| 1 | //! Session-scoped artifact metadata. |
| 2 | //! |
| 3 | //! Large tool outputs are written under the owning session directory and saved |
| 4 | //! sessions keep a durable metadata index for resume/listing flows. |
| 5 | |
| 6 | use std::io; |
| 7 | use std::path::Component; |
| 8 | use std::path::Path; |
| 9 | use std::path::PathBuf; |
| 10 | |
| 11 | use chrono::{DateTime, Utc}; |
| 12 | use serde::{Deserialize, Serialize}; |
| 13 | |
| 14 | pub const ARTIFACTS_DIR_NAME: &str = "artifacts"; |
| 15 | |
| 16 | #[cfg(test)] |
| 17 | static TEST_ARTIFACT_SESSIONS_ROOT: std::sync::Mutex<Option<PathBuf>> = std::sync::Mutex::new(None); |
| 18 | |
| 19 | #[cfg(test)] |
| 20 | pub(crate) static TEST_ARTIFACT_SESSIONS_GUARD: std::sync::Mutex<()> = std::sync::Mutex::new(()); |
| 21 | |
| 22 | #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] |
| 23 | #[serde(rename_all = "snake_case")] |
| 24 | pub enum ArtifactKind { |
| 25 | ToolOutput, |
| 26 | } |
| 27 | |
| 28 | #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] |
| 29 | pub struct ArtifactRecord { |
| 30 | pub id: String, |
| 31 | pub kind: ArtifactKind, |
| 32 | #[serde(default)] |
| 33 | pub session_id: String, |
| 34 | pub tool_call_id: String, |
| 35 | pub tool_name: String, |
| 36 | pub created_at: DateTime<Utc>, |
| 37 | pub byte_size: u64, |
| 38 | pub preview: String, |
| 39 | pub storage_path: PathBuf, |
| 40 | } |
| 41 | |
| 42 | fn sanitize_id_component(input: &str) -> String { |
| 43 | input |
| 44 | .chars() |
| 45 | .map(|c| { |
| 46 | if c.is_ascii_alphanumeric() || c == '-' || c == '_' { |
| 47 | c |
| 48 | } else { |
| 49 | '_' |
| 50 | } |
| 51 | }) |
| 52 | .collect() |
| 53 | } |
| 54 | |
| 55 | pub(crate) fn is_valid_session_id(session_id: &str) -> bool { |
| 56 | !session_id.is_empty() |
| 57 | && session_id |
| 58 | .chars() |
| 59 | .all(|c| c.is_ascii_alphanumeric() || c == '-' || c == '_') |
| 60 | } |
| 61 | |
| 62 | #[must_use] |
| 63 | pub fn artifact_id_for_tool_call(tool_call_id: &str) -> String { |
| 64 | format!("art_{}", sanitize_id_component(tool_call_id)) |
| 65 | } |
| 66 | |
| 67 | #[must_use] |
| 68 | pub fn session_artifact_relative_path(artifact_id: &str) -> PathBuf { |
| 69 | PathBuf::from(ARTIFACTS_DIR_NAME).join(format!("{artifact_id}.txt")) |
| 70 | } |
| 71 | |
| 72 | fn session_artifact_relative_path_with_extension( |
| 73 | artifact_id: &str, |
| 74 | extension: &str, |
| 75 | ) -> io::Result<PathBuf> { |
| 76 | let artifact_id = sanitize_id_component(artifact_id); |
| 77 | let extension = extension.trim_start_matches('.').to_ascii_lowercase(); |
| 78 | if artifact_id.is_empty() |
| 79 | || extension.is_empty() |
| 80 | || !extension |
| 81 | .chars() |
| 82 | .all(|character| character.is_ascii_alphanumeric()) |
| 83 | { |
| 84 | return Err(io::Error::new( |
| 85 | io::ErrorKind::InvalidInput, |
| 86 | "artifact id and extension must contain safe ASCII characters", |
| 87 | )); |
| 88 | } |
| 89 | Ok(PathBuf::from(ARTIFACTS_DIR_NAME).join(format!("{artifact_id}.{extension}"))) |
| 90 | } |
| 91 | |
| 92 | fn artifact_sessions_root() -> Option<PathBuf> { |
| 93 | #[cfg(test)] |
| 94 | if let Some(root) = TEST_ARTIFACT_SESSIONS_ROOT |
| 95 | .lock() |
