| 1 | //! The allowlisted safe-label boundary for `/preview-request` (#1004). |
| 2 | //! |
| 3 | //! Every free-form string that reaches a manifest surface — human table or |
| 4 | //! JSON — crosses this module first. Nothing here is a "scrubber" that tries |
| 5 | //! to find secrets in arbitrary text: a value either matches a narrow |
| 6 | //! allowlist and is published verbatim, or it is replaced by a stable |
| 7 | //! `sha256:<12 hex>` fingerprint. Two previews of the same route still |
| 8 | //! compare equal, and nothing that was not on the allowlist is ever printed. |
| 9 | //! |
| 10 | //! Why this exists at all: the obvious "identifier" fields are not safe by |
| 11 | //! construction. A custom `[providers.<name>]` key is user-authored text, and |
| 12 | //! a model id can be a filesystem path (`/models/llama-3.gguf`), a URL, a URL |
| 13 | //! path, or a deployment id that is itself a credential. Bounding the *shape* |
| 14 | //! of what may be printed is the only way to keep those out of a manifest a |
| 15 | //! user will paste into an issue tracker. |
| 16 | //! |
| 17 | //! Error strings get the same treatment through [`safe_error_text`], which is |
| 18 | //! path- and URL-path-safe: an MCP or request-preparation failure often |
| 19 | //! carries an absolute workspace path or an endpoint URL, and neither may |
| 20 | //! reach the transcript. |
| 21 | |
| 22 | use serde::{Serialize, Serializer}; |
| 23 | |
| 24 | /// Longest identifier published verbatim. Real provider/model/route ids are |
| 25 | /// far shorter; anything longer is treated as opaque payload. |
| 26 | const MAX_IDENTIFIER_LEN: usize = 64; |
| 27 | /// Longest short phrase (labels with spaces, e.g. a billing presentation). |
| 28 | const MAX_PHRASE_LEN: usize = 80; |
| 29 | /// Longest error sentence published. Errors are truncated, never wrapped. |
| 30 | const MAX_ERROR_LEN: usize = 200; |
| 31 | /// A run of this many characters from a single "opaque" alphabet reads as a |
| 32 | /// key, token, or hash rather than as a name. |
| 33 | const OPAQUE_RUN_LEN: usize = 20; |
| 34 | /// Hex prefix length used when a value is replaced by its fingerprint. |
| 35 | const FINGERPRINT_HEX_LEN: usize = 12; |
| 36 | |
| 37 | /// A string that is safe to publish on a manifest surface. |
| 38 | /// |
| 39 | /// Construct with [`SafeLabel::identifier`], [`SafeLabel::catalog_model`], or |
| 40 | /// [`SafeLabel::phrase`]; all fall back to a fingerprint when the input is not |
| 41 | /// on the allowlist. There is deliberately no constructor that takes |
| 42 | /// arbitrary text verbatim. |
| 43 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 44 | pub struct SafeLabel { |
| 45 | text: String, |
| 46 | redacted: bool, |
| 47 | } |
| 48 | |
| 49 | impl SafeLabel { |
| 50 | /// A generic identifier-shaped value: provider id, route id, reasoning |
| 51 | /// tier. Allows `A-Z a-z 0-9 . _ : - + @` and rejects every slash. Model |
| 52 | /// ids with a slash must use [`Self::catalog_model`] instead. |
| 53 | pub fn identifier(raw: &str) -> Self { |
| 54 | let trimmed = raw.trim(); |
| 55 | if identifier_is_allowlisted(trimmed) { |
| 56 | Self { |
| 57 | text: trimmed.to_string(), |
| 58 | redacted: false, |
| 59 | } |
| 60 | } else { |
| 61 | Self::fingerprint(raw) |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | /// A model label. Slash-bearing values are published only when the exact |
