| 1 | use super::{Hook, HookCondition, HookEvent, HooksConfig}; |
| 2 | use chrono::{DateTime, Utc}; |
| 3 | use serde_json::json; |
| 4 | use std::collections::HashMap; |
| 5 | use std::fmt; |
| 6 | use std::io::{Read, Write}; |
| 7 | use std::path::PathBuf; |
| 8 | use std::process::{Child, Command, Stdio}; |
| 9 | use std::sync::mpsc::{self, Receiver, RecvTimeoutError, SyncSender, TrySendError}; |
| 10 | use std::sync::{Arc, Mutex}; |
| 11 | use std::thread::JoinHandle; |
| 12 | use std::time::{Duration, Instant}; |
| 13 | use wait_timeout::ChildExt; |
| 14 | |
| 15 | #[cfg(windows)] |
| 16 | use std::os::windows::io::AsRawHandle; |
| 17 | #[cfg(windows)] |
| 18 | use windows::Win32::Foundation::{CloseHandle, HANDLE}; |
| 19 | #[cfg(windows)] |
| 20 | use windows::Win32::System::Diagnostics::ToolHelp::{ |
| 21 | CreateToolhelp32Snapshot, TH32CS_SNAPTHREAD, THREADENTRY32, Thread32First, Thread32Next, |
| 22 | }; |
| 23 | #[cfg(windows)] |
| 24 | use windows::Win32::System::JobObjects::{ |
| 25 | AssignProcessToJobObject, CreateJobObjectW, JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE, |
| 26 | JOBOBJECT_EXTENDED_LIMIT_INFORMATION, JobObjectExtendedLimitInformation, |
| 27 | SetInformationJobObject, TerminateJobObject, |
| 28 | }; |
| 29 | #[cfg(windows)] |
| 30 | use windows::Win32::System::Threading::{OpenThread, ResumeThread, THREAD_SUSPEND_RESUME}; |
| 31 | #[cfg(windows)] |
| 32 | use windows::core::PCWSTR; |
| 33 | |
| 34 | /// Context passed to hooks via environment variables |
| 35 | #[derive(Debug, Clone, Default)] |
| 36 | pub struct HookContext { |
| 37 | /// Tool name (for ToolCallBefore/After) |
| 38 | pub tool_name: Option<String>, |
| 39 | /// Engine-assigned tool call id, so a `tool_call_before` record and the |
| 40 | /// matching `tool_call_after` / `on_error` record can be correlated. |
| 41 | pub tool_call_id: Option<String>, |
| 42 | /// Tool arguments as JSON string |
| 43 | pub tool_args: Option<String>, |
| 44 | /// Tool result output (truncated) |
| 45 | pub tool_result: Option<String>, |
| 46 | /// Tool exit code if applicable. |
| 47 | /// |
| 48 | /// `i64` end-to-end: a Windows crash code such as `3221225477` |
| 49 | /// (`0xC0000005`) is a real value `exec_shell` reports, and narrowing it |
| 50 | /// to `i32` used to discard exactly the failures a hook most wants to see. |
| 51 | pub tool_exit_code: Option<i64>, |
| 52 | /// Whether tool succeeded |
| 53 | pub tool_success: Option<bool>, |
| 54 | /// Current mode |
| 55 | pub mode: Option<String>, |
| 56 | /// Previous mode (for `ModeChange`) |
| 57 | pub previous_mode: Option<String>, |
| 58 | /// Session ID |
| 59 | pub session_id: Option<String>, |
| 60 | /// User message content |
| 61 | pub message: Option<String>, |
| 62 | /// Error message (for `OnError`) |
| 63 | pub error_message: Option<String>, |
| 64 | /// Workspace path |
| 65 | pub workspace: Option<PathBuf>, |
| 66 | /// Current model name |
| 67 | pub model: Option<String>, |
| 68 | /// Total tokens used |
| 69 | pub total_tokens: Option<u32>, |
| 70 | /// Session cost in USD |
| 71 | pub session_cost: Option<f64>, |
| 72 | } |
| 73 | |
| 74 | impl HookContext { |
| 75 | pub fn new() -> Self { |
| 76 | Self::default() |
| 77 | } |
| 78 | |
| 79 | pub fn with_tool_name(mut self, name: &str) -> Self { |
| 80 | self.tool_name = Some(name.to_string()); |
| 81 | self |
| 82 | } |
| 83 | |
| 84 | pub fn with_tool_call_id(mut self, id: &str) -> Self { |
| 85 | self.tool_call_id = Some(id.to_string()); |
| 86 | self |
| 87 | } |
| 88 | |
| 89 | pub fn with_tool_args(mut self, args: &serde_json::Value) -> Self { |
| 90 | self.tool_args = Some(truncate_env_value( |
| 91 | &args.to_string(), |
| 92 | HOOK_TOOL_ARGS_ENV_MAX_BYTES, |
| 93 | )); |
| 94 | self |
| 95 | } |
| 96 | |
| 97 | pub fn with_tool_result(mut self, result: &str, success: bool, exit_code: Option<i64>) -> Self { |
| 98 | self.tool_result = Some(truncate_env_value( |
| 99 | result, |
| 100 | HOOK_TOOL_RESULT_CONTEXT_MAX_BYTES, |
| 101 | )); |
| 102 | self.tool_success = Some(success); |
| 103 | self.tool_exit_code = exit_code; |
| 104 | self |
| 105 | } |
| 106 | |
| 107 | pub fn with_mode(mut self, mode: &str) -> Self { |
| 108 | self.mode = Some(mode.to_string()); |
| 109 | self |
| 110 | } |
| 111 | |
| 112 | pub fn with_previous_mode(mut self, mode: &str) -> Self { |
| 113 | self.previous_mode = Some(mode.to_string()); |
| 114 | self |
| 115 | } |
| 116 | |
| 117 | pub fn with_workspace(mut self, path: PathBuf) -> Self { |
| 118 | self.workspace = Some(path); |
| 119 | self |
| 120 | } |
| 121 | |
| 122 | pub fn with_model(mut self, model: &str) -> Self { |
| 123 | self.model = Some(model.to_string()); |
| 124 | self |
| 125 | } |
| 126 | |
| 127 | pub fn with_session_id(mut self, session_id: &str) -> Self { |
| 128 | self.session_id = Some(session_id.to_string()); |
| 129 | self |
| 130 | } |
| 131 | |
| 132 | pub fn with_message(mut self, message: &str) -> Self { |
| 133 | self.message = Some(message.to_string()); |
| 134 | self |
| 135 | } |
| 136 | |
| 137 | pub fn with_error(mut self, error: &str) -> Self { |
| 138 | self.error_message = Some(truncate_env_value(error, HOOK_ERROR_CONTEXT_MAX_BYTES)); |
| 139 | self |
| 140 | } |
| 141 | |
| 142 | pub fn with_tokens(mut self, tokens: u32) -> Self { |
| 143 | self.total_tokens = Some(tokens); |
| 144 | self |
| 145 | } |
| 146 | |
| 147 | /// Clamp all observer-owned strings before the context is cloned into a |
| 148 | /// bounded queue. Builders already apply these limits, but fields remain |
| 149 | /// public for compatibility, so the submission boundary must defend |
| 150 | /// itself against a directly-constructed context too. |
| 151 | fn bounded_for_observer(mut self) -> Self { |
| 152 | fn bound(value: &mut Option<String>, max_bytes: usize) { |
| 153 | if let Some(raw) = value.take() { |
| 154 | *value = Some(truncate_env_value(&raw, max_bytes)); |
| 155 | } |
| 156 | } |
| 157 | |
| 158 | bound(&mut self.tool_name, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 159 | bound(&mut self.tool_call_id, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 160 | bound(&mut self.tool_args, HOOK_TOOL_ARGS_ENV_MAX_BYTES); |
| 161 | bound(&mut self.tool_result, HOOK_TOOL_RESULT_CONTEXT_MAX_BYTES); |
| 162 | bound(&mut self.mode, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 163 | bound(&mut self.previous_mode, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 164 | bound(&mut self.session_id, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 165 | bound(&mut self.message, HOOK_MESSAGE_CONTEXT_MAX_BYTES); |
| 166 | bound(&mut self.error_message, HOOK_ERROR_CONTEXT_MAX_BYTES); |
| 167 | bound(&mut self.model, HOOK_OBSERVER_METADATA_MAX_BYTES); |
| 168 | if let Some(workspace) = self.workspace.take() { |
| 169 | self.workspace = Some(PathBuf::from(truncate_env_value( |
| 170 | &workspace.to_string_lossy(), |
| 171 | HOOK_OBSERVER_METADATA_MAX_BYTES, |
| 172 | ))); |
| 173 | } |
| 174 | self |
| 175 | } |
| 176 | |
| 177 | /// Convert to environment variables |
| 178 | pub fn to_env_vars(&self) -> HashMap<String, String> { |
| 179 | let mut env = HashMap::new(); |
| 180 | |
| 181 | if let Some(ref name) = self.tool_name { |
| 182 | env.insert("DEEPSEEK_TOOL_NAME".to_string(), name.clone()); |
| 183 | } |
| 184 | if let Some(ref id) = self.tool_call_id { |
| 185 | env.insert("CODEWHALE_TOOL_CALL_ID".to_string(), id.clone()); |
| 186 | env.insert("DEEPSEEK_TOOL_CALL_ID".to_string(), id.clone()); |
| 187 | } |
| 188 | if let Some(ref args) = self.tool_args { |
| 189 | // Tool arguments can include whole patches or encoded payloads. |
| 190 | // Keep the diagnostic environment surface bounded just like tool |
| 191 | // results; hooks that need the canonical arguments already receive |
| 192 | // the structured tool request at the engine boundary. |
| 193 | env.insert( |
| 194 | "DEEPSEEK_TOOL_ARGS".to_string(), |
| 195 | truncate_env_value(args, HOOK_TOOL_ARGS_ENV_MAX_BYTES), |
| 196 | ); |
| 197 | } |
| 198 | if let Some(ref result) = self.tool_result { |
| 199 | // Truncate result to 10KB to avoid environment variable size limits |
| 200 | env.insert( |
| 201 | "DEEPSEEK_TOOL_RESULT".to_string(), |
| 202 | truncate_env_value(result, 10000), |
| 203 | ); |
| 204 | } |
| 205 | if let Some(code) = self.tool_exit_code { |
| 206 | env.insert("DEEPSEEK_TOOL_EXIT_CODE".to_string(), code.to_string()); |
| 207 | } |
| 208 | if let Some(success) = self.tool_success { |
| 209 | env.insert("DEEPSEEK_TOOL_SUCCESS".to_string(), success.to_string()); |
| 210 | } |
| 211 | if let Some(ref mode) = self.mode { |
| 212 | env.insert("DEEPSEEK_MODE".to_string(), mode.clone()); |
| 213 | } |
| 214 | if let Some(ref prev) = self.previous_mode { |
| 215 | env.insert("DEEPSEEK_PREVIOUS_MODE".to_string(), prev.clone()); |
| 216 | } |
| 217 | if let Some(ref session_id) = self.session_id { |
| 218 | env.insert("CODEWHALE_SESSION_ID".to_string(), session_id.clone()); |
| 219 | env.insert("DEEPSEEK_SESSION_ID".to_string(), session_id.clone()); |
