| 1 | //! Transactional persistence, atomic writes, and secret redaction for the |
| 2 | //! v0.8.67 constitution-first setup lane (#3410). |
| 3 | //! |
| 4 | //! This is the safety layer under every setup step. A setup session may touch |
| 5 | //! several files (the setup-state sidecar, the user-global constitution, and — |
| 6 | //! through the existing comment-preserving `ConfigStore` — `config.toml`). The |
| 7 | //! contract this module guarantees: |
| 8 | //! |
| 9 | //! - **Preview writes nothing.** [`SetupTransaction::preview`] reports what |
| 10 | //! would change without touching the filesystem. |
| 11 | //! - **Cancel leaves files unchanged.** A staged transaction that is dropped |
| 12 | //! without [`SetupTransaction::commit`] never wrote anything. |
| 13 | //! - **Save is atomic.** Each file is written through a temp file + rename |
| 14 | //! ([`atomic_write`]); a multi-file commit either fully applies or fully |
| 15 | //! rolls back, so a partial failure never leaves a half-written file. |
| 16 | //! - **Secrets never leak.** [`redact_secrets`] masks secret-bearing values for |
| 17 | //! any report, log line, or diagnostic that might echo config text. |
| 18 | //! |
| 19 | //! This module deliberately owns only the write / rollback / secret contract. |
| 20 | //! Each setup step owns *which* fields it writes; see [`crate::setup_state`] and |
| 21 | //! [`crate::user_constitution`]. |
| 22 | |
| 23 | use std::fs; |
| 24 | use std::path::{Path, PathBuf}; |
| 25 | |
| 26 | use anyhow::{Context, Result}; |
| 27 | use serde::Serialize; |
| 28 | |
| 29 | #[cfg(unix)] |
| 30 | use std::os::unix::fs::PermissionsExt; |
| 31 | |
| 32 | /// Restrictive file mode for setup-owned files (owner read/write only). |
| 33 | #[cfg(unix)] |
| 34 | const SETUP_FILE_MODE: u32 = 0o600; |
| 35 | |
| 36 | /// Atomically write `bytes` to `path` via a sibling temp file + rename. |
| 37 | /// |
| 38 | /// The temp file is created in the same directory as `path` so the final |
| 39 | /// `rename` is atomic on the same filesystem. On Unix the file is created with |
| 40 | /// `0o600` so setup-owned state never lands world-readable. Parent directories |
| 41 | /// are created as needed. |
| 42 | pub fn atomic_write(path: &Path, bytes: &[u8]) -> Result<()> { |
| 43 | let parent = path.parent().filter(|p| !p.as_os_str().is_empty()); |
| 44 | if let Some(parent) = parent { |
| 45 | fs::create_dir_all(parent) |
| 46 | .with_context(|| format!("failed to create directory {}", parent.display()))?; |
| 47 | } |
| 48 | |
| 49 | let dir = parent.unwrap_or_else(|| Path::new(".")); |
| 50 | let mut tmp = tempfile::NamedTempFile::new_in(dir) |
| 51 | .with_context(|| format!("failed to create temp file in {}", dir.display()))?; |
| 52 | |
| 53 | use std::io::Write as _; |
| 54 | tmp.write_all(bytes) |
| 55 | .with_context(|| format!("failed to write temp file for {}", path.display()))?; |
| 56 | tmp.flush() |
| 57 | .with_context(|| format!("failed to flush temp file for {}", path.display()))?; |
| 58 | |
| 59 | #[cfg(unix)] |
| 60 | { |
| 61 | let perms = fs::Permissions::from_mode(SETUP_FILE_MODE); |
| 62 | tmp.as_file() |
| 63 | .set_permissions(perms) |
| 64 | .with_context(|| format!("failed to set permissions for {}", path.display()))?; |
| 65 | } |
| 66 | |
| 67 | tmp.persist(path) |
| 68 | .map_err(|e| e.error) |
| 69 | .with_context(|| format!("failed to persist {}", path.display()))?; |
| 70 | Ok(()) |
| 71 | } |
| 72 | |
| 73 | /// Atomically write `value` as pretty-printed JSON to `path`. |
| 74 | /// |
| 75 | /// A trailing newline is appended so the file is well-formed for line-oriented |
| 76 | /// tooling and diffs. |
| 77 | pub fn atomic_write_json<T: Serialize>(path: &Path, value: &T) -> Result<()> { |
| 78 | let mut body = serde_json::to_string_pretty(value) |
