import pytest
from ocrag.chunker import chunk_code


@pytest.fixture
def sample_python_code():
    return """def hello():
    print('Hello, World!')

class Test:
    def method(self):
        pass"""


@pytest.fixture
def sample_markdown_content():
    return """# Title

Paragraph 1

## Subtitle

Paragraph 2"""


def test_chunk_code_python(sample_python_code):
    chunks = chunk_code(sample_python_code, "test.py")
    assert len(chunks) > 0
    assert "def hello" in chunks[0]["text"]
    assert chunks[0]["metadata"]["language"] == "python"


def test_chunk_code_markdown(sample_markdown_content):
    chunks = chunk_code(sample_markdown_content, "test.md")
    assert len(chunks) > 0
    assert "# Title" in chunks[0]["text"]
    assert chunks[0]["metadata"]["language"] == "markdown"


def test_chunk_code_text():
    content = "a" * 10000  # Large content to test text fallback
    chunks = chunk_code(content, "large.txt")
    assert len(chunks) > 0
    assert chunks[0]["metadata"]["source_file"] == "large.txt"
    assert chunks[0]["metadata"]["language"] == "text"