSourceAdapter } from './tender-source-adapter.interface';
 
@@ -39,7 +40,21 @@ import type { TenderSourceAdapter } from './tender-source-adapter.interface';
  * read by this adapter, so no raw HTML fragment is ever carried into a
  * RawTenderRecord (V5 — no stored-XSS surface, same text-only discipline
  * as NetServerAdapter/CosinexAdapter).
+ *
+ * `fetchTenders()` (Plan 14-02 Task 2) is an internal-fan-out adapter
+ * (14-RESEARCH.md Pattern 1, same shape as `NetServerAdapter`'s
+ * multi-portal loop): reads every `isActive` `TenderRssFeedSource` row and
+ * fetches+parses each feed independently, catch-per-feed (D-01 discipline)
+ * — one broken/unreachable feed never blocks the others in the same tick.
+ * `_dayCursor` is accepted for interface conformance but NOT used as a
+ * filter — RSS has no day-batch concept; it is polled every scheduler
+ * tick via `pollGranularity: 'tick'` (D-15, wired in
+ * `tender-ingestion.service.ts`), not gated by the day-cursor at all.
  */
+const RSS_FETCH_TIMEOUT_MS = 15_000;
+/** RSS feeds are normally small; an admin-supplied URL is less trusted than a hardcoded portal (RESEARCH.md Security Domain, DoS mitigation). */
+const RSS_RESPONSE_SIZE_CEILING_BYTES = 10 * 1024 * 1024;
+
 @Injectable()
 export class RssAdapter implements TenderSourceAdapter {
   readonly sourceType: SourceType = 'rss';
@@ -53,14 +68,78 @@ export class RssAdapter implements TenderSourceAdapter {
     attributeNamePrefix: '@_',
   });
 
+  constructor(private readonly prisma: PrismaService) {}
+
   /**
-   * Placeholder — wired to the real `TenderRssFeedSource` internal fan-out
+   * Internal fan-out over every active admin-managed feed (D-14/D-08,
-   * (native fetch + AbortController per admin-added feed URL) in Plan
+   * global — no tenantId). Catch-per-feed (D-01): a single feed that fails
-   * 14-02 Task 2. Kept here only so the class satisfies
+   * to fetch/parse is skipped (logger.warn), never aborting the rest.
-   * `TenderSourceAdapter` for this task's fixture-driven `parseFeed` proof.
    */
   async fetchTenders(_dayCursor: string): Promise<RawTenderRecord[]> {
-    return [];
+    const feeds = await this.prisma.tenderRssFeedSource.findMany({
+      where: { isActive: true },
+    });
+
+    const records: RawTenderRecord[] = [];
+
+    for (const feed of feeds) {
+      try {
+        const xml = await this.fetchFeedXml(feed.url);
+        records.push(...this.parseFeed(xml, feed.label));
+      } catch (error) {
+        this.logger.warn(
+          `RSS feed '${feed.label}' (${feed.url}) fetch failed, skipping this feed for this tick: ${(error as Error).message}`,
+        );
+      }
+    }
+
+    return records;
+  }
+
+  /**
+   * Native fetch + AbortController 15s timeout (project-wide convention,
+   * DoeOpenDataAdapter/NetServerAdapter/CosinexAdapter idiom) plus a
+   * response-size ceiling (T-14-02-02, DoS mitigation) — checked both via
+   * the `Content-Length` header (fast-path, may be absent/wrong) and the
+   * actually-received byte count (authoritative).
+   */
+  private async fetchFeedXml(url: string): Promise<string> {
+    const controller = new AbortController();
+    const timeout = setTimeout(
+      () => controller.abort(),
+      RSS_FETCH_TIMEOUT_MS,
+    );
+
+    try {
+      const res = await fetch(url, {
+        signal: controller.signal,
+        redirect: 'follow',
+      });
+      if (!res.ok) {
+        throw new Error(`RSS feed fetch failed: HTTP ${res.status}`);
+      }
+
+      const contentLength = res.headers.get('content-length');
+      if (
+        contentLength &&
+        Number(contentLength) > RSS_RESPONSE_SIZE_CEILING_BYTES
+      ) {
+        throw new Error(
+          `RSS feed exceeds size ceiling (Content-Length: ${contentLength} bytes)`,
+        );
+      }
+
+      const buffer = await res.arrayBuffer();
+      if (buffer.byteLength > RSS_RESPONSE_SIZE_CEILING_BYTES) {
+        throw new Error(
+          `RSS feed exceeds size ceiling (${buffer.byteLength} bytes received)`,
+        );
+      }
+
+      return new TextDecoder('utf-8').decode(buffer);
+    } finally {
+      clearTimeout(timeout);
+    }
   }
 