| 79 | .with_context(|| format!("failed to serialize JSON for {}", path.display()))?; |
| 80 | body.push('\n'); |
| 81 | atomic_write(path, body.as_bytes()) |
| 82 | } |
| 83 | |
| 84 | /// A staged multi-file write that either fully applies or fully rolls back. |
| 85 | /// |
| 86 | /// Stage every file the setup step intends to write, then call [`commit`]. If |
| 87 | /// any single write fails, every already-applied write in the transaction is |
| 88 | /// restored to its pre-commit contents (or removed if it did not previously |
| 89 | /// exist), and the original error is returned. A transaction that is dropped |
| 90 | /// without committing leaves the filesystem untouched. |
| 91 | /// |
| 92 | /// [`commit`]: SetupTransaction::commit |
| 93 | #[derive(Debug, Default)] |
| 94 | pub struct SetupTransaction { |
| 95 | writes: Vec<StagedWrite>, |
| 96 | } |
| 97 | |
| 98 | #[derive(Debug, Clone)] |
| 99 | struct StagedWrite { |
| 100 | path: PathBuf, |
| 101 | bytes: Vec<u8>, |
| 102 | } |
| 103 | |
| 104 | /// A snapshot of a file's pre-commit state, captured so [`SetupTransaction`] |
| 105 | /// can restore it during rollback. |
| 106 | struct Snapshot { |
| 107 | path: PathBuf, |
| 108 | /// Original bytes, or `None` if the file did not exist before commit. |
| 109 | original: Option<Vec<u8>>, |
| 110 | } |
| 111 | |
| 112 | impl SetupTransaction { |
| 113 | /// Create an empty transaction. |
| 114 | #[must_use] |
| 115 | pub fn new() -> Self { |
| 116 | Self::default() |
| 117 | } |
| 118 | |
| 119 | /// Stage `bytes` to be written to `path` on [`commit`](Self::commit). |
| 120 | /// |
| 121 | /// Staging touches nothing on disk. A later stage for the same path |
| 122 | /// replaces an earlier one, so a step can revise its intended output before |
| 123 | /// committing. |
| 124 | pub fn stage(&mut self, path: impl Into<PathBuf>, bytes: impl Into<Vec<u8>>) -> &mut Self { |
| 125 | let path = path.into(); |
| 126 | let bytes = bytes.into(); |
| 127 | if let Some(existing) = self.writes.iter_mut().find(|w| w.path == path) { |
| 128 | existing.bytes = bytes; |
| 129 | } else { |
| 130 | self.writes.push(StagedWrite { path, bytes }); |
| 131 | } |
| 132 | self |
| 133 | } |
| 134 | |
| 135 | /// Stage `value` serialized as pretty JSON (with trailing newline). |
| 136 | pub fn stage_json<T: Serialize>( |
| 137 | &mut self, |
| 138 | path: impl Into<PathBuf>, |
| 139 | value: &T, |
| 140 | ) -> Result<&mut Self> { |
| 141 | let path = path.into(); |
| 142 | let mut body = serde_json::to_string_pretty(value) |
| 143 | .with_context(|| format!("failed to serialize JSON for {}", path.display()))?; |
| 144 | body.push('\n'); |
| 145 | Ok(self.stage(path, body.into_bytes())) |
| 146 | } |
| 147 | |
| 148 | /// The paths that [`commit`](Self::commit) would write, in staging order. |
| 149 | /// Writes nothing — this is the preview surface. |
| 150 | #[must_use] |
| 151 | pub fn preview(&self) -> Vec<&Path> { |
| 152 | self.writes.iter().map(|w| w.path.as_path()).collect() |
| 153 | } |
| 154 | |
| 155 | /// True when nothing is staged. |
| 156 | #[must_use] |
| 157 | pub fn is_empty(&self) -> bool { |
| 158 | self.writes.is_empty() |
| 159 | } |
| 160 | |
| 161 | /// Apply every staged write atomically. |
| 162 | /// |
| 163 | /// On success all files are updated. On the first failure, every write that |
| 164 | /// already landed is rolled back to its captured pre-commit state and the |
| 165 | /// original error is returned (rollback failures are attached as context). |
| 166 | pub fn commit(self) -> Result<()> { |
| 167 | let mut snapshots: Vec<Snapshot> = Vec::with_capacity(self.writes.len()); |
| 168 | |
| 169 | for write in &self.writes { |
| 170 | // Capture the pre-commit state before mutating, so we can restore it. |
| 171 | let original = match fs::read(&write.path) { |
| 172 | Ok(bytes) => Some(bytes), |
