| 1 | use crate::{ |
| 2 | Error, Result, |
| 3 | auth::{Access, Capability}, |
| 4 | model::*, |
| 5 | policy, |
| 6 | }; |
| 7 | use rusqlite::{Connection, OptionalExtension, Row, TransactionBehavior, named_params, params}; |
| 8 | use std::{ |
| 9 | collections::BTreeMap, |
| 10 | fs, |
| 11 | io::Write, |
| 12 | path::Path, |
| 13 | sync::Arc, |
| 14 | time::{Duration, SystemTime, UNIX_EPOCH}, |
| 15 | }; |
| 16 | use uuid::Uuid; |
| 17 | |
| 18 | /// Nesting-safe transaction scope for `&self` methods. rusqlite's |
| 19 | /// `unchecked_transaction` issues a literal `BEGIN`, which fails inside an |
| 20 | /// already-open transaction; engine methods compose inside each other's |
| 21 | /// scopes (e.g. `prepare_context` → `recall`), so this issues a `SAVEPOINT` |
| 22 | /// instead — a top-level savepoint behaves as a deferred transaction, and a |
| 23 | /// nested one rolls back only its own work when dropped uncommitted. |
| 24 | pub(crate) struct Tx<'c> { |
| 25 | conn: &'c Connection, |
| 26 | name: String, |
| 27 | done: bool, |
| 28 | } |
| 29 | static TX_SEQ: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0); |
| 30 | impl<'c> Tx<'c> { |
| 31 | pub(crate) fn begin(conn: &'c Connection) -> Result<Self> { |
| 32 | let name = format!( |
| 33 | "tx_{}", |
| 34 | TX_SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed) |
| 35 | ); |
| 36 | conn.execute_batch(&format!("SAVEPOINT \"{name}\""))?; |
| 37 | Ok(Self { |
| 38 | conn, |
| 39 | name, |
| 40 | done: false, |
| 41 | }) |
| 42 | } |
| 43 | pub(crate) fn commit(mut self) -> Result<()> { |
| 44 | self.conn |
| 45 | .execute_batch(&format!("RELEASE \"{}\"", self.name))?; |
| 46 | self.done = true; |
| 47 | Ok(()) |
| 48 | } |
| 49 | } |
| 50 | impl Drop for Tx<'_> { |
| 51 | fn drop(&mut self) { |
| 52 | if !self.done { |
| 53 | let _ = self |
| 54 | .conn |
| 55 | .execute_batch(&format!("ROLLBACK TO \"{0}\"; RELEASE \"{0}\"", self.name)); |
| 56 | } |
| 57 | } |
| 58 | } |
| 59 | |
| 60 | const APP_ID: i64 = 1129794866; |
| 61 | const SCHEMA: &str = include_str!("sql/schema.sql"); |
| 62 | const ELIGIBLE: &str = include_str!("sql/eligible.sql"); |
| 63 | const SELECT: &str = "m.id,m.revision,m.status,m.payload,m.created_at,m.updated_at,m.content_hash"; |
| 64 | |
| 65 | /// Vendor-neutral lifecycle seam. Transport adapters must preserve authorization, |
| 66 | /// revision checks, freshness, correction and forgetting, not just store/search. |
| 67 | pub trait MemoryBackend: Send { |
| 68 | fn capture(&mut self, access: &Access, request: &str, draft: Draft) -> Result<CaptureReceipt>; |
| 69 | fn get(&self, access: &Access, id: &str) -> Result<Memory>; |
| 70 | fn recall(&self, access: &Access, query: &Recall) -> Result<RecallReport>; |
| 71 | fn approve( |
| 72 | &mut self, |
| 73 | access: &Access, |
| 74 | id: &str, |
| 75 | revision: i64, |
| 76 | validation: Option<&ValidationReceipt>, |
| 77 | snapshot: &Snapshot, |
| 78 | ) -> Result<Memory>; |
| 79 | fn reject(&mut self, access: &Access, id: &str, revision: i64) -> Result<Memory>; |
| 80 | fn supersede( |
| 81 | &mut self, |
| 82 | access: &Access, |
| 83 | old: (&str, i64), |
| 84 | new: (&str, i64), |
| 85 | validation: Option<&ValidationReceipt>, |
| 86 | snapshot: &Snapshot, |
| 87 | ) -> Result<Memory>; |
| 88 | fn forget(&mut self, access: &Access, id: &str, revision: i64) -> Result<ForgetReport>; |
| 89 | fn save_checkpoint( |
| 90 | &mut self, |
| 91 | access: &Access, |
| 92 | draft: CheckpointDraft, |
| 93 | expected_revision: Option<i64>, |
| 94 | snapshot: &Snapshot, |
| 95 | ) -> Result<Checkpoint>; |
| 96 | fn resume( |
| 97 | &self, |
| 98 | access: &Access, |
| 99 | scope: &Scope, |
| 100 | key: &str, |
| 101 | snapshot: &Snapshot, |
| 102 | ) -> Result<Resume>; |
| 103 | } |
| 104 | |
| 105 | pub struct Store { |
| 106 | pub(crate) conn: Connection, |
| 107 | clock: Arc<dyn Fn() -> i64 + Send + Sync>, |
| 108 | } |
| 109 | fn now() -> i64 { |
| 110 | SystemTime::now() |
| 111 | .duration_since(UNIX_EPOCH) |
| 112 | .unwrap_or_default() |
| 113 | .as_secs() as i64 |
| 114 | } |
| 115 | fn row_memory(row: &Row<'_>) -> rusqlite::Result<Memory> { |
| 116 | let json: String = row.get(3)?; |
| 117 | let status: String = row.get(2)?; |
| 118 | let decode = |e: Box<dyn std::error::Error + Send + Sync>| { |
| 119 | rusqlite::Error::FromSqlConversionFailure(3, rusqlite::types::Type::Text, e) |
| 120 | }; |
| 121 | Ok(Memory { |
| 122 | id: row.get(0)?, |
| 123 | revision: row.get(1)?, |
| 124 | status: Status::parse(&status).map_err(|e| decode(Box::new(e)))?, |
| 125 | draft: serde_json::from_str(&json).map_err(|e| decode(Box::new(e)))?, |
| 126 | created_at: row.get(4)?, |
| 127 | updated_at: row.get(5)?, |
| 128 | content_hash: row.get(6)?, |
| 129 | }) |
| 130 | } |
| 131 | fn event( |
| 132 | conn: &Connection, |
| 133 | access: &Access, |
| 134 | id: &str, |
| 135 | action: &str, |
| 136 | revision: i64, |
| 137 | now: i64, |
| 138 | ) -> Result<()> { |
| 139 | conn.execute( |
| 140 | "INSERT INTO events(entity_id,action,actor_hash,revision,at) VALUES(?1,?2,?3,?4,?5)", |
| 141 | params![id, action, access.actor_hash(), revision, now], |
| 142 | )?; |
| 143 | Ok(()) |
| 144 | } |
| 145 | fn get_at(conn: &Connection, access: &Access, id: &str) -> Result<Memory> { |
| 146 | access.require(Capability::Read)?; |
| 147 | let sql = format!( |
| 148 | "SELECT {SELECT} FROM memories m WHERE m.id=?1 AND m.scope IN (SELECT value FROM json_each(?2))" |
| 149 | ); |
| 150 | conn.query_row(&sql, params![id, access.scope_json()?], row_memory) |
| 151 | .optional()? |
| 152 | .ok_or(Error::NotFound) |
| 153 | } |
| 154 | pub(crate) fn effective_freshness( |
| 155 | conn: &Connection, |
| 156 | access: &Access, |
| 157 | memory: &Memory, |
| 158 | snapshot: &Snapshot, |
| 159 | now: i64, |
| 160 | ) -> Result<Freshness> { |
