rose-ash/shared/sx/tests/test_html_to_sx.py

"""Unit tests for html_to_sx converter."""
from __future__ import annotations

import pytest

from shared.sx.html_to_sx import html_to_sx


class TestBasicElements:
    def test_simple_paragraph(self):
        assert html_to_sx("<p>Hello</p>") == '(p "Hello")'

    def test_with_class(self):
        assert html_to_sx('<p class="intro">Hi</p>') == '(p :class "intro" "Hi")'

    def test_multiple_attrs(self):
        result = html_to_sx('<a href="u" class="link">click</a>')
        assert result == '(a :href "u" :class "link" "click")'

    def test_nested_inline(self):
        result = html_to_sx("<p>Hello <em>world</em></p>")
        assert result == '(p "Hello " (em "world"))'

    def test_deeply_nested(self):
        result = html_to_sx("<div><p><strong>bold</strong></p></div>")
        assert result == '(div (p (strong "bold")))'


class TestVoidElements:
    def test_br(self):
        assert html_to_sx("<br>") == "(br)"

    def test_img(self):
        result = html_to_sx('<img src="a.jpg" alt="pic">')
        assert result == '(img :src "a.jpg" :alt "pic")'

    def test_hr(self):
        assert html_to_sx("<hr>") == "(hr)"

    def test_input(self):
        result = html_to_sx('<input type="text" value="hi">')
        assert result == '(input :type "text" :value "hi")'


class TestBooleanAttrs:
    def test_checked(self):
        result = html_to_sx('<input type="checkbox" checked>')
        assert result == '(input :type "checkbox" :checked true)'

    def test_disabled(self):
        result = html_to_sx('<button disabled>No</button>')
        assert result == '(button :disabled true "No")'

    def test_controls(self):
        result = html_to_sx('<video controls></video>')
        assert result == '(video :controls true)'


class TestTopLevel:
    def test_multiple_top_level(self):
        result = html_to_sx("<p>A</p><p>B</p>")
        assert result == '(<> (p "A") (p "B"))'

    def test_single_top_level(self):
        result = html_to_sx("<p>Only</p>")
        assert result == '(p "Only")'

    def test_text_only(self):
        result = html_to_sx("just text")
        assert result == '"just text"'

    def test_empty(self):
        assert html_to_sx("") == '""'

    def test_whitespace_only(self):
        assert html_to_sx("   \n  ") == '""'


class TestWhitespace:
    def test_root_whitespace_stripped(self):
        result = html_to_sx("\n<p>A</p>\n<p>B</p>\n")
        assert result == '(<> (p "A") (p "B"))'


class TestEntities:
    def test_amp(self):
        result = html_to_sx("<p>A &amp; B</p>")
        assert result == '(p "A & B")'

    def test_lt_gt(self):
        result = html_to_sx("<p>&lt;tag&gt;</p>")
        assert result == '(p "<tag>")'

    def test_nbsp(self):
        result = html_to_sx("<p>hello&nbsp;world</p>")
        assert result == '(p "hello\u00a0world")'


class TestEscaping:
    def test_quotes_in_text(self):
        result = html_to_sx('<p>He said "hello"</p>')
        assert result == '(p "He said \\"hello\\"")'

    def test_backslash_in_text(self):
        result = html_to_sx("<p>a\\b</p>")
        assert result == '(p "a\\\\b")'

    def test_quotes_in_attr(self):
        # Attribute values with quotes get escaped
        result = html_to_sx('<div title="a&quot;b">x</div>')
        assert result == '(div :title "a\\"b" "x")'


class TestComments:
    def test_comment_stripped(self):
        result = html_to_sx("<!-- comment --><p>hi</p>")
        assert result == '(p "hi")'


class TestMixedContent:
    def test_caption_with_link(self):
        result = html_to_sx('Photo by <a href="https://x.com">Author</a>')
        assert result == '(<> "Photo by " (a :href "https://x.com" "Author"))'

    def test_caption_plain_text(self):
        result = html_to_sx("Figure 1")
        assert result == '"Figure 1"'


class TestRoundtrip:
    """html_to_sx → parse → render should produce equivalent HTML."""

    def _roundtrip(self, html_in: str) -> str:
        from shared.sx.parser import parse
        from shared.sx.html import render
        sx_src = html_to_sx(html_in)
        expr = parse(sx_src)
        return render(expr)

    def test_simple(self):
        assert self._roundtrip("<p>Hello</p>") == "<p>Hello</p>"

    def test_nested(self):
        assert self._roundtrip("<p>A <em>B</em> C</p>") == "<p>A <em>B</em> C</p>"

    def test_void(self):
        assert self._roundtrip('<img src="a.jpg" alt="">') == '<img src="a.jpg" alt="">'

    def test_link(self):
        html = '<a href="https://example.com">click</a>'
        assert self._roundtrip(html) == html

    def test_entities_roundtrip(self):
        # Entities get decoded by parser, then re-escaped by render
        assert self._roundtrip("<p>A &amp; B</p>") == "<p>A &amp; B</p>"

    def test_multi_block(self):
        html = "<p>A</p><p>B</p>"
        assert self._roundtrip(html) == html