| 220 | } |
| 221 | if let Some(ref message) = self.message { |
| 222 | // Truncate message to prevent env var issues |
| 223 | env.insert( |
| 224 | "DEEPSEEK_MESSAGE".to_string(), |
| 225 | truncate_env_value(message, 5000), |
| 226 | ); |
| 227 | } |
| 228 | if let Some(ref error) = self.error_message { |
| 229 | // Bounded like every other payload field: a tool failure message |
| 230 | // can be the whole of a failed command's output, and an unbounded |
| 231 | // env var is both an exec limit risk and an accidental transcript |
| 232 | // copy in whatever the hook writes it to. |
| 233 | env.insert( |
| 234 | "DEEPSEEK_ERROR".to_string(), |
| 235 | truncate_env_value(error, 5000), |
| 236 | ); |
| 237 | } |
| 238 | if let Some(ref ws) = self.workspace { |
| 239 | env.insert("DEEPSEEK_WORKSPACE".to_string(), ws.display().to_string()); |
| 240 | } |
| 241 | if let Some(ref model) = self.model { |
| 242 | env.insert("DEEPSEEK_MODEL".to_string(), model.clone()); |
| 243 | } |
| 244 | if let Some(tokens) = self.total_tokens { |
| 245 | env.insert("DEEPSEEK_TOTAL_TOKENS".to_string(), tokens.to_string()); |
| 246 | } |
| 247 | if let Some(cost) = self.session_cost { |
| 248 | env.insert("DEEPSEEK_SESSION_COST".to_string(), format!("{cost:.6}")); |
| 249 | } |
| 250 | |
| 251 | env |
| 252 | } |
| 253 | } |
| 254 | |
| 255 | /// Clamp a hook environment value to `max_bytes`, on a UTF-8 boundary, with a |
| 256 | /// visible marker so a hook can tell truncation from a short value. |
| 257 | fn truncate_env_value(value: &str, max_bytes: usize) -> String { |
| 258 | if value.len() <= max_bytes { |
| 259 | return value.to_string(); |
| 260 | } |
| 261 | let safe_end = value |
| 262 | .char_indices() |
| 263 | .take_while(|(i, c)| *i + c.len_utf8() <= max_bytes) |
| 264 | .last() |
| 265 | .map_or(0, |(i, c)| i + c.len_utf8()); |
| 266 | format!("{}...[truncated]", &value[..safe_end]) |
| 267 | } |
| 268 | |
| 269 | /// Result of a hook execution |
| 270 | #[derive(Debug, Clone, Default)] |
| 271 | pub struct HookResult { |
| 272 | /// Hook name (if specified) |
| 273 | pub name: Option<String>, |
| 274 | /// Whether the hook succeeded. |
| 275 | /// |
| 276 | /// For a background hook this is `true` as soon as the bounded supervisor |
| 277 | /// accepts the job: no child outcome has been observed yet. Check |
| 278 | /// [`Self::background`] before reading this as "the command succeeded". |
| 279 | pub success: bool, |
| 280 | /// `true` when this result describes a background submission rather than |
| 281 | /// a completed run. Background results always carry `exit_code: None`, |
| 282 | /// empty `stdout`/`stderr`, and a duration that measures the spawn, not |
| 283 | /// the command. |
| 284 | pub background: bool, |
| 285 | /// `true` when the hook behind this result declared |
| 286 | /// `continue_on_error = false` and ran in the foreground. |
| 287 | /// |
| 288 | /// This travels with the *result*, not with the event, because it is the |
| 289 | /// only way a steering call site can tell "the gate that actually matched |
| 290 | /// this call could not answer" from "some other, unrelated strict hook for |
| 291 | /// the same event exists in config". Background submissions are never |
| 292 | /// strict: nothing is awaited, so there is no answer to withhold. |
| 293 | pub strict: bool, |
| 294 | /// Exit code from the hook command |
| 295 | pub exit_code: Option<i32>, |
| 296 | /// Standard output |
| 297 | pub stdout: String, |
| 298 | /// Standard error |
| 299 | #[allow(dead_code)] // written by prod constructors, read only in tests |
| 300 | pub stderr: String, |
| 301 | /// Time taken to execute |
| 302 | pub duration: Duration, |
| 303 | /// Error message if execution failed |
| 304 | pub error: Option<String>, |
| 305 | } |
| 306 | |
| 307 | impl HookResult { |
| 308 | /// A result that carries an observed exit code, as opposed to a |
| 309 | /// background submission or a spawn failure. |
| 310 | /// |
| 311 | /// Steering paths must gate on this: a background hook's `exit_code` is |
| 312 | /// `None` because nothing was waited for, not because the command exited |
| 313 | /// without a code. |
| 314 | #[must_use] |
| 315 | pub fn observed_exit_code(&self) -> Option<i32> { |
| 316 | if self.background { |
| 317 | return None; |
| 318 | } |
| 319 | self.exit_code |
| 320 | } |
| 321 | } |
| 322 | |
| 323 | /// Result of running mutable `message_submit` hooks. |
| 324 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 325 | pub enum MessageSubmitOutcome { |
| 326 | /// No hook changed the submitted text. |
| 327 | Unchanged { warning: Option<String> }, |
| 328 | /// One or more hooks replaced the submitted text. |
| 329 | Replaced { |
| 330 | text: String, |
| 331 | warning: Option<String>, |
| 332 | }, |
| 333 | /// A hook intentionally blocked the submission. |
| 334 | Blocked { reason: String }, |
| 335 | } |
| 336 | |
| 337 | impl MessageSubmitOutcome { |
| 338 | pub fn unchanged() -> Self { |
| 339 | Self::Unchanged { warning: None } |
| 340 | } |
| 341 | |
| 342 | pub fn replaced(text: String) -> Self { |
| 343 | Self::Replaced { |
| 344 | text, |
| 345 | warning: None, |
| 346 | } |
| 347 | } |
| 348 | |
| 349 | fn with_warning(self, warning: Option<String>) -> Self { |
| 350 | match self { |
| 351 | Self::Unchanged { .. } => Self::Unchanged { warning }, |
| 352 | Self::Replaced { text, .. } => Self::Replaced { text, warning }, |
| 353 | Self::Blocked { reason } => Self::Blocked { reason }, |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | pub fn warning(&self) -> Option<&str> { |
| 358 | match self { |
| 359 | Self::Unchanged { warning } | Self::Replaced { warning, .. } => warning.as_deref(), |
| 360 | Self::Blocked { .. } => None, |
| 361 | } |
| 362 | } |
| 363 | } |
| 364 | |
| 365 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 366 | enum MessageSubmitStdout { |
| 367 | Unchanged, |
| 368 | Replaced(String), |
| 369 | Invalid(String), |
| 370 | } |
| 371 | |
| 372 | /// Maximum characters kept from one text field a `tool_call_before` hook |
| 373 | /// prints (`reason`, `additionalContext`). |
| 374 | /// |
| 375 | /// Both fields end up somewhere unbounded output would be a real problem: |
| 376 | /// `reason` in a TUI denial line, `additionalContext` inside the tool result |
| 377 | /// that is sent to the model and counted against the context budget. A hook |
| 378 | /// that prints a megabyte gets a bounded, marked prefix instead. |
| 379 | pub(crate) const HOOK_TEXT_FIELD_MAX_CHARS: usize = 2_000; |
| 380 | |
| 381 | /// Maximum characters of concatenated `additionalContext` appended to a single |
| 382 | /// tool result, across every hook that contributed to that one call. |
| 383 | pub(crate) const HOOK_CONTEXT_AGGREGATE_MAX_CHARS: usize = 8_000; |
| 384 | |
| 385 | /// Largest tool-argument snapshot exported through `DEEPSEEK_TOOL_ARGS`. |
| 386 | const HOOK_TOOL_ARGS_ENV_MAX_BYTES: usize = 10_000; |
| 387 | |
| 388 | /// Largest raw tool result retained in an observer job before enqueue. |
| 389 | const HOOK_TOOL_RESULT_CONTEXT_MAX_BYTES: usize = 10_000; |
| 390 | |
| 391 | /// Largest error retained in an observer job before enqueue. |
| 392 | const HOOK_ERROR_CONTEXT_MAX_BYTES: usize = 5_000; |
| 393 | |
| 394 | /// Largest user/message preview retained in an observer job before enqueue. |
| 395 | const HOOK_MESSAGE_CONTEXT_MAX_BYTES: usize = 5_000; |
| 396 | |
| 397 | /// Largest identifier or other diagnostic retained in an observer job. |
| 398 | const HOOK_OBSERVER_METADATA_MAX_BYTES: usize = 4_096; |
| 399 | |
| 400 | /// Largest stdout or stderr prefix retained from one foreground hook. Reader |
| 401 | /// threads continue draining after this cap so a verbose child cannot fill its |
| 402 | /// pipe and deadlock before exit; only the in-memory receipt is clipped. |
| 403 | const HOOK_PIPE_CAPTURE_MAX_BYTES: usize = 64 * 1024; |
| 404 | |
| 405 | /// Largest serialized `updatedInput` object accepted from a decision hook. |
| 406 | /// This is intentionally smaller than the pipe cap so the surrounding JSON |
| 407 | /// and other fields still have headroom. |
| 408 | const HOOK_UPDATED_INPUT_MAX_BYTES: usize = 32 * 1024; |
| 409 | |
| 410 | /// Largest replacement message accepted from `message_submit`. |
| 411 | const HOOK_MESSAGE_REPLACEMENT_MAX_CHARS: usize = 32_000; |
| 412 | |
| 413 | /// Hard ceiling for the complete serialized `message_submit` stdin document. |
| 414 | /// The text prefix is fitted beneath this boundary after bounded metadata has |
| 415 | /// been added, so JSON escaping cannot push a producer past the limit. |
| 416 | pub(crate) const HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES: usize = 32 * 1024; |
| 417 | |
| 418 | /// Individual metadata fields in `message_submit` stdin are diagnostic only. |
| 419 | /// Bound them before fitting text so an unusual workspace/model value cannot |
| 420 | /// consume the entire payload budget. |
| 421 | const HOOK_MESSAGE_SUBMIT_METADATA_MAX_BYTES: usize = 4 * 1024; |
| 422 | |