   /**

apps/api/src/tenders/dto/tender-rss-feed.dto.ts

@@ -0,0 +1,41 @@
+import {
+  IsBoolean,
+  IsOptional,
+  IsString,
+  IsUrl,
+  MaxLength,
+  MinLength,
+} from 'class-validator';
+
+/**
+ * DTO for admin-managed RSS feed sources (`TenderRssFeedSource`, D-14/D-08).
+ *
+ * `@IsUrl` here is only a COARSE well-formedness check (http/https,
+ * protocol required). The SUBSTANTIVE SSRF/denylist guard — rejecting
+ * DENYLISTED_PORTALS hostnames and private/loopback hosts — is enforced in
+ * `TenderRssFeedSourceService.assertUrlAllowed()`, NOT here (RESEARCH.md
+ * Pitfall 3: an admin-supplied RSS feed URL is runtime data, added long
+ * after `SourceRegistry.register()`'s DI-boot-time denylist check runs —
+ * this DTO alone provides zero protection against a feed URL pointing at
+ * vergabe24/aumass or an internal host). `require_tld: false` deliberately
+ * lets IP-literal URLs pass THIS validation layer so the service-layer
+ * check can reject them with a clear, domain-specific SSRF error message
+ * instead of a generic "invalid URL" one.
+ */
+export class TenderRssFeedDto {
+  @IsUrl({
+    protocols: ['http', 'https'],
+    require_protocol: true,
+    require_tld: false,
+  })
+  url!: string;
+
+  @IsString()
+  @MinLength(1)
+  @MaxLength(200)
+  label!: string;
+
+  @IsOptional()
+  @IsBoolean()
+  isActive?: boolean;
+}

apps/api/src/tenders/tender-ingestion.service.spec.ts

@@ -299,6 +299,59 @@ describe('TenderIngestionService.pollDueSources — catch-per-source error isola
   });
 });
 
+describe("TenderIngestionService.pollDueSources — pollGranularity 'tick' gate (D-15, Phase 14 Plan 02)", () => {
+  it("fetches a pollGranularity='tick' source on every tick, while a 'day' source with today's next-day gated-out lastIngestedDay is skipped in the SAME tick", async () => {
+    const prisma = makeFakePrisma({
+      'doe-opendata': { lastIngestedDay: dayDate(-1) }, // -> next day is today, gated out ('day' path unchanged)
+      rss: { pollGranularity: 'tick', lastIngestedDay: null }, // 'tick' — no day-cursor gate at all
+    });
+    const doeAdapter = { fetchTenders: vi.fn() };
+    const rssAdapter = {
+      fetchTenders: vi.fn().mockResolvedValue([
+        { ...BASE_RECORD, sourcePortal: 'rss', sourceNoticeId: 'rss-1', ocid: null, dedupKey: 'rss:rss-1' },
+      ]),
+    };
+    const registry = makeFakeRegistry({ 'doe-opendata': doeAdapter, rss: rssAdapter });
+    const normalizer = { normalize: vi.fn((raw: any) => raw) };
+    const service = makeService(prisma, registry, normalizer);
+
+    await service.pollDueSources();
+
+    expect(doeAdapter.fetchTenders).not.toHaveBeenCalled(); // 'day' gate unchanged
+    expect(rssAdapter.fetchTenders).toHaveBeenCalledTimes(1); // 'tick' — fetched regardless
+    expect(prisma.__store.tenders.size).toBe(1);
+  });
+
+  it("does NOT advance or read lastIngestedDay for a 'tick' source — it is fetched again next tick even with lastIngestedDay already set to today", async () => {
+    const prisma = makeFakePrisma({
+      rss: { pollGranularity: 'tick', lastIngestedDay: dayDate(0) }, // today — would gate a 'day' source out entirely
+    });
+    const rssAdapter = { fetchTenders: vi.fn().mockResolvedValue([]) };
+    const registry = makeFakeRegistry({ rss: rssAdapter });
+    const normalizer = { normalize: vi.fn() };
+    const service = makeService(prisma, registry, normalizer);
+
+    await service.pollDueSources();
+    await service.pollDueSources();
+
+    expect(rssAdapter.fetchTenders).toHaveBeenCalledTimes(2);
+    // lastIngestedDay is never touched by the 'tick' path.
+    expect(prisma.__store.configs.get('rss').lastIngestedDay).toEqual(dayDate(0));
+  });
+
+  it("passes berlinToday (not a day-cursor loop) as the single argument to a 'tick' adapter's fetchTenders", async () => {
+    const prisma = makeFakePrisma({ rss: { pollGranularity: 'tick' } });
+    const rssAdapter = { fetchTenders: vi.fn().mockResolvedValue([]) };
+    const registry = makeFakeRegistry({ rss: rssAdapter });
+    const normalizer = { normalize: vi.fn() };
+    const service = makeService(prisma, registry, normalizer);
+
+    await service.pollDueSources();
+
+    expect(rssAdapter.fetchTenders).toHaveBeenCalledWith(berlinTodayStr());
+  });
+});
+
 describe('TenderIngestionService.pollDueSources — dedupActive gate (D-05)', () => {
   it('passes dedupActive=false to resolve() when exactly one config is active', async () => {
     const prisma = makeFakePrisma({ 'doe-opendata': { lastIngestedDay: dayDate(-2) } });

