| 1 | use chrono::{DateTime, Utc}; |
| 2 | use codewhale_models::{ContentBlock, Message, Role}; |
| 3 | use serde::{Deserialize, Serialize}; |
| 4 | use std::collections::{HashMap, HashSet}; |
| 5 | pub const CURRENT_JOURNAL_SCHEMA_VERSION: u32 = 1; |
| 6 | pub type EntryId = String; |
| 7 | #[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, Default)] |
| 8 | pub struct SpawnDepth(pub u32); |
| 9 | impl SpawnDepth { |
| 10 | pub fn next(self) -> Self { |
| 11 | Self(self.0.saturating_add(1)) |
| 12 | } |
| 13 | } |
| 14 | #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)] |
| 15 | #[serde(tag = "kind", rename_all = "snake_case")] |
| 16 | pub enum SessionEntryKind { |
| 17 | Message { |
| 18 | message: Message, |
| 19 | }, |
| 20 | User { |
| 21 | text: String, |
| 22 | }, |
| 23 | Assistant { |
| 24 | text: String, |
| 25 | }, |
| 26 | Compaction { |
| 27 | summary: String, |
| 28 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 29 | tokens_before: Option<u64>, |
| 30 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 31 | tokens_after: Option<u64>, |
| 32 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 33 | model: Option<String>, |
| 34 | }, |
| 35 | BranchSummary { |
| 36 | branch_id: String, |
| 37 | summary: String, |
| 38 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 39 | parent_branch_id: Option<String>, |
| 40 | }, |
| 41 | System { |
| 42 | content: String, |
| 43 | }, |
| 44 | } |
| 45 | impl SessionEntryKind { |
| 46 | pub fn is_contextual(&self) -> bool { |
| 47 | matches!( |
| 48 | self, |
| 49 | Self::Message { .. } | Self::User { .. } | Self::Assistant { .. } |
| 50 | ) |
| 51 | } |
| 52 | /// `self.as_message() == Some(message)` without materializing the |
| 53 | /// projection. Every autosave compares the whole active branch with the |
| 54 | /// live transcript, so the common `Message` entry must not be deep-cloned |
| 55 | /// just to be compared. |
| 56 | pub fn projects_to(&self, message: &Message) -> bool { |
| 57 | match self { |
| 58 | Self::Message { message: own } => own == message, |
| 59 | other => other.as_message().as_ref() == Some(message), |
| 60 | } |
| 61 | } |
| 62 | pub fn as_message(&self) -> Option<Message> { |
| 63 | match self { |
| 64 | Self::Message { message } => Some(message.clone()), |
| 65 | Self::User { text } => Some(Message { |
| 66 | role: Role::User, |
| 67 | content: vec![ContentBlock::Text { |
| 68 | text: text.clone(), |
| 69 | cache_control: None, |
| 70 | }], |
| 71 | }), |
| 72 | Self::Assistant { text } => Some(Message { |
| 73 | role: Role::Assistant, |
| 74 | content: vec![ContentBlock::Text { |
| 75 | text: text.clone(), |
| 76 | cache_control: None, |
| 77 | }], |
| 78 | }), |
| 79 | Self::Compaction { summary, .. } => Some(Message { |
| 80 | role: Role::System, |
| 81 | content: vec![ContentBlock::Text { |
| 82 | text: format!("[compaction summary] {summary}"), |
| 83 | cache_control: None, |
| 84 | }], |
| 85 | }), |
| 86 | Self::BranchSummary { summary, .. } => Some(Message { |
| 87 | role: Role::System, |
| 88 | content: vec![ContentBlock::Text { |
| 89 | text: format!("[branch summary] {summary}"), |
| 90 | cache_control: None, |
| 91 | }], |
| 92 | }), |
| 93 | Self::System { content } => Some(Message { |
| 94 | role: Role::System, |
| 95 | content: vec![ContentBlock::Text { |
| 96 | text: content.clone(), |
| 97 | cache_control: None, |
| 98 | }], |
| 99 | }), |
| 100 | } |
| 101 | } |
| 102 | } |
| 103 | #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)] |
| 104 | pub struct SessionEntry { |
| 105 | pub id: EntryId, |
| 106 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 107 | pub parent_id: Option<EntryId>, |
| 108 | #[serde(flatten)] |
| 109 | pub kind: SessionEntryKind, |
| 110 | pub created_at: DateTime<Utc>, |
| 111 | #[serde(default)] |
| 112 | pub spawn_depth: u32, |
| 113 | } |
| 114 | impl SessionEntry { |
| 115 | pub fn new(kind: SessionEntryKind, parent_id: Option<EntryId>, spawn_depth: u32) -> Self { |
| 116 | Self { |
| 117 | id: uuid::Uuid::new_v4().to_string(), |
| 118 | parent_id, |
| 119 | kind, |
| 120 | created_at: Utc::now(), |
| 121 | spawn_depth, |
| 122 | } |
| 123 | } |
| 124 | pub fn short_id(&self) -> &str { |
| 125 | char_prefix(&self.id, 8) |
| 126 | } |
| 127 | } |
| 128 | #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Default)] |
| 129 | pub struct SessionJournal { |
| 130 | #[serde(default)] |
| 131 | pub entries: Vec<SessionEntry>, |
| 132 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 133 | pub leaf_id: Option<EntryId>, |
| 134 | #[serde(default = "default_journal_schema_version")] |
| 135 | pub schema_version: u32, |
| 136 | #[serde(default)] |
| 137 | pub spawn_depth: u32, |
| 138 | } |
| 139 | fn default_journal_schema_version() -> u32 { |
| 140 | CURRENT_JOURNAL_SCHEMA_VERSION |
| 141 | } |
| 142 | impl SessionJournal { |
| 143 | pub fn new() -> Self { |
| 144 | Self { |
| 145 | entries: Vec::new(), |
| 146 | leaf_id: None, |
| 147 | schema_version: CURRENT_JOURNAL_SCHEMA_VERSION, |
| 148 | spawn_depth: 0, |
| 149 | } |
| 150 | } |
| 151 | pub fn with_spawn_depth(depth: u32) -> Self { |
| 152 | Self { |
| 153 | spawn_depth: depth, |
| 154 | ..Self::new() |
| 155 | } |
| 156 | } |
| 157 | pub fn append(&mut self, kind: SessionEntryKind) -> EntryId { |