| 423 | /// Largest turn error copied into a `turn_end` observer payload. |
| 424 | const HOOK_TURN_ERROR_MAX_CHARS: usize = 2_000; |
| 425 | |
| 426 | /// Largest denial reason persisted into UI/model receipts. |
| 427 | const HOOK_DENIAL_RECEIPT_MAX_CHARS: usize = 240; |
| 428 | |
| 429 | /// Bound and de-fang text a hook printed before it is shown or sent onward. |
| 430 | /// |
| 431 | /// Control characters are removed (`\r`) or flattened to a space so hook |
| 432 | /// stdout cannot repaint the TUI with escape sequences or forge structure in |
| 433 | /// the model-facing transcript; `\n` and `\t` survive because a hook's context |
| 434 | /// is legitimately multi-line. Truncation carries a visible marker so a |
| 435 | /// consumer can tell a clipped value from a short one. |
| 436 | pub(crate) fn sanitize_hook_text(text: &str, max_chars: usize) -> String { |
| 437 | let mut out = String::new(); |
| 438 | let mut kept = 0usize; |
| 439 | let mut truncated = false; |
| 440 | for ch in text.chars() { |
| 441 | let mapped = match ch { |
| 442 | '\n' | '\t' => ch, |
| 443 | '\r' => continue, |
| 444 | c if c.is_control() => ' ', |
| 445 | c => c, |
| 446 | }; |
| 447 | if kept == max_chars { |
| 448 | truncated = true; |
| 449 | break; |
| 450 | } |
| 451 | out.push(mapped); |
| 452 | kept += 1; |
| 453 | } |
| 454 | if truncated { |
| 455 | out.push_str("…[truncated]"); |
| 456 | } |
| 457 | out |
| 458 | } |
| 459 | |
| 460 | /// Longest hook/config name kept in a log line, a `/hooks` row, or a receipt. |
| 461 | /// |
| 462 | /// Names are operator-supplied and otherwise unbounded: nothing stops a |
| 463 | /// `name` from being a megabyte of ANSI escapes, and it is echoed into the |
| 464 | /// TUI, the tracing stream, and the model-facing denial. |
| 465 | pub(crate) const HOOK_LABEL_MAX_CHARS: usize = 64; |
| 466 | |
| 467 | /// [`sanitize_hook_text`], forced onto one line. |
| 468 | /// |
| 469 | /// Labels and previews sit inside a formatted row, so an embedded newline or |
| 470 | /// tab would forge structure in the very listing that is supposed to describe |
| 471 | /// the hook. Everything else [`sanitize_hook_text`] does — control-character |
| 472 | /// removal and the marked truncation — still applies. |
| 473 | pub(crate) fn sanitize_hook_line(text: &str, max_chars: usize) -> String { |
| 474 | sanitize_hook_text(text, max_chars) |
| 475 | .chars() |
| 476 | .map(|c| if c == '\n' || c == '\t' { ' ' } else { c }) |
| 477 | .collect() |
| 478 | } |
| 479 | |
| 480 | /// The display label for a hook, from its optional operator-supplied `name`. |
| 481 | /// |
| 482 | /// One line, bounded, control-free, and never empty — every surface that |
| 483 | /// prints a hook name (logs, `/hooks list`, config problems, no-verdict |
| 484 | /// receipts) goes through here so there is one answer to "what can a `name` |
| 485 | /// put on my screen". |
| 486 | pub(crate) fn sanitize_hook_label(name: Option<&str>) -> String { |
| 487 | let cleaned = name |
| 488 | .map(|name| sanitize_hook_line(name, HOOK_LABEL_MAX_CHARS)) |
| 489 | .unwrap_or_default(); |
| 490 | if cleaned.trim().is_empty() { |
| 491 | "(unnamed)".to_string() |
| 492 | } else { |
| 493 | cleaned.trim().to_string() |
| 494 | } |
| 495 | } |
| 496 | |
| 497 | #[derive(Clone, Copy)] |
| 498 | enum PendingDenialRedaction { |
| 499 | AuthorizationSchemeOrCredential, |
| 500 | SecretValue, |
| 501 | Command, |
| 502 | Path, |
| 503 | } |
| 504 | |
| 505 | /// Split a denial into whitespace-delimited fields while keeping quoted |
| 506 | /// values together. This makes `command="rm -rf"` and |
| 507 | /// `path='/private folder'` one redaction unit even though the value contains |
| 508 | /// spaces. Unterminated quotes are conservatively kept in the final field. |
| 509 | fn denial_fields(line: &str) -> Vec<String> { |
| 510 | let mut fields = Vec::new(); |
| 511 | let mut current = String::new(); |
| 512 | let mut quote = None; |
| 513 | for ch in line.chars() { |
| 514 | match (quote, ch) { |
| 515 | (None, '\'' | '"') => { |
| 516 | quote = Some(ch); |
| 517 | current.push(ch); |
| 518 | } |
| 519 | (Some(open), close) if open == close => { |
| 520 | quote = None; |
| 521 | current.push(ch); |
| 522 | } |
| 523 | (None, ch) if ch.is_whitespace() => { |
| 524 | if !current.is_empty() { |
| 525 | fields.push(std::mem::take(&mut current)); |
| 526 | } |
| 527 | } |
| 528 | _ => current.push(ch), |
| 529 | } |
| 530 | } |
| 531 | if !current.is_empty() { |
| 532 | fields.push(current); |
| 533 | } |
| 534 | fields |
| 535 | } |
| 536 | |
| 537 | fn denial_field_core(field: &str) -> &str { |
| 538 | field.trim_matches(|ch: char| { |
| 539 | matches!( |
| 540 | ch, |
| 541 | '\'' | '"' | '`' | '(' | ')' | '[' | ']' | '{' | '}' | ',' | ';' |
| 542 | ) |
| 543 | }) |
| 544 | } |
| 545 | |
| 546 | fn denial_sensitive_assignment(field: &str) -> Option<(&str, &str)> { |
| 547 | let core = denial_field_core(field); |
| 548 | let separator = core.find([':', '='])?; |
| 549 | let key = denial_field_core(&core[..separator]); |
| 550 | let value = denial_field_core(&core[separator + 1..]); |
| 551 | Some((key, value)) |
| 552 | } |
| 553 | |
| 554 | fn normalized_denial_key(key: &str) -> String { |
| 555 | denial_field_core(key) |
| 556 | .chars() |
| 557 | .map(|ch| match ch { |
| 558 | '-' | '.' => '_', |
| 559 | ch => ch.to_ascii_lowercase(), |
| 560 | }) |
| 561 | .collect() |
| 562 | } |
| 563 | |
| 564 | fn denial_key_is_secret(key: &str) -> bool { |
| 565 | matches!( |
| 566 | key, |
| 567 | "token" |
| 568 | | "secret" |
| 569 | | "password" |
| 570 | | "passwd" |
| 571 | | "api_key" |
| 572 | | "apikey" |
| 573 | | "authorization" |
| 574 | | "bearer" |
| 575 | ) || key.ends_with("_api_key") |
| 576 | || key.ends_with("_token") |
| 577 | || key.ends_with("_secret") |
| 578 | } |
| 579 | |
| 580 | /// Render an explicit hook denial without carrying raw process output into a |
| 581 | /// durable transcript. Structured reasons are useful operator copy, but they |
| 582 | /// still pass through a conservative redaction boundary: path-like tokens, |
| 583 | /// command-line flags, and common secret assignments are replaced rather than |
| 584 | /// persisted. Unstructured stdout/stderr never reaches this function. |
| 585 | pub(crate) fn sanitize_hook_denial_reason(reason: &str) -> String { |
| 586 | let line = sanitize_hook_line(reason, HOOK_DENIAL_RECEIPT_MAX_CHARS); |
| 587 | let mut redacted = Vec::new(); |
| 588 | let mut pending = None; |
| 589 | for field in denial_fields(&line) { |
| 590 | let core = denial_field_core(&field); |
| 591 | let lower = core.to_ascii_lowercase(); |
| 592 | |
| 593 | if matches!(core, "=" | ":") { |
| 594 | continue; |
| 595 | } |
| 596 | |
| 597 | if let Some(expected) = pending { |
| 598 | match expected { |
| 599 | PendingDenialRedaction::AuthorizationSchemeOrCredential => { |
| 600 | redacted.push("[secret]".to_string()); |
| 601 | // Authorization uses `scheme credentials`. Treat the |
| 602 | // first field as a scheme even when it is proprietary; |
| 603 | // over-redacting one following field is safer than |
| 604 | // leaking a credential for a scheme we do not know. |
| 605 | pending = Some(PendingDenialRedaction::SecretValue); |
| 606 | } |
| 607 | PendingDenialRedaction::SecretValue => { |
| 608 | redacted.push("[secret]".to_string()); |
| 609 | pending = None; |
| 610 | } |
| 611 | PendingDenialRedaction::Command => { |
| 612 | redacted.push("[command]".to_string()); |
| 613 | pending = None; |
| 614 | } |
| 615 | PendingDenialRedaction::Path => { |
| 616 | redacted.push("[path]".to_string()); |
| 617 | pending = None; |
| 618 | } |
| 619 | } |
| 620 | continue; |
| 621 | } |
| 622 | |
| 623 | if let Some((key, value)) = denial_sensitive_assignment(&field) { |
| 624 | let key = normalized_denial_key(key); |
| 625 | if denial_key_is_secret(&key) { |
| 626 | redacted.push("[secret]".to_string()); |
| 627 | pending = if key == "authorization" && value.is_empty() { |
| 628 | Some(PendingDenialRedaction::AuthorizationSchemeOrCredential) |
| 629 | } else if key == "authorization" |
| 630 | && !value.chars().any(char::is_whitespace) |
| 631 | && !value.contains(':') |
| 632 | { |
| 633 | // A lone assignment value is normally the scheme |
| 634 | // (`Authorization=Digest <credential>`). Quoted values |
| 635 | // containing whitespace already include both pieces. |
| 636 | Some(PendingDenialRedaction::SecretValue) |
| 637 | } else if value.is_empty() { |
| 638 | Some(PendingDenialRedaction::SecretValue) |
| 639 | } else { |
| 640 | None |
| 641 | }; |
| 642 | continue; |
| 643 | } |
| 644 | if matches!( |
| 645 | key.as_str(), |
| 646 | "path" | "file" | "directory" | "cwd" | "workspace" |
| 647 | ) { |
| 648 | redacted.push("[path]".to_string()); |
| 649 | pending = value.is_empty().then_some(PendingDenialRedaction::Path); |
| 650 | continue; |
| 651 | } |
| 652 | if matches!(key.as_str(), "command" | "cmd" | "argv" | "executable") { |
| 653 | redacted.push("[command]".to_string()); |
