"""normalize_title: de-shout / de-clutter video titles without mangling non-English text."""
from app.titles import normalize_title


def test_empty_and_whitespace_pass_through_unchanged():
    assert normalize_title(None) is None
    assert normalize_title("") == ""
    assert normalize_title("   ") == "   "


def test_plain_title_is_left_alone():
    assert normalize_title("How I built a boat") == "How I built a boat"


def test_shouting_title_is_de_capsed_and_first_letter_kept_upper():
    # >=2 all-caps runs and >=50% caps → treated as shouting: title-case the long words.
    out = normalize_title("THIS IS A HUGE ANNOUNCEMENT")
    assert out is not None
    assert out == out[0].upper() + out[1:]  # first alpha char stays capitalized
    assert "HUGE" not in out and "ANNOUNCEMENT" not in out


def test_trailing_hashtags_are_stripped():
    assert normalize_title("Great cooking video #food #yummy") == "Great cooking video"


def test_emoji_and_symbols_are_dropped():
    assert "🔥" not in (normalize_title("Fire recipe 🔥🔥") or "")


def test_repeated_punctuation_is_collapsed():
    assert normalize_title("Wait what!!!!") == "Wait what!"


def test_accented_non_english_letters_survive():
    # Siftlode is trilingual — Hungarian/German letters must not be ASCII-folded away.
    for title in ["Gulyásleves főzés", "Über die Straße"]:
        out = normalize_title(title) or ""
        for ch in "áöüőűé":
            if ch in title:
                assert ch in out