rag 0.1.5

A Rust library and CLI for Retrieval-Augmented Generation
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

//! Integration tests for the new capabilities:
//! evaluation metrics, recursive/semantic/structural chunking, Int8
//! quantization, SIMD distances, and the write-ahead log.

use rag::{
    chunker::{RecursiveChunker, SemanticChunker, StructuralChunker, TextChunker},
    eval::{
        EvalReport, average_precision, ndcg_at_k, precision_at_k, recall_at_k, reciprocal_rank,
        relevance_set,
    },
    quantize::{QuantizationParams, QuantizedIndex},
    simd,
    vector_store::Document,
    wal::{WalOp, WriteAheadLog, apply_ops},
};
use std::collections::HashSet;
use tempfile::TempDir;

// ---------------- Evaluation metrics ----------------

#[test]
fn eval_metrics_end_to_end() {
    let relevant: HashSet<String> =
        relevance_set(["d1".to_string(), "d3".to_string(), "d7".to_string()]);
    let retrieved = vec![
        "d1".to_string(),
        "d2".to_string(),
        "d7".to_string(),
        "d4".to_string(),
    ];

    assert!((recall_at_k(&retrieved, &relevant, 4) - (2.0 / 3.0)).abs() < 1e-9);
    assert!((precision_at_k(&retrieved, &relevant, 4) - 0.5).abs() < 1e-9);
    assert_eq!(reciprocal_rank(&retrieved, &relevant), 1.0);
    assert!(average_precision(&retrieved, &relevant) > 0.5);

    let grades = vec![3.0, 0.0, 2.0, 0.0];
    let ideal = vec![3.0, 2.0];
    let ndcg = ndcg_at_k(&grades, &ideal, 4);
    assert!(ndcg > 0.0 && ndcg <= 1.0 + 1e-9);
}

#[tokio::test]
async fn eval_evaluate_aggregates_multiple_queries() {
    let queries = vec![
        (
            "rust memory".to_string(),
            vec!["a".to_string(), "b".to_string()],
        ),
        ("paris tower".to_string(), vec!["c".to_string()]),
    ];
    let report = rag::eval::evaluate(queries, 2, |q| async move {
        if q.contains("rust") {
            vec!["a".to_string(), "x".to_string()]
        } else {
            vec!["c".to_string(), "y".to_string()]
        }
    })
    .await;
    assert_eq!(report.n_queries, 2);
    assert!(report.recall > 0.0 && report.recall <= 1.0);
    assert!((report.mrr - 1.0).abs() < 1e-9);
    let empty = EvalReport::default();
    assert_eq!(empty.n_queries, 0);
}

// ---------------- Chunkers ----------------

#[test]
fn recursive_chunker_splits_long_text() {
    let chunker = RecursiveChunker::new(50, 0);
    let text = "alpha beta gamma delta.\n\nepsilon zeta eta theta.\n\niota kappa lambda.";
    let chunks = chunker.chunk(text).unwrap();
    assert!(chunks.len() >= 2);
    for c in &chunks {
        assert!(c.len() <= 55);
    }
}

#[test]
fn recursive_chunker_rejects_bad_overlap() {
    let chunker = RecursiveChunker::new(10, 20);
    assert!(chunker.chunk("some text here").is_err());
}

#[test]
fn semantic_chunker_groups_by_topic() {
    let chunker = SemanticChunker::new(500, 0.1);
    let text = "cats meow. cats purr. cats nap.\n\nquantum physics studies particles. particles make waves.";
    let chunks = chunker.chunk(text).unwrap();
    assert!(chunks.len() >= 2);
}

#[test]
fn structural_chunker_keeps_markdown_sections() {
    let chunker = StructuralChunker::new(1000, 0);
    let text = "# Title\nintro.\n## A\ncontent a.\n## B\ncontent b.";
    let chunks = chunker.chunk(text).unwrap();
    assert!(
        chunks
            .iter()
            .any(|c| c.contains("Section A") || c.contains("content a"))
    );
    assert!(chunks.iter().any(|c| c.contains("content b")));
}

#[test]
fn structural_chunker_code_fence_together() {
    let chunker = StructuralChunker::new(5000, 0);
    let text = "# Intro\nprose.\n```rust\nfn main() {}\n```\n## After\nmore prose.";
    let chunks = chunker.chunk(text).unwrap();
    assert!(chunks.iter().any(|c| c.contains("fn main()")));
}