| 158 | self.append_stamped(kind, Utc::now()) |
| 159 | } |
| 160 | /// Append an entry carrying the time the underlying event happened, not |
| 161 | /// the time the save ran. `created_at` is the journal's timeline; stamping |
| 162 | /// it at append is what keeps a rebuilt journal honest — a save must never |
| 163 | /// rewrite an entry's time to the moment it was written. |
| 164 | pub fn append_stamped(&mut self, kind: SessionEntryKind, created_at: DateTime<Utc>) -> EntryId { |
| 165 | let mut entry = SessionEntry::new(kind, self.leaf_id.clone(), self.spawn_depth); |
| 166 | entry.created_at = created_at; |
| 167 | let id = entry.id.clone(); |
| 168 | self.entries.push(entry); |
| 169 | self.leaf_id = Some(id.clone()); |
| 170 | id |
| 171 | } |
| 172 | pub fn append_message(&mut self, message: Message) -> EntryId { |
| 173 | self.append(SessionEntryKind::Message { message }) |
| 174 | } |
| 175 | pub fn append_compaction( |
| 176 | &mut self, |
| 177 | summary: String, |
| 178 | tokens_before: Option<u64>, |
| 179 | tokens_after: Option<u64>, |
| 180 | model: Option<String>, |
| 181 | ) -> EntryId { |
| 182 | self.append(SessionEntryKind::Compaction { |
| 183 | summary, |
| 184 | tokens_before, |
| 185 | tokens_after, |
| 186 | model, |
| 187 | }) |
| 188 | } |
| 189 | pub fn append_branch_summary( |
| 190 | &mut self, |
| 191 | branch_id: String, |
| 192 | summary: String, |
| 193 | parent_branch_id: Option<String>, |
| 194 | ) -> EntryId { |
| 195 | self.append(SessionEntryKind::BranchSummary { |
| 196 | branch_id, |
| 197 | summary, |
| 198 | parent_branch_id, |
| 199 | }) |
| 200 | } |
| 201 | pub fn branch_to(&mut self, entry_id: &str) -> Result<(), String> { |
| 202 | if self.entries.iter().any(|e| e.id == entry_id) { |
| 203 | self.leaf_id = Some(entry_id.to_string()); |
| 204 | Ok(()) |
| 205 | } else { |
| 206 | Err(format!("entry {entry_id} not found")) |
| 207 | } |
| 208 | } |
| 209 | pub fn fork_from(&self, from_entry_id: Option<&str>) -> Result<Self, String> { |
| 210 | let leaf = if let Some(id) = from_entry_id { |
| 211 | if !self.entries.iter().any(|e| e.id == id) { |
| 212 | return Err(format!("fork source {id} not found")); |
| 213 | } |
| 214 | Some(id.to_string()) |
| 215 | } else { |
| 216 | self.leaf_id.clone() |
| 217 | }; |
| 218 | Ok(Self { |
| 219 | entries: self.entries.clone(), |
| 220 | leaf_id: leaf, |
| 221 | schema_version: self.schema_version, |
| 222 | spawn_depth: self.spawn_depth.saturating_add(1), |
| 223 | }) |
| 224 | } |
| 225 | pub fn index(&self) -> HashMap<&str, &SessionEntry> { |
| 226 | self.entries.iter().map(|e| (e.id.as_str(), e)).collect() |
| 227 | } |
| 228 | pub fn children_of(&self, parent_id: Option<&str>) -> Vec<&SessionEntry> { |
| 229 | self.entries |
| 230 | .iter() |
| 231 | .filter(|e| e.parent_id.as_deref() == parent_id) |
| 232 | .collect() |
| 233 | } |
| 234 | pub fn contains(&self, entry_id: &str) -> bool { |
| 235 | self.entries.iter().any(|e| e.id == entry_id) |
| 236 | } |
| 237 | pub fn leaf(&self) -> Option<&SessionEntry> { |
| 238 | self.leaf_id |
| 239 | .as_deref() |
| 240 | .and_then(|id| self.entries.iter().find(|e| e.id == id)) |
| 241 | } |
| 242 | pub fn root_to_leaf(&self) -> Vec<&SessionEntry> { |
| 243 | let index: HashMap<&str, &SessionEntry> = |
| 244 | self.entries.iter().map(|e| (e.id.as_str(), e)).collect(); |
| 245 | let mut path = Vec::new(); |
| 246 | let mut cur = self.leaf_id.as_deref(); |
| 247 | let mut seen = HashSet::new(); |
| 248 | while let Some(id) = cur { |
| 249 | if !seen.insert(id) { |
| 250 | break; |
| 251 | } |
| 252 | if let Some(entry) = index.get(id) { |
| 253 | path.push(*entry); |
| 254 | cur = entry.parent_id.as_deref(); |
| 255 | } else { |
| 256 | break; |
| 257 | } |
| 258 | } |
| 259 | path.reverse(); |
| 260 | path |
| 261 | } |
| 262 | pub fn active_messages(&self, include_system: bool) -> Vec<Message> { |
| 263 | self.root_to_leaf() |
| 264 | .into_iter() |
| 265 | .filter_map(|e| { |
| 266 | if !include_system && !e.kind.is_contextual() { |
| 267 | return None; |
| 268 | } |
| 269 | e.kind.as_message() |
| 270 | }) |
| 271 | .collect() |
| 272 | } |
| 273 | pub fn leaves(&self) -> Vec<&SessionEntry> { |
| 274 | let parents: HashSet<&str> = self |
| 275 | .entries |
| 276 | .iter() |
| 277 | .filter_map(|e| e.parent_id.as_deref()) |
| 278 | .collect(); |
| 279 | self.entries |
| 280 | .iter() |
| 281 | .filter(|e| !parents.contains(e.id.as_str())) |
| 282 | .collect() |
| 283 | } |
| 284 | pub fn is_empty(&self) -> bool { |
| 285 | self.entries.is_empty() |
| 286 | } |
| 287 | pub fn len(&self) -> usize { |
| 288 | self.entries.len() |
| 289 | } |
| 290 | pub fn validate(&self) -> Result<(), String> { |
| 291 | let ids: HashSet<&str> = self.entries.iter().map(|e| e.id.as_str()).collect(); |
| 292 | for entry in &self.entries { |
| 293 | if let Some(parent) = entry.parent_id.as_deref() |
| 294 | && !ids.contains(parent) |
| 295 | { |
| 296 | return Err(format!("entry {} missing parent {}", entry.id, parent)); |
| 297 | } |
| 298 | } |
| 299 | if let Some(leaf) = self.leaf_id.as_deref() |
| 300 | && !ids.contains(leaf) |
| 301 | { |
| 302 | return Err(format!("leaf {leaf} not found")); |
| 303 | } |
| 304 | Ok(()) |
| 305 | } |
