wapdown/tests/test_wml.py

"""Deck-level WML emission: prolog, head, block mapping, and validity."""
from __future__ import annotations

import re
from xml.etree import ElementTree

import pytest

from wapdown.renderers.wml.renderer import WML_PROLOG


TWO_SECTIONS = "Intro.\n\n{.card One}\nFirst.\n\n{.card Two}\nSecond.\n"


def parse(markup: str) -> ElementTree.Element:
    """Parse the deck, proving it is well-formed XML.

    ElementTree honours the DOCTYPE declaration without trying to fetch the
    external DTD, which matters because wapforum.org has been gone for years.
    """
    return ElementTree.fromstring(markup)


class TestDocumentShape:
    def test_prolog_and_doctype(self, render):
        markup = render("Body.\n")
        assert markup.startswith(WML_PROLOG)
        assert 'PUBLIC "-//WAPFORUM//DTD WML 1.3//EN"' in markup

    def test_root_element_is_wml(self, render):
        assert parse(render("Body.\n")).tag == "wml"

    def test_output_is_well_formed(self, render):
        parse(render(TWO_SECTIONS))

    def test_no_carriage_returns(self, render):
        # WML is XML served over HTTP, not a DOS text file.
        assert "\r" not in render(TWO_SECTIONS)

    def test_ends_with_a_single_newline(self, render):
        markup = render("Body.\n")
        assert markup.endswith("</wml>\n")


class TestCardContentModel:
    """WML 1.3 declares <card> as (onevent*, timer?, (do | p)*)."""

    def test_card_children_are_only_do_and_p(self, render):
        for card in parse(render(TWO_SECTIONS)).findall("card"):
            assert {child.tag for child in card} <= {"do", "p"}

    def test_table_is_wrapped_in_a_paragraph(self, render):
        # Regression: <table> was emitted as a direct child of <card>, which
        # the DTD does not permit -- it is flow content and belongs in a <p>.
        markup = render("| A | B |\n| --- | --- |\n| 1 | 2 |\n")
        assert "<p><table" in markup
        card = parse(markup).find("card")
        assert [child.tag for child in card] == ["p"]
        assert card.find("p/table") is not None

    def test_do_elements_precede_content(self, render):
        card = parse(render(TWO_SECTIONS, menu=False)).find("card")
        tags = [child.tag for child in card]
        assert tags.index("do") < tags.index("p")


class TestBlockMapping:
    @pytest.mark.parametrize(
        ("source", "expected"),
        [
            ("Just text.\n", "<p>Just text.</p>"),
            ("- item\n", "<p>- item</p>"),
            ("> quoted\n", "<p><i>quoted</i></p>"),
        ],
    )
    def test_blocks(self, render, source, expected):
        assert expected in render(source)

    def test_thematic_break_draws_nothing(self, render):
        # WML has no rule element -- its whole %layout entity is <br> -- and
        # a row of ASCII dashes is noise on a small screen.
        assert "---" not in render("Before.\n\n---\n\nAfter.\n")

    def test_first_heading_is_emphasised_larger(self, render):
        markup = render("# Title\n\n## Sub\n")
        assert "<p><b><big>Title</big></b></p>" in markup
        assert "<p><b>Sub</b></p>" in markup

    def test_ordered_list_is_numbered_manually(self, render):
        # WML has no list element, so the numbers become literal text.
        markup = render("1. first\n2. second\n")
        assert "<p>1. first</p>" in markup
        assert "<p>2. second</p>" in markup

    def test_ordered_list_numbering_restarts_after_a_paragraph(self, render):
        markup = render("1. a\n\nBreak.\n\n1. b\n")
        assert markup.count("<p>1. ") == 2

    def test_code_block_lines_join_with_breaks(self, render):
        markup = render("```\none\ntwo\n```\n")
        assert "<p>one<br/>two</p>" in markup

    def test_code_block_content_is_escaped(self, render):
        assert "&lt;wml&gt;" in render("```\n<wml>\n```\n")

    def test_table_header_row_is_bold(self, render):
        markup = render("| A | B |\n| --- | --- |\n| 1 | 2 |\n")
        assert "<td><b>A</b></td>" in markup
        assert "<td>1</td>" in markup

    def test_ragged_table_rows_are_padded(self, render):
        markup = render("| A | B | C |\n| --- | --- | --- |\n| 1 |\n")
        assert '<table columns="3">' in markup
        assert markup.count("<td>") % 3 == 0


class TestHead:
    def test_no_head_by_default(self, render):
        assert "<head>" not in render("Body.\n")

    def test_cache_control(self, render):
        markup = render("Body.\n", cache_control=0)
        assert '<meta http-equiv="Cache-Control" content="max-age=0"/>' in markup
        parse(markup)

    def test_access_control(self, render):
        markup = render("Body.\n", access_domain="example.com", access_path="/wap")
        assert '<access domain="example.com" path="/wap"/>' in markup
        parse(markup)

    def test_head_precedes_template_and_cards(self, render):
        markup = render(TWO_SECTIONS, cache_control=60)
        assert [child.tag for child in parse(markup)][:2] == ["head", "template"]


class TestTitles:
    def test_explicit_title_reaches_the_menu_card(self, render):
        assert '<card id="menu" title="Deck">' in render(TWO_SECTIONS, title="Deck")

    def test_single_card_is_untitled_without_an_explicit_title(self, render):
        # The document's own heading already renders into the card body.
        assert '<card id="card1">' in render("# Heading\n\nBody.\n")

    def test_title_is_escaped(self, render):
        markup = render(TWO_SECTIONS, title='Fish & "Chips"')
        assert 'title="Fish &amp; &quot;Chips&quot;"' in markup
        parse(markup)

    def test_dollar_in_a_title_is_doubled(self, render):
        assert 'title="$$5"' in render(TWO_SECTIONS, title="$5")


class TestMultiDeck:
    def test_one_file_per_section(self, render_output):
        output = render_output(TWO_SECTIONS, deck_per_card=True)
        assert sorted(doc.name for doc in output.documents) == [
            "index.wml",
            "one.wml",
            "two.wml",
        ]

    def test_menu_links_across_files(self, render_output):
        docs = {d.name: d.text for d in render_output(TWO_SECTIONS, deck_per_card=True).documents}
        assert '<a href="one.wml#card1">One</a>' in docs["index.wml"]

    def test_links_within_the_same_file_stay_local(self, render_output):
        docs = {d.name: d.text for d in render_output(TWO_SECTIONS, deck_per_card=True).documents}
        assert "index.wml#" not in docs["one.wml"].split("</template>")[-1]

    def test_every_file_is_well_formed(self, render_output):
        for doc in render_output(TWO_SECTIONS, deck_per_card=True).documents:
            parse(doc.text)

    def test_single_section_document_stays_one_file(self, render_output):
        output = render_output("# Title\n\nBody.\n", deck_per_card=True)
        assert len(output.documents) == 1
        assert output.documents[0].name is None

    def test_duplicate_section_titles_get_distinct_filenames(self, render_output):
        source = "Intro.\n\n{.card Same}\nA.\n\n{.card Same}\nB.\n"
        names = [d.name for d in render_output(source, deck_per_card=True).documents]
        assert len(set(names)) == len(names)