apps/api/src/tenders/tender-ingestion.service.ts

@@ -70,6 +70,15 @@ function addOneDay(day: string): string {
  * structurally prevents a backfill flood. Rows that merely changed
  * (SCHEMA-02 contentHash update) or were merged as an additional source
  * (`created: false`) are NOT included.
+ *
+ * Phase 14, Plan 02 (D-15/RESEARCH.md Pitfall 1): each config now branches
+ * on `pollGranularity` ('day' | 'tick'). 'day' sources (doe-opendata/
+ * ai-netserver/cosinex-dtvp) keep the exact pre-existing nextDayToFetch
+ * gate — zero regression. 'tick' sources (rss) skip the day-cursor gate
+ * entirely and fetch on every active tick, since the day-cursor mechanism
+ * was built for a genuine daily batch-export API and would otherwise
+ * silently cap a finer-grained source to one fetch per calendar day
+ * regardless of its configured `pollIntervalMin`.
  */
 @Injectable()
 export class TenderIngestionService {
@@ -123,6 +132,32 @@ export class TenderIngestionService {
             continue;
           }
 
+          if (config.pollGranularity === 'tick') {
+            // D-15/Pitfall 1: 'tick' sources (rss) are NOT gated on
+            // lastIngestedDay at all — the day-cursor gate was built for a
+            // genuine daily batch-export API (DÖE) and would otherwise
+            // silently cap these sources to one fetch per calendar day
+            // regardless of pollIntervalMin. Fetched unconditionally on
+            // every tick this config is active; lastIngestedDay is never
+            // read or advanced for 'tick' sources.
+            const rawRecords = await adapter.fetchTenders(berlinToday);
+            const normalized = rawRecords.map((r) =>
+              this.normalizer.normalize(r),
+            );
+
+            for (const tender of normalized) {
+              const { tenderId, created } = await this.dedup.resolve(tender, {
+                dedupActive,
+              });
+              if (created) newTenderIds.push(tenderId);
+            }
+
+            anyDayFetched = true; // also triggers pruneExpiredTenders below, same as 'day' sources
+            continue;
+          }
+
+          // 'day' path — UNCHANGED from before this plan (zero regression
+          // for doe-opendata/ai-netserver/cosinex-dtvp, D-15).
           let cursorDay = nextDayToFetch(config.lastIngestedDay);
           if (!cursorDay) {
             this.logger.debug(

