1use serde::{Deserialize, Serialize};
2use std::path::{Path, PathBuf};
3use thiserror::Error;
4
5#[derive(Debug, Error)]
6pub enum CkError {
7 #[error("IO error: {0}")]
8 Io(#[from] std::io::Error),
9
10 #[error("Regex error: {0}")]
11 Regex(#[from] regex::Error),
12
13 #[error("Serialization error: {0}")]
14 Serialization(#[from] bincode::Error),
15
16 #[error("JSON error: {0}")]
17 Json(#[from] serde_json::Error),
18
19 #[error("Index error: {0}")]
20 Index(String),
21
22 #[error("Search error: {0}")]
23 Search(String),
24
25 #[error("Embedding error: {0}")]
26 Embedding(String),
27
28 #[error("Other error: {0}")]
29 Other(String),
30}
31
32pub type Result<T> = std::result::Result<T, CkError>;
33
34#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
35pub enum Language {
36 Rust,
37 Python,
38 JavaScript,
39 TypeScript,
40 Haskell,
41 Go,
42 Java,
43 C,
44 Cpp,
45 CSharp,
46 Ruby,
47 Php,
48 Swift,
49 Kotlin,
50}
51
52impl Language {
53 pub fn from_extension(ext: &str) -> Option<Self> {
54 match ext {
55 "rs" => Some(Language::Rust),
56 "py" => Some(Language::Python),
57 "js" => Some(Language::JavaScript),
58 "ts" | "tsx" => Some(Language::TypeScript),
59 "hs" | "lhs" => Some(Language::Haskell),
60 "go" => Some(Language::Go),
61 "java" => Some(Language::Java),
62 "c" => Some(Language::C),
63 "cpp" | "cc" | "cxx" | "c++" => Some(Language::Cpp),
64 "h" | "hpp" => Some(Language::Cpp), "cs" => Some(Language::CSharp),
66 "rb" => Some(Language::Ruby),
67 "php" => Some(Language::Php),
68 "swift" => Some(Language::Swift),
69 "kt" | "kts" => Some(Language::Kotlin),
70 _ => None,
71 }
72 }
73
74 pub fn from_path(path: &Path) -> Option<Self> {
75 path.extension()
76 .and_then(|ext| ext.to_str())
77 .and_then(Self::from_extension)
78 }
79}
80
81impl std::fmt::Display for Language {
82 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
83 let name = match self {
84 Language::Rust => "rust",
85 Language::Python => "python",
86 Language::JavaScript => "javascript",
87 Language::TypeScript => "typescript",
88 Language::Haskell => "haskell",
89 Language::Go => "go",
90 Language::Java => "java",
91 Language::C => "c",
92 Language::Cpp => "cpp",
93 Language::CSharp => "csharp",
94 Language::Ruby => "ruby",
95 Language::Php => "php",
96 Language::Swift => "swift",
97 Language::Kotlin => "kotlin",
98 };
99 write!(f, "{}", name)
100 }
101}
102
103#[derive(Debug, Clone, Serialize, Deserialize)]
104pub struct Span {
105 pub byte_start: usize,
106 pub byte_end: usize,
107 pub line_start: usize,
108 pub line_end: usize,
109}
110
111#[derive(Debug, Clone, Serialize, Deserialize)]
112pub struct FileMetadata {
113 pub path: PathBuf,
114 pub hash: String,
115 pub last_modified: u64,
116 pub size: u64,
117}
118
119#[derive(Debug, Clone, Serialize, Deserialize)]
120pub struct SearchResult {
121 pub file: PathBuf,
122 pub span: Span,
123 pub score: f32,
124 pub preview: String,
125 #[serde(skip_serializing_if = "Option::is_none")]
126 pub lang: Option<Language>,
127 #[serde(skip_serializing_if = "Option::is_none")]
128 pub symbol: Option<String>,
129 #[serde(skip_serializing_if = "Option::is_none")]
130 pub chunk_hash: Option<String>,
131 #[serde(skip_serializing_if = "Option::is_none")]
132 pub index_epoch: Option<u64>,
133}
134
135#[derive(Debug, Clone, Serialize, Deserialize)]
