merica/frontend/src/utils/yamlCodec.test.js

import { describe, it, expect } from 'vitest';
import * as yaml from 'js-yaml';

/**
 * Helper to serialize canvas topology to Airflow YAML specification.
 * Directly mirrors StudioPage.jsx generateCanvasYaml implementation.
 */
function generateCanvasYaml({ nodes, edges, pipelineId, schedule, bundleName, description }) {
  const taskNodes = nodes.filter((n) => n.type !== 'noteNode');
  if (taskNodes.length === 0) {
    return 'dag_id: ' + (pipelineId || 'example') + '\ntasks: []';
  }

  const tasks = taskNodes.map((n) => {
    const nodeType = n.data?.type || n.type || 'source_db';
    const config = n.data?.params || n.data?.config || {};
    const targetEdges = edges.filter((e) => (e.target || e[1]) === n.id);
    const dependencies = targetEdges.map((e) => e.source || e[0]);
    const isDisabled = n.data?.disabled === true || n.data?.enabled === false;

    return {
      task_id: n.id,
      type: isDisabled ? 'empty' : (n.data?.yamlType || nodeType),
      label: n.data?.label || n.data?.title || n.id,
      enabled: !isDisabled,
      ...(dependencies.length > 0 ? { dependencies } : {}),
      config: isDisabled ? {} : config,
    };
  });

  const pipelineSpec = {
    dag_id: pipelineId || 'example_pipeline',
    schedule: schedule || '@hourly',
    bundle: bundleName || 'dlh-core-ingestion',
    description: description || '',
    tasks,
  };

  return yaml.dump(pipelineSpec);
}

describe('yamlCodec.test.js - Airflow YAML Generator and Codec', () => {
  it('1. Generates clean YAML with dag_id, schedule, and tasks from canvas graph', () => {
    const nodes = [
      {
        id: 'node-db',
        type: 'customNode',
        data: {
          label: 'Source Database',
          type: 'source_db',
          yamlType: 'source_db',
          params: { conn: 'Postgres', table: 'permits' },
        },
      },
      {
        id: 'node-dbt',
        type: 'customNode',
        data: {
          label: 'dbt Transform',
          type: 'dbt_transform',
          yamlType: 'dbt_transform',
          params: { sel: 'gold', tgt: 'postgres' },
        },
      },
    ];

    const edges = [{ source: 'node-db', target: 'node-dbt' }];

    const yamlOutput = generateCanvasYaml({
      nodes,
      edges,
      pipelineId: 'permits_pipeline',
      schedule: '0 */2 * * *',
    });

    expect(yamlOutput).toContain('dag_id: permits_pipeline');
    expect(yamlOutput).toMatch(/schedule:\s*"?0 \*\/2 \* \* \*"?/);
    expect(yamlOutput).toContain('task_id: node-db');
    expect(yamlOutput).toContain('type: source_db');
    expect(yamlOutput).toContain('task_id: node-dbt');
    expect(yamlOutput).toContain('type: dbt_transform');
    expect(yamlOutput).toContain('dependencies:');
    expect(yamlOutput).toContain('- node-db');

    // Parse back to confirm valid YAML syntax
    const parsed = yaml.load(yamlOutput);
    expect(parsed.dag_id).toBe('permits_pipeline');
    expect(parsed.tasks).toHaveLength(2);
    expect(parsed.tasks[1].dependencies).toContain('node-db');
    expect(parsed.tasks[0].config.table).toBe('permits');
  });

  it('2. Maps disabled task to type: empty and enabled: false per Decision #1', () => {
    const nodes = [
      {
        id: 'node-disabled',
        type: 'customNode',
        data: {
          label: 'Disabled Task',
          type: 'ge_quality',
          yamlType: 'ge_quality',
          disabled: true,
          enabled: false,
          params: { suite: 'bronze_suite' },
        },
      },
    ];

    const yamlOutput = generateCanvasYaml({ nodes, edges: [], pipelineId: 'disabled_test' });
    const parsed = yaml.load(yamlOutput);

    expect(parsed.tasks[0].type).toBe('empty');
    expect(parsed.tasks[0].enabled).toBe(false);
    expect(parsed.tasks[0].config).toEqual({});
  });

  it('3. Strictly omits note nodes from YAML output per data-model.md', () => {
    const nodes = [
      {
        id: 'note-1',
        type: 'noteNode',
        data: { text: 'Architecture notes\nInternal details' },
      },
      {
        id: 'node-task',
        type: 'customNode',
        data: { label: 'Real Task', type: 'source_file', params: { path: 'data/' } },
      },
    ];

    const yamlOutput = generateCanvasYaml({ nodes, edges: [], pipelineId: 'notes_test' });
    const parsed = yaml.load(yamlOutput);

    expect(parsed.tasks).toHaveLength(1);
    expect(parsed.tasks[0].task_id).toBe('node-task');
    expect(yamlOutput).not.toContain('note-1');
    expect(yamlOutput).not.toContain('Architecture notes');
  });
});