sqlite_graphrag/commands/
hybrid_search.rs

1use crate::cli::MemoryType;
2use crate::errors::AppError;
3use crate::output::{self, OutputFormat, 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 = OutputFormat::Json)]
32    pub format: OutputFormat,
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    match args.format {
84        OutputFormat::Text | OutputFormat::Markdown => {
85            return Err(AppError::Validation(
86                "formato text/markdown ainda não implementado para hybrid-search — use --format json ou aguarde Tier 2".into(),
87            ));
88        }
89        OutputFormat::Json => {}
90    }
91
92    let namespace = crate::namespace::resolve_namespace(args.namespace.as_deref())?;
93    let paths = AppPaths::resolve(args.db.as_deref())?;
94
95    output::emit_progress_i18n(
96        "Computing query embedding...",
97        "Calculando embedding da consulta...",
98    );
99    let embedder = crate::embedder::get_embedder(&paths.models)?;
100    let embedding = crate::embedder::embed_query(embedder, &args.query)?;
101
102    let conn = open_ro(&paths.db)?;
103
104    let memory_type_str = args.r#type.map(|t| t.as_str());
105
106    let vec_results =
107        memories::knn_search(&conn, &embedding, &namespace, memory_type_str, args.k * 2)?;
108
109    // Mapear posição de ranking vetorial por memory_id (1-indexed conforme schema)
110    let vec_rank_map: HashMap<i64, usize> = vec_results
111        .iter()
112        .enumerate()
113        .map(|(pos, (id, _))| (*id, pos + 1))
114        .collect();
115
116    let fts_results =
117        memories::fts_search(&conn, &args.query, &namespace, memory_type_str, args.k * 2)?;
118
119    // Mapear posição de ranking FTS por memory_id (1-indexed conforme schema)
120    let fts_rank_map: HashMap<i64, usize> = fts_results
121        .iter()
122        .enumerate()
123        .map(|(pos, row)| (row.id, pos + 1))
124        .collect();
125
126    let rrf_k = args.rrf_k as f64;
127
128    // Acumular scores RRF combinados
129    let mut combined_scores: HashMap<i64, f64> = HashMap::new();
130
131    for (rank, (memory_id, _)) in vec_results.iter().enumerate() {
132        let score = args.weight_vec as f64 * (1.0 / (rrf_k + rank as f64 + 1.0));
133        *combined_scores.entry(*memory_id).or_insert(0.0) += score;
134    }
135
136    for (rank, row) in fts_results.iter().enumerate() {
137        let score = args.weight_fts as f64 * (1.0 / (rrf_k + rank as f64 + 1.0));
138        *combined_scores.entry(row.id).or_insert(0.0) += score;
139    }
140
141    // Ordenar por score descendente e tomar os top-k
142    let mut ranked: Vec<(i64, f64)> = combined_scores.into_iter().collect();
143    ranked.sort_by(|a, b| b.1.partial_cmp(&a.1).unwrap_or(std::cmp::Ordering::Equal));
144    ranked.truncate(args.k);
145
146    // Coletar todos os IDs para busca em batch (evitar N+1)
147    let top_ids: Vec<i64> = ranked.iter().map(|(id, _)| *id).collect();
148
149    // Buscar dados completos das memórias top
150    let mut memory_data: HashMap<i64, memories::MemoryRow> = HashMap::new();
151    for id in &top_ids {
152        if let Some(row) = memories::read_full(&conn, *id)? {
153            memory_data.insert(*id, row);
154        }
155    }
156
157    // Construir resultados finais na ordem de ranking
158    let results: Vec<HybridSearchItem> = ranked
159        .into_iter()
160        .filter_map(|(memory_id, combined_score)| {
161            memory_data.remove(&memory_id).map(|row| HybridSearchItem {
162                memory_id: row.id,
163                name: row.name,
164                namespace: row.namespace,
165                memory_type: row.memory_type,
166                description: row.description,
167                body: row.body,
168                combined_score,
169                score: combined_score,
170                source: "hybrid".to_string(),
171                vec_rank: vec_rank_map.get(&memory_id).copied(),
172                fts_rank: fts_rank_map.get(&memory_id).copied(),
173            })
174        })
175        .collect();
176
177    output::emit_json(&HybridSearchResponse {
178        query: args.query,
179        k: args.k,
180        rrf_k: args.rrf_k,
181        weights: Weights {
182            vec: args.weight_vec,
183            fts: args.weight_fts,
184        },
185        results,
186        graph_matches: vec![],
187        elapsed_ms: start.elapsed().as_millis() as u64,
188    })?;
189
190    Ok(())
191}
192
193#[cfg(test)]
194mod testes {
195    use super::*;
196
197    fn resposta_vazia(
198        k: usize,
199        rrf_k: u32,
200        weight_vec: f32,
201        weight_fts: f32,
202    ) -> HybridSearchResponse {
203        HybridSearchResponse {
204            query: "busca teste".to_string(),
205            k,
206            rrf_k,
207            weights: Weights {
208                vec: weight_vec,
209                fts: weight_fts,
210            },
211            results: vec![],
212            graph_matches: vec![],
213            elapsed_ms: 0,
214        }
215    }
216
217    #[test]
218    fn hybrid_search_response_vazia_serializa_campos_corretos() {
219        let resp = resposta_vazia(10, 60, 1.0, 1.0);