| 306 | pub fn from_messages(messages: Vec<Message>, spawn_depth: u32) -> Self { |
| 307 | let mut j = Self::with_spawn_depth(spawn_depth); |
| 308 | for msg in messages { |
| 309 | j.append(SessionEntryKind::Message { message: msg }); |
| 310 | } |
| 311 | j |
| 312 | } |
| 313 | /// Build the journal honoring a per-message append stamp. `stamps[i]` is |
| 314 | /// the time `messages[i]` entered the conversation; a missing stamp falls |
| 315 | /// back to now, so a drifted caller degrades to save-time ordering rather |
| 316 | /// than dropping the message. |
| 317 | pub fn from_messages_stamped( |
| 318 | messages: Vec<Message>, |
| 319 | stamps: &[DateTime<Utc>], |
| 320 | spawn_depth: u32, |
| 321 | ) -> Self { |
| 322 | let mut j = Self::with_spawn_depth(spawn_depth); |
| 323 | for (index, msg) in messages.into_iter().enumerate() { |
| 324 | let created_at = stamps.get(index).copied().unwrap_or_else(Utc::now); |
| 325 | j.append_stamped(SessionEntryKind::Message { message: msg }, created_at); |
| 326 | } |
| 327 | j |
| 328 | } |
| 329 | pub fn to_messages(&self) -> Vec<Message> { |
| 330 | self.active_messages(true) |
| 331 | } |
| 332 | |
| 333 | /// Make `messages` the active projection without rewriting the journal. |
| 334 | /// |
| 335 | /// The existing active branch remains as evidence. We reuse its longest |
| 336 | /// unchanged prefix, then append the repaired suffix as a sibling branch. |
| 337 | pub fn rebranch_active_messages(&mut self, messages: &[Message]) { |
| 338 | self.rebranch_active_messages_stamped(messages, &[]); |
| 339 | } |
| 340 | |
| 341 | /// Preserve existing entry identity and timestamps; only append the changed |
| 342 | /// suffix, keeping the previous branch reachable. |
| 343 | pub fn rebranch_active_messages_stamped( |
| 344 | &mut self, |
| 345 | messages: &[Message], |
| 346 | stamps: &[DateTime<Utc>], |
| 347 | ) { |
| 348 | let active_path = self.root_to_leaf(); |
| 349 | let shared_prefix = active_path |
| 350 | .iter() |
| 351 | .zip(messages) |
| 352 | .take_while(|(entry, message)| entry.kind.projects_to(message)) |
| 353 | .count(); |
| 354 | self.leaf_id = shared_prefix |
| 355 | .checked_sub(1) |
| 356 | .map(|index| active_path[index].id.clone()); |
| 357 | for (index, message) in messages.iter().enumerate().skip(shared_prefix) { |
| 358 | self.append_stamped( |
| 359 | SessionEntryKind::Message { |
| 360 | message: message.clone(), |
| 361 | }, |
| 362 | stamps.get(index).copied().unwrap_or_else(Utc::now), |
| 363 | ); |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | /// Ids a bounded save may move out of this journal, in journal order. |
| 368 | /// |
| 369 | /// Empty until more than `dead_threshold` entries sit off the active |
| 370 | /// branch — superseded versions left behind by compaction, undo and |
| 371 | /// edits (#6842). Past it, the plan keeps the root→leaf path, then whole |
| 372 | /// off-branch chains (an entry with every ancestor) newest first, with |
| 373 | /// `Compaction`/`BranchSummary` chains ahead of plain leaves, while the |
| 374 | /// kept off-branch total stays within `dead_budget`. A kept entry never |
| 375 | /// loses its parent, so the remainder always passes [`Self::validate`]. |
| 376 | /// |
| 377 | /// Pure: callers archive the planned entries durably before removing |
| 378 | /// them with [`Self::remove_entries`]. Known limit: the active branch |
| 379 | /// itself is never pruned, so a session that never compacts stays as |
| 380 | /// long as its conversation. |
| 381 | pub fn prune_plan(&self, dead_threshold: usize, dead_budget: usize) -> Vec<EntryId> { |
| 382 | let active = self.root_to_leaf(); |
| 383 | if self.entries.len().saturating_sub(active.len()) <= dead_threshold { |
| 384 | return Vec::new(); |
| 385 | } |
| 386 | let index: HashMap<&str, usize> = self |
| 387 | .entries |
| 388 | .iter() |
| 389 | .enumerate() |
| 390 | .map(|(i, e)| (e.id.as_str(), i)) |
| 391 | .collect(); |
| 392 | let mut keep = vec![false; self.entries.len()]; |
| 393 | for entry in &active { |
| 394 | if let Some(&i) = index.get(entry.id.as_str()) { |
| 395 | keep[i] = true; |
| 396 | } |
| 397 | } |
| 398 | let parents: HashSet<&str> = self |
| 399 | .entries |
| 400 | .iter() |
| 401 | .filter_map(|e| e.parent_id.as_deref()) |
| 402 | .collect(); |
| 403 | let summary = |e: &SessionEntry| { |
| 404 | matches!( |
| 405 | e.kind, |
| 406 | SessionEntryKind::Compaction { .. } | SessionEntryKind::BranchSummary { .. } |
| 407 | ) |
| 408 | }; |
| 409 | // Newest first: journal order is append order. |
| 410 | let candidates = self |
| 411 | .entries |
| 412 | .iter() |
| 413 | .enumerate() |
| 414 | .rev() |
| 415 | .filter(|(_, e)| summary(e)) |
| 416 | .chain( |
| 417 | self.entries |
| 418 | .iter() |
| 419 | .enumerate() |
| 420 | .rev() |
| 421 | .filter(|(_, e)| !parents.contains(e.id.as_str())), |
| 422 | ); |
| 423 | let mut used = 0usize; |
| 424 | for (start, _) in candidates { |
| 425 | if keep[start] { |
| 426 | continue; |
| 427 | } |
| 428 | let mut chain = Vec::new(); |
| 429 | let mut seen = HashSet::new(); |
| 430 | let mut cursor = Some(start); |
| 431 | let mut closed = true; |
| 432 | while let Some(i) = cursor { |
| 433 | if keep[i] { |
| 434 | break; |
| 435 | } |
