1use std::path::Path;
2
3use crate::core::bm25_index::BM25Index;
4#[cfg(feature = "qdrant")]
5use crate::core::bm25_index::ChunkKind;
6use crate::core::hybrid_search::{DenseSearchResult, HybridConfig, HybridResult};
7
8#[derive(Debug, Clone, Copy, PartialEq, Eq)]
9pub enum DenseBackendKind {
10 Local,
11 #[cfg(feature = "qdrant")]
12 Qdrant,
13}
14
15impl DenseBackendKind {
16 pub fn try_from_env() -> Result<Self, String> {
17 let explicit = std::env::var("LEANCTX_DENSE_BACKEND")
18 .ok()
19 .map(|v| v.trim().to_ascii_lowercase())
20 .filter(|v| !v.is_empty());
21
22 let inferred_qdrant = std::env::var("LEANCTX_QDRANT_URL")
23 .ok()
24 .is_some_and(|v| !v.trim().is_empty());
25
26 let requested = explicit.or_else(|| inferred_qdrant.then_some("qdrant".to_string()));
27
28 match requested.as_deref() {
29 None | Some("local") => Ok(Self::Local),
30 Some("qdrant") => {
31 #[cfg(feature = "qdrant")]
32 {
33 Ok(Self::Qdrant)
34 }
35 #[cfg(not(feature = "qdrant"))]
36 {
37 Err("Dense backend 'qdrant' requested, but feature 'qdrant' is not enabled. Rebuild with --features qdrant.".to_string())
38 }
39 }
40 Some(other) => Err(format!(
41 "Unknown LEANCTX_DENSE_BACKEND={other:?} (expected 'local' or 'qdrant')"
42 )),
43 }
44 }
45
46 pub fn label(&self) -> &'static str {
47 match self {
48 Self::Local => "local",
49 #[cfg(feature = "qdrant")]
50 Self::Qdrant => "qdrant",
51 }
52 }
53}
54
55#[cfg(feature = "embeddings")]
56#[allow(clippy::too_many_arguments)]
57pub fn dense_results_as_hybrid(
58 backend: DenseBackendKind,
59 root: &Path,
60 index: &BM25Index,
61 engine: &crate::core::embeddings::EmbeddingEngine,
62 aligned_embeddings: &[Vec<f32>],
63 changed_files: &[String],
64 query: &str,
65 top_k: usize,
66 filter: Option<&dyn Fn(&str) -> bool>,
67) -> Result<Vec<HybridResult>, String> {
68 let dense = dense_results(
69 backend,
70 root,
71 index,
72 engine,
73 aligned_embeddings,
74 changed_files,
75 query,
76 top_k,
77 filter,
78 )?;
79
80 Ok(dense
81 .into_iter()
82 .map(|d| HybridResult {
83 file_path: d.file_path,
84 symbol_name: d.symbol_name,
85 kind: d.kind,
86 start_line: d.start_line,
87 end_line: d.end_line,
88 snippet: d.snippet,
89 rrf_score: d.similarity as f64,
90 bm25_score: None,
91 dense_score: Some(d.similarity),
92 bm25_rank: None,
93 dense_rank: None,
94 })
95 .collect())
96}
97
98#[cfg(feature = "embeddings")]
99#[allow(clippy::too_many_arguments)]
100pub fn hybrid_results(
101 backend: DenseBackendKind,
102 root: &Path,
103 index: &BM25Index,
104 engine: &crate::core::embeddings::EmbeddingEngine,
105 aligned_embeddings: &[Vec<f32>],
106 changed_files: &[String],
107 query: &str,
108 top_k: usize,
109 config: &HybridConfig,
110 filter: Option<&dyn Fn(&str) -> bool>,
111 graph_file_ranks: Option<&std::collections::HashMap<String, usize>>,
112) -> Result<Vec<HybridResult>, String> {
113 match backend {
114 DenseBackendKind::Local => {
115 let _ = (root, changed_files);
116 let mut results = crate::core::hybrid_search::hybrid_search(
117 query,
118 index,
119 Some(engine),
120 Some(aligned_embeddings),
121 top_k,
122 config,
123 graph_file_ranks,
124 );
125 if let Some(pred) = filter {
126 results.retain(|r| pred(&r.file_path));
127 }
128 results.truncate(top_k);
129 Ok(results)
130 }
131 #[cfg(feature = "qdrant")]
132 DenseBackendKind::Qdrant => {
133 let bm25_k = config.bm25_candidates.max(top_k);
134 let dense_k = config.dense_candidates.max(top_k);
135
136 let mut bm25 = index.search(query, bm25_k);
137 if let Some(pred) = filter {
138 bm25.retain(|r| pred(&r.file_path));
139 }
140
141 let dense = dense_results(
142 backend,
143 root,
144 index,
145 engine,
146 aligned_embeddings,
147 changed_files,
148 query,
149 dense_k,
150 filter,
151 )?;
152
153 let mut fused = crate::core::hybrid_search::reciprocal_rank_fusion(
154 &bm25,
155 &dense,
156 config,
157 top_k,
158 graph_file_ranks,
159 );
