| 1 | //! @-mention frecency tracking (#441). |
| 2 | //! |
| 3 | //! Records every file the user @-mentions with a timestamp and click count, |
| 4 | //! decays the score over time so a file that was hot last week ranks below |
| 5 | //! one mentioned 5 minutes ago, and re-orders mention-popup completions by |
| 6 | //! the resulting score. Persisted as a single JSONL file at |
| 7 | //! `$CODEWHALE_HOME/file-frecency.jsonl` (normally |
| 8 | //! `~/.codewhale/file-frecency.jsonl`) so frecency survives restarts. |
| 9 | //! |
| 10 | //! Append-only on the wire, compacted in memory: the loader replays every |
| 11 | //! line into a `HashMap<String, FrecencyEntry>` keyed by repo-relative path, |
| 12 | //! folding duplicates into the last record. We cap the in-memory map at |
| 13 | //! 1000 entries and evict the lowest-scored on overflow — same heuristic |
| 14 | //! the OPENCODE source uses. |
| 15 | |
| 16 | use std::collections::HashMap; |
| 17 | use std::fs::OpenOptions; |
| 18 | use std::io::Write; |
| 19 | use std::path::PathBuf; |
| 20 | use std::sync::{Mutex, OnceLock}; |
| 21 | use std::time::{SystemTime, UNIX_EPOCH}; |
| 22 | |
| 23 | use serde::{Deserialize, Serialize}; |
| 24 | |
| 25 | /// Hard cap on the number of paths we track (the acceptance criterion for |
| 26 | /// #441). Older / lower-scored entries are evicted when the map exceeds |
| 27 | /// this. |
| 28 | const FRECENCY_CAP: usize = 1000; |
| 29 | |
| 30 | /// Half-life of a frecency score, in seconds. After this many seconds the |
| 31 | /// score has decayed to ½ of its peak. 7 days is OPENCODE's default — long |
| 32 | /// enough that a commonly-edited file stays sticky across a workweek but |
| 33 | /// short enough that yesterday's deep-dive doesn't haunt you forever. |
| 34 | const HALF_LIFE_SECS: f64 = 7.0 * 24.0 * 60.0 * 60.0; |
| 35 | |
| 36 | #[derive(Debug, Clone, Serialize, Deserialize)] |
| 37 | struct FrecencyRecord { |
| 38 | /// Workspace-relative path string. |
| 39 | path: String, |
| 40 | /// Total mentions over the lifetime of the entry. |
| 41 | count: u32, |
| 42 | /// Unix timestamp (seconds) of the last mention. |
| 43 | last_used: u64, |
| 44 | } |
| 45 | |
| 46 | #[derive(Debug, Default)] |
| 47 | struct Store { |
| 48 | by_path: HashMap<String, FrecencyRecord>, |
| 49 | persisted_path: Option<PathBuf>, |
| 50 | loaded: bool, |
| 51 | } |
| 52 | |
| 53 | fn store() -> &'static Mutex<Store> { |
| 54 | static STORE: OnceLock<Mutex<Store>> = OnceLock::new(); |
| 55 | STORE.get_or_init(|| Mutex::new(Store::default())) |
| 56 | } |
| 57 | |
| 58 | fn default_path() -> Option<PathBuf> { |
| 59 | // Unit tests exercise mention selection heavily and must never read from or |
| 60 | // append to a developer's real Codewhale home. Integration and release QA |
| 61 | // processes use an explicit isolated CODEWHALE_HOME instead. |
| 62 | #[cfg(test)] |
| 63 | { |
| 64 | None |
| 65 | } |
| 66 | |
| 67 | #[cfg(not(test))] |
| 68 | { |
| 69 | codewhale_config::codewhale_home() |
| 70 | .ok() |
| 71 | .map(|home| home.join("file-frecency.jsonl")) |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | fn now_secs() -> u64 { |
| 76 | SystemTime::now() |
| 77 | .duration_since(UNIX_EPOCH) |
| 78 | .map(|d| d.as_secs()) |
| 79 | .unwrap_or(0) |
| 80 | } |
| 81 | |
| 82 | /// Time-decayed frecency score for a record, in arbitrary units. Mentions |
| 83 | /// count linearly; the whole sum is multiplied by an exponential decay |
| 84 | /// factor based on time since `last_used`. Records older than ~5 half-lives |
| 85 | /// score effectively zero. |
