ck_core/
lib.rs

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), // Assume C++ for headers
65            "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}
204
205impl JsonlSearchResult {
206    pub fn from_search_result(result: &SearchResult, include_snippet: bool) -> Self {
207        Self {
208            path: result.file.to_string_lossy().to_string(),
209            span: result.span.clone(),
210            language: result.lang.as_ref().map(|l| l.to_string()),
211            snippet: if include_snippet {
212                Some(result.preview.clone())
213            } else {
214                None
215            },
216            score: if result.score >= 0.0 {
217                Some(result.score)
218            } else {
219                None
220            },
221            chunk_hash: result.chunk_hash.clone(),
222            index_epoch: result.index_epoch,
223        }
224    }
225}
226
227impl Default for SearchOptions {
228    fn default() -> Self {
229        Self {
230            mode: SearchMode::Regex,
231            query: String::new(),
232            path: PathBuf::from("."),
233            top_k: None,
234            threshold: None,
235            case_insensitive: false,
236            whole_word: false,
237            fixed_string: false,
238            line_numbers: false,
239            context_lines: 0,
240            before_context_lines: 0,
241            after_context_lines: 0,
242            recursive: true,
243            json_output: false,
244            jsonl_output: false,
245            no_snippet: false,
246            reindex: false,
247            show_scores: false,
248            show_filenames: false,
249            files_with_matches: false,
250            files_without_matches: false,
251            exclude_patterns: get_default_exclude_patterns(),
252            respect_gitignore: true,
253            full_section: false,
254        }
255    }
256}
257
258/// Get default exclusion patterns for directories that should be skipped during search.
259/// These are common cache, build, and system directories that rarely contain user code.
260pub fn get_default_exclude_patterns() -> Vec<String> {
261    vec![
262        // ck's own index directory
263        ".ck".to_string(),
264        // AI/ML model cache directories
265        ".fastembed_cache".to_string(),
266        ".cache".to_string(),
267        "__pycache__".to_string(),
268        // Version control
269        ".git".to_string(),
270        ".svn".to_string(),
271        ".hg".to_string(),
272        // Build directories
273        "target".to_string(),       // Rust
274        "build".to_string(),        // Various
275        "dist".to_string(),         // JavaScript/Python
276        "node_modules".to_string(), // JavaScript
277        ".gradle".to_string(),      // Java
278        ".mvn".to_string(),         // Maven
279        "bin".to_string(),          // Various
280        "obj".to_string(),          // .NET
281        // Python virtual environments
282        "venv".to_string(),
283        ".venv".to_string(),
284        "env".to_string(),
285        ".env".to_string(),
286        "virtualenv".to_string(),
287        // IDE/Editor directories
288        ".vscode".to_string(),
289        ".idea".to_string(),
290        ".eclipse".to_string(),
291        // Temporary directories
292        "tmp".to_string(),
293        "temp".to_string(),
294        ".tmp".to_string(),
295    ]
296}
297
298pub fn get_sidecar_path(repo_root: &Path, file_path: &Path) -> PathBuf {
299    let relative = file_path.strip_prefix(repo_root).unwrap_or(file_path);
300    let mut sidecar = repo_root.join(".ck");
301    sidecar.push(relative);
302    let ext = relative
303        .extension()
304        .map(|e| format!("{}.ck", e.to_string_lossy()))
305        .unwrap_or_else(|| "ck".to_string());
306    sidecar.set_extension(ext);
307    sidecar
308}
309
310pub fn compute_file_hash(path: &Path) -> Result<String> {
311    let data = std::fs::read(path)?;
312    let hash = blake3::hash(&data);
313    Ok(hash.to_hex().to_string())
314}
315
316#[cfg(test)]
317mod tests {
318    use super::*;
319    use std::fs;
320    use tempfile::TempDir;
321
322    #[test]
323    fn test_span_creation() {
324        let span = Span {
325            byte_start: 0,
326            byte_end: 10,
327            line_start: 1,
328            line_end: 2,
329        };
330
331        assert_eq!(span.byte_start, 0);
332        assert_eq!(span.byte_end, 10);
333        assert_eq!(span.line_start, 1);
334        assert_eq!(span.line_end, 2);
335    }
336
337    #[test]
338    fn test_search_options_default() {
339        let options = SearchOptions::default();
340        assert!(matches!(options.mode, SearchMode::Regex));
341        assert_eq!(options.query, "");
342        assert_eq!(options.path, PathBuf::from("."));
343        assert_eq!(options.top_k, None);
344        assert_eq!(options.threshold, None);
