use mem_store::{VectorStore, VectorRecord};

#[test]
fn a1_insert_and_search() {
    let mut store = VectorStore::new();
    
    // Insert two similar vectors
    let v1 = vec![1.0, 0.0, 0.0];
    let v2 = vec![0.99, 0.1, 0.0];
    let v3 = vec![0.0, 0.0, 1.0];  // orthogonal
    
    store.insert(VectorRecord {
        id: "r1".to_string(),
        chunk_id: "c1".to_string(),
        kind: "text".to_string(),
        embedding: v1,
        tokens: 100,
    }).unwrap();
    
    store.insert(VectorRecord {
        id: "r2".to_string(),
        chunk_id: "c2".to_string(),
        kind: "text".to_string(),
        embedding: v2,
        tokens: 100,
    }).unwrap();
    
    store.insert(VectorRecord {
        id: "r3".to_string(),
        chunk_id: "c3".to_string(),
        kind: "text".to_string(),
        embedding: v3,
        tokens: 100,
    }).unwrap();
    
    // Search for vectors similar to v1
    let results = store.search(&[1.0, 0.0, 0.0], 3, 0.0).unwrap();
    
    // r1 should be first (identical)
    assert_eq!(results[0].0, "r1");
    assert!((results[0].1 - 1.0).abs() < 0.01);
    
    // r2 should be second (similar)
    assert_eq!(results[1].0, "r2");
    assert!(results[1].1 > 0.9);
    
    // r3 should be last (orthogonal)
    assert_eq!(results[2].0, "r3");
    assert!(results[2].1 < 0.1);
}

#[test]
fn a2_min_score_filter() {
    let mut store = VectorStore::new();
    
    store.insert(VectorRecord {
        id: "r1".to_string(),
        chunk_id: "c1".to_string(),
        kind: "text".to_string(),
        embedding: vec![1.0, 0.0],
        tokens: 100,
    }).unwrap();
    
    store.insert(VectorRecord {
        id: "r2".to_string(),
        chunk_id: "c2".to_string(),
        kind: "text".to_string(),
        embedding: vec![0.0, 1.0],
        tokens: 100,
    }).unwrap();
    
    // Search with high threshold - only perfect match
    let results = store.search(&[1.0, 0.0], 10, 0.99).unwrap();
    assert_eq!(results.len(), 1);
    assert_eq!(results[0].0, "r1");
}