voyage/tests/validators/plan-validator-v18.test.mjs
Kjell Tore Guttormsen c56f852ab6
fix(validate): a check that can never run is caught when the brief or plan is written
voyage-lekerepo end-to-end trial (2026-10-06), report points 2, 3, 4, 9:
11 of 13 brief criteria were NOT RUN at review with no warning in /trekbrief;
a plan `git grep` criterion passed validator, gate and three critic rounds,
then felled the run in Phase 7; the End-to-end field and Phase 7's text
search never met; a prose Verify left the test command undefined.

- criteria-runner: runnableCheck() screens a declared check without running
  it (same allowlist; an npm run script is not looked up at write time);
  /bin/bash tests/<script> allowed; the plan report carries end_to_end,
  verified when a passed criterion runs a step's End-to-end command (the
  older text form still counts).
- plan-validator 1.8: PLAN_VERIFICATION_NOT_RUNNABLE,
  PLAN_STEP_VERIFY_NOT_RUNNABLE, PLAN_END_TO_END_NOT_RUNNABLE,
  PLAN_END_TO_END_NOT_VERIFIED (the gate inherits them).
- brief-validator: BRIEF_SC_NOT_RUNNABLE warning; /trekbrief offers a
  tests/ script; brief-reviewer no longer proposes unrunnable commands.
- trekexecute: Phase 7 copies end_to_end from the runner; Verify's test
  command is its first backticked span.

Measured on the trial's own brief and plan: 11 warnings (the 11 NOT RUN),
and the plan stops on V5 git grep and the unlisted End-to-end command.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-08 06:42:33 +02:00

252 lines
12 KiB
JavaScript

// tests/validators/plan-validator-v18.test.mjs
// plan_version 1.8 (wave 1, order 3): `Test first:` is required on every step,
// every plan carries one "run the thing" step (`End-to-end:`), the K4 step
// fields (`Interface in:`, `Interface out:`, `Review focus:`) are optional but
// never empty, and an architectural plan carries the design part. 1.7 plans
// keep validating exactly as before.
import { test } from 'node:test';
import { strict as assert } from 'node:assert';
import { execFileSync } from 'node:child_process';
import { mkdtempSync, writeFileSync, rmSync, symlinkSync } from 'node:fs';
import { join, dirname } from 'node:path';
import { tmpdir } from 'node:os';
import { fileURLToPath } from 'node:url';
import { validatePlanContent } from '../../lib/validators/plan-validator.mjs';
import {
parseStepFields, ceremonyFromBrief, resolveCeremony, findModuleCycle, splitSections,
} from '../../lib/parsers/plan-schema.mjs';
import { buildPlan, DESIGN } from '../helpers/plan-v18-builder.mjs';
const HERE = dirname(fileURLToPath(import.meta.url));
const VALIDATOR = join(HERE, '..', '..', 'lib', 'validators', 'plan-validator.mjs');
const codes = r => r.errors.map(e => e.code);
test('a complete 1.8 architectural plan validates strict', () => {
const r = validatePlanContent(buildPlan());
assert.deepEqual(r.errors, [], JSON.stringify(r.errors, null, 2));
assert.equal(r.valid, true);
assert.equal(r.parsed.planVersion, '1.8');
assert.equal(r.parsed.ceremony, 'architectural');
});
test('1.8: a step without `Test first:` is rejected', () => {
const r = validatePlanContent(buildPlan({ steps: [{ testFirst: false }, { endToEnd: 'bash tests/e2e.sh' }] }));
assert.equal(r.valid, false);
assert.ok(codes(r).includes('PLAN_STEP_NO_TEST_FIRST'), codes(r).join(','));
assert.match(r.errors.find(e => e.code === 'PLAN_STEP_NO_TEST_FIRST').message, /Step 1/);
});
test('1.8: `Test first:` must name a File and what it Verifies', () => {