| 654 | pending = value.is_empty().then_some(PendingDenialRedaction::Command); |
| 655 | continue; |
| 656 | } |
| 657 | } |
| 658 | |
| 659 | let secret_prefix = lower.starts_with("sk-") |
| 660 | || lower.starts_with("ghp_") |
| 661 | || lower.starts_with("github_pat_"); |
| 662 | let path_like = core.starts_with('/') |
| 663 | || core.starts_with("~/") |
| 664 | || core.starts_with("./") |
| 665 | || core.starts_with("../") |
| 666 | || core.contains('/') |
| 667 | || core.contains('\\') |
| 668 | || core |
| 669 | .as_bytes() |
| 670 | .get(1) |
| 671 | .is_some_and(|separator| *separator == b':'); |
| 672 | let command_flag = core.starts_with('-'); |
| 673 | let label = lower.trim_end_matches([':', '=']); |
| 674 | if matches!(label, "command" | "cmd" | "argv" | "executable") { |
| 675 | redacted.push("[command]".to_string()); |
| 676 | pending = Some(PendingDenialRedaction::Command); |
| 677 | } else if label == "authorization" { |
| 678 | redacted.push("[secret]".to_string()); |
| 679 | pending = Some(PendingDenialRedaction::AuthorizationSchemeOrCredential); |
| 680 | } else if matches!(label, "bearer" | "token" | "secret" | "password" | "passwd") { |
| 681 | redacted.push("[secret]".to_string()); |
| 682 | pending = Some(PendingDenialRedaction::SecretValue); |
| 683 | } else if matches!(label, "path" | "file" | "directory" | "cwd" | "workspace") { |
| 684 | redacted.push("[path]".to_string()); |
| 685 | pending = Some(PendingDenialRedaction::Path); |
| 686 | } else if secret_prefix { |
| 687 | redacted.push("[secret]".to_string()); |
| 688 | } else if path_like { |
| 689 | redacted.push("[path]".to_string()); |
| 690 | } else if command_flag { |
| 691 | redacted.push("[argument]".to_string()); |
| 692 | } else { |
| 693 | redacted.push(field); |
| 694 | } |
| 695 | } |
| 696 | let rendered = sanitize_hook_line(&redacted.join(" "), HOOK_DENIAL_RECEIPT_MAX_CHARS); |
| 697 | if rendered.is_empty() { |
| 698 | "hook denied the action".to_string() |
| 699 | } else { |
| 700 | rendered |
| 701 | } |
| 702 | } |
| 703 | |
| 704 | /// Render a foreground hook's failure as a detail string that is safe to show. |
| 705 | /// |
| 706 | /// The executor already writes generic errors, but this is the *boundary*, not |
| 707 | /// a restatement of that habit: only the shapes recognized here survive, and |
| 708 | /// each is re-rendered from parts rather than passed through. A future code |
| 709 | /// path that stuffs a command line, a resolved interpreter path, or hook |
| 710 | /// output into `HookResult::error` therefore cannot leak it into a receipt |
| 711 | /// merely by skipping the genericization at the producer — it degrades to the |
| 712 | /// catch-all instead, and the raw text is discarded. |
| 713 | pub(crate) fn generic_unavailable_detail(error: Option<&str>) -> String { |
| 714 | const GENERIC: &str = "hook returned no verdict"; |
| 715 | let Some(error) = error else { |
| 716 | return GENERIC.to_string(); |
| 717 | }; |
| 718 | if let Some(rest) = error.strip_prefix("Hook timed out after ") { |
| 719 | let secs: String = rest.chars().take_while(char::is_ascii_digit).collect(); |
| 720 | return if secs.is_empty() { |
| 721 | "hook timed out".to_string() |
| 722 | } else { |
| 723 | format!("hook timed out after {secs}s") |
| 724 | }; |
| 725 | } |
| 726 | if let Some(rest) = error.strip_prefix("hook process could not be started (") { |
| 727 | // Only the `std::io::ErrorKind` debug name, and only if it really is |
| 728 | // one: bare ASCII letters, nothing else. |
| 729 | let kind: String = rest.chars().take_while(char::is_ascii_alphabetic).collect(); |
| 730 | return if kind.is_empty() { |
| 731 | "hook process could not be started".to_string() |
| 732 | } else { |
| 733 | format!("hook process could not be started ({kind})") |
| 734 | }; |
| 735 | } |
| 736 | if error.starts_with("failed to contain hook process tree") |
| 737 | || error.starts_with("failed to resume contained hook process") |
| 738 | { |
| 739 | return "hook process could not be contained".to_string(); |
| 740 | } |
| 741 | if error.starts_with("hook executor did not run") { |
| 742 | return "hook executor did not run".to_string(); |
| 743 | } |
| 744 | if error.starts_with("Failed to submit background hook") |
| 745 | || error.starts_with("background hook supervisor could not be started") |
| 746 | || error.starts_with("background hook supervisor queue is full") |
| 747 | || error.starts_with("background hook supervisor is unavailable") |
| 748 | { |
| 749 | return "hook could not be submitted".to_string(); |
| 750 | } |
| 751 | if error.starts_with("Failed to wait for hook") |
| 752 | || error.starts_with("hook could not be reaped") |
| 753 | || error.starts_with("Failed to encode hook stdin") |
| 754 | || error.starts_with("hook stdout reader could not be started") |
| 755 | || error.starts_with("hook stderr reader could not be started") |
| 756 | || error.starts_with("hook stdin writer could not be started") |
| 757 | || error.starts_with("background hook process could not be started") |
| 758 | || error.starts_with("background hook stdin writer could not be started") |
| 759 | || error.starts_with("background hook setup") |
| 760 | { |
| 761 | return "hook did not complete cleanly".to_string(); |
| 762 | } |
| 763 | tracing::debug!(target: "hooks", "hook failure had no recognized shape; reporting it generically"); |
| 764 | GENERIC.to_string() |
| 765 | } |
| 766 | |
| 767 | /// [`sanitize_hook_text`], dropping the value entirely when nothing |
| 768 | /// meaningful survives. |
| 769 | fn sanitized_hook_field(text: &str) -> Option<String> { |
| 770 | let cleaned = sanitize_hook_text(text, HOOK_TEXT_FIELD_MAX_CHARS); |
| 771 | if cleaned.trim().is_empty() { |
| 772 | None |
| 773 | } else { |
| 774 | Some(cleaned) |
| 775 | } |
| 776 | } |
| 777 | |
| 778 | /// Parsed stdout from a `tool_call_before` hook (#3026). |
| 779 | /// |
| 780 | /// Hooks may emit a JSON decision on stdout: |
| 781 | /// `{"decision": "allow"|"deny"|"ask", "reason": "...", |
| 782 | /// "updatedInput": {...}, "additionalContext": "..."}` |
| 783 | /// Non-JSON or empty stdout → legacy passthrough (allow). |
| 784 | /// |
| 785 | /// `reason` and `additional_context` are sanitized and bounded here, at the |
| 786 | /// only door hook stdout comes through, so no downstream consumer has to |
| 787 | /// remember to do it. |
| 788 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 789 | pub struct ToolCallBeforeStdout { |
| 790 | pub decision: Option<ToolCallDecision>, |
| 791 | pub reason: Option<String>, |
| 792 | pub updated_input: Option<serde_json::Value>, |
| 793 | pub additional_context: Option<String>, |
| 794 | } |
| 795 | |
| 796 | /// Decision a hook can return for a tool call. |
| 797 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 798 | pub enum ToolCallDecision { |
| 799 | Allow, |
| 800 | Deny, |
| 801 | Ask, |
| 802 | } |
| 803 | |
| 804 | pub(crate) fn parse_tool_call_before_stdout(stdout: &str) -> ToolCallBeforeStdout { |
| 805 | let passthrough = ToolCallBeforeStdout { |
| 806 | decision: None, |
| 807 | reason: None, |
| 808 | updated_input: None, |
| 809 | additional_context: None, |
| 810 | }; |
| 811 | let trimmed = stdout.trim(); |
| 812 | if trimmed.is_empty() { |
| 813 | return passthrough; |
| 814 | } |
| 815 | let value: serde_json::Value = match serde_json::from_str(trimmed) { |
| 816 | Ok(v) => v, |
| 817 | // Non-JSON stdout → legacy passthrough (allow). |
| 818 | Err(_) => return passthrough, |
| 819 | }; |
| 820 | let Some(obj) = value.as_object() else { |
| 821 | tracing::warn!( |
| 822 | "tool_call_before hook stdout is JSON but not an object; \ |
| 823 | ignoring it (legacy passthrough)" |
| 824 | ); |
| 825 | return passthrough; |
| 826 | }; |
| 827 | let decision = obj |
| 828 | .get("decision") |
| 829 | .and_then(|v| v.as_str()) |
| 830 | .and_then(|s| match s { |
| 831 | "allow" => Some(ToolCallDecision::Allow), |
| 832 | "deny" => Some(ToolCallDecision::Deny), |
| 833 | "ask" => Some(ToolCallDecision::Ask), |
| 834 | _ => { |
| 835 | tracing::warn!( |
| 836 | "tool_call_before hook returned unrecognized decision \ |
| 837 | (expected allow|deny|ask); treating as allow" |
| 838 | ); |
| 839 | None |
| 840 | } |
| 841 | }); |
| 842 | let reason = obj |
| 843 | .get("reason") |
| 844 | .and_then(|v| v.as_str()) |
| 845 | .and_then(sanitized_hook_field); |
| 846 | let updated_input = obj.get("updatedInput").cloned().filter(|v| { |
| 847 | if !v.is_object() { |
| 848 | tracing::warn!("tool_call_before hook updatedInput must be a JSON object; ignoring"); |
| 849 | return false; |
| 850 | } |
| 851 | let serialized_len = serde_json::to_vec(v).map_or(usize::MAX, |bytes| bytes.len()); |
| 852 | if serialized_len > HOOK_UPDATED_INPUT_MAX_BYTES { |
| 853 | tracing::warn!( |
| 854 | serialized_len, |
| 855 | max_bytes = HOOK_UPDATED_INPUT_MAX_BYTES, |
| 856 | "tool_call_before hook updatedInput exceeded the size limit; ignoring" |
| 857 | ); |
| 858 | return false; |
| 859 | } |
| 860 | true |
| 861 | }); |
| 862 | let additional_context = obj |