| 173 | Err(e) if e.kind() == std::io::ErrorKind::NotFound => None, |
| 174 | Err(e) => { |
| 175 | rollback(&snapshots); |
| 176 | return Err(e).with_context(|| { |
| 177 | format!( |
| 178 | "failed to read existing {} before write; rolled back {} prior change(s)", |
| 179 | write.path.display(), |
| 180 | snapshots.len() |
| 181 | ) |
| 182 | }); |
| 183 | } |
| 184 | }; |
| 185 | |
| 186 | match atomic_write(&write.path, &write.bytes) { |
| 187 | Ok(()) => snapshots.push(Snapshot { |
| 188 | path: write.path.clone(), |
| 189 | original, |
| 190 | }), |
| 191 | Err(err) => { |
| 192 | // This write did not land (atomic_write is all-or-nothing), |
| 193 | // so roll back only the writes that came before it. |
| 194 | rollback(&snapshots); |
| 195 | return Err(err).with_context(|| { |
| 196 | format!( |
| 197 | "setup transaction failed writing {}; rolled back {} prior change(s)", |
| 198 | write.path.display(), |
| 199 | snapshots.len() |
| 200 | ) |
| 201 | }); |
| 202 | } |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | Ok(()) |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | /// Restore every snapshot to its captured pre-commit state. Best-effort: a |
| 211 | /// rollback error is logged but does not abort the remaining restores, because |
| 212 | /// leaving as many files as possible in their original state is the goal. |
| 213 | fn rollback(snapshots: &[Snapshot]) { |
| 214 | for snap in snapshots.iter().rev() { |
| 215 | let result = match &snap.original { |
| 216 | Some(bytes) => atomic_write(&snap.path, bytes), |
| 217 | None => match fs::remove_file(&snap.path) { |
| 218 | Ok(()) => Ok(()), |
| 219 | Err(e) if e.kind() == std::io::ErrorKind::NotFound => Ok(()), |
| 220 | Err(e) => Err(e.into()), |
| 221 | }, |
| 222 | }; |
| 223 | if let Err(e) = result { |
| 224 | tracing::error!( |
| 225 | target: "config::persistence", |
| 226 | "failed to roll back {} during setup transaction: {e:#}", |
| 227 | snap.path.display() |
| 228 | ); |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | |
| 233 | // FEAT-025 D4: the pure secret-redaction primitives moved to |
| 234 | // `codewhale-secrets::redact` so portable command helpers can share one |
| 235 | // implementation without depending on this crate. Re-exported here to keep |
| 236 | // the existing `codewhale_config::persistence::*` public API stable. |
| 237 | pub use codewhale_secrets::redact::{ |
| 238 | REDACTED, RedactionPolicy, redact_json_model_bound_secrets, redact_json_secrets, |
| 239 | redact_model_bound_secrets, redact_secrets, redact_secrets_with, |
| 240 | }; |
| 241 | |
| 242 | #[cfg(test)] |
| 243 | mod tests { |
| 244 | use super::*; |
| 245 | |
| 246 | fn read(path: &Path) -> String { |
| 247 | fs::read_to_string(path).unwrap() |
| 248 | } |
| 249 | |
| 250 | fn synthetic_secret_fixture() -> String { |
| 251 | ["abc123", "def456", "ghi"].concat() |
| 252 | } |
| 253 | |
| 254 | #[test] |
| 255 | fn atomic_write_creates_parent_dirs_and_content() { |
| 256 | let tmp = tempfile::tempdir().unwrap(); |
| 257 | let path = tmp.path().join("nested/dir/state.json"); |
| 258 | atomic_write(&path, b"hello").unwrap(); |
| 259 | assert_eq!(read(&path), "hello"); |
| 260 | } |
| 261 | |
| 262 | #[cfg(unix)] |
| 263 | #[test] |
| 264 | fn atomic_write_uses_owner_only_permissions() { |
| 265 | let tmp = tempfile::tempdir().unwrap(); |
| 266 | let path = tmp.path().join("state.json"); |
| 267 | atomic_write(&path, b"x").unwrap(); |
| 268 | let mode = fs::metadata(&path).unwrap().permissions().mode() & 0o777; |
| 269 | assert_eq!(mode, SETUP_FILE_MODE); |
| 270 | } |
| 271 | |
| 272 | #[test] |
| 273 | fn atomic_write_replaces_existing_atomically() { |
| 274 | let tmp = tempfile::tempdir().unwrap(); |
| 275 | let path = tmp.path().join("state.json"); |
| 276 | atomic_write(&path, b"old").unwrap(); |