136pub struct JsonSearchResult {
137 pub file: String,
138 pub span: Span,
139 pub lang: Option<Language>,
140 pub symbol: Option<String>,
141 pub score: f32,
142 pub signals: SearchSignals,
143 pub preview: String,
144 pub model: String,
145}
146
147#[derive(Debug, Clone, Serialize, Deserialize)]
148pub struct JsonlSearchResult {
149 pub path: String,
150 pub span: Span,
151 pub language: Option<String>,
152 #[serde(skip_serializing_if = "Option::is_none")]
153 pub snippet: Option<String>,
154 #[serde(skip_serializing_if = "Option::is_none")]
155 pub score: Option<f32>,
156 #[serde(skip_serializing_if = "Option::is_none")]
157 pub chunk_hash: Option<String>,
158 #[serde(skip_serializing_if = "Option::is_none")]
159 pub index_epoch: Option<u64>,
160}
161
162#[derive(Debug, Clone, Serialize, Deserialize)]
163pub struct SearchSignals {
164 pub lex_rank: Option<usize>,
165 pub vec_rank: Option<usize>,
166 pub rrf_score: f32,
167}
168
169#[derive(Debug, Clone, PartialEq)]
170pub enum SearchMode {
171 Regex,
172 Lexical,
173 Semantic,
174 Hybrid,
175}
176
177#[derive(Debug, Clone)]
178pub struct SearchOptions {
179 pub mode: SearchMode,
180 pub query: String,
181 pub path: PathBuf,
182 pub top_k: Option<usize>,
183 pub threshold: Option<f32>,
184 pub case_insensitive: bool,
185 pub whole_word: bool,
186 pub fixed_string: bool,
187 pub line_numbers: bool,
188 pub context_lines: usize,
189 pub before_context_lines: usize,
190 pub after_context_lines: usize,
191 pub recursive: bool,
192 pub json_output: bool,
193 pub jsonl_output: bool,
194 pub no_snippet: bool,
195 pub reindex: bool,
196 pub show_scores: bool,
197 pub show_filenames: bool,
198 pub files_with_matches: bool,
199 pub files_without_matches: bool,
200 pub exclude_patterns: Vec<String>,
201 pub respect_gitignore: bool,
202 pub full_section: bool,
203 pub rerank: bool,
205 pub rerank_model: Option<String>,
206}
207
208impl JsonlSearchResult {
209 pub fn from_search_result(result: &SearchResult, include_snippet: bool) -> Self {
210 Self {
211 path: result.file.to_string_lossy().to_string(),
212 span: result.span.clone(),
213 language: result.lang.as_ref().map(|l| l.to_string()),
214 snippet: if include_snippet {
215 Some(result.preview.clone())
216 } else {
217 None
218 },
219 score: if result.score >= 0.0 {
220 Some(result.score)
221 } else {
222 None
223 },
224 chunk_hash: result.chunk_hash.clone(),
225 index_epoch: result.index_epoch,
226 }
227 }
228}
229
230impl Default for SearchOptions {
231 fn default() -> Self {
232 Self {
233 mode: SearchMode::Regex,
234 query: String::new(),
235 path: PathBuf::from("."),
236 top_k: None,
237 threshold: None,
238 case_insensitive: false,
239 whole_word: false,
240 fixed_string: false,
241 line_numbers: false,
242 context_lines: 0,
243 before_context_lines: 0,
244 after_context_lines: 0,
245 recursive: true,
246 json_output: false,
247 jsonl_output: false,
248 no_snippet: false,
249 reindex: false,
250 show_scores: false,
251 show_filenames: false,
252 files_with_matches: false,
253 files_without_matches: false,
254 exclude_patterns: get_default_exclude_patterns(),
255 respect_gitignore: true,
256 full_section: false,
257 rerank: false,
259 rerank_model: None,
260 }
261 }
262}
263
264pub fn get_default_exclude_patterns() -> Vec<String> {
267 vec![
