vicinity 0.3.1

Approximate Nearest Neighbor Search: HNSW, DiskANN, IVF-PQ, ScaNN, quantization
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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302

//! Robustness tests for vicinity.
//!
//! These tests verify graceful error handling for edge cases: NaN/Inf inputs,
//! empty indexes, dimension mismatches, SIMD odd dimensions, and misuse patterns.
//! Tests should not panic -- they assert that errors are returned or results are valid.

use vicinity::distance::{normalize, DistanceMetric};
use vicinity::hnsw::HNSWIndex;

// ---------------------------------------------------------------------------
// 1. NaN / Inf rejection
// ---------------------------------------------------------------------------

#[test]
fn nan_vector_rejected_or_handled_gracefully() {
    // NaN vectors should ideally be rejected at add time. This test documents
    // actual behavior and ensures no unrecoverable panic in release builds.
    //
    // FINDING: In debug builds, a debug_assert! in add_slice fires on NaN
    // vectors (norm^2 check). In release builds, NaN is silently accepted.
    // Neither path corrupts the index in a way that causes subsequent panics.
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();

    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();

    let nan_vec = vec![f32::NAN; 4];
    // Wrap in catch_unwind because debug_assert! fires in debug builds.
    let add_result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
        index.add_slice(1, &nan_vec)
    }));

    match add_result {
        Err(_) => {
            // debug_assert! fired -- acceptable in debug builds.
        }
        Ok(Err(_)) => {
            // Explicit error returned -- ideal behavior.
        }
        Ok(Ok(())) => {
            // Silently accepted (release builds). Verify search still works.
            let _ = index.build();
        }
    }
}

#[test]
fn inf_vector_rejected_or_handled_gracefully() {
    // Same pattern as NaN: debug_assert! fires in debug builds on Inf vectors
    // because norm^2 is Inf. Release builds accept silently.
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();

    let v = normalize(&[0.0, 1.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();

    let inf_vec = vec![f32::INFINITY; 4];
    let add_result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
        index.add_slice(1, &inf_vec)
    }));

    match add_result {
        Err(_) => {
            // debug_assert! fired -- acceptable in debug builds.
        }
        Ok(Err(_)) => {
            // Explicit error returned -- ideal behavior.
        }
        Ok(Ok(())) => {
            // Silently accepted (release builds). Verify search still works.
            let _ = index.build();
        }
    }
}

// ---------------------------------------------------------------------------
// 2. Empty index search
// ---------------------------------------------------------------------------

#[test]
fn search_empty_index_returns_error() {
    let index = HNSWIndex::new(4, 16, 32).unwrap();
    // Index is not built, so search should return Err (not built).
    let result = index.search(&[1.0, 0.0, 0.0, 0.0], 5, 10);
    assert!(
        result.is_err(),
        "searching an unbuilt index should return Err"
    );
}

#[test]
fn build_empty_index_returns_error() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let result = index.build();
    assert!(result.is_err(), "building an empty index should return Err");
}

// ---------------------------------------------------------------------------
// 3. Zero-dimension index
// ---------------------------------------------------------------------------

#[test]
fn zero_dimension_rejected() {
    let result = HNSWIndex::new(0, 16, 32);
    assert!(
        result.is_err(),
        "creating an index with dimension=0 should fail"
    );
}

// ---------------------------------------------------------------------------
// 4. Dimension mismatch on search
// ---------------------------------------------------------------------------

#[test]
fn search_dimension_mismatch_returns_error() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();
    index.build().unwrap();

    // Query with wrong dimension (2 instead of 4).
    let result = index.search(&[1.0, 0.0], 1, 10);
    assert!(
        result.is_err(),
        "search with mismatched query dimension should return Err"
    );
}

#[test]
fn add_dimension_mismatch_returns_error() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let result = index.add_slice(0, &[1.0, 0.0]);
    assert!(
        result.is_err(),
        "adding a vector with wrong dimension should return Err"
    );
}

// ---------------------------------------------------------------------------
// 5. Search before build
// ---------------------------------------------------------------------------

#[test]
fn search_before_build_returns_error() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();

    let result = index.search(&[1.0, 0.0, 0.0, 0.0], 1, 10);
    assert!(result.is_err(), "searching before build should return Err");
}

// ---------------------------------------------------------------------------
// 6. SIMD odd dimensions
// ---------------------------------------------------------------------------