apps/api/src/tenders/tender-rss-feed.service.spec.ts

@@ -0,0 +1,217 @@
+import { BadRequestException } from '@nestjs/common';
+import { describe, expect, it } from 'vitest';
+import { TenderRssFeedSourceService } from './tender-rss-feed.service';
+
+/**
+ * TenderRssFeedSourceService.spec — proves the save-time hostname/SSRF
+ * guard (T-14-02-01, D-14/RESEARCH.md Pitfall 3): a runtime-admin-supplied
+ * RSS feed URL is NOT covered by the code-level SourceRegistry denylist
+ * gate (that only checks an adapter's statically-declared `portals` array
+ * at DI-boot time) — this service is the separate, independent
+ * enforcement point.
+ *
+ * Uses the same hand-rolled prisma-shaped fake convention as
+ * tender-notification-pref.service.spec.ts / tender-saved-search.service.spec.ts
+ * (in-memory Map, no live DB connection).
+ */
+function makeFakePrisma() {
+  const rows = new Map<string, any>();
+  let seq = 0;
+
+  return {
+    tenderRssFeedSource: {
+      findMany: async ({ orderBy }: any) => {
+        const all = [...rows.values()];
+        if (orderBy?.createdAt === 'asc') {
+          all.sort((a, b) => a.createdAt.getTime() - b.createdAt.getTime());
+        }
+        return all;
+      },
+      create: async ({ data }: any) => {
+        seq += 1;
+        const row = {
+          id: `feed-${seq}`,
+          createdAt: new Date(Date.now() + seq),
+          updatedAt: new Date(Date.now() + seq),
+          ...data,
+        };
+        rows.set(row.id, row);
+        return row;
+      },
+      delete: async ({ where }: any) => {
+        const existing = rows.get(where.id);
+        rows.delete(where.id);
+        return existing;
+      },
+    },
+    __rows: rows,
+  };
+}
+
+describe('TenderRssFeedSourceService', () => {
+  describe('create() — save-time hostname/SSRF guard (T-14-02-01)', () => {
+    it('rejects a vergabe24.de feed URL (denylisted portal, D-14)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({
+          url: 'https://www.vergabe24.de/rss.xml',
+          label: 'vergabe24',
+        }),
+      ).rejects.toThrow(BadRequestException);
+      expect(prisma.__rows.size).toBe(0);
+    });
+
+    it('rejects an aumass.de feed URL (denylisted portal, D-14)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'https://aumass.de/feed', label: 'aumass' }),
+      ).rejects.toThrow(BadRequestException);
+      expect(prisma.__rows.size).toBe(0);
+    });
+
+    it('rejects a subdomain of a denylisted host (substring match on hostname)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({
+          url: 'https://feeds.vergabe24.de/rss.xml',
+          label: 'vergabe24-sub',
+        }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('accepts a valid service.bund.de feed URL', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      const created = await service.create({
+        url: 'https://www.service.bund.de/Content/Globals/Functions/RSSFeed/RSSGenerator_Ausschreibungen.xml',
+        label: 'service-bund',
+      });
+
+      expect(created.url).toBe(
+        'https://www.service.bund.de/Content/Globals/Functions/RSSFeed/RSSGenerator_Ausschreibungen.xml',
+      );
+      expect(created.isActive).toBe(true); // default when omitted
+      expect(prisma.__rows.size).toBe(1);
+    });
+
+    it('rejects a non-http(s) scheme (e.g. file://)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'file:///etc/passwd', label: 'local-file' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects a malformed URL', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'not-a-url', label: 'broken' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects a loopback host (127.0.0.1, SSRF)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({
+          url: 'http://127.0.0.1:8080/internal-feed.xml',
+          label: 'internal',
+        }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects "localhost" (SSRF)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'http://localhost:3000/feed', label: 'x' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects a private 10.x.x.x host (SSRF)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'http://10.0.0.5/feed', label: 'x' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects a private 192.168.x.x host (SSRF)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'http://192.168.1.1/feed', label: 'x' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('rejects an IPv6 loopback host ([::1], SSRF)', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await expect(
+        service.create({ url: 'http://[::1]/feed', label: 'x' }),
+      ).rejects.toThrow(BadRequestException);
+    });
+
+    it('creates isActive=false when explicitly supplied', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      const created = await service.create({
+        url: 'https://example-tenders.invalid/rss.xml',
+        label: 'inactive-feed',
+        isActive: false,
+      });
+
+      expect(created.isActive).toBe(false);
+    });
+  });
+
+  describe('list()/remove()', () => {
+    it('list() returns created feeds ordered by createdAt asc', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      await service.create({
+        url: 'https://a.example-tenders.invalid/rss.xml',
+        label: 'a',
+      });
+      await service.create({
+        url: 'https://b.example-tenders.invalid/rss.xml',
+        label: 'b',
+      });
+
+      const list = await service.list();
+      expect(list.map((f: any) => f.label)).toEqual(['a', 'b']);
+    });
+
+    it('remove() deletes the feed by id', async () => {
+      const prisma = makeFakePrisma();
+      const service = new TenderRssFeedSourceService(prisma as any);
+
+      const created = await service.create({
+        url: 'https://c.example-tenders.invalid/rss.xml',
+        label: 'c',
+      });
+
+      const result = await service.remove(created.id);
+
+      expect(result).toEqual({ success: true });
+      expect(prisma.__rows.size).toBe(0);
+    });
+  });
+});

