tests/test_parser.py

import pytest

from app.services.parser import EmptyDocument, UnsupportedFormat, page_for_offset, parse


def test_parse_text_basic():
    doc = parse("note.txt", b"Hello world\n\nSecond paragraph.")
    assert "Hello world" in doc.text
    assert doc.source_format == "text"
    assert doc.char_count > 0
    assert doc.page_count == 1


def test_parse_empty_raises():
    with pytest.raises(EmptyDocument):
        parse("blank.txt", b"   \n  ")


def test_unknown_format_raises():
    with pytest.raises(UnsupportedFormat):
        parse("mystery.xyz", b"data")


def test_page_for_offset_returns_correct_page():
    from app.services.parser import PageSpan

    pages = [PageSpan(1, 0, 50), PageSpan(2, 50, 120)]
    assert page_for_offset(pages, 10) == 1
    assert page_for_offset(pages, 75) == 2
    assert page_for_offset(pages, 500) == 2