| 1 | //! Side-git repository wrapper for workspace snapshots. |
| 2 | //! |
| 3 | //! `SnapshotRepo` shells out to the system `git` binary (we deliberately |
| 4 | //! avoid `git2` to dodge its LGPL surface). The two paths that matter: |
| 5 | //! |
| 6 | //! - `git_dir` → `~/.deepseek/snapshots/<project_hash>/<worktree_hash>/.git` |
| 7 | //! - `work_tree` → the user's actual workspace |
| 8 | //! |
| 9 | //! Every git invocation passes both `--git-dir` AND `--work-tree`. That is |
| 10 | //! the single biggest safety mechanism: it guarantees we never accidentally |
| 11 | //! mutate the user's own `.git` directory. If git can't find the side |
| 12 | //! repo, the command fails fast instead of falling back to "current |
| 13 | //! directory". |
| 14 | |
| 15 | use std::collections::HashSet; |
| 16 | use std::io; |
| 17 | use std::path::{Component, Path, PathBuf}; |
| 18 | use std::process::Output; |
| 19 | use std::time::{Duration, SystemTime, UNIX_EPOCH}; |
| 20 | |
| 21 | use crate::dependencies::ExternalTool; |
| 22 | |
| 23 | use super::paths::{ensure_snapshot_dir, snapshot_git_dir}; |
| 24 | |
| 25 | /// Identifier for a snapshot — currently the underlying git commit SHA. |
| 26 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 27 | pub struct SnapshotId(pub String); |
| 28 | |
| 29 | impl SnapshotId { |
| 30 | /// Borrow the SHA as a string slice. |
| 31 | pub fn as_str(&self) -> &str { |
| 32 | &self.0 |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | /// A single snapshot record (one row in `git log`). |
| 37 | #[derive(Debug, Clone)] |
| 38 | pub struct Snapshot { |
| 39 | /// Commit SHA inside the side repo. |
| 40 | pub id: SnapshotId, |
| 41 | /// Subject line — the label passed to [`SnapshotRepo::snapshot`]. |
| 42 | pub label: String, |
| 43 | /// Author timestamp (Unix seconds). |
| 44 | pub timestamp: i64, |
| 45 | /// Session this snapshot belongs to, when recorded (encoded as a |
| 46 | /// `[sid=...] ` label prefix). `None` for legacy snapshots taken |
| 47 | /// before session tagging existed. |
| 48 | pub session_id: Option<String>, |
| 49 | } |
| 50 | |
| 51 | /// What a file-scoped restore did to one path, relative to the working tree |
| 52 | /// it was applied to. |
| 53 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 54 | pub enum PathRestoreAction { |
| 55 | /// Both the snapshot and the working tree had the path; its content came |
| 56 | /// back from the snapshot. |
| 57 | Modified, |
| 58 | /// The snapshot had the path and the working tree no longer did, so the |
| 59 | /// restore recreated the file. |
| 60 | Recreated, |
| 61 | /// The working tree had the path and the snapshot did not, so the restore |
| 62 | /// removed the file. |
| 63 | Removed, |
| 64 | } |
| 65 | |
| 66 | impl PathRestoreAction { |
| 67 | /// Stable wire name, also used by the runtime API response. |
| 68 | pub fn as_str(self) -> &'static str { |
| 69 | match self { |
| 70 | Self::Modified => "modified", |
| 71 | Self::Recreated => "recreated", |
| 72 | Self::Removed => "removed", |
| 73 | } |
| 74 | } |
| 75 | } |
| 76 | |
| 77 | /// Report of what [`SnapshotRepo::restore_paths`] did to one path. |
| 78 | #[derive(Debug, Clone)] |
| 79 | pub struct PathRestoreOutcome { |
| 80 | /// Workspace-relative path that was restored. |
| 81 | pub path: PathBuf, |
| 82 | /// How the working tree changed. |
| 83 | pub action: PathRestoreAction, |
| 84 | } |
| 85 | |
| 86 | /// Wrapper around the per-workspace side-git repo. |
| 87 | pub struct SnapshotRepo { |
| 88 | git_dir: PathBuf, |
| 89 | work_tree: PathBuf, |
| 90 | } |
| 91 | |
| 92 | const STALE_TMP_PACK_AGE: Duration = Duration::from_secs(60 * 60); |
| 93 | |
| 94 | /// Maximum total snapshot storage in megabytes before pruning kicks in at |
| 95 | /// snapshot time. Keeps the side repo from blowing up the user's disk during |
| 96 | /// long-running or high-churn sessions (#1112). |
| 97 | const MAX_SNAPSHOT_SIZE_MB: u64 = 500; |
| 98 | |
| 99 | const BYTES_PER_MB: u64 = 1024 * 1024; |
| 100 | |
| 101 | /// Grace margin below `MAX_SNAPSHOT_SIZE_MB` used as the prune target |
| 102 | /// so the repo doesn't hit the limit again one snapshot later. |
| 103 | const PRUNE_TARGET_MB: u64 = 400; |
| 104 | |
| 105 | /// Default workspace-size ceiling above which snapshots self-disable |
| 106 | /// on first use (2 GB of non-excluded content). Reports from users with |
| 107 | /// multi-hundred-GB project directories — datasets, model weights, |
| 108 | /// docker image dumps that fall outside the built-in excludes — |
| 109 | /// surfaced that `git add -A` on first init would hang the TUI for |
| 110 | /// minutes-to-hours while indexing the workspace. Snapshots are a |
| 111 | /// rollback safety net, not a backup tool; bailing out on workspaces |
| 112 | /// that big is the right tradeoff. Users with legitimate large |
| 113 | /// monorepos can raise `[snapshots] max_workspace_gb` (or set it to |
| 114 | /// `0` to disable the cap entirely). |
| 115 | pub const DEFAULT_MAX_WORKSPACE_BYTES_FOR_SNAPSHOT: u64 = 2 * 1024 * 1024 * 1024; |
| 116 | |
| 117 | /// Hard cap on the number of file entries the bounded size estimator |
| 118 | /// will inspect before declaring the workspace "too large". Protects |
| 119 | /// against a workspace with millions of tiny files (no individual |
| 120 | /// file is large, but `git add -A` would still take forever). |
| 121 | pub const SIZE_WALK_MAX_ENTRIES: usize = 200_000; |
| 122 | |
| 123 | /// Which snapshot gate refused a workspace. The recovery differs per gate — |
| 124 | /// raising `[snapshots] max_workspace_gb` lifts only [`WorkspaceGate::TooLarge`] |
| 125 | /// — so callers must not offer one gate's remedy for another's failure. |
| 126 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 127 | pub enum WorkspaceGate { |
| 128 | /// Snapshot-eligible content exceeds the configured byte cap. |
| 129 | TooLarge, |
| 130 | /// The bounded walk hit [`SIZE_WALK_MAX_ENTRIES`]. This bound is |
| 131 | /// independent of the byte cap: `max_workspace_gb = 0` does not lift it. |
| 132 | TooManyEntries, |
| 133 | } |
| 134 | |
| 135 | /// Leading text of the `io::Error` each gate produces. `core::turn` matches on |
| 136 | /// these to pick the right consequence/recovery notice, so they are one |
| 137 | /// declaration shared by producer and matcher rather than two literals. |
| 138 | pub const GATE_TOO_LARGE_MARKER: &str = "workspace too large for snapshots"; |
| 139 | pub const GATE_TOO_MANY_ENTRIES_MARKER: &str = "workspace has too many files for snapshots"; |
| 140 | pub const GATE_UNSAFE_LOCATION_MARKER: &str = "workspace snapshots are disabled"; |
| 141 | |
| 142 | /// Display a workspace path in gate diagnostics. The diagnostic names the |
| 143 | /// path the caller passed in — canonicalization is for filesystem and |
| 144 | /// security logic, not for the message. On Windows `Path::canonicalize` |
| 145 | /// rewrites more than the verbatim (`\\?\`) prefix (case, 8.3 names), so a |
| 146 | /// canonical spelling can never be relied on to match what users name; the |
| 147 | /// verbatim prefix is still stripped when present for readability. |
| 148 | fn display_workspace_for_gate(workspace: &Path) -> String { |
| 149 | let raw = workspace.display().to_string(); |
| 150 | raw.strip_prefix(r"\\?\") |
| 151 | .or_else(|| raw.strip_prefix("//?/")) |
| 152 | .unwrap_or(&raw) |
| 153 | .to_string() |
| 154 | } |
| 155 | |
| 156 | impl WorkspaceGate { |
| 157 | /// One-line English diagnostic for logs, `/undo`, and the gate matcher. |
| 158 | /// The user-facing consequence and recovery are localized by the notice |
| 159 | /// surfaces; this string must not restate them. |
| 160 | fn describe(self, cap_bytes: u64, workspace: &Path) -> String { |
| 161 | let workspace = display_workspace_for_gate(workspace); |
| 162 | match self { |
| 163 | Self::TooLarge => format!( |
| 164 | "{GATE_TOO_LARGE_MARKER}: over {} bytes of snapshot-eligible content in {workspace}", |
| 165 | cap_bytes, |
| 166 | ), |
| 167 | Self::TooManyEntries => format!( |
| 168 | "{GATE_TOO_MANY_ENTRIES_MARKER}: over {SIZE_WALK_MAX_ENTRIES} snapshot-eligible entries in {workspace}" |
| 169 | ), |
| 170 | } |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | /// Top-level directory and extension patterns that the snapshot path |
| 175 | /// already excludes via `BUILTIN_EXCLUDES`. The estimator skips these |
| 176 | /// up front so the size walk reflects what would actually land in the |
| 177 | /// snapshot commit. Kept narrow to common build-output dirs — anything |
| 178 | /// else falls back to the `.gitignore` filter. |
| 179 | const SIZE_WALK_SKIP_DIRS: &[&str] = &[ |
| 180 | "node_modules", |
| 181 | "target", |
| 182 | "dist", |
| 183 | "build", |
| 184 | ".build", |
| 185 | ".next", |
| 186 | ".nuxt", |
| 187 | ".svelte-kit", |
| 188 | ".turbo", |
| 189 | ".parcel-cache", |
| 190 | "vendor", |
| 191 | ".cargo", |
| 192 | ".rustup", |
| 193 | ".npm", |
| 194 | ".bun", |
| 195 | ".yarn", |
| 196 | ".pnpm-store", |
| 197 | ".cache", |
| 198 | ".venv", |
| 199 | "venv", |
| 200 | ".tox", |
| 201 | "__pycache__", |
| 202 | ".mypy_cache", |
| 203 | ".pytest_cache", |
| 204 | ".ruff_cache", |
| 205 | ".gradle", |
| 206 | ".m2", |
| 207 | ".local", |
| 208 | ".git", |
| 209 | ]; |
| 210 | |
| 211 | const BUILTIN_EXCLUDES: &str = "\ |
| 212 | # CodeWhale built-in snapshot exclusions |
| 213 | node_modules/ |
| 214 | target/ |
| 215 | dist/ |
| 216 | build/ |
| 217 | .build/ |
| 218 | .next/ |
| 219 | .nuxt/ |
| 220 | .svelte-kit/ |
| 221 | .turbo/ |
| 222 | .parcel-cache/ |
| 223 | vendor/ |
| 224 | .cargo/ |
| 225 | .rustup/ |
| 226 | .npm/ |
| 227 | .bun/ |
| 228 | .yarn/ |
| 229 | .pnpm-store/ |
| 230 | .cache/ |
| 231 | .venv/ |
| 232 | venv/ |
| 233 | .tox/ |
| 234 | __pycache__/ |
| 235 | *.pyc |
| 236 | .mypy_cache/ |
| 237 | .pytest_cache/ |
| 238 | .ruff_cache/ |
| 239 | .gradle/ |
| 240 | .m2/ |
| 241 | .local/ |
| 242 | .DS_Store |
| 243 | |
| 244 | # Binary and generated artifacts. Snapshots are source rollback checkpoints, |
| 245 | # not a full binary backup; keeping these out avoids side-repo bloat. |
| 246 | *.exe |
| 247 | *.dll |
| 248 | *.so |
| 249 | *.dylib |
| 250 | *.wasm |
| 251 | *.o |
| 252 | *.obj |
| 253 | *.class |
| 254 | *.pdb |
| 255 | *.dSYM |
| 256 | *.zip |
| 257 | *.tar |
| 258 | *.tar.gz |
| 259 | *.tgz |
| 260 | *.tar.bz2 |
| 261 | *.tar.xz |
| 262 | *.7z |
| 263 | *.rar |
| 264 | *.iso |
| 265 | *.dmg |
| 266 | *.bin |
| 267 | *.mp4 |
| 268 | *.mov |
| 269 | *.mkv |
| 270 | *.avi |
| 271 | *.webm |
| 272 | *.mp3 |
| 273 | *.wav |
| 274 | *.flac |
| 275 | *.aac |
| 276 | "; |
| 277 | |
| 278 | impl SnapshotRepo { |
| 279 | /// Open an existing snapshot repo for `workspace` without creating or |
| 280 | /// initializing anything on disk. |
| 281 | /// |
| 282 | /// This is useful for read-only UI surfaces that want to report checkpoint |
| 283 | /// availability without paying the first-init size walk or surprising the |
| 284 | /// user by creating a side repo from a view action. |
| 285 | pub fn open_existing(workspace: &Path) -> io::Result<Option<Self>> { |
| 286 | let work_tree = workspace |
| 287 | .canonicalize() |
| 288 | .unwrap_or_else(|_| workspace.to_path_buf()); |
| 289 | let git_dir = snapshot_git_dir(&work_tree); |
| 290 | if !git_dir.exists() || !git_dir.join("HEAD").exists() { |
| 291 | return Ok(None); |
| 292 | } |
| 293 | Ok(Some(Self { git_dir, work_tree })) |
| 294 | } |
| 295 | |
| 296 | /// Open or initialize the snapshot repo for `workspace`. |
| 297 | /// |
| 298 | /// On first use this: |
| 299 | /// 1. Creates the `~/.deepseek/snapshots/<…>/.git` dir. |
| 300 | /// 2. Runs `git init --bare=false --quiet`. |
| 301 | /// 3. Sets a fixed `user.name` / `user.email` so commits don't pick up |
| 302 | /// the user's global git identity (we don't want our snapshots to |
| 303 | /// look like they came from the user). |
| 304 | pub fn open_or_init(workspace: &Path) -> io::Result<Self> { |
| 305 | Self::open_or_init_with_cap(workspace, DEFAULT_MAX_WORKSPACE_BYTES_FOR_SNAPSHOT) |
| 306 | } |
| 307 | |
| 308 | /// Variant of [`Self::open_or_init`] that accepts an explicit |
| 309 | /// workspace-size cap. `cap_bytes = 0` disables the cap entirely |
| 310 | /// (always snapshot, regardless of size). |
| 311 | /// |
| 312 | /// When the workspace exceeds the cap and the side repo hasn't |
| 313 | /// been initialized yet, returns `Err(InvalidInput)` with a |
| 314 | /// "workspace too large" reason. Subsequent calls (after the user |
| 315 | /// shrinks the workspace or raises the cap via config) succeed. |
| 316 | pub fn open_or_init_with_cap(workspace: &Path, cap_bytes: u64) -> io::Result<Self> { |
| 317 | let work_tree = workspace |
| 318 | .canonicalize() |
| 319 | .unwrap_or_else(|_| workspace.to_path_buf()); |
| 320 | if let Some(reason) = unsafe_workspace_snapshot_reason( |
| 321 | &work_tree, |
| 322 | crate::config::effective_home_dir().as_deref(), |
| 323 | ) { |
| 324 | return Err(io::Error::new( |
| 325 | io::ErrorKind::InvalidInput, |
| 326 | format!( |
| 327 | "{GATE_UNSAFE_LOCATION_MARKER} for {reason}: {}", |
| 328 | display_workspace_for_gate(workspace) |
| 329 | ), |
| 330 | )); |
| 331 | } |
| 332 | |
| 333 | let _ = ensure_snapshot_dir(&work_tree)?; |
| 334 | let git_dir = snapshot_git_dir(&work_tree); |
| 335 | |
| 336 | let needs_init = !git_dir.exists(); |
| 337 | if needs_init { |
| 338 | // First-init size guard. Skipping this on subsequent opens |
| 339 | // is intentional: paying a workspace walk on every snapshot |
| 340 | // would defeat the purpose of the cap, and a workspace |
| 341 | // that fit on first init is allowed to grow within the |
| 342 | // existing repo's `MAX_SNAPSHOT_SIZE_MB` budget. Users on |
| 343 | // workspaces that grew past the cap mid-session get the |
| 344 | // existing aggressive-pruning path in `snapshot()`. |
| 345 | if let Err(gate) = |
| 346 | estimate_workspace_size_bounded(&work_tree, cap_bytes, SIZE_WALK_MAX_ENTRIES) |
| 347 | { |
| 348 | return Err(io::Error::new( |
| 349 | io::ErrorKind::InvalidInput, |
| 350 | gate.describe(cap_bytes, workspace), |
| 351 | )); |
| 352 | } |
| 353 | let parent = git_dir.parent().ok_or_else(|| { |
| 354 | io::Error::new(io::ErrorKind::InvalidInput, "snapshot dir has no parent") |
| 355 | })?; |
| 356 | std::fs::create_dir_all(parent)?; |
| 357 | // `git init` here uses the parent directory as the work tree |
| 358 | // and stores metadata in `.git`. We then continue to use |
| 359 | // explicit `--git-dir` / `--work-tree` flags for every other |
| 360 | // command so behaviour is invariant of cwd. |
| 361 | let init = crate::dependencies::Git::command() |
| 362 | .ok_or_else(|| io_other("git not found on PATH"))? |
| 363 | .arg("init") |
| 364 | .arg("--quiet") |
| 365 | .arg(parent) |
| 366 | .output() |
| 367 | .map_err(|e| io_other(format!("failed to spawn git init: {e}")))?; |
| 368 | if !init.status.success() { |
| 369 | return Err(io_other(format!( |
| 370 | "git init failed: {}", |
| 371 | String::from_utf8_lossy(&init.stderr).trim() |
| 372 | ))); |
| 373 | } |
| 374 | |
| 375 | // Pin a stable identity so snapshot commits are recognisable |
| 376 | // and don't bleed into the user's git config. |
| 377 | let _ = run_git( |
| 378 | &git_dir, |
| 379 | &work_tree, |
| 380 | &["config", "user.name", "deepseek-snapshots"], |
| 381 | ); |
| 382 | let _ = run_git( |
| 383 | &git_dir, |
| 384 | &work_tree, |
| 385 | &["config", "user.email", "snapshots@codewhale.local"], |
| 386 | ); |
| 387 | // Don't auto-gc on every commit; we manage pruning ourselves. |
| 388 | let _ = run_git(&git_dir, &work_tree, &["config", "gc.auto", "0"]); |
