From 190a75a0e103a0724f08be96e74c92a97c8b28b5 Mon Sep 17 00:00:00 2001 From: Vuong Hoang Date: Mon, 28 Sep 2026 10:37:36 -0700 Subject: [PATCH] fix(docs): a posted doc cannot run script on the Booth's origin Found by design-dev's impeccable run and confirmed at source. Python-Markdown passes raw HTML through, and doc.html and booth.html render the result |safe. A \n\nafter", + "inline here", + "", + "| a |\n|---|\n| |", +]) +def test_raw_html_in_a_doc_is_text_never_markup(src): + html, s = _scan(src) + assert not {"script", "img", "iframe", "svg"} & set(s.tags), (s.tags, html) + assert "<" in html + + +def test_a_quoted_pre_is_shown_not_opened_and_the_doc_goes_on(): + html, s = _scan("a contract that says 
 opens one\n\nnext paragraph")
+    assert "pre" not in s.tags, html
+    assert "
next paragraph
" in html
+
+
+def test_fenced_code_is_still_a_code_block():
+    """Positive control: escaping raw HTML must not cost the code block."""
+    html, s = _scan("```html\n\n```")
+    assert s.tags == ["pre", "code"], html
+    assert "<script>" in html
+
+
+def test_inline_html_a_doc_meant_is_now_literal_text():
+    """The declared cost of the ruling:  and 
 show as tags."""
+    html, s = _scan("H2O")
+    assert "sub" not in s.tags and "<sub>" in html
+
+
+@pytest.mark.parametrize("dest", [
+    "[x](javascript:alert(1))",
+    "[x](JaVaScRiPt:alert(1))",
+    "[x](javascript:alert(1))",
+    "[x](javascript:alert(1))",
+    "[x](javascript:alert`1`)",
+    "[x](java	script:alert`1`)",
+    "[x](javascript:alert`1`)",
+    "[x]()",
+    "[x](\x01javascript:alert`1`)",
+    "[r]: javascript:alert`1`\n\n[go][r]",
+    "[x](data:text/html,)",
+])
+def test_a_link_that_would_run_code_keeps_its_text_and_loses_its_href(dest):
+    html, s = _scan(dest)
+    assert not any(_navigates_to_script(h) for h in s.hrefs), (s.hrefs, html)
+    assert "a" in s.tags, html                    # the words are still there
+
+
+def test_ordinary_links_survive():
+    html, s = _scan("[a](https://example.com) [b](http://x.test/p) [c](other.md) [d](#frag)")
+    assert s.hrefs == ["https://example.com", "http://x.test/p", "other.md", "#frag"], html
+
+
+# Markdown reads `\\` as an escaped backslash, so FOUR in the source put two
+# in the href: `\\evil.test`, which a browser reads as `//evil.test`.
+@pytest.mark.parametrize("dest", ["[x](//evil.test/p)", "[x](/\\evil.test/p)", "[x](\\\\\\\\evil.test/p)"])
+def test_a_link_that_leaves_the_origin_by_backslash_is_refused(dest):
+    """heid bug-hunt, groa: `is_safe_href` refused `//host` but not its
+    backslash twin — a browser reads `\\` as `/` in an http(s) URL, so
+    `/\\evil.test` is `//evil.test`."""
+    html, s = _scan(dest)
+    assert s.hrefs == [], (s.hrefs, html)
+
+
+def test_a_renderer_failure_costs_the_doc_its_formatting_never_the_page(monkeypatch):
+    """heid bug-hunt, 3 of 4 arms: nothing bounded the render. A doc that
+    makes Python-Markdown raise — deep nesting, or an upgrade that renames the
+    processors this module deregisters — must fall back to escaped raw text,
+    never raise out of the page."""
+    import booth.items as items_mod
+
+    def boom():
+        raise RecursionError("too deep")
+    monkeypatch.setattr(items_mod, "_markdown_renderer", boom)
+    assert items_mod.render_doc("# t\n\n", "markdown") == \
+        ("# t\n\n", False)