| 161 | let own = memory.freshness(snapshot, now); |
| 162 | if own != Freshness::Current { |
| 163 | return Ok(own); |
| 164 | } |
| 165 | let mut pending = memory.draft.parent_ids.clone(); |
| 166 | let mut seen = std::collections::BTreeSet::new(); |
| 167 | while let Some(id) = pending.pop() { |
| 168 | if !seen.insert(id.clone()) { |
| 169 | continue; |
| 170 | } |
| 171 | if seen.len() > 4096 { |
| 172 | return Ok(Freshness::Unknown); |
| 173 | } |
| 174 | let parent = match get_at(conn, access, &id) { |
| 175 | Ok(m) => m, |
| 176 | Err(Error::NotFound) => return Ok(Freshness::Unknown), |
| 177 | Err(e) => return Err(e), |
| 178 | }; |
| 179 | let freshness = parent.freshness(snapshot, now); |
| 180 | if freshness != Freshness::Current { |
| 181 | return Ok(if freshness == Freshness::Inactive { |
| 182 | Freshness::Changed |
| 183 | } else { |
| 184 | freshness |
| 185 | }); |
| 186 | } |
| 187 | pending.extend(parent.draft.parent_ids); |
| 188 | } |
| 189 | Ok(Freshness::Current) |
| 190 | } |
| 191 | fn require_revision(m: &Memory, expected: i64) -> Result<()> { |
| 192 | if m.revision == expected { |
| 193 | Ok(()) |
| 194 | } else { |
| 195 | Err(Error::RevisionConflict) |
| 196 | } |
| 197 | } |
| 198 | fn ensure_key_available( |
| 199 | conn: &Connection, |
| 200 | access: &Access, |
| 201 | draft: &Draft, |
| 202 | except: Option<&str>, |
| 203 | now: i64, |
| 204 | ) -> Result<()> { |
| 205 | if let Some(key) = &draft.key { |
| 206 | let old: Option<(String,i64,Option<i64>)> = conn.query_row( |
| 207 | "SELECT id,revision,expires_at FROM memories WHERE scope=?1 AND semantic_key=?2 AND status='active'", |
| 208 | params![draft.scope.key()?,key],|r| Ok((r.get(0)?,r.get(1)?,r.get(2)?))).optional()?; |
| 209 | if let Some((id, rev, expiry)) = old { |
| 210 | if Some(id.as_str()) == except { |
| 211 | return Ok(()); |
| 212 | } |
| 213 | if expiry.is_some_and(|t| t <= now) { |
| 214 | conn.execute("UPDATE memories SET status='stale',revision=revision+1,updated_at=?2 WHERE id=?1",params![id,now])?; |
| 215 | event(conn, access, &id, "expired", rev + 1, now)?; |
| 216 | invalidate_descendants(conn, access, &id, now)?; |
| 217 | } else { |
| 218 | return Err(Error::KeyConflict); |
| 219 | } |
| 220 | } |
| 221 | } |
| 222 | Ok(()) |
| 223 | } |
| 224 | fn descendants(conn: &Connection, id: &str) -> Result<Vec<String>> { |
| 225 | let mut q = conn.prepare("WITH RECURSIVE children(id) AS (SELECT child_id FROM lineage WHERE parent_id=?1 UNION SELECT l.child_id FROM lineage l JOIN children c ON l.parent_id=c.id) SELECT id FROM children ORDER BY id")?; |
| 226 | Ok(q.query_map([id], |r| r.get(0))? |
| 227 | .collect::<rusqlite::Result<Vec<_>>>()?) |
| 228 | } |
| 229 | fn invalidate_descendants(conn: &Connection, access: &Access, id: &str, now: i64) -> Result<()> { |
| 230 | for child in descendants(conn, id)? { |
| 231 | let m = get_at(conn, access, &child)?; |
| 232 | if m.status == Status::Active { |
| 233 | conn.execute( |
| 234 | "UPDATE memories SET status='stale',revision=revision+1,updated_at=?2 WHERE id=?1", |
| 235 | params![child, now], |
| 236 | )?; |
| 237 | event( |
| 238 | conn, |
| 239 | access, |
| 240 | &child, |
| 241 | "parent_invalidated", |
| 242 | m.revision + 1, |
| 243 | now, |
| 244 | )?; |
| 245 | } |
| 246 | } |
| 247 | Ok(()) |
| 248 | } |
| 249 | fn check_activation( |
| 250 | conn: &Connection, |
| 251 | access: &Access, |
| 252 | m: &Memory, |
| 253 | receipt: Option<&ValidationReceipt>, |
| 254 | snapshot: &Snapshot, |
| 255 | now: i64, |
| 256 | ) -> Result<()> { |
| 257 | if m.status != Status::Candidate { |
| 258 | return Err(Error::InvalidState); |
| 259 | } |
| 260 | let mut projected = m.clone(); |
| 261 | projected.status = Status::Active; |
| 262 | if effective_freshness(conn, access, &projected, snapshot, now)? != Freshness::Current { |
| 263 | return Err(Error::InvalidState); |
| 264 | } |
| 265 | for parent in &m.draft.parent_ids { |
| 266 | let p = get_at(conn, access, parent)?; |
| 267 | if p.status != Status::Active |
| 268 | || effective_freshness(conn, access, &p, snapshot, now)? != Freshness::Current |
| 269 | { |
| 270 | return Err(Error::InvalidParent); |
| 271 | } |
| 272 | } |
| 273 | if matches!(m.draft.kind, Kind::Procedure | Kind::Lesson) && receipt.is_none() { |
| 274 | return Err(Error::ValidationRequired); |
| 275 | } |
| 276 | if let Some(receipt) = receipt { |
| 277 | if !receipt.passed || receipt.content_hash != m.content_hash { |
| 278 | return Err(Error::ValidationRequired); |
| 279 | } |
| 280 | policy::bounded(&receipt.validator, "validator", 256, true)?; |
| 281 | policy::bounded(&receipt.evidence_uri, "validation evidence URI", 1024, true)?; |
| 282 | policy::ensure_no_secret(&serde_json::to_string(receipt)?)?; |
| 283 | } |
| 284 | Ok(()) |
| 285 | } |
| 286 | fn activate( |
| 287 | conn: &Connection, |
| 288 | access: &Access, |
| 289 | m: &Memory, |
| 290 | receipt: Option<&ValidationReceipt>, |
| 291 | now: i64, |
| 292 | ) -> Result<()> { |
| 293 | conn.execute( |
| 294 | "UPDATE memories SET status='active',revision=revision+1,updated_at=?2 WHERE id=?1", |
| 295 | params![m.id, now], |
| 296 | )?; |
| 297 | if let Some(r) = receipt { |
| 298 | conn.execute( |
| 299 | "INSERT INTO validations(memory_id,receipt) VALUES(?1,?2)", |
| 300 | params![m.id, serde_json::to_string(r)?], |
| 301 | )?; |
| 302 | } |
| 303 | event(conn, access, &m.id, "approved", m.revision + 1, now)?; |
| 304 | Ok(()) |
| 305 | } |
| 306 | impl Store { |
| 307 | pub fn open(path: impl AsRef<Path>) -> Result<Self> { |
| 308 | let path = path.as_ref(); |
| 309 | if let Some(parent) = path.parent().filter(|p| !p.as_os_str().is_empty()) |
| 310 | && !parent.exists() |
| 311 | { |
| 312 | #[cfg(unix)] |
| 313 | { |
| 314 | use std::os::unix::fs::DirBuilderExt; |
| 315 | fs::DirBuilder::new() |
