okr/tests/innboks-ingest.test.mjs
2026-07-16 20:32:50 +02:00

246 lines
11 KiB
JavaScript

// innboks-ingest.test.mjs
// Step 12 (A2): end-to-end-test av pipeline-orkestratoren scripts/innboks-ingest.mjs.
// Kjoeres i mkdtempSync UTENFOR .claude/ med innboksen NESTET under bundle-rota
// (<tmp>/innboks inni <tmp>) saa walk-skippet ekserseres end-to-end. Verifiserer
// (plan.md Step 12, verifies 1-8):
// (1) idempotens: 2 kjoeringer med ULIK veggklokke -> sha256-manifest deepEqual
// (by-construction-determinisme via original-mtime, ikke cache)
// (2) konformitet: okf-check subprosess BAADE default OG --strict-ingest exit 0
// (3) index-integritet: generateIndexes 2x byte-identisk OG rot-index lister
// ikke innboks/
// (4) original-bevaring: drop-zone-originalene er byte-uendret (sha256)
// (5) >= 1 relasjon: kryss-dokument tittel-omtale (dok-a -> dok-b) emitteres
// som bundle-root-relativ .md-lenke (SC-regex)
// (6) .txt-only full-pipeline kjoerer groent zero-dep (ingen npm-deps)
// (7) strict-scoping (B2, laast A0): pre-eksisterende kuratert fil med ekstern
// lenke + out-of-vocab-type feller IKKE kjoeringen; kuratert fil uroert
// (8) per-dokument-staging/gate (laast A0): multi-doc-drop der ett dok feiler
// gaten -> KUN det dokumentet discardes (ingen spor i tre/index), det
// andre bestaar
// Binaer-originaler (.docx/.pdf) testes conditional { skip: !engines } -- B1-
// aksept: en ren kjoering med ikke-.md-original passerer strict.
// Moenster: tests/okf-check.test.mjs (subprosess-exit) + innboks-write.test.mjs.
import { test } from 'node:test';
import assert from 'node:assert/strict';
import { execFileSync } from 'node:child_process';
import { createHash } from 'node:crypto';
import {
mkdtempSync, mkdirSync, writeFileSync, readFileSync, readdirSync, copyFileSync,
existsSync, rmSync,
} from 'node:fs';
import { tmpdir } from 'node:os';
import { join, dirname, relative } from 'node:path';
import { fileURLToPath } from 'node:url';
import { ingestInbox } from '../scripts/innboks-ingest.mjs';
const ROOT = join(dirname(fileURLToPath(import.meta.url)), '..');
const CHECK = join(ROOT, 'scripts', 'okf-check.mjs');
const SAMPLE = join(ROOT, 'tests', 'fixtures', 'inbox-sample');
// Er binaer-konverterings-motorene installert? (npm-deps = dokumentert prerequisite.)
async function enginesAvailable() {
try {
await import('mammoth');
await import('turndown');
await import('unpdf');
return true;
} catch {
return false;
}
}
const engines = await enginesAvailable();
// Temp-bundle: bundle-rot = <tmp>, drop-zone = <tmp>/innboks (nestet, som prod).
async function withBundle(fn) {
const bundleRoot = mkdtempSync(join(tmpdir(), 'okringest-'));
const inbox = join(bundleRoot, 'innboks');
mkdirSync(inbox, { recursive: true });
try {
await fn({ bundleRoot, inbox });
} finally {
rmSync(bundleRoot, { recursive: true, force: true });
}
}
function sha256(buf) {
return createHash('sha256').update(buf).digest('hex');
}
// Rekursivt sha256-manifest av bundlen: relativ sti -> hash (deterministisk sortert).
function manifest(root) {
const out = {};
const walk = (dir) => {
for (const e of readdirSync(dir, { withFileTypes: true }).sort((a, b) => a.name.localeCompare(b.name))) {
const p = join(dir, e.name);
if (e.isDirectory()) walk(p);
else out[relative(root, p)] = sha256(readFileSync(p));
}
};
walk(root);
return out;
}
// Alle .md-filer under root (rekursivt), relativ sti.
