Statum/tests/markdown/regressions.js

/**
 * Hand-written regression suite for the markdown engine.
 *
 * Covers the three reported bugs (emphasis across a line break, single
 * newline must not start a new paragraph, wrapped list-item lines must stay
 * in the item) plus the security posture (raw HTML neutralised, URL scheme
 * filtering, entity smuggling) and edge cases (CRLF, empty input, tight vs
 * loose lists, setext headings).
 *
 * Usage: qjs tests/markdown/regressions.js
 */
import * as std from 'std';
import { installShim } from './shim.js';

function fail(msg) {
  std.err.puts(msg + '\n');
  std.exit(1);
}

installShim();
(0, eval)(std.loadFile('js/statum.js'));

var render = globalThis.statum._testMarkdown;
if (typeof render !== 'function') fail('statum._testMarkdown is not exposed');

var passed = 0, failed = 0;

function eq(name, input, expected) {
  var got;
  try {
    got = render(input);
  } catch (e) {
    got = '<<threw ' + e + '>>';
  }
  if (got === expected) {
    passed++;
  } else {
    failed++;
    print('FAIL: ' + name);
    print('  input:    ' + JSON.stringify(input));
    print('  expected: ' + JSON.stringify(expected));
    print('  got:      ' + JSON.stringify(got));
  }
}

function ok(name, input, test, description) {
  var got;
  try {
    got = render(input);
  } catch (e) {
    got = '<<threw ' + e + '>>';
  }
  if (test(got)) {
    passed++;
  } else {
    failed++;
    print('FAIL: ' + name + (description ? ' — ' + description : ''));
    print('  input: ' + JSON.stringify(input));
    print('  got:   ' + JSON.stringify(got));
  }
}

// ---------------------------------------------------------------------------
// The three reported bugs
// ---------------------------------------------------------------------------

eq('emphasis wraps across a newline',
   '**bold text\nthat wraps**\n',
   '<p><strong>bold text\nthat wraps</strong></p>\n');

eq('single newline is a soft break, not a new paragraph',
   'one\ntwo\n',
   '<p>one\ntwo</p>\n');

eq('paragraphs only split on blank lines',
   'aaa\nbbb\n\nccc\nddd\n',
   '<p>aaa\nbbb</p>\n<p>ccc\nddd</p>\n');

eq('wrapped line stays inside the list item',
   '- item one\n  continued line\n- item two\n',
   '<ul>\n<li>item one\ncontinued line</li>\n<li>item two</li>\n</ul>\n');

eq('wrapped line in ordered list stays in the item',
   '1. first\n   wrapped\n2. second\n',
   '<ol>\n<li>first\nwrapped</li>\n<li>second</li>\n</ol>\n');

// ---------------------------------------------------------------------------
// Block constructs
// ---------------------------------------------------------------------------

eq('ATX headings', '# Title\n###### Deep\n', '<h1>Title</h1>\n<h6>Deep</h6>\n');
eq('setext headings', 'Title\n=====\nSub\n---\n', '<h1>Title</h1>\n<h2>Sub</h2>\n');
eq('thematic break', '---\n', '<hr />\n');
eq('fenced code with language', '```js\nvar x = 1;\n```\n',
   '<pre><code class="language-js">var x = 1;\n</code></pre>\n');
eq('tilde fence', '~~~\nplain\n~~~\n', '<pre><code>plain\n</code></pre>\n');
eq('indented code', '    code\n', '<pre><code>code\n</code></pre>\n');
eq('blockquote with lazy continuation', '> quoted\nlazy\n',
   '<blockquote>\n<p>quoted\nlazy</p>\n</blockquote>\n');
eq('tight list stays tight', '- a\n- b\n', '<ul>\n<li>a</li>\n<li>b</li>\n</ul>\n');
eq('loose list gets paragraphs', '- a\n\n- b\n',
   '<ul>\n<li>\n<p>a</p>\n</li>\n<li>\n<p>b</p>\n</li>\n</ul>\n');
eq('nested list', '- a\n  - b\n', '<ul>\n<li>a\n<ul>\n<li>b</li>\n</ul>\n</li>\n</ul>\n');
eq('ordered list start attribute', '3. c\n4. d\n',
   '<ol start="3">\n<li>c</li>\n<li>d</li>\n</ol>\n');