| 277 | atomic_write(&path, b"new").unwrap(); |
| 278 | assert_eq!(read(&path), "new"); |
| 279 | // No stray temp files left behind. |
| 280 | let leftovers: Vec<_> = fs::read_dir(tmp.path()) |
| 281 | .unwrap() |
| 282 | .filter_map(Result::ok) |
| 283 | .filter(|e| e.file_name() != "state.json") |
| 284 | .collect(); |
| 285 | assert!(leftovers.is_empty(), "stray temp files: {leftovers:?}"); |
| 286 | } |
| 287 | |
| 288 | #[test] |
| 289 | fn transaction_preview_writes_nothing() { |
| 290 | let tmp = tempfile::tempdir().unwrap(); |
| 291 | let a = tmp.path().join("a.json"); |
| 292 | let b = tmp.path().join("b.json"); |
| 293 | let mut tx = SetupTransaction::new(); |
| 294 | tx.stage(a.clone(), b"1".to_vec()) |
| 295 | .stage(b.clone(), b"2".to_vec()); |
| 296 | let preview = tx.preview(); |
| 297 | assert_eq!(preview, vec![a.as_path(), b.as_path()]); |
| 298 | assert!(!a.exists()); |
| 299 | assert!(!b.exists()); |
| 300 | } |
| 301 | |
| 302 | #[test] |
| 303 | fn dropped_transaction_leaves_files_unchanged() { |
| 304 | let tmp = tempfile::tempdir().unwrap(); |
| 305 | let a = tmp.path().join("a.json"); |
| 306 | { |
| 307 | let mut tx = SetupTransaction::new(); |
| 308 | tx.stage(a.clone(), b"staged".to_vec()); |
| 309 | // tx dropped here without commit |
| 310 | } |
| 311 | assert!(!a.exists()); |
| 312 | } |
| 313 | |
| 314 | #[test] |
| 315 | fn transaction_commit_applies_all() { |
| 316 | let tmp = tempfile::tempdir().unwrap(); |
| 317 | let a = tmp.path().join("a.json"); |
| 318 | let b = tmp.path().join("sub/b.json"); |
| 319 | let mut tx = SetupTransaction::new(); |
| 320 | tx.stage(a.clone(), b"A".to_vec()) |
| 321 | .stage(b.clone(), b"B".to_vec()); |
| 322 | tx.commit().unwrap(); |
| 323 | assert_eq!(read(&a), "A"); |
| 324 | assert_eq!(read(&b), "B"); |
| 325 | } |
| 326 | |
| 327 | #[test] |
| 328 | fn transaction_rolls_back_on_partial_failure() { |
| 329 | let tmp = tempfile::tempdir().unwrap(); |
| 330 | let good = tmp.path().join("good.json"); |
| 331 | fs::write(&good, "ORIGINAL").unwrap(); |
| 332 | |
| 333 | // Second target is unwritable: a path whose parent is an existing file. |
| 334 | let blocker = tmp.path().join("blocker"); |
| 335 | fs::write(&blocker, "i am a file").unwrap(); |
| 336 | let bad = blocker.join("child.json"); // parent is a file → create_dir_all fails |
| 337 | |
| 338 | let mut tx = SetupTransaction::new(); |
| 339 | tx.stage(good.clone(), b"UPDATED".to_vec()) |
| 340 | .stage(bad.clone(), b"NOPE".to_vec()); |
| 341 | let err = tx.commit().unwrap_err(); |
| 342 | assert!(format!("{err:#}").contains("rolled back")); |
| 343 | |
| 344 | // The first file must be restored to its original contents. |
| 345 | assert_eq!(read(&good), "ORIGINAL"); |
| 346 | assert!(!bad.exists()); |
| 347 | } |
| 348 | |
| 349 | #[test] |
| 350 | fn transaction_rollback_removes_newly_created_file() { |
| 351 | let tmp = tempfile::tempdir().unwrap(); |
| 352 | let fresh = tmp.path().join("fresh.json"); // did not exist before |
| 353 | let blocker = tmp.path().join("blocker"); |
| 354 | fs::write(&blocker, "file").unwrap(); |
| 355 | let bad = blocker.join("child.json"); |
| 356 | |
| 357 | let mut tx = SetupTransaction::new(); |
| 358 | tx.stage(fresh.clone(), b"created".to_vec()) |
| 359 | .stage(bad, b"x".to_vec()); |
| 360 | assert!(tx.commit().is_err()); |
| 361 | // The newly created file must be removed on rollback, not left behind. |
| 362 | assert!(!fresh.exists()); |
| 363 | } |
| 364 | |
| 365 | #[test] |
| 366 | fn compact_json_and_query_strings_mask_their_credentials() { |
| 367 | // A synthetic, unsigned test value, assembled from its three |
| 368 | // segments so the source holds no complete token literal. |
| 369 | let jwt = [ |
| 370 | "eyJhbGciOiJSUzI1NiJ9", |
| 371 | "eyJzdWIiOiJ1c2VyIn0", |
| 372 | "c2lnbmF0dXJlLXZhbHVl", |
| 373 | ] |