220        let json = serde_json::to_string(&resp).unwrap();
221        assert!(json.contains("\"results\""), "deve conter campo results");
222        assert!(json.contains("\"query\""), "deve conter campo query");
223        assert!(json.contains("\"k\""), "deve conter campo k");
224        assert!(
225            json.contains("\"graph_matches\""),
226            "deve conter campo graph_matches"
227        );
228        assert!(
229            !json.contains("\"combined_rank\""),
230            "NÃO deve conter combined_rank"
231        );
232        assert!(
233            !json.contains("\"vec_rank_list\""),
234            "NÃO deve conter vec_rank_list"
235        );
236        assert!(
237            !json.contains("\"fts_rank_list\""),
238            "NÃO deve conter fts_rank_list"
239        );
240    }
241
242    #[test]
243    fn hybrid_search_response_serializa_rrf_k_e_weights() {
244        let resp = resposta_vazia(5, 60, 0.7, 0.3);
245        let json = serde_json::to_string(&resp).unwrap();
246        assert!(json.contains("\"rrf_k\""), "deve conter campo rrf_k");
247        assert!(json.contains("\"weights\""), "deve conter campo weights");
248        assert!(json.contains("\"vec\""), "deve conter campo weights.vec");
249        assert!(json.contains("\"fts\""), "deve conter campo weights.fts");
250    }
251
252    #[test]
253    fn hybrid_search_response_serializa_elapsed_ms() {
254        let mut resp = resposta_vazia(5, 60, 1.0, 1.0);
255        resp.elapsed_ms = 123;
256        let json = serde_json::to_string(&resp).unwrap();
257        assert!(
258            json.contains("\"elapsed_ms\""),
259            "deve conter campo elapsed_ms"
260        );
261        assert!(json.contains("123"), "deve serializar valor de elapsed_ms");
262    }
263
264    #[test]
265    fn weights_struct_serializa_corretamente() {
266        let w = Weights { vec: 0.6, fts: 0.4 };
267        let json = serde_json::to_string(&w).unwrap();
268        assert!(json.contains("\"vec\""));
269        assert!(json.contains("\"fts\""));
270    }
271
272    #[test]
273    fn hybrid_search_item_omite_fts_rank_quando_none() {
274        let item = HybridSearchItem {
275            memory_id: 1,
276            name: "mem".to_string(),
277            namespace: "default".to_string(),
278            memory_type: "user".to_string(),
279            description: "desc".to_string(),
280            body: "conteúdo".to_string(),
281            combined_score: 0.0328,
282            score: 0.0328,
283            source: "hybrid".to_string(),
284            vec_rank: Some(1),
285            fts_rank: None,
286        };
287        let json = serde_json::to_string(&item).unwrap();
288        assert!(
289            json.contains("\"vec_rank\""),
290            "deve conter vec_rank quando Some"
291        );
292        assert!(
293            !json.contains("\"fts_rank\""),
294            "NÃO deve conter fts_rank quando None"
295        );
296    }
297
298    #[test]
299    fn hybrid_search_item_omite_vec_rank_quando_none() {
300        let item = HybridSearchItem {
301            memory_id: 2,
302            name: "mem2".to_string(),
303            namespace: "default".to_string(),
304            memory_type: "fact".to_string(),
305            description: "desc2".to_string(),
306            body: "corpo2".to_string(),
307            combined_score: 0.016,
308            score: 0.016,
309            source: "hybrid".to_string(),
310            vec_rank: None,
311            fts_rank: Some(2),
312        };
313        let json = serde_json::to_string(&item).unwrap();
314        assert!(
315            !json.contains("\"vec_rank\""),
316            "NÃO deve conter vec_rank quando None"
317        );
318        assert!(
319            json.contains("\"fts_rank\""),
320            "deve conter fts_rank quando Some"
321        );
322    }
323
324    #[test]
325    fn hybrid_search_item_serializa_ambos_ranks_quando_some() {
326        let item = HybridSearchItem {
327            memory_id: 3,
328            name: "mem3".to_string(),
329            namespace: "ns".to_string(),
330            memory_type: "entity".to_string(),
331            description: "desc3".to_string(),
332            body: "corpo3".to_string(),
333            combined_score: 0.05,
334            score: 0.05,
335            source: "hybrid".to_string(),
336            vec_rank: Some(3),
337            fts_rank: Some(1),
338        };
339        let json = serde_json::to_string(&item).unwrap();
340        assert!(json.contains("\"vec_rank\""), "deve conter vec_rank");
341        assert!(json.contains("\"fts_rank\""), "deve conter fts_rank");
342        assert!(json.contains("\"type\""), "deve serializar type renomeado");
343        assert!(!json.contains("memory_type"), "NÃO deve expor memory_type");
344    }
345
346    #[test]
347    fn hybrid_search_response_serializa_k_corretamente() {
348        let resp = resposta_vazia(5, 60, 1.0, 1.0);
349        let json = serde_json::to_string(&resp).unwrap();
350        assert!(json.contains("\"k\":5"), "deve serializar k=5");
351    }
352}
sqlite_graphrag/commands/hybrid_search.rs

sqlite_graphrag/commands/
hybrid_search.rs
