Skip to main content

lean_ctx/core/
mode_predictor.rs

1use std::collections::HashMap;
2use std::sync::Mutex;
3use std::time::Instant;
4
5const STATS_FILE: &str = "mode_stats.json";
6const PREDICTOR_FLUSH_SECS: u64 = 60;
7
8static PREDICTOR_BUFFER: Mutex<Option<(ModePredictor, Instant)>> = Mutex::new(None);
9
10#[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
11pub struct ModeOutcome {
12    pub mode: String,
13    pub tokens_in: usize,
14    pub tokens_out: usize,
15    pub density: f64,
16}
17
18impl ModeOutcome {
19    pub fn efficiency(&self) -> f64 {
20        if self.tokens_out == 0 {
21            return 0.0;
22        }
23        self.density / (self.tokens_out as f64 / self.tokens_in.max(1) as f64)
24    }
25}
26
27#[derive(Clone, Debug, Hash, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
28pub struct FileSignature {
29    pub ext: String,
30    pub size_bucket: u8,
31}
32
33impl FileSignature {
34    pub fn from_path(path: &str, token_count: usize) -> Self {
35        let ext = std::path::Path::new(path)
36            .extension()
37            .and_then(|e| e.to_str())
38            .unwrap_or("")
39            .to_string();
40        let size_bucket = match token_count {
41            0..=500 => 0,
42            501..=2000 => 1,
43            2001..=5000 => 2,
44            5001..=20000 => 3,
45            _ => 4,
46        };
47        Self { ext, size_bucket }
48    }
49}
50
51#[derive(Debug, Default, Clone, serde::Serialize, serde::Deserialize)]
52pub struct ModePredictor {
53    history: HashMap<FileSignature, Vec<ModeOutcome>>,
54}
55
56impl ModePredictor {
57    pub fn new() -> Self {
58        let mut guard = PREDICTOR_BUFFER.lock().unwrap_or_else(|e| e.into_inner());
59        if let Some((ref predictor, _)) = *guard {
60            return predictor.clone();
61        }
62        let loaded = Self::load_from_disk().unwrap_or_default();
63        *guard = Some((loaded.clone(), Instant::now()));
64        loaded
65    }
66
67    pub fn record(&mut self, sig: FileSignature, outcome: ModeOutcome) {
68        let entries = self.history.entry(sig).or_default();
69        entries.push(outcome);
70        if entries.len() > 100 {
71            entries.drain(0..50);
72        }
73    }
74
75    /// Returns the best mode based on historical efficiency.
76    /// Chain: local history -> Pro adaptive models -> built-in defaults.
77    pub fn predict_best_mode(&self, sig: &FileSignature) -> Option<String> {
78        if let Some(local) = self.predict_from_local(sig) {
79            return Some(local);
80        }
81        if let Some(pro) = self.predict_from_pro(sig) {
82            return Some(pro);
83        }
84        Self::predict_from_defaults(sig)
85    }
86
87    fn predict_from_local(&self, sig: &FileSignature) -> Option<String> {
88        let entries = self.history.get(sig)?;
89        if entries.len() < 3 {
90            return None;
91        }
92
93        let mut mode_scores: HashMap<&str, (f64, usize)> = HashMap::new();
94        for entry in entries {
95            let (sum, count) = mode_scores.entry(&entry.mode).or_insert((0.0, 0));
96            *sum += entry.efficiency();
97            *count += 1;
98        }
99
100        mode_scores
101            .into_iter()
102            .max_by(|a, b| {
103                let avg_a = a.1 .0 / a.1 .1 as f64;
104                let avg_b = b.1 .0 / b.1 .1 as f64;
105                avg_a
106                    .partial_cmp(&avg_b)
107                    .unwrap_or(std::cmp::Ordering::Equal)
108            })
109            .map(|(mode, _)| mode.to_string())
110    }
111
112    /// Loads Pro adaptive models (requires Pro subscription).
113    /// Pro models are cached locally and auto-updated for Pro users.