| 96 | .unwrap_or_else(|err| err.into_inner()) |
| 97 | .clone() |
| 98 | { |
| 99 | return Some(root); |
| 100 | } |
| 101 | |
| 102 | // Use the same state-root authority as saved sessions, including an explicit |
| 103 | // CODEWHALE_HOME and legacy read fallback. |
| 104 | codewhale_config::resolve_state_dir("sessions").ok() |
| 105 | } |
| 106 | |
| 107 | #[cfg(test)] |
| 108 | pub(crate) fn set_test_artifact_sessions_root(root: Option<PathBuf>) -> Option<PathBuf> { |
| 109 | let mut guard = TEST_ARTIFACT_SESSIONS_ROOT |
| 110 | .lock() |
| 111 | .unwrap_or_else(|err| err.into_inner()); |
| 112 | std::mem::replace(&mut *guard, root) |
| 113 | } |
| 114 | |
| 115 | #[must_use] |
| 116 | pub fn session_artifact_absolute_path(session_id: &str, relative_path: &Path) -> Option<PathBuf> { |
| 117 | if !is_valid_session_id(session_id) { |
| 118 | return None; |
| 119 | } |
| 120 | if relative_path.is_absolute() |
| 121 | || relative_path |
| 122 | .components() |
| 123 | .any(|component| matches!(component, Component::ParentDir)) |
| 124 | { |
| 125 | return None; |
| 126 | } |
| 127 | Some( |
| 128 | artifact_sessions_root()? |
| 129 | .join(session_id) |
| 130 | .join(relative_path), |
| 131 | ) |
| 132 | } |
| 133 | |
| 134 | pub fn write_session_artifact( |
| 135 | session_id: &str, |
| 136 | artifact_id: &str, |
| 137 | content: &str, |
| 138 | ) -> io::Result<(PathBuf, PathBuf)> { |
| 139 | let relative_path = session_artifact_relative_path(artifact_id); |
| 140 | let absolute_path = |
| 141 | session_artifact_absolute_path(session_id, &relative_path).ok_or_else(|| { |
| 142 | io::Error::new( |
| 143 | io::ErrorKind::InvalidInput, |
| 144 | "could not resolve session artifact path (missing home directory)", |
| 145 | ) |
| 146 | })?; |
| 147 | if let Some(parent) = absolute_path.parent() { |
| 148 | std::fs::create_dir_all(parent)?; |
| 149 | } |
| 150 | crate::utils::write_atomic(&absolute_path, content.as_bytes())?; |
| 151 | Ok((absolute_path, relative_path)) |
| 152 | } |
| 153 | |
| 154 | /// Publish immutable session-owned bytes without replacing an earlier handle. |
| 155 | /// A duplicate replay with identical bytes is idempotent; a different payload |
| 156 | /// for the same relative path fails closed. |
| 157 | pub fn write_session_relative_immutable( |
| 158 | session_id: &str, |
| 159 | relative_path: &Path, |
| 160 | content: &[u8], |
| 161 | ) -> io::Result<PathBuf> { |
| 162 | let absolute_path = |
| 163 | session_artifact_absolute_path(session_id, relative_path).ok_or_else(|| { |
| 164 | io::Error::new(io::ErrorKind::InvalidInput, "invalid session artifact path") |
| 165 | })?; |
| 166 | let destination = open_session_relative(session_id, relative_path, true)?; |
| 167 | match destination.publish(content) { |
| 168 | Ok(()) => {} |
| 169 | Err(err) if err.kind() == io::ErrorKind::AlreadyExists => { |
| 170 | use std::io::Read; |
| 171 | let mut existing = Vec::new(); |
| 172 | destination |
| 173 | .open_file()? |
| 174 | .take(content.len() as u64 + 1) |
| 175 | .read_to_end(&mut existing)?; |
| 176 | if existing != content { |
| 177 | return Err(io::Error::new( |
| 178 | io::ErrorKind::AlreadyExists, |
