sqlite_graphrag/commands/
hybrid_search.rs

1use crate::cli::MemoryType;
2use crate::errors::AppError;
3use crate::output::{self, JsonOutputFormat, RecallItem};
4use crate::paths::AppPaths;
5use crate::storage::connection::open_ro;
6use crate::storage::memories;
7
8use std::collections::HashMap;
9
10#[derive(clap::Args)]
11pub struct HybridSearchArgs {
12    pub query: String,
13    #[arg(short = 'k', long, default_value = "10")]
14    pub k: usize,
15    #[arg(long, default_value = "60")]
16    pub rrf_k: u32,
17    #[arg(long, default_value = "1.0")]
18    pub weight_vec: f32,
19    #[arg(long, default_value = "1.0")]
20    pub weight_fts: f32,
21    #[arg(long, value_enum)]
22    pub r#type: Option<MemoryType>,
23    #[arg(long)]
24    pub namespace: Option<String>,
25    #[arg(long)]
26    pub with_graph: bool,
27    #[arg(long, default_value = "2")]
28    pub max_hops: u32,
29    #[arg(long, default_value = "0.3")]
30    pub min_weight: f64,
31    #[arg(long, value_enum, default_value_t = JsonOutputFormat::Json)]
32    pub format: JsonOutputFormat,
33    #[arg(long, env = "SQLITE_GRAPHRAG_DB_PATH")]
34    pub db: Option<String>,
35    /// Accept `--json` as a no-op because output is already JSON by default.
36    #[arg(long, help = "No-op; JSON is always emitted on stdout")]
37    pub json: bool,
38}
39
40#[derive(serde::Serialize)]
41pub struct HybridSearchItem {
42    pub memory_id: i64,
43    pub name: String,
44    pub namespace: String,
45    #[serde(rename = "type")]
46    pub memory_type: String,
47    pub description: String,
48    pub body: String,
49    pub combined_score: f64,
50    /// Alias de `combined_score` para contrato documentado em SKILL.md.
51    pub score: f64,
52    /// Fonte do match: sempre "hybrid" (RRF de vec + fts). Adicionado em v2.0.1.
53    pub source: String,
54    #[serde(skip_serializing_if = "Option::is_none")]
55    pub vec_rank: Option<usize>,
56    #[serde(skip_serializing_if = "Option::is_none")]
57    pub fts_rank: Option<usize>,
58    /// Score RRF combinado — alias explícito de `combined_score` para contratos de integração.
59    #[serde(skip_serializing_if = "Option::is_none")]
60    pub rrf_score: Option<f64>,
61}
62
63/// Pesos RRF usados na busca híbrida: vec (vetorial) e fts (texto).
64#[derive(serde::Serialize)]
65pub struct Weights {
66    pub vec: f32,
67    pub fts: f32,
68}
69
70#[derive(serde::Serialize)]
71pub struct HybridSearchResponse {
72    pub query: String,
73    pub k: usize,
74    /// Parâmetro k do RRF usado no ranking combinado.
75    pub rrf_k: u32,
76    /// Pesos aplicados às fontes vec e fts no RRF.
77    pub weights: Weights,
78    pub results: Vec<HybridSearchItem>,
79    pub graph_matches: Vec<RecallItem>,
80    /// Tempo total de execução em milissegundos desde início do handler até serialização.