| 374 | .join("."); |
| 375 | let jwt = jwt.as_str(); |
| 376 | let refresh = "rt_Q2x9fK3mZ7pL1vB8nH4sT6wY0aE5cR2d"; |
| 377 | let cases = [ |
| 378 | format!( |
| 379 | r#"{{"tokens":{{"id_token":"{jwt}","access_token":"{jwt}","refresh_token":"{refresh}"}}}}"# |
| 380 | ), |
| 381 | format!(r#"{{"token_type":"Bearer","access_token":"{jwt}"}}"#), |
| 382 | format!(r#"[{{"Authorization":"Bearer {jwt}"}}]"#), |
| 383 | format!("https://example.test/cb?state=abc&access_token={jwt}&x=1"), |
| 384 | ]; |
| 385 | for policy in [RedactionPolicy::CredentialShaped, RedactionPolicy::KeyBased] { |
| 386 | for case in &cases { |
| 387 | let out = redact_secrets_with(case, policy); |
| 388 | assert!(!out.contains(jwt), "{policy:?} leaked a JWT: {out}"); |
| 389 | assert!( |
| 390 | !out.contains(refresh), |
| 391 | "{policy:?} leaked a refresh token: {out}" |
| 392 | ); |
| 393 | } |
| 394 | } |
| 395 | // Structure and harmless members stay byte-exact. |
| 396 | assert_eq!( |
| 397 | redact_model_bound_secrets(&cases[1]), |
| 398 | r#"{"token_type":"Bearer","access_token":"[redacted]"}"# |
| 399 | ); |
| 400 | let json = serde_json::json!({ |
| 401 | "summary": format!("{{\"access_token\":\"{jwt}\"}}"), |
| 402 | "stdout_summary": "token = make_token()", |
| 403 | "api_key": "short", |
| 404 | "access_token": jwt, |
| 405 | }); |
| 406 | let masked = redact_json_model_bound_secrets(&json); |
| 407 | assert!(!masked.to_string().contains(jwt), "{masked}"); |
| 408 | assert_eq!(masked["stdout_summary"], "token = make_token()"); |
| 409 | // A sensitive key's value is masked wholesale, credential-shaped or not. |
| 410 | assert_eq!(masked["api_key"], REDACTED); |
| 411 | } |
| 412 | |
| 413 | #[test] |
| 414 | fn model_bound_redaction_keeps_code_and_config_byte_exact() { |
| 415 | // Every line here is code or configuration that a model must be able |
| 416 | // to quote back for an exact-match edit (#5546). |
| 417 | let lines = [ |
| 418 | " \"jsonwebtoken\": \"^9.0.2\",", |
| 419 | " \"@types/jsonwebtoken\": \"^9.0.5\",", |
| 420 | " \"password-validator\": \"^5.3.0\",", |
| 421 | " \"authorization\": \"1.0.0\",", |
| 422 | " password: credentials?.password,", |
| 423 | " token = generate_verification_token()", |
| 424 | " secret: process.env.NEXTAUTH_SECRET!,", |
| 425 | " password?: string;", |
| 426 | " token: string;", |
| 427 | " \"tokenizer\": \"gpt2\",", |
| 428 | "max tokens = 8192", |
| 429 | "export const AUTH_TOKEN_HEADER = 'x-auth-token';", |
| 430 | "{\"id\":1, \"password\": \"x\", \"language\": \"en\"}", |
| 431 | "{\"id\":1, \"password\":\"x\", \"language\":\"en\"}", |
| 432 | "password = hunter2", |
| 433 | "api_key = os.environ[\"OPENAI_API_KEY\"]", |
| 434 | "let token = ${TOKEN_FROM_ENV}", |
| 435 | "auth_url = https://example.test/oauth/token", |
| 436 | ]; |
| 437 | for line in lines { |
| 438 | assert_eq!( |
| 439 | redact_model_bound_secrets(line), |
| 440 | line, |
| 441 | "line changed: {line}" |
| 442 | ); |
| 443 | } |
| 444 | let file = lines.join("\n"); |
| 445 | assert_eq!(redact_model_bound_secrets(&file), file); |
| 446 | } |
| 447 | |
| 448 | #[test] |
| 449 | fn model_bound_redaction_masks_credential_shaped_values() { |
| 450 | let hex40 = ["0123456789abcdef", "0123456789abcdef", "01234567"].concat(); |
| 451 | let sk = ["sk-", "abcdef1234567890abcdef"].concat(); |
| 452 | let jwt = [ |
| 453 | "eyJhbGciOiJIUzI1NiJ9", |
| 454 | ".", |
| 455 | "eyJzdWIiOiIxMjM0NTY3ODkwIn0", |
| 456 | ".", |
| 457 | "c2lnbmF0dXJlLXNpZ25hdHVyZQ", |
| 458 | ] |
| 459 | .concat(); |
| 460 | let ya = ["ya29.", "a0AfH6SMBx1234567890abcdefghij"].concat(); |
| 461 | let cases = [ |
| 462 | ( |
| 463 | format!("NEXTAUTH_SECRET={hex40}"), |
| 464 | "NEXTAUTH_SECRET=[redacted]".to_string(), |
| 465 | ), |
| 466 | ( |
| 467 | format!("api_key = \"{sk}\""), |
| 468 | "api_key = \"[redacted]\"".to_string(), |