| 389 | // Ignore CRLF rewriting — we want byte-for-byte fidelity. |
| 390 | let _ = run_git(&git_dir, &work_tree, &["config", "core.autocrlf", "false"]); |
| 391 | } |
| 392 | |
| 393 | write_builtin_excludes(&git_dir)?; |
| 394 | if let Err(err) = cleanup_stale_pack_temps(&git_dir, STALE_TMP_PACK_AGE) { |
| 395 | tracing::debug!( |
| 396 | target: "snapshot", |
| 397 | "failed to clean stale snapshot tmp_pack files: {err}" |
| 398 | ); |
| 399 | } |
| 400 | Ok(Self { git_dir, work_tree }) |
| 401 | } |
| 402 | |
| 403 | /// Take a snapshot of the current working tree. |
| 404 | /// |
| 405 | /// Internally: `git add -A`, `git write-tree`, `git commit-tree`, then |
| 406 | /// `git update-ref HEAD <commit>`. |
| 407 | /// `git add -A` honours the user's workspace ignore rules while staging |
| 408 | /// into the side repo's index. |
| 409 | /// |
| 410 | /// Before committing, checks whether the snapshot directory exceeds |
| 411 | /// [`MAX_SNAPSHOT_SIZE_MB`] and prunes the oldest snapshots if it does. |
| 412 | /// |
| 413 | /// Returns the snapshot's commit SHA. |
| 414 | #[allow(dead_code)] // convenience entry kept for tests and legacy callers; production writes go through snapshot_with_session |
| 415 | pub fn snapshot(&self, label: &str) -> io::Result<SnapshotId> { |
| 416 | self.snapshot_with_session(label, None) |
| 417 | } |
| 418 | |
| 419 | /// Take a snapshot, tagging it with the owning session id. |
| 420 | /// |
| 421 | /// The session id is encoded into the commit message as a `[sid=...] ` |
| 422 | /// label prefix. [`Self::list`] decodes it back into |
| 423 | /// [`Snapshot::session_id`] and strips the prefix from the visible |
| 424 | /// label, so existing listing surfaces keep showing the plain label. |
| 425 | /// Legacy snapshots taken through [`Self::snapshot`] carry no prefix |
| 426 | /// and decode with `session_id == None`. |
| 427 | pub fn snapshot_with_session( |
| 428 | &self, |
| 429 | label: &str, |
| 430 | session_id: Option<&str>, |
| 431 | ) -> io::Result<SnapshotId> { |
| 432 | // Guard against disk blowup (#1112): if the snapshot directory has |
| 433 | // grown beyond the limit, prune aggressively before adding more. |
| 434 | // When the prune actually destroys restore points the user is told |
| 435 | // once per workspace — losing undo history to a log line is the S5 |
| 436 | // failure mode (2026-08-04 snapshot hunt). |
| 437 | if let Ok(removed) = self.prune_size_pressure( |
| 438 | MAX_SNAPSHOT_SIZE_MB * BYTES_PER_MB, |
| 439 | PRUNE_TARGET_MB * BYTES_PER_MB, |
| 440 | ) && removed > 0 |
| 441 | { |
| 442 | notify_snapshot_history_pruned_once(&self.work_tree, removed); |
| 443 | } |
| 444 | // Stage every tracked + untracked path the workspace exposes. |
| 445 | // `--all` here means `add` + `update` + `remove` — the same set |
| 446 | // `git status` would show. |
| 447 | let add = run_git(&self.git_dir, &self.work_tree, &["add", "-A"])?; |
| 448 | if !add.status.success() { |
| 449 | return Err(io_other(format!( |
| 450 | "git add -A failed: {}", |
| 451 | String::from_utf8_lossy(&add.stderr).trim() |
| 452 | ))); |
| 453 | } |
| 454 | |
| 455 | let tree = run_git(&self.git_dir, &self.work_tree, &["write-tree"])?; |
| 456 | if !tree.status.success() { |
| 457 | return Err(io_other(format!( |
| 458 | "git write-tree failed: {}", |
| 459 | String::from_utf8_lossy(&tree.stderr).trim() |
| 460 | ))); |
| 461 | } |
| 462 | let tree = String::from_utf8_lossy(&tree.stdout).trim().to_string(); |
| 463 | |
| 464 | let parent = run_git( |
| 465 | &self.git_dir, |
| 466 | &self.work_tree, |
| 467 | &["rev-parse", "--verify", "HEAD"], |
| 468 | )?; |
| 469 | let parent = parent |
| 470 | .status |
| 471 | .success() |
| 472 | .then(|| String::from_utf8_lossy(&parent.stdout).trim().to_string()) |
| 473 | .filter(|s| !s.is_empty()); |
| 474 | |
| 475 | let mut args = vec!["commit-tree".to_string(), tree]; |
| 476 | if let Some(parent) = parent { |
| 477 | args.push("-p".to_string()); |
| 478 | args.push(parent); |
| 479 | } |
| 480 | args.push("-m".to_string()); |
| 481 | args.push(Self::encode_session_label(label, session_id)); |
| 482 | let arg_refs: Vec<&str> = args.iter().map(String::as_str).collect(); |
| 483 | |
| 484 | // `commit-tree` creates marker commits even when the tree matches its |
| 485 | // parent, and it does not run user/global commit hooks. |
| 486 | let commit = run_git(&self.git_dir, &self.work_tree, &arg_refs)?; |
| 487 | if !commit.status.success() { |
| 488 | return Err(io_other(format!( |
| 489 | "git commit-tree failed: {}", |
| 490 | String::from_utf8_lossy(&commit.stderr).trim() |
| 491 | ))); |
| 492 | } |
| 493 | let sha = String::from_utf8_lossy(&commit.stdout).trim().to_string(); |
| 494 | |
| 495 | let update = run_git( |
| 496 | &self.git_dir, |
| 497 | &self.work_tree, |
| 498 | &["update-ref", "HEAD", &sha], |
| 499 | )?; |
| 500 | if !update.status.success() { |
| 501 | return Err(io_other(format!( |
| 502 | "git update-ref HEAD failed: {}", |
| 503 | String::from_utf8_lossy(&update.stderr).trim() |
| 504 | ))); |
| 505 | } |
| 506 | |
| 507 | Ok(SnapshotId(sha)) |
| 508 | } |
| 509 | |
| 510 | /// Prefix a snapshot label with its owning session id, if any. |
| 511 | fn encode_session_label(label: &str, session_id: Option<&str>) -> String { |
| 512 | match session_id { |
| 513 | Some(sid) if !sid.is_empty() => format!("[sid={sid}] {label}"), |
| 514 | _ => label.to_string(), |
| 515 | } |
| 516 | } |
| 517 | |
| 518 | /// Split a possibly session-tagged label back into `(session_id, label)`. |
| 519 | /// |
| 520 | /// Returns `(None, label)` for untagged labels. The decoded label is |
| 521 | /// the original one without the `[sid=...] ` prefix, so consumers that |
| 522 | /// match on `pre-turn:`/`tool:`/`redo:` prefixes keep working unchanged. |
| 523 | fn decode_session_label(label: &str) -> (Option<String>, String) { |
| 524 | let Some(rest) = label.strip_prefix("[sid=") else { |
| 525 | return (None, label.to_string()); |
| 526 | }; |
| 527 | let Some(end) = rest.find("] ") else { |
| 528 | return (None, label.to_string()); |
| 529 | }; |
| 530 | let sid = &rest[..end]; |
| 531 | let plain = &rest[end + 2..]; |
| 532 | if sid.is_empty() || plain.is_empty() { |
| 533 | return (None, label.to_string()); |
| 534 | } |
| 535 | (Some(sid.to_string()), plain.to_string()) |
| 536 | } |
| 537 | /// Size-pressure prune (#1112): if the side repo exceeds `max_bytes`, |
| 538 | /// walk backward from a 1-second retention toward zero until the store is |
| 539 | /// at or under `target_bytes`, escalating to a full wipe when nothing |
| 540 | /// else helps. Returns the total number of snapshots destroyed, so the |
| 541 | /// caller can tell the user their undo history shrank (S5 — the wipe was |
| 542 | /// previously announced only by a `tracing::warn`). |
| 543 | fn prune_size_pressure(&self, max_bytes: u64, target_bytes: u64) -> io::Result<usize> { |
| 544 | let current_bytes = dir_size_bytes(&self.git_dir)?; |
| 545 | if current_bytes <= max_bytes { |
| 546 | return Ok(0); |
| 547 | } |
| 548 | tracing::warn!( |
| 549 | target: "snapshot", |
| 550 | current_mb = current_bytes / BYTES_PER_MB, |
| 551 | limit_mb = max_bytes / BYTES_PER_MB, |
| 552 | "snapshot storage approaching limit — pruning aggressively" |
| 553 | ); |
| 554 | let mut removed_total: usize = 0; |
| 555 | // Walk backward from a 1-second retention to zero until |
| 556 | // we're under the target, or until there's nothing left. |
| 557 | let mut age = Duration::from_secs(1); |
| 558 | for _ in 0..10 { |
| 559 | if let Ok(removed) = self.prune_older_than(age) { |
| 560 | removed_total = removed_total.saturating_add(removed); |
| 561 | } |
| 562 | if let Ok(new_size) = dir_size_bytes(&self.git_dir) |
| 563 | && new_size <= target_bytes |
| 564 | { |
| 565 | tracing::info!( |
| 566 | target: "snapshot", |
| 567 | new_size_mb = new_size / BYTES_PER_MB, |
| 568 | "pruned snapshot storage back under limit" |
| 569 | ); |
| 570 | break; |
| 571 | } |
| 572 | age = age.saturating_sub(Duration::from_millis(100)); |
| 573 | } |
| 574 | // Fallback: if even 0-second pruning didn't help (shouldn't |
| 575 | // happen but belt-and-suspenders), nuke the refs so the next |
| 576 | // snapshot starts a fresh history. |
| 577 | if let Ok(final_size) = dir_size_bytes(&self.git_dir) |
| 578 | && final_size > max_bytes |
| 579 | { |
| 580 | tracing::warn!( |
| 581 | target: "snapshot", |
| 582 | "snapshot storage still over limit after pruning; wiping history" |
| 583 | ); |
| 584 | if let Ok(removed) = self.prune_older_than(Duration::ZERO) { |
| 585 | removed_total = removed_total.saturating_add(removed); |
| 586 | } |
| 587 | let _ = self.prune_unreachable_objects(); |
| 588 | } |
| 589 | Ok(removed_total) |
| 590 | } |
| 591 | |
| 592 | /// Restore the workspace to the state at `id`. |
| 593 | /// |
| 594 | /// Uses `git checkout <sha> -- :/` which checks out every path in the |
| 595 | /// snapshot tree relative to the workspace root. We do NOT touch the |
| 596 | /// user's own `.git` — snapshots only contain working-tree files. |
| 597 | pub fn restore(&self, id: &SnapshotId) -> io::Result<()> { |
| 598 | // Restore is the one destructive operation with no undo of its own. |
| 599 | // Capture the pre-restore state first so the restore itself can be |
| 600 | // reversed (2026-08-04 snapshot hunt: makes several other findings |
| 601 | // recoverable instead of final). The `pre-restore:` prefix is |
| 602 | // deliberately not a `/undo` or `revert_turn` candidate label, so the |
| 603 | // safety net never changes snapshot selection. Best-effort: a failed |
| 604 | // safety snapshot must never block the restore the user asked for. |
| 605 | let target_short = &id.as_str()[..id.as_str().len().min(12)]; |
| 606 | if let Err(e) = self.snapshot_with_session(&format!("pre-restore:{target_short}"), None) { |
| 607 | tracing::warn!( |
| 608 | target: "snapshot", |
| 609 | "pre-restore safety snapshot failed (restore will proceed): {e}" |
| 610 | ); |
| 611 | } |
| 612 | let current_paths = self.tree_paths("HEAD")?; |
| 613 | let target_paths = self.tree_paths(id.as_str())?; |
| 614 | let checkout = run_git( |
| 615 | &self.git_dir, |
| 616 | &self.work_tree, |
| 617 | &["checkout", id.as_str(), "--", ":/"], |
| 618 | )?; |
| 619 | if !checkout.status.success() { |
| 620 | return Err(io_other(format!( |
| 621 | "git checkout failed: {}", |
| 622 | String::from_utf8_lossy(&checkout.stderr).trim() |
| 623 | ))); |
| 624 | } |
| 625 | self.remove_paths_missing_from_target(¤t_paths, &target_paths)?; |
| 626 | Ok(()) |
| 627 | } |
| 628 | |
| 629 | /// File restore never traverses symlinks, directories, or Git metadata. |
| 630 | /// Validate every existing component before reading, backing up or writing. |
| 631 | pub fn validate_restore_file(&self, rel: &Path) -> io::Result<bool> { |
| 632 | if !is_safe_relative_path(rel) |
| 633 | || rel.components().any(|part| { |
| 634 | part.as_os_str() |
| 635 | .as_encoded_bytes() |
| 636 | .eq_ignore_ascii_case(b".git") |
| 637 | }) |
| 638 | { |
| 639 | return Err(io::Error::new( |
| 640 | io::ErrorKind::InvalidInput, |
| 641 | format!( |
| 642 | "refusing to restore unsafe path '{}': restore requires a regular workspace file", |
| 643 | rel.display() |
| 644 | ), |
| 645 | )); |
| 646 | } |
| 647 | let mut path = self.work_tree.clone(); |
| 648 | for part in rel.components() { |
| 649 | path.push(part); |
| 650 | match std::fs::symlink_metadata(&path) { |
| 651 | Ok(meta) |
| 652 | if meta.file_type().is_symlink() |
| 653 | || (path == self.work_tree.join(rel) && !meta.is_file()) |
| 654 | || (path != self.work_tree.join(rel) && !meta.is_dir()) => |
| 655 | { |
| 656 | return Err(io::Error::new( |
| 657 | io::ErrorKind::InvalidInput, |
| 658 | "restore refuses directories, symlinks and non-regular files", |
| 659 | )); |
| 660 | } |
| 661 | Ok(_) => {} |
| 662 | Err(error) if error.kind() == io::ErrorKind::NotFound => return Ok(false), |
| 663 | Err(error) => return Err(error), |
| 664 | } |
| 665 | } |
| 666 | Ok(true) |
| 667 | } |
| 668 | |
| 669 | fn snapshot_contains_regular_file(&self, id: &SnapshotId, rel: &Path) -> io::Result<bool> { |
| 670 | let entry = run_git( |
| 671 | &self.git_dir, |
| 672 | &self.work_tree, |
| 673 | &[ |
| 674 | "--literal-pathspecs", |
| 675 | "ls-tree", |
| 676 | "-z", |
| 677 | id.as_str(), |
| 678 | "--", |
| 679 | rel.to_str() |
| 680 | .ok_or_else(|| io_other("restore path must be UTF-8"))?, |
| 681 | ], |
| 682 | )?; |
| 683 | if !entry.status.success() { |
| 684 | return Err(io_other(format!( |
| 685 | "Failed to inspect snapshot file: {}", |
| 686 | String::from_utf8_lossy(&entry.stderr).trim() |
| 687 | ))); |
| 688 | } |
| 689 | if entry.stdout.is_empty() { |
| 690 | return Ok(false); |
| 691 | } |
| 692 | if !(entry.stdout.starts_with(b"100644 blob ") || entry.stdout.starts_with(b"100755 blob ")) |
| 693 | { |
| 694 | return Err(io::Error::new( |
| 695 | io::ErrorKind::InvalidInput, |
| 696 | "snapshot path is not a regular file", |
| 697 | )); |
| 698 | } |
| 699 | Ok(true) |
| 700 | } |
| 701 | |
| 702 | /// Return whether `rel` differs between snapshot `id` and the current |
| 703 | /// working tree. |
| 704 | /// |
| 705 | /// This is the single-path counterpart of |
| 706 | /// [`Self::work_tree_matches_snapshot`]: it answers "would restoring just |
| 707 | /// this file change anything?", which is what file-scoped revert |
| 708 | /// cursoring needs. A path that exists in neither the snapshot nor the |
| 709 | /// working tree does not differ. |
| 710 | pub fn path_differs_from_snapshot(&self, id: &SnapshotId, rel: &Path) -> io::Result<bool> { |
| 711 | let in_work = self.validate_restore_file(rel)?; |
| 712 | let in_target = self.snapshot_contains_regular_file(id, rel)?; |
| 713 | match (in_target, in_work) { |
| 714 | // Neither side has it: nothing to restore and nothing to remove. |
| 715 | (false, false) => Ok(false), |
| 716 | // The snapshot has it and the working tree lost it. |
| 717 | (true, false) => Ok(true), |
| 718 | // The path was created after the snapshot. |
| 719 | (false, true) => Ok(true), |
| 720 | (true, true) => { |
| 721 | let rel = rel.to_string_lossy().into_owned(); |
| 722 | let diff = run_git( |
| 723 | &self.git_dir, |
| 724 | &self.work_tree, |
| 725 | &[ |
| 726 | "--literal-pathspecs", |
| 727 | "diff", |
| 728 | "--quiet", |
| 729 | id.as_str(), |
| 730 | "--", |
| 731 | rel.as_str(), |
| 732 | ], |
| 733 | )?; |
| 734 | git_diff_matches(diff).map(|matches| !matches) |
| 735 | } |
| 736 | } |
| 737 | } |
| 738 | |
| 739 | /// Restore only `rel_paths` from snapshot `id`. |
| 740 | /// |
| 741 | /// This is the file-scoped counterpart of [`Self::restore`]. The |
| 742 | /// difference that matters: the whole-tree `git checkout <sha> -- :/` is |
| 743 | /// replaced by a pathspec-limited checkout, so a working-tree path outside |
| 744 | /// `rel_paths` is never written or deleted. The safety backup reads the workspace. |
| 745 | /// |
| 746 | /// A path the snapshot does not track is removed from the working tree |
| 747 | /// (that is how a file created after the snapshot is reverted), and a path |
| 748 | /// the snapshot tracks but the working tree lost is recreated. A path that |
| 749 | /// exists in neither side produces no outcome at all, rather than a |
| 750 | /// report claiming a change that did not happen. |
| 751 | #[cfg(test)] |
| 752 | pub fn restore_paths( |
| 753 | &self, |
| 754 | id: &SnapshotId, |
| 755 | rel_paths: &[PathBuf], |
| 756 | ) -> io::Result<Vec<PathRestoreOutcome>> { |
| 757 | self.restore_paths_checked(id, rel_paths, || Ok(())) |
| 758 | } |
| 759 | |
| 760 | pub fn restore_file_if_unchanged( |
