// Tests for the retired-reference gate. The unit tests use synthetic terms; only the last
// test reads the real, local-only term list, and it SKIPS loudly when that file is absent.
import { test } from 'node:test';
import assert from 'node:assert/strict';
import { execFileSync, spawnSync } from 'node:child_process';
import { mkdtempSync, mkdirSync, writeFileSync } from 'node:fs';
import { tmpdir } from 'node:os';
import { dirname, join } from 'node:path';
import { fileURLToPath } from 'node:url';
import {
  DEFAULT_TERMS_FILE,
  EXEMPT_PATHS,
  loadTerms,
  buildPattern,
  scanEntries,
  runCheck,
} from './check-retired-refs.mjs';

const HERE = dirname(fileURLToPath(import.meta.url));
const REPO = join(HERE, '..');
const CLI = join(HERE, 'check-retired-refs.mjs');
const TERMS = ['retired-alpha', 'Beta.Tool', 'gamma'];
const PATTERN = buildPattern(TERMS);

function tmpDir(prefix) {
  return mkdtempSync(join(tmpdir(), prefix));
}

function termsFile(lines) {
  const file = join(tmpDir('retired-terms-'), 'terms.local.md');
  writeFileSync(file, lines.join('\n'));
  return file;
}

test('loadTerms: one term per line, blanks and # lines ignored, missing file -> null', () => {
  const file = termsFile(['# comment', '', '  retired-alpha  ', 'gamma', '']);
  assert.deepEqual(loadTerms(file), ['retired-alpha', 'gamma']);
  assert.equal(loadTerms(join(tmpDir('retired-none-'), 'absent.md')), null);
});

test('buildPattern: case-insensitive plain substrings, regex metacharacters escaped', () => {
  assert.match('uses RETIRED-ALPHA here', PATTERN);
  assert.match('beta.tool', PATTERN);
  assert.doesNotMatch('betaXtool', PATTERN, '"." is literal, not a wildcard');
  assert.doesNotMatch('repo-mailbox, check-versions', PATTERN);
  assert.throws(() => buildPattern([]), /empty term list/);
});

test('scanEntries reports content hits with line numbers and path hits', () => {
  const findings = scanEntries(
    [
      { path: 'clean.md', content: 'nothing here\n' },
      { path: 'a.md', content: 'line one\nuses retired-alpha/retired-alpha\n' },
      { path: 'docs/gamma.md', content: 'clean body\n' },
    ],
    { pattern: PATTERN },
  );
  assert.deepEqual(
    findings.map((f) => [f.path, f.kind, f.line]),
    [
      ['a.md', 'content', 2],
      ['docs/gamma.md', 'path', 0],
    ],
  );
});

test('scanEntries skips binary content but still checks its path', () => {
  const findings = scanEntries(
    [
      { path: 'img.png', content: null },
      { path: 'gamma.bin', content: null },
    ],
    { pattern: PATTERN },
  );
  assert.deepEqual(findings.map((f) => f.path), ['gamma.bin']);
});

test('scanEntries honours the named exemption list', () => {
  const entries = [{ path: 'keep.md', content: 'gamma' }];
  assert.equal(scanEntries(entries, { pattern: PATTERN }).length, 1);
  assert.equal(scanEntries(entries, { pattern: PATTERN, exempt: ['keep.md'] }).length, 0);
  assert.deepEqual(EXEMPT_PATHS, [], 'the shipped exemption list is empty');
});

function tmpRepo(files) {
  const root = tmpDir('retired-refs-');
  execFileSync('git', ['init', '-q', root]);
  for (const [path, body] of Object.entries(files)) {
    mkdirSync(dirname(join(root, path)), { recursive: true });
    writeFileSync(join(root, path), body);
  }
  execFileSync('git', ['-C', root, 'add', '-A']);
  return root;
}

test('runCheck reads the git index: a staged hit is found, an untracked one is not', () => {
  const root = tmpRepo({ 'README.md': 'points at gamma\n' });
  writeFileSync(join(root, 'untracked.md'), 'retired-alpha');
  const { checked, findings } = runCheck(root, TERMS);
  assert.equal(checked, 1);
  assert.deepEqual(findings.map((f) => `${f.path}:${f.line}`), ['README.md:1']);
});

test('CLI exits 1 on a hit and 0 on a clean tree, always printing the denominator', () => {
  const terms = termsFile(TERMS);
  const run = (root) => spawnSync(process.execPath, [CLI, root, '--terms', terms], { encoding: 'utf8' });
  const dirty = run(tmpRepo({ 'x.md': 'retired-alpha' }));
  assert.equal(dirty.status, 1);
  assert.match(dirty.stdout, /\[ERROR\] retired reference — x\.md:1/);
  assert.match(dirty.stdout, /checked 1 tracked files against 3 terms — 1 hit/);
  const clean = run(tmpRepo({ 'x.md': 'clean\n' }));
  assert.equal(clean.status, 0);
  assert.match(clean.stdout, /checked 1 tracked files against 3 terms — 0 hit/);
});

test('CLI fails on an empty index: 0 files checked verified nothing', () => {
  const empty = tmpDir('retired-refs-empty-');
  execFileSync('git', ['init', '-q', empty]);
  const r = spawnSync(process.execPath, [CLI, empty, '--terms', termsFile(TERMS)], { encoding: 'utf8' });
  assert.equal(r.status, 1);
  assert.match(r.stdout, /verified nothing/);
});

test('CLI reports NOT CHECKED (exit 2) for a missing or empty term list, never a pass', () => {
  const root = tmpRepo({ 'x.md': 'retired-alpha' });
  const missing = spawnSync(process.execPath, [CLI, root, '--terms', join(root, 'absent.md')], { encoding: 'utf8' });
  assert.equal(missing.status, 2);
  assert.match(missing.stdout, /NOT CHECKED — term list missing/);
  const empty = spawnSync(process.execPath, [CLI, root, '--terms', termsFile(['# only a comment'])], { encoding: 'utf8' });
  assert.equal(empty.status, 2);
  assert.match(empty.stdout, /NOT CHECKED — term list empty/);
});

// The FERDIG criterion: this repository's own tracked files carry zero retired references.
// The real term list is local-only by design; without it this test cannot measure anything,
// so it skips with the reason instead of passing silently.
test('this repository: 0 retired references across its tracked files', (t) => {
  const terms = loadTerms();
  if (!terms || terms.length === 0) {
    t.skip(`NOT CHECKED — local term list missing or empty: ${DEFAULT_TERMS_FILE}`);
    return;
  }
  const { checked, findings } = runCheck(REPO, terms);
  assert.ok(checked > 0, 'checked 0 files — verified nothing');
  assert.deepEqual(findings.map((f) => `${f.path}:${f.line}`), []);
});