| 436 | if !seen.insert(i) || used + chain.len() >= dead_budget { |
| 437 | closed = false; |
| 438 | break; |
| 439 | } |
| 440 | chain.push(i); |
| 441 | cursor = match self.entries[i].parent_id.as_deref() { |
| 442 | None => None, |
| 443 | Some(parent) => match index.get(parent) { |
| 444 | Some(&p) => Some(p), |
| 445 | None => { |
| 446 | closed = false; |
| 447 | None |
| 448 | } |
| 449 | }, |
| 450 | }; |
| 451 | } |
| 452 | if closed { |
| 453 | used += chain.len(); |
| 454 | for i in chain { |
| 455 | keep[i] = true; |
| 456 | } |
| 457 | } |
| 458 | } |
| 459 | self.entries |
| 460 | .iter() |
| 461 | .zip(keep) |
| 462 | .filter(|(_, kept)| !kept) |
| 463 | .map(|(e, _)| e.id.clone()) |
| 464 | .collect() |
| 465 | } |
| 466 | |
| 467 | /// Remove `ids` atomically: either every listed entry leaves and the |
| 468 | /// remainder still validates, or the journal is left exactly as it was |
| 469 | /// and the validation error is returned. Returns how many were removed. |
| 470 | pub fn remove_entries(&mut self, ids: &HashSet<EntryId>) -> Result<usize, String> { |
| 471 | if ids.is_empty() { |
| 472 | return Ok(0); |
| 473 | } |
| 474 | let original = std::mem::take(&mut self.entries); |
| 475 | let mut removed = Vec::new(); |
| 476 | for (position, entry) in original.into_iter().enumerate() { |
| 477 | if ids.contains(&entry.id) { |
| 478 | removed.push((position, entry)); |
| 479 | } else { |
| 480 | self.entries.push(entry); |
| 481 | } |
| 482 | } |
| 483 | if let Err(error) = self.validate() { |
| 484 | // Ascending positions put every entry back where it was. |
| 485 | for (position, entry) in removed { |
| 486 | self.entries.insert(position, entry); |
| 487 | } |
| 488 | return Err(error); |
| 489 | } |
| 490 | Ok(removed.len()) |
| 491 | } |
| 492 | } |
| 493 | #[derive(Debug, Clone, Serialize, Deserialize, PartialEq)] |
| 494 | pub struct SessionImportContainer { |
| 495 | pub format_version: u32, |
| 496 | pub source: String, |
| 497 | pub metadata: Option<serde_json::Value>, |
| 498 | pub entries: Vec<SessionEntry>, |
| 499 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 500 | pub leaf_id: Option<EntryId>, |
| 501 | pub exported_at: DateTime<Utc>, |
| 502 | #[serde(default)] |
| 503 | pub spawn_depth: u32, |
| 504 | } |
| 505 | impl SessionImportContainer { |
| 506 | pub fn new( |
| 507 | source: String, |
| 508 | journal: &SessionJournal, |
| 509 | metadata: Option<serde_json::Value>, |
| 510 | ) -> Self { |
| 511 | Self { |
| 512 | format_version: CURRENT_JOURNAL_SCHEMA_VERSION, |
| 513 | source, |
| 514 | metadata, |
| 515 | entries: journal.entries.clone(), |
| 516 | leaf_id: journal.leaf_id.clone(), |
| 517 | exported_at: Utc::now(), |
| 518 | spawn_depth: journal.spawn_depth, |
| 519 | } |
| 520 | } |
| 521 | pub fn into_journal(self) -> Result<SessionJournal, String> { |
| 522 | let j = SessionJournal { |
| 523 | entries: self.entries, |
| 524 | leaf_id: self.leaf_id, |
| 525 | schema_version: self.format_version, |
| 526 | spawn_depth: self.spawn_depth, |
| 527 | }; |
| 528 | j.validate()?; |
| 529 | Ok(j) |
| 530 | } |
| 531 | pub fn to_json(&self) -> Result<String, serde_json::Error> { |
| 532 | serde_json::to_string_pretty(self) |
| 533 | } |
| 534 | pub fn from_json(json: &str) -> Result<Self, serde_json::Error> { |
| 535 | serde_json::from_str(json) |
| 536 | } |
| 537 | } |
| 538 | /// The first `chars` characters of `text`. Ids come from imported journals as |
| 539 | /// well as from this process, so a byte slice could split a character. |
| 540 | fn char_prefix(text: &str, chars: usize) -> &str { |
| 541 | let end = text |
| 542 | .char_indices() |
| 543 | .nth(chars) |
| 544 | .map_or(text.len(), |(index, _)| index); |
| 545 | &text[..end] |
| 546 | } |
| 547 | |
| 548 | /// Render the journal as a tree, one line per reachable entry. |
| 549 | /// |
| 550 | /// Indentation grows only where the history forks: an entry with a single |
| 551 | /// child is followed by that child at the same depth, so a long linear |
| 552 | /// session stays flat instead of drawing entry N at depth N. Each fork on the |
| 553 | /// way down still adds one guide column. Several roots (a rewind to before |
| 554 | /// the first message starts a new one) are drawn as siblings. The walk uses |
| 555 | /// an explicit stack, so no session length can exhaust the thread's stack. |
| 556 | pub fn render_tree(journal: &SessionJournal) -> String { |
| 557 | if journal.entries.is_empty() { |
| 558 | return "(empty session — no entries yet)".to_string(); |
| 559 | } |
| 560 | let mut out = String::new(); |
| 561 | let mut children: HashMap<Option<&str>, Vec<usize>> = HashMap::new(); |
| 562 | for (index, entry) in journal.entries.iter().enumerate() { |
| 563 | children |
| 564 | .entry(entry.parent_id.as_deref()) |
| 565 | .or_default() |
| 566 | .push(index); |
| 567 | } |
| 568 | let active_ids: HashSet<&str> = journal |
| 569 | .root_to_leaf() |
| 570 | .iter() |
| 571 | .map(|e| e.id.as_str()) |
| 572 | .collect(); |
| 573 | let leaf = journal.leaf_id.as_deref(); |
| 574 | // (entry index, guide columns inherited from forks above, connector) |
| 575 | let mut stack: Vec<(usize, String, &'static str)> = Vec::new(); |
| 576 | push_tree_children(&mut stack, children.get(&None), ""); |