| 86 | fn decayed_score(record: &FrecencyRecord, now: u64) -> f64 { |
| 87 | let age_secs = now.saturating_sub(record.last_used) as f64; |
| 88 | let lambda = std::f64::consts::LN_2 / HALF_LIFE_SECS; |
| 89 | (record.count as f64) * (-lambda * age_secs).exp() |
| 90 | } |
| 91 | |
| 92 | fn ensure_loaded(store: &mut Store) { |
| 93 | if store.loaded { |
| 94 | return; |
| 95 | } |
| 96 | store.loaded = true; |
| 97 | let Some(path) = default_path() else { |
| 98 | return; |
| 99 | }; |
| 100 | store.persisted_path = Some(path.clone()); |
| 101 | let Ok(text) = std::fs::read_to_string(&path) else { |
| 102 | return; |
| 103 | }; |
| 104 | for line in text.lines() { |
| 105 | if line.trim().is_empty() { |
| 106 | continue; |
| 107 | } |
| 108 | let Ok(record) = serde_json::from_str::<FrecencyRecord>(line) else { |
| 109 | continue; |
| 110 | }; |
| 111 | store.by_path.insert(record.path.clone(), record); |
| 112 | } |
| 113 | } |
| 114 | |
| 115 | fn evict_to_cap(store: &mut Store, now: u64) { |
| 116 | if store.by_path.len() <= FRECENCY_CAP { |
| 117 | return; |
| 118 | } |
| 119 | let target = FRECENCY_CAP; |
| 120 | let mut scored: Vec<(String, f64)> = store |
| 121 | .by_path |
| 122 | .iter() |
| 123 | .map(|(k, v)| (k.clone(), decayed_score(v, now))) |
| 124 | .collect(); |
| 125 | scored.sort_by(|a, b| a.1.partial_cmp(&b.1).unwrap_or(std::cmp::Ordering::Equal)); |
| 126 | let drop_count = store.by_path.len().saturating_sub(target); |
| 127 | for (key, _) in scored.iter().take(drop_count) { |
| 128 | store.by_path.remove(key); |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | fn append_record_line(path: &PathBuf, record: &FrecencyRecord) -> std::io::Result<()> { |
| 133 | if let Some(parent) = path.parent() { |
| 134 | std::fs::create_dir_all(parent)?; |
| 135 | } |
| 136 | let mut file = OpenOptions::new().create(true).append(true).open(path)?; |
| 137 | let line = serde_json::to_string(record).map_err(std::io::Error::other)?; |
| 138 | writeln!(file, "{line}")?; |
| 139 | Ok(()) |
| 140 | } |
| 141 | |
| 142 | /// Record one mention of `path` (a workspace-relative path string). Updates |
| 143 | /// the in-memory store, persists a single JSONL line, and evicts the lowest- |
| 144 | /// scored entry if we just exceeded the cap. Best-effort: I/O failures are |
| 145 | /// logged and swallowed — losing a frecency datapoint is never worth |
| 146 | /// failing the user's `@` autocomplete. |
| 147 | pub fn record_mention(path: &str) { |
| 148 | if path.is_empty() { |
| 149 | return; |
| 150 | } |
| 151 | let store = store(); |
| 152 | let Ok(mut store) = store.lock() else { |
| 153 | return; |
| 154 | }; |
| 155 | ensure_loaded(&mut store); |
| 156 | let now = now_secs(); |
| 157 | let entry = store |
| 158 | .by_path |
| 159 | .entry(path.to_string()) |
| 160 | .or_insert_with(|| FrecencyRecord { |
| 161 | path: path.to_string(), |
| 162 | count: 0, |
| 163 | last_used: now, |
| 164 | }); |
| 165 | entry.count = entry.count.saturating_add(1); |
| 166 | entry.last_used = now; |
| 167 | let snapshot = entry.clone(); |
| 168 | if let Some(persisted_path) = store.persisted_path.clone() |
| 169 | && let Err(err) = append_record_line(&persisted_path, &snapshot) |
| 170 | { |
| 171 | tracing::debug!(target: "frecency", "persist failed: {err}"); |
| 172 | } |
| 173 | evict_to_cap(&mut store, now); |
| 174 | } |
| 175 | |
| 176 | /// Re-sort a candidate list by frecency score (highest first), preserving |
| 177 | /// the original order for ties so the underlying ranker's choices aren't |