| 863 | .get("additionalContext") |
| 864 | .and_then(|v| v.as_str()) |
| 865 | .and_then(sanitized_hook_field); |
| 866 | ToolCallBeforeStdout { |
| 867 | decision, |
| 868 | reason, |
| 869 | updated_input, |
| 870 | additional_context, |
| 871 | } |
| 872 | } |
| 873 | |
| 874 | /// Post-turn accumulated totals included in the `turn_end` observer payload. |
| 875 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 876 | pub struct TurnEndTotals { |
| 877 | pub session_tokens: u32, |
| 878 | pub conversation_tokens: u32, |
| 879 | pub input_tokens: u32, |
| 880 | pub output_tokens: u32, |
| 881 | } |
| 882 | |
| 883 | /// Input used to build the structured `turn_end` observer payload. |
| 884 | pub struct TurnEndPayloadInput<'a> { |
| 885 | pub context: &'a HookContext, |
| 886 | pub created_at: DateTime<Utc>, |
| 887 | pub model_backed: bool, |
| 888 | pub provider: Option<&'a str>, |
| 889 | pub billing_surface: Option<&'a str>, |
| 890 | pub model: Option<&'a str>, |
| 891 | pub turn_id: &'a str, |
| 892 | pub status: &'a str, |
| 893 | pub error: Option<&'a str>, |
| 894 | pub duration: Duration, |
| 895 | pub usage: &'a codewhale_models::Usage, |
| 896 | pub totals: TurnEndTotals, |
| 897 | pub tool_count: usize, |
| 898 | pub queued_message_count: usize, |
| 899 | } |
| 900 | |
| 901 | /// Owns the process tree created for one hook invocation. |
| 902 | /// |
| 903 | /// Hooks run through a shell, so killing only the immediate `sh`/`cmd.exe` |
| 904 | /// child can leave the actual hook runtime alive. Unix hooks get their own |
| 905 | /// process group and Windows hooks are attached to a kill-on-close Job Object. |
| 906 | /// Dropping this guard after the shell exits also closes inherited stdout and |
| 907 | /// stderr pipes held by any lingering descendants. |
| 908 | struct HookProcessTree { |
| 909 | #[cfg(unix)] |
| 910 | pgid: libc::pid_t, |
| 911 | #[cfg(windows)] |
| 912 | job: WindowsHookJob, |
| 913 | } |
| 914 | |
| 915 | impl HookProcessTree { |
| 916 | fn attach(child: &Child) -> std::io::Result<Self> { |
| 917 | #[cfg(unix)] |
| 918 | { |
| 919 | Ok(Self { |
| 920 | pgid: child.id() as libc::pid_t, |
| 921 | }) |
| 922 | } |
| 923 | |
| 924 | #[cfg(windows)] |
| 925 | { |
| 926 | Ok(Self { |
| 927 | job: WindowsHookJob::attach(child)?, |
| 928 | }) |
| 929 | } |
| 930 | |
| 931 | #[cfg(not(any(unix, windows)))] |
| 932 | { |
| 933 | Ok(Self {}) |
| 934 | } |
| 935 | } |
| 936 | |
| 937 | fn terminate(&self, child: &mut Child) { |
| 938 | #[cfg(unix)] |
| 939 | { |
| 940 | // SAFETY: kill(2) dereferences no pointers. |
| 941 | let result = unsafe { libc::kill(-self.pgid, libc::SIGKILL) }; |
| 942 | if result != 0 { |
| 943 | let error = std::io::Error::last_os_error(); |
| 944 | if error.raw_os_error() != Some(libc::ESRCH) { |
| 945 | tracing::warn!(?error, "failed to terminate hook process group"); |
| 946 | let _ = child.kill(); |
| 947 | } |
| 948 | } |
| 949 | } |
| 950 | |
| 951 | #[cfg(windows)] |
| 952 | { |
| 953 | let result = self |
| 954 | .job |
| 955 | .terminate() |
| 956 | .or_else(|_| kill_windows_process_tree(child.id())); |
| 957 | if let Err(error) = result { |
| 958 | tracing::warn!( |
| 959 | ?error, |
| 960 | "failed to terminate hook process tree; killing immediate child" |
| 961 | ); |
| 962 | let _ = child.kill(); |
| 963 | } |
| 964 | } |
| 965 | |
| 966 | #[cfg(not(any(unix, windows)))] |
| 967 | { |
| 968 | let _ = child.kill(); |
| 969 | } |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | impl Drop for HookProcessTree { |
| 974 | fn drop(&mut self) { |
| 975 | #[cfg(unix)] |
| 976 | // SAFETY: kill(2) dereferences no pointers. |
| 977 | unsafe { |
| 978 | // The shell may have exited while one of its descendants still |
| 979 | // holds a captured pipe. Reaping the process group keeps hook |
| 980 | // lifetimes bounded and lets the reader threads finish. |
| 981 | let _ = libc::kill(-self.pgid, libc::SIGKILL); |
| 982 | } |
| 983 | // On Windows, dropping WindowsHookJob closes a Job Object configured |
| 984 | // with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE. |
| 985 | } |
| 986 | } |
| 987 | |
| 988 | #[cfg(windows)] |
| 989 | struct WindowsHookJob { |
| 990 | handle: HANDLE, |
| 991 | } |
| 992 | |
| 993 | #[cfg(windows)] |
| 994 | impl WindowsHookJob { |
| 995 | fn attach(child: &Child) -> std::io::Result<Self> { |
| 996 | // SAFETY: returned handle is owned by the new wrapper. |
| 997 | let handle = unsafe { CreateJobObjectW(None, PCWSTR::null()).map_err(windows_io_error)? }; |
| 998 | let job = Self { handle }; |
| 999 | let mut limits = JOBOBJECT_EXTENDED_LIMIT_INFORMATION::default(); |
| 1000 | limits.BasicLimitInformation.LimitFlags = JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE; |
| 1001 | |
| 1002 | // SAFETY: `limits` is live with matching size; both handles are live. |
| 1003 | unsafe { |
| 1004 | SetInformationJobObject( |
| 1005 | job.handle, |
| 1006 | JobObjectExtendedLimitInformation, |
| 1007 | &limits as *const _ as *const core::ffi::c_void, |
| 1008 | std::mem::size_of::<JOBOBJECT_EXTENDED_LIMIT_INFORMATION>() as u32, |
| 1009 | ) |
| 1010 | .map_err(windows_io_error)?; |
| 1011 | AssignProcessToJobObject(job.handle, HANDLE(child.as_raw_handle())) |
| 1012 | .map_err(windows_io_error)?; |
| 1013 | } |
| 1014 | Ok(job) |
| 1015 | } |
| 1016 | |
| 1017 | fn terminate(&self) -> std::io::Result<()> { |
| 1018 | // SAFETY: `self.handle` is a live owned job handle. |
| 1019 | unsafe { TerminateJobObject(self.handle, 1).map_err(windows_io_error) } |
| 1020 | } |
| 1021 | } |
| 1022 | |
| 1023 | #[cfg(windows)] |
| 1024 | impl Drop for WindowsHookJob { |
| 1025 | fn drop(&mut self) { |
| 1026 | // SAFETY: `self.handle` is owned here; Drop runs once. |
| 1027 | unsafe { |
| 1028 | let _ = CloseHandle(self.handle); |
| 1029 | } |
| 1030 | } |
| 1031 | } |
| 1032 | |
| 1033 | #[cfg(windows)] |
| 1034 | fn windows_io_error(error: windows::core::Error) -> std::io::Error { |
| 1035 | std::io::Error::other(error) |
| 1036 | } |
| 1037 | |
| 1038 | #[cfg(windows)] |
| 1039 | fn resume_windows_process(child: &Child) -> std::io::Result<()> { |
| 1040 | let snapshot = |
| 1041 | // SAFETY: returned handle is owned here; closed before return. |
| 1042 | unsafe { CreateToolhelp32Snapshot(TH32CS_SNAPTHREAD, 0).map_err(windows_io_error)? }; |
| 1043 | let result = (|| { |
| 1044 | let mut entry = THREADENTRY32 { |
| 1045 | dwSize: std::mem::size_of::<THREADENTRY32>() as u32, |
| 1046 | ..Default::default() |
| 1047 | }; |
| 1048 | // SAFETY: `entry` is live with dwSize initialized above. |
| 1049 | let mut next = unsafe { Thread32First(snapshot, &mut entry) }; |
| 1050 | let mut resumed = 0usize; |
| 1051 | while next.is_ok() { |
| 1052 | if entry.th32OwnerProcessID == child.id() { |
| 1053 | // SAFETY: returned handle is owned here; closed below. |
| 1054 | let thread = unsafe { |
| 1055 | OpenThread(THREAD_SUSPEND_RESUME, false, entry.th32ThreadID) |
| 1056 | .map_err(windows_io_error)? |
| 1057 | }; |
| 1058 | // SAFETY: `thread` is a live owned handle. |
| 1059 | let resume_result = unsafe { ResumeThread(thread) }; |
| 1060 | // SAFETY: `thread` is owned here and not used after. |
| 1061 | let close_result = unsafe { CloseHandle(thread).map_err(windows_io_error) }; |
| 1062 | if resume_result == u32::MAX { |
| 1063 | return Err(std::io::Error::last_os_error()); |
| 1064 | } |
| 1065 | close_result?; |
| 1066 | resumed += 1; |
| 1067 | } |
| 1068 | // SAFETY: `entry` is live with dwSize initialized above. |
| 1069 | next = unsafe { Thread32Next(snapshot, &mut entry) }; |
| 1070 | } |
| 1071 | if resumed == 0 { |
| 1072 | return Err(std::io::Error::other( |
| 1073 | "suspended hook process had no resumable thread", |
| 1074 | )); |
| 1075 | } |
| 1076 | Ok(()) |
| 1077 | })(); |
| 1078 | // SAFETY: `snapshot` is owned here and not used after. |
| 1079 | let close_result = unsafe { CloseHandle(snapshot).map_err(windows_io_error) }; |
| 1080 | result?; |
| 1081 | close_result |
| 1082 | } |
| 1083 | |
| 1084 | #[cfg(windows)] |
| 1085 | fn kill_windows_process_tree(pid: u32) -> std::io::Result<()> { |
| 1086 | let mut command = Command::new("taskkill"); |
| 1087 | crate::utils::suppress_console_window(&mut command); |
| 1088 | let pid = pid.to_string(); |
| 1089 | let mut child = command |
| 1090 | .args(["/F", "/T", "/PID", pid.as_str()]) |
| 1091 | .stdin(Stdio::null()) |
| 1092 | .stdout(Stdio::null()) |
| 1093 | .stderr(Stdio::null()) |
| 1094 | .spawn()?; |
| 1095 | let status = wait_for_helper_status(&mut child, WINDOWS_TASKKILL_TIMEOUT)?; |
| 1096 | if status.success() { |
| 1097 | Ok(()) |
| 1098 | } else { |
| 1099 | Err(std::io::Error::other(format!( |
| 1100 | "taskkill exited with {status}" |
| 1101 | ))) |
| 1102 | } |
| 1103 | } |
| 1104 | |
| 1105 | #[cfg(any(windows, test))] |
| 1106 | fn wait_for_helper_status( |
| 1107 | child: &mut Child, |
| 1108 | timeout: Duration, |
| 1109 | ) -> std::io::Result<std::process::ExitStatus> { |
| 1110 | match child.wait_timeout(timeout)? { |