| 577 | // Duplicate ids in an imported journal would otherwise revisit a subtree |
| 578 | // forever; each entry is drawn at most once. |
| 579 | let mut drawn = vec![false; journal.entries.len()]; |
| 580 | while let Some((index, guide, connector)) = stack.pop() { |
| 581 | if std::mem::replace(&mut drawn[index], true) { |
| 582 | continue; |
| 583 | } |
| 584 | let entry = &journal.entries[index]; |
| 585 | let marker = if Some(entry.id.as_str()) == leaf { |
| 586 | "*" |
| 587 | } else if active_ids.contains(entry.id.as_str()) { |
| 588 | "●" |
| 589 | } else { |
| 590 | "○" |
| 591 | }; |
| 592 | out.push_str(&format!( |
| 593 | "{guide}{connector}{marker} {} [{}] {}\n", |
| 594 | entry.short_id(), |
| 595 | entry.id, |
| 596 | tree_label(&entry.kind) |
| 597 | )); |
| 598 | let child_guide = match connector { |
| 599 | "├─ " => format!("{guide}│ "), |
| 600 | "└─ " => format!("{guide} "), |
| 601 | _ => guide, |
| 602 | }; |
| 603 | push_tree_children( |
| 604 | &mut stack, |
| 605 | children.get(&Some(entry.id.as_str())), |
| 606 | &child_guide, |
| 607 | ); |
| 608 | } |
| 609 | if let Some(leaf_id) = leaf { |
| 610 | out.push_str(&format!( |
| 611 | "\nleaf: {leaf_id} (active, {} entries)\n", |
| 612 | journal.entries.len() |
| 613 | )); |
| 614 | } |
| 615 | // Entries whose parents form a loop hang off no root and cannot be drawn; |
| 616 | // say so instead of letting the total above silently disagree. |
| 617 | let unreachable = drawn.iter().filter(|drawn| !**drawn).count(); |
| 618 | if unreachable > 0 { |
| 619 | out.push_str(&format!( |
| 620 | "{unreachable} entries not reachable from a root\n" |
| 621 | )); |
| 622 | } |
| 623 | out |
| 624 | } |
| 625 | |
| 626 | /// Queue `nodes` so they pop in journal order. A lone child (or a lone root) |
| 627 | /// continues its line of history without a connector; siblings get one each. |
| 628 | fn push_tree_children( |
| 629 | stack: &mut Vec<(usize, String, &'static str)>, |
| 630 | nodes: Option<&Vec<usize>>, |
| 631 | guide: &str, |
| 632 | ) { |
| 633 | let Some(nodes) = nodes else { |
| 634 | return; |
| 635 | }; |
| 636 | let forked = nodes.len() > 1; |
| 637 | for (position, &index) in nodes.iter().enumerate().rev() { |
| 638 | let connector = if !forked { |
| 639 | "" |
| 640 | } else if position + 1 == nodes.len() { |
| 641 | "└─ " |
| 642 | } else { |
| 643 | "├─ " |
| 644 | }; |
| 645 | stack.push((index, guide.to_string(), connector)); |
| 646 | } |
| 647 | } |
| 648 | |
| 649 | fn tree_label(kind: &SessionEntryKind) -> String { |
| 650 | match kind { |
| 651 | SessionEntryKind::Message { message } => { |
| 652 | let role = &message.role; |
| 653 | let snippet: String = message |
| 654 | .content |
| 655 | .iter() |
| 656 | .filter_map(|b| match b { |
| 657 | ContentBlock::Text { text, .. } => Some(text.as_str()), |
| 658 | _ => None, |
| 659 | }) |
| 660 | .collect::<Vec<_>>() |
| 661 | .join(" "); |
| 662 | format!("{role}: {}", label_snippet(&snippet)) |
| 663 | } |
| 664 | SessionEntryKind::User { text } => format!("user: {}", label_snippet(text)), |
| 665 | SessionEntryKind::Assistant { text } => format!("assistant: {}", label_snippet(text)), |
| 666 | SessionEntryKind::Compaction { summary, .. } => { |
| 667 | format!("compaction: {}", label_snippet(summary)) |
| 668 | } |
| 669 | SessionEntryKind::BranchSummary { |
| 670 | branch_id, summary, .. |
| 671 | } => format!( |
| 672 | "branch:{} {}", |
| 673 | char_prefix(branch_id, 8), |
| 674 | label_snippet(summary) |
| 675 | ), |
| 676 | SessionEntryKind::System { content } => format!("system: {}", label_snippet(content)), |
| 677 | } |
| 678 | } |
| 679 | |
| 680 | /// The first 60 characters of an entry, kept on its one tree line: a line |
| 681 | /// break, tab or other control character reads as a space, and bidirectional |
| 682 | /// format characters, which would reorder the rest of the row, are dropped. |
| 683 | fn label_snippet(text: &str) -> String { |
| 684 | text.chars() |
| 685 | .filter(|ch| { |
| 686 | !matches!( |
| 687 | ch, |
| 688 | '\u{200E}' | '\u{200F}' | '\u{202A}'..='\u{202E}' | '\u{2066}'..='\u{2069}' |
| 689 | ) |
| 690 | }) |
| 691 | .map(|ch| { |
| 692 | if ch.is_control() || matches!(ch, '\u{2028}' | '\u{2029}') { |
| 693 | ' ' |
| 694 | } else { |
| 695 | ch |
| 696 | } |
| 697 | }) |
| 698 | .take(60) |
| 699 | .collect() |
| 700 | } |
| 701 | #[cfg(test)] |
| 702 | mod tests { |
| 703 | use super::*; |
| 704 | use codewhale_models::{ContentBlock, Message, Role}; |
| 705 | fn msg(role: &str, text: &str) -> Message { |
| 706 | Message { |
| 707 | role: Role::from(role), |
| 708 | content: vec![ContentBlock::Text { |
| 709 | text: text.to_string(), |
| 710 | cache_control: None, |
| 711 | }], |
| 712 | } |
| 713 | } |
| 714 | #[test] |
| 715 | fn append_creates_child_of_leaf() { |
| 716 | let mut j = SessionJournal::new(); |
| 717 | let a = j.append(SessionEntryKind::User { |
| 718 | text: "hello".into(), |
| 719 | }); |
| 720 | let b = j.append(SessionEntryKind::Assistant { |
| 721 | text: "world".into(), |
| 722 | }); |
| 723 | assert_eq!(j.leaf_id.as_deref(), Some(b.as_str())); |
| 724 | let be = j.entries.iter().find(|e| e.id == b).unwrap(); |
| 725 | assert_eq!(be.parent_id.as_deref(), Some(a.as_str())); |
| 726 | } |
| 727 | #[test] |