| 66 | /// id exists in the active local model catalog; a vendor-looking prefix is |
| 67 | /// never authority by itself. Non-slash ids retain the generic identifier |
| 68 | /// boundary for custom compatible deployments. |
| 69 | pub fn catalog_model(raw: &str) -> Self { |
| 70 | let trimmed = raw.trim(); |
| 71 | if !trimmed.contains('/') { |
| 72 | return Self::identifier(raw); |
| 73 | } |
| 74 | if catalog_model_identifier_is_allowlisted(trimmed) { |
| 75 | Self { |
| 76 | text: trimmed.to_string(), |
| 77 | redacted: false, |
| 78 | } |
| 79 | } else { |
| 80 | Self::fingerprint(raw) |
| 81 | } |
| 82 | } |
| 83 | |
| 84 | /// A short human phrase: the same allowlist plus spaces, parentheses, and |
| 85 | /// commas, for host-supplied presentation labels. |
| 86 | pub fn phrase(raw: &str) -> Self { |
| 87 | let trimmed = raw.trim(); |
| 88 | if phrase_is_allowlisted(trimmed) { |
| 89 | Self { |
| 90 | text: trimmed.to_string(), |
| 91 | redacted: false, |
| 92 | } |
| 93 | } else { |
| 94 | Self::fingerprint(raw) |
| 95 | } |
| 96 | } |
| 97 | |
| 98 | /// Replace a value with a stable fingerprint of its exact bytes. |
| 99 | fn fingerprint(raw: &str) -> Self { |
| 100 | let digest = crate::hashing::sha256_hex(raw.as_bytes()); |
| 101 | Self { |
| 102 | text: format!("sha256:{}", &digest[..FINGERPRINT_HEX_LEN]), |
| 103 | redacted: true, |
| 104 | } |
| 105 | } |
| 106 | |
| 107 | pub fn as_str(&self) -> &str { |
| 108 | &self.text |
| 109 | } |
| 110 | |
| 111 | /// True when the original value failed the allowlist and only its |
| 112 | /// fingerprint is being published. |
| 113 | #[cfg(any(test, feature = "test-support"))] |
| 114 | pub fn is_redacted(&self) -> bool { |
| 115 | self.redacted |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | impl std::fmt::Display for SafeLabel { |
| 120 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 121 | f.write_str(&self.text) |
| 122 | } |
| 123 | } |
| 124 | |
| 125 | impl Serialize for SafeLabel { |
| 126 | fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> { |
| 127 | serializer.serialize_str(&self.text) |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | fn identifier_is_allowlisted(value: &str) -> bool { |
| 132 | if value.is_empty() || value.len() > MAX_IDENTIFIER_LEN { |
| 133 | return false; |
| 134 | } |
| 135 | if value.starts_with('/') || value.starts_with('~') || value.starts_with('.') { |
| 136 | return false; |
| 137 | } |
| 138 | if value.contains("//") || value.contains("..") || value.contains(':') && value.contains('/') { |
| 139 | return false; |
| 140 | } |
| 141 | if !value.chars().all(is_identifier_char) { |
| 142 | return false; |
| 143 | } |
| 144 | if value.contains('/') { |
| 145 | return false; |
| 146 | } |
| 147 | !looks_opaque(value) |
| 148 | } |
| 149 | |
| 150 | fn catalog_model_identifier_is_allowlisted(value: &str) -> bool { |
| 151 | if value.is_empty() |
| 152 | || value.len() > MAX_IDENTIFIER_LEN |
| 153 | || value.starts_with('/') |
| 154 | || value.starts_with('~') |
| 155 | || value.starts_with('.') |
| 156 | || value.contains("//") |
| 157 | || value.contains("..") |
| 158 | || value.contains(':') |
| 159 | || !value.chars().all(is_identifier_char) |
| 160 | || looks_opaque(value) |
| 161 | { |
| 162 | return false; |
| 163 | } |
| 164 | codewhale_config::catalog::reviewed::public_model_identifier(value) |
| 165 | } |
| 166 | |