| 316 | .recursive(true) |
| 317 | .mode(0o700) |
| 318 | .create(parent)?; |
| 319 | } |
| 320 | #[cfg(not(unix))] |
| 321 | fs::create_dir_all(parent)?; |
| 322 | } |
| 323 | // Explicit local file security. No URI filenames and no extension loading. |
| 324 | match fs::symlink_metadata(path) { |
| 325 | Ok(meta) => { |
| 326 | if meta.file_type().is_symlink() || !meta.is_file() { |
| 327 | return Err(Error::Invalid( |
| 328 | "database must be a regular, non-symlink file".into(), |
| 329 | )); |
| 330 | } |
| 331 | #[cfg(unix)] |
| 332 | { |
| 333 | use std::os::unix::fs::PermissionsExt; |
| 334 | if meta.permissions().mode() & 0o077 != 0 { |
| 335 | return Err(Error::Invalid( |
| 336 | "database permissions must exclude group and world access".into(), |
| 337 | )); |
| 338 | } |
| 339 | } |
| 340 | } |
| 341 | Err(e) if e.kind() == std::io::ErrorKind::NotFound => { |
| 342 | let mut opts = fs::OpenOptions::new(); |
| 343 | opts.write(true).create_new(true); |
| 344 | #[cfg(unix)] |
| 345 | { |
| 346 | use std::os::unix::fs::OpenOptionsExt; |
| 347 | opts.mode(0o600); |
| 348 | } |
| 349 | match opts.open(path) { |
| 350 | Ok(_) => (), |
| 351 | Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => { |
| 352 | return Self::open(path); |
| 353 | } |
| 354 | Err(e) => return Err(e.into()), |
| 355 | } |
| 356 | } |
| 357 | Err(e) => return Err(e.into()), |
| 358 | } |
| 359 | let conn = Connection::open_with_flags( |
| 360 | path, |
| 361 | rusqlite::OpenFlags::SQLITE_OPEN_READ_WRITE | rusqlite::OpenFlags::SQLITE_OPEN_NO_MUTEX, |
| 362 | )?; |
| 363 | Self::from_connection(conn, Arc::new(now)) |
| 364 | } |
| 365 | pub fn in_memory() -> Result<Self> { |
| 366 | Self::from_connection(Connection::open_in_memory()?, Arc::new(now)) |
| 367 | } |
| 368 | pub fn in_memory_with_clock(clock: impl Fn() -> i64 + Send + Sync + 'static) -> Result<Self> { |
| 369 | Self::from_connection(Connection::open_in_memory()?, Arc::new(clock)) |
| 370 | } |
| 371 | fn from_connection( |
| 372 | mut conn: Connection, |
| 373 | clock: Arc<dyn Fn() -> i64 + Send + Sync>, |
| 374 | ) -> Result<Self> { |
| 375 | let app: i64 = conn.query_row("PRAGMA application_id", [], |r| r.get(0))?; |
| 376 | let version: i64 = conn.query_row("PRAGMA user_version", [], |r| r.get(0))?; |
| 377 | let count: i64 = conn.query_row( |
| 378 | "SELECT count(*) FROM sqlite_master WHERE name NOT LIKE 'sqlite_%'", |
| 379 | [], |
| 380 | |r| r.get(0), |
| 381 | )?; |
| 382 | if !((app == 0 && version == 0 && count == 0) |
| 383 | || (app == APP_ID && (1..=2).contains(&version))) |
| 384 | { |
| 385 | return Err(Error::DatabaseMismatch); |
| 386 | } |
| 387 | conn.busy_timeout(Duration::from_secs(5))?; |
| 388 | conn.execute_batch("PRAGMA foreign_keys=ON; PRAGMA secure_delete=ON; PRAGMA temp_store=MEMORY; PRAGMA journal_mode=WAL; PRAGMA synchronous=FULL;")?; |
| 389 | if version < 2 { |
| 390 | let tx = conn.transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 391 | let v: i64 = tx.query_row("PRAGMA user_version", [], |r| r.get(0))?; |
| 392 | if v == 0 { |
| 393 | tx.execute_batch(SCHEMA)?; |
| 394 | } else if !(1..=2).contains(&v) { |
| 395 | return Err(Error::DatabaseMismatch); |
| 396 | } |
| 397 | if v < 2 { |
| 398 | tx.execute_batch(include_str!("sql/migrate_002.sql"))?; |
| 399 | } |
| 400 | tx.commit()?; |
| 401 | } |
| 402 | Ok(Self { conn, clock }) |
| 403 | } |
| 404 | pub fn timestamp(&self) -> i64 { |
| 405 | (self.clock)() |
| 406 | } |
| 407 | /// Validate the complete provenance chain, not just this record's own hashes. |
| 408 | pub fn freshness( |
| 409 | &self, |
| 410 | access: &Access, |
| 411 | memory: &Memory, |
| 412 | snapshot: &Snapshot, |
| 413 | ) -> Result<Freshness> { |
| 414 | access.read(&memory.draft.scope)?; |
| 415 | effective_freshness(&self.conn, access, memory, snapshot, self.timestamp()) |
| 416 | } |
| 417 | pub fn list(&self, access: &Access, after: Option<&str>, limit: usize) -> Result<Vec<Memory>> { |
| 418 | access.require(Capability::Read)?; |
| 419 | let sql = format!( |
| 420 | "SELECT {SELECT} FROM memories m WHERE m.scope IN (SELECT value FROM json_each(?1)) AND m.id>?2 ORDER BY m.id LIMIT ?3" |
| 421 | ); |
| 422 | let mut stmt = self.conn.prepare(&sql)?; |
| 423 | Ok(stmt |
| 424 | .query_map( |
| 425 | params![ |
| 426 | access.scope_json()?, |
| 427 | after.unwrap_or(""), |
| 428 | limit.clamp(1, 500) as i64 |
| 429 | ], |
| 430 | row_memory, |
| 431 | )? |
| 432 | .collect::<rusqlite::Result<Vec<_>>>()?) |
| 433 | } |
| 434 | pub fn dependency_paths(&self, access: &Access) -> Result<Vec<String>> { |
| 435 | access.require(Capability::Read)?; |
| 436 | let mut q = self.conn.prepare("SELECT DISTINCT d.path FROM dependencies d JOIN memories m ON m.id=d.memory_id WHERE m.scope IN (SELECT value FROM json_each(?1)) ORDER BY d.path LIMIT 4096")?; |
| 437 | Ok(q.query_map([access.scope_json()?], |r| r.get(0))? |
| 438 | .collect::<rusqlite::Result<Vec<_>>>()?) |
| 439 | } |
| 440 | pub fn status(&self, access: &Access) -> Result<serde_json::Value> { |
| 441 | access.require(Capability::Read)?; |
| 442 | let mut q = self.conn.prepare("SELECT status,count(*) FROM memories WHERE scope IN (SELECT value FROM json_each(?1)) GROUP BY status")?; |
| 443 | let counts = q |
| 444 | .query_map([access.scope_json()?], |r| { |
| 445 | Ok((r.get::<_, String>(0)?, r.get::<_, i64>(1)?)) |
| 446 | })? |
| 447 | .collect::<rusqlite::Result<BTreeMap<_, _>>>()?; |
| 448 | Ok( |
| 449 | serde_json::json!({"schema_version":2,"backend":"sqlite-local-v2","counts":counts,"network_calls":false,"encrypted_at_rest":false}), |
| 450 | ) |
| 451 | } |
| 452 | pub fn link( |
| 453 | &mut self, |
| 454 | access: &Access, |
| 455 | from: &str, |
| 456 | to: &str, |
| 457 | relation: Relation, |
| 458 | ) -> Result<()> { |
| 459 | let now = self.timestamp(); |
| 460 | let tx = self |