| 728 | fn branch_moves_leaf_only() { |
| 729 | let mut j = SessionJournal::new(); |
| 730 | let a = j.append(SessionEntryKind::User { text: "a".into() }); |
| 731 | let b = j.append(SessionEntryKind::User { text: "b".into() }); |
| 732 | let _c = j.append(SessionEntryKind::User { text: "c".into() }); |
| 733 | j.branch_to(&a).unwrap(); |
| 734 | let d = j.append(SessionEntryKind::User { text: "d".into() }); |
| 735 | assert_eq!(j.entries.len(), 4); |
| 736 | assert_eq!(j.leaf_id.as_deref(), Some(d.as_str())); |
| 737 | let path: Vec<String> = j.root_to_leaf().iter().map(|e| e.id.clone()).collect(); |
| 738 | assert_eq!(path, vec![a.clone(), d.clone()]); |
| 739 | assert!(j.entries.iter().any(|e| e.id == b)); |
| 740 | } |
| 741 | #[test] |
| 742 | fn from_messages_migrates() { |
| 743 | let msgs = vec![msg("user", "hi"), msg("assistant", "hello")]; |
| 744 | let j = SessionJournal::from_messages(msgs, 0); |
| 745 | assert_eq!(j.entries.len(), 2); |
| 746 | assert!(j.validate().is_ok()); |
| 747 | assert_eq!(j.root_to_leaf().len(), 2); |
| 748 | } |
| 749 | #[test] |
| 750 | fn repaired_messages_form_an_append_only_sibling_branch() { |
| 751 | let original = vec![ |
| 752 | msg("user", "shared"), |
| 753 | msg("assistant", "broken"), |
| 754 | msg("user", "old tail"), |
| 755 | ]; |
| 756 | let mut journal = SessionJournal::from_messages(original, 0); |
| 757 | let old_leaf = journal.leaf_id.clone().expect("old leaf"); |
| 758 | let repaired = vec![ |
| 759 | msg("user", "shared"), |
| 760 | msg("assistant", "repaired"), |
| 761 | msg("user", "new tail"), |
| 762 | ]; |
| 763 | |
| 764 | journal.rebranch_active_messages(&repaired); |
| 765 | |
| 766 | assert_eq!(journal.to_messages(), repaired); |
| 767 | assert!(journal.contains(&old_leaf), "old evidence must remain"); |
| 768 | assert_eq!( |
| 769 | journal.entries.len(), |
| 770 | 5, |
| 771 | "one shared entry plus two branches" |
| 772 | ); |
| 773 | } |
| 774 | #[test] |
| 775 | fn stamped_rebranch_keeps_prefix_identity_and_suffix_stamps() { |
| 776 | let stamp = |secs: i64| DateTime::from_timestamp(secs, 0).expect("stamp"); |
| 777 | let mut j = SessionJournal::new(); |
| 778 | j.append_stamped( |
| 779 | SessionEntryKind::Message { |
| 780 | message: msg("user", "a"), |
| 781 | }, |
| 782 | stamp(100), |
| 783 | ); |
| 784 | j.append_stamped( |
| 785 | SessionEntryKind::Message { |
| 786 | message: msg("assistant", "b"), |
| 787 | }, |
| 788 | stamp(200), |
| 789 | ); |
| 790 | let a_id = j.entries[0].id.clone(); |
| 791 | j.rebranch_active_messages_stamped( |
| 792 | &[msg("user", "a"), msg("assistant", "b2")], |
| 793 | &[stamp(100), stamp(300)], |
| 794 | ); |
| 795 | assert_eq!(j.entries.len(), 3); |
| 796 | assert_eq!(j.entries[0].id, a_id, "shared prefix keeps its id"); |
| 797 | assert_eq!(j.entries[0].created_at, stamp(100)); |
| 798 | let path = j.root_to_leaf(); |
| 799 | assert_eq!(path.len(), 2); |
| 800 | assert_eq!( |
| 801 | path[1].created_at, |
| 802 | stamp(300), |
| 803 | "suffix keeps the live stamp" |
| 804 | ); |
| 805 | assert!( |
| 806 | j.entries |
| 807 | .iter() |
| 808 | .any(|e| e.kind.as_message().as_ref() == Some(&msg("assistant", "b"))), |
| 809 | "replaced suffix survives as a sibling" |
| 810 | ); |
| 811 | } |
| 812 | #[test] |
| 813 | fn compaction_fits() { |
| 814 | let mut j = SessionJournal::new(); |
| 815 | let id = j.append_compaction("summary".into(), Some(1000), Some(100), None); |
| 816 | assert!(j.contains(&id)); |
| 817 | assert!(matches!( |
| 818 | j.leaf().unwrap().kind, |
| 819 | SessionEntryKind::Compaction { .. } |
| 820 | )); |
| 821 | } |
| 822 | #[test] |
| 823 | fn branch_summary_fits() { |
| 824 | let mut j = SessionJournal::new(); |
| 825 | let a = j.append(SessionEntryKind::User { |
| 826 | text: "root".into(), |
| 827 | }); |
| 828 | let b = j.append_branch_summary(a.clone(), "branch summary".into(), None); |
| 829 | assert!(j.contains(&b)); |
| 830 | } |
| 831 | #[test] |
| 832 | fn spawn_depth_fork() { |
| 833 | let mut j = SessionJournal::with_spawn_depth(1); |
| 834 | let forked = j.fork_from(None).unwrap(); |
| 835 | assert_eq!(forked.spawn_depth, 2); |
| 836 | let a = j.append(SessionEntryKind::User { |
| 837 | text: "root".into(), |
| 838 | }); |
| 839 | let fork2 = j.fork_from(Some(&a)).unwrap(); |
| 840 | assert_eq!(fork2.spawn_depth, 2); |
| 841 | } |
| 842 | #[test] |
| 843 | fn foreign_roundtrip() { |
| 844 | let mut j = SessionJournal::new(); |
| 845 | j.append(SessionEntryKind::User { |
| 846 | text: "hello".into(), |
| 847 | }); |
| 848 | let c = SessionImportContainer::new("codewhale".into(), &j, None); |
| 849 | let json = c.to_json().unwrap(); |
| 850 | let back = SessionImportContainer::from_json(&json).unwrap(); |
| 851 | let j2 = back.into_journal().unwrap(); |
| 852 | assert_eq!(j.entries.len(), j2.entries.len()); |
| 853 | } |
| 854 | #[test] |
| 855 | fn render_marks_active() { |
| 856 | let mut j = SessionJournal::new(); |
| 857 | j.append(SessionEntryKind::User { |
| 858 | text: "root".into(), |
| 859 | }); |
| 860 | let tree = render_tree(&j); |
| 861 | assert!(tree.contains('*')); |
| 862 | } |
| 863 | #[test] |
| 864 | fn render_keeps_a_linear_session_flat() { |
| 865 | let mut j = SessionJournal::new(); |
| 866 | for i in 0..200 { |