114    fn predict_from_pro(&self, sig: &FileSignature) -> Option<String> {
115        let data = crate::cloud_client::load_pro_models()?;
116        let models = data["models"].as_array()?;
117
118        let ext_with_dot = format!(".{}", sig.ext);
119        let bucket_name = match sig.size_bucket {
120            0 => "0-500",
121            1 => "500-2k",
122            2 => "2k-10k",
123            3 => "10k+",
124            _ => "10k+",
125        };
126
127        let mut best: Option<(&str, f64)> = None;
128
129        for model in models {
130            let m_ext = model["file_ext"].as_str().unwrap_or("");
131            let m_bucket = model["size_bucket"].as_str().unwrap_or("");
132            let confidence = model["confidence"].as_f64().unwrap_or(0.0);
133
134            if m_ext == ext_with_dot && m_bucket == bucket_name && confidence > 0.5 {
135                if let Some(mode) = model["recommended_mode"].as_str() {
136                    if best.is_none() || confidence > best.unwrap().1 {
137                        best = Some((mode, confidence));
138                    }
139                }
140            }
141        }
142
143        if let Some((mode, _)) = best {
144            return Some(mode.to_string());
145        }
146
147        for model in models {
148            let m_ext = model["file_ext"].as_str().unwrap_or("");
149            let confidence = model["confidence"].as_f64().unwrap_or(0.0);
150            if m_ext == ext_with_dot && confidence > 0.5 {
151                return model["recommended_mode"].as_str().map(|s| s.to_string());
152            }
153        }
154
155        None
156    }
157
158    /// Built-in defaults for common file types and sizes.
159    /// Ensures reasonable compression even without local history or Pro models.
160    /// Respects Kolmogorov-Gate: files with K>0.7 skip aggressive modes.
161    fn predict_from_defaults(sig: &FileSignature) -> Option<String> {
162        let mode = match (sig.ext.as_str(), sig.size_bucket) {
163            // Tiny files (0-500 tokens): always full — compression overhead not worth it
164            (_, 0) => return None,
165
166            // Config / data files: aggressive strips comments and whitespace
167            ("json" | "yaml" | "yml" | "toml" | "xml" | "csv", _) => "aggressive",
168
169            // Lock files: signatures only (just versions matter)
170            ("lock", _) => "signatures",
171
172            // Code files by size bucket
173            // 500-2k tokens: full is fine
174            (
175                "rs" | "ts" | "tsx" | "js" | "jsx" | "py" | "go" | "java" | "c" | "cpp" | "rb"
176                | "swift" | "kt" | "cs" | "vue" | "svelte",
177                1,
178            ) => return None,
179
180            // 2k-5k tokens: map gives structure without bloat
181            (
182                "rs" | "ts" | "tsx" | "js" | "jsx" | "py" | "go" | "java" | "c" | "cpp" | "rb"
183                | "swift" | "kt" | "cs" | "vue" | "svelte",
184                2,
185            ) => "map",
186
187            // 5k-20k tokens: map is strongly preferred
188            (
189                "rs" | "ts" | "tsx" | "js" | "jsx" | "py" | "go" | "java" | "c" | "cpp" | "rb"
190                | "swift" | "kt" | "cs" | "vue" | "svelte",
191                3,
192            ) => "map",
193
194            // 20k+ tokens: signatures only — too large for full context
195            (
196                "rs" | "ts" | "tsx" | "js" | "jsx" | "py" | "go" | "java" | "c" | "cpp" | "rb"
197                | "swift" | "kt" | "cs" | "vue" | "svelte",
198                4..,
199            ) => "signatures",
200
201            // Markup / docs: aggressive for large, map for medium
202            ("md" | "mdx" | "rst" | "txt" | "html" | "astro", 1..=2) => return None,
203            ("md" | "mdx" | "rst" | "txt" | "html" | "astro", 3..) => "aggressive",
204
205            // CSS / styles: aggressive strips whitespace well
206            ("css" | "scss" | "less" | "sass", 2..) => "aggressive",
207
208            // SQL: map for medium+
209            ("sql", 2..) => "map",
210
211            // Unknown large files: aggressive as safe fallback
212            (_, 3..) => "aggressive",
213
214            _ => return None,
215        };
216        Some(mode.to_string())
217    }
218
219    pub fn save(&self) {
220        let mut guard = PREDICTOR_BUFFER.lock().unwrap_or_else(|e| e.into_inner());
221        let should_flush = match *guard {
222            Some((_, ref last_flush)) => last_flush.elapsed().as_secs() >= PREDICTOR_FLUSH_SECS,
223            None => true,
224        };
225        *guard = Some((
226            self.clone(),
227            guard.as_ref().map_or_else(Instant::now, |(_, t)| *t),
228        ));
229        if should_flush {
230            self.save_to_disk();
231            if let Some((_, ref mut t)) = *guard {
232                *t = Instant::now();
233            }
234        }
235    }
236
237    fn save_to_disk(&self) {
238        let dir = match dirs::home_dir() {
239            Some(d) => d.join(".lean-ctx"),
240            None => return,
241        };
242        let _ = std::fs::create_dir_all(&dir);
