fix: preserve domain classification catalogs and align audit regressions

Author: docgen-ci

global-template/docgen/lib/audit-guard.mjs

@@ -125,11 +125,16 @@ function sanitizeSemanticObject(root, value, inventory, sourceCache) {
     const evidence = dedupeEvidence(evidenceFromAliases(value)
       .map((entry) => canonicalEvidence(root, entry, inventory, sourceCache))
       .filter(Boolean));
-    const requested = normalizeClassification(value.classification ?? value.claimClassification ?? value.certainty);
+    const rawClassification = value.classification ?? value.claimClassification ?? value.certainty;
+    const scalarClassification = rawClassification === undefined || typeof rawClassification === 'string' || typeof rawClassification === 'number';
+    const requested = normalizeClassification(rawClassification);
     const hasLineEvidence = evidence.some((entry) => entry.startLine && !entry.__stale);
-    value.classification = requested === 'FACT' && !hasLineEvidence ? 'INFERENCE' : requested;
-    value.confidence = normalizeConfidence(value.confidence ?? value.confidenceScore, value.classification);
-    if (value.classification !== 'FACT') value.confidence = Math.min(value.confidence, 0.7);
+    const normalizedClassification = requested === 'FACT' && !hasLineEvidence ? 'INFERENCE' : requested;
+    // A field named `classification` can be a domain catalog rather than semantic
+    // metadata. Preserve arrays/objects and only normalize scalar metadata.
+    if (scalarClassification) value.classification = normalizedClassification;
+    value.confidence = normalizeConfidence(value.confidence ?? value.confidenceScore, normalizedClassification);
+    if (scalarClassification && normalizedClassification !== 'FACT') value.confidence = Math.min(value.confidence, 0.7);
     value.evidence = evidence.map(({ __stale, ...entry }) => entry);
     removeEvidenceAliases(value);
     delete value.claimClassification;

global-template/docgen/test/semantic-index.test.mjs

@@ -189,14 +189,16 @@ test('recovery never accepts stale pre-existing plan artifacts when provider wri
   assert.equal(readJson(p.state).stages.plan.status, 'failed');
 });
 
-test('audit rejects unknown model references and publish rejects stale source artifacts', async () => {
+test('audit sanitizes unknown model references and publish rejects stale source artifacts', async () => {
   const root = fixture(); const p = projectPaths(root); const provider = installFakeProvider(root);
   fs.writeFileSync(path.join(root, 'src', 'Resource.java'), 'class Resource {}\n');
   const config = readJson(p.config); config.commandCode = { executable: provider, trust: false, skipOnboarding: false, yolo: false, maxTurns: { default: 30 } }; writeJson(p.config, config);
   const run = spawnSync(process.execPath, [cli, 'all'], { cwd: root, encoding: 'utf8', env: { ...process.env, DOCGEN_PROGRESS: '0' } }); assert.equal(run.status, 0, run.stderr || run.stdout);
   const traceFile = path.join(p.traceability, 'pages', 'overview.json'); const trace = readJson(traceFile); trace.claims[0].sourceModelRefs = ['system:does-not-exist']; writeJson(traceFile, trace);
-  await assert.rejects(() => audit(root), /Quality failed/); const report = readJson(path.join(p.audit, 'deterministic.json')); assert(report.errors.some((error) => /unknown sourceModelRef/.test(error)));
-  trace.claims[0].sourceModelRefs = ['system:resource']; writeJson(traceFile, trace); await audit(root);
+  const summary = await audit(root);
+  const sanitizedTrace = readJson(traceFile);
+  assert.deepEqual(sanitizedTrace.claims[0].sourceModelRefs, []);
+  assert.equal(summary.deterministicFailures, 0);
   fs.appendFileSync(path.join(root, 'src', 'Resource.java'), '// stale\n');
   assert.throws(() => publish(root), /stale relative to current source/);
 });
@@ -221,7 +223,7 @@ test('enterprise model bundle repairs a missing decisions object without discard
   const config = readJson(p.config); config.commandCode = { executable: provider, trust: false, skipOnboarding: false, yolo: false, maxTurns: { default: 30 } }; config.budget.maxProviderCalls = 12; writeJson(p.config, config);
   const run = spawnSync(process.execPath, [cli, 'all'], { cwd: root, encoding: 'utf8', env: { ...process.env, DOCGEN_PROGRESS: '0', DOCGEN_TEST_OMIT_MODEL_ONCE: 'decisions' } });
   assert.equal(run.status, 0, `STDERR:\n${run.stderr}\nSTDOUT:\n${run.stdout}`);
-  assert.match(run.stderr, /modelEnterprise REPAIR \| bundle omitted decisions/);
+  assert.match(run.stderr, /modelEnterprise REPAIR \| unresolved: decisions/);
   assert.equal(fs.existsSync(path.join(p.model, 'decisions.json')), true);
   assert.equal(readJson(p.state).stages.modelEnterprise.status, 'completed');
   const budget = readJson(p.budget); assert.equal(budget.usage.providerCalls, 5); assert.equal(budget.usage.failedCalls, 0);