function mdFiles(root) {
const out = [];
const walk = (dir) => {
for (const e of readdirSync(dir, { withFileTypes: true })) {
const p = join(dir, e.name);
if (e.isDirectory()) walk(p);
else if (e.name.endsWith('.md')) out.push(relative(root, p));
}
};
walk(root);
return out;
}
function runCheck(root, ...extra) {
try {
const stdout = execFileSync('node', [CHECK, root, ...extra], { encoding: 'utf8' });
return { status: 0, stdout };
} catch (e) {
return { status: e.status ?? 1, stdout: `${e.stdout || ''}${e.stderr || ''}` };
}
}
function dropTxtFixtures(inbox) {
copyFileSync(join(SAMPLE, 'dok-a.txt'), join(inbox, 'dok-a.txt'));
copyFileSync(join(SAMPLE, 'dok-b.txt'), join(inbox, 'dok-b.txt'));
}
test('ingest: .txt-only full pipeline zero-dep + konformitet + relasjon + original-bevaring (1,2,4,5,6)', async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
dropTxtFixtures(inbox);
const originalsBefore = {
a: sha256(readFileSync(join(inbox, 'dok-a.txt'))),
b: sha256(readFileSync(join(inbox, 'dok-b.txt'))),
};
const run1 = await ingestInbox(inbox, bundleRoot);
assert.equal(run1.failed.length, 0, `ingen feilede dokumenter: ${JSON.stringify(run1.failed)}`);
assert.equal(run1.ingested.length, 2, 'begge .txt-dokumenter ingested');
const m1 = manifest(bundleRoot);
// (1) Idempotens: run2 paa ULIK veggklokke -> byte-identisk bundle.
const run2 = await ingestInbox(inbox, bundleRoot);
assert.equal(run2.failed.length, 0, 'run2 feiler ikke');
assert.deepEqual(manifest(bundleRoot), m1, 'sha256-manifest run1 vs run2 identisk (idempotens by construction)');
// (2) Konformitet: default OG strict-ingest exit 0 (subprosess, kontrakt).
assert.equal(runCheck(bundleRoot).status, 0, 'okf-check default exit 0');
assert.equal(runCheck(bundleRoot, '--strict-ingest').status, 0, 'okf-check --strict-ingest exit 0');
// (4) Original-bevaring: drop-zonen er byte-uendret.
assert.equal(sha256(readFileSync(join(inbox, 'dok-a.txt'))), originalsBefore.a, 'dok-a.txt uendret');
assert.equal(sha256(readFileSync(join(inbox, 'dok-b.txt'))), originalsBefore.b, 'dok-b.txt uendret');
// (5) >= 1 relasjon: kryss-dokument tittel-omtale emitteres som trygg lenke.
const all = mdFiles(bundleRoot)
.map((f) => readFileSync(join(bundleRoot, f), 'utf8'))
.join('\n');
assert.match(all, /\]\(\/[^)]*\.md\)/, 'minst en bundle-root-relativ .md-relasjon emittert');
assert.match(all, /## Relaterte dokumenter/, 'relasjons-seksjon skrevet');
});
});
test('ingest: index-integritet -- regen byte-identisk + rot-index lister ikke innboks/ (3)', async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
dropTxtFixtures(inbox);
await ingestInbox(inbox, bundleRoot);
const rootIndex = readFileSync(join(bundleRoot, 'index.md'), 'utf8');
assert.doesNotMatch(rootIndex, /innboks/, 'rot-index lister ikke drop-zonen');
const { generateIndexes } = await import('../scripts/okf-index.mjs');
const before = manifest(bundleRoot);
generateIndexes(bundleRoot);
assert.deepEqual(manifest(bundleRoot), before, 'generateIndexes re-kjoert er byte-identisk');
});
});
test('ingest: strict-gaten scopes til kjoeringens filer -- kuratert innhold feller ikke (7)', async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
// Pre-eksisterende kuratert fil: out-of-vocab-type + ekstern lenke er LOVLIG
// paa lese-siden -- den skal aldri felle ingestion (B2).