apps/api/src/tenders/tender-rss-feed.service.ts

@@ -0,0 +1,127 @@
+import { BadRequestException, Injectable } from '@nestjs/common';
+import { PrismaService } from '../prisma/prisma.service';
+import type { TenderRssFeedDto } from './dto/tender-rss-feed.dto';
+import { DENYLISTED_PORTALS } from './source-registry';
+
+/**
+ * TenderRssFeedSourceService — admin CRUD for the GLOBAL RSS feed list
+ * (`TenderRssFeedSource`, D-14/D-08). No `forTenant()`/RLS — this is
+ * platform-wide config, mirroring `TenderSourcePollConfig`'s stance.
+ *
+ * Save-time hostname/SSRF guard (T-14-02-01, RESEARCH.md Pitfall 3): RSS
+ * feed URLs are RUNTIME admin input, added long after
+ * `SourceRegistry.register()`'s DI-boot-time `DENYLISTED_PORTALS` check
+ * runs — that gate provides ZERO protection here. This service is the
+ * SEPARATE, independent enforcement point: reject non-http(s) schemes,
+ * reject any hostname containing a `DENYLISTED_PORTALS` entry (same
+ * constant as the code-level gate — single source of truth, D-14), and
+ * reject private/loopback hosts (SSRF guard, analog to T-10-06). Runs on
+ * BOTH create and update paths.
+ */
+@Injectable()
+export class TenderRssFeedSourceService {
+  constructor(private readonly prisma: PrismaService) {}
+
+  async list() {
+    return this.prisma.tenderRssFeedSource.findMany({
+      orderBy: { createdAt: 'asc' },
+    });
+  }
+
+  async create(dto: TenderRssFeedDto) {
+    this.assertUrlAllowed(dto.url);
+
+    return this.prisma.tenderRssFeedSource.create({
+      data: {
+        url: dto.url,
+        label: dto.label,
+        isActive: dto.isActive ?? true,
+      },
+    });
+  }
+
+  async remove(id: string) {
+    await this.prisma.tenderRssFeedSource.delete({ where: { id } });
+    return { success: true };
+  }
+
+  /**
+   * Rejects (BadRequestException, HTTP 400):
+   * - non-http(s) schemes (e.g. `file://`, `ftp://`, `javascript:`)
+   * - hostnames containing a DENYLISTED_PORTALS entry (vergabe24/aumass)
+   * - private/loopback/link-local IP-literal hosts (SSRF)
+   *
+   * Deliberately string/IP-literal-based, not DNS-resolution-based — same
+   * scope as the existing SSRF-guard pattern documented for
+   * NETSERVER_PORTALS/COSINEX_BASE_URL (T-10-06); resolving a hostname to
+   * check its IP at save-time would itself be an SSRF-adjacent action and
+   * is out of scope for this task.
+   */
+  private assertUrlAllowed(rawUrl: string): void {
+    let parsed: URL;
+    try {
+      parsed = new URL(rawUrl);
+    } catch {
+      throw new BadRequestException('Ungültige URL.');
+    }
+
+    if (parsed.protocol !== 'http:' && parsed.protocol !== 'https:') {
+      throw new BadRequestException(
+        'Nur http(s)-URLs sind für RSS-Feeds erlaubt.',
+      );
+    }
+
+    const hostname = parsed.hostname.toLowerCase();
+
+    if (
+      (DENYLISTED_PORTALS as readonly string[]).some((portal) =>
+        hostname.includes(portal),
+      )
+    ) {
+      throw new BadRequestException(
+        `Der Host '${hostname}' ist AGB-seitig für automatisierten Zugriff gesperrt (Denylist) und darf nicht als RSS-Feed hinterlegt werden.`,
+      );
+    }
+
+    if (isPrivateOrLoopbackHost(hostname)) {
+      throw new BadRequestException(
+        `Der Host '${hostname}' ist ein privater/loopback-Host und darf nicht als RSS-Feed hinterlegt werden (SSRF-Schutz).`,
+      );
+    }
+  }
+}
+
+/**
+ * Best-effort, literal-only private/loopback/link-local host check (SSRF
+ * guard, T-14-02-01) — matches on the hostname string as supplied, no DNS
+ * resolution (see assertUrlAllowed docstring).
+ */
+function isPrivateOrLoopbackHost(rawHostname: string): boolean {
+  // WHATWG URL keeps IPv6 literals bracketed (e.g. `new URL('http://[::1]/').hostname === '[::1]'`) — strip for the checks below.
+  const hostname = rawHostname.replace(/^\[|\]$/g, '');
+
+  if (hostname === 'localhost' || hostname.endsWith('.localhost')) {
+    return true;
+  }
+  if (hostname === '::1' || hostname === '0.0.0.0') {
+    return true;
+  }
+
+  const ipv4Match = hostname.match(/^(\d{1,3})\.(\d{1,3})\.(\d{1,3})\.(\d{1,3})$/);
+  if (ipv4Match) {
+    const [a, b] = ipv4Match.slice(1, 3).map(Number);
+    if (a === 127) return true; // loopback (127.0.0.0/8)
+    if (a === 10) return true; // private (10.0.0.0/8)
+    if (a === 172 && b >= 16 && b <= 31) return true; // private (172.16.0.0/12)
+    if (a === 192 && b === 168) return true; // private (192.168.0.0/16)
+    if (a === 169 && b === 254) return true; // link-local (169.254.0.0/16)
+    if (a === 0) return true; // "this network"
+    return false;
+  }
+
+  // Bare IPv6 literal ranges (unique-local fc00::/7, link-local fe80::/10).
+  if (hostname.startsWith('fc') || hostname.startsWith('fd')) return true;
+  if (hostname.startsWith('fe80:')) return true;
+
+  return false;
+}