| 761 | &self, |
| 762 | id: &SnapshotId, |
| 763 | rel: &Path, |
| 764 | expected_hash: &str, |
| 765 | ) -> io::Result<Vec<PathRestoreOutcome>> { |
| 766 | let verify = || { |
| 767 | let actual = if self.validate_restore_file(rel)? { |
| 768 | let bytes = std::fs::read(self.work_tree.join(rel))?; |
| 769 | format!("sha256:{}", crate::hashing::sha256_hex(bytes)) |
| 770 | } else { |
| 771 | "absent".to_string() |
| 772 | }; |
| 773 | if actual != expected_hash { |
| 774 | return Err(io::Error::new( |
| 775 | io::ErrorKind::WouldBlock, |
| 776 | "The file changed after the selected change record. Refresh and review it before restoring; nothing was changed.", |
| 777 | )); |
| 778 | } |
| 779 | Ok(()) |
| 780 | }; |
| 781 | verify()?; |
| 782 | self.restore_paths_checked(id, &[rel.to_path_buf()], verify) |
| 783 | } |
| 784 | |
| 785 | fn restore_paths_checked( |
| 786 | &self, |
| 787 | id: &SnapshotId, |
| 788 | rel_paths: &[PathBuf], |
| 789 | preflight: impl FnOnce() -> io::Result<()>, |
| 790 | ) -> io::Result<Vec<PathRestoreOutcome>> { |
| 791 | if rel_paths.is_empty() { |
| 792 | return Ok(Vec::new()); |
| 793 | } |
| 794 | // Validate the entire request before any mutation or backup. A snapshot |
| 795 | // directory entry must not turn a file action into recursive checkout. |
| 796 | let mut pre_state = Vec::with_capacity(rel_paths.len()); |
| 797 | for rel in rel_paths { |
| 798 | let in_work = self.validate_restore_file(rel)?; |
| 799 | let in_target = self.snapshot_contains_regular_file(id, rel)?; |
| 800 | pre_state.push((rel.clone(), in_target, in_work)); |
| 801 | } |
| 802 | |
| 803 | // A durable backup is required for this new destructive API. Ignored |
| 804 | // files cannot be removed/overwritten if the snapshot cannot retain them. |
| 805 | let target_short = &id.as_str()[..id.as_str().len().min(12)]; |
| 806 | let backup = self.snapshot_with_session(&format!("pre-restore:{target_short}"), None)?; |
| 807 | for (rel, _, in_work) in &pre_state { |
| 808 | if *in_work && !self.snapshot_contains_regular_file(&backup, rel)? { |
| 809 | return Err(io_other( |
| 810 | "File was excluded from the safety snapshot; nothing was restored", |
| 811 | )); |
| 812 | } |
| 813 | self.validate_restore_file(rel)?; |
| 814 | } |
| 815 | |
| 816 | // Recheck after the potentially slow safety snapshot, immediately |
| 817 | // before checkout/removal. New editor work is retained in the backup. |
| 818 | preflight()?; |
| 819 | |
| 820 | let tracked: Vec<String> = pre_state |
| 821 | .iter() |
| 822 | .filter(|(_, in_target, _)| *in_target) |
| 823 | .map(|(rel, _, _)| rel.to_string_lossy().into_owned()) |
| 824 | .collect(); |
| 825 | if !tracked.is_empty() { |
| 826 | let mut args: Vec<String> = vec![ |
| 827 | "--literal-pathspecs".to_string(), |
| 828 | "checkout".to_string(), |
| 829 | id.as_str().to_string(), |
| 830 | "--".to_string(), |
| 831 | ]; |
| 832 | args.extend(tracked); |
| 833 | let arg_refs: Vec<&str> = args.iter().map(String::as_str).collect(); |
| 834 | let checkout = run_git(&self.git_dir, &self.work_tree, &arg_refs)?; |
| 835 | if !checkout.status.success() { |
| 836 | return Err(io_other(format!( |
| 837 | "git checkout failed: {} (safety snapshot {} holds the previous files)", |
| 838 | String::from_utf8_lossy(&checkout.stderr).trim(), |
| 839 | backup.as_str() |
| 840 | ))); |
| 841 | } |
| 842 | } |
| 843 | |
| 844 | let mut outcomes = Vec::new(); |
| 845 | for (rel, in_target, was_in_work) in pre_state { |
| 846 | match (in_target, was_in_work) { |
| 847 | (true, true) => outcomes.push(PathRestoreOutcome { |
| 848 | path: rel, |
| 849 | action: PathRestoreAction::Modified, |
| 850 | }), |
| 851 | (true, false) => outcomes.push(PathRestoreOutcome { |
| 852 | path: rel, |
| 853 | action: PathRestoreAction::Recreated, |
| 854 | }), |
| 855 | (false, true) => { |
| 856 | let path = self.work_tree.join(&rel); |
| 857 | self.validate_restore_file(&rel)?; |
| 858 | // Only the requested file goes; its parent directories |
| 859 | // stay even when emptied, because the request named a |
| 860 | // file, not a tree. |
| 861 | std::fs::remove_file(&path).map_err(|error| { |
| 862 | io_other(format!( |
| 863 | "removing '{}' failed: {error} (safety snapshot {} holds the previous files)", |
| 864 | rel.display(), |
| 865 | backup.as_str() |
| 866 | )) |
| 867 | })?; |
| 868 | outcomes.push(PathRestoreOutcome { |
| 869 | path: rel, |
| 870 | action: PathRestoreAction::Removed, |
| 871 | }); |
| 872 | } |
| 873 | // Already in the snapshot's state. |
| 874 | (false, false) => {} |
| 875 | } |
| 876 | } |
| 877 | Ok(outcomes) |
| 878 | } |
| 879 | |
| 880 | /// `git diff --stat` between snapshot `id` and the current working tree, |
| 881 | /// computed inside the side repo. |
| 882 | /// |
| 883 | /// This is what restoring `id` *would* change, so it must be captured |
| 884 | /// before the restore runs — afterwards the work tree matches the snapshot |
| 885 | /// and the diff is empty by construction. |
| 886 | /// |
| 887 | /// It deliberately runs against the side repo rather than the user's. The |
| 888 | /// previous summary ran `git diff --stat` in the workspace with the user's |
| 889 | /// `.git`, which reports the user's own uncommitted work: it listed files |
| 890 | /// the restore had not touched, and reported nothing when that work |
| 891 | /// happened to be committed. Returns `None` when nothing differs. |
| 892 | pub fn snapshot_diff_stat(&self, id: &SnapshotId) -> io::Result<Option<String>> { |
| 893 | let diff = run_git( |
| 894 | &self.git_dir, |
| 895 | &self.work_tree, |
| 896 | &["diff", "--stat", id.as_str(), "--", ":/"], |
| 897 | )?; |
| 898 | if !diff.status.success() { |
| 899 | return Err(io_other(format!( |
| 900 | "git diff --stat failed: {}", |
| 901 | String::from_utf8_lossy(&diff.stderr).trim() |
| 902 | ))); |
| 903 | } |
| 904 | let stat = String::from_utf8_lossy(&diff.stdout).trim().to_string(); |
| 905 | Ok((!stat.is_empty()).then_some(stat)) |
| 906 | } |
| 907 | |
| 908 | /// Return whether the current workspace matches the given snapshot's |
| 909 | /// tracked file content. |
| 910 | /// |
| 911 | /// This is intentionally narrower than a full "workspace identical" |
| 912 | /// claim: it compares the current working tree against the snapshot's |
| 913 | /// tracked paths via git's diff machinery. That is sufficient for |
| 914 | /// `/undo` cursoring — if the diff is empty, restoring this snapshot |
| 915 | /// again would be a no-op, so the caller should continue scanning |
| 916 | /// older snapshots. |
| 917 | pub fn work_tree_matches_snapshot(&self, id: &SnapshotId) -> io::Result<bool> { |
| 918 | let diff = run_git( |
| 919 | &self.git_dir, |
| 920 | &self.work_tree, |
| 921 | &["diff", "--quiet", id.as_str(), "--", ":/"], |
| 922 | )?; |
| 923 | git_diff_matches(diff) |
| 924 | } |
| 925 | |
| 926 | fn tree_paths(&self, treeish: &str) -> io::Result<HashSet<PathBuf>> { |
| 927 | let ls = run_git( |
| 928 | &self.git_dir, |
| 929 | &self.work_tree, |
| 930 | &["ls-tree", "-r", "-z", "--name-only", treeish], |
| 931 | )?; |
| 932 | if !ls.status.success() { |
| 933 | return Err(io_other(format!( |
| 934 | "git ls-tree failed: {}", |
| 935 | String::from_utf8_lossy(&ls.stderr).trim() |
| 936 | ))); |
| 937 | } |
| 938 | Ok(parse_nul_paths(&ls.stdout)) |
| 939 | } |
| 940 | |
| 941 | fn remove_paths_missing_from_target( |
| 942 | &self, |
| 943 | current_paths: &HashSet<PathBuf>, |
| 944 | target_paths: &HashSet<PathBuf>, |
| 945 | ) -> io::Result<()> { |
| 946 | for rel in current_paths.difference(target_paths) { |
| 947 | if !is_safe_relative_path(rel) { |
| 948 | continue; |
| 949 | } |
| 950 | let path = self.work_tree.join(rel); |
| 951 | let Ok(metadata) = std::fs::symlink_metadata(&path) else { |
| 952 | continue; |
| 953 | }; |
| 954 | if metadata.file_type().is_dir() { |
| 955 | let _ = std::fs::remove_dir(&path); |
| 956 | } else { |
| 957 | std::fs::remove_file(&path)?; |
| 958 | } |
| 959 | self.prune_empty_parent_dirs(path.parent()); |
| 960 | } |
| 961 | Ok(()) |
| 962 | } |
| 963 | |
| 964 | fn prune_empty_parent_dirs(&self, mut dir: Option<&Path>) { |
| 965 | while let Some(path) = dir { |
| 966 | if path == self.work_tree { |
| 967 | break; |
| 968 | } |
| 969 | if std::fs::remove_dir(path).is_err() { |
| 970 | break; |
| 971 | } |
| 972 | dir = path.parent(); |
| 973 | } |
| 974 | } |
| 975 | |
| 976 | /// List up to `limit` most-recent snapshots, newest first. |
| 977 | pub fn list(&self, limit: usize) -> io::Result<Vec<Snapshot>> { |
| 978 | // `git log -<n>` is the short form of `--max-count=<n>`; if `limit` |
| 979 | // is `usize::MAX` (caller asked for "everything") we pass an empty |
| 980 | // count so git defaults to no upper bound. |
| 981 | let mut args: Vec<String> = vec!["log".to_string()]; |
| 982 | if limit < usize::MAX { |
| 983 | args.push(format!("--max-count={limit}")); |
| 984 | } |
| 985 | args.push("--pretty=format:%H%x09%at%x09%s".to_string()); |
| 986 | args.push("--no-color".to_string()); |
| 987 | let arg_refs: Vec<&str> = args.iter().map(String::as_str).collect(); |
| 988 | let log = run_git(&self.git_dir, &self.work_tree, &arg_refs)?; |
| 989 | if !log.status.success() { |
| 990 | let head = run_git( |
| 991 | &self.git_dir, |
| 992 | &self.work_tree, |
| 993 | &["symbolic-ref", "-q", "HEAD"], |
| 994 | )?; |
| 995 | if head.status.success() { |
| 996 | let reference = String::from_utf8_lossy(&head.stdout); |
| 997 | let exists = run_git( |
| 998 | &self.git_dir, |
| 999 | &self.work_tree, |
| 1000 | &["show-ref", "--verify", "--quiet", reference.trim()], |
| 1001 | )?; |
| 1002 | if exists.status.code() == Some(1) { |
| 1003 | return Ok(Vec::new()); |
| 1004 | } |
| 1005 | } |
| 1006 | return Err(io_other(format!( |
| 1007 | "git log failed: {}", |
| 1008 | String::from_utf8_lossy(&log.stderr).trim() |
| 1009 | ))); |
| 1010 | } |
| 1011 | let stdout = String::from_utf8_lossy(&log.stdout); |
| 1012 | let mut out = Vec::new(); |
| 1013 | for line in stdout.lines() { |
| 1014 | let mut parts = line.splitn(3, '\t'); |
| 1015 | let sha = parts.next().unwrap_or("").to_string(); |
| 1016 | let ts = parts |
| 1017 | .next() |
| 1018 | .and_then(|s| s.parse::<i64>().ok()) |
| 1019 | .unwrap_or(0); |
| 1020 | let subject = parts.next().unwrap_or("").to_string(); |
| 1021 | if sha.is_empty() { |
| 1022 | continue; |
| 1023 | } |
| 1024 | let (session_id, label) = Self::decode_session_label(&subject); |
| 1025 | out.push(Snapshot { |
| 1026 | id: SnapshotId(sha), |
| 1027 | label, |
| 1028 | timestamp: ts, |
| 1029 | session_id, |
| 1030 | }); |
| 1031 | } |
| 1032 | Ok(out) |
| 1033 | } |
| 1034 | |
| 1035 | /// Drop snapshots older than `max_age`, returning the count removed. |
| 1036 | /// |
| 1037 | /// Strategy: identify keepable commits (younger than the cutoff), |
| 1038 | /// reset HEAD to the oldest survivor, then `git reflog expire` + |
| 1039 | /// `git gc --prune=now` to actually reclaim space. Cheap and avoids |
| 1040 | /// rewriting history when nothing has aged out. |
| 1041 | pub fn prune_older_than(&self, max_age: Duration) -> io::Result<usize> { |
| 1042 | let now = SystemTime::now() |
| 1043 | .duration_since(UNIX_EPOCH) |
| 1044 | .map_err(|e| io_other(format!("clock error: {e}")))? |
| 1045 | .as_secs() as i64; |
| 1046 | let cutoff = now - max_age.as_secs() as i64; |
| 1047 | |
| 1048 | let snapshots = self.list(usize::MAX)?; |
| 1049 | if snapshots.is_empty() { |
| 1050 | return Ok(0); |
| 1051 | } |
| 1052 | |
| 1053 | // Snapshots are newest-first. Find the index of the first one |
| 1054 | // at-or-older than the cutoff — every entry from that index |
| 1055 | // onward is a candidate for removal. We use `<=` so a 0-second |
| 1056 | // retention drops same-second commits (otherwise tests calling |
| 1057 | // `prune_older_than(Duration::ZERO)` immediately after creating |
| 1058 | // a snapshot would never prune anything). |
| 1059 | let cut_index = snapshots.iter().position(|s| s.timestamp <= cutoff); |
| 1060 | let Some(cut) = cut_index else { |
| 1061 | return Ok(0); |
| 1062 | }; |
| 1063 | let removed = snapshots.len() - cut; |
| 1064 | if removed == 0 { |
| 1065 | return Ok(0); |
| 1066 | } |
| 1067 | |
| 1068 | if cut == 0 { |
| 1069 | // Every snapshot is older than the cutoff — wipe the repo |
| 1070 | // entirely so the next snapshot starts a fresh history. |
| 1071 | // Removing `.git/refs/heads/*` is enough to orphan the old |
| 1072 | // commits, then gc reclaims them. |
| 1073 | let refs_dir = self.git_dir.join("refs").join("heads"); |
| 1074 | if refs_dir.exists() { |
| 1075 | for entry in std::fs::read_dir(&refs_dir)? { |
| 1076 | let path = entry?.path(); |
| 1077 | if path.is_file() { |
| 1078 | let _ = std::fs::remove_file(&path); |
| 1079 | } |
| 1080 | } |
| 1081 | } |
| 1082 | // Also drop HEAD's packed refs so `git log` returns nothing. |
| 1083 | let packed = self.git_dir.join("packed-refs"); |
| 1084 | if packed.exists() { |
| 1085 | let _ = std::fs::remove_file(&packed); |
| 1086 | } |
| 1087 | } else { |
| 1088 | // Keep the newest `cut` snapshots (indices [0..cut], newest-first) |
| 1089 | // and drop the older tail. This MUST rebuild the survivors as a |
| 1090 | // fresh orphan chain, not `update-ref HEAD <oldest survivor>`: |
| 1091 | // the snapshots are a parent-linked commit chain with the newest |
| 1092 | // at HEAD, so pointing HEAD at the oldest survivor orphaned every |
| 1093 | // NEWER snapshot (gc then destroyed them) while keeping the very |
| 1094 | // snapshots we meant to remove as its ancestors — the exact |
| 1095 | // inverse of the intent (2026-08-04 review, reproduced). |
| 1096 | self.rebuild_survivor_chain(&snapshots[..cut])?; |
| 1097 | } |
| 1098 | |
| 1099 | // Reclaim space. |
| 1100 | let _ = run_git( |
| 1101 | &self.git_dir, |
| 1102 | &self.work_tree, |
| 1103 | &["reflog", "expire", "--expire=now", "--all"], |
| 1104 | ); |
| 1105 | let _ = run_git( |
| 1106 | &self.git_dir, |
| 1107 | &self.work_tree, |
| 1108 | &["gc", "--prune=now", "--quiet"], |
| 1109 | ); |
| 1110 | |
| 1111 | Ok(removed) |
| 1112 | } |
| 1113 | |
| 1114 | /// Rebuild `survivors` (newest-first) as a fresh orphan commit chain and |
| 1115 | /// point HEAD at its tip, so every snapshot NOT in `survivors` becomes |
| 1116 | /// unreachable for gc to reclaim. Each survivor's tree, label, session |
| 1117 | /// id, and author/committer timestamp are preserved, so ages do not lie |
| 1118 | /// after a prune (finding: `prune_keep_last_n` previously reset them to |
| 1119 | /// "now"). Assumes `survivors` is non-empty. |
| 1120 | fn rebuild_survivor_chain(&self, survivors: &[Snapshot]) -> io::Result<()> { |
| 1121 | let mut prev_sha: Option<String> = None; |
| 1122 | for s in survivors.iter().rev() { |
| 1123 | let tree = run_git( |
| 1124 | &self.git_dir, |
| 1125 | &self.work_tree, |
| 1126 | &["rev-parse", &format!("{}^{{tree}}", s.id.as_str())], |
| 1127 | )?; |
| 1128 | if !tree.status.success() { |
| 1129 | return Err(io_other(format!( |
| 1130 | "rev-parse {}^{{tree}} failed: {}", |
| 1131 | s.id.as_str(), |
| 1132 | String::from_utf8_lossy(&tree.stderr).trim() |
| 1133 | ))); |
| 1134 | } |
| 1135 | let tree_hash = String::from_utf8_lossy(&tree.stdout).trim().to_string(); |
| 1136 | |
| 1137 | let mut args = vec![ |
| 1138 | "commit-tree".to_string(), |
| 1139 | "-m".to_string(), |
| 1140 | Self::encode_session_label(&s.label, s.session_id.as_deref()), |
| 1141 | tree_hash, |
| 1142 | ]; |
| 1143 | if let Some(ref p) = prev_sha { |
| 1144 | args.push("-p".to_string()); |
| 1145 | args.push(p.clone()); |
| 1146 | } |
| 1147 | let arg_refs: Vec<&str> = args.iter().map(String::as_str).collect(); |