#[test]
fn simd_odd_dimensions_produce_finite_nonnegative_distances() {
    for dim in [1, 3, 7, 15, 17, 31, 33] {
        let a: Vec<f32> = (0..dim).map(|i| i as f32 * 0.1 + 0.01).collect();
        let b: Vec<f32> = (0..dim).map(|i| (dim - i) as f32 * 0.1 + 0.01).collect();
        let a_norm = normalize(&a);
        let b_norm = normalize(&b);

        let dist = DistanceMetric::Cosine.distance(&a_norm, &b_norm);
        assert!(
            dist.is_finite(),
            "dim={dim}: cosine distance is not finite: {dist}"
        );
        assert!(
            dist >= -1e-6,
            "dim={dim}: cosine distance is unexpectedly negative: {dist}"
        );

        let l2 = DistanceMetric::L2.distance(&a_norm, &b_norm);
        assert!(l2.is_finite(), "dim={dim}: L2 distance is not finite: {l2}");
        assert!(l2 >= 0.0, "dim={dim}: L2 distance is negative: {l2}");
    }
}

// ---------------------------------------------------------------------------
// 7. add_batch dimension mismatch
// ---------------------------------------------------------------------------

#[test]
fn add_batch_dimension_mismatch_returns_error() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    // 3 floats for 2 vectors of dim 4 -- should fail.
    let result = index.add_batch(&[0, 1], &[1.0, 2.0, 3.0]);
    assert!(
        result.is_err(),
        "add_batch with mismatched vector/id count should return Err"
    );
}

#[test]
fn add_batch_correct_dimensions_succeeds() {
    let mut index = HNSWIndex::new(2, 4, 4).unwrap();
    let v0 = normalize(&[1.0, 0.0]);
    let v1 = normalize(&[0.0, 1.0]);
    let mut flat = Vec::new();
    flat.extend_from_slice(&v0);
    flat.extend_from_slice(&v1);

    let result = index.add_batch(&[0, 1], &flat);
    assert!(
        result.is_ok(),
        "add_batch with correct dimensions should succeed"
    );
}

// ---------------------------------------------------------------------------
// 8. Additional edge cases
// ---------------------------------------------------------------------------

#[test]
fn duplicate_doc_id_rejected() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();

    let result = index.add_slice(0, &v);
    assert!(
        result.is_err(),
        "adding a duplicate doc_id should return Err"
    );
}

#[test]
fn search_k_zero_returns_empty() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();
    index.build().unwrap();

    let result = index.search(&normalize(&[1.0, 0.0, 0.0, 0.0]), 0, 10);
    match result {
        Ok(r) => assert!(r.is_empty(), "k=0 should return empty results"),
        Err(_) => {} // also acceptable
    }
}

#[test]
fn search_k_larger_than_index_size() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();
    index.build().unwrap();

    // Ask for 100 neighbors when only 1 vector exists.
    let result = index.search(&normalize(&[1.0, 0.0, 0.0, 0.0]), 100, 200);
    match result {
        Ok(r) => assert!(
            r.len() <= 1,
            "should return at most 1 result, got {}",
            r.len()
        ),
        Err(_) => {} // also acceptable
    }
}

#[test]
fn add_after_build_rejected() {
    let mut index = HNSWIndex::new(4, 16, 32).unwrap();
    let v = normalize(&[1.0, 0.0, 0.0, 0.0]);
    index.add_slice(0, &v).unwrap();
    index.build().unwrap();

    let result = index.add_slice(1, &normalize(&[0.0, 1.0, 0.0, 0.0]));
    assert!(
        result.is_err(),
        "adding vectors after build should return Err"
    );
}

#[test]
fn distance_metric_dimension_mismatch_returns_infinity() {
    let a = [1.0, 0.0, 0.0];
    let b = [1.0, 0.0];
    let dist = DistanceMetric::L2.distance(&a, &b);
    assert!(
        dist == f32::INFINITY,
        "L2 distance on mismatched dims should be INFINITY, got {dist}"
    );
    let dist = DistanceMetric::Cosine.distance(&a, &b);
    assert!(
        dist == f32::INFINITY,
        "Cosine distance on mismatched dims should be INFINITY, got {dist}"
    );
}

#[test]
fn zero_m_rejected() {
    let result = HNSWIndex::new(4, 0, 32);
    assert!(result.is_err(), "m=0 should be rejected");
}

#[test]
fn zero_m_max_rejected() {
    let result = HNSWIndex::new(4, 16, 0);
    assert!(result.is_err(), "m_max=0 should be rejected");
}