| 867 | j.append(SessionEntryKind::User { |
| 868 | text: format!("turn {i}"), |
| 869 | }); |
| 870 | } |
| 871 | let tree = render_tree(&j); |
| 872 | let entry_lines: Vec<&str> = tree.lines().take(200).collect(); |
| 873 | for (i, line) in entry_lines.iter().enumerate() { |
| 874 | assert!( |
| 875 | line.ends_with(&format!("user: turn {i}")), |
| 876 | "entry {i} out of order: {line:?}" |
| 877 | ); |
| 878 | assert!( |
| 879 | line.starts_with(['●', '*']), |
| 880 | "a linear history must not indent: {line:?}" |
| 881 | ); |
| 882 | } |
| 883 | } |
| 884 | |
| 885 | #[test] |
| 886 | fn render_indents_only_at_forks() { |
| 887 | let mut j = SessionJournal::new(); |
| 888 | let a = j.append(SessionEntryKind::User { text: "a".into() }); |
| 889 | j.append(SessionEntryKind::User { text: "b".into() }); |
| 890 | j.append(SessionEntryKind::User { text: "c".into() }); |
| 891 | j.branch_to(&a).unwrap(); |
| 892 | j.append(SessionEntryKind::User { text: "d".into() }); |
| 893 | j.append(SessionEntryKind::User { text: "e".into() }); |
| 894 | let tree = render_tree(&j); |
| 895 | let shape: Vec<String> = tree |
| 896 | .lines() |
| 897 | .take(5) |
| 898 | .map(|line| { |
| 899 | let (head, label) = line.split_once(" [").unwrap(); |
| 900 | let prefix: String = head.chars().take_while(|c| !c.is_alphanumeric()).collect(); |
| 901 | format!("{prefix}{}", label.rsplit(": ").next().unwrap()) |
| 902 | }) |
| 903 | .collect(); |
| 904 | assert_eq!( |
| 905 | shape, |
| 906 | ["● a", "├─ ○ b", "│ ○ c", "└─ ● d", " * e"], |
| 907 | "{tree}" |
| 908 | ); |
| 909 | } |
| 910 | |
| 911 | #[test] |
| 912 | fn render_draws_several_roots_as_siblings() { |
| 913 | let mut j = SessionJournal::new(); |
| 914 | j.append(SessionEntryKind::User { text: "r1".into() }); |
| 915 | j.append(SessionEntryKind::User { text: "c1".into() }); |
| 916 | // A rewind to before the first message appends a second root. |
| 917 | j.leaf_id = None; |
| 918 | j.append(SessionEntryKind::User { text: "r2".into() }); |
| 919 | j.append(SessionEntryKind::User { text: "d1".into() }); |
| 920 | let tree = render_tree(&j); |
| 921 | let shape: Vec<String> = tree |
| 922 | .lines() |
| 923 | .take(4) |
| 924 | .map(|line| { |
| 925 | let (head, label) = line.split_once(" [").unwrap(); |
| 926 | let prefix: String = head.chars().take_while(|c| !c.is_alphanumeric()).collect(); |
| 927 | format!("{prefix}{}", label.rsplit(": ").next().unwrap()) |
| 928 | }) |
| 929 | .collect(); |
| 930 | assert_eq!( |
| 931 | shape, |
| 932 | ["├─ ○ r1", "│ ○ c1", "└─ ● r2", " * d1"], |
| 933 | "{tree}" |
| 934 | ); |
| 935 | } |
| 936 | |
| 937 | #[test] |
| 938 | fn render_keeps_each_entry_on_one_line_whatever_its_text_or_id() { |
| 939 | let mut j = SessionJournal::new(); |
| 940 | j.append(SessionEntryKind::User { |
| 941 | text: "first line\nsecond line\r\n\tthird\u{202E}".into(), |
| 942 | }); |
| 943 | // Imported ids need not be ASCII; a byte slice would split a char. |
| 944 | j.entries[0].id = "会話会話会話会話".into(); |
| 945 | j.leaf_id = Some(j.entries[0].id.clone()); |
| 946 | j.append(SessionEntryKind::BranchSummary { |
| 947 | branch_id: "枝枝枝枝枝枝枝枝枝".into(), |
| 948 | summary: "multi\nline".into(), |
| 949 | parent_branch_id: None, |
| 950 | }); |
| 951 | let tree = render_tree(&j); |
| 952 | let lines: Vec<&str> = tree.lines().collect(); |
| 953 | assert_eq!( |
| 954 | lines[0], "● 会話会話会話会話 [会話会話会話会話] user: first line second line third", |
| 955 | "{tree}" |
| 956 | ); |
| 957 | assert!( |
| 958 | lines[1].ends_with("branch:枝枝枝枝枝枝枝枝 multi line"), |
| 959 | "{tree}" |
| 960 | ); |
| 961 | assert_eq!(lines.len(), 2 + 2, "two entries plus the footer: {tree}"); |
| 962 | } |
| 963 | |
| 964 | #[test] |
| 965 | fn render_draws_a_repeated_id_once_and_counts_unreachable_entries() { |
| 966 | let mut j = SessionJournal::new(); |
| 967 | let a = j.append(SessionEntryKind::User { text: "a".into() }); |
| 968 | j.append(SessionEntryKind::User { text: "b".into() }); |
| 969 | // An imported journal repeating `a`'s id under `a` itself: following |
| 970 | // children by id would revisit `a`'s subtree forever. |
| 971 | let mut repeat = j.entries[0].clone(); |
| 972 | repeat.parent_id = Some(a.clone()); |
| 973 | j.entries.push(repeat); |
| 974 | // Two entries that are each other's parent hang off no root. |
| 975 | let mut x = SessionEntry::new(SessionEntryKind::User { text: "x".into() }, None, 0); |
| 976 | let mut y = SessionEntry::new(SessionEntryKind::User { text: "y".into() }, None, 0); |
| 977 | x.parent_id = Some(y.id.clone()); |
| 978 | y.parent_id = Some(x.id.clone()); |
| 979 | j.entries.extend([x, y]); |
| 980 | |
| 981 | let (tx, rx) = std::sync::mpsc::channel(); |
| 982 | std::thread::spawn(move || tx.send(render_tree(&j)).ok()); |
| 983 | let tree = rx |
| 984 | .recv_timeout(std::time::Duration::from_secs(2)) |
| 985 | .expect("a repeated id must not make the walk revisit a subtree forever"); |
| 986 | let entry_lines = tree.lines().take_while(|line| !line.is_empty()).count(); |
| 987 | assert_eq!( |
| 988 | entry_lines, 3, |
| 989 | "each journal entry is drawn at most once: {tree}" |
| 990 | ); |
| 991 | assert!( |
| 992 | tree.ends_with("2 entries not reachable from a root\n"), |
| 993 | "{tree}" |
| 994 | ); |
| 995 | } |
| 996 | |
| 997 | #[test] |