| 167 | fn phrase_is_allowlisted(value: &str) -> bool { |
| 168 | if value.is_empty() || value.len() > MAX_PHRASE_LEN { |
| 169 | return false; |
| 170 | } |
| 171 | if value.contains('/') || value.contains('\\') || value.contains('~') { |
| 172 | return false; |
| 173 | } |
| 174 | if !value |
| 175 | .chars() |
| 176 | .all(|ch| is_identifier_char(ch) || matches!(ch, ' ' | '(' | ')' | ',')) |
| 177 | { |
| 178 | return false; |
| 179 | } |
| 180 | !looks_opaque(value) |
| 181 | } |
| 182 | |
| 183 | fn is_identifier_char(ch: char) -> bool { |
| 184 | ch.is_ascii_alphanumeric() || matches!(ch, '.' | '_' | '-' | '+' | '@' | ':' | '/') |
| 185 | } |
| 186 | |
| 187 | /// Whether a value carries a key-, token-, or hash-shaped run. |
| 188 | /// |
| 189 | /// Deliberately shape-based rather than a keyword list: `sk-`-style prefixes |
| 190 | /// are only one of the ways a deployment id can be a credential. |
| 191 | fn looks_opaque(value: &str) -> bool { |
| 192 | let lower = value.to_ascii_lowercase(); |
| 193 | for marker in ["sk-", "api_key", "apikey", "secret", "password", "bearer"] { |
| 194 | if lower.contains(marker) { |
| 195 | return true; |
| 196 | } |
| 197 | } |
| 198 | let mut run = 0usize; |
| 199 | for ch in value.chars() { |
| 200 | // A long unbroken alphanumeric run with no separator is what base64 |
| 201 | // and hex payloads look like; real ids use `-`, `.`, or `/`. |
| 202 | if ch.is_ascii_alphanumeric() { |
| 203 | run += 1; |
| 204 | if run >= OPAQUE_RUN_LEN { |
| 205 | return true; |
| 206 | } |
| 207 | } else { |
| 208 | run = 0; |
| 209 | } |
| 210 | } |
| 211 | false |
| 212 | } |
| 213 | |
| 214 | /// Longest single word published verbatim inside an error sentence. |
| 215 | const MAX_ERROR_WORD_LEN: usize = 40; |
| 216 | /// Longest scheme published from a URL-shaped token. |
| 217 | const MAX_SCHEME_LEN: usize = 16; |
| 218 | /// Longest `host[:port]` published from a URL-shaped token. |
| 219 | const MAX_HOST_LEN: usize = 80; |
| 220 | /// Stand-in for a word that is not on the error allowlist. |
| 221 | const REDACTED_WORD: &str = "<redacted>"; |
| 222 | /// Stand-in for anything path-shaped. |
| 223 | const REDACTED_PATH: &str = "<path-redacted>"; |
| 224 | |
| 225 | /// Bound an error string so it can be shown in the transcript. |
| 226 | /// |
| 227 | /// This is an **allowlist**, not a scrubber. Host error text is arbitrary: it |
| 228 | /// can interpolate a route id, a model id, a deployment path, a quoted server |
| 229 | /// name, a URL with a secret in its path, or a raw credential. Rather than |
| 230 | /// hunting for the bad parts, every whitespace-separated token must earn its |
| 231 | /// place: |
| 232 | /// |
| 233 | /// - the config crate's secret redaction runs first; |
| 234 | /// - a token containing a control character is dropped entirely; |
| 235 | /// - a URL-shaped token keeps only `scheme://host[:port]`, and only when both |
| 236 | /// are themselves allowlisted — the path, query, fragment, and userinfo are |
| 237 | /// never published, because a deployment path can *be* the credential; |
| 238 | /// - a path-shaped token (POSIX absolute, `~/`, Windows drive, or anything |
| 239 | /// containing a backslash) collapses to `REDACTED_PATH`; |
| 240 | /// - a token carrying a quote character (`"`, `'`, or a backtick) is replaced |
| 241 | /// wholesale: quoted spans are where hostile identifiers hide; |