| 469 | ), |
| 470 | ( |
| 471 | format!(" \"access_token\": \"{ya}\","), |
| 472 | " \"access_token\": \"[redacted]\",".to_string(), |
| 473 | ), |
| 474 | ( |
| 475 | format!("Authorization: Bearer {jwt}"), |
| 476 | "Authorization: [redacted]".to_string(), |
| 477 | ), |
| 478 | ( |
| 479 | format!("curl -H \"Authorization: Bearer {jwt}\" https://api.test"), |
| 480 | "curl -H \"Authorization: Bearer [redacted]\" https://api.test".to_string(), |
| 481 | ), |
| 482 | ( |
| 483 | format!("password = {hex40}, retries = 3"), |
| 484 | "password = [redacted], retries = 3".to_string(), |
| 485 | ), |
| 486 | ( |
| 487 | format!("found key {sk} in the log"), |
| 488 | "found key [redacted] in the log".to_string(), |
| 489 | ), |
| 490 | ]; |
| 491 | for (input, expected) in cases { |
| 492 | assert_eq!( |
| 493 | redact_model_bound_secrets(&input), |
| 494 | expected, |
| 495 | "input: {input}" |
| 496 | ); |
| 497 | } |
| 498 | } |
| 499 | |
| 500 | #[test] |
| 501 | fn private_key_blocks_are_masked_between_pem_markers() { |
| 502 | // Assemble the PEM markers at runtime so the source file never |
| 503 | // contains a literal private-key header for a scanner to match; the |
| 504 | // runtime strings are identical to a real block. |
| 505 | let begin = ["-----BEGIN RSA", " PRIVATE KEY-----"].concat(); |
| 506 | let end = ["-----END RSA", " PRIVATE KEY-----"].concat(); |
| 507 | let body = "MIIEpAIBAAKCAQEA0Z3VS5JJcds3xfn\nabcdefghijklmnopqrstuvwxyz012345"; |
| 508 | let pem = format!("{begin}\n{body}\n{end}\nnext_line = ok\n"); |
| 509 | let expected = format!("{begin}\n[redacted]\n[redacted]\n{end}\nnext_line = ok\n"); |
| 510 | assert_eq!(redact_model_bound_secrets(&pem), expected); |
| 511 | assert_eq!(redact_secrets(&pem), expected); |
| 512 | } |
| 513 | |
| 514 | #[test] |
| 515 | fn key_based_policy_is_unchanged_by_the_model_bound_mode() { |
| 516 | // The broad scrubber for logs/previews keeps masking key-only hits. |
| 517 | assert_eq!( |
| 518 | redact_secrets(" password: credentials?.password,"), |
| 519 | " password: [redacted]" |
| 520 | ); |
| 521 | assert_eq!( |
| 522 | redact_secrets(" \"password-validator\": \"^5.3.0\","), |
| 523 | " \"password-validator\": \"[redacted]\"" |
| 524 | ); |
| 525 | } |
| 526 | |
| 527 | #[test] |
| 528 | fn redact_masks_keyed_secrets_toml_and_json() { |
| 529 | let synthetic_secret = synthetic_secret_fixture(); |
| 530 | let input = format!( |
| 531 | "\ |
| 532 | api_key = \"sk-supersecretvalue123\" |
| 533 | provider = \"openai\" |
| 534 | \"token\": \"{synthetic_secret}\", |
| 535 | model = \"mimo-ultraspeed\" |
| 536 | PASSWORD=hunter2hunter2" |
| 537 | ); |
| 538 | let out = redact_secrets(&input); |
| 539 | assert!(!out.contains("sk-supersecretvalue123"), "{out}"); |
| 540 | assert!(!out.contains(&synthetic_secret), "{out}"); |
| 541 | assert!(!out.contains("hunter2hunter2"), "{out}"); |
| 542 | // Non-secret values survive untouched. |
| 543 | assert!(out.contains("provider = \"openai\"")); |
| 544 | assert!(out.contains("model = \"mimo-ultraspeed\"")); |
| 545 | assert!(out.matches(REDACTED).count() >= 3, "{out}"); |
| 546 | } |
| 547 | |
| 548 | #[test] |
| 549 | fn redact_json_masks_camel_case_and_dotted_secret_keys() { |
| 550 | let synthetic_secret = synthetic_secret_fixture(); |
| 551 | let input = serde_json::json!({ |
| 552 | "accessToken": synthetic_secret.clone(), |
| 553 | "refreshToken": synthetic_secret_fixture(), |
| 554 | "oauth.token": synthetic_secret_fixture(), |
| 555 | "APIKey": synthetic_secret_fixture(), |
| 556 | "maxTokens": 8192, |
| 557 | "tokenBudget": 4096, |
| 558 | "tokenCount": 1024, |
| 559 | "token_count": 512, |
| 560 | "tokenizer": "sentencepiece", |
| 561 | }); |
| 562 | |
| 563 | let out = redact_json_secrets(&input); |
| 564 | for key in ["accessToken", "refreshToken", "oauth.token", "APIKey"] { |
