tests/test_align.py

"""Tests for pairwise alignment (dee/core/align.py)."""
import pytest

from dee.core import align as A


def test_identical_global():
    r = A.align("ACGTACGT", "ACGTACGT")
    assert r["identity"] == 100.0
    assert r["gaps"] == 0
    assert r["aligned_a"] == r["aligned_b"] == "ACGTACGT"
    assert r["mode"] == "global"


def test_single_mismatch():
    r = A.align("ACGTACGT", "ACGTTCGT")
    assert r["gaps"] == 0
    assert r["length"] == 8
    assert r["matches"] == 7
    assert 80.0 < r["identity"] < 100.0
    assert r["midline"].count(".") == 1   # one mismatch column


def test_insertion_makes_a_gap():
    # b has an extra base → a gap appears in the alignment
    r = A.align("ACGTACGT", "ACGTAACGT")
    assert r["gaps"] >= 1
    assert "-" in r["aligned_a"]
    # all original bases still match where aligned
    assert r["matches"] == 8


def test_local_finds_embedded_region():
    a = "GGGGGGACGTACGTGGGGGG"
    b = "TTTTACGTACGTTTTT"
    r = A.align(a, b, mode="local")
    assert r["mode"] == "local"
    # the shared ACGTACGT core should align at 100% identity, no gaps
    assert "ACGTACGT" in r["aligned_a"].replace("-", "")
    assert r["identity"] == 100.0
    assert r["gaps"] == 0


def test_whitespace_and_case_normalised():
    r = A.align("ac gt\nac gt", "ACGTACGT")
    assert r["identity"] == 100.0


def test_empty_raises():
    with pytest.raises(ValueError):
        A.align("", "ACGT")
    with pytest.raises(ValueError):
        A.align("ACGT", "   ")


def test_oversize_raises():
    big = "A" * 2000
    with pytest.raises(ValueError):
        A.align(big, big)   # 4,000,000 cells > MAX_CELLS


def test_midline_lengths_consistent():
    r = A.align("ACGTACGTAC", "ACGAACGTTC")
    assert len(r["aligned_a"]) == len(r["aligned_b"]) == len(r["midline"]) == r["length"]