SoulStream/crates/metadata/examples/cold_audit.rs

//! Cold-start bug audit for the artist-page pipeline. Companion to
//! `artist_page_bench` (latency) — this one checks INVARIANTS across a spread
//! of artist shapes and prints a statistics row per artist:
//!
//!   - overview / page / identity wall-clock, run the way the frontend runs
//!     them (all three overlapped, one fresh client per artist, no disk cache)
//!   - album and popular-row counts at each stage
//!   - listeners coverage (the Last.fm anchor working end-to-end)
//!   - release-date coverage (the MusicBrainz dates pass reaching the page)
//!   - BUG lines for anything that violates an invariant
//!
//! Artists are audited SEQUENTIALLY — each fresh client has its own
//! MusicBrainz pacer, and overlapping artists would multiply request rate
//! against the 1 req/s policy (and contaminate the numbers, see the
//! paced-probe rule).
//!
//! Usage:
//!   cargo run -p metadata --release --example cold_audit -- ["Artist" ...]
//!   (no args = the built-in spread)

use metadata::MetadataClient;
use std::sync::Arc;
use std::time::Instant;

#[tokio::main]
async fn main() {
    let mut artists: Vec<String> = std::env::args().skip(1).collect();
    if artists.is_empty() {
        artists = [
            "Alvvays",                            // small, clean MB identity
            "Weezer",                             // medium, deep catalog
            "Queen",                              // huge, 1119 MB pressings
            "Death Grips",                        // mixtape/street edge
            "King Gizzard & The Lizard Wizard",   // Last.fm mbid-index miss + '&'
            "Boards of Canada",                   // electronic, sparse Deezer
            "Duster",                             // namesake-collision shape
        ]
        .into_iter()
        .map(String::from)
        .collect();
    }

    let mut bugs: Vec<String> = Vec::new();
    println!(
        "{:<36} {:>8} {:>8} {:>8}  {:>6} {:>6} {:>7}  {:>9} {:>8}",
        "artist", "overview", "page", "identity", "ov-alb", "pg-alb", "popular", "listeners", "dates"
    );

    for artist in &artists {
        let client = Arc::new(MetadataClient::new());
        let search = client.catalog_search(artist).await;
        let Some(found) = search
            .artists
            .iter()
            .find(|a| normalize(&a.name) == normalize(artist))
        else {
            bugs.push(format!("{artist}: no exact-name artist in catalog search"));
            continue;
        };
        let id = found.id;

        // The frontend shape: overview, page and the identity warm-up all in
        // flight together on one client.
        let (overview, page, identity) = tokio::join!(
            async {
                let t = Instant::now();
                let overview = client.catalog_artist_overview(id).await;
                (t.elapsed().as_millis(), overview)
            },
            async {
                let t = Instant::now();
                let page = client.catalog_artist(id).await;
                (t.elapsed().as_millis(), page)
            },
            async {
                let t = Instant::now();
                let identity = client.catalog_artist_identity(id, artist).await;
                (t.elapsed().as_millis(), identity)
            },
        );

        let (ov_ms, overview) = overview;
        let (pg_ms, page) = page;
        let (id_ms, identity) = identity;

        let mut flag = |what: &str| bugs.push(format!("{artist}: {what}"));

        let (ov_albums, _ov_popular) = match &overview {
            Some(ov) => {
                if ov.artist.id != id {
                    flag("overview echoed a different artist id");
                }
                if ov.artist.name.is_empty() {
                    flag("overview artist name empty");
                }
                if ov.albums.is_empty() {
                    flag("overview albums empty");
                }
                if ov.popular_tracks.is_empty() {
                    flag("overview popular empty");
                }
                (ov.albums.len(), ov.popular_tracks.len())
            }
            None => {
                flag("overview returned None");
                (0, 0)
            }
        };

        let (pg_albums, pg_popular, with_listeners, with_dates) = match &page {
            Some(pg) => {
                if pg.artist.id != id {
                    flag("page echoed a different artist id");
                }
                if pg.albums.is_empty() {
                    flag("page albums empty");
                }
                if pg.popular_tracks.is_empty() {
                    flag("page popular empty");
                }
                let listeners = pg
                    .popular_tracks
                    .iter()
                    .filter(|t| t.listeners.is_some())
                    .count();
                let dates = pg
                    .albums
                    .iter()
                    .filter(|a| a.release_date.is_some())
                    .count();
                // The page waits for real supplements; zero listener counts on
                // a popular row set means the Last.fm anchor died end-to-end.
                if !pg.popular_tracks.is_empty() && listeners == 0 {
                    flag("page popular has zero Last.fm listener counts");
                }
                (pg.albums.len(), pg.popular_tracks.len(), listeners, dates)
            }
            None => {
                flag("page returned None");
                (0, 0, 0, 0)
            }
        };

        if identity.artist_id != id {
            flag("identity echoed a different artist id");
        }

        println!(
            "{:<36} {:>6}ms {:>6}ms {:>6}ms  {:>6} {:>6} {:>7}  {:>6}/{:<2} {:>4}/{:<3}",
            artist, ov_ms, pg_ms, id_ms, ov_albums, pg_albums, pg_popular,
            with_listeners, pg_popular, with_dates, pg_albums
        );
    }

    println!();
    if bugs.is_empty() {
        println!("NO INVARIANT FAILURES");
    } else {
        println!("FAILURES ({}):", bugs.len());
        for bug in &bugs {
            println!("  BUG {bug}");
        }
    }
}

fn normalize(s: &str) -> String {
    s.to_lowercase()
        .chars()
        .filter(|c| c.is_alphanumeric())
        .collect()
}