Structural validation in CI

validateArchiModel is a natural pipeline gate: run it in a pre-commit hook or a CI step and let the exit code block merges.

A reusable check script

scripts/validate-model.mts

import { readFile } from 'node:fs/promises';
import { parseArchiModel, validateArchiModel } from '@cda/archi-semantic-core';
import { extractArchiModelXml } from '@cda/archi-semantic-core/archive';

const filePath = process.argv[2];
if (!filePath) {
  console.error('usage: node validate-model.mts <model.archimate>');
  process.exit(2);
}

const bytes = await readFile(filePath);
const model = parseArchiModel(extractArchiModelXml(bytes));
const { valid, errors } = validateArchiModel(model);

for (const issue of errors) {
  console.error(`[${issue.code}] ${issue.path} — ${issue.message}`);
}

if (!valid) {
  console.error(`Model is not valid: ${errors.length} issue(s).`);
  process.exit(1);
}

console.log(`Model OK: ${model.elements.length} elements, ` +
  `${model.relationships.length} relationships, ${model.views.length} views.`);

Wire it into your pipeline

Pre-commit hook (e.g. via lefthook, husky or a plain git hook):

node scripts/validate-model.mts "models/CoreModel.archimate"

CI step (GitHub Actions example):

- name: Validate architecture model
  run: node scripts/validate-model.mts models/CoreModel.archimate

Why extractArchiModelXml first

The .archimate file may be a zip archive (see Working with .archimate archives); reading raw bytes and passing them through extractArchiModelXml covers both shapes before parsing. The /archive subpath import is fine here — CI runs on Node.

Output conventions

Use the exit code for gating and the report for humans: every ArchiValidationIssue already carries a path locator into the model (e.g. relationships[rel-3].sourceId) plus a code you can match on (broken-relationship-source, duplicate-id, …). Print them all — the pipeline may fail, but the developer should know exactly what to fix.