345        assert!(!options.case_insensitive);
346        assert!(!options.whole_word);
347        assert!(!options.fixed_string);
348        assert!(!options.line_numbers);
349        assert_eq!(options.context_lines, 0);
350        assert!(options.recursive);
351        assert!(!options.json_output);
352        assert!(!options.reindex);
353        assert!(!options.show_scores);
354        assert!(!options.show_filenames);
355    }
356
357    #[test]
358    fn test_file_metadata_serialization() {
359        let metadata = FileMetadata {
360            path: PathBuf::from("test.txt"),
361            hash: "abc123".to_string(),
362            last_modified: 1234567890,
363            size: 1024,
364        };
365
366        let json = serde_json::to_string(&metadata).unwrap();
367        let deserialized: FileMetadata = serde_json::from_str(&json).unwrap();
368
369        assert_eq!(metadata.path, deserialized.path);
370        assert_eq!(metadata.hash, deserialized.hash);
371        assert_eq!(metadata.last_modified, deserialized.last_modified);
372        assert_eq!(metadata.size, deserialized.size);
373    }
374
375    #[test]
376    fn test_search_result_serialization() {
377        let result = SearchResult {
378            file: PathBuf::from("test.txt"),
379            span: Span {
380                byte_start: 0,
381                byte_end: 10,
382                line_start: 1,
383                line_end: 1,
384            },
385            score: 0.95,
386            preview: "hello world".to_string(),
387            lang: Some(Language::Rust),
388            symbol: Some("main".to_string()),
389            chunk_hash: Some("abc123".to_string()),
390            index_epoch: Some(1699123456),
391        };
392
393        let json = serde_json::to_string(&result).unwrap();
394        let deserialized: SearchResult = serde_json::from_str(&json).unwrap();
395
396        assert_eq!(result.file, deserialized.file);
397        assert_eq!(result.score, deserialized.score);
398        assert_eq!(result.preview, deserialized.preview);
399        assert_eq!(result.lang, deserialized.lang);
400        assert_eq!(result.symbol, deserialized.symbol);
401        assert_eq!(result.chunk_hash, deserialized.chunk_hash);
402        assert_eq!(result.index_epoch, deserialized.index_epoch);
403    }
404
405    #[test]
406    fn test_jsonl_search_result_conversion() {
407        let result = SearchResult {
408            file: PathBuf::from("src/auth.rs"),
409            span: Span {
410                byte_start: 1203,
411                byte_end: 1456,
412                line_start: 42,
413                line_end: 58,
414            },
415            score: 0.89,
416            preview: "function authenticate(user) {...}".to_string(),
417            lang: Some(Language::Rust),
418            symbol: Some("authenticate".to_string()),
419            chunk_hash: Some("abc123def456".to_string()),
420            index_epoch: Some(1699123456),
421        };
422
423        // Test with snippet
424        let jsonl_with_snippet = JsonlSearchResult::from_search_result(&result, true);
425        assert_eq!(jsonl_with_snippet.path, "src/auth.rs");
426        assert_eq!(jsonl_with_snippet.span.line_start, 42);
427        assert_eq!(jsonl_with_snippet.language, Some("rust".to_string()));
428        assert_eq!(
429            jsonl_with_snippet.snippet,
430            Some("function authenticate(user) {...}".to_string())
431        );
432        assert_eq!(jsonl_with_snippet.score, Some(0.89));
433        assert_eq!(
434            jsonl_with_snippet.chunk_hash,
435            Some("abc123def456".to_string())
436        );
437        assert_eq!(jsonl_with_snippet.index_epoch, Some(1699123456));
438
439        // Test without snippet
440        let jsonl_no_snippet = JsonlSearchResult::from_search_result(&result, false);
441        assert_eq!(jsonl_no_snippet.snippet, None);
442        assert_eq!(jsonl_no_snippet.path, "src/auth.rs");
443    }
444
445    #[test]
446    fn test_get_sidecar_path() {
447        let repo_root = PathBuf::from("/home/user/project");
448        let file_path = PathBuf::from("/home/user/project/src/main.rs");
449
450        let sidecar = get_sidecar_path(&repo_root, &file_path);
451        let expected = PathBuf::from("/home/user/project/.ck/src/main.rs.ck");
452
453        assert_eq!(sidecar, expected);
454    }
455
456    #[test]
457    fn test_get_sidecar_path_no_extension() {
458        let repo_root = PathBuf::from("/project");
459        let file_path = PathBuf::from("/project/README");
460
461        let sidecar = get_sidecar_path(&repo_root, &file_path);
462        let expected = PathBuf::from("/project/.ck/README.ck");
463
464        assert_eq!(sidecar, expected);
465    }
466
467    #[test]
468    fn test_compute_file_hash() {
469        let temp_dir = TempDir::new().unwrap();
470        let file_path = temp_dir.path().join("test.txt");