| 179 | "immutable artifact handle already contains different bytes", |
| 180 | )); |
| 181 | } |
| 182 | } |
| 183 | Err(err) => return Err(err), |
| 184 | } |
| 185 | Ok(absolute_path) |
| 186 | } |
| 187 | |
| 188 | /// The same confined session directory for mutable sidecars and immutable |
| 189 | /// artifacts. The saved-session owner remains responsible for session state. |
| 190 | pub(crate) fn open_session_relative( |
| 191 | session_id: &str, |
| 192 | relative_path: &Path, |
| 193 | create: bool, |
| 194 | ) -> io::Result<crate::fleet::files::WorkspaceFile> { |
| 195 | session_artifact_absolute_path(session_id, relative_path).ok_or_else(|| { |
| 196 | io::Error::new(io::ErrorKind::InvalidInput, "invalid session artifact path") |
| 197 | })?; |
| 198 | let root = artifact_sessions_root() |
| 199 | .ok_or_else(|| io::Error::other("session artifact root unavailable"))?; |
| 200 | if create { |
| 201 | std::fs::create_dir_all(&root)?; |
| 202 | } |
| 203 | crate::fleet::files::WorkspaceFile::open( |
| 204 | &root, |
| 205 | &PathBuf::from(session_id).join(relative_path), |
| 206 | create, |
| 207 | ) |
| 208 | } |
| 209 | |
| 210 | pub fn write_session_artifact_immutable( |
| 211 | session_id: &str, |
| 212 | artifact_id: &str, |
| 213 | content: &[u8], |
| 214 | ) -> io::Result<(PathBuf, PathBuf)> { |
| 215 | let relative_path = session_artifact_relative_path(artifact_id); |
| 216 | let absolute_path = write_session_relative_immutable(session_id, &relative_path, content)?; |
| 217 | Ok((absolute_path, relative_path)) |
| 218 | } |
| 219 | |
| 220 | /// Write arbitrary fetched bytes into a session artifact with a validated |
| 221 | /// extension. Media fetches use this after magic-byte validation. |
| 222 | pub fn write_session_artifact_bytes( |
| 223 | session_id: &str, |
| 224 | artifact_id: &str, |
| 225 | extension: &str, |
| 226 | content: &[u8], |
| 227 | ) -> io::Result<(PathBuf, PathBuf)> { |
| 228 | let relative_path = session_artifact_relative_path_with_extension(artifact_id, extension)?; |
| 229 | let absolute_path = |
| 230 | session_artifact_absolute_path(session_id, &relative_path).ok_or_else(|| { |
| 231 | io::Error::new( |
| 232 | io::ErrorKind::InvalidInput, |
| 233 | "could not resolve session artifact path (missing home directory)", |
| 234 | ) |
| 235 | })?; |
| 236 | if let Some(parent) = absolute_path.parent() { |
| 237 | std::fs::create_dir_all(parent)?; |
| 238 | } |
| 239 | crate::utils::write_atomic(&absolute_path, content)?; |
| 240 | Ok((absolute_path, relative_path)) |
| 241 | } |
| 242 | |
| 243 | fn preview_text(content: &str, max_chars: usize) -> String { |
| 244 | let mut preview: String = content.chars().take(max_chars).collect(); |
| 245 | if content.chars().count() > max_chars { |
| 246 | preview.push_str("..."); |
| 247 | } |
| 248 | preview |
| 249 | } |
| 250 | |
| 251 | pub fn record_tool_output_artifact( |
| 252 | session_id: &str, |
| 253 | tool_call_id: &str, |
| 254 | tool_name: &str, |
| 255 | storage_path: impl Into<PathBuf>, |
| 256 | content: &str, |
| 257 | ) -> ArtifactRecord { |
| 258 | let storage_path = storage_path.into(); |
| 259 | let byte_size = std::fs::metadata(&storage_path) |
| 260 | .map(|metadata| metadata.len()) |