// ---------------- Quantization ----------------

#[test]
fn quantization_roundtrip_preserves_order() {
    let vectors = vec![
        vec![1.0, 0.0, 0.0],
        vec![0.0, 1.0, 0.0],
        vec![1.0, 1.0, 0.0],
        vec![0.1, 0.1, 0.1],
    ];
    let params = QuantizationParams::fit(&vectors).unwrap();
    let mut index = QuantizedIndex::new(params);
    index.add("x", &[1.0, 0.0, 0.0]);
    index.add("y", &[0.0, 1.0, 0.0]);
    index.add("z", &[1.0, 1.0, 0.0]);

    let res = index.search(&[1.0, 0.0, 0.0], 3);
    assert_eq!(res.len(), 3);
    // "y" is orthogonal to the query and must rank last; the aligned
    // vectors "x"/"z" tie after dequantization, so only require top-2 set.
    assert_eq!(res.last().unwrap().0, "y");
    let top2: Vec<&str> = res.iter().take(2).map(|(id, _)| id.as_str()).collect();
    assert!(top2.contains(&"x"));
}

#[test]
fn quantization_errors_on_empty() {
    assert!(QuantizationParams::fit(&Vec::<Vec<f32>>::new()).is_err());
}

// ---------------- SIMD distances ----------------

#[test]
fn simd_matches_scalar_results() {
    let a: Vec<f32> = (0..300).map(|i| (i as f32) * 0.1 - 15.0).collect();
    let b: Vec<f32> = (0..300).map(|i| (i as f32) * 0.05).collect();

    let scalar_dot: f32 = a.iter().zip(b.iter()).map(|(x, y)| x * y).sum();
    assert!((simd::dot_product(&a, &b) - scalar_dot).abs() < 1e-2);

    let scalar_euc: f32 = a
        .iter()
        .zip(b.iter())
        .map(|(x, y)| (x - y) * (x - y))
        .sum::<f32>()
        .sqrt();
    assert!((simd::euclidean_distance(&a, &b) - scalar_euc).abs() < 1e-2);

    assert!(simd::cosine_similarity(&[1.0, 0.0], &[0.0, 1.0]).abs() < 1e-6);
}

// ---------------- Write-ahead log ----------------

#[test]
fn wal_append_replay_reconstructs_state() {
    let dir = TempDir::new().unwrap();
    let wal = WriteAheadLog::new(dir.path().join("wal.jsonl"));
    wal.append(&WalOp::Put(Document::with_id(
        "a".to_string(),
        "one".to_string(),
    )))
    .unwrap();
    wal.append_batch(&[
        WalOp::Put(Document::with_id("b".to_string(), "two".to_string())),
        WalOp::Delete("a".to_string()),
    ])
    .unwrap();

    let ops = wal.replay().unwrap();
    let state = apply_ops(&ops);
    assert!(!state.contains_key("a"));
    assert_eq!(state.get("b").unwrap().content, "two");
}

#[test]
fn wal_truncate_after_checkpoint() {
    let dir = TempDir::new().unwrap();
    let wal = WriteAheadLog::new(dir.path().join("wal.jsonl"));
    wal.append(&WalOp::Put(Document::with_id(
        "a".to_string(),
        "one".to_string(),
    )))
    .unwrap();
    assert!(wal.size() > 0);
    wal.truncate().unwrap();
    assert_eq!(wal.size(), 0);
    assert!(wal.replay().unwrap().is_empty());
}

#[test]
fn wal_missing_file_replays_empty() {
    let wal = WriteAheadLog::new("/nonexistent/path/wal.jsonl");
    assert!(wal.replay().unwrap().is_empty());
}

// ---------------- Compressed persistence (feature-gated) ----------------

#[cfg(feature = "compress")]
#[test]
fn compress_roundtrip_via_lib() {
    use rag::compress::{compress_bytes, decompress_bytes};
    let data = b"repeating repeating repeating".repeat(500);
    let compressed = compress_bytes(&data, 19).unwrap();
    assert!(compressed.len() < data.len());
    let back = decompress_bytes(&compressed).unwrap();
    assert_eq!(back, data);
}