471
472        fs::write(&file_path, "hello world").unwrap();
473
474        let hash1 = compute_file_hash(&file_path).unwrap();
475        let hash2 = compute_file_hash(&file_path).unwrap();
476
477        // Same content should produce same hash
478        assert_eq!(hash1, hash2);
479        assert!(!hash1.is_empty());
480
481        // Different content should produce different hash
482        fs::write(&file_path, "hello rust").unwrap();
483        let hash3 = compute_file_hash(&file_path).unwrap();
484        assert_ne!(hash1, hash3);
485    }
486
487    #[test]
488    fn test_compute_file_hash_nonexistent() {
489        let result = compute_file_hash(&PathBuf::from("nonexistent.txt"));
490        assert!(result.is_err());
491    }
492
493    #[test]
494    fn test_json_search_result_serialization() {
495        let signals = SearchSignals {
496            lex_rank: Some(1),
497            vec_rank: Some(2),
498            rrf_score: 0.85,
499        };
500
501        let result = JsonSearchResult {
502            file: "test.txt".to_string(),
503            span: Span {
504                byte_start: 0,
505                byte_end: 5,
506                line_start: 1,
507                line_end: 1,
508            },
509            lang: None, // txt is not a supported language
510            symbol: None,
511            score: 0.95,
512            signals,
513            preview: "hello".to_string(),
514            model: "bge-small".to_string(),
515        };
516
517        let json = serde_json::to_string(&result).unwrap();
518        let deserialized: JsonSearchResult = serde_json::from_str(&json).unwrap();
519
520        assert_eq!(result.file, deserialized.file);
521        assert_eq!(result.score, deserialized.score);
522        assert_eq!(result.signals.rrf_score, deserialized.signals.rrf_score);
523        assert_eq!(result.model, deserialized.model);
524    }
525
526    #[test]
527    fn test_language_from_extension() {
528        assert_eq!(Language::from_extension("rs"), Some(Language::Rust));
529        assert_eq!(Language::from_extension("py"), Some(Language::Python));
530        assert_eq!(Language::from_extension("js"), Some(Language::JavaScript));
531        assert_eq!(Language::from_extension("ts"), Some(Language::TypeScript));
532        assert_eq!(Language::from_extension("tsx"), Some(Language::TypeScript));
533        assert_eq!(Language::from_extension("hs"), Some(Language::Haskell));
534        assert_eq!(Language::from_extension("lhs"), Some(Language::Haskell));
535        assert_eq!(Language::from_extension("go"), Some(Language::Go));
536        assert_eq!(Language::from_extension("java"), Some(Language::Java));
537        assert_eq!(Language::from_extension("c"), Some(Language::C));
538        assert_eq!(Language::from_extension("cpp"), Some(Language::Cpp));
539        assert_eq!(Language::from_extension("cs"), Some(Language::CSharp));
540        assert_eq!(Language::from_extension("rb"), Some(Language::Ruby));
541        assert_eq!(Language::from_extension("php"), Some(Language::Php));
542        assert_eq!(Language::from_extension("swift"), Some(Language::Swift));
543        assert_eq!(Language::from_extension("kt"), Some(Language::Kotlin));
544        assert_eq!(Language::from_extension("kts"), Some(Language::Kotlin));
545        assert_eq!(Language::from_extension("unknown"), None);
546    }
547
548    #[test]
549    fn test_language_from_path() {
550        assert_eq!(
551            Language::from_path(&PathBuf::from("test.rs")),
552            Some(Language::Rust)
553        );
554        assert_eq!(
555            Language::from_path(&PathBuf::from("test.py")),
556            Some(Language::Python)
557        );
558        assert_eq!(
559            Language::from_path(&PathBuf::from("test.js")),
560            Some(Language::JavaScript)
561        );
562        assert_eq!(
563            Language::from_path(&PathBuf::from("test.hs")),
564            Some(Language::Haskell)
565        );
566        assert_eq!(
567            Language::from_path(&PathBuf::from("test.lhs")),
568            Some(Language::Haskell)
569        );
570        assert_eq!(
571            Language::from_path(&PathBuf::from("test.go")),
572            Some(Language::Go)
573        );
574        assert_eq!(Language::from_path(&PathBuf::from("test.unknown")), None); // unknown extensions return None
575        assert_eq!(Language::from_path(&PathBuf::from("noext")), None); // no extension
576    }
577
578    #[test]
579    fn test_language_display() {
580        assert_eq!(Language::Rust.to_string(), "rust");
581        assert_eq!(Language::Python.to_string(), "python");
582        assert_eq!(Language::JavaScript.to_string(), "javascript");
583        assert_eq!(Language::TypeScript.to_string(), "typescript");
584        assert_eq!(Language::Go.to_string(), "go");
585        assert_eq!(Language::Java.to_string(), "java");
586    }
587}