| 261 | .unwrap_or_else(|_| content.len() as u64); |
| 262 | record_tool_output_artifact_with_size( |
| 263 | session_id, |
| 264 | tool_call_id, |
| 265 | tool_name, |
| 266 | storage_path, |
| 267 | byte_size, |
| 268 | &preview_text(content, 200), |
| 269 | ) |
| 270 | } |
| 271 | |
| 272 | pub fn record_tool_output_artifact_with_size( |
| 273 | session_id: &str, |
| 274 | tool_call_id: &str, |
| 275 | tool_name: &str, |
| 276 | storage_path: impl Into<PathBuf>, |
| 277 | byte_size: u64, |
| 278 | preview: &str, |
| 279 | ) -> ArtifactRecord { |
| 280 | ArtifactRecord { |
| 281 | id: artifact_id_for_tool_call(tool_call_id), |
| 282 | kind: ArtifactKind::ToolOutput, |
| 283 | session_id: session_id.to_string(), |
| 284 | tool_call_id: tool_call_id.to_string(), |
| 285 | tool_name: tool_name.to_string(), |
| 286 | created_at: Utc::now(), |
| 287 | byte_size, |
| 288 | preview: preview_text(preview, 200), |
| 289 | storage_path: storage_path.into(), |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | #[must_use] |
| 294 | pub fn format_artifact_relative_path(path: &Path) -> String { |
| 295 | path.display().to_string().replace('\\', "/") |
| 296 | } |
| 297 | |
| 298 | #[must_use] |
| 299 | pub fn format_byte_size(bytes: u64) -> String { |
| 300 | const KIB: u64 = 1024; |
| 301 | const MIB: u64 = KIB * 1024; |
| 302 | if bytes >= MIB { |
| 303 | format!("{} MB", bytes.div_ceil(MIB)) |
| 304 | } else if bytes >= KIB { |
| 305 | format!("{} KB", bytes.div_ceil(KIB)) |
| 306 | } else { |
| 307 | format!("{bytes} B") |
| 308 | } |
| 309 | } |
| 310 | |
| 311 | #[cfg(test)] |
| 312 | mod tests { |
| 313 | use super::*; |
| 314 | |
| 315 | struct TestArtifactSessionsRoot { |
| 316 | prior: Option<PathBuf>, |
| 317 | } |
| 318 | |
| 319 | impl Drop for TestArtifactSessionsRoot { |
| 320 | fn drop(&mut self) { |
| 321 | set_test_artifact_sessions_root(self.prior.take()); |
| 322 | } |
| 323 | } |
| 324 | |
| 325 | fn set_test_sessions_root(root: PathBuf) -> TestArtifactSessionsRoot { |
| 326 | TestArtifactSessionsRoot { |
| 327 | prior: set_test_artifact_sessions_root(Some(root)), |
| 328 | } |
| 329 | } |
| 330 | |
| 331 | #[test] |
| 332 | fn session_artifact_absolute_path_uses_test_sessions_root() { |
| 333 | let _guard = TEST_ARTIFACT_SESSIONS_GUARD |
| 334 | .lock() |
| 335 | .unwrap_or_else(|err| err.into_inner()); |
| 336 | let tmp = tempfile::tempdir().unwrap(); |
| 337 | let _root = set_test_sessions_root(tmp.path().join("sessions")); |
| 338 | |
| 339 | let path = session_artifact_absolute_path( |
| 340 | "session-123", |
| 341 | &PathBuf::from("artifacts").join("art_call-big.txt"), |
| 342 | ) |
| 343 | .expect("path"); |
| 344 | |
| 345 | assert_eq!( |
| 346 | path, |
| 347 | tmp.path() |
| 348 | .join("sessions") |
| 349 | .join("session-123") |
| 350 | .join("artifacts") |
| 351 | .join("art_call-big.txt") |
| 352 | ); |
| 353 | } |
| 354 | |
| 355 | #[test] |
| 356 | fn binary_session_artifact_uses_validated_extension_and_exact_bytes() { |
| 357 | let _guard = TEST_ARTIFACT_SESSIONS_GUARD |
| 358 | .lock() |
| 359 | .unwrap_or_else(|err| err.into_inner()); |
| 360 | let tmp = tempfile::tempdir().unwrap(); |
| 361 | let _root = set_test_sessions_root(tmp.path().join("sessions")); |