apps/api/src/tenders/tenders.module.ts

@@ -6,6 +6,7 @@ import { SettingsModule } from '../settings/settings.module';
 import { CosinexAdapter } from './adapters/cosinex.adapter';
 import { DoeOpenDataAdapter } from './adapters/doe-opendata.adapter';
 import { NetServerAdapter } from './adapters/netserver.adapter';
+import { RssAdapter } from './adapters/rss.adapter';
 import { SourceRegistry } from './source-registry';
 import { seedTendersModule } from './tenders.seed';
 import { TenderDedupService } from './tender-dedup.service';
@@ -15,6 +16,7 @@ import { TenderMailService } from './tender-mail.service';
 import { TenderMatchingService } from './tender-matching.service';
 import { TenderNormalizerService } from './tender-normalizer.service';
 import { TenderNotificationPrefService } from './tender-notification-pref.service';
+import { TenderRssFeedSourceService } from './tender-rss-feed.service';
 import { TenderSavedSearchService } from './tender-saved-search.service';
 import { TenderSchedulerService } from './tender-scheduler.service';
 import { TenderTriageService } from './tender-triage.service';
@@ -93,6 +95,15 @@ import { TendersController } from './tenders.controller';
  * final Wave-4 additive `registry.register(...)` call. Its
  * `TenderSourcePollConfig` row is likewise seeded with `isActive: false`
  * (D-02: activation is a later admin/seed decision, Phase 14 UI).
+ *
+ * Phase 14, Plan 02 (INGEST-04): adds `RssAdapter` (registered alongside
+ * the existing adapters — `rss` is not denylisted) and
+ * `TenderRssFeedSourceService` (admin CRUD for the global feed list,
+ * D-14). Unlike ai-netserver/cosinex-dtvp, the `rss` `TenderSourcePollConfig`
+ * seed is `isActive: true` with `pollGranularity: 'tick'` (D-15) —
+ * service.bund.de is seeded as a default-active `TenderRssFeedSource` row
+ * (RESEARCH.md Open Question 3), so RSS ingestion is live out of the box,
+ * not "framework ready, activation deferred" like the Phase 13 sources.
  */
 @Module({
   imports: [ModuleRegistryModule, SettingsModule],
@@ -101,6 +112,7 @@ import { TendersController } from './tenders.controller';
     DoeOpenDataAdapter,
     NetServerAdapter,
     CosinexAdapter,
+    RssAdapter,
     SourceRegistry,
     TenderNormalizerService,
     TenderDedupService,
@@ -112,6 +124,7 @@ import { TendersController } from './tenders.controller';
     TenderMailService,
     TenderDigestScheduler,
     TenderNotificationPrefService,
+    TenderRssFeedSourceService,
   ],
 })
 export class TendersModule implements OnModuleInit {
@@ -124,18 +137,24 @@ export class TendersModule implements OnModuleInit {
     private readonly doeAdapter: DoeOpenDataAdapter,
     private readonly netServerAdapter: NetServerAdapter,
     private readonly cosinexAdapter: CosinexAdapter,
+    private readonly rssAdapter: RssAdapter,
   ) {}
 
