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

sqlite_graphrag/commands/
hybrid_search.rs