| 362 | let bytes = b"\x89PNG\r\n\x1a\nfixture"; |
| 363 | |
| 364 | let (absolute, relative) = |
| 365 | write_session_artifact_bytes("session-123", "web/media", ".PNG", bytes) |
| 366 | .expect("write binary artifact"); |
| 367 | |
| 368 | assert_eq!(relative, PathBuf::from("artifacts/web_media.png")); |
| 369 | assert_eq!(std::fs::read(absolute).unwrap(), bytes); |
| 370 | assert!(write_session_artifact_bytes("session-123", "bad", "../png", bytes).is_err()); |
| 371 | } |
| 372 | |
| 373 | #[test] |
| 374 | fn adaptive_evidence_publication_is_immutable_and_replay_safe() { |
| 375 | let _guard = TEST_ARTIFACT_SESSIONS_GUARD |
| 376 | .lock() |
| 377 | .unwrap_or_else(|err| err.into_inner()); |
| 378 | let tmp = tempfile::tempdir().unwrap(); |
| 379 | let _root = set_test_sessions_root(tmp.path().join("sessions")); |
| 380 | let relative = PathBuf::from("artifacts/art_call.txt"); |
| 381 | let bytes = b"first exact payload"; |
| 382 | |
| 383 | let first = write_session_relative_immutable("session-a", &relative, bytes).unwrap(); |
| 384 | let replay = write_session_relative_immutable("session-a", &relative, bytes).unwrap(); |
| 385 | assert_eq!(first, replay); |
| 386 | let err = write_session_relative_immutable("session-a", &relative, b"different") |
| 387 | .expect_err("handle aliasing must fail"); |
| 388 | assert_eq!(err.kind(), io::ErrorKind::AlreadyExists); |
| 389 | assert_eq!(std::fs::read(first).unwrap(), bytes); |
| 390 | } |
| 391 | |
| 392 | #[cfg(unix)] |
| 393 | #[test] |
| 394 | fn immutable_session_artifact_rejects_symlinked_parent() { |
| 395 | let _guard = TEST_ARTIFACT_SESSIONS_GUARD |
| 396 | .lock() |
| 397 | .unwrap_or_else(|err| err.into_inner()); |
| 398 | let tmp = tempfile::tempdir().unwrap(); |
| 399 | let sessions = tmp.path().join("sessions"); |
| 400 | let _root = set_test_sessions_root(sessions.clone()); |
| 401 | std::fs::create_dir_all(&sessions).unwrap(); |
| 402 | let outside = tempfile::tempdir().unwrap(); |
| 403 | std::os::unix::fs::symlink(outside.path(), sessions.join("session-a")).unwrap(); |
| 404 | assert!( |
| 405 | write_session_relative_immutable( |
| 406 | "session-a", |
| 407 | Path::new("artifacts/handoff.json"), |
| 408 | b"private" |
| 409 | ) |
| 410 | .is_err() |
| 411 | ); |
| 412 | assert!(!outside.path().join("artifacts").exists()); |
| 413 | } |
| 414 | |
| 415 | #[test] |
| 416 | fn adaptive_evidence_failed_publication_creates_no_handle() { |
| 417 | let _guard = TEST_ARTIFACT_SESSIONS_GUARD |
| 418 | .lock() |
| 419 | .unwrap_or_else(|err| err.into_inner()); |
| 420 | let tmp = tempfile::tempdir().unwrap(); |
| 421 | let sessions = tmp.path().join("sessions"); |
| 422 | let _root = set_test_sessions_root(sessions.clone()); |
| 423 | std::fs::create_dir_all(sessions.join("session-a")).unwrap(); |
| 424 | std::fs::write(sessions.join("session-a/artifacts"), b"block directory").unwrap(); |
| 425 | let relative = PathBuf::from("artifacts/art_failed.txt"); |
| 426 | |
| 427 | assert!(write_session_relative_immutable("session-a", &relative, b"payload").is_err()); |
| 428 | assert!(!sessions.join("session-a/artifacts/art_failed.txt").exists()); |
| 429 | } |
| 430 | } |
| 431 |