| 242 | /// - anything else must be a short, ordinary word — ASCII alphanumerics plus |
| 243 | /// `-`, `_`, `.`, bounded by `MAX_ERROR_WORD_LEN` and rejected by |
| 244 | /// `looks_opaque` — with only a small set of sentence punctuation allowed |
| 245 | /// at its edges. Everything else becomes `REDACTED_WORD`. |
| 246 | /// |
| 247 | /// The result therefore contains no filesystem path, no URL path, no quoted |
| 248 | /// span, no token-shaped run, and no control character, and is truncated to |
| 249 | /// `MAX_ERROR_LEN`. |
| 250 | pub fn safe_error_text(raw: &str) -> String { |
| 251 | let redacted = codewhale_config::persistence::redact_secrets(raw); |
| 252 | let mut out = String::with_capacity(redacted.len().min(MAX_ERROR_LEN)); |
| 253 | let mut last_was_redacted = false; |
| 254 | for token in redacted.split_whitespace() { |
| 255 | let safe = safe_error_token(token); |
| 256 | if safe.is_empty() { |
| 257 | continue; |
| 258 | } |
| 259 | // Collapse runs of redactions: `<redacted> <redacted> <redacted>` is |
| 260 | // noise, and its length would leak the shape of what was removed. |
| 261 | let is_redacted = safe == REDACTED_WORD; |
| 262 | if is_redacted && last_was_redacted { |
| 263 | continue; |
| 264 | } |
| 265 | last_was_redacted = is_redacted; |
| 266 | if !out.is_empty() { |
| 267 | out.push(' '); |
| 268 | } |
| 269 | out.push_str(&safe); |
| 270 | } |
| 271 | if out.is_empty() { |
| 272 | out.push_str("<unavailable>"); |
| 273 | } |
| 274 | if out.len() > MAX_ERROR_LEN { |
| 275 | out.truncate( |
| 276 | (0..=MAX_ERROR_LEN) |
| 277 | .rev() |
| 278 | .find(|index| out.is_char_boundary(*index)) |
| 279 | .unwrap_or(0), |
| 280 | ); |
| 281 | out.push('…'); |
| 282 | } |
| 283 | out |
| 284 | } |
| 285 | |
| 286 | fn safe_error_token(token: &str) -> String { |
| 287 | if token.chars().any(char::is_control) { |
| 288 | return REDACTED_WORD.to_string(); |
| 289 | } |
| 290 | // A URL keeps its scheme and host and loses everything after it — but only |
| 291 | // when the scheme and host are themselves ordinary. |
| 292 | if let Some(scheme_end) = token.find("://") { |
| 293 | return safe_url_token(token, scheme_end); |
| 294 | } |
| 295 | // Absolute and home-relative paths, plus Windows drive paths, collapse |
| 296 | // entirely: a workspace path names the user's machine and project. |
| 297 | let looks_like_path = token.starts_with('/') |
| 298 | || token.starts_with("~/") |
| 299 | || token.contains('\\') |
| 300 | || (token.len() > 2 && token.as_bytes()[1] == b':' && token.contains('\\')); |
| 301 | if looks_like_path { |
| 302 | return REDACTED_PATH.to_string(); |
| 303 | } |
| 304 | // Quoted spans are the classic carrier for a hostile server, route, or |
| 305 | // model id. Never republish one, even partially. |
| 306 | if token.contains(['"', '\'', '`']) { |
| 307 | return REDACTED_WORD.to_string(); |
| 308 | } |
| 309 | |
| 310 | let (lead, core, trail) = split_sentence_punctuation(token); |
| 311 | if core.is_empty() { |
| 312 | // Pure punctuation: keep it only if every character is on the small |
| 313 | // sentence-punctuation allowlist, which `split` already guaranteed. |
| 314 | return format!("{lead}{trail}"); |
| 315 | } |
| 316 | if error_word_is_allowlisted(core) { |
| 317 | format!("{lead}{core}{trail}") |
| 318 | } else { |
| 319 | REDACTED_WORD.to_string() |
| 320 | } |
| 321 | } |
| 322 | |
