synadb 1.3.0

An AI-native embedded database
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
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130

//! Property-based tests for Vector serialization round-trip.
//!
//! **Feature: Syna-ai-native, Property 17: Vector Serialization Round-Trip**
//! **Validates: Requirements 1.1, 1.2**

use proptest::prelude::*;
use synadb::{Atom, SynaDB};
use tempfile::tempdir;

/// Generator for valid vector dimensions (64-8192).
/// These cover common embedding model dimensions.
fn arb_dimensions() -> impl Strategy<Value = u16> {
    prop_oneof![
        Just(64u16),
        Just(128u16),
        Just(256u16),
        Just(384u16), // MiniLM
        Just(512u16),
        Just(768u16),  // BERT base
        Just(1024u16), // BERT large
        Just(1536u16), // OpenAI ada-002
        Just(3072u16), // OpenAI text-embedding-3-large
        Just(4096u16),
    ]
}

proptest! {
    #![proptest_config(ProptestConfig::with_cases(100))]

    /// **Feature: Syna-ai-native, Property 17: Vector Serialization Round-Trip**
    ///
    /// For any valid vector with dimensions in the range 64-8192,
    /// storing it in the database and retrieving it SHALL produce
    /// a value equal to the original (within floating-point tolerance).
    ///
    /// **Validates: Requirements 1.1, 1.2**
    #[test]
    fn prop_vector_roundtrip(dims in arb_dimensions()) {
        let dir = tempdir().unwrap();
        let db_path = dir.path().join("test.db");

        // Generate vector with correct dimensions using deterministic values
        let vector: Vec<f32> = (0..dims).map(|i| i as f32 * 0.001).collect();

        let mut db = SynaDB::new(&db_path).unwrap();
        db.append("test_vec", Atom::Vector(vector.clone(), dims)).unwrap();

        let result = db.get("test_vec").unwrap();
        prop_assert!(result.is_some(), "Expected to retrieve stored vector");

        if let Some(Atom::Vector(data, d)) = result {
            prop_assert_eq!(d, dims, "Dimensions should match");
            prop_assert_eq!(data.len(), dims as usize, "Vector length should match dimensions");
            for (a, b) in data.iter().zip(vector.iter()) {
                prop_assert!((a - b).abs() < 1e-6 || (a.is_nan() && b.is_nan()),
                    "Vector elements should match: got {} expected {}", a, b);
            }
        } else {
            prop_assert!(false, "Expected Vector atom, got {:?}", result);
        }
    }

    /// Test vector round-trip with random f32 values (including edge cases).
    #[test]
    fn prop_vector_roundtrip_random_values(dims in arb_dimensions()) {
        let dir = tempdir().unwrap();
        let db_path = dir.path().join("test_random.db");

        // Generate random vector values
        let vector: Vec<f32> = (0..dims)
            .map(|i| {
                // Mix of positive, negative, and small values
                let base = (i as f32 - dims as f32 / 2.0) / dims as f32;
                base * 10.0
            })
            .collect();

        let mut db = SynaDB::new(&db_path).unwrap();
        db.append("random_vec", Atom::Vector(vector.clone(), dims)).unwrap();

        let result = db.get("random_vec").unwrap();
        prop_assert!(result.is_some(), "Expected to retrieve stored vector");

        if let Some(Atom::Vector(data, d)) = result {
            prop_assert_eq!(d, dims, "Dimensions should match");
            prop_assert_eq!(data.len(), dims as usize, "Vector length should match dimensions");
            for (i, (a, b)) in data.iter().zip(vector.iter()).enumerate() {
                prop_assert!((a - b).abs() < 1e-6 || (a.is_nan() && b.is_nan()),
                    "Vector element {} mismatch: got {} expected {}", i, a, b);
            }
        } else {
            prop_assert!(false, "Expected Vector atom, got {:?}", result);
        }
    }

    /// Test that vector persists across database reopen.
    #[test]
    fn prop_vector_persistence_roundtrip(dims in arb_dimensions()) {
        let dir = tempdir().unwrap();
        let db_path = dir.path().join("test_persist.db");

        // Generate vector
        let vector: Vec<f32> = (0..dims).map(|i| (i as f32).sin()).collect();

        // Write and close
        {
            let mut db = SynaDB::new(&db_path).unwrap();
            db.append("persist_vec", Atom::Vector(vector.clone(), dims)).unwrap();
            // db is dropped here, closing the file
        }

        // Reopen and read
        {
            let mut db = SynaDB::new(&db_path).unwrap();
            let result = db.get("persist_vec").unwrap();
            prop_assert!(result.is_some(), "Expected to retrieve stored vector after reopen");

            if let Some(Atom::Vector(data, d)) = result {
                prop_assert_eq!(d, dims, "Dimensions should match after reopen");
                prop_assert_eq!(data.len(), dims as usize, "Vector length should match after reopen");
                for (i, (a, b)) in data.iter().zip(vector.iter()).enumerate() {
                    prop_assert!((a - b).abs() < 1e-6 || (a.is_nan() && b.is_nan()),
                        "Vector element {} mismatch after reopen: got {} expected {}", i, a, b);
                }
            } else {
                prop_assert!(false, "Expected Vector atom after reopen, got {:?}", result);
            }
        }
    }
}