const curatedPath = join(bundleRoot, 'syklus', 'kuratert.md');
mkdirSync(dirname(curatedPath), { recursive: true });
const curated = '---\ntype: Egendefinert\ntitle: Kuratert notat\n---\n'
+ 'Se [veilederen](https://www.regjeringen.no/veileder) for detaljer.\n';
writeFileSync(curatedPath, curated);
dropTxtFixtures(inbox);
const run = await ingestInbox(inbox, bundleRoot);
assert.equal(run.failed.length, 0, 'kuratert fil feller ikke kjoeringen (scoped strict)');
assert.equal(run.ingested.length, 2, 'begge dokumenter ingested');
assert.equal(readFileSync(curatedPath, 'utf8'), curated, 'kuratert fil byte-uroert');
});
});
test('ingest: per-dokument-staging/gate -- fiendtlig dok discardes alene, uten spor (8)', async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
dropTxtFixtures(inbox);
// Dangling bundle-lenke overlever noeytralisering (trygg FORM, mangler paa
// disk) -> strict-gaten feller dokumentet deterministisk.
writeFileSync(
join(inbox, 'fiendtlig.txt'),
'# Fiendtlig notat\n\nSe [detaljer](/dokumenter/finnes-ikke.md) for mer.\n',
);
const run = await ingestInbox(inbox, bundleRoot);
assert.equal(run.failed.length, 1, 'noeyaktig ett dokument discardet');
assert.match(run.failed[0].source, /fiendtlig/, 'det fiendtlige dokumentet');
assert.equal(run.ingested.length, 2, 'de to legitime dokumentene bestaar (Map-Reduce-isolasjon)');
// Ingen spor: verken konsept-fil, peker-fil eller index-oppfoering.
const rest = mdFiles(bundleRoot).map((f) => `${f}\n${readFileSync(join(bundleRoot, f), 'utf8')}`).join('\n');
assert.doesNotMatch(rest, /fiendtlig/i, 'ingen spor av discardet dokument i tre/index');
// Treet som bestaar er fortsatt konformt (default + strict).
assert.equal(runCheck(bundleRoot).status, 0, 'okf-check default exit 0 etter discard');
assert.equal(runCheck(bundleRoot, '--strict-ingest').status, 0, 'strict exit 0 etter discard');
});
});
test('ingest: ukjent extension skippes med notice, feller ikke kjoeringen', async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
dropTxtFixtures(inbox);
writeFileSync(join(inbox, 'regneark.xlsx'), 'ikke-stoettet');
const notices = [];
const run = await ingestInbox(inbox, bundleRoot, { onNotice: (m) => notices.push(m) });
assert.equal(run.skipped.length, 1, 'xlsx skippet');
assert.equal(run.failed.length, 0, 'skip er ikke feil');
assert.equal(run.ingested.length, 2, 'txt-dokumentene ingested');
assert.ok(notices.some((n) => n.includes('xlsx')), 'norsk notice om ukjent filtype');
});
});
test('ingest: binaer-original (.docx/.pdf) passerer strict -- pekerfil uten lenkeform (B1-aksept)', { skip: !engines }, async () => {
await withBundle(async ({ bundleRoot, inbox }) => {
copyFileSync(join(SAMPLE, 'dok.docx'), join(inbox, 'dok.docx'));
copyFileSync(join(SAMPLE, 'dok.pdf'), join(inbox, 'dok.pdf'));
const run = await ingestInbox(inbox, bundleRoot);
assert.equal(run.failed.length, 0, `binaer-ingest feiler ikke: ${JSON.stringify(run.failed)}`);
assert.equal(run.ingested.length, 2, 'begge binaer-dokumenter ingested');
// B1-aksept: hele bundlen passerer strict med ikke-.md-originaler.
assert.equal(runCheck(bundleRoot, '--strict-ingest').status, 0, 'strict exit 0 med .docx/.pdf-originaler');
// Pekerfiler finnes og baerer original-stien i resource (uten lenkeform).
const pointers = mdFiles(bundleRoot).filter((f) => f.endsWith('.kilde.md'));
assert.equal(pointers.length, 2, 'en pekerfil per original');
for (const p of pointers) {
const content = readFileSync(join(bundleRoot, p), 'utf8');
assert.match(content, /^resource: innboks\/dok[^\n]*$/m, 'resource baerer original-stien');
assert.doesNotMatch(content, /\]\(/, 'ingen lenkeform i pekerfil (B1)');
}
// Idempotens gjelder ogsaa binaer-stien (convert-twice-identical + mtime).
const m1 = manifest(bundleRoot);
await ingestInbox(inbox, bundleRoot);
assert.deepEqual(manifest(bundleRoot), m1, 'binaer-ingest idempotent');
});
});