| 323 | /// Collapse a URL-shaped token to `scheme://host[:port]/<path-redacted>`. |
| 324 | /// |
| 325 | /// Userinfo, path, query, and fragment are dropped unconditionally. A scheme |
| 326 | /// or host that is not itself ordinary makes the whole token opaque rather |
| 327 | /// than publishing a hostile "host". |
| 328 | fn safe_url_token(token: &str, scheme_end: usize) -> String { |
| 329 | let scheme = &token[..scheme_end]; |
| 330 | let rest = &token[scheme_end + 3..]; |
| 331 | let authority_end = rest.find(['/', '?', '#']).unwrap_or(rest.len()); |
| 332 | let host = rest[..authority_end].rsplit('@').next().unwrap_or(""); |
| 333 | |
| 334 | let scheme_ok = !scheme.is_empty() |
| 335 | && scheme.len() <= MAX_SCHEME_LEN |
| 336 | && scheme |
| 337 | .chars() |
| 338 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '+' | '-' | '.')); |
| 339 | let host_ok = !host.is_empty() |
| 340 | && host.len() <= MAX_HOST_LEN |
| 341 | && host |
| 342 | .chars() |
| 343 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '.' | '-' | ':')) |
| 344 | && !looks_opaque(host); |
| 345 | if scheme_ok && host_ok { |
| 346 | format!("{scheme}://{host}/{REDACTED_PATH}") |
| 347 | } else { |
| 348 | REDACTED_WORD.to_string() |
| 349 | } |
| 350 | } |
| 351 | |
| 352 | /// Sentence punctuation that may bracket an allowlisted word. Deliberately |
| 353 | /// excludes every quote character. |
| 354 | fn is_edge_punctuation(ch: char) -> bool { |
| 355 | matches!(ch, '.' | ',' | ';' | ':' | '!' | '?' | '(' | ')') |
| 356 | } |
| 357 | |
| 358 | /// Split leading/trailing sentence punctuation off a token. |
| 359 | /// |
| 360 | /// Returns `("", token, "")` when the token carries punctuation that is not on |
| 361 | /// the edge allowlist, so the caller rejects it as a whole. |
| 362 | fn split_sentence_punctuation(token: &str) -> (&str, &str, &str) { |
| 363 | let start = token |
| 364 | .char_indices() |
| 365 | .find(|(_, ch)| !is_edge_punctuation(*ch)) |
| 366 | .map_or(token.len(), |(index, _)| index); |
| 367 | let end = token |
| 368 | .char_indices() |
| 369 | .rev() |
| 370 | .find(|(_, ch)| !is_edge_punctuation(*ch)) |
| 371 | .map_or(start, |(index, ch)| index + ch.len_utf8()); |
| 372 | ( |
| 373 | &token[..start], |
| 374 | &token[start..end.max(start)], |
| 375 | &token[end.max(start)..], |
| 376 | ) |
| 377 | } |
| 378 | |
| 379 | /// Whether a bare word inside an error sentence may be published verbatim. |
| 380 | fn error_word_is_allowlisted(word: &str) -> bool { |
| 381 | if word.is_empty() || word.len() > MAX_ERROR_WORD_LEN { |
| 382 | return false; |
| 383 | } |
| 384 | if word.contains("..") { |
| 385 | return false; |
| 386 | } |
| 387 | if !word |
| 388 | .chars() |
| 389 | .all(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '-' | '_' | '.')) |
| 390 | { |
| 391 | return false; |
| 392 | } |
| 393 | !looks_opaque(word) |
| 394 | } |
| 395 | |
| 396 | #[cfg(test)] |
| 397 | mod tests { |
| 398 | use super::*; |
| 399 | |
| 400 | #[test] |
| 401 | fn ordinary_identifiers_pass_through_verbatim() { |
| 402 | for value in [ |
| 403 | "deepseek-chat", |
| 404 | "claude-sonnet-4-5", |
| 405 | "gpt-5-codex", |
| 406 | "MiniMax-M3", |
| 407 | "my-gateway", |
| 408 | "kimi-k2-0905-preview", |
| 409 | ] { |
| 410 | let label = SafeLabel::identifier(value); |
| 411 | assert_eq!(label.as_str(), value, "{value} must publish verbatim"); |