| 998 | fn render_handles_a_long_session_on_a_small_stack() { |
| 999 | let mut j = SessionJournal::new(); |
| 1000 | for i in 0..20_000 { |
| 1001 | j.append(SessionEntryKind::User { |
| 1002 | text: format!("turn {i}"), |
| 1003 | }); |
| 1004 | } |
| 1005 | // One recursion frame per entry overflows this long before the end. |
| 1006 | let lines = std::thread::Builder::new() |
| 1007 | .stack_size(256 * 1024) |
| 1008 | .spawn(move || render_tree(&j).lines().count()) |
| 1009 | .unwrap() |
| 1010 | .join() |
| 1011 | .expect("rendering must not exhaust the stack"); |
| 1012 | assert_eq!(lines, 20_000 + 2); |
| 1013 | } |
| 1014 | |
| 1015 | #[test] |
| 1016 | fn active_messages_root_to_leaf() { |
| 1017 | let mut j = SessionJournal::new(); |
| 1018 | j.append(SessionEntryKind::User { text: "a".into() }); |
| 1019 | j.append(SessionEntryKind::User { text: "b".into() }); |
| 1020 | let msgs = j.active_messages(false); |
| 1021 | assert_eq!(msgs.len(), 2); |
| 1022 | j.branch_to(&j.entries[0].id.clone()).unwrap(); |
| 1023 | j.append(SessionEntryKind::User { text: "c".into() }); |
| 1024 | let msgs2 = j.active_messages(false); |
| 1025 | assert_eq!(msgs2.len(), 2); |
| 1026 | } |
| 1027 | |
| 1028 | #[test] |
| 1029 | fn projects_to_matches_as_message_equality_for_every_kind() { |
| 1030 | let kinds = [ |
| 1031 | SessionEntryKind::Message { |
| 1032 | message: msg("assistant", "hi"), |
| 1033 | }, |
| 1034 | SessionEntryKind::User { |
| 1035 | text: "hi".to_string(), |
| 1036 | }, |
| 1037 | SessionEntryKind::Assistant { |
| 1038 | text: "hi".to_string(), |
| 1039 | }, |
| 1040 | SessionEntryKind::System { |
| 1041 | content: "hi".to_string(), |
| 1042 | }, |
| 1043 | ]; |
| 1044 | let probes = [ |
| 1045 | msg("assistant", "hi"), |
| 1046 | msg("user", "hi"), |
| 1047 | msg("system", "hi"), |
| 1048 | msg("assistant", "other"), |
| 1049 | ]; |
| 1050 | for kind in &kinds { |
| 1051 | for probe in &probes { |
| 1052 | assert_eq!( |
| 1053 | kind.projects_to(probe), |
| 1054 | kind.as_message().as_ref() == Some(probe), |
| 1055 | "{kind:?} vs {probe:?}" |
| 1056 | ); |
| 1057 | } |
| 1058 | } |
| 1059 | } |
| 1060 | |
| 1061 | /// A compaction-heavy session: each round rebranches the whole tail, so |
| 1062 | /// superseded copies pile up off the active branch (#6842). |
| 1063 | fn compacted_journal(rounds: usize, tail: usize) -> SessionJournal { |
| 1064 | let mut j = SessionJournal::new(); |
| 1065 | for round in 0..rounds { |
| 1066 | let messages: Vec<Message> = (0..tail) |
| 1067 | .map(|i| msg("user", &format!("round {round} message {i}"))) |
| 1068 | .collect(); |
| 1069 | j.rebranch_active_messages(&messages); |
| 1070 | } |
| 1071 | j |
| 1072 | } |
| 1073 | |
| 1074 | #[test] |
| 1075 | fn prune_plan_is_empty_under_the_threshold() { |
| 1076 | let j = compacted_journal(4, 10); |
| 1077 | assert_eq!(j.len(), 40); |
| 1078 | assert!(j.prune_plan(30, 5).is_empty()); |
| 1079 | } |
| 1080 | |
| 1081 | #[test] |
| 1082 | fn prune_plan_keeps_active_branch_and_whole_recent_chains() { |
| 1083 | let mut j = compacted_journal(50, 10); |
| 1084 | let active_before = j.active_messages(true); |
| 1085 | let leaf = j.leaf_id.clone(); |
| 1086 | let plan = j.prune_plan(100, 25); |
| 1087 | // 490 dead entries; 2 whole 10-entry chains fit a 25 budget. |
| 1088 | assert_eq!(plan.len(), 490 - 20); |
| 1089 | let ids: HashSet<EntryId> = plan.into_iter().collect(); |
| 1090 | assert_eq!(j.remove_entries(&ids).unwrap(), 470); |
| 1091 | assert_eq!(j.len(), 30); |
| 1092 | j.validate().unwrap(); |
| 1093 | assert_eq!(j.leaf_id, leaf); |
| 1094 | assert_eq!(j.active_messages(true), active_before); |
| 1095 | // The newest superseded round survives, not an arbitrary one. |
| 1096 | let kept: Vec<String> = j |
| 1097 | .entries |
| 1098 | .iter() |
| 1099 | .filter_map(|e| e.kind.as_message()) |
| 1100 | .map(|m| match &m.content[0] { |
| 1101 | ContentBlock::Text { text, .. } => text.clone(), |
| 1102 | _ => String::new(), |
| 1103 | }) |
| 1104 | .collect(); |
| 1105 | assert!(kept.iter().any(|t| t.starts_with("round 48 ")), "{kept:?}"); |
| 1106 | assert!(!kept.iter().any(|t| t.starts_with("round 0 ")), "{kept:?}"); |
| 1107 | } |
| 1108 | |
| 1109 | #[test] |
| 1110 | fn prune_plan_prefers_summary_chains() { |
| 1111 | let mut j = compacted_journal(30, 4); |
| 1112 | let leaf = j.leaf_id.clone().unwrap(); |
| 1113 | // An old compaction hanging off the very first entry. |
| 1114 | let first = j.entries[0].id.clone(); |
| 1115 | j.branch_to(&first).unwrap(); |
| 1116 | let summary = j.append_compaction("old summary".into(), None, None, None); |
| 1117 | j.branch_to(&leaf).unwrap(); |
| 1118 | let plan = j.prune_plan(10, 2); |
| 1119 | assert!(!plan.contains(&summary)); |
| 1120 | assert!(!plan.contains(&first)); |
| 1121 | } |
| 1122 | |
| 1123 | #[test] |
| 1124 | fn remove_entries_refuses_to_orphan_and_leaves_journal_untouched() { |
| 1125 | let mut j = compacted_journal(3, 3); |
| 1126 | let before = j.clone(); |
| 1127 | let active: HashSet<EntryId> = j |
| 1128 | .root_to_leaf() |
| 1129 | .iter() |
| 1130 | .map(|e| e.id.clone()) |
| 1131 | .take(1) |
| 1132 | .collect(); |
| 1133 | assert!(j.remove_entries(&active).is_err()); |
| 1134 | assert_eq!(j, before); |
| 1135 | } |
| 1136 | } |
| 1137 |