81    pub elapsed_ms: u64,
82}
83
84pub fn run(args: HybridSearchArgs) -> Result<(), AppError> {
85    let start = std::time::Instant::now();
86    let _ = args.format;
87
88    let namespace = crate::namespace::resolve_namespace(args.namespace.as_deref())?;
89    let paths = AppPaths::resolve(args.db.as_deref())?;
90
91    output::emit_progress_i18n(
92        "Computing query embedding...",
93        "Calculando embedding da consulta...",
94    );
95    let embedding = crate::daemon::embed_query_or_local(&paths.models, &args.query)?;
96
97    let conn = open_ro(&paths.db)?;
98
99    let memory_type_str = args.r#type.map(|t| t.as_str());
100
101    let vec_results =
102        memories::knn_search(&conn, &embedding, &namespace, memory_type_str, args.k * 2)?;
103
104    // Mapear posição de ranking vetorial por memory_id (1-indexed conforme schema)
105    let vec_rank_map: HashMap<i64, usize> = vec_results
106        .iter()
107        .enumerate()
108        .map(|(pos, (id, _))| (*id, pos + 1))
109        .collect();
110
111    let fts_results =
112        memories::fts_search(&conn, &args.query, &namespace, memory_type_str, args.k * 2)?;
113
114    // Mapear posição de ranking FTS por memory_id (1-indexed conforme schema)
115    let fts_rank_map: HashMap<i64, usize> = fts_results
116        .iter()
117        .enumerate()
118        .map(|(pos, row)| (row.id, pos + 1))
119        .collect();
120
121    let rrf_k = args.rrf_k as f64;
122
123    // Acumular scores RRF combinados
124    let mut combined_scores: HashMap<i64, f64> = HashMap::new();
125
126    for (rank, (memory_id, _)) in vec_results.iter().enumerate() {
127        let score = args.weight_vec as f64 * (1.0 / (rrf_k + rank as f64 + 1.0));
128        *combined_scores.entry(*memory_id).or_insert(0.0) += score;
129    }
130
131    for (rank, row) in fts_results.iter().enumerate() {
132        let score = args.weight_fts as f64 * (1.0 / (rrf_k + rank as f64 + 1.0));
133        *combined_scores.entry(row.id).or_insert(0.0) += score;
134    }
135
136    // Ordenar por score descendente e tomar os top-k
137    let mut ranked: Vec<(i64, f64)> = combined_scores.into_iter().collect();
138    ranked.sort_by(|a, b| b.1.partial_cmp(&a.1).unwrap_or(std::cmp::Ordering::Equal));
139    ranked.truncate(args.k);
140
141    // Coletar todos os IDs para busca em batch (evitar N+1)
142    let top_ids: Vec<i64> = ranked.iter().map(|(id, _)| *id).collect();
143
144    // Buscar dados completos das memórias top
145    let mut memory_data: HashMap<i64, memories::MemoryRow> = HashMap::new();
146    for id in &top_ids {
147        if let Some(row) = memories::read_full(&conn, *id)? {
148            memory_data.insert(*id, row);
149        }
150    }
151
152    // Construir resultados finais na ordem de ranking
153    let results: Vec<HybridSearchItem> = ranked
154        .into_iter()
155        .filter_map(|(memory_id, combined_score)| {
156            memory_data.remove(&memory_id).map(|row| HybridSearchItem {
157                memory_id: row.id,
158                name: row.name,
159                namespace: row.namespace,
160                memory_type: row.memory_type,
161                description: row.description,
162                body: row.body,
163                combined_score,
164                score: combined_score,
165                source: "hybrid".to_string(),
166                vec_rank: vec_rank_map.get(&memory_id).copied(),
167                fts_rank: fts_rank_map.get(&memory_id).copied(),
168                rrf_score: Some(combined_score),
169            })
170        })
171        .collect();
172
173    output::emit_json(&HybridSearchResponse {
174        query: args.query,
175        k: args.k,
176        rrf_k: args.rrf_k,
177        weights: Weights {
178            vec: args.weight_vec,
179            fts: args.weight_fts,
180        },
181        results,
182        graph_matches: vec![],
183        elapsed_ms: start.elapsed().as_millis() as u64,
184    })?;
185
186    Ok(())
187}
188
189#[cfg(test)]
190mod testes {
191    use super::*;
192
193    fn resposta_vazia(
194        k: usize,
195        rrf_k: u32,
196        weight_vec: f32,
197        weight_fts: f32,
198    ) -> HybridSearchResponse {
199        HybridSearchResponse {
200            query: "busca teste".to_string(),
201            k,
202            rrf_k,
203            weights: Weights {
204                vec: weight_vec,
205                fts: weight_fts,
206            },
207            results: vec![],
208            graph_matches: vec![],
209            elapsed_ms: 0,
210        }
211    }
212
213    #[test]
214    fn hybrid_search_response_vazia_serializa_campos_corretos() {
215        let resp = resposta_vazia(10, 60, 1.0, 1.0);
216        let json = serde_json::to_string(&resp).unwrap();
217        assert!(json.contains("\"results\""), "deve conter campo results");
218        assert!(json.contains("\"query\""), "deve conter campo query");
219        assert!(json.contains("\"k\""), "deve conter campo k");