| 565 | assert_eq!(out[key], REDACTED, "{key}: {out}"); |
| 566 | } |
| 567 | assert_eq!(out["maxTokens"], 8192); |
| 568 | assert_eq!(out["tokenBudget"], 4096); |
| 569 | assert_eq!(out["tokenCount"], 1024); |
| 570 | assert_eq!(out["token_count"], 512); |
| 571 | assert_eq!(out["tokenizer"], "sentencepiece"); |
| 572 | assert!(!out.to_string().contains(&synthetic_secret), "{out}"); |
| 573 | } |
| 574 | |
| 575 | #[test] |
| 576 | fn redact_json_truncates_pathological_nesting() { |
| 577 | let mut value = serde_json::Value::String("leaf".to_string()); |
| 578 | for _ in 0..150 { |
| 579 | let mut map = serde_json::Map::new(); |
| 580 | map.insert("t".to_string(), value); |
| 581 | value = serde_json::Value::Object(map); |
| 582 | } |
| 583 | let out = redact_json_secrets(&value); |
| 584 | let mut cursor = &out; |
| 585 | let mut descended = 0; |
| 586 | while let serde_json::Value::Object(map) = cursor { |
| 587 | cursor = map.values().next().expect("single-key nesting"); |
| 588 | descended += 1; |
| 589 | } |
| 590 | assert_eq!( |
| 591 | cursor, |
| 592 | &serde_json::Value::String(REDACTED.to_string()), |
| 593 | "over-deep value must be redacted, not traversed" |
| 594 | ); |
| 595 | assert!(descended < 150, "guard must fire before the leaf"); |
| 596 | } |
| 597 | |
| 598 | #[test] |
| 599 | fn redact_text_masks_camel_case_and_dotted_secret_assignments() { |
| 600 | let synthetic_secret = synthetic_secret_fixture(); |
| 601 | for key in ["accessToken", "refreshToken", "oauth.token", "APIKey"] { |
| 602 | let out = redact_secrets(&format!("request failed: {key} = {synthetic_secret}")); |
| 603 | assert!(!out.contains(&synthetic_secret), "{key}: {out}"); |
| 604 | assert!(out.contains(REDACTED), "{key}: {out}"); |
| 605 | } |
| 606 | for key in ["tokenBudget", "tokenCount", "token_count"] { |
| 607 | let input = format!("model usage: {key} = 8192"); |
| 608 | assert_eq!(redact_secrets(&input), input, "{key}"); |
| 609 | } |
| 610 | } |
| 611 | |
| 612 | #[test] |
| 613 | fn redact_masks_bare_token_prefixes() { |
| 614 | let out = redact_secrets("the leaked key sk-abcdef1234567890 appeared in a log"); |
| 615 | assert!(!out.contains("sk-abcdef1234567890"), "{out}"); |
| 616 | assert!(out.contains(REDACTED)); |
| 617 | assert!(out.contains("appeared in a log")); |
| 618 | } |
| 619 | |
| 620 | #[test] |
| 621 | fn redact_masks_inline_sensitive_assignments_after_prose_prefixes() { |
| 622 | let out = redact_secrets( |
| 623 | "Decision: use token=plain-secret-value and api_key:another-secret-value", |
| 624 | ); |
| 625 | assert!(!out.contains("plain-secret-value"), "{out}"); |
| 626 | assert!(!out.contains("another-secret-value"), "{out}"); |
| 627 | assert_eq!(out.matches(REDACTED).count(), 2, "{out}"); |
| 628 | assert!(out.starts_with("Decision: use "), "{out}"); |
| 629 | } |
| 630 | |
| 631 | #[test] |
| 632 | fn redact_masks_spaced_assignment_that_is_not_the_first_separator() { |
| 633 | // The shape `redact_secrets(&format!("{error:#}"))` produces: an |
| 634 | // anyhow chain puts prose and its own `: ` separators in front of the |
| 635 | // assignment, so the sensitive key never owns the line's first |
| 636 | // separator and the whole-line pass declines the line. |
| 637 | let out = redact_secrets("request failed: api_key = AIzaSyDeadBeefLeak"); |
| 638 | assert!(!out.contains("AIzaSyDeadBeefLeak"), "{out}"); |
| 639 | assert!(out.contains(REDACTED), "{out}"); |
| 640 | |
| 641 | let synthetic_secret = synthetic_secret_fixture(); |
| 642 | let out = redact_secrets(&format!("note: the token = {synthetic_secret}")); |
| 643 | assert!(!out.contains(&synthetic_secret), "{out}"); |
| 644 | assert!(out.contains(REDACTED), "{out}"); |
| 645 | } |
| 646 | |
| 647 | #[test] |
| 648 | fn redact_masks_whole_multi_word_value_of_a_spaced_assignment() { |