   async onModuleInit(): Promise<void> {
     try {
       // D-06/INGEST-07: registration itself is the denylist-gate enforcement
       // point — SourceRegistry.register() throws for any adapter serving a
-      // denylisted portal. DoeOpenDataAdapter, NetServerAdapter, and
+      // denylisted portal. DoeOpenDataAdapter, NetServerAdapter,
-      // CosinexAdapter are all legitimate (none of tender24/lhs-vpbw/
+      // CosinexAdapter, and RssAdapter are all legitimate (none of
-      // vergabe.landbw/cosinex-dtvp are on the denylist).
+      // tender24/lhs-vpbw/vergabe.landbw/cosinex-dtvp/rss are on the
+      // denylist). Note: RssAdapter's `portals: ['rss']` is a SYMBOLIC
+      // placeholder — the actual admin-supplied feed hostnames are NOT
+      // covered by this gate at all (RESEARCH.md Pitfall 3); that runtime
+      // check lives in TenderRssFeedSourceService instead (D-14).
       this.sourceRegistry.register(this.doeAdapter);
       this.sourceRegistry.register(this.netServerAdapter);
       this.sourceRegistry.register(this.cosinexAdapter);
+      this.sourceRegistry.register(this.rssAdapter);
       this.logger.log(
         `Registered source adapters: ${this.sourceRegistry
           .activeAdapters()
@@ -211,5 +230,51 @@ export class TendersModule implements OnModuleInit {
     } catch (error) {
       this.logger.error('Failed to seed cosinex-dtvp poll config', error);
     }
+
+    try {
+      // Phase 14, Plan 02 (INGEST-04, D-15): seed the rss poll config with
+      // pollGranularity: 'tick' (fetched every active scheduler tick, NOT
+      // gated by lastIngestedDay — see tender-ingestion.service.ts) and
+      // isActive: true — unlike ai-netserver/cosinex-dtvp, RSS is live by
+      // default (RESEARCH.md Open Question 3: service.bund.de is a safe,
+      // genuinely national default feed, seeded below).
+      await this.prisma.tenderSourcePollConfig.upsert({
+        where: { sourceType: 'rss' },
+        update: {},
+        create: {
+          sourceType: 'rss',
+          pollIntervalMin: 60,
+          isActive: true,
+          pollGranularity: 'tick',
+        },
+      });
+      this.logger.log("rss poll config seeded (active, pollGranularity='tick')");
+    } catch (error) {
+      this.logger.error('Failed to seed rss poll config', error);
+    }
+
+    try {
+      // Phase 14, Plan 02 (D-14, RESEARCH.md Open Question 3): seed a
+      // single default-active service.bund.de feed row — the one genuinely
+      // national/global RSS source. subreport-elvis has no single
+      // canonical URL (per-municipality instances, RESEARCH.md Pitfall 2)
+      // — deliberately ZERO subreport-elvis rows seeded; admins add the
+      // municipality feeds relevant to them via the RSS-Feeds admin UI
+      // (Plan 14-02 Task 3).
+      await this.prisma.tenderRssFeedSource.upsert({
+        where: {
+          url: 'https://www.service.bund.de/Content/Globals/Functions/RSSFeed/RSSGenerator_Ausschreibungen.xml',
+        },
+        update: {},
+        create: {
+          url: 'https://www.service.bund.de/Content/Globals/Functions/RSSFeed/RSSGenerator_Ausschreibungen.xml',
+          label: 'service-bund',
+          isActive: true,
+        },
+      });
+      this.logger.log('service.bund.de default RSS feed seeded (active)');
+    } catch (error) {
+      this.logger.error('Failed to seed service.bund.de RSS feed', error);
+    }
   }
 }