tests/test_dictation.py

import pytest

from tutor.services.dictation import normalize_words, score_dictation


def test_normalize_folds_case_punctuation_contractions_spelling() -> None:
    assert normalize_words("Hello, World!") == ["hello", "world"]
    assert normalize_words("I don't know") == ["i", "do", "not", "know"]
    assert normalize_words("The colour of the centre") == ["the", "color", "of", "the", "center"]


def test_perfect_despite_legitimate_variants() -> None:
    result = score_dictation("I don't like the colour.", "I do not like the color")
    assert result.is_perfect
    assert result.wer == 0.0
    assert result.substitutions == result.deletions == result.insertions == 0


def test_catches_phonetic_confusion() -> None:
    result = score_dictation("I saw a sheep on the hill", "I saw a ship on the hill")
    assert result.substitutions == 1
    subs = [(op.ref_word, op.hyp_word) for op in result.ops if op.op == "substitute"]
    assert subs == [("sheep", "ship")]


def test_deletion_and_alignment_order() -> None:
    result = score_dictation("the quick brown fox jumps", "the quik brown fox")
    assert result.substitutions == 1  # quick/quik
    assert result.deletions == 1  # jumps missing
    assert [op.op for op in result.ops] == ["equal", "substitute", "equal", "equal", "delete"]


def test_insertion_is_flagged() -> None:
    result = score_dictation("the cat sat", "the big cat sat")
    assert result.insertions == 1
    inserted = [op.hyp_word for op in result.ops if op.op == "insert"]
    assert inserted == ["big"]


def test_empty_hypothesis_is_all_deletions() -> None:
    result = score_dictation("hello world", "   ")
    assert result.wer == 1.0
    assert result.deletions == 2
    assert all(op.op == "delete" for op in result.ops)


def test_empty_reference_raises() -> None:
    with pytest.raises(ValueError, match="reference is empty"):
        score_dictation("!!!", "anything")


def test_render_feedback_marks_errors() -> None:
    from tutor.app.main import render_dictation_feedback

    perfect = render_dictation_feedback(score_dictation("the cat sat", "the cat sat"))
    assert "100% correct 🎉" in perfect

    marked = render_dictation_feedback(score_dictation("I saw a sheep", "I saw a ship"))
    assert "**sheep**" in marked  # correct word in bold
    assert "~~ship~~" in marked  # learner's wrong version struck through