sqlite-vector-rs 0.2.1

SQLite extension providing PGVector-like native vector types with HNSW indexing
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114

use sqlite_vector_rs::distance::DistanceMetric;
use sqlite_vector_rs::types::VectorType;
use sqlite_vector_rs::vtab::config::VectorTableConfig;

#[test]
fn parse_basic_args() {
    let args = vec![
        "vector",
        "main",
        "embeddings",
        "dim=3",
        "type=float4",
        "metric=l2",
    ];
    let config = VectorTableConfig::parse(&args).unwrap();
    assert_eq!(config.dim, 3);
    assert_eq!(config.vtype, VectorType::Float4);
    assert_eq!(config.metric, DistanceMetric::L2);
    assert_eq!(config.table_name, "embeddings");
    assert!(config.metadata_columns.is_empty());
}

#[test]
fn parse_with_hnsw_params() {
    let args = vec![
        "vector",
        "main",
        "emb",
        "dim=768",
        "type=float4",
        "metric=cosine",
        "m=32",
        "ef_construction=400",
        "ef_search=128",
    ];
    let config = VectorTableConfig::parse(&args).unwrap();
    assert_eq!(config.hnsw_params.m, 32);
    assert_eq!(config.hnsw_params.ef_construction, 400);
    assert_eq!(config.hnsw_params.ef_search, 128);
}

#[test]
fn parse_with_metadata() {
    let args = vec![
        "vector",
        "main",
        "emb",
        "dim=3",
        "type=float4",
        "metric=l2",
        "metadata=\"label TEXT, category INTEGER\"",
    ];
    let config = VectorTableConfig::parse(&args).unwrap();
    assert_eq!(config.metadata_columns.len(), 2);
    assert_eq!(
        config.metadata_columns[0],
        ("label".to_string(), "TEXT".to_string())
    );
    assert_eq!(
        config.metadata_columns[1],
        ("category".to_string(), "INTEGER".to_string())
    );
}

#[test]
fn parse_defaults() {
    let args = vec!["vector", "main", "emb", "dim=3"];
    let config = VectorTableConfig::parse(&args).unwrap();
    assert_eq!(config.vtype, VectorType::Float4);
    assert_eq!(config.metric, DistanceMetric::L2);
    assert_eq!(config.hnsw_params.m, 16);
}

#[test]
fn parse_missing_dim_fails() {
    let args = vec!["vector", "main", "emb", "type=float4"];
    assert!(VectorTableConfig::parse(&args).is_err());
}

#[test]
fn parse_invalid_dim_fails() {
    let args = vec!["vector", "main", "emb", "dim=0"];
    assert!(VectorTableConfig::parse(&args).is_err());

    let args = vec!["vector", "main", "emb", "dim=-5"];
    assert!(VectorTableConfig::parse(&args).is_err());
}

#[test]
fn generates_create_table_sql() {
    let args = vec!["vector", "main", "emb", "dim=3", "type=float4", "metric=l2"];
    let config = VectorTableConfig::parse(&args).unwrap();
    let sql = config.vtab_schema();
    assert!(sql.contains("id INTEGER PRIMARY KEY"));
    assert!(sql.contains("vector BLOB"));
    assert!(sql.contains("distance REAL"));
}

#[test]
fn generates_schema_with_metadata() {
    let args = vec![
        "vector",
        "main",
        "emb",
        "dim=3",
        "type=float4",
        "metric=l2",
        "metadata=\"label TEXT, category INTEGER\"",
    ];
    let config = VectorTableConfig::parse(&args).unwrap();
    let sql = config.vtab_schema();
    assert!(sql.contains("label TEXT"));
    assert!(sql.contains("category INTEGER"));
}