268 ".ck".to_string(),
270 ".fastembed_cache".to_string(),
272 ".cache".to_string(),
273 "__pycache__".to_string(),
274 ".git".to_string(),
276 ".svn".to_string(),
277 ".hg".to_string(),
278 "target".to_string(), "build".to_string(), "dist".to_string(), "node_modules".to_string(), ".gradle".to_string(), ".mvn".to_string(), "bin".to_string(), "obj".to_string(), "venv".to_string(),
289 ".venv".to_string(),
290 "env".to_string(),
291 ".env".to_string(),
292 "virtualenv".to_string(),
293 ".vscode".to_string(),
295 ".idea".to_string(),
296 ".eclipse".to_string(),
297 "tmp".to_string(),
299 "temp".to_string(),
300 ".tmp".to_string(),
301 ]
302}
303
304pub fn get_sidecar_path(repo_root: &Path, file_path: &Path) -> PathBuf {
305 let relative = file_path.strip_prefix(repo_root).unwrap_or(file_path);
306 let mut sidecar = repo_root.join(".ck");
307 sidecar.push(relative);
308 let ext = relative
309 .extension()
310 .map(|e| format!("{}.ck", e.to_string_lossy()))
311 .unwrap_or_else(|| "ck".to_string());
312 sidecar.set_extension(ext);
313 sidecar
314}
315
316pub fn compute_file_hash(path: &Path) -> Result<String> {
317 let data = std::fs::read(path)?;
318 let hash = blake3::hash(&data);
319 Ok(hash.to_hex().to_string())
320}
321
322#[cfg(test)]
323mod tests {
324 use super::*;
325 use std::fs;
326 use tempfile::TempDir;
327
328 #[test]
329 fn test_span_creation() {
330 let span = Span {
331 byte_start: 0,
332 byte_end: 10,
333 line_start: 1,
334 line_end: 2,
335 };
336
337 assert_eq!(span.byte_start, 0);
338 assert_eq!(span.byte_end, 10);
339 assert_eq!(span.line_start, 1);
340 assert_eq!(span.line_end, 2);
341 }
342
343 #[test]
344 fn test_search_options_default() {
345 let options = SearchOptions::default();
346 assert!(matches!(options.mode, SearchMode::Regex));
347 assert_eq!(options.query, "");
348 assert_eq!(options.path, PathBuf::from("."));
349 assert_eq!(options.top_k, None);
350 assert_eq!(options.threshold, None);
351 assert!(!options.case_insensitive);
352 assert!(!options.whole_word);
353 assert!(!options.fixed_string);
354 assert!(!options.line_numbers);
355 assert_eq!(options.context_lines, 0);
356 assert!(options.recursive);
357 assert!(!options.json_output);
358 assert!(!options.reindex);
359 assert!(!options.show_scores);
360 assert!(!options.show_filenames);
361 }
362
363 #[test]
364 fn test_file_metadata_serialization() {
365 let metadata = FileMetadata {
366 path: PathBuf::from("test.txt"),
367 hash: "abc123".to_string(),
368 last_modified: 1234567890,
369 size: 1024,
370 };
371
372 let json = serde_json::to_string(&metadata).unwrap();
373 let deserialized: FileMetadata = serde_json::from_str(&json).unwrap();
374
375 assert_eq!(metadata.path, deserialized.path);
376 assert_eq!(metadata.hash, deserialized.hash);
377 assert_eq!(metadata.last_modified, deserialized.last_modified);
378 assert_eq!(metadata.size, deserialized.size);
379 }
380
381 #[test]
382 fn test_search_result_serialization() {
383 let result = SearchResult {
384 file: PathBuf::from("test.txt"),
385 span: Span {
386 byte_start: 0,
387 byte_end: 10,
388 line_start: 1,
389 line_end: 1,
390 },
391 score: 0.95,
392 preview: "hello world".to_string(),
393 lang: Some(Language::Rust),
394 symbol: Some("main".to_string()),
395 chunk_hash: Some("abc123".to_string()),
