wiki-kb/tests/test_html2md.py

"""Tests for HTML -> markdown conversion."""

from __future__ import annotations

from wikikb import html2md

_HTML = """
<div class="mw-parser-output">
<div class="mw-editsection">[edit]</div>
<p>A <b>count</b> can hold up to 4 counties.</p>
<table><tr><th>Trait</th><th>Diplomacy</th></tr>
<tr><td>Genius</td><td>+5</td></tr></table>
<div class="navbox">navigation</div>
</div>
"""


def test_clean_html_strips_chrome_keeps_content() -> None:
    cleaned = html2md.clean_html(_HTML)
    assert "editsection" not in cleaned
    assert "navbox" not in cleaned
    assert "count" in cleaned


def test_clean_html_without_parser_output_wrapper() -> None:
    cleaned = html2md.clean_html("<p>bare</p>")
    assert "bare" in cleaned


def test_html_to_markdown_preserves_table_numbers() -> None:
    md = html2md.html_to_markdown(_HTML)
    assert "Genius" in md
    assert "+5" in md
    assert "\n\n\n" not in md  # blank lines collapsed


def test_build_document_frontmatter_and_quote_escaping() -> None:
    doc = html2md.build_document(
        title='War of the "Roses"',
        url="https://wiki.example/War",
        revid=42,
        categories=["Warfare", "1.19"],
        html="<p>hello</p>",
    )
    assert doc.startswith("---\n")
    assert 'title: "War of the \\"Roses\\""' in doc
    assert "revid: 42" in doc
    assert 'categories: ["Warfare", "1.19"]' in doc
    assert "# War of the" in doc
    assert "hello" in doc


# A cut-down Fandom PortableInfobox: labels are <h3> siblings of their values, not
# table cells, and the tier variants are inlined behind tabs.
_INFOBOX = """
<div class="mw-parser-output">
<aside class="portable-infobox">
  <h2 class="pi-item pi-title">Common longsword</h2>
  <section class="pi-item pi-group">
    <h2 class="pi-item pi-header">Information</h2>
    <div class="pi-item pi-data"><h3 class="pi-data-label">Classification</h3>
      <div class="pi-data-value">Weapons</div></div>
  </section>
  <h2 class="pi-item pi-header">Attributes</h2>
  <section class="pi-item pi-panel wds-tabber">
    <div class="wds-tabs__wrapper"><ul class="wds-tabs">
      <li class="wds-tabs__tab"><div class="wds-tabs__tab-label">I</div></li>
      <li class="wds-tabs__tab"><div class="wds-tabs__tab-label">II</div></li>
    </ul></div>
    <div class="wds-tab__content">
      <section class="pi-item pi-smart-group">
        <section class="pi-smart-group-head">
          <h3 class="pi-smart-data-label">Stab dmg.</h3>
          <h3 class="pi-smart-data-label">Slash dmg.</h3>
        </section>
        <section class="pi-smart-group-body">
          <div class="pi-smart-data-value">74</div>
          <div class="pi-smart-data-value">71</div>
        </section>
      </section>
    </div>
    <div class="wds-tab__content">
      <section class="pi-item pi-smart-group">
        <section class="pi-smart-group-head">
          <h3 class="pi-smart-data-label">Stab dmg.</h3>
          <h3 class="pi-smart-data-label">Slash dmg.</h3>
        </section>
        <section class="pi-smart-group-body">
          <div class="pi-smart-data-value">92</div>
          <div class="pi-smart-data-value">88</div>
        </section>
      </section>
    </div>
  </section>
</aside>
<p>The lead sentence.</p>
</div>
"""


def test_infobox_becomes_a_table_with_labels_bound_to_values() -> None:
    md = html2md.html_to_markdown(_INFOBOX)
    # The point of the transform: not merely that "74" survives, but that it stays
    # attached to "Stab dmg." Plain markdownify emits headings then orphaned numbers.
    assert "| Stab dmg. | 74 |" in md
    assert "| Slash dmg. | 71 |" in md
    assert "| Classification | Weapons |" in md


def test_infobox_tier_tabs_are_labelled() -> None:
    md = html2md.html_to_markdown(_INFOBOX)
    # Without the tab label, both tiers' stat blocks are indistinguishable.
    assert "## Infobox: Attributes (I)" in md
    assert "## Infobox: Attributes (II)" in md
    assert "| Stab dmg. | 92 |" in md


def test_infobox_lead_text_gets_its_own_section() -> None:
    md = html2md.html_to_markdown(_INFOBOX)
    # The lead paragraph must not be filed under the last infobox group.
    assert md.index("## Overview") < md.index("The lead sentence.")


def test_infobox_without_a_group_header_is_named_plainly() -> None:
    html = (
        '<aside class="portable-infobox"><div class="pi-item pi-data">'
        '<h3 class="pi-data-label">ID</h3><div class="pi-data-value">4</div></div></aside>'
    )
    md = html2md.html_to_markdown(html)
    assert "## Infobox\n" in md
    assert "| ID | 4 |" in md


def test_empty_infobox_is_dropped() -> None:
    # Theme/dummy infoboxes hold no data and would otherwise leave a stray heading.
    md = html2md.html_to_markdown(
        '<div><aside class="portable-infobox pi-theme-dummy"></aside><p>body</p></div>'
    )
    assert "Infobox" not in md
    assert "body" in md


def test_mismatched_smart_group_keeps_the_pairs_that_line_up() -> None:
    html = (
        '<aside class="portable-infobox"><section class="pi-smart-group">'
        '<section class="pi-smart-group-head"><h3 class="pi-smart-data-label">A</h3>'
        '<h3 class="pi-smart-data-label">B</h3></section>'
        '<section class="pi-smart-group-body"><div class="pi-smart-data-value">1</div>'
        "</section></section></aside>"
    )
    md = html2md.html_to_markdown(html)
    assert "| A | 1 |" in md
    assert "| B |" not in md


def test_incomplete_pi_data_is_skipped_without_losing_the_rest() -> None:
    # A hand-edited infobox row can carry a label with no value. Skip that row and
    # keep going rather than emitting a label bound to nothing.
    html = (
        '<aside class="portable-infobox">'
        '<div class="pi-item pi-data"><h3 class="pi-data-label">Orphan</h3></div>'
        '<div class="pi-item pi-data"><h3 class="pi-data-label">Weight</h3>'
        '<div class="pi-data-value">3.9</div></div>'
        "</aside>"
    )
    md = html2md.html_to_markdown(html)
    assert "| Weight | 3.9 |" in md
    assert "Orphan" not in md


def test_page_tabber_switcher_is_stripped() -> None:
    # It links to the other game's page; keeping it would put a bare "KCD" link at
    # the top of every dual-game article.
    html = (
        '<div class="mw-parser-output">'
        '<div class="pagetabber-absolute"><a href="/wiki/Alchemy">KCD</a></div>'
        "<p>real content</p></div>"
    )
    md = html2md.html_to_markdown(html)
    assert "real content" in md
    assert "/wiki/Alchemy" not in md