| 461 | .conn |
| 462 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 463 | let a = get_at(&tx, access, from)?; |
| 464 | let b = get_at(&tx, access, to)?; |
| 465 | access.write(&a.draft.scope, Capability::Link)?; |
| 466 | access.write(&b.draft.scope, Capability::Link)?; |
| 467 | if a.draft.scope != b.draft.scope || from == to { |
| 468 | return Err(Error::Invalid( |
| 469 | "relations must join distinct memories in one exact scope".into(), |
| 470 | )); |
| 471 | } |
| 472 | if tx.execute( |
| 473 | "INSERT OR IGNORE INTO links(from_id,to_id,relation) VALUES(?1,?2,?3)", |
| 474 | params![from, to, relation.as_str()], |
| 475 | )? > 0 |
| 476 | { |
| 477 | event(&tx, access, from, "linked", a.revision, now)?; |
| 478 | } |
| 479 | tx.commit()?; |
| 480 | Ok(()) |
| 481 | } |
| 482 | pub fn set_embedding( |
| 483 | &mut self, |
| 484 | access: &Access, |
| 485 | id: &str, |
| 486 | expected_hash: &str, |
| 487 | embedding: &Embedding, |
| 488 | ) -> Result<()> { |
| 489 | let normalized = policy::normalize_embedding(embedding)?; |
| 490 | let bytes: Vec<u8> = normalized.iter().flat_map(|v| v.to_le_bytes()).collect(); |
| 491 | let now = self.timestamp(); |
| 492 | let tx = self |
| 493 | .conn |
| 494 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 495 | let m = get_at(&tx, access, id)?; |
| 496 | access.write(&m.draft.scope, Capability::Index)?; |
| 497 | if m.content_hash != expected_hash { |
| 498 | return Err(Error::RevisionConflict); |
| 499 | } |
| 500 | tx.execute("INSERT INTO embeddings(memory_id,model,dimensions,content_hash,vector) VALUES(?1,?2,?3,?4,?5) ON CONFLICT(memory_id) DO UPDATE SET model=excluded.model,dimensions=excluded.dimensions,content_hash=excluded.content_hash,vector=excluded.vector",params![id,embedding.model,normalized.len() as i64,expected_hash,bytes])?; |
| 501 | event(&tx, access, id, "embedded", m.revision, now)?; |
| 502 | tx.commit()?; |
| 503 | Ok(()) |
| 504 | } |
| 505 | pub fn reindex(&mut self, access: &Access) -> Result<()> { |
| 506 | access.require(Capability::Maintenance)?; |
| 507 | let tx = self |
| 508 | .conn |
| 509 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 510 | tx.execute_batch("INSERT INTO memory_fts(memory_fts) VALUES('rebuild'); INSERT INTO memory_grams(memory_grams) VALUES('rebuild'); INSERT INTO memory_fts(memory_fts,rank) VALUES('integrity-check',1); INSERT INTO memory_grams(memory_grams,rank) VALUES('integrity-check',1);")?; |
| 511 | tx.commit()?; |
| 512 | Ok(()) |
| 513 | } |
| 514 | pub fn export_jsonl(&self, access: &Access, mut output: impl Write) -> Result<usize> { |
| 515 | access.require(Capability::Export)?; |
| 516 | let _read_snapshot = Tx::begin(&self.conn)?; |
| 517 | let mut count = 0; |
| 518 | let mut after = None; |
| 519 | loop { |
| 520 | let batch = self.list(access, after.as_deref(), 500)?; |
| 521 | if batch.is_empty() { |
| 522 | break; |
| 523 | } |
| 524 | for memory in &batch { |
| 525 | serde_json::to_writer( |
| 526 | &mut output, |
| 527 | &serde_json::json!({"schema":"codewhale.memory.export.v1","memory":memory}), |
| 528 | )?; |
| 529 | output.write_all(b"\n")?; |
| 530 | count += 1; |
| 531 | } |
| 532 | after = batch.last().map(|m| m.id.clone()); |
| 533 | } |
| 534 | Ok(count) |
| 535 | } |
| 536 | /// Mark active expired records and their dependents stale. Expiry is already |
| 537 | /// enforced in recall; correctness does not depend on scheduling this sweep. |
| 538 | pub fn expire(&mut self, access: &Access) -> Result<usize> { |
| 539 | access.require(Capability::Maintenance)?; |
| 540 | let now = self.timestamp(); |
| 541 | let tx = self |
| 542 | .conn |
| 543 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 544 | let ids: Vec<String> = { |
| 545 | let mut q = tx.prepare("SELECT id FROM memories WHERE scope IN (SELECT value FROM json_each(?1)) AND status='active' AND expires_at<=?2")?; |
| 546 | q.query_map(params![access.scope_json()?, now], |r| r.get(0))? |
| 547 | .collect::<rusqlite::Result<Vec<_>>>()? |
| 548 | }; |
| 549 | for id in &ids { |
| 550 | let m = get_at(&tx, access, id)?; |
| 551 | access.write(&m.draft.scope, Capability::Maintenance)?; |
| 552 | tx.execute( |
| 553 | "UPDATE memories SET status='stale',revision=revision+1,updated_at=?2 WHERE id=?1", |
| 554 | params![id, now], |
| 555 | )?; |
| 556 | invalidate_descendants(&tx, access, id, now)?; |
| 557 | event(&tx, access, id, "expired", m.revision + 1, now)?; |
| 558 | } |
| 559 | tx.commit()?; |
| 560 | Ok(ids.len()) |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | impl MemoryBackend for Store { |
| 565 | fn capture( |
| 566 | &mut self, |
| 567 | access: &Access, |
| 568 | request: &str, |
| 569 | mut draft: Draft, |
| 570 | ) -> Result<CaptureReceipt> { |
| 571 | access.write(&draft.scope, Capability::Propose)?; |
| 572 | policy::bounded(request, "request id", 1024, true)?; |
| 573 | let now = self.timestamp(); |
| 574 | draft.title = draft.title.trim().to_owned(); |
| 575 | draft.body = draft.body.trim().to_owned(); |
| 576 | draft.tags.sort(); |
| 577 | draft.tags.dedup(); |
| 578 | draft.parent_ids.sort(); |
| 579 | draft.parent_ids.dedup(); |
| 580 | policy::validate_draft(&draft, now)?; |
| 581 | let scope = draft.scope.key()?; |
| 582 | let request_hash = policy::sha256(request.as_bytes()); |
| 583 | let request_draft_hash = policy::sha256(&serde_json::to_vec(&draft)?); |
| 584 | let tx = self |
| 585 | .conn |
| 586 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 587 | let existing: Option<(String, Option<String>)> = tx |
| 588 | .query_row( |
| 589 | "SELECT draft_hash,memory_id FROM requests WHERE scope=?1 AND request_hash=?2", |
| 590 | params![scope, request_hash], |
| 591 | |r| Ok((r.get(0)?, r.get(1)?)), |
| 592 | ) |
| 593 | .optional()?; |
| 594 | if let Some((old_hash, id)) = existing { |
| 595 | if old_hash != request_draft_hash { |
