//! Byte-index safety for the news scorer.
//!
//! Google's headlines are full of curly quotes and en/em dashes. Every one of
//! these panicked at some point during development because a byte offset was
//! advanced by 1 past a multi-byte character. A panic inside an async Tauri
//! command never resolves its promise, so the symptom was a hung News band —
//! and only for artists whose name appears mid-headline, which is why it was
//! invisible on the first artist tried.
use metadata::{news_relevance, news_tier, ArtistNewsItem};

fn score(title: &str, artist: &str) -> Option<i32> {
    let item = ArtistNewsItem {
        title: title.to_string(),
        url: String::new(),
        source: String::new(),
        date: 1_785_000_000,
        host: "example.com".to_string(),
        others: 0,
    };
    news_relevance(&item, artist, news_tier(artist))
}

#[test]
fn multibyte_punctuation_before_the_name_does_not_panic() {
    for (title, artist) in [
        ("Listen \u{2013} toe announce a new album", "toe"),
        ("Review \u{2014} Pond announce 'Terrestrials'", "Pond"),
        ("Watch: toe share \u{2018}Nothing\u{2019} video", "toe"),
        ("\u{201c}Terrestrials\u{201d} \u{2013} Pond return", "Pond"),
        ("Interview \u{2013} King Gizzard & The Lizard Wizard on Spotify", "King Gizzard & The Lizard Wizard"),
        ("\u{2013}\u{2014}\u{2018}\u{2019}\u{201c}\u{201d} toe", "toe"),
    ] {
        let _ = score(title, artist); // must not panic
    }
}

#[test]
fn non_ascii_headlines_score_without_panicking() {
    for (title, artist) in [
        ("Elephant Gym\u{2019}s Math Rock journey from Kaohsiung", "Elephant Gym"),
        ("\u{6771}\u{4EAC}: toe announce October shows", "toe"),
        ("R\u{00F6}yksopp \u{2013} toe remix premiere", "toe"),
    ] {
        let _ = score(title, artist);
    }
}

#[test]
fn a_multibyte_dash_marker_behaves_exactly_like_an_ascii_colon() {
    // The fix must not merely avoid the panic — the masthead signal has to
    // still fire. Comparing the two spellings of the same headline tests that
    // directly, without depending on where a tier's threshold happens to sit.
    // The lead word must NOT be one of the "Review:/Listen:" prefix markers —
    // a colon would then earn a second, unrelated bonus and the two spellings
    // would legitimately differ.
    for (colon, dash) in [
        ("Stereogum: toe announce a new album", "Stereogum \u{2013} toe announce a new album"),
        ("Mojo: Pond return with 'Terrestrials'", "Mojo \u{2014} Pond return with 'Terrestrials'"),
    ] {
        let artist = if colon.contains("toe") { "toe" } else { "Pond" };
        assert_eq!(
            score(colon, artist),
            score(dash, artist),
            "marker spelling changed the score for: {dash}",
        );
    }
    // Whether that shape clears a tier's bar is the relevance suite's job —
    // it depends on the publisher host, which this file deliberately holds
    // neutral so only the byte handling is under test.
}