| 1148 | let new_sha = self.commit_tree_preserving_date(&arg_refs, s.timestamp)?; |
| 1149 | prev_sha = Some(new_sha); |
| 1150 | } |
| 1151 | |
| 1152 | if let Some(final_sha) = prev_sha { |
| 1153 | let up = run_git( |
| 1154 | &self.git_dir, |
| 1155 | &self.work_tree, |
| 1156 | &["update-ref", "HEAD", &final_sha], |
| 1157 | )?; |
| 1158 | if !up.status.success() { |
| 1159 | return Err(io_other(format!( |
| 1160 | "update-ref HEAD failed: {}", |
| 1161 | String::from_utf8_lossy(&up.stderr).trim() |
| 1162 | ))); |
| 1163 | } |
| 1164 | } |
| 1165 | Ok(()) |
| 1166 | } |
| 1167 | |
| 1168 | /// Run a `commit-tree` invocation with the author/committer dates pinned |
| 1169 | /// to `timestamp` (Unix seconds), so a rebuilt survivor keeps its real |
| 1170 | /// age instead of stamping "now". |
| 1171 | fn commit_tree_preserving_date(&self, args: &[&str], timestamp: i64) -> io::Result<String> { |
| 1172 | let date = format!("{timestamp} +0000"); |
| 1173 | let out = crate::dependencies::Git::command() |
| 1174 | .ok_or_else(|| io::Error::new(io::ErrorKind::NotFound, "git not found on PATH"))? |
| 1175 | .arg("--git-dir") |
| 1176 | .arg(&self.git_dir) |
| 1177 | .arg("--work-tree") |
| 1178 | .arg(&self.work_tree) |
| 1179 | .env("GIT_AUTHOR_DATE", &date) |
| 1180 | .env("GIT_COMMITTER_DATE", &date) |
| 1181 | .args(args) |
| 1182 | .output()?; |
| 1183 | if !out.status.success() { |
| 1184 | return Err(io_other(format!( |
| 1185 | "commit-tree failed: {}", |
| 1186 | String::from_utf8_lossy(&out.stderr).trim() |
| 1187 | ))); |
| 1188 | } |
| 1189 | Ok(String::from_utf8_lossy(&out.stdout).trim().to_string()) |
| 1190 | } |
| 1191 | |
| 1192 | /// Keep only the latest `max_count` snapshots, dropping older ones. |
| 1193 | /// |
| 1194 | /// Uses `commit-tree` with no `-p` to create a true orphan commit at |
| 1195 | /// the eldest survivor's tree, preserving its label. The old chain |
| 1196 | /// has zero refs after gc and is physically reclaimed. |
| 1197 | /// Keep only the latest `max_count` snapshots by rebuilding the |
| 1198 | /// survivor chain as orphan commits. Each survivor's tree and label |
| 1199 | /// are preserved — only the parent chain to older snapshots is cut. |
| 1200 | /// Old objects become unreachable and gc reclaims them. |
| 1201 | pub fn prune_keep_last_n(&self, max_count: usize) -> io::Result<usize> { |
| 1202 | let snapshots = self.list(usize::MAX)?; |
| 1203 | if snapshots.len() <= max_count { |
| 1204 | return Ok(0); |
| 1205 | } |
| 1206 | let keep = max_count; |
| 1207 | let removed = snapshots.len() - keep; |
| 1208 | // snapshots are newest-first: [0..keep] are the survivors. Rebuild |
| 1209 | // them as an orphan chain so the older tail is reclaimed. |
| 1210 | self.rebuild_survivor_chain(&snapshots[..keep])?; |
| 1211 | let _ = run_git( |
| 1212 | &self.git_dir, |
| 1213 | &self.work_tree, |
| 1214 | &["reflog", "expire", "--expire=now", "--all"], |
| 1215 | ); |
| 1216 | let _ = run_git( |
| 1217 | &self.git_dir, |
| 1218 | &self.work_tree, |
| 1219 | &["gc", "--prune=now", "--quiet"], |
| 1220 | ); |
| 1221 | Ok(removed) |
| 1222 | } |
| 1223 | |
| 1224 | /// Drop unreachable loose objects left behind by interrupted or |
| 1225 | /// orphaned side-repo operations. |
| 1226 | pub fn prune_unreachable_objects(&self) -> io::Result<()> { |
| 1227 | let prune = run_git(&self.git_dir, &self.work_tree, &["prune", "--expire=now"])?; |
| 1228 | if !prune.status.success() { |
| 1229 | return Err(io_other(format!( |
| 1230 | "git prune failed: {}", |
| 1231 | String::from_utf8_lossy(&prune.stderr).trim() |
| 1232 | ))); |
| 1233 | } |
| 1234 | Ok(()) |
| 1235 | } |
| 1236 | |
| 1237 | /// Return the side-repo's `.git` directory (for diagnostics). |
| 1238 | #[cfg_attr(not(test), expect(dead_code))] |
| 1239 | pub fn git_dir(&self) -> &Path { |
| 1240 | &self.git_dir |
| 1241 | } |
| 1242 | |
| 1243 | /// Return the work tree path (for diagnostics). |
| 1244 | #[cfg_attr(not(test), expect(dead_code))] |
| 1245 | pub fn work_tree(&self) -> &Path { |
| 1246 | &self.work_tree |
| 1247 | } |
| 1248 | } |
| 1249 | |
| 1250 | fn write_builtin_excludes(git_dir: &Path) -> io::Result<()> { |
| 1251 | let info_dir = git_dir.join("info"); |
| 1252 | std::fs::create_dir_all(&info_dir)?; |
| 1253 | std::fs::write(info_dir.join("exclude"), BUILTIN_EXCLUDES) |
| 1254 | } |
| 1255 | |
| 1256 | /// Recursively compute the total size of a directory in bytes. |
| 1257 | fn dir_size_bytes(root: &Path) -> io::Result<u64> { |
| 1258 | fn walk(dir: &Path, total: &mut u64) -> io::Result<()> { |
| 1259 | if !dir.is_dir() { |
| 1260 | return Ok(()); |
| 1261 | } |
| 1262 | for entry in std::fs::read_dir(dir)? { |
| 1263 | let entry = entry?; |
| 1264 | let path = entry.path(); |
| 1265 | let ft = entry.file_type()?; |
| 1266 | if ft.is_symlink() { |
| 1267 | continue; |
| 1268 | } |
| 1269 | if ft.is_dir() { |
| 1270 | walk(&path, total)?; |
| 1271 | } else if ft.is_file() { |
| 1272 | *total = total.saturating_add(entry.metadata().map(|m| m.len()).unwrap_or(0)); |
| 1273 | } |
| 1274 | } |
| 1275 | Ok(()) |
| 1276 | } |
| 1277 | let mut total: u64 = 0; |
| 1278 | walk(root, &mut total)?; |
| 1279 | Ok(total) |
| 1280 | } |
| 1281 | |
| 1282 | /// One prominent notice per workspace per process when the size-pressure |
| 1283 | /// prune destroys restore points — silent loss of undo history is the S5 |
| 1284 | /// failure mode (2026-08-04 snapshot hunt). The stderr print is deliberate: |
| 1285 | /// headless/CLI stderr is the user surface for once-per-workspace snapshot |
| 1286 | /// warnings, matching `maybe_notify_snapshots_disabled_once` in |
| 1287 | /// `core/turn.rs`. |
| 1288 | #[allow(clippy::print_stderr)] |
| 1289 | fn notify_snapshot_history_pruned_once(workspace: &Path, removed: usize) { |
| 1290 | use std::collections::HashSet; |
| 1291 | use std::sync::{Mutex, OnceLock}; |
| 1292 | static NOTIFIED: OnceLock<Mutex<HashSet<String>>> = OnceLock::new(); |
| 1293 | let key = workspace.to_string_lossy().into_owned(); |
| 1294 | let set = NOTIFIED.get_or_init(|| Mutex::new(HashSet::new())); |
| 1295 | let Ok(mut guard) = set.lock() else { |
| 1296 | return; |
| 1297 | }; |
| 1298 | if !guard.insert(key) { |
| 1299 | return; |
| 1300 | } |
| 1301 | drop(guard); |
| 1302 | eprint!("{}", snapshot_history_pruned_message(workspace, removed)); |
| 1303 | } |
| 1304 | |
| 1305 | /// Build the user-visible notice for a size-pressure prune. Kept pure and |
| 1306 | /// separate from the emit/dedup shell so the content is unit-testable. |
| 1307 | fn snapshot_history_pruned_message(workspace: &Path, removed: usize) -> String { |
| 1308 | format!( |
| 1309 | "warning: snapshot/undo history for {} was pruned to stay under the {} MB snapshot storage cap. |
| 1310 | {} snapshot(s) were removed and can no longer be restored. |
| 1311 | The cap bounds the undo side-repo's disk use; high-churn or large workspaces hit it sooner. |
| 1312 | ", |
| 1313 | workspace.display(), |
| 1314 | MAX_SNAPSHOT_SIZE_MB, |
| 1315 | removed |
| 1316 | ) |
| 1317 | } |
| 1318 | |
| 1319 | fn cleanup_stale_pack_temps(git_dir: &Path, stale_age: Duration) -> io::Result<usize> { |
| 1320 | let pack_dir = git_dir.join("objects").join("pack"); |
| 1321 | if !pack_dir.exists() { |
| 1322 | return Ok(0); |
| 1323 | } |
| 1324 | cleanup_stale_pack_temps_in(&pack_dir, stale_age, SystemTime::now()) |
| 1325 | } |
| 1326 | |
| 1327 | fn cleanup_stale_pack_temps_in( |
| 1328 | pack_dir: &Path, |
| 1329 | stale_age: Duration, |
| 1330 | now: SystemTime, |
| 1331 | ) -> io::Result<usize> { |
| 1332 | let mut removed = 0; |
| 1333 | for entry in std::fs::read_dir(pack_dir)? { |
| 1334 | let entry = entry?; |
| 1335 | let name = entry.file_name(); |
| 1336 | let Some(name) = name.to_str() else { |
| 1337 | continue; |
| 1338 | }; |
| 1339 | if !name.starts_with("tmp_pack_") { |
| 1340 | continue; |
| 1341 | } |
| 1342 | if !entry.file_type()?.is_file() { |
| 1343 | continue; |
| 1344 | } |
| 1345 | |
| 1346 | let metadata = entry.metadata()?; |
| 1347 | let Ok(modified) = metadata.modified() else { |
| 1348 | continue; |
| 1349 | }; |
| 1350 | let Ok(age) = now.duration_since(modified) else { |
| 1351 | continue; |
| 1352 | }; |
| 1353 | if age < stale_age { |
| 1354 | continue; |
| 1355 | } |
| 1356 | |
| 1357 | match std::fs::remove_file(entry.path()) { |
| 1358 | Ok(()) => removed += 1, |
| 1359 | Err(err) if err.kind() == io::ErrorKind::NotFound => {} |
| 1360 | Err(err) => return Err(err), |
| 1361 | } |
| 1362 | } |
| 1363 | Ok(removed) |
| 1364 | } |
| 1365 | |
| 1366 | fn run_git(git_dir: &Path, work_tree: &Path, args: &[&str]) -> io::Result<Output> { |
| 1367 | crate::dependencies::Git::command() |
| 1368 | .ok_or_else(|| io::Error::new(io::ErrorKind::NotFound, "git not found on PATH"))? |
| 1369 | .arg("--git-dir") |
| 1370 | .arg(git_dir) |
| 1371 | .arg("--work-tree") |
| 1372 | .arg(work_tree) |
| 1373 | .args(args) |
| 1374 | .output() |
| 1375 | } |
| 1376 | |
| 1377 | fn git_diff_matches(output: Output) -> io::Result<bool> { |
| 1378 | match output.status.code() { |
| 1379 | Some(0) => Ok(true), |
| 1380 | Some(1) => Ok(false), |
| 1381 | _ => Err(io_other(format!( |
| 1382 | "git diff failed: {}", |
| 1383 | String::from_utf8_lossy(&output.stderr).trim() |
| 1384 | ))), |
| 1385 | } |
| 1386 | } |
| 1387 | |
| 1388 | fn io_other(msg: impl Into<String>) -> io::Error { |
| 1389 | io::Error::other(msg.into()) |
| 1390 | } |
| 1391 | |
| 1392 | /// Walk `workspace` and accumulate file sizes, returning `Ok(total)` |
| 1393 | /// when the workspace fits under `cap_bytes` and `Err(gate)` naming the |
| 1394 | /// bound that tripped. Honors `.gitignore` — whether or not the |
| 1395 | /// workspace is itself a git repo, matching the `git add -A` that the |
| 1396 | /// snapshot commit actually runs against this work tree — and the |
| 1397 | /// snapshot-specific skip list above, so the measured size reflects |
| 1398 | /// what would land in a snapshot commit rather than the raw `du -sh` |
| 1399 | /// total. |
| 1400 | /// |
| 1401 | /// The walk is bounded by both `cap_bytes` and `max_entries`, and the |
| 1402 | /// two bounds are reported separately because they have different |
| 1403 | /// recoveries. A `cap_bytes` of `0` disables the byte cap entirely (so |
| 1404 | /// config can opt out) but not the entry bound. |
| 1405 | /// |
| 1406 | /// Production passes [`SIZE_WALK_MAX_ENTRIES`] for `max_entries`; it is |
| 1407 | /// a parameter only so the entry bound is reachable in a test without |
| 1408 | /// creating 200,000 inodes. It must not be threaded up through |
| 1409 | /// [`SnapshotRepo::open_or_init_with_cap`]: [`WorkspaceGate::describe`] |
| 1410 | /// interpolates the constant into the user-facing message, so a weaker |
| 1411 | /// injected bound would report a number that did not trip. |
| 1412 | pub fn estimate_workspace_size_bounded( |
| 1413 | workspace: &Path, |
| 1414 | cap_bytes: u64, |
| 1415 | max_entries: usize, |
| 1416 | ) -> Result<u64, WorkspaceGate> { |
| 1417 | use ignore::WalkBuilder; |
| 1418 | let mut total: u64 = 0; |
| 1419 | let mut entries: usize = 0; |
| 1420 | let skip: HashSet<&'static str> = SIZE_WALK_SKIP_DIRS.iter().copied().collect(); |
| 1421 | let walker = WalkBuilder::new(workspace) |
| 1422 | .hidden(false) |
| 1423 | // `ignore` defaults to `require_git(true)`, which silently disables |
| 1424 | // every gitignore rule when the workspace is not inside a git repo. |
| 1425 | // The snapshot's own `git add -A` honors `.gitignore` regardless, so |
| 1426 | // without this the estimator over-counts a non-git workspace and can |
| 1427 | // refuse it while offering a `.gitignore` remedy that cannot work. |
| 1428 | .require_git(false) |
| 1429 | .follow_links(false) |
| 1430 | .filter_entry(move |entry| { |
| 1431 | // Skip the well-known build-output directories at any depth. |
| 1432 | // The `ignore` crate calls `filter_entry` once per dir/file; |
| 1433 | // returning `false` here prunes the whole subtree. |
| 1434 | entry |
| 1435 | .file_name() |
| 1436 | .to_str() |
| 1437 | .is_none_or(|name| !skip.contains(name)) |
| 1438 | }) |
| 1439 | .build(); |
| 1440 | for entry in walker.flatten() { |
| 1441 | entries += 1; |
| 1442 | if entries > max_entries { |
| 1443 | return Err(WorkspaceGate::TooManyEntries); |
| 1444 | } |
| 1445 | if let Ok(meta) = entry.metadata() |
| 1446 | && meta.is_file() |
| 1447 | { |
| 1448 | total = total.saturating_add(meta.len()); |
| 1449 | if cap_bytes > 0 && total > cap_bytes { |
| 1450 | return Err(WorkspaceGate::TooLarge); |
| 1451 | } |
| 1452 | } |
| 1453 | } |
| 1454 | Ok(total) |
| 1455 | } |
| 1456 | |
| 1457 | fn unsafe_workspace_snapshot_reason(workspace: &Path, home: Option<&Path>) -> Option<&'static str> { |
| 1458 | let workspace = normalize_path_for_safety(workspace); |
| 1459 | if is_filesystem_root(&workspace) { |
| 1460 | return Some("filesystem root"); |
| 1461 | } |
| 1462 | |
| 1463 | if is_home_directory(&workspace, home) { |
| 1464 | return Some("home directory"); |
| 1465 | } |
| 1466 | |
| 1467 | let home = home.map(normalize_path_for_safety)?; |
| 1468 | if workspace.parent() == Some(home.as_path()) { |
| 1469 | let name = workspace.file_name().and_then(|name| name.to_str()); |
| 1470 | if matches!( |
| 1471 | name, |
| 1472 | Some( |
| 1473 | "Desktop" | "Documents" | "Downloads" | "Library" | "Movies" | "Music" | "Pictures" |
| 1474 | ) |
| 1475 | ) { |
| 1476 | return Some("home collection directory"); |
| 1477 | } |
| 1478 | } |
| 1479 | |
| 1480 | None |
| 1481 | } |
| 1482 | |
| 1483 | fn normalize_path_for_safety(path: &Path) -> PathBuf { |
| 1484 | path.canonicalize().unwrap_or_else(|_| path.to_path_buf()) |
| 1485 | } |
| 1486 | |
| 1487 | fn is_filesystem_root(path: &Path) -> bool { |
| 1488 | path.parent().is_none() |
| 1489 | } |
| 1490 | |
| 1491 | fn is_home_directory(work_tree: &Path, home: Option<&Path>) -> bool { |
| 1492 | let Some(home) = home else { |
| 1493 | return false; |
| 1494 | }; |
| 1495 | |
| 1496 | let home_canonical = home.canonicalize().unwrap_or_else(|_| home.to_path_buf()); |
| 1497 | work_tree == home_canonical |
| 1498 | } |
| 1499 | |
| 1500 | fn parse_nul_paths(bytes: &[u8]) -> HashSet<PathBuf> { |
| 1501 | bytes |
| 1502 | .split(|b| *b == 0) |
| 1503 | .filter(|chunk| !chunk.is_empty()) |
| 1504 | .map(|chunk| PathBuf::from(String::from_utf8_lossy(chunk).into_owned())) |
| 1505 | .collect() |
| 1506 | } |
| 1507 | |
| 1508 | fn is_safe_relative_path(path: &Path) -> bool { |
| 1509 | !path.as_os_str().is_empty() |
| 1510 | && path |
| 1511 | .components() |
| 1512 | .all(|component| matches!(component, Component::Normal(_))) |
| 1513 | } |
| 1514 | |
| 1515 | /// Normalize a caller-supplied path into a safe workspace-relative path. |
| 1516 | /// |
| 1517 | /// Accepts either a workspace-relative path or an absolute path inside |
| 1518 | /// `workspace`. Returns `None` when the result is not a plain relative path — |
| 1519 | /// absolute, empty, containing `..`, or pointing outside the workspace. Every |
| 1520 | /// file-scoped restore path passes through here, so a caller never gets to |
| 1521 | /// name a path the snapshot repo would resolve outside the work tree. |
| 1522 | /// |
| 1523 | /// The name is literal: leading or trailing spaces, brackets and glob |
| 1524 | /// characters are filename bytes, never trimmed and never patterns. Git is |
| 1525 | /// invoked with `--literal-pathspecs` for every file-scoped operation. |
| 1526 | pub fn workspace_relative_path(workspace: &Path, raw: &str) -> Option<PathBuf> { |
| 1527 | if raw.is_empty() { |
| 1528 | return None; |
| 1529 | } |
| 1530 | let candidate = Path::new(raw); |