243        let path = dir.join(STATS_FILE);
244        if let Ok(json) = serde_json::to_string_pretty(self) {
245            let _ = std::fs::write(path, json);
246        }
247    }
248
249    pub fn flush() {
250        let guard = PREDICTOR_BUFFER.lock().unwrap_or_else(|e| e.into_inner());
251        if let Some((ref predictor, _)) = *guard {
252            predictor.save_to_disk();
253        }
254    }
255
256    fn load_from_disk() -> Option<Self> {
257        let path = dirs::home_dir()?.join(".lean-ctx").join(STATS_FILE);
258        let data = std::fs::read_to_string(path).ok()?;
259        serde_json::from_str(&data).ok()
260    }
261}
262
263#[cfg(test)]
264mod tests {
265    use super::*;
266
267    #[test]
268    fn file_signature_buckets() {
269        assert_eq!(FileSignature::from_path("main.rs", 100).size_bucket, 0);
270        assert_eq!(FileSignature::from_path("main.rs", 1000).size_bucket, 1);
271        assert_eq!(FileSignature::from_path("main.rs", 3000).size_bucket, 2);
272        assert_eq!(FileSignature::from_path("main.rs", 10000).size_bucket, 3);
273        assert_eq!(FileSignature::from_path("main.rs", 50000).size_bucket, 4);
274    }
275
276    #[test]
277    fn predict_returns_none_without_history() {
278        let predictor = ModePredictor::default();
279        let sig = FileSignature::from_path("test.zzz", 500);
280        assert!(predictor.predict_from_local(&sig).is_none());
281    }
282
283    #[test]
284    fn predict_returns_none_with_too_few_entries() {
285        let mut predictor = ModePredictor::default();
286        let sig = FileSignature::from_path("test.zzz", 500);
287        predictor.record(
288            sig.clone(),
289            ModeOutcome {
290                mode: "full".to_string(),
291                tokens_in: 100,
292                tokens_out: 100,
293                density: 0.5,
294            },
295        );
296        assert!(predictor.predict_from_local(&sig).is_none());
297    }
298
299    #[test]
300    fn predict_learns_best_mode() {
301        let mut predictor = ModePredictor::default();
302        let sig = FileSignature::from_path("big.rs", 5000);
303        for _ in 0..5 {
304            predictor.record(
305                sig.clone(),
306                ModeOutcome {
307                    mode: "full".to_string(),
308                    tokens_in: 5000,
309                    tokens_out: 5000,
310                    density: 0.3,
311                },
312            );
313            predictor.record(
314                sig.clone(),
315                ModeOutcome {
316                    mode: "map".to_string(),
317                    tokens_in: 5000,
318                    tokens_out: 800,
319                    density: 0.6,
320                },
321            );
322        }
323        let best = predictor.predict_best_mode(&sig);
324        assert_eq!(best, Some("map".to_string()));
325    }
326
327    #[test]
328    fn history_caps_at_100() {
329        let mut predictor = ModePredictor::default();
330        let sig = FileSignature::from_path("test.rs", 100);
331        for _ in 0..120 {
332            predictor.record(
333                sig.clone(),
334                ModeOutcome {
335                    mode: "full".to_string(),
336                    tokens_in: 100,
337                    tokens_out: 100,
338                    density: 0.5,
339                },
340            );
341        }
342        assert!(predictor.history.get(&sig).unwrap().len() <= 100);
343    }
344
345    #[test]
346    fn defaults_return_none_for_small_files() {
347        let sig = FileSignature::from_path("small.rs", 200);
348        assert!(ModePredictor::predict_from_defaults(&sig).is_none());
349    }
350
351    #[test]
352    fn defaults_recommend_map_for_medium_code() {
353        let sig = FileSignature::from_path("medium.rs", 3000);
354        assert_eq!(
355            ModePredictor::predict_from_defaults(&sig),
356            Some("map".to_string())
357        );
358    }
359
360    #[test]
361    fn defaults_recommend_aggressive_for_json() {
362        let sig = FileSignature::from_path("config.json", 1000);
363        assert_eq!(
364            ModePredictor::predict_from_defaults(&sig),
365            Some("aggressive".to_string())
366        );
367    }
368
369    #[test]
370    fn defaults_recommend_signatures_for_huge_code() {
371        let sig = FileSignature::from_path("huge.ts", 25000);
372        assert_eq!(
373            ModePredictor::predict_from_defaults(&sig),
374            Some("signatures".to_string())
375        );
376    }
377
378    #[test]
379    fn defaults_recommend_aggressive_for_large_unknown() {
380        let sig = FileSignature::from_path("data.xyz", 8000);
381        assert_eq!(
382            ModePredictor::predict_from_defaults(&sig),
383            Some("aggressive".to_string())
384        );
385    }
386
387    #[test]
388    fn mode_outcome_efficiency() {
389        let o = ModeOutcome {
390            mode: "map".to_string(),
391            tokens_in: 1000,
392            tokens_out: 200,
393            density: 0.6,
394        };
395        assert!(o.efficiency() > 0.0);
396    }
397}