"""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_numeric_episode_marker_is_not_treated_as_a_hashtag():
    # A bare "#3" is an episode marker, not an SEO hashtag (no letters) → it must survive.
    assert normalize_title("My show #3") == "My show #3"


def test_midstring_hashtags_are_fast_and_kept_redos_regression():
    # Facebook prepends an engagement prefix and appends a page name, leaving a long hashtag cluster
    # in the MIDDLE. That is not a trailing SEO cluster, so the tags stay — and, critically, this
    # exact shape used to trigger catastrophic backtracking (ReDoS) in `_TRAILING_HASHTAGS` that
    # pinned a download-worker thread at 100% CPU forever inside the yt-dlp progress hook.
    import time

    raw = (
        "398K views · 6.1K reactions | The money should follow 🙄 #reel #reels #meme #memes "
        "#fun #funny #work #worklife #office #viral #trending #fyp #foryou #foryoupage | Memebonanza4U"
    )
    t = time.perf_counter()
    out = normalize_title(raw) or ""
    assert time.perf_counter() - t < 1.0  # linear now; the old regex ran for minutes / never returned
    assert "#foryoupage" in out and "#reel" in out  # mid-string hashtags preserved (not trailing)
    assert "views" not in out  # the engagement-count prefix was still stripped


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