| 1531 | let rel = if candidate.is_absolute() { |
| 1532 | candidate.strip_prefix(workspace).ok()?.to_path_buf() |
| 1533 | } else { |
| 1534 | candidate.to_path_buf() |
| 1535 | }; |
| 1536 | is_safe_relative_path(&rel).then_some(rel) |
| 1537 | } |
| 1538 | |
| 1539 | #[cfg(test)] |
| 1540 | mod tests { |
| 1541 | use super::*; |
| 1542 | use crate::test_support::lock_test_env; |
| 1543 | use std::fs::{File, FileTimes}; |
| 1544 | use tempfile::tempdir; |
| 1545 | |
| 1546 | /// Holds the home directory pinned to a tempdir for the lifetime of a test. Also |
| 1547 | /// owns the process-wide env-var mutex so tests across modules |
| 1548 | /// don't trample each other's home env vars. |
| 1549 | pub(super) struct ScopedHome { |
| 1550 | prev_vars: Vec<(&'static str, Option<std::ffi::OsString>)>, |
| 1551 | _guard: crate::test_support::TestEnvLock, |
| 1552 | } |
| 1553 | impl Drop for ScopedHome { |
| 1554 | fn drop(&mut self) { |
| 1555 | // SAFETY: process-wide lock still held. |
| 1556 | unsafe { |
| 1557 | for (key, prev) in self.prev_vars.drain(..) { |
| 1558 | match prev { |
| 1559 | Some(value) => std::env::set_var(key, value), |
| 1560 | None => std::env::remove_var(key), |
| 1561 | } |
| 1562 | } |
| 1563 | } |
| 1564 | } |
| 1565 | } |
| 1566 | pub(super) fn scoped_home(home: &Path) -> ScopedHome { |
| 1567 | let guard = lock_test_env(); |
| 1568 | let prev_vars = ["HOME", "USERPROFILE", "HOMEDRIVE", "HOMEPATH"] |
| 1569 | .into_iter() |
| 1570 | .map(|key| (key, std::env::var_os(key))) |
| 1571 | .collect(); |
| 1572 | // SAFETY: serialised by the global env lock. |
| 1573 | unsafe { |
| 1574 | std::env::set_var("HOME", home); |
| 1575 | std::env::set_var("USERPROFILE", home); |
| 1576 | std::env::remove_var("HOMEDRIVE"); |
| 1577 | std::env::remove_var("HOMEPATH"); |
| 1578 | } |
| 1579 | ScopedHome { |
| 1580 | prev_vars, |
| 1581 | _guard: guard, |
| 1582 | } |
| 1583 | } |
| 1584 | |
| 1585 | /// Build a side-repo whose snapshot dir lives under the same |
| 1586 | /// tempdir we're using for `HOME` — so the inner `crate::config::effective_home_dir()` |
| 1587 | /// lookup stays inside our sandbox. Returns the guard alongside so |
| 1588 | /// the caller can keep HOME pinned for the rest of the test. |
| 1589 | fn make_repo(tmp: &Path) -> (SnapshotRepo, ScopedHome) { |
| 1590 | let workspace = tmp.join("workspace"); |
| 1591 | std::fs::create_dir_all(&workspace).unwrap(); |
| 1592 | let guard = scoped_home(tmp); |
| 1593 | let repo = SnapshotRepo::open_or_init(&workspace).expect("open_or_init"); |
| 1594 | (repo, guard) |
| 1595 | } |
| 1596 | |
| 1597 | #[test] |
| 1598 | fn snapshot_creates_commit_in_side_repo_only() { |
| 1599 | let tmp = tempdir().unwrap(); |
| 1600 | let (repo, _home) = make_repo(tmp.path()); |
| 1601 | std::fs::write(repo.work_tree().join("a.txt"), b"alpha").unwrap(); |
| 1602 | |
| 1603 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1604 | assert_eq!(id.as_str().len(), 40); |
| 1605 | |
| 1606 | let list = repo.list(10).expect("list"); |
| 1607 | assert_eq!(list.len(), 1); |
| 1608 | assert_eq!(list[0].label, "pre-turn:1"); |
| 1609 | |
| 1610 | // The user's workspace must NOT have a real `.git` because we |
| 1611 | // never created one in their workspace — only in the side dir. |
| 1612 | assert!(!repo.work_tree().join(".git").exists()); |
| 1613 | } |
| 1614 | |
| 1615 | #[test] |
| 1616 | fn open_existing_is_read_only_and_does_not_initialize() { |
| 1617 | let tmp = tempdir().unwrap(); |
| 1618 | let workspace = tmp.path().join("workspace"); |
| 1619 | std::fs::create_dir_all(&workspace).unwrap(); |
| 1620 | let _home = scoped_home(tmp.path()); |
| 1621 | |
| 1622 | let before = SnapshotRepo::open_existing(&workspace).expect("open existing"); |
| 1623 | assert!(before.is_none()); |
| 1624 | assert!( |
| 1625 | !snapshot_git_dir(&workspace).exists(), |
| 1626 | "read-only open must not create the side repo" |
| 1627 | ); |
| 1628 | |
| 1629 | let repo = SnapshotRepo::open_or_init(&workspace).expect("open_or_init"); |
| 1630 | std::fs::write(repo.work_tree().join("a.txt"), b"alpha").unwrap(); |
| 1631 | repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1632 | |
| 1633 | let after = SnapshotRepo::open_existing(&workspace).expect("open existing"); |
| 1634 | assert!(after.is_some()); |
| 1635 | } |
| 1636 | |
| 1637 | #[test] |
| 1638 | fn restore_reverts_workspace_files() { |
| 1639 | let tmp = tempdir().unwrap(); |
| 1640 | let (repo, _home) = make_repo(tmp.path()); |
| 1641 | let f = repo.work_tree().join("file.txt"); |
| 1642 | |
| 1643 | std::fs::write(&f, b"original").unwrap(); |
| 1644 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1645 | |
| 1646 | std::fs::write(&f, b"clobbered").unwrap(); |
| 1647 | repo.snapshot("post-turn:1").expect("snapshot 2"); |
| 1648 | |
| 1649 | repo.restore(&id).expect("restore"); |
| 1650 | let after = std::fs::read_to_string(&f).unwrap(); |
| 1651 | assert_eq!(after, "original"); |
| 1652 | } |
| 1653 | |
| 1654 | #[test] |
| 1655 | fn restore_removes_files_added_after_target_snapshot() { |
| 1656 | let tmp = tempdir().unwrap(); |
| 1657 | let (repo, _home) = make_repo(tmp.path()); |
| 1658 | let original = repo.work_tree().join("original.txt"); |
| 1659 | let added = repo.work_tree().join("added.txt"); |
| 1660 | |
| 1661 | std::fs::write(&original, b"original").unwrap(); |
| 1662 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1663 | |
| 1664 | std::fs::write(&added, b"new file").unwrap(); |
| 1665 | repo.snapshot("post-turn:1").expect("snapshot 2"); |
| 1666 | |
| 1667 | repo.restore(&id).expect("restore"); |
| 1668 | assert!(original.exists()); |
| 1669 | assert!(!added.exists(), "restore must remove tracked added files"); |
| 1670 | } |
| 1671 | |
| 1672 | #[test] |
| 1673 | fn restore_paths_leaves_unrelated_files_alone() { |
| 1674 | let tmp = tempdir().unwrap(); |
| 1675 | let (repo, _home) = make_repo(tmp.path()); |
| 1676 | let wanted = repo.work_tree().join("wanted.txt"); |
| 1677 | let unrelated = repo.work_tree().join("unrelated.txt"); |
| 1678 | |
| 1679 | std::fs::write(&wanted, b"original").unwrap(); |
| 1680 | std::fs::write(&unrelated, b"original").unwrap(); |
| 1681 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1682 | |
| 1683 | std::fs::write(&wanted, b"clobbered").unwrap(); |
| 1684 | std::fs::write(&unrelated, b"also clobbered").unwrap(); |
| 1685 | repo.snapshot("post-turn:1").expect("snapshot 2"); |
| 1686 | |
| 1687 | let outcomes = repo |
| 1688 | .restore_paths(&id, &[PathBuf::from("wanted.txt")]) |
| 1689 | .expect("scoped restore"); |
| 1690 | |
| 1691 | assert_eq!(std::fs::read_to_string(&wanted).unwrap(), "original"); |
| 1692 | assert_eq!( |
| 1693 | std::fs::read_to_string(&unrelated).unwrap(), |
| 1694 | "also clobbered", |
| 1695 | "a file-scoped restore must not touch a path it was not given" |
| 1696 | ); |
| 1697 | assert_eq!(outcomes.len(), 1); |
| 1698 | assert_eq!(outcomes[0].path, PathBuf::from("wanted.txt")); |
| 1699 | assert_eq!(outcomes[0].action, PathRestoreAction::Modified); |
| 1700 | } |
| 1701 | |
| 1702 | #[test] |
| 1703 | fn only_restore_paths_is_safe_for_a_single_file_action() { |
| 1704 | // Characterizes the difference the per-file Revert control depends on. |
| 1705 | // `restore()` is what the TUI's `patch_undo()` and the runtime's |
| 1706 | // `patch-undo` endpoint both call. It is scoped in *snapshot selection* |
| 1707 | // (it picks a recent `tool:` snapshot) but not in *effect*: it checks |
| 1708 | // out the whole tree, so it also rolls back a working-tree path that no |
| 1709 | // tool touched. That is the data loss #2 removed the control over, and |
| 1710 | // it is why a per-file action cannot be built on top of it. |
| 1711 | let tmp = tempdir().unwrap(); |
| 1712 | let (repo, _home) = make_repo(tmp.path()); |
| 1713 | let touched = repo.work_tree().join("touched.txt"); |
| 1714 | let unrelated = repo.work_tree().join("unrelated.txt"); |
| 1715 | |
| 1716 | std::fs::write(&touched, b"v1").unwrap(); |
| 1717 | std::fs::write(&unrelated, b"snapshot-time").unwrap(); |
| 1718 | let id = repo.snapshot("tool:call-1").expect("snapshot"); |
| 1719 | |
| 1720 | // The tool edits one file; something else — the user, another editor — |
| 1721 | // changes the other one after the snapshot was taken. |
| 1722 | std::fs::write(&touched, b"v2").unwrap(); |
| 1723 | std::fs::write(&unrelated, b"user-work-in-progress").unwrap(); |
| 1724 | |
| 1725 | repo.restore(&id).expect("whole-tree restore"); |
| 1726 | assert_eq!(std::fs::read_to_string(&touched).unwrap(), "v1"); |
| 1727 | assert_eq!( |
| 1728 | std::fs::read_to_string(&unrelated).unwrap(), |
| 1729 | "snapshot-time", |
| 1730 | "whole-tree restore rolls back a file no tool touched" |
| 1731 | ); |
| 1732 | |
| 1733 | // Same situation again, but through the file-scoped path the |
| 1734 | // `file-revert` endpoint uses. |
| 1735 | std::fs::write(&touched, b"v1").unwrap(); |
| 1736 | std::fs::write(&unrelated, b"snapshot-time").unwrap(); |
| 1737 | let id2 = repo.snapshot("tool:call-2").expect("snapshot 2"); |
| 1738 | std::fs::write(&touched, b"v2").unwrap(); |
| 1739 | std::fs::write(&unrelated, b"user-work-in-progress").unwrap(); |
| 1740 | |
| 1741 | repo.restore_paths(&id2, &[PathBuf::from("touched.txt")]) |
| 1742 | .expect("scoped restore"); |
| 1743 | assert_eq!(std::fs::read_to_string(&touched).unwrap(), "v1"); |
| 1744 | assert_eq!( |
| 1745 | std::fs::read_to_string(&unrelated).unwrap(), |
| 1746 | "user-work-in-progress", |
| 1747 | "the scoped restore must leave the unrelated edit alone" |
| 1748 | ); |
| 1749 | } |
| 1750 | |
| 1751 | #[test] |
| 1752 | fn snapshot_diff_stat_describes_what_a_restore_would_change() { |
| 1753 | let tmp = tempdir().unwrap(); |
| 1754 | let (repo, _home) = make_repo(tmp.path()); |
| 1755 | let changed = repo.work_tree().join("changed.txt"); |
| 1756 | let untouched = repo.work_tree().join("untouched.txt"); |
| 1757 | |
| 1758 | std::fs::write(&changed, b"v1").unwrap(); |
| 1759 | std::fs::write(&untouched, b"stable").unwrap(); |
| 1760 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1761 | |
| 1762 | std::fs::write(&changed, b"v2").unwrap(); |
| 1763 | |
| 1764 | let stat = repo |
| 1765 | .snapshot_diff_stat(&id) |
| 1766 | .expect("diff stat") |
| 1767 | .expect("the snapshot differs, so something must be reported"); |
| 1768 | assert!(stat.contains("changed.txt"), "got: {stat}"); |
| 1769 | // The stat describes the restore's effect, not the workspace's whole |
| 1770 | // uncommitted state — a file the restore will not touch must not appear. |
| 1771 | assert!(!stat.contains("untouched.txt"), "got: {stat}"); |
| 1772 | |
| 1773 | // After restoring, the two sides agree: nothing left to report. (Which |
| 1774 | // is why the caller must capture this *before* the restore runs.) |
| 1775 | repo.restore(&id).expect("restore"); |
| 1776 | assert_eq!(repo.snapshot_diff_stat(&id).expect("diff stat"), None); |
| 1777 | } |
| 1778 | |
| 1779 | #[test] |
| 1780 | fn restore_paths_removes_a_file_created_after_the_snapshot() { |
| 1781 | let tmp = tempdir().unwrap(); |
| 1782 | let (repo, _home) = make_repo(tmp.path()); |
| 1783 | let kept = repo.work_tree().join("kept.txt"); |
| 1784 | let created = repo.work_tree().join("created.txt"); |
| 1785 | |
| 1786 | std::fs::write(&kept, b"kept").unwrap(); |
| 1787 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1788 | |
| 1789 | std::fs::write(&created, b"new file").unwrap(); |
| 1790 | repo.snapshot("post-turn:1").expect("snapshot 2"); |
| 1791 | |
| 1792 | let outcomes = repo |
| 1793 | .restore_paths(&id, &[PathBuf::from("created.txt")]) |
| 1794 | .expect("scoped restore"); |
| 1795 | |
| 1796 | assert!( |
| 1797 | !created.exists(), |
| 1798 | "a created file must be removed by revert" |
| 1799 | ); |
| 1800 | assert!(kept.exists(), "the untouched file must survive"); |
| 1801 | assert_eq!(outcomes[0].action, PathRestoreAction::Removed); |
| 1802 | } |
| 1803 | |
| 1804 | #[test] |
| 1805 | fn restore_paths_recreates_a_file_deleted_after_the_snapshot() { |
| 1806 | let tmp = tempdir().unwrap(); |
| 1807 | let (repo, _home) = make_repo(tmp.path()); |
| 1808 | let deleted = repo.work_tree().join("deleted.txt"); |
| 1809 | |
| 1810 | std::fs::write(&deleted, b"content").unwrap(); |
| 1811 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1812 | |
| 1813 | std::fs::remove_file(&deleted).unwrap(); |
| 1814 | repo.snapshot("post-turn:1").expect("snapshot 2"); |
| 1815 | |
| 1816 | let outcomes = repo |
| 1817 | .restore_paths(&id, &[PathBuf::from("deleted.txt")]) |
| 1818 | .expect("scoped restore"); |
| 1819 | |
| 1820 | assert_eq!(std::fs::read_to_string(&deleted).unwrap(), "content"); |
| 1821 | assert_eq!(outcomes[0].action, PathRestoreAction::Recreated); |
| 1822 | } |
| 1823 | |
| 1824 | #[test] |
| 1825 | fn restore_paths_rejects_parent_traversal() { |
| 1826 | let tmp = tempdir().unwrap(); |
| 1827 | let (repo, _home) = make_repo(tmp.path()); |
| 1828 | std::fs::write(repo.work_tree().join("a.txt"), b"a").unwrap(); |
| 1829 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1830 | |
| 1831 | let err = repo |
| 1832 | .restore_paths(&id, &[PathBuf::from("../escape.txt")]) |
| 1833 | .expect_err("traversal must be refused"); |
| 1834 | assert!(err.to_string().contains("unsafe path"), "got: {err}"); |
| 1835 | } |
| 1836 | |
| 1837 | #[test] |
| 1838 | fn path_differs_from_snapshot_is_scoped_to_the_named_path() { |
| 1839 | let tmp = tempdir().unwrap(); |
| 1840 | let (repo, _home) = make_repo(tmp.path()); |
| 1841 | let touched = repo.work_tree().join("touched.txt"); |
| 1842 | let untouched = repo.work_tree().join("untouched.txt"); |
| 1843 | |
| 1844 | std::fs::write(&touched, b"v1").unwrap(); |
| 1845 | std::fs::write(&untouched, b"v1").unwrap(); |
| 1846 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1847 | |
| 1848 | std::fs::write(&touched, b"v2").unwrap(); |
| 1849 | |
| 1850 | assert!( |
| 1851 | repo.path_differs_from_snapshot(&id, Path::new("touched.txt")) |
| 1852 | .expect("differs") |
| 1853 | ); |
| 1854 | assert!( |
| 1855 | !repo |
| 1856 | .path_differs_from_snapshot(&id, Path::new("untouched.txt")) |
| 1857 | .expect("differs") |
| 1858 | ); |
| 1859 | } |
| 1860 | |
| 1861 | #[test] |
| 1862 | fn workspace_relative_path_accepts_inside_paths_and_refuses_outside_ones() { |
| 1863 | // A real absolute temp path so the fixture is absolute on Windows too |
| 1864 | // (`/tmp/ws` is a relative path with a root-dir component there). |
| 1865 | let temp = std::env::temp_dir(); |
| 1866 | let workspace = temp.join("ws"); |
| 1867 | let other = temp.join("other"); |
| 1868 | |
| 1869 | assert_eq!( |
| 1870 | workspace_relative_path(&workspace, "src/lib.rs"), |
| 1871 | Some(PathBuf::from("src/lib.rs")) |
| 1872 | ); |
| 1873 | let inside = workspace.join("src").join("lib.rs"); |
| 1874 | assert_eq!( |
| 1875 | workspace_relative_path(&workspace, &inside.to_string_lossy()), |
| 1876 | Some(PathBuf::from("src").join("lib.rs")) |
| 1877 | ); |
| 1878 | let outside = other.join("lib.rs"); |
| 1879 | assert_eq!( |
| 1880 | workspace_relative_path(&workspace, &outside.to_string_lossy()), |
| 1881 | None |
| 1882 | ); |
| 1883 | assert_eq!(workspace_relative_path(&workspace, "../escape"), None); |
| 1884 | assert_eq!( |
| 1885 | workspace_relative_path(&workspace, "src/../../escape"), |
| 1886 | None |
| 1887 | ); |
| 1888 | assert_eq!(workspace_relative_path(&workspace, ""), None); |
| 1889 | // Whitespace and glob characters are literal filename bytes. |
| 1890 | assert_eq!( |
| 1891 | workspace_relative_path(&workspace, " padded.txt "), |
| 1892 | Some(PathBuf::from(" padded.txt ")) |
| 1893 | ); |
| 1894 | assert_eq!( |