160 if let Some(pred) = filter {
161 fused.retain(|r| pred(&r.file_path));
162 }
163 fused.truncate(top_k);
164 Ok(fused)
165 }
166 }
167}
168
169#[cfg(feature = "embeddings")]
170#[allow(clippy::too_many_arguments)]
171fn dense_results(
172 backend: DenseBackendKind,
173 root: &Path,
174 index: &BM25Index,
175 engine: &crate::core::embeddings::EmbeddingEngine,
176 aligned_embeddings: &[Vec<f32>],
177 changed_files: &[String],
178 query: &str,
179 top_k: usize,
180 filter: Option<&dyn Fn(&str) -> bool>,
181) -> Result<Vec<DenseSearchResult>, String> {
182 match backend {
183 DenseBackendKind::Local => {
184 let _ = (root, changed_files);
185 dense_results_local(index, engine, aligned_embeddings, query, top_k, filter)
186 }
187 #[cfg(feature = "qdrant")]
188 DenseBackendKind::Qdrant => dense_results_qdrant(
189 root,
190 index,
191 engine,
192 aligned_embeddings,
193 changed_files,
194 query,
195 top_k,
196 filter,
197 ),
198 }
199}
200
201#[cfg(feature = "embeddings")]
202fn dense_results_local(
203 index: &BM25Index,
204 engine: &crate::core::embeddings::EmbeddingEngine,
205 aligned_embeddings: &[Vec<f32>],
206 query: &str,
207 top_k: usize,
208 filter: Option<&dyn Fn(&str) -> bool>,
209) -> Result<Vec<DenseSearchResult>, String> {
210 use crate::core::embeddings::cosine_similarity;
211
212 let query_embedding = engine
213 .embed(query)
214 .map_err(|e| format!("embedding failed: {e}"))?;
215
216 let mut scored: Vec<(usize, f32)> = aligned_embeddings
217 .iter()
218 .enumerate()
219 .filter(|(i, _)| {
220 let Some(pred) = filter else { return true };
221 index.chunks.get(*i).is_some_and(|c| pred(&c.file_path))
222 })
223 .map(|(i, emb)| (i, cosine_similarity(&query_embedding, emb)))
224 .collect();
225
226 scored.sort_by(|a, b| b.1.partial_cmp(&a.1).unwrap_or(std::cmp::Ordering::Equal));
227 scored.truncate(top_k);
228
229 Ok(scored
230 .into_iter()
231 .filter_map(|(idx, sim)| {
232 let chunk = index.chunks.get(idx)?;
233 let snippet = chunk.content.lines().take(5).collect::<Vec<_>>().join("\n");
234 Some(DenseSearchResult {
235 chunk_idx: idx,
236 similarity: sim,
237 file_path: chunk.file_path.clone(),
238 symbol_name: chunk.symbol_name.clone(),
239 kind: chunk.kind.clone(),
240 start_line: chunk.start_line,
241 end_line: chunk.end_line,
242 snippet,
243 })
244 })
245 .collect())
246}
247
248#[cfg(feature = "qdrant")]
249#[cfg(feature = "embeddings")]
250fn dense_results_qdrant(
251 root: &Path,
252 index: &BM25Index,
253 engine: &crate::core::embeddings::EmbeddingEngine,
254 aligned_embeddings: &[Vec<f32>],
255 changed_files: &[String],
256 query: &str,
257 top_k: usize,
258 filter: Option<&dyn Fn(&str) -> bool>,
259) -> Result<Vec<DenseSearchResult>, String> {
260 let store = crate::core::qdrant_store::QdrantStore::from_env()?;
261 let collection = store.collection_name(root, engine.dimensions())?;
262 let created_new = store.ensure_collection(&collection, engine.dimensions())?;
263 store.sync_index(
264 &collection,
265 index,
266 aligned_embeddings,
267 changed_files,
268 created_new,
269 )?;
270
271 let query_vec = engine
272 .embed(query)
273 .map_err(|e| format!("embedding failed: {e}"))?;
274
275 let hits = store.search(&collection, &query_vec, top_k)?;
276 let mut out = Vec::with_capacity(hits.len());
277 for hit in hits {
278 if let Some(pred) = filter {
279 if !pred(&hit.file_path) {
280 continue;
281 }
282 }
283 let snippet = snippet_from_disk(root, &hit.file_path, hit.start_line, hit.end_line, 5);
284 out.push(DenseSearchResult {
285 chunk_idx: 0,
286 similarity: hit.score,
287 file_path: hit.file_path,
288 symbol_name: hit.symbol_name,
289 kind: hit.kind,
290 start_line: hit.start_line,
291 end_line: hit.end_line,
292 snippet,
293 });
294 }
295 Ok(out)
296}
297
298#[cfg(feature = "qdrant")]
299fn snippet_from_disk(
300 root: &Path,
301 rel_path: &str,
302 start_line: usize,
303 end_line: usize,
