test(10-03): add failing specs + real DÖE fixtures for adapter/normalizer

- Capture real, trimmed DÖE day-export fixtures (pubDay=2026-07-19, 8
  notices spanning tender/award/planning/untagged-with-awards tag classes)
  under tenders/__fixtures__/, live-downloaded from oeffentlichevergabe.de
- Define RawTenderRecord/NormalizedTenderFields/SourceType (tender.types.ts)
  and TenderSourceAdapter (day-cursor fetchTenders signature per RESEARCH
  Pattern 1)
- RED: doe-opendata.adapter.spec.ts asserts D-02 exact-count filtering,
  HTTP-400 no-op, and a zip-bomb ceiling guard (T-10-07)
- RED: tender-normalizer.service.spec.ts asserts eForms-primary deadline
  recovery where OCDS is null (RESEARCH Pattern 3), nullable deadline/value,
  dedupKey priority (ocid -> sourcePortal:sourceNoticeId), and a stable
  sha256 contentHash
This commit is contained in:
2026-07-21 10:56:13 +02:00
parent f80d491adc
commit f88e2a8141
6 changed files with 388 additions and 0 deletions
@@ -0,0 +1,146 @@
import AdmZip from 'adm-zip';
import { XMLParser } from 'fast-xml-parser';
import { readFileSync } from 'fs';
import { join } from 'path';
import { describe, expect, it } from 'vitest';
import type { RawTenderRecord } from './tender.types';
import { TenderNormalizerService } from './tender-normalizer.service';
/**
* Loads a real notice pair (eForms-DE XML + OCDS JSON) directly from the
* committed fixture ZIPs and assembles it into a RawTenderRecord — the same
* real-fixture precedent as the DKV PDF parser, kept independent of
* DoeOpenDataAdapter so this unit test exercises TenderNormalizerService in
* isolation as a pure mapping function.
*/
const FIXTURES_DIR = join(__dirname, '__fixtures__');
const xmlParser = new XMLParser({
removeNSPrefix: true,
ignoreAttributes: false,
attributeNamePrefix: '@_',
});
function loadRawNoticePair(basename: string): RawTenderRecord {
const eformsZip = new AdmZip(
readFileSync(join(FIXTURES_DIR, 'doe-eforms-sample.zip')),
);
const ocdsZip = new AdmZip(
readFileSync(join(FIXTURES_DIR, 'doe-ocds-sample.zip')),
);
const xmlEntry = eformsZip.getEntry(`${basename}.xml`);
const jsonEntry = ocdsZip.getEntry(`${basename}.json`);
if (!xmlEntry || !jsonEntry) {
throw new Error(`Fixture entry missing for ${basename}`);
}
const eformsPayload = xmlParser.parse(xmlEntry.getData().toString('utf8'));
const ocdsDocument = JSON.parse(jsonEntry.getData().toString('utf8'));
const release = ocdsDocument.releases[0];
return {
sourceType: 'doe-opendata',
sourcePortal: 'doe-opendata',
sourceNoticeId: release.id,
ocid: release.ocid,
sourceUrl: ocdsDocument.uri,
fetchedAt: new Date(),
publishedAt: ocdsDocument.publishedDate
? new Date(ocdsDocument.publishedDate)
: null,
eformsPayload,
ocdsPayload: release,
};
}
// Real notice: eForms XML has a structured
// ProcurementProjectLot/TenderingProcess/TenderSubmissionDeadlinePeriod/EndDate
// (2026-08-18), while the paired OCDS release has no `tender.tenderPeriod` at
// all (RESEARCH.md Pattern 3 live cross-check).
const NOTICE_WITH_EFORMS_ONLY_DEADLINE =
'019f64ec-cd64-4c09-8fd7-4fd12d8eea64-01';
// Real notice: neither the eForms XML nor the OCDS release carries a
// deadline or estimated value (RESEARCH.md Pattern 4 — the common case).
const NOTICE_WITH_NO_DEADLINE_OR_VALUE =
'001c68dd-258a-4bf7-a264-478604ad9c3e-01';
describe('TenderNormalizerService', () => {
const service = new TenderNormalizerService();
it('recovers deadlineAt from eForms-DE XML when the OCDS release has no tenderPeriod (SCHEMA-01 key assertion)', () => {
const raw = loadRawNoticePair(NOTICE_WITH_EFORMS_ONLY_DEADLINE);
const normalized = service.normalize(raw);
expect(normalized.deadlineAt).not.toBeNull();
expect(normalized.deadlineAt?.getUTCFullYear()).toBe(2026);
expect(normalized.deadlineAt?.getUTCMonth()).toBe(7); // August (0-indexed)
expect(normalized.deadlineAt?.getUTCDate()).toBe(18);
});
it('normalizes missing deadline/value to null, not thrown or zero', () => {
const raw = loadRawNoticePair(NOTICE_WITH_NO_DEADLINE_OR_VALUE);
const normalized = service.normalize(raw);
expect(normalized.deadlineAt).toBeNull();
expect(normalized.estimatedValue).toBeNull();
expect(normalized.title).toBeTruthy();
});
it('dedupKey = ocid when present', () => {
const raw = loadRawNoticePair(NOTICE_WITH_EFORMS_ONLY_DEADLINE);
const normalized = service.normalize(raw);
expect(raw.ocid).toBeTruthy();
expect(normalized.dedupKey).toBe(raw.ocid);
});
it('dedupKey falls back to sourcePortal:sourceNoticeId when ocid is absent', () => {
const raw = loadRawNoticePair(NOTICE_WITH_EFORMS_ONLY_DEADLINE);
const rawWithoutOcid: RawTenderRecord = { ...raw, ocid: undefined };
const normalized = service.normalize(rawWithoutOcid);
expect(normalized.dedupKey).toBe(
`${raw.sourcePortal}:${raw.sourceNoticeId}`,
);
});
it('contentHash is a stable sha256 over title+deadline+value+status', () => {
const raw = loadRawNoticePair(NOTICE_WITH_EFORMS_ONLY_DEADLINE);
const first = service.normalize(raw);
const second = service.normalize(raw);
expect(first.contentHash).toBe(second.contentHash);
expect(first.contentHash).toMatch(/^[a-f0-9]{64}$/);
});
it('contentHash changes when the deadline changes', () => {
const raw = loadRawNoticePair(NOTICE_WITH_EFORMS_ONLY_DEADLINE);
const original = service.normalize(raw);
// Mutate the eForms deadline to simulate a Fristverlängerung (SCHEMA-02).
const mutatedEformsPayload = JSON.parse(JSON.stringify(raw.eformsPayload));
const root =
mutatedEformsPayload[Object.keys(mutatedEformsPayload)[0]];
const lots = Array.isArray(root.ProcurementProjectLot)
? root.ProcurementProjectLot
: [root.ProcurementProjectLot];
for (const lot of lots) {
lot.TenderingProcess.TenderSubmissionDeadlinePeriod.EndDate =
'2026-12-31+01:00';
}
const mutatedRaw: RawTenderRecord = {
...raw,
eformsPayload: mutatedEformsPayload,
};
const mutated = service.normalize(mutatedRaw);
expect(mutated.contentHash).not.toBe(original.contentHash);
});
});