396 index_epoch: Some(1699123456),
397 };
398
399 let json = serde_json::to_string(&result).unwrap();
400 let deserialized: SearchResult = serde_json::from_str(&json).unwrap();
401
402 assert_eq!(result.file, deserialized.file);
403 assert_eq!(result.score, deserialized.score);
404 assert_eq!(result.preview, deserialized.preview);
405 assert_eq!(result.lang, deserialized.lang);
406 assert_eq!(result.symbol, deserialized.symbol);
407 assert_eq!(result.chunk_hash, deserialized.chunk_hash);
408 assert_eq!(result.index_epoch, deserialized.index_epoch);
409 }
410
411 #[test]
412 fn test_jsonl_search_result_conversion() {
413 let result = SearchResult {
414 file: PathBuf::from("src/auth.rs"),
415 span: Span {
416 byte_start: 1203,
417 byte_end: 1456,
418 line_start: 42,
419 line_end: 58,
420 },
421 score: 0.89,
422 preview: "function authenticate(user) {...}".to_string(),
423 lang: Some(Language::Rust),
424 symbol: Some("authenticate".to_string()),
425 chunk_hash: Some("abc123def456".to_string()),
426 index_epoch: Some(1699123456),
427 };
428
429 let jsonl_with_snippet = JsonlSearchResult::from_search_result(&result, true);
431 assert_eq!(jsonl_with_snippet.path, "src/auth.rs");
432 assert_eq!(jsonl_with_snippet.span.line_start, 42);
433 assert_eq!(jsonl_with_snippet.language, Some("rust".to_string()));
434 assert_eq!(
435 jsonl_with_snippet.snippet,
436 Some("function authenticate(user) {...}".to_string())
437 );
438 assert_eq!(jsonl_with_snippet.score, Some(0.89));
439 assert_eq!(
440 jsonl_with_snippet.chunk_hash,
441 Some("abc123def456".to_string())
442 );
443 assert_eq!(jsonl_with_snippet.index_epoch, Some(1699123456));
444
445 let jsonl_no_snippet = JsonlSearchResult::from_search_result(&result, false);
447 assert_eq!(jsonl_no_snippet.snippet, None);
448 assert_eq!(jsonl_no_snippet.path, "src/auth.rs");
449 }
450
451 #[test]
452 fn test_get_sidecar_path() {
453 let repo_root = PathBuf::from("/home/user/project");
454 let file_path = PathBuf::from("/home/user/project/src/main.rs");
455
456 let sidecar = get_sidecar_path(&repo_root, &file_path);
457 let expected = PathBuf::from("/home/user/project/.ck/src/main.rs.ck");
458
459 assert_eq!(sidecar, expected);
460 }
461
462 #[test]
463 fn test_get_sidecar_path_no_extension() {
464 let repo_root = PathBuf::from("/project");
465 let file_path = PathBuf::from("/project/README");
466
467 let sidecar = get_sidecar_path(&repo_root, &file_path);
468 let expected = PathBuf::from("/project/.ck/README.ck");
469
470 assert_eq!(sidecar, expected);
471 }
472
473 #[test]
474 fn test_compute_file_hash() {
475 let temp_dir = TempDir::new().unwrap();
476 let file_path = temp_dir.path().join("test.txt");
477
478 fs::write(&file_path, "hello world").unwrap();
479
480 let hash1 = compute_file_hash(&file_path).unwrap();
481 let hash2 = compute_file_hash(&file_path).unwrap();
482
483 assert_eq!(hash1, hash2);
485 assert!(!hash1.is_empty());
486
487 fs::write(&file_path, "hello rust").unwrap();
489 let hash3 = compute_file_hash(&file_path).unwrap();
490 assert_ne!(hash1, hash3);
491 }
492
493 #[test]
494 fn test_compute_file_hash_nonexistent() {
495 let result = compute_file_hash(&PathBuf::from("nonexistent.txt"));
496 assert!(result.is_err());
497 }
498
499 #[test]
500 fn test_json_search_result_serialization() {
501 let signals = SearchSignals {