| 596 | return Err(Error::IdempotencyConflict); |
| 597 | } |
| 598 | let id = id.ok_or(Error::Forgotten)?; |
| 599 | let memory = get_at(&tx, access, &id)?; |
| 600 | return Ok(CaptureReceipt { |
| 601 | memory, |
| 602 | created: false, |
| 603 | }); |
| 604 | } |
| 605 | let content_hash = policy::content_hash(&draft)?; |
| 606 | let forgotten: bool = tx.query_row( |
| 607 | "SELECT EXISTS(SELECT 1 FROM tombstones WHERE scope=?1 AND content_hash=?2)", |
| 608 | params![scope, content_hash], |
| 609 | |r| r.get(0), |
| 610 | )?; |
| 611 | if forgotten { |
| 612 | return Err(Error::Forgotten); |
| 613 | } |
| 614 | for parent in &draft.parent_ids { |
| 615 | let p = get_at(&tx, access, parent)?; |
| 616 | if p.draft.scope != draft.scope { |
| 617 | return Err(Error::Invalid( |
| 618 | "derived memories cannot cross scope boundaries".into(), |
| 619 | )); |
| 620 | } |
| 621 | if let Some(expiry) = p.draft.expires_at { |
| 622 | draft.expires_at = Some(draft.expires_at.map_or(expiry, |e| e.min(expiry))); |
| 623 | } |
| 624 | // Inherit dependency fingerprints, not merely prose. Otherwise a |
| 625 | // summary could remain "fresh" after its supporting file changed. |
| 626 | for (path, digest) in &p.draft.dependencies { |
| 627 | if draft.dependencies.get(path).is_some_and(|h| h != digest) { |
| 628 | return Err(Error::InvalidParent); |
| 629 | } |
| 630 | draft.dependencies.insert(path.clone(), digest.clone()); |
| 631 | } |
| 632 | if draft.repository_revision.is_none() { |
| 633 | draft.repository_revision = p.draft.repository_revision.clone(); |
| 634 | } else if p.draft.dependencies.is_empty() |
| 635 | && p.draft.repository_revision.is_some() |
| 636 | && p.draft.repository_revision != draft.repository_revision |
| 637 | { |
| 638 | return Err(Error::InvalidParent); |
| 639 | } |
| 640 | } |
| 641 | if draft.expires_at.is_some_and(|t| t <= now) { |
| 642 | return Err(Error::InvalidParent); |
| 643 | } |
| 644 | policy::validate_draft(&draft, now)?; |
| 645 | let id = Uuid::new_v4().to_string(); |
| 646 | tx.execute("INSERT INTO memories(id,scope,kind,semantic_key,status,revision,title,body,tags,payload,content_hash,repo_revision,importance,confidence,created_at,updated_at,expires_at) VALUES(?1,?2,?3,?4,'candidate',1,?5,?6,?7,?8,?9,?10,?11,?12,?13,?13,?14)", |
| 647 | params![id,scope,draft.kind.as_str(),draft.key,draft.title,draft.body,draft.tags.join(" "),serde_json::to_string(&draft)?,content_hash,draft.repository_revision,draft.importance,draft.confidence,now,draft.expires_at])?; |
| 648 | for (path, digest) in &draft.dependencies { |
| 649 | tx.execute( |
| 650 | "INSERT INTO dependencies(memory_id,path,digest) VALUES(?1,?2,?3)", |
| 651 | params![id, path, digest], |
| 652 | )?; |
| 653 | } |
| 654 | for parent in &draft.parent_ids { |
| 655 | tx.execute( |
| 656 | "INSERT INTO lineage(parent_id,child_id) VALUES(?1,?2)", |
| 657 | params![parent, id], |
| 658 | )?; |
| 659 | } |
| 660 | tx.execute( |
| 661 | "INSERT INTO requests(scope,request_hash,draft_hash,memory_id) VALUES(?1,?2,?3,?4)", |
| 662 | params![scope, request_hash, request_draft_hash, id], |
| 663 | )?; |
| 664 | event(&tx, access, &id, "proposed", 1, now)?; |
| 665 | let memory = get_at(&tx, access, &id)?; |
| 666 | tx.commit()?; |
| 667 | Ok(CaptureReceipt { |
| 668 | memory, |
| 669 | created: true, |
| 670 | }) |
| 671 | } |
| 672 | fn get(&self, access: &Access, id: &str) -> Result<Memory> { |
| 673 | get_at(&self.conn, access, id) |
| 674 | } |
| 675 | fn approve( |
| 676 | &mut self, |
| 677 | access: &Access, |
| 678 | id: &str, |
| 679 | revision: i64, |
| 680 | validation: Option<&ValidationReceipt>, |
| 681 | snapshot: &Snapshot, |
| 682 | ) -> Result<Memory> { |
| 683 | let now = self.timestamp(); |
| 684 | let tx = self |
| 685 | .conn |
| 686 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 687 | let m = get_at(&tx, access, id)?; |
| 688 | access.write(&m.draft.scope, Capability::Review)?; |
| 689 | require_revision(&m, revision)?; |
| 690 | check_activation(&tx, access, &m, validation, snapshot, now)?; |
| 691 | ensure_key_available(&tx, access, &m.draft, None, now)?; |
| 692 | activate(&tx, access, &m, validation, now)?; |
| 693 | let updated = get_at(&tx, access, id)?; |
| 694 | tx.commit()?; |
| 695 | Ok(updated) |
| 696 | } |
| 697 | fn reject(&mut self, access: &Access, id: &str, revision: i64) -> Result<Memory> { |
| 698 | let now = self.timestamp(); |
| 699 | let tx = self |
| 700 | .conn |
| 701 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 702 | let m = get_at(&tx, access, id)?; |
| 703 | access.write(&m.draft.scope, Capability::Review)?; |
| 704 | require_revision(&m, revision)?; |
| 705 | if m.status != Status::Candidate { |
| 706 | return Err(Error::InvalidState); |
| 707 | } |
| 708 | tx.execute( |
| 709 | "UPDATE memories SET status='rejected',revision=revision+1,updated_at=?2 WHERE id=?1", |
| 710 | params![id, now], |
| 711 | )?; |
| 712 | event(&tx, access, id, "rejected", m.revision + 1, now)?; |
| 713 | let updated = get_at(&tx, access, id)?; |
| 714 | tx.commit()?; |
| 715 | Ok(updated) |
| 716 | } |
| 717 | fn supersede( |
| 718 | &mut self, |
| 719 | access: &Access, |
| 720 | old: (&str, i64), |
| 721 | new: (&str, i64), |
| 722 | validation: Option<&ValidationReceipt>, |
| 723 | snapshot: &Snapshot, |
| 724 | ) -> Result<Memory> { |
| 725 | let now = self.timestamp(); |
| 726 | let tx = self |
| 727 | .conn |
| 728 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 729 | let a = get_at(&tx, access, old.0)?; |
| 730 | let b = get_at(&tx, access, new.0)?; |
| 731 | access.write(&a.draft.scope, Capability::Correct)?; |
| 732 | access.write(&b.draft.scope, Capability::Review)?; |
| 733 | require_revision(&a, old.1)?; |
| 734 | require_revision(&b, new.1)?; |
| 735 | if old.0 == new.0 |