| 1111 | Some(status) => Ok(status), |
| 1112 | None => { |
| 1113 | let _ = kill_and_reap_immediate_child(child, HOOK_REAP_TIMEOUT); |
| 1114 | Err(std::io::Error::new( |
| 1115 | std::io::ErrorKind::TimedOut, |
| 1116 | "hook helper did not finish within its timeout", |
| 1117 | )) |
| 1118 | } |
| 1119 | } |
| 1120 | } |
| 1121 | |
| 1122 | fn kill_and_reap_immediate_child(child: &mut Child, timeout: Duration) -> bool { |
| 1123 | let _ = child.kill(); |
| 1124 | matches!(child.wait_timeout(timeout), Ok(Some(_))) |
| 1125 | } |
| 1126 | |
| 1127 | /// Spawn a contained hook child. |
| 1128 | /// |
| 1129 | /// Errors returned here are deliberately free of the hook command, the |
| 1130 | /// resolved interpreter path, and the OS message: the caller turns them into a |
| 1131 | /// user-visible "hook could not answer" receipt, and on Windows a raw spawn |
| 1132 | /// error echoes the whole command line back. The detail is logged instead. |
| 1133 | fn spawn_hook_child(command: &mut Command) -> std::io::Result<(Child, HookProcessTree)> { |
| 1134 | let mut child = command.spawn()?; |
| 1135 | let process_tree = match HookProcessTree::attach(&child) { |
| 1136 | Ok(process_tree) => process_tree, |
| 1137 | Err(error) => { |
| 1138 | // Windows hooks are created suspended, so a containment failure |
| 1139 | // cannot race with a descendant spawn. Fail closed without ever |
| 1140 | // running the uncontained hook. |
| 1141 | let _ = kill_and_reap_immediate_child(&mut child, HOOK_REAP_TIMEOUT); |
| 1142 | tracing::warn!(target: "hooks", %error, "failed to contain hook process tree"); |
| 1143 | return Err(std::io::Error::other("failed to contain hook process tree")); |
| 1144 | } |
| 1145 | }; |
| 1146 | |
| 1147 | #[cfg(windows)] |
| 1148 | if let Err(error) = resume_windows_process(&child) { |
| 1149 | let _ = terminate_and_reap(None, &mut child, process_tree); |
| 1150 | tracing::warn!(target: "hooks", %error, "failed to resume contained hook process"); |
| 1151 | return Err(std::io::Error::other( |
| 1152 | "failed to resume contained hook process", |
| 1153 | )); |
| 1154 | } |
| 1155 | |
| 1156 | Ok((child, process_tree)) |
| 1157 | } |
| 1158 | |
| 1159 | /// A spawn failure rendered without the command, the path, or the OS message. |
| 1160 | /// |
| 1161 | /// The error kind is the useful, non-identifying part (`NotFound`, |
| 1162 | /// `PermissionDenied`, …); everything else is logged, not surfaced. |
| 1163 | fn spawn_failure_message(error: &std::io::Error) -> String { |
| 1164 | format!("hook process could not be started ({:?})", error.kind()) |
| 1165 | } |
| 1166 | |
| 1167 | const OBSERVER_DISPATCH_QUEUE_CAPACITY: usize = 32; |
| 1168 | const OBSERVER_DISPATCH_WORKERS: usize = 2; |
| 1169 | |
| 1170 | #[derive(Debug, Clone, Copy)] |
| 1171 | enum ObserverDispatchFailure { |
| 1172 | Full, |
| 1173 | Disconnected, |
| 1174 | } |
| 1175 | |
| 1176 | /// Bounded, persistent submission path for observer-only events. |
| 1177 | /// |
| 1178 | /// The terminal loop never creates a thread per event. Two long-lived workers |
| 1179 | /// drain a fixed-capacity channel, and `try_send` makes saturation observable |
| 1180 | /// without ever parking the caller. |
| 1181 | #[derive(Clone)] |
| 1182 | struct ObserverDispatcher { |
| 1183 | sender: Option<SyncSender<ObserverJob>>, |
| 1184 | #[cfg(test)] |
| 1185 | held_receiver: Option<Arc<Mutex<Receiver<ObserverJob>>>>, |
| 1186 | } |
| 1187 | |
| 1188 | impl fmt::Debug for ObserverDispatcher { |
| 1189 | fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 1190 | formatter |
| 1191 | .debug_struct("ObserverDispatcher") |
| 1192 | .field("available", &self.sender.is_some()) |
| 1193 | .finish_non_exhaustive() |
| 1194 | } |
| 1195 | } |
| 1196 | |
| 1197 | impl ObserverDispatcher { |
| 1198 | fn new() -> Self { |
| 1199 | let (sender, receiver) = mpsc::sync_channel(OBSERVER_DISPATCH_QUEUE_CAPACITY); |
| 1200 | let receiver = Arc::new(Mutex::new(receiver)); |
| 1201 | |
| 1202 | for worker_index in 0..OBSERVER_DISPATCH_WORKERS { |
| 1203 | let worker_receiver = Arc::clone(&receiver); |
| 1204 | let spawned = std::thread::Builder::new() |
| 1205 | .name(format!("hook-observer-{worker_index}")) |
| 1206 | .spawn(move || observer_worker_loop(worker_receiver)); |
| 1207 | if let Err(error) = spawned { |
| 1208 | tracing::warn!( |
| 1209 | target: "hooks", |
| 1210 | worker_index, |
| 1211 | error_kind = ?error.kind(), |
| 1212 | "failed to start observer hook dispatcher" |
| 1213 | ); |
| 1214 | // Dropping the only sender disconnects any workers that did |
| 1215 | // start. A partially-created pool is not presented as healthy. |
| 1216 | drop(sender); |
| 1217 | return Self { |
| 1218 | sender: None, |
| 1219 | #[cfg(test)] |
| 1220 | held_receiver: None, |
| 1221 | }; |
| 1222 | } |
| 1223 | } |
| 1224 | |
| 1225 | Self { |
| 1226 | sender: Some(sender), |
| 1227 | #[cfg(test)] |
| 1228 | held_receiver: None, |
| 1229 | } |
| 1230 | } |
| 1231 | |
| 1232 | fn submit(&self, event: HookEvent, job: ObserverJob) -> Result<(), String> { |
| 1233 | let Some(sender) = &self.sender else { |
| 1234 | return Err(observer_dispatch_failure_message( |
| 1235 | event, |
| 1236 | ObserverDispatchFailure::Disconnected, |
| 1237 | )); |
| 1238 | }; |
| 1239 | match sender.try_send(job) { |
| 1240 | Ok(()) => Ok(()), |
| 1241 | Err(TrySendError::Full(_)) => Err(observer_dispatch_failure_message( |
| 1242 | event, |
| 1243 | ObserverDispatchFailure::Full, |
| 1244 | )), |
| 1245 | Err(TrySendError::Disconnected(_)) => Err(observer_dispatch_failure_message( |
| 1246 | event, |
| 1247 | ObserverDispatchFailure::Disconnected, |
| 1248 | )), |
| 1249 | } |
| 1250 | } |
| 1251 | } |
| 1252 | |
| 1253 | fn observer_dispatch_failure_message(event: HookEvent, failure: ObserverDispatchFailure) -> String { |
| 1254 | match failure { |
| 1255 | ObserverDispatchFailure::Full => format!( |
| 1256 | "{} observer hook queue is full; event was not submitted", |
| 1257 | event.as_str() |
| 1258 | ), |
| 1259 | ObserverDispatchFailure::Disconnected => format!( |
| 1260 | "{} observer hook dispatcher is unavailable; event was not submitted", |
| 1261 | event.as_str() |
| 1262 | ), |
| 1263 | } |
| 1264 | } |
| 1265 | |
| 1266 | enum ObserverJob { |
| 1267 | Environment { |
| 1268 | hooks: HookExecutor, |
| 1269 | event: HookEvent, |
| 1270 | context: HookContext, |
| 1271 | }, |
| 1272 | Json { |
| 1273 | hooks: HookExecutor, |
| 1274 | event: HookEvent, |
| 1275 | context: HookContext, |
| 1276 | payload: serde_json::Value, |
| 1277 | }, |
| 1278 | } |
| 1279 | |
| 1280 | impl ObserverJob { |
| 1281 | fn run(self) { |
| 1282 | match self { |
| 1283 | Self::Environment { |
| 1284 | hooks, |
| 1285 | event, |
| 1286 | context, |
| 1287 | } => { |
| 1288 | let _ = hooks.execute(event, &context); |
| 1289 | } |
| 1290 | Self::Json { |
| 1291 | hooks, |
| 1292 | event, |
| 1293 | context, |
| 1294 | payload, |
| 1295 | } => { |
| 1296 | let _ = hooks.execute_json_observer(event, &context, &payload); |
| 1297 | } |
| 1298 | } |
| 1299 | } |
| 1300 | } |
| 1301 | |
| 1302 | fn observer_worker_loop(receiver: Arc<Mutex<Receiver<ObserverJob>>>) { |
| 1303 | loop { |
| 1304 | let received = match receiver.lock() { |
| 1305 | Ok(receiver) => receiver.recv(), |
| 1306 | Err(_) => { |
| 1307 | tracing::warn!(target: "hooks", "observer hook dispatcher lock was poisoned"); |
| 1308 | return; |
| 1309 | } |
| 1310 | }; |
| 1311 | match received { |
| 1312 | Ok(job) => job.run(), |
| 1313 | Err(_) => return, |
| 1314 | } |
| 1315 | } |
| 1316 | } |
| 1317 | |
| 1318 | const BACKGROUND_SUPERVISOR_QUEUE_CAPACITY: usize = 32; |
| 1319 | const BACKGROUND_SUPERVISOR_WORKERS: usize = 2; |
| 1320 | |
| 1321 | #[derive(Debug, Clone, Copy)] |
| 1322 | enum BackgroundSupervisorFailure { |
| 1323 | Full, |
| 1324 | Disconnected, |
| 1325 | } |
| 1326 | |
| 1327 | /// Bounded pool that owns background-child setup, timeout, tree kill, and |
| 1328 | /// reap. Observer workers enqueue here instead of creating one detached |
| 1329 | /// supervisor thread per invocation. |
| 1330 | #[derive(Clone)] |
| 1331 | struct BackgroundSupervisor { |
| 1332 | sender: Option<SyncSender<BackgroundHookJob>>, |
| 1333 | #[cfg(test)] |
| 1334 | held_receiver: Option<Arc<Mutex<Receiver<BackgroundHookJob>>>>, |
| 1335 | } |
| 1336 | |
| 1337 | impl fmt::Debug for BackgroundSupervisor { |
| 1338 | fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 1339 | formatter |
| 1340 | .debug_struct("BackgroundSupervisor") |
| 1341 | .field("available", &self.sender.is_some()) |
| 1342 | .finish_non_exhaustive() |
| 1343 | } |
| 1344 | } |
| 1345 | |
| 1346 | impl BackgroundSupervisor { |
| 1347 | fn new() -> Self { |
| 1348 | let (sender, receiver) = mpsc::sync_channel(BACKGROUND_SUPERVISOR_QUEUE_CAPACITY); |
| 1349 | let receiver = Arc::new(Mutex::new(receiver)); |
| 1350 | |
| 1351 | for worker_index in 0..BACKGROUND_SUPERVISOR_WORKERS { |
| 1352 | let worker_receiver = Arc::clone(&receiver); |
| 1353 | let spawned = std::thread::Builder::new() |