| 1895 | workspace_relative_path(&workspace, "file[12].txt"), |
| 1896 | Some(PathBuf::from("file[12].txt")) |
| 1897 | ); |
| 1898 | } |
| 1899 | |
| 1900 | fn sha256_hash(path: &Path) -> String { |
| 1901 | format!( |
| 1902 | "sha256:{}", |
| 1903 | crate::hashing::sha256_hex(std::fs::read(path).unwrap()) |
| 1904 | ) |
| 1905 | } |
| 1906 | |
| 1907 | /// The Git primitive treats `[12]` as a pattern even after `--`; the |
| 1908 | /// file-scoped restore must not. `file[12].txt` and `file1.txt` both exist |
| 1909 | /// in the snapshot, so only literal pathspecs keep the second one intact. |
| 1910 | #[test] |
| 1911 | fn restore_file_if_unchanged_treats_glob_characters_literally() { |
| 1912 | let tmp = tempdir().unwrap(); |
| 1913 | let (repo, _home) = make_repo(tmp.path()); |
| 1914 | let literal = repo.work_tree().join("file[12].txt"); |
| 1915 | let sibling = repo.work_tree().join("file1.txt"); |
| 1916 | std::fs::write(&literal, b"literal-before").unwrap(); |
| 1917 | std::fs::write(&sibling, b"sibling-before").unwrap(); |
| 1918 | let id = repo.snapshot("tool:call-1").expect("snapshot"); |
| 1919 | std::fs::write(&literal, b"literal-after").unwrap(); |
| 1920 | std::fs::write(&sibling, b"sibling-after").unwrap(); |
| 1921 | |
| 1922 | assert!( |
| 1923 | repo.path_differs_from_snapshot(&id, Path::new("file[12].txt")) |
| 1924 | .unwrap() |
| 1925 | ); |
| 1926 | let outcomes = repo |
| 1927 | .restore_file_if_unchanged(&id, Path::new("file[12].txt"), &sha256_hash(&literal)) |
| 1928 | .expect("literal restore"); |
| 1929 | assert_eq!(outcomes.len(), 1); |
| 1930 | assert_eq!(outcomes[0].action, PathRestoreAction::Modified); |
| 1931 | assert_eq!(std::fs::read_to_string(&literal).unwrap(), "literal-before"); |
| 1932 | assert_eq!( |
| 1933 | std::fs::read_to_string(&sibling).unwrap(), |
| 1934 | "sibling-after", |
| 1935 | "a bracketed filename must never restore its glob siblings" |
| 1936 | ); |
| 1937 | } |
| 1938 | |
| 1939 | #[test] |
| 1940 | fn restore_file_if_unchanged_refuses_when_the_reviewed_bytes_changed() { |
| 1941 | let tmp = tempdir().unwrap(); |
| 1942 | let (repo, _home) = make_repo(tmp.path()); |
| 1943 | let file = repo.work_tree().join("a.txt"); |
| 1944 | std::fs::write(&file, b"v1").unwrap(); |
| 1945 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1946 | std::fs::write(&file, b"v2").unwrap(); |
| 1947 | let reviewed = sha256_hash(&file); |
| 1948 | // The user edits again after the client captured its change record. |
| 1949 | std::fs::write(&file, b"v3-user-edit").unwrap(); |
| 1950 | |
| 1951 | let err = repo |
| 1952 | .restore_file_if_unchanged(&id, Path::new("a.txt"), &reviewed) |
| 1953 | .expect_err("stale hash must refuse"); |
| 1954 | assert_eq!(err.kind(), io::ErrorKind::WouldBlock); |
| 1955 | assert_eq!(std::fs::read_to_string(&file).unwrap(), "v3-user-edit"); |
| 1956 | // `absent` is only valid for a file the client saw as deleted. |
| 1957 | let err = repo |
| 1958 | .restore_file_if_unchanged(&id, Path::new("a.txt"), "absent") |
| 1959 | .expect_err("absent must not match an existing file"); |
| 1960 | assert_eq!(err.kind(), io::ErrorKind::WouldBlock); |
| 1961 | // The exact current bytes restore. |
| 1962 | let outcomes = repo |
| 1963 | .restore_file_if_unchanged(&id, Path::new("a.txt"), &sha256_hash(&file)) |
| 1964 | .expect("current hash restores"); |
| 1965 | assert_eq!(outcomes[0].action, PathRestoreAction::Modified); |
| 1966 | assert_eq!(std::fs::read_to_string(&file).unwrap(), "v1"); |
| 1967 | } |
| 1968 | |
| 1969 | #[test] |
| 1970 | fn restore_file_if_unchanged_handles_deleted_and_created_files() { |
| 1971 | let tmp = tempdir().unwrap(); |
| 1972 | let (repo, _home) = make_repo(tmp.path()); |
| 1973 | let deleted = repo.work_tree().join("deleted.txt"); |
| 1974 | std::fs::write(&deleted, b"content").unwrap(); |
| 1975 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 1976 | std::fs::remove_file(&deleted).unwrap(); |
| 1977 | let created = repo.work_tree().join("created.txt"); |
| 1978 | std::fs::write(&created, b"new").unwrap(); |
| 1979 | |
| 1980 | let outcomes = repo |
| 1981 | .restore_file_if_unchanged(&id, Path::new("deleted.txt"), "absent") |
| 1982 | .expect("recreate"); |
| 1983 | assert_eq!(outcomes[0].action, PathRestoreAction::Recreated); |
| 1984 | assert_eq!(std::fs::read_to_string(&deleted).unwrap(), "content"); |
| 1985 | |
| 1986 | let outcomes = repo |
| 1987 | .restore_file_if_unchanged(&id, Path::new("created.txt"), &sha256_hash(&created)) |
| 1988 | .expect("remove"); |
| 1989 | assert_eq!(outcomes[0].action, PathRestoreAction::Removed); |
| 1990 | assert!(!created.exists()); |
| 1991 | // A created file inside a new directory is removed alone; the |
| 1992 | // directory the user made stays. |
| 1993 | let nested_dir = repo.work_tree().join("newdir"); |
| 1994 | std::fs::create_dir_all(&nested_dir).unwrap(); |
| 1995 | let nested = nested_dir.join("only.txt"); |
| 1996 | std::fs::write(&nested, b"n").unwrap(); |
| 1997 | let outcomes = repo |
| 1998 | .restore_file_if_unchanged(&id, Path::new("newdir/only.txt"), &sha256_hash(&nested)) |
| 1999 | .expect("remove nested"); |
| 2000 | assert_eq!(outcomes[0].action, PathRestoreAction::Removed); |
| 2001 | assert!(!nested.exists()); |
| 2002 | assert!(nested_dir.is_dir(), "the parent directory is not pruned"); |
| 2003 | // A path missing on both sides is not a change and reports nothing. |
| 2004 | assert!( |
| 2005 | !repo |
| 2006 | .path_differs_from_snapshot(&id, Path::new("never.txt")) |
| 2007 | .unwrap() |
| 2008 | ); |
| 2009 | } |
| 2010 | |
| 2011 | #[test] |
| 2012 | fn restore_file_if_unchanged_refuses_directories_git_metadata_and_ignored_files() { |
| 2013 | let tmp = tempdir().unwrap(); |
| 2014 | let (repo, _home) = make_repo(tmp.path()); |
| 2015 | let dir = repo.work_tree().join("src"); |
| 2016 | std::fs::create_dir_all(&dir).unwrap(); |
| 2017 | std::fs::write(dir.join("lib.rs"), b"fn a() {}").unwrap(); |
| 2018 | std::fs::write( |
| 2019 | repo.work_tree().join(".gitignore"), |
| 2020 | "ignored.txt |
| 2021 | ", |
| 2022 | ) |
| 2023 | .unwrap(); |
| 2024 | std::fs::write(repo.work_tree().join("ignored.txt"), b"secret").unwrap(); |
| 2025 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2026 | std::fs::write(dir.join("lib.rs"), b"fn b() {}").unwrap(); |
| 2027 | |
| 2028 | for rel in ["src", ".git/config", "src/.GIT/x", ".git"] { |
| 2029 | let err = repo.validate_restore_file(Path::new(rel)).expect_err(rel); |
| 2030 | assert_eq!(err.kind(), io::ErrorKind::InvalidInput, "{rel}"); |
| 2031 | } |
| 2032 | let err = repo |
| 2033 | .restore_file_if_unchanged(&id, Path::new("src"), "absent") |
| 2034 | .expect_err("directories are refused"); |
| 2035 | assert_eq!(err.kind(), io::ErrorKind::InvalidInput); |
| 2036 | assert_eq!( |
| 2037 | std::fs::read_to_string(dir.join("lib.rs")).unwrap(), |
| 2038 | "fn b() {}" |
| 2039 | ); |
| 2040 | |
| 2041 | // A gitignored file is excluded from the safety backup, so removing |
| 2042 | // it would be unrecoverable: refuse and leave it in place. |
| 2043 | let ignored = repo.work_tree().join("ignored.txt"); |
| 2044 | let err = repo |
| 2045 | .restore_file_if_unchanged(&id, Path::new("ignored.txt"), &sha256_hash(&ignored)) |
| 2046 | .expect_err("ignored files are refused"); |
| 2047 | assert!(err.to_string().contains("safety snapshot"), "got: {err}"); |
| 2048 | assert_eq!(std::fs::read_to_string(&ignored).unwrap(), "secret"); |
| 2049 | } |
| 2050 | |
| 2051 | #[cfg(unix)] |
| 2052 | #[test] |
| 2053 | fn restore_file_if_unchanged_refuses_symlinks_anywhere_in_the_path() { |
| 2054 | let tmp = tempdir().unwrap(); |
| 2055 | let (repo, _home) = make_repo(tmp.path()); |
| 2056 | let outside = tmp.path().join("outside"); |
| 2057 | std::fs::create_dir_all(&outside).unwrap(); |
| 2058 | std::fs::write(outside.join("target.txt"), b"outside").unwrap(); |
| 2059 | std::fs::write(repo.work_tree().join("real.txt"), b"real").unwrap(); |
| 2060 | std::os::unix::fs::symlink(&outside, repo.work_tree().join("linkdir")).unwrap(); |
| 2061 | std::os::unix::fs::symlink( |
| 2062 | outside.join("target.txt"), |
| 2063 | repo.work_tree().join("link.txt"), |
| 2064 | ) |
| 2065 | .unwrap(); |
| 2066 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2067 | |
| 2068 | for rel in ["link.txt", "linkdir/target.txt"] { |
| 2069 | let err = repo |
| 2070 | .restore_file_if_unchanged(&id, Path::new(rel), "absent") |
| 2071 | .expect_err(rel); |
| 2072 | assert_eq!(err.kind(), io::ErrorKind::InvalidInput, "{rel}"); |
| 2073 | } |
| 2074 | assert_eq!( |
| 2075 | std::fs::read_to_string(outside.join("target.txt")).unwrap(), |
| 2076 | "outside" |
| 2077 | ); |
| 2078 | // The snapshot side is checked too: a symlink entry in the tree is |
| 2079 | // not a regular file even when the work tree copy is gone. |
| 2080 | std::fs::remove_file(repo.work_tree().join("link.txt")).unwrap(); |
| 2081 | let err = repo |
| 2082 | .restore_file_if_unchanged(&id, Path::new("link.txt"), "absent") |
| 2083 | .expect_err("snapshot symlink entry"); |
| 2084 | assert_eq!(err.kind(), io::ErrorKind::InvalidInput); |
| 2085 | assert!(!repo.work_tree().join("link.txt").exists()); |
| 2086 | } |
| 2087 | |
| 2088 | #[test] |
| 2089 | fn list_distinguishes_an_unborn_head_from_broken_history() { |
| 2090 | let tmp = tempdir().unwrap(); |
| 2091 | let (repo, _home) = make_repo(tmp.path()); |
| 2092 | assert!(repo.list(10).expect("unborn HEAD lists nothing").is_empty()); |
| 2093 | |
| 2094 | std::fs::write(repo.work_tree().join("a.txt"), b"a").unwrap(); |
| 2095 | repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2096 | assert_eq!(repo.list(10).unwrap().len(), 1); |
| 2097 | |
| 2098 | // Point the branch at an object that does not exist: the history is |
| 2099 | // now broken, which must surface as an error rather than "no |
| 2100 | // snapshots" (an empty list would let patch-undo drop a turn). |
| 2101 | let head = String::from_utf8( |
| 2102 | run_git(repo.git_dir(), repo.work_tree(), &["symbolic-ref", "HEAD"]) |
| 2103 | .unwrap() |
| 2104 | .stdout, |
| 2105 | ) |
| 2106 | .unwrap(); |
| 2107 | std::fs::write( |
| 2108 | repo.git_dir().join(head.trim()), |
| 2109 | "0123456789abcdef0123456789abcdef01234567\n", |
| 2110 | ) |
| 2111 | .unwrap(); |
| 2112 | let err = repo.list(10).expect_err("broken history must error"); |
| 2113 | assert!(err.to_string().contains("git log failed"), "got: {err}"); |
| 2114 | } |
| 2115 | |
| 2116 | #[test] |
| 2117 | fn restore_takes_a_pre_restore_safety_snapshot_that_round_trips() { |
| 2118 | let tmp = tempdir().unwrap(); |
| 2119 | let (repo, _home) = make_repo(tmp.path()); |
| 2120 | let f = repo.work_tree().join("file.txt"); |
| 2121 | |
| 2122 | std::fs::write(&f, b"v1").unwrap(); |
| 2123 | let id1 = repo.snapshot("pre-turn:1").expect("snapshot v1"); |
| 2124 | |
| 2125 | std::fs::write(&f, b"v2").unwrap(); |
| 2126 | repo.snapshot("post-turn:1").expect("snapshot v2"); |
| 2127 | |
| 2128 | repo.restore(&id1).expect("restore to v1"); |
| 2129 | assert_eq!(std::fs::read_to_string(&f).unwrap(), "v1"); |
| 2130 | |
| 2131 | // The restore must have captured the pre-restore state (v2) under a |
| 2132 | // `pre-restore:` label naming its target, so the destructive op is |
| 2133 | // itself reversible (2026-08-04 snapshot hunt). |
| 2134 | let snapshots = repo.list(usize::MAX).expect("list"); |
| 2135 | let safety = snapshots |
| 2136 | .iter() |
| 2137 | .find(|s| s.label.starts_with("pre-restore:")) |
| 2138 | .expect("a pre-restore safety snapshot must exist"); |
| 2139 | assert!( |
| 2140 | safety.label.ends_with(&id1.as_str()[..12]), |
| 2141 | "safety label should name the restore target: {}", |
| 2142 | safety.label |
| 2143 | ); |
| 2144 | |
| 2145 | repo.restore(&safety.id) |
| 2146 | .expect("restore the safety snapshot"); |
| 2147 | assert_eq!( |
| 2148 | std::fs::read_to_string(&f).unwrap(), |
| 2149 | "v2", |
| 2150 | "the safety snapshot must bring back the pre-restore state" |
| 2151 | ); |
| 2152 | } |
| 2153 | |
| 2154 | #[test] |
| 2155 | fn snapshot_and_restore_do_not_move_user_git_head() { |
| 2156 | let tmp = tempdir().unwrap(); |
| 2157 | let workspace = tmp.path().join("workspace"); |
| 2158 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2159 | crate::dependencies::Git::command() |
| 2160 | .expect("git not found") |
| 2161 | .arg("-C") |
| 2162 | .arg(&workspace) |
| 2163 | .arg("init") |
| 2164 | .arg("--quiet") |
| 2165 | .status() |
| 2166 | .unwrap(); |
| 2167 | std::fs::write(workspace.join("tracked.txt"), b"committed").unwrap(); |
| 2168 | crate::dependencies::Git::command() |
| 2169 | .expect("git not found") |
| 2170 | .arg("-C") |
| 2171 | .arg(&workspace) |
| 2172 | .arg("add") |
| 2173 | .arg("tracked.txt") |
| 2174 | .status() |
| 2175 | .unwrap(); |
| 2176 | crate::dependencies::Git::command() |
| 2177 | .expect("git not found") |
| 2178 | .arg("-C") |
| 2179 | .arg(&workspace) |
| 2180 | .arg("-c") |
| 2181 | .arg("user.name=user") |
| 2182 | .arg("-c") |
| 2183 | .arg("user.email=user@example.test") |
| 2184 | .arg("commit") |
| 2185 | .arg("--quiet") |
| 2186 | .arg("-m") |
| 2187 | .arg("init") |
| 2188 | .status() |
| 2189 | .unwrap(); |
| 2190 | let user_head_before = crate::dependencies::Git::command() |
| 2191 | .expect("git not found") |
| 2192 | .arg("-C") |
| 2193 | .arg(&workspace) |
| 2194 | .args(["rev-parse", "HEAD"]) |
| 2195 | .output() |
| 2196 | .unwrap() |
| 2197 | .stdout; |
| 2198 | |
| 2199 | let _home = scoped_home(tmp.path()); |
| 2200 | let repo = SnapshotRepo::open_or_init(&workspace).unwrap(); |
| 2201 | std::fs::write(workspace.join("tracked.txt"), b"dirty-before").unwrap(); |
| 2202 | let id = repo.snapshot("pre-turn:1").unwrap(); |
| 2203 | std::fs::write(workspace.join("tracked.txt"), b"dirty-after").unwrap(); |
| 2204 | repo.snapshot("post-turn:1").unwrap(); |
| 2205 | repo.restore(&id).unwrap(); |
| 2206 | |
| 2207 | let user_head_after = crate::dependencies::Git::command() |
| 2208 | .expect("git not found") |
| 2209 | .arg("-C") |
| 2210 | .arg(&workspace) |
| 2211 | .args(["rev-parse", "HEAD"]) |
| 2212 | .output() |
| 2213 | .unwrap() |
| 2214 | .stdout; |
| 2215 | assert_eq!(user_head_after, user_head_before); |
| 2216 | assert_eq!( |
| 2217 | std::fs::read_to_string(workspace.join("tracked.txt")).unwrap(), |
| 2218 | "dirty-before" |
| 2219 | ); |
| 2220 | } |
| 2221 | |
| 2222 | #[test] |
| 2223 | fn list_respects_limit() { |
| 2224 | let tmp = tempdir().unwrap(); |
| 2225 | let (repo, _home) = make_repo(tmp.path()); |
| 2226 | for i in 0..5 { |
| 2227 | std::fs::write(repo.work_tree().join("f.txt"), format!("v{i}")).unwrap(); |
| 2228 | repo.snapshot(&format!("turn:{i}")).unwrap(); |
| 2229 | } |
| 2230 | let three = repo.list(3).unwrap(); |
| 2231 | assert_eq!(three.len(), 3); |
| 2232 | // Newest first. |
| 2233 | assert_eq!(three[0].label, "turn:4"); |
| 2234 | } |
| 2235 | |
| 2236 | #[test] |
| 2237 | fn prune_drops_snapshots_older_than_threshold() { |
| 2238 | let tmp = tempdir().unwrap(); |
| 2239 | let (repo, _home) = make_repo(tmp.path()); |
| 2240 | std::fs::write(repo.work_tree().join("f.txt"), "v0").unwrap(); |
| 2241 | repo.snapshot("turn:0").unwrap(); |
| 2242 | |
| 2243 | // Wait one second so the snapshot's commit timestamp is strictly |
| 2244 | // in the past relative to the prune call's "now" — otherwise |
| 2245 | // same-second comparisons make the assertion flaky. |
| 2246 | std::thread::sleep(Duration::from_millis(1100)); |
| 2247 | |
| 2248 | let removed = repo.prune_older_than(Duration::from_secs(0)).unwrap(); |
| 2249 | assert!(removed >= 1, "expected at least 1 pruned, got {removed}"); |
| 2250 | |
| 2251 | // After pruning everything, the next snapshot should start a |