| 736 | || a.draft.scope != b.draft.scope |
| 737 | || a.draft.key != b.draft.key |
| 738 | || a.draft.kind != b.draft.kind |
| 739 | || !matches!(a.status, Status::Active | Status::Stale) |
| 740 | { |
| 741 | return Err(Error::InvalidState); |
| 742 | } |
| 743 | if descendants(&tx, old.0)?.iter().any(|id| id == new.0) { |
| 744 | return Err(Error::Invalid( |
| 745 | "a correction cannot depend on the memory it invalidates".into(), |
| 746 | )); |
| 747 | } |
| 748 | check_activation(&tx, access, &b, validation, snapshot, now)?; |
| 749 | ensure_key_available(&tx, access, &b.draft, Some(old.0), now)?; |
| 750 | tx.execute( |
| 751 | "UPDATE memories SET status='superseded',revision=revision+1,updated_at=?2 WHERE id=?1", |
| 752 | params![old.0, now], |
| 753 | )?; |
| 754 | invalidate_descendants(&tx, access, old.0, now)?; |
| 755 | tx.execute( |
| 756 | "INSERT INTO replacements(old_id,new_id) VALUES(?1,?2)", |
| 757 | params![old.0, new.0], |
| 758 | )?; |
| 759 | activate(&tx, access, &b, validation, now)?; |
| 760 | event(&tx, access, old.0, "superseded", a.revision + 1, now)?; |
| 761 | let updated = get_at(&tx, access, new.0)?; |
| 762 | tx.commit()?; |
| 763 | Ok(updated) |
| 764 | } |
| 765 | fn forget(&mut self, access: &Access, id: &str, revision: i64) -> Result<ForgetReport> { |
| 766 | let now = self.timestamp(); |
| 767 | let tx = self |
| 768 | .conn |
| 769 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 770 | let root = get_at(&tx, access, id)?; |
| 771 | access.write(&root.draft.scope, Capability::Forget)?; |
| 772 | require_revision(&root, revision)?; |
| 773 | let ids: Vec<String> = { |
| 774 | let mut q = tx.prepare(include_str!("sql/forget_closure.sql"))?; |
| 775 | q.query_map(named_params! {":id":id}, |r| r.get(0))? |
| 776 | .collect::<rusqlite::Result<Vec<_>>>()? |
| 777 | }; |
| 778 | for id in &ids { |
| 779 | let m = get_at(&tx, access, id)?; |
| 780 | access.write(&m.draft.scope, Capability::Forget)?; |
| 781 | tx.execute("INSERT OR IGNORE INTO tombstones(scope,content_hash,forgotten_at) VALUES(?1,?2,?3)",params![m.draft.scope.key()?,m.content_hash,now])?; |
| 782 | event(&tx, access, id, "forgotten", m.revision + 1, now)?; |
| 783 | } |
| 784 | let ids_json = serde_json::to_string(&ids)?; |
| 785 | // A cited memory retracts dependent checkpoints, including their summaries. |
| 786 | // Scope-crossing checkpoint references are same-tenant/user only (Access). |
| 787 | let checkpoints_deleted=tx.execute("DELETE FROM checkpoints WHERE id IN (SELECT checkpoint_id FROM checkpoint_refs WHERE memory_id IN (SELECT value FROM json_each(?1)))",[&ids_json])?; |
| 788 | let memories_deleted = tx.execute( |
| 789 | "DELETE FROM memories WHERE id IN (SELECT value FROM json_each(?1))", |
| 790 | [&ids_json], |
| 791 | )?; |
| 792 | tx.commit()?; |
| 793 | // A busy reader may retain the WAL. Logical deletion still succeeded; |
| 794 | // report that condition instead of claiming physical erasure or failing |
| 795 | // the already-committed operation. |
| 796 | let wal_truncated = self |
| 797 | .conn |
| 798 | .query_row("PRAGMA wal_checkpoint(TRUNCATE)", [], |r| { |
| 799 | r.get::<_, i64>(0) |
| 800 | }) |
| 801 | .map(|busy| busy == 0) |
| 802 | .unwrap_or(false); |
| 803 | Ok(ForgetReport { |
| 804 | memories_deleted, |
| 805 | checkpoints_deleted, |
| 806 | wal_truncated, |
| 807 | physical_erasure_guaranteed: false, |
| 808 | }) |
| 809 | } |
| 810 | fn recall(&self, access: &Access, query: &Recall) -> Result<RecallReport> { |
| 811 | self.recall_impl(access, query) |
| 812 | } |
| 813 | fn save_checkpoint( |
| 814 | &mut self, |
| 815 | access: &Access, |
| 816 | mut draft: CheckpointDraft, |
| 817 | expected_revision: Option<i64>, |
| 818 | snapshot: &Snapshot, |
| 819 | ) -> Result<Checkpoint> { |
| 820 | access.write(&draft.scope, Capability::Checkpoint)?; |
| 821 | if draft.scope.session.is_none() { |
| 822 | return Err(Error::Invalid( |
| 823 | "working state requires a session scope".into(), |
| 824 | )); |
| 825 | } |
| 826 | policy::bounded(&draft.key, "checkpoint key", 256, true)?; |
| 827 | policy::bounded(&draft.state.summary, "checkpoint summary", 8192, false)?; |
| 828 | if draft.memory_ids.len() > 128 |
| 829 | || draft.state.next_steps.len() > 64 |
| 830 | || draft.state.artifact_refs.len() > 64 |
| 831 | || draft.state.pending_operations.len() > 64 |
| 832 | { |
| 833 | return Err(Error::Invalid( |
| 834 | "checkpoint collection limit exceeded".into(), |
| 835 | )); |
| 836 | } |
| 837 | for item in draft |
| 838 | .state |
| 839 | .next_steps |
| 840 | .iter() |
| 841 | .chain(draft.state.artifact_refs.iter()) |
| 842 | { |
| 843 | policy::bounded(item, "checkpoint item", 1024, true)?; |
| 844 | } |
| 845 | for op in &draft.state.pending_operations { |
| 846 | policy::bounded(&op.operation_id, "operation id", 256, true)?; |
| 847 | policy::bounded(&op.state, "operation state", 128, true)?; |
| 848 | } |
| 849 | let encoded = serde_json::to_string(&draft)?; |
| 850 | policy::bounded(&encoded, "checkpoint", 32768, false)?; |
| 851 | policy::ensure_no_secret(&encoded)?; |
| 852 | draft.memory_ids.sort(); |
| 853 | draft.memory_ids.dedup(); |
| 854 | let now = self.timestamp(); |
| 855 | let expires = draft.expires_at.unwrap_or(now.saturating_add(7 * 86400)); |
| 856 | if expires <= now { |
| 857 | return Err(Error::Invalid( |
| 858 | "checkpoint expiry must be in the future".into(), |
| 859 | )); |
| 860 | } |
| 861 | let tx = self |
| 862 | .conn |
| 863 | .transaction_with_behavior(TransactionBehavior::Immediate)?; |
| 864 | let old: Option<(String, i64)> = tx |
| 865 | .query_row( |
| 866 | "SELECT id,revision FROM checkpoints WHERE scope=?1 AND checkpoint_key=?2", |
| 867 | params![draft.scope.key()?, draft.key], |
| 868 | |r| Ok((r.get(0)?, r.get(1)?)), |
| 869 | ) |
| 870 | .optional()?; |
| 871 | let (id, revision) = match old { |