502 lex_rank: Some(1),
503 vec_rank: Some(2),
504 rrf_score: 0.85,
505 };
506
507 let result = JsonSearchResult {
508 file: "test.txt".to_string(),
509 span: Span {
510 byte_start: 0,
511 byte_end: 5,
512 line_start: 1,
513 line_end: 1,
514 },
515 lang: None, symbol: None,
517 score: 0.95,
518 signals,
519 preview: "hello".to_string(),
520 model: "bge-small".to_string(),
521 };
522
523 let json = serde_json::to_string(&result).unwrap();
524 let deserialized: JsonSearchResult = serde_json::from_str(&json).unwrap();
525
526 assert_eq!(result.file, deserialized.file);
527 assert_eq!(result.score, deserialized.score);
528 assert_eq!(result.signals.rrf_score, deserialized.signals.rrf_score);
529 assert_eq!(result.model, deserialized.model);
530 }
531
532 #[test]
533 fn test_language_from_extension() {
534 assert_eq!(Language::from_extension("rs"), Some(Language::Rust));
535 assert_eq!(Language::from_extension("py"), Some(Language::Python));
536 assert_eq!(Language::from_extension("js"), Some(Language::JavaScript));
537 assert_eq!(Language::from_extension("ts"), Some(Language::TypeScript));
538 assert_eq!(Language::from_extension("tsx"), Some(Language::TypeScript));
539 assert_eq!(Language::from_extension("hs"), Some(Language::Haskell));
540 assert_eq!(Language::from_extension("lhs"), Some(Language::Haskell));
541 assert_eq!(Language::from_extension("go"), Some(Language::Go));
542 assert_eq!(Language::from_extension("java"), Some(Language::Java));
543 assert_eq!(Language::from_extension("c"), Some(Language::C));
544 assert_eq!(Language::from_extension("cpp"), Some(Language::Cpp));
545 assert_eq!(Language::from_extension("cs"), Some(Language::CSharp));
546 assert_eq!(Language::from_extension("rb"), Some(Language::Ruby));
547 assert_eq!(Language::from_extension("php"), Some(Language::Php));
548 assert_eq!(Language::from_extension("swift"), Some(Language::Swift));
549 assert_eq!(Language::from_extension("kt"), Some(Language::Kotlin));
550 assert_eq!(Language::from_extension("kts"), Some(Language::Kotlin));
551 assert_eq!(Language::from_extension("unknown"), None);
552 }
553
554 #[test]
555 fn test_language_from_path() {
556 assert_eq!(
557 Language::from_path(&PathBuf::from("test.rs")),
558 Some(Language::Rust)
559 );
560 assert_eq!(
561 Language::from_path(&PathBuf::from("test.py")),
562 Some(Language::Python)
563 );
564 assert_eq!(
565 Language::from_path(&PathBuf::from("test.js")),
566 Some(Language::JavaScript)
567 );
568 assert_eq!(
569 Language::from_path(&PathBuf::from("test.hs")),
570 Some(Language::Haskell)
571 );
572 assert_eq!(
573 Language::from_path(&PathBuf::from("test.lhs")),
574 Some(Language::Haskell)
575 );
576 assert_eq!(
577 Language::from_path(&PathBuf::from("test.go")),
578 Some(Language::Go)
579 );
580 assert_eq!(Language::from_path(&PathBuf::from("test.unknown")), None); assert_eq!(Language::from_path(&PathBuf::from("noext")), None); }
583
584 #[test]
585 fn test_language_display() {
586 assert_eq!(Language::Rust.to_string(), "rust");
587 assert_eq!(Language::Python.to_string(), "python");
588 assert_eq!(Language::JavaScript.to_string(), "javascript");
589 assert_eq!(Language::TypeScript.to_string(), "typescript");
590 assert_eq!(Language::Go.to_string(), "go");
591 assert_eq!(Language::Java.to_string(), "java");
592 }
593}