| 2252 | // fresh history. |
| 2253 | std::fs::write(repo.work_tree().join("f.txt"), "v1").unwrap(); |
| 2254 | repo.snapshot("turn:1").unwrap(); |
| 2255 | let list = repo.list(10).unwrap(); |
| 2256 | assert_eq!(list.len(), 1); |
| 2257 | assert_eq!(list[0].label, "turn:1"); |
| 2258 | } |
| 2259 | |
| 2260 | /// The 2026-08-04 regression: with a cut in the MIDDLE of history, |
| 2261 | /// `prune_older_than` used to `update-ref HEAD <oldest survivor>`, which |
| 2262 | /// orphaned (and gc destroyed) the NEWEST snapshots while keeping the |
| 2263 | /// old ones as ancestors — the inverse of the intent, firing on every |
| 2264 | /// boot. This pins the correct partial-cut behavior. |
| 2265 | #[test] |
| 2266 | fn prune_older_than_keeps_the_newest_and_drops_only_the_old_tail() { |
| 2267 | let tmp = tempdir().unwrap(); |
| 2268 | let (repo, _home) = make_repo(tmp.path()); |
| 2269 | |
| 2270 | // Two "old" snapshots, then a pause, then two "new" ones. |
| 2271 | for i in 0..2 { |
| 2272 | std::fs::write(repo.work_tree().join("f.txt"), format!("old{i}")).unwrap(); |
| 2273 | repo.snapshot(&format!("old:{i}")).unwrap(); |
| 2274 | std::thread::sleep(Duration::from_millis(1100)); |
| 2275 | } |
| 2276 | // A wide gap so git's whole-second commit timestamps land the cut |
| 2277 | // unambiguously between the old and new pairs. The margins are |
| 2278 | // deliberately generous: this test runs under full-suite parallelism |
| 2279 | // where a sleep can overrun, and the cut is wall-clock. At prune time |
| 2280 | // the newest pair is ~0-1.2s old against a 6s cutoff, and the old |
| 2281 | // pair is ~9s old — ~5s of slack in both directions. |
| 2282 | std::thread::sleep(Duration::from_secs(8)); |
| 2283 | for i in 0..2 { |
| 2284 | std::fs::write(repo.work_tree().join("f.txt"), format!("new{i}")).unwrap(); |
| 2285 | repo.snapshot(&format!("new:{i}")).unwrap(); |
| 2286 | if i == 0 { |
| 2287 | std::thread::sleep(Duration::from_millis(1100)); |
| 2288 | } |
| 2289 | } |
| 2290 | let before = repo.list(usize::MAX).unwrap(); |
| 2291 | assert_eq!(before.len(), 4); |
| 2292 | // Derive the cut from the timestamps actually recorded rather than a |
| 2293 | // fixed 6s. A fixed cut assumes `repo.snapshot()` is fast: `new:0` is |
| 2294 | // only ~1.2s plus one git subprocess older than prune time, so on a |
| 2295 | // loaded Windows runner that subprocess alone pushed it past 6s and |
| 2296 | // three snapshots were pruned instead of two. (The old fixture guard |
| 2297 | // could not catch it either — it checked `before[0]` and `before[2]`, |
| 2298 | // and `before[1]` is the entry that drifts.) |
| 2299 | let now = std::time::SystemTime::now() |
| 2300 | .duration_since(std::time::UNIX_EPOCH) |
| 2301 | .unwrap() |
| 2302 | .as_secs() as i64; |
| 2303 | // Newest-first: [new:1, new:0, old:1, old:0]. The cut must land |
| 2304 | // strictly between the pairs, so aim at the midpoint of the 8s gap — |
| 2305 | // that leaves ~4s of slack against clock drift and a slow runner in |
| 2306 | // both directions. |
| 2307 | let survivor = before[1].timestamp; |
| 2308 | let victim = before[2].timestamp; |
| 2309 | assert!( |
| 2310 | survivor - victim >= 8, |
| 2311 | "fixture needs an 8s gap between the pairs (survivor {survivor}, victim {victim})" |
| 2312 | ); |
| 2313 | let midpoint = victim + (survivor - victim) / 2; |
| 2314 | assert!( |
| 2315 | now > midpoint, |
| 2316 | "fixture cutoff is not before the current time" |
| 2317 | ); |
| 2318 | let max_age = Duration::from_secs((now - midpoint) as u64); |
| 2319 | |
| 2320 | // The two old snapshots drop, the two new ones survive. |
| 2321 | let removed = repo.prune_older_than(max_age).unwrap(); |
| 2322 | assert_eq!(removed, 2, "only the old tail should be removed"); |
| 2323 | |
| 2324 | let remaining = repo.list(usize::MAX).unwrap(); |
| 2325 | assert_eq!(remaining.len(), 2, "the two newest must survive"); |
| 2326 | assert_eq!( |
| 2327 | remaining[0].label, "new:1", |
| 2328 | "newest survives (was destroyed before)" |
| 2329 | ); |
| 2330 | assert_eq!(remaining[1].label, "new:0"); |
| 2331 | assert!( |
| 2332 | !remaining.iter().any(|s| s.label.starts_with("old:")), |
| 2333 | "old snapshots must be gone, not kept as ancestors: {:?}", |
| 2334 | remaining.iter().map(|s| &s.label).collect::<Vec<_>>() |
| 2335 | ); |
| 2336 | |
| 2337 | // The survivors' contents are intact and restorable. |
| 2338 | repo.restore(&remaining[0].id).unwrap(); |
| 2339 | assert_eq!( |
| 2340 | std::fs::read_to_string(repo.work_tree().join("f.txt")).unwrap(), |
| 2341 | "new1" |
| 2342 | ); |
| 2343 | } |
| 2344 | |
| 2345 | #[test] |
| 2346 | fn prune_keep_last_n_keeps_latest_and_gc_reclaims_rest() { |
| 2347 | let tmp = tempdir().unwrap(); |
| 2348 | let (repo, _home) = make_repo(tmp.path()); |
| 2349 | |
| 2350 | for i in 0..3 { |
| 2351 | std::fs::write(repo.work_tree().join("f.txt"), format!("v{i}")).unwrap(); |
| 2352 | repo.snapshot(&format!("turn:{i}")).unwrap(); |
| 2353 | std::thread::sleep(Duration::from_millis(1100)); |
| 2354 | } |
| 2355 | |
| 2356 | assert_eq!(repo.list(usize::MAX).unwrap().len(), 3); |
| 2357 | |
| 2358 | let removed = repo.prune_keep_last_n(1).unwrap(); |
| 2359 | assert_eq!(removed, 2); |
| 2360 | |
| 2361 | let remaining = repo.list(usize::MAX).unwrap(); |
| 2362 | assert_eq!(remaining.len(), 1); |
| 2363 | assert_eq!(remaining[0].label, "turn:2"); |
| 2364 | |
| 2365 | // New snapshot starts a clean chain (not appending to old). |
| 2366 | std::fs::write(repo.work_tree().join("f.txt"), "fresh").unwrap(); |
| 2367 | repo.snapshot("turn:new").unwrap(); |
| 2368 | assert_eq!(repo.list(usize::MAX).unwrap().len(), 2); |
| 2369 | } |
| 2370 | |
| 2371 | #[test] |
| 2372 | fn prune_keep_last_n_preserves_multiple_snapshots_in_order() { |
| 2373 | let tmp = tempdir().unwrap(); |
| 2374 | let (repo, _home) = make_repo(tmp.path()); |
| 2375 | |
| 2376 | for i in 0..4 { |
| 2377 | std::fs::write(repo.work_tree().join("f.txt"), format!("v{i}")).unwrap(); |
| 2378 | repo.snapshot(&format!("turn:{i}")).unwrap(); |
| 2379 | std::thread::sleep(Duration::from_millis(1100)); |
| 2380 | } |
| 2381 | |
| 2382 | assert_eq!(repo.list(usize::MAX).unwrap().len(), 4); |
| 2383 | |
| 2384 | let removed = repo.prune_keep_last_n(2).unwrap(); |
| 2385 | assert_eq!(removed, 2); |
| 2386 | |
| 2387 | let remaining = repo.list(usize::MAX).unwrap(); |
| 2388 | assert_eq!(remaining.len(), 2); |
| 2389 | // Should be newest-first: turn:3 (newest), turn:2 (second newest) |
| 2390 | assert_eq!(remaining[0].label, "turn:3"); |
| 2391 | assert_eq!(remaining[1].label, "turn:2"); |
| 2392 | |
| 2393 | // New snapshot continues the chain. |
| 2394 | std::fs::write(repo.work_tree().join("f.txt"), "fresh").unwrap(); |
| 2395 | repo.snapshot("turn:new").unwrap(); |
| 2396 | let after = repo.list(usize::MAX).unwrap(); |
| 2397 | assert_eq!(after.len(), 3); |
| 2398 | assert_eq!(after[0].label, "turn:new"); |
| 2399 | } |
| 2400 | |
| 2401 | #[test] |
| 2402 | fn open_or_init_removes_stale_tmp_pack_files_only() { |
| 2403 | let tmp = tempdir().unwrap(); |
| 2404 | let (repo, _home) = make_repo(tmp.path()); |
| 2405 | let workspace = repo.work_tree().to_path_buf(); |
| 2406 | let pack_dir = repo.git_dir().join("objects").join("pack"); |
| 2407 | std::fs::create_dir_all(&pack_dir).unwrap(); |
| 2408 | |
| 2409 | let stale = pack_dir.join("tmp_pack_stale"); |
| 2410 | let fresh = pack_dir.join("tmp_pack_fresh"); |
| 2411 | let ordinary_pack = pack_dir.join("pack-kept.pack"); |
| 2412 | std::fs::write(&stale, b"stale").unwrap(); |
| 2413 | std::fs::write(&fresh, b"fresh").unwrap(); |
| 2414 | std::fs::write(&ordinary_pack, b"pack").unwrap(); |
| 2415 | |
| 2416 | let old_time = SystemTime::now() - STALE_TMP_PACK_AGE - Duration::from_secs(60); |
| 2417 | { |
| 2418 | let file = File::options().write(true).open(&stale).unwrap(); |
| 2419 | file.set_times(FileTimes::new().set_modified(old_time)) |
| 2420 | .unwrap(); |
| 2421 | } |
| 2422 | |
| 2423 | SnapshotRepo::open_or_init(&workspace).unwrap(); |
| 2424 | |
| 2425 | assert!(!stale.exists(), "stale tmp_pack file should be removed"); |
| 2426 | assert!(fresh.exists(), "fresh tmp_pack file should be kept"); |
| 2427 | assert!(ordinary_pack.exists(), "non-temp pack file should be kept"); |
| 2428 | } |
| 2429 | |
| 2430 | #[test] |
| 2431 | fn snapshot_respects_workspace_gitignore() { |
| 2432 | let tmp = tempdir().unwrap(); |
| 2433 | let (repo, _home) = make_repo(tmp.path()); |
| 2434 | std::fs::write(repo.work_tree().join(".gitignore"), "ignored.txt\n").unwrap(); |
| 2435 | std::fs::write(repo.work_tree().join("ignored.txt"), b"secret").unwrap(); |
| 2436 | std::fs::write(repo.work_tree().join("kept.txt"), b"public").unwrap(); |
| 2437 | |
| 2438 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2439 | |
| 2440 | // `git ls-tree` against the snapshot's commit shouldn't list ignored.txt. |
| 2441 | let ls = run_git( |
| 2442 | repo.git_dir(), |
| 2443 | repo.work_tree(), |
| 2444 | &["ls-tree", "-r", "--name-only", id.as_str()], |
| 2445 | ) |
| 2446 | .expect("ls-tree"); |
| 2447 | let names = String::from_utf8_lossy(&ls.stdout); |
| 2448 | assert!(names.contains("kept.txt"), "kept.txt missing: {names}"); |
| 2449 | assert!( |
| 2450 | !names.contains("ignored.txt"), |
| 2451 | "ignored.txt should not be in snapshot: {names}", |
| 2452 | ); |
| 2453 | } |
| 2454 | |
| 2455 | #[test] |
| 2456 | fn unsafe_workspace_rejects_home_directory_workspace() { |
| 2457 | let tmp = tempdir().unwrap(); |
| 2458 | let home = tmp.path(); |
| 2459 | |
| 2460 | assert_eq!( |
| 2461 | unsafe_workspace_snapshot_reason(home, Some(home)), |
| 2462 | Some("home directory") |
| 2463 | ); |
| 2464 | } |
| 2465 | |
| 2466 | #[test] |
| 2467 | fn unsafe_workspace_rejects_home_collection_directories() { |
| 2468 | let tmp = tempdir().unwrap(); |
| 2469 | let home = tmp.path(); |
| 2470 | let desktop = tmp.path().join("Desktop"); |
| 2471 | std::fs::create_dir_all(&desktop).unwrap(); |
| 2472 | |
| 2473 | assert_eq!( |
| 2474 | unsafe_workspace_snapshot_reason(&desktop, Some(home)), |
| 2475 | Some("home collection directory") |
| 2476 | ); |
| 2477 | } |
| 2478 | |
| 2479 | #[test] |
| 2480 | fn unsafe_workspace_allows_project_directories_under_home() { |
| 2481 | let tmp = tempdir().unwrap(); |
| 2482 | let home = tmp.path(); |
| 2483 | let workspace = tmp.path().join("code").join("project"); |
| 2484 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2485 | |
| 2486 | assert_eq!( |
| 2487 | unsafe_workspace_snapshot_reason(&workspace, Some(home)), |
| 2488 | None |
| 2489 | ); |
| 2490 | } |
| 2491 | |
| 2492 | #[test] |
| 2493 | fn snapshot_respects_builtin_excludes() { |
| 2494 | let tmp = tempdir().unwrap(); |
| 2495 | let (repo, _home) = make_repo(tmp.path()); |
| 2496 | std::fs::create_dir_all(repo.work_tree().join("node_modules/pkg")).unwrap(); |
| 2497 | std::fs::create_dir_all(repo.work_tree().join(".next/cache")).unwrap(); |
| 2498 | std::fs::create_dir_all(repo.work_tree().join("src")).unwrap(); |
| 2499 | std::fs::write( |
| 2500 | repo.work_tree().join("node_modules/pkg/index.js"), |
| 2501 | b"generated", |
| 2502 | ) |
| 2503 | .unwrap(); |
| 2504 | std::fs::write(repo.work_tree().join(".next/cache/chunk.bin"), b"generated").unwrap(); |
| 2505 | std::fs::write(repo.work_tree().join("debug.wasm"), b"binary").unwrap(); |
| 2506 | std::fs::write(repo.work_tree().join("src/main.rs"), b"fn main() {}").unwrap(); |
| 2507 | |
| 2508 | let excludes = std::fs::read_to_string(repo.git_dir().join("info/exclude")).unwrap(); |
| 2509 | assert!(excludes.contains("node_modules/")); |
| 2510 | assert!(excludes.contains(".next/")); |
| 2511 | assert!(excludes.contains("*.wasm")); |
| 2512 | |
| 2513 | let id = repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2514 | let ls = run_git( |
| 2515 | repo.git_dir(), |
| 2516 | repo.work_tree(), |
| 2517 | &["ls-tree", "-r", "--name-only", id.as_str()], |
| 2518 | ) |
| 2519 | .expect("ls-tree"); |
| 2520 | let names = String::from_utf8_lossy(&ls.stdout); |
| 2521 | assert!( |
| 2522 | names.contains("src/main.rs"), |
| 2523 | "src/main.rs missing: {names}" |
| 2524 | ); |
| 2525 | assert!( |
| 2526 | !names.contains("node_modules"), |
| 2527 | "node_modules should not be in snapshot: {names}", |
| 2528 | ); |
| 2529 | assert!( |
| 2530 | !names.contains(".next"), |
| 2531 | ".next should not be in snapshot: {names}", |
| 2532 | ); |
| 2533 | assert!( |
| 2534 | !names.contains("debug.wasm"), |
| 2535 | "binary artifacts should not be in snapshot: {names}", |
| 2536 | ); |
| 2537 | } |
| 2538 | |
| 2539 | #[test] |
| 2540 | fn open_or_init_is_idempotent() { |
| 2541 | let tmp = tempdir().unwrap(); |
| 2542 | let (_r, _h) = make_repo(tmp.path()); |
| 2543 | // Second open should not panic and should reuse the existing |
| 2544 | // `.git`. We re-open via the public API rather than make_repo to |
| 2545 | // avoid double-acquiring HOME (the guard would deadlock). |
| 2546 | drop((_r, _h)); |
| 2547 | let (_r2, _h2) = make_repo(tmp.path()); |
| 2548 | } |
| 2549 | |
| 2550 | #[test] |
| 2551 | fn home_directory_guard_matches_canonical_paths() { |
| 2552 | let tmp = tempdir().unwrap(); |
| 2553 | let home = tmp.path(); |
| 2554 | let home_canonical = home.canonicalize().unwrap(); |
| 2555 | let workspace = home.join("workspace"); |
| 2556 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2557 | let workspace_canonical = workspace.canonicalize().unwrap(); |
| 2558 | |
| 2559 | assert!(is_home_directory(&home_canonical, Some(home))); |
| 2560 | assert!(!is_home_directory(&workspace_canonical, Some(home))); |
| 2561 | assert!(!is_home_directory(&home_canonical, None)); |
| 2562 | } |
| 2563 | |
| 2564 | #[test] |
| 2565 | fn dir_size_bytes_measures_directory_bytes() { |
| 2566 | let tmp = tempdir().unwrap(); |
| 2567 | let dir = tmp.path().join("sizedir"); |
| 2568 | std::fs::create_dir_all(dir.join("sub")).unwrap(); |
| 2569 | // 3 bytes per file. |
| 2570 | std::fs::write(dir.join("a.txt"), b"abc").unwrap(); |
| 2571 | std::fs::write(dir.join("sub/b.txt"), b"xyz").unwrap(); |
| 2572 | |
| 2573 | let size = dir_size_bytes(&dir).expect("dir_size_bytes"); |
| 2574 | assert_eq!(size, 6, "two 3-byte files should measure 6 bytes"); |
| 2575 | |
| 2576 | // Write 2 MB of data. |
| 2577 | let big = dir.join("big.bin"); |
| 2578 | std::fs::write(&big, vec![0u8; 2 * 1024 * 1024]).unwrap(); |
| 2579 | let size = dir_size_bytes(&dir).expect("dir_size_bytes after big write"); |
| 2580 | assert_eq!( |
| 2581 | size, |
| 2582 | 2 * 1024 * 1024 + 6, |
| 2583 | "expected 2 MB + 6 bytes after writing a 2 MB file" |
| 2584 | ); |
| 2585 | } |
| 2586 | |
| 2587 | /// Regression: snapshot size cap (#1112). When the snapshot dir grows, |
| 2588 | /// `snapshot()` must prune old snapshots to stay under the limit. |
| 2589 | /// This test uses the real size constants, which are 500/400 MB — |
| 2590 | /// we can't easily blow up a temp dir to 500 MB in a unit test. |
| 2591 | /// Instead we verify the guard logic doesn't panic or error on a |
| 2592 | /// small repo (well under the cap), and that `snapshot()` still works. |
| 2593 | #[test] |
| 2594 | fn snapshot_succeeds_when_under_size_cap() { |
| 2595 | let tmp = tempdir().unwrap(); |
| 2596 | let (repo, _home) = make_repo(tmp.path()); |
| 2597 | // The side repo is tiny — well under 500 MB. Snapshot should work. |
| 2598 | std::fs::write(repo.work_tree().join("f.txt"), b"hello").unwrap(); |
| 2599 | let id = repo.snapshot("pre-turn:1").expect("snapshot under cap"); |
| 2600 | assert_eq!(id.as_str().len(), 40); |
| 2601 | } |
| 2602 | |
| 2603 | #[test] |
| 2604 | fn prune_size_pressure_counts_and_removes_history_when_over_limit() { |