| 412 | assert!(!label.is_redacted(), "{value}"); |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | #[test] |
| 417 | fn hostile_route_and_model_identifiers_never_reach_a_surface() { |
| 418 | let hostile = [ |
| 419 | "/Users/someone/models/private-weights.gguf".to_string(), |
| 420 | "~/.codewhale/config.toml".to_string(), |
| 421 | "https://internal.example.com/v1/deployments/prod".to_string(), |
| 422 | "C:\\Users\\someone\\models\\weights.bin".to_string(), |
| 423 | ["sk", "-fixture-not-a-real-key-00000000"].concat(), |
| 424 | "deployments/9f8e7d6c5b4a39281706abcdef012345".to_string(), |
| 425 | "../../etc/passwd".to_string(), |
| 426 | "model with spaces and a /path/inside".to_string(), |
| 427 | ["api_key=sk", "-live-1234567890"].concat(), |
| 428 | "src/lib.rs".to_string(), |
| 429 | "config/prod".to_string(), |
| 430 | "models/weights.gguf".to_string(), |
| 431 | "foo/bar-baz".to_string(), |
| 432 | ]; |
| 433 | for value in hostile { |
| 434 | let label = SafeLabel::identifier(&value); |
| 435 | assert!(label.is_redacted(), "`{value}` must not publish verbatim"); |
| 436 | assert!(label.as_str().starts_with("sha256:"), "{}", label.as_str()); |
| 437 | assert!(!label.as_str().contains('/'), "{}", label.as_str()); |
| 438 | assert!(!label.as_str().contains(' '), "{}", label.as_str()); |
| 439 | } |
| 440 | } |
| 441 | |
| 442 | #[test] |
| 443 | fn generic_identifiers_reject_all_slashes_and_catalog_models_require_exact_ids() { |
| 444 | for path in [ |
| 445 | "src/lib.rs", |
| 446 | "docs/PREVIEW_REQUEST.md", |
| 447 | "config/prod", |
| 448 | "models/llama-3.gguf", |
| 449 | "foo/bar-baz", |
| 450 | ] { |
| 451 | assert!( |
| 452 | SafeLabel::identifier(path).is_redacted(), |
| 453 | "relative path `{path}` must not be published" |
| 454 | ); |
| 455 | } |
| 456 | |
| 457 | let known = "qwen/qwen3.6-flash"; |
| 458 | assert!(SafeLabel::identifier(known).is_redacted()); |
| 459 | assert_eq!(SafeLabel::catalog_model(known).as_str(), known); |
| 460 | for hostile in ["openai/secrets/config", "qwen/src/lib.rs"] { |
| 461 | assert!(SafeLabel::identifier(hostile).is_redacted()); |
| 462 | assert!(SafeLabel::catalog_model(hostile).is_redacted()); |
| 463 | } |
| 464 | } |
| 465 | |
| 466 | #[test] |
| 467 | fn fingerprints_are_stable_and_distinguishing() { |
| 468 | let first = SafeLabel::identifier("/models/a.gguf"); |
| 469 | let second = SafeLabel::identifier("/models/a.gguf"); |
| 470 | let other = SafeLabel::identifier("/models/b.gguf"); |
| 471 | assert_eq!(first, second); |
| 472 | assert_ne!(first, other); |
| 473 | } |
| 474 | |
| 475 | #[test] |
| 476 | fn phrases_allow_spaces_but_not_paths() { |
| 477 | assert_eq!( |
| 478 | SafeLabel::phrase("Codex OAuth quota").as_str(), |
| 479 | "Codex OAuth quota" |
| 480 | ); |
| 481 | assert!(SafeLabel::phrase("/opt/quota/plan").is_redacted()); |
| 482 | } |
| 483 | |
| 484 | #[test] |
| 485 | fn error_text_is_path_and_url_path_safe() { |
| 486 | let raw = "MCP server 'x' failed: cannot spawn /Users/someone/work/repo/bin/server \ |
| 487 | while calling https://gateway.internal.example.com/v1/secret-deployment/messages"; |
| 488 | let safe = safe_error_text(raw); |
| 489 | assert!(!safe.contains("/Users/someone"), "{safe}"); |
| 490 | assert!(!safe.contains("/v1/secret-deployment"), "{safe}"); |
| 491 | assert!(safe.contains("<path-redacted>"), "{safe}"); |
| 492 | assert!( |
| 493 | safe.contains("https://gateway.internal.example.com/<path-redacted>"), |
