// tests/kb-update/test-excluded-terms.test.mjs
// Preventive gate: no tracked text file may contain a term from the
// maintainer's excluded-terms list. Examples, fixtures and placeholders use
// generic, fictitious material instead.
//
// The list lives in `tests/.excluded-terms.local.json` (gitignored via
// `*.local.json`) and holds `pattern`, `flags`, `positives` and `negatives`.
// Without that file the gate is skipped with a message, so a fresh clone runs
// green; with it, the known positives prove the pattern can find before the
// empty result over the tree is trusted.
//
// Population: `git ls-files` (tracked files only). Binary files (NUL byte in
// the first 8000 bytes, as git decides) are skipped: text in a screenshot is
// rendered pixels, so a byte match there says nothing about what it shows.

import { test } from 'node:test';
import assert from 'node:assert/strict';
import { execFileSync } from 'node:child_process';
import { existsSync, readFileSync } from 'node:fs';
import { fileURLToPath } from 'node:url';

const pluginRoot = fileURLToPath(new URL('../../', import.meta.url));
const listPath = fileURLToPath(new URL('../.excluded-terms.local.json', import.meta.url));

const skip = existsSync(listPath)
  ? false
  : 'tests/.excluded-terms.local.json not present (local-only list) — gate skipped';

function loadList() {
  const list = JSON.parse(readFileSync(listPath, 'utf8'));
  return { ...list, regex: new RegExp(list.pattern, list.flags ?? '') };
}

function trackedFiles() {
  return execFileSync('git', ['ls-files', '-z'], { cwd: pluginRoot, encoding: 'utf8' })
    .split('\0')
    .filter(Boolean);
}

function isBinary(buf) {
  return buf.subarray(0, 8000).includes(0);
}

test('pattern matches every known positive and no known negative', { skip }, () => {
  const { regex, positives = [], negatives = [] } = loadList();
  assert.ok(positives.length > 0, 'the list must carry known positives');
  for (const s of positives) assert.match(s, regex, s);
  for (const s of negatives) assert.doesNotMatch(s, regex, s);
});

test('no tracked text file contains an excluded term', { skip }, () => {
  const { regex } = loadList();
  const files = trackedFiles();
  assert.ok(files.length > 500, `git ls-files returned ${files.length} files — population too small to trust`);
  const hits = [];
  for (const rel of files) {
    let buf;
    try { buf = readFileSync(pluginRoot + rel); } catch { continue; }
    if (isBinary(buf)) continue;
    buf.toString('utf8').split('\n').forEach((line, i) => {
      if (regex.test(line)) hits.push(`${rel}:${i + 1}`);
    });
  }
  assert.deepEqual(hits, [], `excluded terms in tracked files:\n  ${hits.join('\n  ')}`);
});