scan4llm/tests/test_markdown.py

from __future__ import annotations

from pathlib import Path

from src.markdown.generator import (
    _validate_dollar_signs,
    _validate_latex_braces,
    _validate_left_right,
    generate_markdown,
    validate_markdown,
)
from src.ocr.paddle_engine import ParsedBlock


class TestValidateBraces:
    def test_balanced(self) -> None:
        assert _validate_latex_braces(r"a{b}c") == []

    def test_unbalanced_open(self) -> None:
        warnings = _validate_latex_braces(r"a{b")
        assert len(warnings) > 0

    def test_unbalanced_close(self) -> None:
        warnings = _validate_latex_braces(r"a}b")
        assert len(warnings) > 0

    def test_nested_balanced(self) -> None:
        assert _validate_latex_braces(r"{a{b}c}") == []

    def test_complex_formula(self) -> None:
        assert _validate_latex_braces(r"\frac{a}{b}") == []


class TestValidateDollar:
    def test_balanced_inline(self) -> None:
        assert _validate_dollar_signs("$a + b$") == []

    def test_balanced_display(self) -> None:
        assert _validate_dollar_signs(r"$$a + b$$") == []

    def test_unbalanced_inline(self) -> None:
        warnings = _validate_dollar_signs("$a + b")
        assert len(warnings) > 0

    def test_balanced_multiple(self) -> None:
        assert _validate_dollar_signs("$x$ and $y$") == []

    def test_unbalanced_multiple(self) -> None:
        warnings = _validate_dollar_signs("$x and $y$")
        assert len(warnings) > 0


class TestValidateLeftRight:
    def test_balanced(self) -> None:
        assert _validate_left_right(r"\left( x \right)") == []

    def test_unbalanced(self) -> None:
        warnings = _validate_left_right(r"\left( x")
        assert len(warnings) > 0


class TestValidateMarkdown:
    def test_valid(self) -> None:
        assert validate_markdown("text\n\n$$\na + b\n$$\n") == []

    def test_unbalanced_dollar(self) -> None:
        warnings = validate_markdown("$$\n$a\n$$\n")
        assert len(warnings) > 0

    def test_unbalanced_brace_in_formula(self) -> None:
        warnings = validate_markdown("$$\na{\nb\n$$\n")
        assert len(warnings) > 0


class TestGenerateMarkdown:
    def test_text_block(self) -> None:
        import tempfile

        blocks = [ParsedBlock(label="text", content="Hello world", bbox=(0, 0, 100, 20))]
        with tempfile.TemporaryDirectory() as tmp:
            p = generate_markdown(blocks, Path(tmp) / "output.md")
            assert p.exists()
            content = p.read_text()
            assert "Hello world" in content

    def test_formula_block(self) -> None:
        import tempfile

        blocks = [ParsedBlock(label="formula", content=r"a + b = c", bbox=(0, 0, 100, 20))]
        with tempfile.TemporaryDirectory() as tmp:
            p = generate_markdown(blocks, Path(tmp) / "output.md")
            content = p.read_text()
            assert "$$\n" in content
            assert "a + b = c" in content

    def test_title_block(self) -> None:
        import tempfile

        blocks = [ParsedBlock(label="paragraph_title", content="Section 1", bbox=(0, 0, 100, 20))]
        with tempfile.TemporaryDirectory() as tmp:
            p = generate_markdown(blocks, Path(tmp) / "output.md")
            content = p.read_text()
            assert "## Section 1" in content

    def test_no_escaped_latex(self) -> None:
        import tempfile

        blocks = [ParsedBlock(label="text", content=r"alpha \sim", bbox=(0, 0, 100, 20))]
        with tempfile.TemporaryDirectory() as tmp:
            p = generate_markdown(blocks, Path(tmp) / "output.md")
            content = p.read_text()
            assert r"\textbackslash" not in content
            assert r"alpha" in content