| 494 | "{safe}" |
| 495 | ); |
| 496 | } |
| 497 | |
| 498 | /// The error surface is where hostile text most easily reaches a |
| 499 | /// transcript: preflight, MCP, and request-preparation failures all |
| 500 | /// interpolate route ids, model ids, server names, and endpoints. |
| 501 | #[test] |
| 502 | fn hostile_error_text_never_publishes_the_hostile_part() { |
| 503 | let home = std::env::var("HOME").unwrap_or_else(|_| "/root".to_string()); |
| 504 | let hostile: Vec<String> = vec![ |
| 505 | "route 'prod-key-8f2a' rejected key sk-live-abcdef0123456789abcdef".to_string(), |
| 506 | "cannot read C:\\Users\\someone\\.codewhale\\config.toml".to_string(), |
| 507 | format!("cannot read {home}/.codewhale/config.toml"), |
| 508 | "GET https://gw.example.com/v1/deployments/prod-key-8f2a?api_key=sk-1234567890abcdef failed".to_string(), |
| 509 | "server \"my secret server\" refused: password=hunter2".to_string(), |
| 510 | "model /Users/someone/models/private.gguf is unavailable".to_string(), |
| 511 | "authorization: Bearer eyJhbGciFAKEFIXTUREnotasecret".to_string(), |
| 512 | "endpoint http://10.0.0.5:8443/internal/deploy-9f8e7d6c5b4a3928 timed out".to_string(), |
| 513 | format!("crash{}oops", '\u{7}'), |
| 514 | ]; |
| 515 | for raw in &hostile { |
| 516 | let safe = safe_error_text(raw); |
| 517 | for forbidden in [ |
| 518 | "prod-key-8f2a", |
| 519 | "sk-live-", |
| 520 | "sk-1234567890", |
| 521 | "/Users/someone", |
| 522 | "C:\\Users", |
| 523 | ".codewhale", |
| 524 | "api_key=", |
| 525 | "hunter2", |
| 526 | "password=", |
| 527 | "eyJhbGci", |
| 528 | "/v1/deployments", |
| 529 | "/internal/deploy", |
| 530 | "private.gguf", |
| 531 | "my secret server", |
| 532 | ] { |
| 533 | assert!( |
| 534 | !safe.contains(forbidden), |
| 535 | "`{forbidden}` leaked from `{raw}`:\n{safe}" |
| 536 | ); |
| 537 | } |
| 538 | assert!(!safe.contains(&home), "home leaked from `{raw}`:\n{safe}"); |
| 539 | assert!(!safe.contains('"'), "{safe}"); |
| 540 | assert!(!safe.contains('\''), "{safe}"); |
| 541 | assert!(!safe.contains('`'), "{safe}"); |
| 542 | assert!(!safe.chars().any(char::is_control), "{safe}"); |
| 543 | } |
| 544 | } |
| 545 | |
| 546 | #[test] |
| 547 | fn ordinary_error_words_survive_so_the_message_stays_useful() { |
| 548 | let safe = safe_error_text("the shared route planner could not resolve this turn."); |
| 549 | assert_eq!( |
| 550 | safe, "the shared route planner could not resolve this turn.", |
| 551 | "an allowlisted sentence must survive intact" |
| 552 | ); |
| 553 | } |
| 554 | |
| 555 | #[test] |
| 556 | fn a_url_with_a_hostile_authority_is_dropped_rather_than_half_published() { |
| 557 | // The "host" here is a long opaque run — republishing it would be |
| 558 | // republishing the secret the path redaction exists to remove. |
| 559 | let safe = safe_error_text("calling https://9f8e7d6c5b4a39281706abcdef012345.example/x"); |
| 560 | assert!(!safe.contains("9f8e7d6c5b4a3928"), "{safe}"); |
| 561 | assert!(safe.contains("<redacted>"), "{safe}"); |
| 562 | } |
| 563 | |
| 564 | #[test] |
| 565 | fn error_text_is_bounded_and_single_line() { |
| 566 | let raw = format!("failure {}", "x".repeat(4_000)); |
| 567 | let safe = safe_error_text(&raw); |
| 568 | assert!(safe.chars().count() <= MAX_ERROR_LEN + 1, "{}", safe.len()); |
| 569 | assert!(!safe.contains('\n')); |
| 570 | } |
| 571 | } |
| 572 |