const plan = buildPlan().replace(/ - Verifies: run1\(\)[^\n]*\n/, '');
const r = validatePlanContent(plan);
assert.ok(codes(r).includes('PLAN_STEP_NO_TEST_FIRST'), codes(r).join(','));
});
test('1.8: a plan without a "run the thing" step is rejected', () => {
const r = validatePlanContent(buildPlan({ steps: [{}, {}] }));
assert.equal(r.valid, false);
assert.ok(codes(r).includes('PLAN_NO_END_TO_END'), codes(r).join(','));
});
test('1.8: `End-to-end:` without a backticked command does not count', () => {
const plan = buildPlan().replace(/- \*\*End-to-end:\*\* `[^`]+`/, '- **End-to-end:** run it and look');
assert.ok(codes(validatePlanContent(plan)).includes('PLAN_NO_END_TO_END'));
});
test('1.8: K4 fields are optional, but an empty one is rejected', () => {
const withFields = buildPlan({ steps: [
{ interfaceOut: '`run1(): number` read by Step 2', reviewFocus: 'empty input — pinned by `tests/mod1.test.mjs`' },
{ interfaceIn: '`run1()` from Step 1', endToEnd: 'bash tests/e2e.sh' },
] });
assert.deepEqual(validatePlanContent(withFields).errors, []);
const empty = buildPlan({ steps: [{ interfaceIn: '' }, { endToEnd: 'bash tests/e2e.sh' }] });
assert.ok(codes(validatePlanContent(empty)).includes('PLAN_STEP_FIELD_EMPTY'));
});
test('1.7 plans stay valid: no Test first, no End-to-end, no design required', () => {
const plan = buildPlan({ version: '1.7', ceremony: null, design: false, steps: [{ testFirst: false }] });
const r = validatePlanContent(plan);
assert.deepEqual(r.errors, [], JSON.stringify(r.errors));
assert.equal(r.warnings.find(w => w.code === 'PLAN_VERSION_MISMATCH'), undefined);
});
test('a CRLF plan validates like its LF twin (the manifest parser used to fail on \\r)', () => {
const r = validatePlanContent(buildPlan().replace(/\n/g, '\r\n'));
assert.deepEqual(r.errors, [], JSON.stringify(r.errors));
});
test('1.8: --soft downgrades the 1.8 rules to warnings', () => {
const r = validatePlanContent(buildPlan({ steps: [{ testFirst: false }, {}] }), { strict: false });
assert.equal(r.valid, true);
const w = r.warnings.map(x => x.code);
assert.ok(w.includes('PLAN_STEP_NO_TEST_FIRST') && w.includes('PLAN_NO_END_TO_END'), w.join(','));
});
// --- design part (P2) gated on ceremony (K1) ---
test('ceremony: absent means architectural (heaviest on doubt), so the design part is required', () => {
const r = validatePlanContent(buildPlan({ ceremony: null, design: false }));
assert.ok(codes(r).includes('PLAN_DESIGN_MISSING'), codes(r).join(','));
assert.equal(r.parsed.ceremony, 'architectural');
});
test('ceremony: small and bounded plans need no design part', () => {
for (const c of ['small', 'bounded']) {
const r = validatePlanContent(buildPlan({ ceremony: c, design: false }));
assert.deepEqual(r.errors, [], `${c}: ${JSON.stringify(r.errors)}`);
}
});
test('ceremony: an unknown class is rejected', () => {
assert.ok(codes(validatePlanContent(buildPlan({ ceremony: 'huge' }))).includes('PLAN_CEREMONY_INVALID'));
});
test('design part: every required subsection must be present', () => {
const r = validatePlanContent(buildPlan({ designText: DESIGN.replace('### Data Flow', '### Flow of data') }));
const e = r.errors.find(x => x.code === 'PLAN_DESIGN_SECTION_MISSING');
assert.ok(e, codes(r).join(','));
assert.match(e.message, /Data Flow/);
});
test('design part: a cycle in the module map is rejected', () => {
const cyclic = DESIGN.replace(' parser --> util[src/util.mjs]', ' parser --> util[src/util.mjs]\n util --> cli');
