epub2md/tests/test_qa_regressions.py

"""Regression tests for defects found in the independent QA run (.qa/QA_REPORT.md, IDs QA2-…)."""

import re

from conftest import Book, Doc


def _pipe_rows(text: str) -> list[str]:
    return [ln for ln in text.splitlines() if ln.startswith("| ")]


def test_pipe_in_inline_code_does_not_split_gfm_table(convert):
    """QA2-01: GFM splits table cells at every unescaped '|', also inside code spans and link targets."""
    body = ("<h1>Operatoren</h1><table><thead><tr><th>Operator</th><th>Bedeutung</th></tr></thead><tbody>"
            "<tr><td><code>||</code></td><td>logisches ODER</td></tr>"
            "<tr><td><code>a | b</code></td><td><a href=\"https://example.org/a|b\">Link</a></td></tr>"
            "</tbody></table>")
    run = convert(Book(docs=[Doc("c.xhtml", body)]))
    text = run.chapter("001")
    for row in _pipe_rows(text):
        cells = re.split(r"(?<!\\)\|", row.strip()[1:-1])
        assert len(cells) == 2, f"Zeile zerfällt in GFM in {len(cells)} Zellen: {row}"
    assert "||" in text and "a | b" in text and "logisches ODER" in text


NOTE_WITH_CODE = ('<h1>K</h1><p>Text<a epub:type="noteref" href="#n1" id="r1">1</a>.</p>'
                  '<aside epub:type="footnote" id="n1"><p>Note mit Code:</p><pre>code()\n    weiter()</pre></aside>'
                  '<p>Zweite<a epub:type="noteref" href="#n2" id="r2">2</a>.</p>'
                  '<aside epub:type="footnote" id="n2"><pre>nur_code()</pre></aside>'
                  "<p>Danach kommt normaler Text.</p>")


def _fences_closed(text: str) -> bool:
    """Every opening ``` fence has a closing line that consists only of backticks."""
    open_fence = None
    for line in text.splitlines():
        s = line.strip()
        if open_fence is None:
            m = re.match(r"(`{3,})", s)
            if m:
                open_fence = m.group(1)
        elif re.fullmatch(r"`{3,}", s) and len(s) >= len(open_fence):
            open_fence = None
    return open_fence is None


def test_backlink_after_code_block_does_not_break_fence_commonmark(convert):
    """QA2-02: the ↩ backlink was appended to the closing ``` line, leaving the fence open."""
    run = convert(Book(docs=[Doc("c.xhtml", NOTE_WITH_CODE)]), "--format", "commonmark")
    text = run.chapter("001")
    assert _fences_closed(text), text
    assert not re.search(r"\*\*\d+[:.]\*\* ```", text), "Fence hinter Notennummer ist kein Codeblock"
    assert "[↩](#fnref-001)" in text


def test_backlink_after_code_block_does_not_break_fence_end_of_book(convert):
    run = convert(Book(docs=[Doc("c.xhtml", NOTE_WITH_CODE)]), "--footnotes", "end-of-book")
    notes = run.chapter("999")
    assert _fences_closed(notes), notes
    assert not re.search(r"\*\*\d+[:.]\*\* ```", notes), "Fence hinter Notennummer ist kein Codeblock"
    assert "[↩](001-k.md#fnref-001)" in notes


NOTE_IN_HTML_TABLE = ('<h1>K</h1><table><tr><td colspan="2">Zelle<a epub:type="noteref" href="#n1" id="r1">1</a></td></tr>'
                      "<tr><td>a</td><td>b</td></tr></table>"
                      '<p>Text<a epub:type="noteref" href="#n1" id="r1b">1</a>.</p>'
                      '<aside epub:type="footnote" id="n1"><p>Die Note.</p></aside>')


def test_note_referenced_from_html_table_and_text_has_target(convert):
    """QA2-03: GFM: the HTML-table reference pointed to #fn-001, but only a [^fn-001] definition existed."""
    run = convert(Book(docs=[Doc("c.xhtml", NOTE_IN_HTML_TABLE)]))
    text = run.chapter("001")
    assert 'href="#fn-001"' in text and "Text[^fn-001]." in text
    assert re.search(r'id="fn-001"', text), text
    assert run.report["output_link_check"]["broken_count"] == 0


def _html_table(text: str) -> str:
    return text[text.index("<table"): text.index("</table>")]


def test_end_of_book_reference_inside_html_table_is_html(convert):
    """QA2-04: a Markdown link inside an HTML block is not rendered (shows up as literal text)."""
    run = convert(Book(docs=[Doc("c.xhtml", NOTE_IN_HTML_TABLE)]), "--footnotes", "end-of-book")
    table = _html_table(run.chapter("001"))
    assert "](" not in table, table
    assert '<a href="999-anmerkungen.md#fn-001">1</a>' in table


def test_inline_svg_inside_html_table_is_html_img(convert):
    """QA2-05: inline SVG in a complex table was written as Markdown image inside the HTML block."""
    body = ('<h1>T</h1><table><tr><th colspan="2">Kopf</th></tr><tr><td>A</td><td>'
            '<svg xmlns="http://www.w3.org/2000/svg" width="5" height="5" aria-label="Kreis"><circle r="2"/></svg>'
            "</td></tr></table>")
    run = convert(Book(docs=[Doc("c.xhtml", body)]))
    table = _html_table(run.chapter("001"))
    assert "![" not in table, table
    assert re.search(r'<img src="\.\./media/images/\d{4}-c-svg\.svg" alt="Kreis"/>', table), table
    assert run.report["output_link_check"]["broken_count"] == 0


def test_multiple_blank_lines_inside_code_are_kept(convert):
    """QA2-06: runs of blank lines inside code blocks were collapsed to one (PEP 8: two between functions)."""
    code = "def f():\n    pass\n\n\nclass K:\n    pass\n\n\n\nx = 1"
    body = f"<h1>K</h1><p>Vorher</p><pre>{code}</pre><p>Nachher</p>"
    text = convert(Book(docs=[Doc("c.xhtml", body)])).chapter("001")
    assert f"```python\n{code}\n```" in text, text
    assert "\n\n\n" not in text.replace(code, "")  # outside code blank lines are still normalized