220        assert!(
221            json.contains("\"graph_matches\""),
222            "deve conter campo graph_matches"
223        );
224        assert!(
225            !json.contains("\"combined_rank\""),
226            "NÃO deve conter combined_rank"
227        );
228        assert!(
229            !json.contains("\"vec_rank_list\""),
230            "NÃO deve conter vec_rank_list"
231        );
232        assert!(
233            !json.contains("\"fts_rank_list\""),
234            "NÃO deve conter fts_rank_list"
235        );
236    }
237
238    #[test]
239    fn hybrid_search_response_serializa_rrf_k_e_weights() {
240        let resp = resposta_vazia(5, 60, 0.7, 0.3);
241        let json = serde_json::to_string(&resp).unwrap();
242        assert!(json.contains("\"rrf_k\""), "deve conter campo rrf_k");
243        assert!(json.contains("\"weights\""), "deve conter campo weights");
244        assert!(json.contains("\"vec\""), "deve conter campo weights.vec");
245        assert!(json.contains("\"fts\""), "deve conter campo weights.fts");
246    }
247
248    #[test]
249    fn hybrid_search_response_serializa_elapsed_ms() {
250        let mut resp = resposta_vazia(5, 60, 1.0, 1.0);
251        resp.elapsed_ms = 123;
252        let json = serde_json::to_string(&resp).unwrap();
253        assert!(
254            json.contains("\"elapsed_ms\""),
255            "deve conter campo elapsed_ms"
256        );
257        assert!(json.contains("123"), "deve serializar valor de elapsed_ms");
258    }
259
260    #[test]
261    fn weights_struct_serializa_corretamente() {
262        let w = Weights { vec: 0.6, fts: 0.4 };
263        let json = serde_json::to_string(&w).unwrap();
264        assert!(json.contains("\"vec\""));
265        assert!(json.contains("\"fts\""));
266    }
267
268    #[test]
269    fn hybrid_search_item_omite_fts_rank_quando_none() {
270        let item = HybridSearchItem {
271            memory_id: 1,
272            name: "mem".to_string(),
273            namespace: "default".to_string(),
274            memory_type: "user".to_string(),
275            description: "desc".to_string(),
276            body: "conteúdo".to_string(),
277            combined_score: 0.0328,
278            score: 0.0328,
279            source: "hybrid".to_string(),
280            vec_rank: Some(1),
281            fts_rank: None,
282            rrf_score: Some(0.0328),
283        };
284        let json = serde_json::to_string(&item).unwrap();
285        assert!(
286            json.contains("\"vec_rank\""),
287            "deve conter vec_rank quando Some"
288        );
289        assert!(
290            !json.contains("\"fts_rank\""),
291            "NÃO deve conter fts_rank quando None"
292        );
293    }
294
295    #[test]
296    fn hybrid_search_item_omite_vec_rank_quando_none() {
297        let item = HybridSearchItem {
298            memory_id: 2,
299            name: "mem2".to_string(),
300            namespace: "default".to_string(),
301            memory_type: "fact".to_string(),
302            description: "desc2".to_string(),
303            body: "corpo2".to_string(),
304            combined_score: 0.016,
305            score: 0.016,
306            source: "hybrid".to_string(),
307            vec_rank: None,
308            fts_rank: Some(2),
309            rrf_score: Some(0.016),
310        };
311        let json = serde_json::to_string(&item).unwrap();
312        assert!(
313            !json.contains("\"vec_rank\""),
314            "NÃO deve conter vec_rank quando None"
315        );
316        assert!(
317            json.contains("\"fts_rank\""),
318            "deve conter fts_rank quando Some"
319        );
320    }
321
322    #[test]
323    fn hybrid_search_item_serializa_ambos_ranks_quando_some() {
324        let item = HybridSearchItem {
325            memory_id: 3,
326            name: "mem3".to_string(),
327            namespace: "ns".to_string(),
328            memory_type: "entity".to_string(),
329            description: "desc3".to_string(),
330            body: "corpo3".to_string(),
331            combined_score: 0.05,
332            score: 0.05,
333            source: "hybrid".to_string(),
334            vec_rank: Some(3),
335            fts_rank: Some(1),
336            rrf_score: Some(0.05),
337        };
338        let json = serde_json::to_string(&item).unwrap();
339        assert!(json.contains("\"vec_rank\""), "deve conter vec_rank");
340        assert!(json.contains("\"fts_rank\""), "deve conter fts_rank");
341        assert!(json.contains("\"type\""), "deve serializar type renomeado");
342        assert!(!json.contains("memory_type"), "NÃO deve expor memory_type");
343    }
344
345    #[test]
346    fn hybrid_search_response_serializa_k_corretamente() {
347        let resp = resposta_vazia(5, 60, 1.0, 1.0);
348        let json = serde_json::to_string(&resp).unwrap();
349        assert!(json.contains("\"k\":5"), "deve serializar k=5");
350    }
351}
sqlite_graphrag/commands/hybrid_search.rs

sqlite_graphrag/commands/
hybrid_search.rs