| 872 | None if expected_revision.is_none() => (Uuid::new_v4().to_string(), 1), |
| 873 | Some((id, rev)) if expected_revision == Some(rev) => (id, rev + 1), |
| 874 | _ => return Err(Error::RevisionConflict), |
| 875 | }; |
| 876 | let mut memories = Vec::new(); |
| 877 | for mid in &draft.memory_ids { |
| 878 | let m = get_at(&tx, access, mid)?; |
| 879 | if effective_freshness(&tx, access, &m, snapshot, now)? != Freshness::Current { |
| 880 | return Err(Error::InvalidParent); |
| 881 | } |
| 882 | memories.push(MemoryRef { |
| 883 | id: m.id, |
| 884 | revision: m.revision, |
| 885 | content_hash: m.content_hash, |
| 886 | }); |
| 887 | } |
| 888 | tx.execute("INSERT INTO checkpoints(id,scope,checkpoint_key,revision,payload,updated_at,expires_at) VALUES(?1,?2,?3,?4,?5,?6,?7) ON CONFLICT(scope,checkpoint_key) DO UPDATE SET revision=excluded.revision,payload=excluded.payload,updated_at=excluded.updated_at,expires_at=excluded.expires_at",params![id,draft.scope.key()?,draft.key,revision,serde_json::to_string(&draft)?,now,expires])?; |
| 889 | tx.execute("DELETE FROM checkpoint_refs WHERE checkpoint_id=?1", [&id])?; |
| 890 | for m in &memories { |
| 891 | tx.execute("INSERT INTO checkpoint_refs(checkpoint_id,memory_id,revision,content_hash) VALUES(?1,?2,?3,?4)",params![id,m.id,m.revision,m.content_hash])?; |
| 892 | } |
| 893 | event(&tx, access, &id, "checkpoint_saved", revision, now)?; |
| 894 | tx.commit()?; |
| 895 | Ok(Checkpoint { |
| 896 | id, |
| 897 | revision, |
| 898 | draft, |
| 899 | memories, |
| 900 | updated_at: now, |
| 901 | expires_at: expires, |
| 902 | }) |
| 903 | } |
| 904 | fn resume( |
| 905 | &self, |
| 906 | access: &Access, |
| 907 | scope: &Scope, |
| 908 | key: &str, |
| 909 | snapshot: &Snapshot, |
| 910 | ) -> Result<Resume> { |
| 911 | access.read(scope)?; |
| 912 | let _read_snapshot = Tx::begin(&self.conn)?; |
| 913 | let row: Option<(String,i64,String,i64,i64)>=self.conn.query_row("SELECT id,revision,payload,updated_at,expires_at FROM checkpoints WHERE scope=?1 AND checkpoint_key=?2 AND expires_at>?3",params![scope.key()?,key,self.timestamp()],|r|Ok((r.get(0)?,r.get(1)?,r.get(2)?,r.get(3)?,r.get(4)?))).optional()?; |
| 914 | let (id, revision, payload, updated_at, expires_at) = row.ok_or(Error::NotFound)?; |
| 915 | let mut q=self.conn.prepare("SELECT memory_id,revision,content_hash FROM checkpoint_refs WHERE checkpoint_id=?1 ORDER BY memory_id")?; |
| 916 | let memories = q |
| 917 | .query_map([&id], |r| { |
| 918 | Ok(MemoryRef { |
| 919 | id: r.get(0)?, |
| 920 | revision: r.get(1)?, |
| 921 | content_hash: r.get(2)?, |
| 922 | }) |
| 923 | })? |
| 924 | .collect::<rusqlite::Result<Vec<_>>>()?; |
| 925 | let mut invalidated = Vec::new(); |
| 926 | for reference in &memories { |
| 927 | match self.get(access, &reference.id) { |
| 928 | Ok(m) => { |
| 929 | let current = m.revision == reference.revision |
| 930 | && m.content_hash == reference.content_hash |
| 931 | && effective_freshness(&self.conn, access, &m, snapshot, self.timestamp())? |
| 932 | == Freshness::Current; |
| 933 | if !current { |
| 934 | invalidated.push(reference.id.clone()); |
| 935 | } |
| 936 | } |
| 937 | Err(Error::NotFound) => invalidated.push(reference.id.clone()), |
| 938 | Err(error) => return Err(error), |
| 939 | } |
| 940 | } |
| 941 | Ok(Resume { |
| 942 | checkpoint: Checkpoint { |
| 943 | id, |
| 944 | revision, |
| 945 | draft: serde_json::from_str(&payload)?, |
| 946 | memories, |
| 947 | updated_at, |
| 948 | expires_at, |
| 949 | }, |
| 950 | invalidated_memory_ids: invalidated, |
| 951 | reconcile_pending_operations: true, |
| 952 | }) |
| 953 | } |
| 954 | } |
| 955 | |
| 956 | fn add_ranking( |
| 957 | target: &mut BTreeMap<String, Hit>, |
| 958 | rows: Vec<Memory>, |
| 959 | reason: &str, |
| 960 | weight: f64, |
| 961 | snapshot: &Snapshot, |
| 962 | now: i64, |
| 963 | ) { |
| 964 | for (rank, memory) in rows.into_iter().enumerate() { |
| 965 | let freshness = memory.freshness(snapshot, now); |
| 966 | let score = weight / (60.0 + rank as f64 + 1.0); |
| 967 | let hit = target.entry(memory.id.clone()).or_insert(Hit { |
| 968 | memory, |
| 969 | freshness, |
| 970 | score: 0.0, |
| 971 | reasons: Vec::new(), |
| 972 | }); |
| 973 | hit.score += score; |
| 974 | hit.reasons.push(reason.to_string()); |
| 975 | } |
| 976 | } |
| 977 | impl Store { |
| 978 | fn recall_impl(&self, access: &Access, query: &Recall) -> Result<RecallReport> { |
| 979 | access.require(Capability::Read)?; |
| 980 | policy::bounded(&query.query, "query", 1024, false)?; |
| 981 | let scopes = access.scope_json()?; |
| 982 | let files = serde_json::to_string(&query.snapshot.files)?; |
| 983 | let now = self.timestamp(); |
| 984 | let limit = query.limit.clamp(1, 64); |
| 985 | let candidates = (limit * 8).clamp(64, 512) as i64; |
| 986 | let include_stale = if query.include_stale { 1_i64 } else { 0_i64 }; |
| 987 | // All retrieval branches share a consistent read transaction. A later |
| 988 | // use still needs a revision/freshness check at the host execution boundary. |
| 989 | let _read_snapshot = Tx::begin(&self.conn)?; |
| 990 | let mut hits = BTreeMap::new(); |
| 991 | let run_text = |sql: &str, text: &str| -> Result<Vec<Memory>> { |
| 992 | let mut q = self.conn.prepare(sql)?; |
| 993 | Ok(q.query_map(named_params!{":scopes":scopes,":now":now,":include_stale":include_stale,":files":files,":repo":query.snapshot.revision,":query":text,":limit":candidates},row_memory)?.collect::<rusqlite::Result<Vec<_>>>()?) |
| 994 | }; |
| 995 | if let Some(fts) = policy::fts_query(&query.query)? { |
| 996 | let sql = format!( |
| 997 | "SELECT {SELECT} FROM memory_fts JOIN memories m ON m.rowid=memory_fts.rowid WHERE ({ELIGIBLE}) AND memory_fts MATCH :query ORDER BY bm25(memory_fts,3.0,1.0,2.0),m.id LIMIT :limit" |
| 998 | ); |