const r = validatePlanContent(buildPlan({ designText: cyclic }));
const e = r.errors.find(x => x.code === 'PLAN_MODULE_CYCLE');
assert.ok(e, codes(r).join(','));
assert.match(e.message, /cli -> parser -> util -> cli/);
});
test('design part: every architecture rule names its enforcer', () => {
const noEnforcer = DESIGN.replace('| Prefer pure functions in the parser | advisory — `.claude/rules/parser.md` | — |',
'| Prefer pure functions in the parser | we will try | — |');
assert.ok(codes(validatePlanContent(buildPlan({ designText: noEnforcer }))).includes('PLAN_ARCH_RULE_NO_ENFORCER'));
});
test('design part: a checked rule names the step that proves it, and the step exists', () => {
const noProof = DESIGN.replace('src` | Step 1 |', 'src` | — |');
assert.ok(codes(validatePlanContent(buildPlan({ designText: noProof }))).includes('PLAN_ARCH_RULE_NO_PROOF'));
const ghost = DESIGN.replace('src` | Step 1 |', 'src` | Step 9 |');
assert.ok(codes(validatePlanContent(buildPlan({ designText: ghost }))).includes('PLAN_ARCH_RULE_NO_PROOF'));
});
test('design part: an advisory rule says which .claude/rules/ file carries it', () => {
const nowhere = DESIGN.replace('advisory — `.claude/rules/parser.md`', 'advisory');
assert.ok(codes(validatePlanContent(buildPlan({ designText: nowhere }))).includes('PLAN_ARCH_RULE_NO_ENFORCER'));
});
// --- parser units ---
test('parseStepFields reads bold and plain field bullets with their blocks', () => {
const f = parseStepFields('### Step 1: x\n\n- **Files:** `a.ts`\n- Test first:\n - File: `t.ts`\n - Verifies: y\n- **Interface in:** `f()`\n');
assert.equal(f.get('Files').inline, '`a.ts`');
assert.match(f.get('Test first').text, /File: `t\.ts`/);
assert.equal(f.get('Interface in').inline, '`f()`');
});
test('parseStepFields stops at the next heading: a later section cannot lend the last step a field', () => {
const f = parseStepFields('### Step 2: x\n\n- **Files:** `a.ts`\n\n## Verification\n\n- **End-to-end:** `npm start`\n');
assert.equal(f.has('End-to-end'), false);
const plan = buildPlan({ steps: [{}, {}] }).replace('## Verification\n', '## Verification\n\n- **End-to-end:** `npm start` → expected: up\n');
assert.ok(codes(validatePlanContent(plan)).includes('PLAN_NO_END_TO_END'), 'only a step can be the run step');
});
test('findModuleCycle finds a cycle through labelled and chained edges', () => {
assert.equal(findModuleCycle('graph TD\n a --> b\n b -->|uses| c\n'), null);
assert.deepEqual(findModuleCycle('graph LR\n a --> b --> c\n c -.-> a\n'), ['a', 'b', 'c', 'a']);
});
// --- review findings (independent code review, 06.10) ---
test('ceremony is read only from frontmatter or the header line, never from prose', () => {
const plan = buildPlan({ ceremony: null, design: false })
.replace('Why: the brief asks for a parser CLI.', 'Why: the brief once said `ceremony: small`.\nceremony: small');
const r = validatePlanContent(plan);
assert.equal(r.parsed.ceremony, 'architectural');
assert.ok(codes(r).includes('PLAN_DESIGN_MISSING'));
});
test('module map: the edge forms mermaid allows are all read, so their cycles are caught', () => {
const cyclic = [
'a-->b\nb-->a',
'a <--> b',
'a ---> b\nb ===> a',
'a -. uses .-> b\nb == calls ==> a',
'a -- calls --> b\nb --> a',
'a & c --> b\nb --> a',
'a --> b; b --> a',
'a:::hot --> b\nb --> a',
'a{{hex}} --> b[[sub]]\nb --> a',
'end-user --> api\napi --> end-user',
'graph.mjs --> x\nx --> graph.mjs',
];
for (const body of cyclic) {