| 1354 | .name(format!("hook-supervisor-{worker_index}")) |
| 1355 | .spawn(move || background_supervisor_worker_loop(worker_receiver)); |
| 1356 | if let Err(error) = spawned { |
| 1357 | tracing::warn!( |
| 1358 | target: "hooks", |
| 1359 | worker_index, |
| 1360 | error_kind = ?error.kind(), |
| 1361 | "failed to start background hook supervisor pool" |
| 1362 | ); |
| 1363 | drop(sender); |
| 1364 | return Self { |
| 1365 | sender: None, |
| 1366 | #[cfg(test)] |
| 1367 | held_receiver: None, |
| 1368 | }; |
| 1369 | } |
| 1370 | } |
| 1371 | |
| 1372 | Self { |
| 1373 | sender: Some(sender), |
| 1374 | #[cfg(test)] |
| 1375 | held_receiver: None, |
| 1376 | } |
| 1377 | } |
| 1378 | |
| 1379 | fn submit(&self, job: BackgroundHookJob) -> Result<(), BackgroundSupervisorFailure> { |
| 1380 | let Some(sender) = &self.sender else { |
| 1381 | return Err(BackgroundSupervisorFailure::Disconnected); |
| 1382 | }; |
| 1383 | match sender.try_send(job) { |
| 1384 | Ok(()) => Ok(()), |
| 1385 | Err(TrySendError::Full(_)) => Err(BackgroundSupervisorFailure::Full), |
| 1386 | Err(TrySendError::Disconnected(_)) => Err(BackgroundSupervisorFailure::Disconnected), |
| 1387 | } |
| 1388 | } |
| 1389 | } |
| 1390 | |
| 1391 | struct BackgroundHookJob { |
| 1392 | command: String, |
| 1393 | env: HashMap<String, String>, |
| 1394 | working_dir: PathBuf, |
| 1395 | stdin_bytes: Option<Vec<u8>>, |
| 1396 | label: String, |
| 1397 | timeout: Duration, |
| 1398 | plugin_authority: Option<crate::plugins::types::PluginAuthority>, |
| 1399 | project_authority: Option<super::authority::ProjectHookAuthority>, |
| 1400 | } |
| 1401 | |
| 1402 | impl BackgroundHookJob { |
| 1403 | fn run(self) { |
| 1404 | let Self { |
| 1405 | command: command_text, |
| 1406 | env, |
| 1407 | working_dir, |
| 1408 | stdin_bytes, |
| 1409 | label, |
| 1410 | timeout, |
| 1411 | plugin_authority, |
| 1412 | project_authority, |
| 1413 | } = self; |
| 1414 | if let Err(error) = super::authority::verify_hook_authorities( |
| 1415 | plugin_authority.as_ref(), |
| 1416 | project_authority.as_ref(), |
| 1417 | ) { |
| 1418 | tracing::warn!( |
| 1419 | target: "hooks", |
| 1420 | hook = %label, |
| 1421 | error = %error, |
| 1422 | "denied queued hook after authority changed" |
| 1423 | ); |
| 1424 | return; |
| 1425 | } |
| 1426 | let timeout_secs = timeout.as_secs(); |
| 1427 | let mut command = HookExecutor::build_shell_command(&command_text); |
| 1428 | command |
| 1429 | .current_dir(&working_dir) |
| 1430 | .envs(&env) |
| 1431 | .stdout(Stdio::null()) |
| 1432 | .stderr(Stdio::null()) |
| 1433 | // Always pipe stdin so dropping it delivers EOF through shell |
| 1434 | // wrappers even when there is no structured payload. |
| 1435 | .stdin(Stdio::piped()); |
| 1436 | |
| 1437 | let (mut child, process_tree) = match spawn_hook_child(&mut command) { |
| 1438 | Ok(child) => child, |
| 1439 | Err(error) => { |
| 1440 | tracing::warn!( |
| 1441 | target: "hooks", |
| 1442 | hook = %label, |
| 1443 | error_kind = ?error.kind(), |
| 1444 | "failed to start background hook" |
| 1445 | ); |
| 1446 | return; |
| 1447 | } |
| 1448 | }; |
| 1449 | |
| 1450 | let _stdin_writer = match (stdin_bytes, child.stdin.take()) { |
| 1451 | (Some(bytes), Some(stdin)) => match spawn_stdin_writer(stdin, bytes) { |
| 1452 | Ok(writer) => Some(writer), |
| 1453 | Err(error) => { |
| 1454 | tracing::warn!( |
| 1455 | target: "hooks", |
| 1456 | hook = %label, |
| 1457 | error_kind = ?error.kind(), |
| 1458 | "failed to start background hook stdin writer" |
| 1459 | ); |
| 1460 | terminate_and_reap(Some(label.as_str()), &mut child, process_tree); |
| 1461 | return; |
| 1462 | } |
| 1463 | }, |
| 1464 | _ => None, |
| 1465 | }; |
| 1466 | |
| 1467 | match child.wait_timeout(timeout) { |
| 1468 | Ok(Some(status)) => { |
| 1469 | if !status.success() { |
| 1470 | tracing::warn!( |
| 1471 | target: "hooks", |
| 1472 | hook = %label, |
| 1473 | exit_code = ?status.code(), |
| 1474 | "background hook exited non-zero" |
| 1475 | ); |
| 1476 | } |
| 1477 | } |
| 1478 | Ok(None) => { |
| 1479 | let reaped = terminate_and_reap(Some(label.as_str()), &mut child, process_tree); |
| 1480 | tracing::warn!( |
| 1481 | target: "hooks", |
| 1482 | hook = %label, |
| 1483 | timeout_secs, |
| 1484 | reaped, |
| 1485 | "background hook timed out; process tree killed" |
| 1486 | ); |
| 1487 | } |
| 1488 | Err(error) => { |
| 1489 | terminate_and_reap(Some(label.as_str()), &mut child, process_tree); |
| 1490 | tracing::warn!( |
| 1491 | target: "hooks", |
| 1492 | hook = %label, |
| 1493 | ?error, |
| 1494 | "failed to wait for background hook; process tree killed" |
| 1495 | ); |
| 1496 | } |
| 1497 | } |
| 1498 | } |
| 1499 | } |
| 1500 | |
| 1501 | fn background_supervisor_worker_loop(receiver: Arc<Mutex<Receiver<BackgroundHookJob>>>) { |
| 1502 | loop { |
| 1503 | let received = match receiver.lock() { |
| 1504 | Ok(receiver) => receiver.recv(), |
| 1505 | Err(_) => { |
| 1506 | tracing::warn!(target: "hooks", "background supervisor lock was poisoned"); |
| 1507 | return; |
| 1508 | } |
| 1509 | }; |
| 1510 | match received { |
| 1511 | Ok(job) => job.run(), |
| 1512 | Err(_) => return, |
| 1513 | } |
| 1514 | } |
| 1515 | } |
| 1516 | |
| 1517 | /// Executor for running hooks |
| 1518 | #[derive(Debug, Clone)] |
| 1519 | pub struct HookExecutor { |
| 1520 | config: HooksConfig, |
| 1521 | default_working_dir: PathBuf, |
| 1522 | session_id: String, |
| 1523 | observer_dispatcher: ObserverDispatcher, |
| 1524 | background_supervisor: BackgroundSupervisor, |
| 1525 | #[cfg(test)] |
| 1526 | lose_message_submit_executor: bool, |
| 1527 | } |
| 1528 | |
| 1529 | impl HookExecutor { |
| 1530 | fn build_shell_command(command: &str) -> Command { |
| 1531 | #[cfg(windows)] |
| 1532 | { |
| 1533 | use std::os::windows::process::CommandExt as _; |
| 1534 | let mut cmd = Command::new("cmd"); |
| 1535 | const CREATE_SUSPENDED: u32 = 0x0000_0004; |
| 1536 | const CREATE_NO_WINDOW: u32 = 0x0800_0000; |
| 1537 | cmd.creation_flags(CREATE_SUSPENDED | CREATE_NO_WINDOW); |
| 1538 | // raw_arg: cmd.exe does not parse the CRT-style \" escapes that |
| 1539 | // Command::arg would insert, so pass the command line verbatim. |
| 1540 | cmd.arg("/C").raw_arg(command); |
| 1541 | cmd |
| 1542 | } |
| 1543 | #[cfg(not(windows))] |
| 1544 | { |
| 1545 | let mut cmd = Command::new("sh"); |
| 1546 | cmd.arg("-c").arg(command); |
| 1547 | #[cfg(unix)] |
| 1548 | { |
| 1549 | use std::os::unix::process::CommandExt as _; |
| 1550 | cmd.process_group(0); |
| 1551 | } |
| 1552 | cmd |
| 1553 | } |
| 1554 | } |
| 1555 | |
| 1556 | /// Create a new `HookExecutor` with configuration. |
| 1557 | /// |
| 1558 | /// This mints the hook session identity for the whole TUI session. Call it |
| 1559 | /// **once per launch**; every later reload (workspace switch, trust |
| 1560 | /// onboarding) must go through [`Self::rebind`] so the id every hook has |
| 1561 | /// already seen stays valid. Regenerating it mid-session would break |
| 1562 | /// correlation for anything that grouped records by `CODEWHALE_SESSION_ID` |
| 1563 | /// or its `DEEPSEEK_SESSION_ID` compatibility alias. |
| 1564 | pub fn new(config: HooksConfig, default_working_dir: PathBuf) -> Self { |
| 1565 | // Generate a session ID |
| 1566 | let session_id = format!("sess_{}", &uuid::Uuid::new_v4().to_string()[..8]); |
| 1567 | Self { |
| 1568 | config, |
| 1569 | default_working_dir, |
| 1570 | session_id, |
| 1571 | observer_dispatcher: ObserverDispatcher::new(), |
| 1572 | background_supervisor: BackgroundSupervisor::new(), |
| 1573 | #[cfg(test)] |
| 1574 | lose_message_submit_executor: false, |
| 1575 | } |
| 1576 | } |
| 1577 | |
| 1578 | /// Rebuild the executor with new configuration and working directory while |
| 1579 | /// preserving the session identity minted at launch. |
| 1580 | /// |
| 1581 | /// Used when the workspace changes or a trust decision makes project hooks |
| 1582 | /// eligible: the hook set may change, the session does not. |
| 1583 | #[must_use] |
| 1584 | pub fn rebind(&self, config: HooksConfig, default_working_dir: PathBuf) -> Self { |
| 1585 | Self { |
| 1586 | config, |
| 1587 | default_working_dir, |
| 1588 | session_id: self.session_id.clone(), |
| 1589 | observer_dispatcher: self.observer_dispatcher.clone(), |
| 1590 | background_supervisor: self.background_supervisor.clone(), |
| 1591 | #[cfg(test)] |
| 1592 | lose_message_submit_executor: self.lose_message_submit_executor, |
| 1593 | } |
| 1594 | } |
| 1595 | |
| 1596 | /// Create a disabled `HookExecutor` (no hooks will run) |
| 1597 | #[cfg(test)] |
| 1598 | pub fn disabled() -> Self { |
| 1599 | Self { |
| 1600 | config: HooksConfig { |
| 1601 | enabled: false, |
| 1602 | ..Default::default() |
| 1603 | }, |
| 1604 | default_working_dir: PathBuf::from("."), |
| 1605 | session_id: String::new(), |
| 1606 | observer_dispatcher: ObserverDispatcher::new(), |
| 1607 | background_supervisor: BackgroundSupervisor::new(), |
| 1608 | #[cfg(test)] |
| 1609 | lose_message_submit_executor: false, |