// ---------------------------------------------------------------------------
// Inlines
// ---------------------------------------------------------------------------

eq('code span', 'a `code` b\n', '<p>a <code>code</code> b</p>\n');
eq('code span with backslash inside', '``a\\`b``\n', '<p><code>a\\`b</code></p>\n');
eq('backslash escape', '\\*not em\\*\n', '<p>*not em*</p>\n');
eq('named entity', '&copy; 2024\n', '<p>\u00A9 2024</p>\n');
eq('numeric entity', '&#35;\n', '<p>#</p>\n');
eq('invalid entity stays literal', '&madeup;\n', '<p>&amp;madeup;</p>\n');
eq('intraword emphasis with underscores is not emphasis', 'foo_bar_baz\n',
   '<p>foo_bar_baz</p>\n');
eq('intraword emphasis with asterisks', 'foo*bar*baz\n',
   '<p>foo<em>bar</em>baz</p>\n');
eq('triple emphasis', '***both***\n', '<p><em><strong>both</strong></em></p>\n');
eq('hard break with two spaces', 'foo  \nbar\n', '<p>foo<br />\nbar</p>\n');
eq('hard break with backslash', 'foo\\\nbar\n', '<p>foo<br />\nbar</p>\n');
eq('inline link with title', '[t](/u "title")\n',
   '<p><a href="/u" title="title">t</a></p>\n');
eq('image', '![alt](/img.png "cap")\n',
   '<p><img src="/img.png" alt="alt" title="cap" /></p>\n');
eq('reference link', '[t][ref]\n\n[ref]: /u "ti"\n',
   '<p><a href="/u" title="ti">t</a></p>\n');
eq('shortcut reference link', '[ref]\n\n[ref]: /u\n',
   '<p><a href="/u">ref</a></p>\n');
eq('autolink http', '<https://example.com/>\n',
   '<p><a href="https://example.com/">https://example.com/</a></p>\n');
eq('autolink mail', '<mail@example.com>\n',
   '<p><a href="mailto:mail@example.com">mail@example.com</a></p>\n');

// ---------------------------------------------------------------------------
// Security posture
// ---------------------------------------------------------------------------

ok('script tag is neutralised', '<script>alert(1)</script>\n',
   function (h) { return h.indexOf('<script') < 0 && h.indexOf('&lt;script') >= 0; });

ok('img onerror is neutralised', '<img src=x onerror=alert(1)>\n',
   function (h) { return h.indexOf('<img') < 0 && h.indexOf('&lt;img') >= 0; },
   'the tag must render as escaped text, not markup');

ok('javascript: link href is replaced', '[click](javascript:alert(1))\n',
   function (h) { return h.indexOf('javascript:') < 0; },
   'javascript: URLs must be filtered');

ok('data: link href is replaced', '[x](data:text/html,<script>)\n',
   function (h) { return h.indexOf('data:') < 0; });

ok('javascript: image src is replaced', '![x](javascript:alert(1))\n',
   function (h) { return h.indexOf('javascript:') < 0; });

ok('vbscript href is replaced', '[x](vbscript:msgbox)\n',
   function (h) { return h.indexOf('vbscript:') < 0; });

eq('allowed schemes pass through', '[a](https://x) [b](http://x) [c](mailto:a@b) [d](ftp://x) /rel\n',
   '<p><a href="https://x">a</a> <a href="http://x">b</a> <a href="mailto:a@b">c</a> <a href="ftp://x">d</a> /rel</p>\n');

ok('entity smuggling is decoded then escaped', '&#60;script&#62;alert(1)&#60;/script&#62;\n',
   function (h) { return h.indexOf('<script') < 0 && h.indexOf('&lt;script') >= 0; },
   'numeric refs must decode to text, then be escaped on output');

ok('hex entity smuggling is neutralised', '&#x3c;script&#x3e;\n',
   function (h) { return h.indexOf('<script') < 0 && h.indexOf('&lt;script') >= 0; });

ok('uppercase scheme javascript is filtered', '[x](JAVASCRIPT:alert(1))\n',
   function (h) { return h.toLowerCase().indexOf('javascript:') < 0; });

ok('title attribute is escaped', '[x](/u ""><script>")\n',
   function (h) { return h.indexOf('<script') < 0; });

// ---------------------------------------------------------------------------
// Edge cases
// ---------------------------------------------------------------------------

eq('empty input renders empty', '', '');
eq('blank-only input renders empty', '\n\n\n', '');
eq('CRLF line endings', 'a\r\nb\r\n\r\nc\r\n', '<p>a\nb</p>\n<p>c</p>\n');
eq('lone CR line endings', 'a\rb\r\rc\r', '<p>a\nb</p>\n<p>c</p>\n');
// CommonMark: NUL is replaced by U+FFFD before parsing
eq('NUL bytes are replaced', 'a\u0000b\n', '<p>a\uFFFDb</p>\n');
eq('heading with trailing hashes', '## Title ##\n', '<h2>Title</h2>\n');
eq('link label case-insensitive reference', '[Foo]\n\n[foo]: /u\n',
   '<p><a href="/u">Foo</a></p>\n');

print('regressions: ' + passed + ' passed, ' + failed + ' failed');
std.exit(failed ? 1 : 0);