fix(research): typeset math in visual reports with the shared renderer (#6929)

tests/test_visual_report_math.py

@@ -0,0 +1,76 @@
+"""Research reports preserve TeX for the shared, vendored math renderer."""
+from bs4 import BeautifulSoup
+import pytest
+
+from src.visual_report import _md_to_html, generate_visual_report
+
+
+@pytest.mark.parametrize("source,tex,display", [
+    (r"Value $x_i^2$.", "x_i^2", "false"),
+    (r"Value \(x_i^2\).", "x_i^2", "false"),
+    (r"Value $$\sum_i x_i$$.", r"\sum_i x_i", "true"),
+    (r"Value \[\sum_i x_i\].", r"\sum_i x_i", "true"),
+    ("$$a_i\n\nb_i$$", "a_i\n\nb_i", "true"),
+    ("\\[a_i\n\nb_i\\]", "a_i\n\nb_i", "true"),
+])
+def test_math_source_survives_markdown(source, tex, display):
+    soup = BeautifulSoup(_md_to_html(source), "html.parser")
+    pending = soup.select(".ody-math-pending")
+    assert len(pending) == 1
+    assert pending[0].get_text() == tex
+    assert pending[0]["data-display"] == display
+    assert pending[0].find("em") is None
+
+
+@pytest.mark.parametrize("source", [
+    "`$x_i$`", "```text\n$x_i$\n```", "~~~text\n$x_i$\n~~~",
+    "    $x_i$", "<code>$x_i$</code>", "<pre>$x_i$</pre>",
+    "Prices range from $5 to $10.", "An escaped \\$5 costs less than $10.",
+])
+def test_code_and_currency_are_not_math(source):
+    soup = BeautifulSoup(_md_to_html(source), "html.parser")
+    assert not soup.select(".ody-math-pending")
+    assert "$" in soup.get_text()
+
+
+def test_html_attributes_are_not_rewritten_as_math():
+    soup = BeautifulSoup(_md_to_html('<a title="$literal$">Value $x_i$</a>'), "html.parser")
+    assert soup.a["title"] == "$literal$"
+    assert soup.select_one("a .ody-math-pending").get_text() == "x_i"
+
+
+def test_link_destinations_stay_literal_while_link_labels_can_contain_math():
+    soup = BeautifulSoup(_md_to_html('[$x_i$](https://example.com/$literal$)'), "html.parser")
+    assert soup.a["href"] == "https://example.com/$literal$"
+    assert soup.select_one("a .ody-math-pending").get_text() == "x_i"
+
+
+def test_math_source_is_inert_and_report_sanitizer_stays_active():
+    soup = BeautifulSoup(_md_to_html(
+        r'$\text{<img src=x onerror="alert(1)">}$'
+        '\n\n<script>alert(2)</script><a href="javascript:alert(3)">link</a>'
+    ), "html.parser")
+    assert soup.select_one(".ody-math-pending").get_text() == r'\text{<img src=x onerror="alert(1)">}'
+    assert soup.find("img") is None
+    assert soup.find("script") is None
+    assert not soup.select('a[href^="javascript:"]')
+
+
+def test_heading_math_reaches_body_and_toc_without_changing_anchor_contract():
+    soup = BeautifulSoup(generate_visual_report(
+        "Formula report", "# Formula report\n\n## Formula $x_i$\n\nValue $x_i$.",
+        session_id="rp-math",
+    ), "html.parser")
+    link = soup.select_one(".toc-sidebar nav a")
+    heading = soup.find(id=link["href"][1:])
+    assert heading["id"] == "formula-xi"
+    for node in (link, heading):
+        assert node.select_one(".ody-math-pending").get_text() == "x_i"
+    assert soup.select_one("#mobile-section-select option").get_text() == "Formula x_i"
+
+
+def test_unclosed_display_math_stays_readable():
+    soup = BeautifulSoup(_md_to_html("$$unfinished\n\nA later paragraph."), "html.parser")
+    assert not soup.select(".ody-math-pending")
+    assert "unfinished" in soup.get_text()
+    assert "A later paragraph." in soup.get_text()