| 649 | // Assemble the placeholder at runtime so secret scanners do not |
| 650 | // mistake a redaction fixture for a committed credential. |
| 651 | let bearer = ["Bear", "er"].concat(); |
| 652 | let credential = ["abc123", "def456", "ghi"].concat(); |
| 653 | let out = redact_secrets(&format!( |
| 654 | "mcp call failed: authorization = {bearer} {credential}" |
| 655 | )); |
| 656 | assert!(!out.contains(&credential), "{out}"); |
| 657 | assert!(!out.contains(&bearer), "{out}"); |
| 658 | assert!( |
| 659 | out.starts_with("mcp call failed: authorization = "), |
| 660 | "{out}" |
| 661 | ); |
| 662 | } |
| 663 | |
| 664 | #[test] |
| 665 | fn redact_spaced_pass_leaves_ordinary_prose_alone() { |
| 666 | // No sensitive key, so the spaced-assignment state machine must not |
| 667 | // start swallowing the rest of the line. |
| 668 | let input = "the quick brown fox = jumps over the lazy dog"; |
| 669 | assert_eq!(redact_secrets(input), input); |
| 670 | let input = "note: the model = deepseek-v4-pro and the seed = 7"; |
| 671 | assert_eq!(redact_secrets(input), input); |
| 672 | } |
| 673 | |
| 674 | #[test] |
| 675 | fn redact_leaves_token_count_diagnostics_intact() { |
| 676 | // "tokens" is the English plural of a usage metric, not a credential |
| 677 | // key. The spaced-assignment pass used to treat the "token" hint as a |
| 678 | // substring and then drop the rest of the line, which made the exact |
| 679 | // class of error people paste into issues unreadable. |
| 680 | for input in [ |
| 681 | "stream error: max tokens = 8192 but budget = 4096", |
| 682 | "error: token expired", |
| 683 | "request failed: token count: 4096 exceeds the model limit", |
| 684 | "http 401: authorization header rejected", |
| 685 | "warning: password policy requires 12 characters", |
| 686 | "note: secret scanning found 3 issues", |
| 687 | ] { |
| 688 | assert_eq!(redact_secrets(input), input, "{input}"); |
| 689 | } |
| 690 | } |
| 691 | |
| 692 | #[test] |
| 693 | fn redact_still_masks_a_bearer_token_assignment() { |
| 694 | // Counterpart of the diagnostic test above: a real credential keyed |
| 695 | // as `token` (or `api_token`) must still be dropped, including a |
| 696 | // multi-word Bearer value that is not a known bare-token prefix. |
| 697 | // The JWT is assembled at runtime so no scanner-shaped literal sits |
| 698 | // in the source tree — same precedent as the AWS fixture in |
| 699 | // `crates/workflow/src/redaction.rs`. |
| 700 | let jwt = ["eyJhbGciOiJIUzI1NiJ9", "e30", "c2lnbmF0dXJl"].join("."); |
| 701 | let out = redact_secrets(&format!("stream error: token = Bearer {jwt}")); |
| 702 | assert!(!out.contains(&jwt), "{out}"); |
| 703 | assert!(!out.contains("Bearer"), "{out}"); |
| 704 | assert!(out.starts_with("stream error: token = "), "{out}"); |
| 705 | assert!(out.contains(REDACTED), "{out}"); |
| 706 | |
| 707 | let synthetic_secret = synthetic_secret_fixture(); |
| 708 | let out = redact_secrets(&format!("note: api_token = {synthetic_secret}")); |
| 709 | assert!(!out.contains(&synthetic_secret), "{out}"); |
| 710 | assert!(out.contains(REDACTED), "{out}"); |
| 711 | } |
| 712 | |
| 713 | #[test] |
| 714 | fn redact_preserves_line_structure() { |
| 715 | let input = "line1\nsecret = \"xyzsecretvalue\"\nline3"; |
| 716 | let out = redact_secrets(input); |
| 717 | let lines: Vec<&str> = out.lines().collect(); |
| 718 | assert_eq!(lines.len(), 3); |
| 719 | assert_eq!(lines[0], "line1"); |
| 720 | assert_eq!(lines[2], "line3"); |
| 721 | assert!(lines[1].contains(REDACTED)); |
| 722 | } |
| 723 | |
| 724 | #[test] |
| 725 | fn redact_leaves_plain_text_untouched() { |
| 726 | let input = "the quick brown fox = jumps over"; |
| 727 | // `fox` key has no sensitive hint → unchanged. |
| 728 | assert_eq!(redact_secrets(input), input); |
| 729 | } |
| 730 | } |
| 731 |