assert.ok(findModuleCycle(`graph TD\n${body}\n`), `cycle not found in: ${JSON.stringify(body)}`);
}
assert.equal(findModuleCycle('graph TD\n a --- b\n b --> a\n'), null, 'an undirected link is not a dependency');
});
test('module map: a link line the parser cannot read is an error, not silently dropped', () => {
const odd = DESIGN.replace(' parser --> util[src/util.mjs]', ' parser --> util[src/util.mjs]\n util --> @weird id');
assert.ok(codes(validatePlanContent(buildPlan({ designText: odd }))).includes('PLAN_MODULE_MAP_UNPARSED'));
});
test('module map: a cycle across two mermaid blocks is caught', () => {
const two = DESIGN.replace(' parser --> util[src/util.mjs]\n```',
' parser --> util[src/util.mjs]\n```\n\n```mermaid\ngraph TD\n util --> cli\n```');
assert.ok(codes(validatePlanContent(buildPlan({ designText: two }))).includes('PLAN_MODULE_CYCLE'));
});
test('findModuleCycle survives a 20 000-node chain (iterative, no stack overflow)', () => {
const edges = [];
for (let i = 0; i < 20000; i++) edges.push([`n${i}`, `n${i + 1}`]);
assert.equal(findModuleCycle(edges), null);
edges.push(['n20000', 'n0']);
assert.equal(findModuleCycle(edges).length, 20002);
});
test('Test first with an empty File: or Verifies: does not count', () => {
const noPath = buildPlan().replace(' - File: `tests/mod1.test.mjs` (new)', ' - File:');
assert.ok(codes(validatePlanContent(noPath)).includes('PLAN_STEP_NO_TEST_FIRST'));
const noWhat = buildPlan().replace(/ - Verifies: run1\(\)[^\n]*/, ' - Verifies:');
assert.ok(codes(validatePlanContent(noWhat)).includes('PLAN_STEP_NO_TEST_FIRST'));
});
test('a column-0 # comment inside a fenced block does not end the step fields', () => {
const f = parseStepFields('### Step 1: x\n\n- **Changes:** run\n ```bash\n# a comment\nnpm test\n ```\n- **Test first:**\n - File: `t.mjs`\n - Verifies: y\n');
assert.ok(f.has('Test first'));
});
test('a ~~~ line inside a ``` block does not end the fence (headings in code stay code)', () => {
const t = '## A\n\n```md\n~~~\n## Not a heading\n~~~\n```\n\n## B\n';
assert.deepEqual(splitSections(t, 2).filter(s => s.title).map(s => s.title), ['A', 'B']);
});
test('K1 stub: ceremonyFromBrief reads the brief frontmatter; absent → architectural', () => {
assert.equal(ceremonyFromBrief('---\ntype: brief\nceremony: small\n---\n# B\n').value, 'small');
const none = ceremonyFromBrief('---\ntype: brief\n---\n# B\n');
assert.equal(none.value, 'architectural');
assert.equal(none.defaulted, true);
assert.equal(resolveCeremony('huge').valid, false);
});
// --- CLI: strict is the default, and the main-module guard survives a symlink ---
test('CLI --strict exits 1 on a 1.8 plan missing Test first; exits 0 through a symlinked path with a space', () => {
const dir = mkdtempSync(join(tmpdir(), 'plan v18 '));
try {
const bad = join(dir, 'bad.md');
writeFileSync(bad, buildPlan({ steps: [{ testFirst: false }, { endToEnd: 'bash tests/e2e.sh' }] }));
let code = 0;
try { execFileSync('node', [VALIDATOR, '--strict', '--json', bad], { stdio: 'pipe' }); } catch (e) { code = e.status; }
assert.equal(code, 1);
const good = join(dir, 'good.md');
writeFileSync(good, buildPlan());
const link = join(dir, 'linked validator.mjs');
symlinkSync(VALIDATOR, link);
const out = execFileSync('node', [link, '--json', good], { encoding: 'utf-8' });
assert.equal(JSON.parse(out).valid, true, 'symlinked CLI must run, not silently no-op');
} finally {
rmSync(dir, { recursive: true, force: true });
}
});