| 178 | /// upended. Candidates the store has never seen score zero — they end up |
| 179 | /// at the bottom of the sort, which means a one-time mention will start |
| 180 | /// floating to the top after first use. |
| 181 | #[must_use] |
| 182 | pub fn rerank_by_frecency(candidates: Vec<String>) -> Vec<String> { |
| 183 | if candidates.len() <= 1 { |
| 184 | return candidates; |
| 185 | } |
| 186 | let store = store(); |
| 187 | let Ok(mut store) = store.lock() else { |
| 188 | return candidates; |
| 189 | }; |
| 190 | ensure_loaded(&mut store); |
| 191 | let now = now_secs(); |
| 192 | let mut scored: Vec<(usize, String, f64)> = candidates |
| 193 | .into_iter() |
| 194 | .enumerate() |
| 195 | .map(|(idx, path)| { |
| 196 | let score = store |
| 197 | .by_path |
| 198 | .get(&path) |
| 199 | .map(|r| decayed_score(r, now)) |
| 200 | .unwrap_or(0.0); |
| 201 | (idx, path, score) |
| 202 | }) |
| 203 | .collect(); |
| 204 | // Stable sort on (-score, original-index): ties keep the underlying |
| 205 | // ranker's order. |
| 206 | scored.sort_by(|a, b| { |
| 207 | b.2.partial_cmp(&a.2) |
| 208 | .unwrap_or(std::cmp::Ordering::Equal) |
| 209 | .then_with(|| a.0.cmp(&b.0)) |
| 210 | }); |
| 211 | scored.into_iter().map(|(_, path, _)| path).collect() |
| 212 | } |
| 213 | |
| 214 | #[cfg(test)] |
| 215 | mod tests { |
| 216 | use super::*; |
| 217 | |
| 218 | /// Recently mentioned paths win against never-mentioned ones; never-mentioned |
| 219 | /// preserve their original ranker order. |
| 220 | #[test] |
| 221 | fn rerank_floats_recent_paths_to_the_top() { |
| 222 | // Use the global store; reset its state so we don't leak across tests. |
| 223 | let store = super::store(); |
| 224 | let mut s = store.lock().unwrap(); |
| 225 | s.by_path.clear(); |
| 226 | s.loaded = true; // skip on-disk replay |
| 227 | s.persisted_path = None; // skip persistence |
| 228 | let now = super::now_secs(); |
| 229 | s.by_path.insert( |
| 230 | "src/popular.rs".into(), |
| 231 | FrecencyRecord { |
| 232 | path: "src/popular.rs".into(), |
| 233 | count: 8, |
| 234 | last_used: now, |
| 235 | }, |
| 236 | ); |
| 237 | drop(s); |
| 238 | |
| 239 | let order = super::rerank_by_frecency(vec![ |
| 240 | "README.md".to_string(), |
| 241 | "src/popular.rs".to_string(), |
| 242 | "Cargo.toml".to_string(), |
| 243 | ]); |
| 244 | assert_eq!(order[0], "src/popular.rs"); |
| 245 | // README.md was first in original order; Cargo.toml second. Both score 0 |
| 246 | // so the original relative order survives. |
| 247 | assert_eq!(order[1], "README.md"); |
| 248 | assert_eq!(order[2], "Cargo.toml"); |
| 249 | } |
| 250 | |
| 251 | /// Decayed score drops below a freshly-used entry after enough half-lives |
| 252 | /// that count alone can't carry the older one. With a 7-day half-life, |
| 253 | /// 8 weeks gives 8 half-lives → ~256× decay; an entry mentioned twice |
| 254 | /// today comfortably beats one mentioned 50× two months ago. |
| 255 | #[test] |
| 256 | fn old_entries_decay_below_recent_ones() { |
| 257 | let now: u64 = 7 * 24 * 60 * 60 * 8; // 8 weeks (8 half-lives) |
| 258 | let stale = FrecencyRecord { |
| 259 | path: "x".into(), |
| 260 | count: 50, |
| 261 | last_used: 0, |
| 262 | }; |
| 263 | let fresh = FrecencyRecord { |
| 264 | path: "y".into(), |
| 265 | count: 2, |
| 266 | last_used: now, |
| 267 | }; |
| 268 | assert!( |
| 269 | super::decayed_score(&fresh, now) > super::decayed_score(&stale, now), |
| 270 | "fresh={}, stale={}", |
| 271 | super::decayed_score(&fresh, now), |
| 272 | super::decayed_score(&stale, now) |
| 273 | ); |
| 274 | } |
| 275 | } |
| 276 |