304 max_lines: usize,
305) -> String {
306 let Ok(path) = crate::core::pathjail::jail_path(&root.join(rel_path), root) else {
307 return String::new();
308 };
309 let Ok(content) = std::fs::read_to_string(path) else {
310 return String::new();
311 };
312 let lines: Vec<&str> = content.lines().collect();
313 if lines.is_empty() {
314 return String::new();
315 }
316 let start = start_line.saturating_sub(1).min(lines.len());
317 let end = end_line.max(start_line).min(lines.len());
318 let mut slice = &lines[start..end];
319 if slice.len() > max_lines {
320 slice = &slice[..max_lines];
321 }
322 slice.join("\n")
323}
324
325#[cfg(feature = "qdrant")]
326fn chunk_kind_str(kind: &ChunkKind) -> &'static str {
327 match kind {
328 ChunkKind::Function => "Function",
329 ChunkKind::Struct => "Struct",
330 ChunkKind::Impl => "Impl",
331 ChunkKind::Module => "Module",
332 ChunkKind::Class => "Class",
333 ChunkKind::Method => "Method",
334 ChunkKind::Other => "Other",
335 }
336}
337
338#[cfg(feature = "qdrant")]
339pub(crate) fn kind_from_str(s: &str) -> ChunkKind {
340 match s {
341 "Function" => ChunkKind::Function,
342 "Struct" => ChunkKind::Struct,
343 "Impl" => ChunkKind::Impl,
344 "Module" => ChunkKind::Module,
345 "Class" => ChunkKind::Class,
346 "Method" => ChunkKind::Method,
347 _ => ChunkKind::Other,
348 }
349}
350
351#[cfg(feature = "qdrant")]
352pub(crate) fn kind_to_str(kind: &ChunkKind) -> &'static str {
353 chunk_kind_str(kind)
354}
355
356#[cfg(test)]
357mod tests {
358 use super::*;
359 use std::sync::Mutex;
360
361 static ENV_LOCK: Mutex<()> = Mutex::new(());
362
363 fn set_env(key: &str, value: Option<&str>) -> Option<String> {
364 let old = std::env::var(key).ok();
365 match value {
366 Some(v) => std::env::set_var(key, v),
367 None => std::env::remove_var(key),
368 }
369 old
370 }
371
372 fn restore_env(key: &str, old: Option<String>) {
373 match old {
374 Some(v) => std::env::set_var(key, v),
375 None => std::env::remove_var(key),
376 }
377 }
378
379 #[test]
380 fn dense_backend_defaults_to_local() {
381 let _g = ENV_LOCK.lock().unwrap();
382 let old_backend = set_env("LEANCTX_DENSE_BACKEND", None);
383 let old_url = set_env("LEANCTX_QDRANT_URL", None);
384
385 let got = DenseBackendKind::try_from_env().unwrap();
386 assert_eq!(got, DenseBackendKind::Local);
387
388 restore_env("LEANCTX_DENSE_BACKEND", old_backend);
389 restore_env("LEANCTX_QDRANT_URL", old_url);
390 }
391
392 #[test]
393 fn dense_backend_unknown_value_errors() {
394 let _g = ENV_LOCK.lock().unwrap();
395 let old_backend = set_env("LEANCTX_DENSE_BACKEND", Some("wat"));
396 let old_url = set_env("LEANCTX_QDRANT_URL", None);
397
398 let err = DenseBackendKind::try_from_env().unwrap_err();
399 assert!(err.contains("Unknown LEANCTX_DENSE_BACKEND"));
400
401 restore_env("LEANCTX_DENSE_BACKEND", old_backend);
402 restore_env("LEANCTX_QDRANT_URL", old_url);
403 }
404
405 #[cfg(feature = "qdrant")]
406 #[test]
407 fn dense_backend_infers_qdrant_from_url() {
408 let _g = ENV_LOCK.lock().unwrap();
409 let old_backend = set_env("LEANCTX_DENSE_BACKEND", None);
410 let old_url = set_env("LEANCTX_QDRANT_URL", Some("http://127.0.0.1:6333"));
411
412 let got = DenseBackendKind::try_from_env().unwrap();
413 assert_eq!(got, DenseBackendKind::Qdrant);
414
415 restore_env("LEANCTX_DENSE_BACKEND", old_backend);
416 restore_env("LEANCTX_QDRANT_URL", old_url);
417 }
418
419 #[cfg(not(feature = "qdrant"))]
420 #[test]
421 fn dense_backend_qdrant_requires_feature() {
422 let _g = ENV_LOCK.lock().unwrap();
423 let old_backend = set_env("LEANCTX_DENSE_BACKEND", Some("qdrant"));
424 let old_url = set_env("LEANCTX_QDRANT_URL", None);
425
426 let err = DenseBackendKind::try_from_env().unwrap_err();
427 assert!(err.contains("feature 'qdrant' is not enabled"));
428
429 restore_env("LEANCTX_DENSE_BACKEND", old_backend);
430 restore_env("LEANCTX_QDRANT_URL", old_url);
431 }
432}