| 2605 | let tmp = tempdir().unwrap(); |
| 2606 | let (repo, _home) = make_repo(tmp.path()); |
| 2607 | for i in 0..3 { |
| 2608 | std::fs::write(repo.work_tree().join("f.txt"), format!("v{i}")).unwrap(); |
| 2609 | repo.snapshot(&format!("pre-turn:{i}")).expect("snapshot"); |
| 2610 | } |
| 2611 | assert_eq!(repo.list(usize::MAX).unwrap().len(), 3); |
| 2612 | // A zero byte limit makes any non-empty side repo "over limit", so the |
| 2613 | // prune must run and report exactly what it destroyed. This is the S5 |
| 2614 | // wipe path; the count is what the user-visible notice is built from. |
| 2615 | let removed = repo.prune_size_pressure(0, 0).expect("prune_size_pressure"); |
| 2616 | assert_eq!(removed, 3, "every snapshot must be reported as removed"); |
| 2617 | assert!( |
| 2618 | repo.list(usize::MAX).unwrap().is_empty(), |
| 2619 | "history should be empty after the forced wipe" |
| 2620 | ); |
| 2621 | } |
| 2622 | |
| 2623 | #[test] |
| 2624 | fn prune_size_pressure_is_a_noop_under_the_limit() { |
| 2625 | let tmp = tempdir().unwrap(); |
| 2626 | let (repo, _home) = make_repo(tmp.path()); |
| 2627 | std::fs::write(repo.work_tree().join("f.txt"), b"v0").unwrap(); |
| 2628 | repo.snapshot("pre-turn:0").expect("snapshot"); |
| 2629 | let removed = repo |
| 2630 | .prune_size_pressure(u64::MAX, u64::MAX) |
| 2631 | .expect("prune_size_pressure"); |
| 2632 | assert_eq!(removed, 0, "under the limit nothing may be removed"); |
| 2633 | assert_eq!(repo.list(usize::MAX).unwrap().len(), 1); |
| 2634 | } |
| 2635 | |
| 2636 | #[test] |
| 2637 | fn snapshot_history_pruned_message_names_workspace_count_and_cap() { |
| 2638 | let msg = snapshot_history_pruned_message(Path::new("/tmp/ws"), 7); |
| 2639 | assert!(msg.contains("/tmp/ws"), "message must name the workspace"); |
| 2640 | assert!(msg.contains("7"), "message must state the removed count"); |
| 2641 | assert!( |
| 2642 | msg.contains(&MAX_SNAPSHOT_SIZE_MB.to_string()), |
| 2643 | "message must state the storage cap" |
| 2644 | ); |
| 2645 | } |
| 2646 | |
| 2647 | #[test] |
| 2648 | fn estimate_workspace_size_bounded_returns_total_when_under_cap() { |
| 2649 | let tmp = tempdir().unwrap(); |
| 2650 | let workspace = tmp.path().join("workspace"); |
| 2651 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2652 | std::fs::write(workspace.join("a.txt"), vec![b'a'; 100]).unwrap(); |
| 2653 | std::fs::write(workspace.join("b.txt"), vec![b'b'; 50]).unwrap(); |
| 2654 | let total = estimate_workspace_size_bounded(&workspace, 10_000, SIZE_WALK_MAX_ENTRIES) |
| 2655 | .expect("under-cap walk must return a total"); |
| 2656 | assert!( |
| 2657 | total >= 150, |
| 2658 | "total ({total}) must include both files (≥150 bytes)" |
| 2659 | ); |
| 2660 | } |
| 2661 | |
| 2662 | #[test] |
| 2663 | fn estimate_workspace_size_bounded_reports_the_size_gate_when_over_cap() { |
| 2664 | let tmp = tempdir().unwrap(); |
| 2665 | let workspace = tmp.path().join("workspace"); |
| 2666 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2667 | // Two 1 KB files, cap at 1 KB — second file should trip the cap. |
| 2668 | std::fs::write(workspace.join("a.bin"), vec![b'a'; 1024]).unwrap(); |
| 2669 | std::fs::write(workspace.join("b.bin"), vec![b'b'; 1024]).unwrap(); |
| 2670 | assert_eq!( |
| 2671 | estimate_workspace_size_bounded(&workspace, 1024, SIZE_WALK_MAX_ENTRIES), |
| 2672 | Err(WorkspaceGate::TooLarge), |
| 2673 | "over-cap walk must name the size gate for early bailout" |
| 2674 | ); |
| 2675 | } |
| 2676 | |
| 2677 | #[test] |
| 2678 | fn oversize_gate_message_states_the_byte_cap_without_a_remedy() { |
| 2679 | // The remedy is localized by the notice surfaces; repeating it here is |
| 2680 | // what produced the doubled warning. |
| 2681 | let message = |
| 2682 | WorkspaceGate::TooLarge.describe(2 * 1024 * 1024 * 1024, Path::new("/tmp/ws")); |
| 2683 | assert!(message.starts_with(GATE_TOO_LARGE_MARKER)); |
| 2684 | assert!(message.contains("/tmp/ws")); |
| 2685 | assert!(!message.contains("max_workspace_gb")); |
| 2686 | assert_eq!(message.lines().count(), 1, "the gate message is one line"); |
| 2687 | } |
| 2688 | |
| 2689 | #[test] |
| 2690 | fn entry_gate_message_is_distinct_and_never_blames_the_size_cap() { |
| 2691 | let message = WorkspaceGate::TooManyEntries.describe(0, Path::new("/tmp/ws")); |
| 2692 | assert!(message.starts_with(GATE_TOO_MANY_ENTRIES_MARKER)); |
| 2693 | assert!( |
| 2694 | !message.contains(GATE_TOO_LARGE_MARKER), |
| 2695 | "the entry gate must not be reported as a size trip" |
| 2696 | ); |
| 2697 | assert!(message.contains(&SIZE_WALK_MAX_ENTRIES.to_string())); |
| 2698 | assert!(!message.contains("max_workspace_gb")); |
| 2699 | } |
| 2700 | |
| 2701 | #[test] |
| 2702 | fn estimate_workspace_size_bounded_skips_builtin_excluded_dirs() { |
| 2703 | let tmp = tempdir().unwrap(); |
| 2704 | let workspace = tmp.path().join("workspace"); |
| 2705 | std::fs::create_dir_all(workspace.join("node_modules")).unwrap(); |
| 2706 | std::fs::create_dir_all(workspace.join("target")).unwrap(); |
| 2707 | std::fs::create_dir_all(workspace.join("src")).unwrap(); |
| 2708 | // 2 MB of "build output" in excluded dirs — must not count toward |
| 2709 | // the cap. |
| 2710 | std::fs::write(workspace.join("node_modules/big.bin"), vec![0u8; 1_000_000]).unwrap(); |
| 2711 | std::fs::write(workspace.join("target/big.bin"), vec![0u8; 1_000_000]).unwrap(); |
| 2712 | std::fs::write(workspace.join("src/lib.rs"), b"// real source").unwrap(); |
| 2713 | let total = estimate_workspace_size_bounded(&workspace, 500_000, SIZE_WALK_MAX_ENTRIES) |
| 2714 | .expect("walk must succeed since real source is tiny"); |
| 2715 | assert!( |
| 2716 | total < 1_000, |
| 2717 | "total ({total}) must reflect only src/, not node_modules/ or target/" |
| 2718 | ); |
| 2719 | } |
| 2720 | |
| 2721 | #[test] |
| 2722 | fn estimate_workspace_size_bounded_cap_zero_disables_cap() { |
| 2723 | let tmp = tempdir().unwrap(); |
| 2724 | let workspace = tmp.path().join("workspace"); |
| 2725 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2726 | // 10 KB file — would trip a 1 KB cap, but cap=0 means no cap. |
| 2727 | std::fs::write(workspace.join("big.bin"), vec![0u8; 10 * 1024]).unwrap(); |
| 2728 | let total = estimate_workspace_size_bounded(&workspace, 0, SIZE_WALK_MAX_ENTRIES) |
| 2729 | .expect("cap=0 must always return a total"); |
| 2730 | assert!( |
| 2731 | total >= 10 * 1024, |
| 2732 | "total ({total}) must include the 10 KB file when cap is disabled" |
| 2733 | ); |
| 2734 | } |
| 2735 | |
| 2736 | /// The entry ceiling is the bound that no test could reach before |
| 2737 | /// `max_entries` became a parameter: 200,000 inodes per run is not a |
| 2738 | /// price a unit test should pay. A byte-cheap workspace must still be |
| 2739 | /// refused, and refused as the *entry* gate — reporting `TooLarge` here |
| 2740 | /// would offer `max_workspace_gb` as a remedy that cannot lift it. |
| 2741 | #[test] |
| 2742 | fn entry_ceiling_refuses_a_byte_cheap_workspace_with_too_many_entries() { |
| 2743 | let tmp = tempdir().unwrap(); |
| 2744 | let workspace = tmp.path().join("workspace"); |
| 2745 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2746 | for i in 0..10 { |
| 2747 | std::fs::write(workspace.join(format!("f{i}.txt")), b"x").unwrap(); |
| 2748 | } |
| 2749 | assert_eq!( |
| 2750 | estimate_workspace_size_bounded(&workspace, 10_000_000, 3), |
| 2751 | Err(WorkspaceGate::TooManyEntries), |
| 2752 | "ten tiny files under a 10 MB cap must trip the entry bound, not the byte cap" |
| 2753 | ); |
| 2754 | } |
| 2755 | |
| 2756 | /// The invariant documented on `WorkspaceGate::TooManyEntries` and on the |
| 2757 | /// estimator: `max_workspace_gb = 0` opts out of the byte cap only. A |
| 2758 | /// future "if `cap_bytes == 0`, skip the walk" shortcut would satisfy |
| 2759 | /// every other test here and silently delete the ceiling that exists to |
| 2760 | /// stop a multi-minute `git add -A`. |
| 2761 | #[test] |
| 2762 | fn cap_zero_does_not_lift_the_entry_ceiling() { |
| 2763 | let tmp = tempdir().unwrap(); |
| 2764 | let workspace = tmp.path().join("workspace"); |
| 2765 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2766 | for i in 0..10 { |
| 2767 | std::fs::write(workspace.join(format!("f{i}.txt")), b"x").unwrap(); |
| 2768 | } |
| 2769 | assert_eq!( |
| 2770 | estimate_workspace_size_bounded(&workspace, 0, 3), |
| 2771 | Err(WorkspaceGate::TooManyEntries), |
| 2772 | "cap_bytes = 0 disables the byte cap, never the entry ceiling" |
| 2773 | ); |
| 2774 | } |
| 2775 | |
| 2776 | /// `ignore` disables gitignore matching entirely when no ancestor holds a |
| 2777 | /// `.git` (`require_git` defaults to true), but the snapshot's own |
| 2778 | /// `git add -A --work-tree <workspace>` reads `.gitignore` either way. A |
| 2779 | /// non-git workspace was therefore measured on content that would never |
| 2780 | /// be staged — and then told to fix it by editing `.gitignore`. |
| 2781 | /// |
| 2782 | /// Deliberately creates no `.git`: the point is the non-repo case. |
| 2783 | #[test] |
| 2784 | fn gitignored_content_is_excluded_outside_a_git_repo() { |
| 2785 | let tmp = tempdir().unwrap(); |
| 2786 | let workspace = tmp.path().join("workspace"); |
| 2787 | std::fs::create_dir_all(workspace.join("src")).unwrap(); |
| 2788 | std::fs::write(workspace.join(".gitignore"), "big.bin\n").unwrap(); |
| 2789 | std::fs::write(workspace.join("big.bin"), vec![0u8; 1_000_000]).unwrap(); |
| 2790 | std::fs::write(workspace.join("src/lib.rs"), b"// real source").unwrap(); |
| 2791 | assert!( |
| 2792 | !workspace.join(".git").exists(), |
| 2793 | "this test is only meaningful outside a git repo" |
| 2794 | ); |
| 2795 | let total = estimate_workspace_size_bounded(&workspace, 500_000, SIZE_WALK_MAX_ENTRIES) |
| 2796 | .expect("the only large file is gitignored, so the walk must fit under the cap"); |
| 2797 | assert!( |
| 2798 | total < 1_000, |
| 2799 | "total ({total}) must exclude the gitignored 1 MB file" |
| 2800 | ); |
| 2801 | } |
| 2802 | |
| 2803 | #[test] |
| 2804 | fn open_or_init_with_cap_rejects_oversized_workspace() { |
| 2805 | let tmp = tempdir().unwrap(); |
| 2806 | let workspace = tmp.path().join("workspace"); |
| 2807 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2808 | let _home = scoped_home(tmp.path()); |
| 2809 | // Drop a 4 KB file under a 1 KB cap. |
| 2810 | std::fs::write(workspace.join("big.bin"), vec![0u8; 4096]).unwrap(); |
| 2811 | let outcome = SnapshotRepo::open_or_init_with_cap(&workspace, 1024); |
| 2812 | let err = match outcome { |
| 2813 | Ok(_) => panic!("oversized workspace must fail open_or_init_with_cap"), |
| 2814 | Err(e) => e, |
| 2815 | }; |
| 2816 | let msg = err.to_string(); |
| 2817 | assert!( |
| 2818 | msg.contains(GATE_TOO_LARGE_MARKER), |
| 2819 | "error must call out the size cap; got: {msg}" |
| 2820 | ); |
| 2821 | let named_owned = workspace.display().to_string(); |
| 2822 | let named = named_owned |
| 2823 | .strip_prefix(r"\\?\") |
| 2824 | .or_else(|| named_owned.strip_prefix("//?/")) |
| 2825 | .unwrap_or(named_owned.as_str()); |
| 2826 | assert!( |
| 2827 | msg.contains(named), |
| 2828 | "error must name the workspace it refused; got: {msg}" |
| 2829 | ); |
| 2830 | // The remedy belongs to the localized notice. Repeating it here is |
| 2831 | // what produced the doubled, three-line warning users saw. |
| 2832 | assert!( |
| 2833 | !msg.contains("max_workspace_gb"), |
| 2834 | "gate error must not carry its own remedy copy; got: {msg}" |
| 2835 | ); |
| 2836 | } |
| 2837 | |
| 2838 | #[test] |
| 2839 | fn open_or_init_with_cap_zero_disables_size_check() { |
| 2840 | let tmp = tempdir().unwrap(); |
| 2841 | let workspace = tmp.path().join("workspace"); |
| 2842 | std::fs::create_dir_all(&workspace).unwrap(); |
| 2843 | let _home = scoped_home(tmp.path()); |
| 2844 | // 4 KB file but cap=0 → should still succeed. |
| 2845 | std::fs::write(workspace.join("big.bin"), vec![0u8; 4096]).unwrap(); |
| 2846 | let repo = SnapshotRepo::open_or_init_with_cap(&workspace, 0) |
| 2847 | .expect("cap=0 must skip the size check"); |
| 2848 | let id = repo |
| 2849 | .snapshot("pre-turn:1") |
| 2850 | .expect("snapshot under disabled cap"); |
| 2851 | assert_eq!(id.as_str().len(), 40); |
| 2852 | } |
| 2853 | |
| 2854 | #[test] |
| 2855 | fn session_tagged_snapshot_round_trips_through_list() { |
| 2856 | let tmp = tempdir().unwrap(); |
| 2857 | let (repo, _home) = make_repo(tmp.path()); |
| 2858 | std::fs::write(repo.work_tree().join("a.txt"), b"x").unwrap(); |
| 2859 | |
| 2860 | repo.snapshot_with_session("pre-turn:1", Some("sess-42")) |
| 2861 | .expect("snapshot with session"); |
| 2862 | |
| 2863 | let list = repo.list(10).expect("list"); |
| 2864 | assert_eq!(list.len(), 1); |
| 2865 | // The visible label stays clean; the session id is decoded separately. |
| 2866 | assert_eq!(list[0].label, "pre-turn:1"); |
| 2867 | assert_eq!(list[0].session_id.as_deref(), Some("sess-42")); |
| 2868 | } |
| 2869 | |
| 2870 | #[test] |
| 2871 | fn untagged_snapshot_decodes_without_session() { |
| 2872 | let tmp = tempdir().unwrap(); |
| 2873 | let (repo, _home) = make_repo(tmp.path()); |
| 2874 | std::fs::write(repo.work_tree().join("a.txt"), b"x").unwrap(); |
| 2875 | |
| 2876 | repo.snapshot("pre-turn:1").expect("snapshot"); |
| 2877 | |
| 2878 | let list = repo.list(10).expect("list"); |
| 2879 | assert_eq!(list.len(), 1); |
| 2880 | assert_eq!(list[0].label, "pre-turn:1"); |
| 2881 | assert_eq!(list[0].session_id, None); |
| 2882 | } |
| 2883 | |
| 2884 | #[test] |
| 2885 | fn prune_keep_last_n_preserves_session_tags() { |
| 2886 | let tmp = tempdir().unwrap(); |
| 2887 | let (repo, _home) = make_repo(tmp.path()); |
| 2888 | let file = repo.work_tree().join("a.txt"); |
| 2889 | |
| 2890 | // More snapshots than DEFAULT_MAX_SNAPSHOTS (50) so the survivor |
| 2891 | // chain is rebuilt as orphan commits — the path that previously |
| 2892 | // dropped the [sid=...] label prefix and turned every surviving |
| 2893 | // snapshot into a "legacy" (untagged) one. |
| 2894 | for i in 0..55 { |
| 2895 | std::fs::write(&file, format!("v{i}")).unwrap(); |
| 2896 | repo.snapshot_with_session(&format!("pre-turn:{i}"), Some("sess-p")) |
| 2897 | .expect("tagged snapshot"); |
| 2898 | } |
| 2899 | |
| 2900 | let removed = repo.prune_keep_last_n(50).expect("prune"); |
| 2901 | assert!(removed > 0, "expected prune to drop older snapshots"); |
| 2902 | |
| 2903 | let list = repo.list(usize::MAX).expect("list"); |
| 2904 | assert_eq!(list.len(), 50); |
| 2905 | assert!( |
| 2906 | list.iter() |
| 2907 | .all(|s| s.session_id.as_deref() == Some("sess-p")), |
| 2908 | "prune must preserve [sid=...] prefixes; got untagged survivors" |
| 2909 | ); |
| 2910 | } |
| 2911 | |
| 2912 | #[test] |
| 2913 | fn tagged_and_untagged_snapshots_coexist_in_one_chain() { |
| 2914 | let tmp = tempdir().unwrap(); |
| 2915 | let (repo, _home) = make_repo(tmp.path()); |
| 2916 | std::fs::write(repo.work_tree().join("a.txt"), b"v1").unwrap(); |
| 2917 | |
| 2918 | // Legacy untagged snapshot, then a session-tagged one. |
| 2919 | repo.snapshot("pre-turn:1").expect("legacy snapshot"); |
| 2920 | std::fs::write(repo.work_tree().join("a.txt"), b"v2").unwrap(); |
| 2921 | repo.snapshot_with_session("pre-turn:1", Some("sess-a")) |
| 2922 | .expect("tagged snapshot"); |
| 2923 | |
| 2924 | let list = repo.list(10).expect("list"); |
| 2925 | assert_eq!(list.len(), 2); |
| 2926 | // Newest first. |
| 2927 | assert_eq!(list[0].session_id.as_deref(), Some("sess-a")); |
| 2928 | assert_eq!(list[1].session_id, None); |
| 2929 | assert_eq!(list[1].label, "pre-turn:1"); |
| 2930 | } |
| 2931 | } |
| 2932 |