| 999 | add_ranking( |
| 1000 | &mut hits, |
| 1001 | run_text(&sql, &fts)?, |
| 1002 | "lexical", |
| 1003 | 1.0, |
| 1004 | &query.snapshot, |
| 1005 | now, |
| 1006 | ); |
| 1007 | let chars = query.query.trim().chars().count(); |
| 1008 | if (3..=256).contains(&chars) { |
| 1009 | let sql = format!( |
| 1010 | "SELECT {SELECT} FROM memory_grams JOIN memories m ON m.rowid=memory_grams.rowid WHERE ({ELIGIBLE}) AND memory_grams MATCH :query ORDER BY bm25(memory_grams,2.0,1.0),m.id LIMIT :limit" |
| 1011 | ); |
| 1012 | let literal = format!("\"{}\"", query.query.trim().replace('"', "\"\"")); |
| 1013 | add_ranking( |
| 1014 | &mut hits, |
| 1015 | run_text(&sql, &literal)?, |
| 1016 | "substring", |
| 1017 | 0.6, |
| 1018 | &query.snapshot, |
| 1019 | now, |
| 1020 | ); |
| 1021 | } else if chars <= 2 { |
| 1022 | let sql = format!( |
| 1023 | "SELECT {SELECT} FROM memories m WHERE ({ELIGIBLE}) AND instr(lower(m.title || ' ' || m.body),lower(:query))>0 ORDER BY m.importance DESC,m.id LIMIT :limit" |
| 1024 | ); |
| 1025 | add_ranking( |
| 1026 | &mut hits, |
| 1027 | run_text(&sql, query.query.trim())?, |
| 1028 | "short_substring", |
| 1029 | 0.5, |
| 1030 | &query.snapshot, |
| 1031 | now, |
| 1032 | ); |
| 1033 | } |
| 1034 | } else if query.query.trim().is_empty() && query.embedding.is_none() { |
| 1035 | let sql = format!( |
| 1036 | "SELECT {SELECT} FROM memories m WHERE ({ELIGIBLE}) AND :query='' ORDER BY m.importance DESC,m.updated_at DESC,m.id LIMIT :limit" |
| 1037 | ); |
| 1038 | add_ranking( |
| 1039 | &mut hits, |
| 1040 | run_text(&sql, "")?, |
| 1041 | "working_set", |
| 1042 | 1.0, |
| 1043 | &query.snapshot, |
| 1044 | now, |
| 1045 | ); |
| 1046 | } |
| 1047 | let mut vector_candidates = 0; |
| 1048 | let mut vector_scan_truncated = false; |
| 1049 | if let Some(embedding) = &query.embedding { |
| 1050 | let normalized = policy::normalize_embedding(embedding)?; |
| 1051 | let scan_limit = query.vector_scan_limit.clamp(1, 20_000); |
| 1052 | let sql = format!( |
| 1053 | "SELECT {SELECT},e.vector FROM embeddings e JOIN memories m ON m.id=e.memory_id WHERE ({ELIGIBLE}) AND e.model=:model AND e.dimensions=:dimensions AND e.content_hash=m.content_hash ORDER BY m.id LIMIT :limit" |
| 1054 | ); |
| 1055 | let mut q = self.conn.prepare(&sql)?; |
| 1056 | let rows=q.query_map(named_params!{":scopes":scopes,":now":now,":include_stale":include_stale,":files":files,":repo":query.snapshot.revision,":model":embedding.model,":dimensions":normalized.len() as i64,":limit":(scan_limit+1) as i64},|r|Ok((row_memory(r)?,r.get::<_,Vec<u8>>(7)?)))?.collect::<rusqlite::Result<Vec<_>>>()?; |
| 1057 | vector_scan_truncated = rows.len() > scan_limit; |
| 1058 | let mut dense = Vec::new(); |
| 1059 | for (m, bytes) in rows.into_iter().take(scan_limit) { |
| 1060 | vector_candidates += 1; |
| 1061 | if bytes.len() != normalized.len() * 4 { |
| 1062 | return Err(Error::Invalid("stored embedding length mismatch".into())); |
| 1063 | } |
| 1064 | let mut cosine = 0.0_f64; |
| 1065 | for (chunk, q) in bytes.as_chunks::<4>().0.iter().zip(&normalized) { |
| 1066 | let v = f32::from_le_bytes([chunk[0], chunk[1], chunk[2], chunk[3]]); |
| 1067 | if !v.is_finite() { |
| 1068 | return Err(Error::Invalid("stored embedding is non-finite".into())); |
| 1069 | } |
| 1070 | cosine += v as f64 * *q as f64; |
| 1071 | } |
| 1072 | if cosine > 0.0 { |
| 1073 | dense.push((m, cosine)); |
| 1074 | } |
| 1075 | } |
| 1076 | dense.sort_by(|a, b| b.1.total_cmp(&a.1).then_with(|| a.0.id.cmp(&b.0.id))); |
| 1077 | add_ranking( |
| 1078 | &mut hits, |
| 1079 | dense |
| 1080 | .into_iter() |
| 1081 | .take(candidates as usize) |
| 1082 | .map(|(m, _)| m) |
| 1083 | .collect(), |
| 1084 | "semantic", |
| 1085 | 1.0, |
| 1086 | &query.snapshot, |
| 1087 | now, |
| 1088 | ); |
| 1089 | } |
| 1090 | if query.expand_graph && !hits.is_empty() { |
| 1091 | let mut ordered: Vec<_> = hits.values().collect(); |
| 1092 | ordered.sort_by(|a, b| { |
| 1093 | b.score |
| 1094 | .total_cmp(&a.score) |
| 1095 | .then_with(|| a.memory.id.cmp(&b.memory.id)) |
| 1096 | }); |
| 1097 | let seeds = serde_json::to_string( |
| 1098 | &ordered |
| 1099 | .iter() |
| 1100 | .take(8) |
| 1101 | .map(|h| &h.memory.id) |
| 1102 | .collect::<Vec<_>>(), |
| 1103 | )?; |
| 1104 | let sql = format!( |
| 1105 | "SELECT DISTINCT {SELECT} FROM memories m JOIN links l ON ((l.from_id IN (SELECT value FROM json_each(:seeds)) AND l.to_id=m.id) OR (l.to_id IN (SELECT value FROM json_each(:seeds)) AND l.from_id=m.id)) WHERE ({ELIGIBLE}) ORDER BY m.id LIMIT 32" |
| 1106 | ); |
| 1107 | let mut q = self.conn.prepare(&sql)?; |
| 1108 | let rows=q.query_map(named_params!{":scopes":scopes,":now":now,":include_stale":include_stale,":files":files,":repo":query.snapshot.revision,":seeds":seeds},row_memory)?.collect::<rusqlite::Result<Vec<_>>>()?; |
| 1109 | add_ranking( |
| 1110 | &mut hits, |
| 1111 | rows, |
| 1112 | "graph_neighbor", |
| 1113 | 0.25, |
| 1114 | &query.snapshot, |
| 1115 | now, |
| 1116 | ); |
| 1117 | } |
| 1118 | let mut hits: Vec<_> = hits.into_values().collect(); |
| 1119 | for hit in &mut hits { |
| 1120 | hit.freshness = |
| 1121 | effective_freshness(&self.conn, access, &hit.memory, &query.snapshot, now)?; |
| 1122 | // Small, explainable tie-breakers. Retrieval never writes reinforcement |
| 1123 | // or turns model confidence into a probability of correctness. |
| 1124 | hit.score += 0.002 * hit.memory.draft.importance + 0.0005 * hit.memory.draft.confidence; |
| 1125 | } |
| 1126 | hits.sort_by(|a, b| { |
| 1127 | b.score |
| 1128 | .total_cmp(&a.score) |
| 1129 | .then_with(|| a.memory.id.cmp(&b.memory.id)) |
| 1130 | }); |
| 1131 | hits.truncate(limit); |
| 1132 | Ok(RecallReport { |
| 1133 | hits, |
| 1134 | vector_candidates, |
| 1135 | vector_scan_truncated, |
| 1136 | }) |
| 1137 | } |
| 1138 | } |
| 1139 |