| 1610 | } |
| 1611 | } |
| 1612 | |
| 1613 | /// Check if hooks are enabled |
| 1614 | #[cfg(test)] |
| 1615 | pub fn is_enabled(&self) -> bool { |
| 1616 | self.config.enabled |
| 1617 | } |
| 1618 | |
| 1619 | /// Get the session ID |
| 1620 | /// Read-only access to the underlying configuration. Used by |
| 1621 | /// `/hooks` (#460 read-only MVP) so the user can list configured |
| 1622 | /// hooks without reaching for `cat ~/.deepseek/config.toml`. |
| 1623 | pub fn config(&self) -> &HooksConfig { |
| 1624 | &self.config |
| 1625 | } |
| 1626 | |
| 1627 | pub fn session_id(&self) -> &str { |
| 1628 | &self.session_id |
| 1629 | } |
| 1630 | |
| 1631 | /// Cheap pre-check: are there any enabled hooks for this event? |
| 1632 | /// Lets call sites avoid building a [`HookContext`] (which allocates |
| 1633 | /// for `workspace`, `model`, `session_id`, …) on every tool call |
| 1634 | /// when the user hasn't configured any hooks. The cost matters |
| 1635 | /// because `ToolCallBefore` / `ToolCallAfter` fire from |
| 1636 | /// `tool_routing.rs` on every tool dispatch (#455). |
| 1637 | #[must_use] |
| 1638 | pub fn has_hooks_for_event(&self, event: HookEvent) -> bool { |
| 1639 | self.config.enabled && self.config.hooks.iter().any(|h| h.event == event) |
| 1640 | } |
| 1641 | |
| 1642 | /// Check if there are any background hooks configured for a specific event. |
| 1643 | /// |
| 1644 | /// Background hooks fire and forget — their `exit_code` is always `None`, |
| 1645 | /// so they cannot deny tool calls. This is a known limitation; the check |
| 1646 | /// is used to warn operators when a `ToolCallBefore` hook is configured |
| 1647 | /// as background but expects to block a tool. |
| 1648 | #[must_use] |
| 1649 | pub fn has_background_hooks_for_event(&self, event: HookEvent) -> bool { |
| 1650 | if !self.config.enabled { |
| 1651 | return false; |
| 1652 | } |
| 1653 | self.config |
| 1654 | .hooks |
| 1655 | .iter() |
| 1656 | .any(|h| h.event == event && h.background) |
| 1657 | } |
| 1658 | |
| 1659 | /// Sanitized labels of the strict foreground gates that *would* run for |
| 1660 | /// this event and context. |
| 1661 | /// |
| 1662 | /// "Strict" is `continue_on_error = false` on a foreground hook: an |
| 1663 | /// operator instruction that the action must not proceed without this |
| 1664 | /// hook's answer. The caller collects these **before** dispatching the |
| 1665 | /// executor so it can still honor them if the execution itself is lost — |
| 1666 | /// a panicked or cancelled `spawn_blocking` returns no results at all, and |
| 1667 | /// an empty result set is indistinguishable from "every hook allowed it". |
| 1668 | /// |
| 1669 | /// Condition matching is the same predicate [`Self::execute`] uses, so |
| 1670 | /// this never names a hook that would not have run: a strict `write_file` |
| 1671 | /// gate has no say over an `exec_shell` call it never matched. |
| 1672 | #[must_use] |
| 1673 | pub fn matched_strict_gate_labels( |
| 1674 | &self, |
| 1675 | event: HookEvent, |
| 1676 | context: &HookContext, |
| 1677 | ) -> Vec<String> { |
| 1678 | if !self.config.enabled { |
| 1679 | return Vec::new(); |
| 1680 | } |
| 1681 | self.config |
| 1682 | .hooks_for_event(event) |
| 1683 | .into_iter() |
| 1684 | .filter(|hook| { |
| 1685 | // A background hook is never awaited, so it is not a gate no |
| 1686 | // matter what `continue_on_error` says. |
| 1687 | let foreground = !hook.background || !hook.event.honors_background(); |
| 1688 | foreground && !hook.continue_on_error && self.matches_condition(hook, context) |
| 1689 | }) |
| 1690 | .map(|hook| sanitize_hook_label(hook.name.as_deref())) |
| 1691 | .collect() |
| 1692 | } |
| 1693 | |
| 1694 | /// Run configured `message_submit` hooks as a mutable submit pipeline. |
| 1695 | /// |
| 1696 | /// This is deliberately separate from [`Self::execute`]: most hook events |
| 1697 | /// are observer-only, while `message_submit` has a narrow stdout JSON |
| 1698 | /// contract that can replace or block the submitted text. |
| 1699 | pub fn execute_message_submit_transform( |
| 1700 | &self, |
| 1701 | context: &HookContext, |
| 1702 | original_text: &str, |
| 1703 | ) -> MessageSubmitOutcome { |
| 1704 | if !self.config.enabled { |
| 1705 | return MessageSubmitOutcome::unchanged(); |
| 1706 | } |
| 1707 | |
| 1708 | let hooks = self.config.hooks_for_event(HookEvent::MessageSubmit); |
| 1709 | if hooks.is_empty() { |
| 1710 | return MessageSubmitOutcome::unchanged(); |
| 1711 | } |
| 1712 | |
| 1713 | let mut current_text = original_text.to_string(); |
| 1714 | let mut warning = None; |
| 1715 | |
| 1716 | for hook in hooks { |
| 1717 | let hook_context = context.clone().with_message(¤t_text); |
| 1718 | if !self.matches_condition(hook, &hook_context) { |
| 1719 | continue; |
| 1720 | } |
| 1721 | |
| 1722 | let env_vars = hook_context.to_env_vars(); |
| 1723 | let payload = message_submit_payload(&hook_context, ¤t_text); |
| 1724 | if hook.background { |
| 1725 | // A background `message_submit` hook cannot steer, but it must |
| 1726 | // still receive the documented stdin payload — the contract is |
| 1727 | // the same JSON, only the steering is dropped. |
| 1728 | let submitted = self.execute_background_with_stdin(hook, &env_vars, &payload); |
| 1729 | // Submission itself can fail (thread spawn refused, payload not |
| 1730 | // encodable). Discarding that silently is the one outcome an |
| 1731 | // operator cannot debug: the hook is configured, nothing runs, |
| 1732 | // and nothing says so. Still non-blocking — the submit proceeds. |
| 1733 | if !submitted.success { |
| 1734 | tracing::warn!( |
| 1735 | target: "hooks", |
| 1736 | hook = %sanitize_hook_label(submitted.name.as_deref()), |
| 1737 | event = "message_submit", |
| 1738 | error = %generic_unavailable_detail(submitted.error.as_deref()), |
| 1739 | "background message_submit hook was not submitted; it will not run" |
| 1740 | ); |
| 1741 | } |
| 1742 | continue; |
| 1743 | } |
| 1744 | |
| 1745 | let result = self.execute_sync_with_stdin(hook, &env_vars, &payload); |
| 1746 | |
| 1747 | if result.exit_code == Some(2) { |
| 1748 | return MessageSubmitOutcome::Blocked { |
| 1749 | reason: message_submit_block_reason( |
| 1750 | &result, |
| 1751 | "message_submit hook blocked submission", |
| 1752 | ), |
| 1753 | }; |
| 1754 | } |
| 1755 | |
| 1756 | if !result.success { |
| 1757 | let label = sanitize_hook_label(result.name.as_deref()); |
| 1758 | tracing::warn!( |
| 1759 | target: "hooks", |
| 1760 | hook = %label, |
| 1761 | event = "message_submit", |
| 1762 | exit_code = ?result.exit_code, |
| 1763 | duration_ms = result.duration.as_millis() as u64, |
| 1764 | detail = %generic_unavailable_detail(result.error.as_deref()), |
| 1765 | "message_submit hook failed" |
| 1766 | ); |
| 1767 | |
| 1768 | if hook.continue_on_error { |
| 1769 | warning = message_submit_continue_warning(&result).or(warning); |
| 1770 | continue; |
| 1771 | } |
| 1772 | |
| 1773 | return MessageSubmitOutcome::Blocked { |
| 1774 | reason: message_submit_block_reason( |
| 1775 | &result, |
| 1776 | "message_submit hook failed and blocked submission", |
| 1777 | ), |
| 1778 | }; |
| 1779 | } |
| 1780 | |
| 1781 | match parse_message_submit_stdout(&result.stdout) { |
| 1782 | MessageSubmitStdout::Unchanged => {} |
| 1783 | MessageSubmitStdout::Replaced(text) => { |
| 1784 | current_text = text; |
| 1785 | } |
| 1786 | MessageSubmitStdout::Invalid(reason) => { |
| 1787 | tracing::warn!( |
| 1788 | target: "hooks", |
| 1789 | hook = %sanitize_hook_label(result.name.as_deref()), |
| 1790 | event = "message_submit", |
| 1791 | reason = %reason, |
| 1792 | "ignored invalid message_submit hook stdout" |
| 1793 | ); |
| 1794 | } |
| 1795 | } |
| 1796 | } |
| 1797 | |
| 1798 | if current_text == original_text { |
| 1799 | MessageSubmitOutcome::unchanged().with_warning(warning) |
| 1800 | } else { |
| 1801 | MessageSubmitOutcome::replaced(current_text).with_warning(warning) |
| 1802 | } |
| 1803 | } |
| 1804 | |
| 1805 | /// Dispatch-bound entry point for the mutable submit gate. |
| 1806 | /// |
| 1807 | /// Keeping this wrapper distinct gives the production dispatch path a |
| 1808 | /// deterministic test seam for a lost blocking task. Normal hook tests use |
| 1809 | /// [`Self::execute_message_submit_transform`] directly. |
| 1810 | pub(crate) fn execute_message_submit_transform_for_dispatch( |
| 1811 | &self, |
| 1812 | context: &HookContext, |
| 1813 | original_text: &str, |
| 1814 | ) -> MessageSubmitOutcome { |
| 1815 | #[cfg(test)] |
| 1816 | if self.lose_message_submit_executor { |
| 1817 | panic!("injected message_submit executor loss"); |
| 1818 | } |
| 1819 | self.execute_message_submit_transform(context, original_text) |
| 1820 | } |
| 1821 | |
| 1822 | #[cfg(test)] |
| 1823 | pub(crate) fn inject_message_submit_executor_loss_for_test(&mut self) { |
| 1824 | self.lose_message_submit_executor = true; |
| 1825 | } |
| 1826 | |
| 1827 | #[cfg(test)] |
| 1828 | pub(crate) fn inject_observer_dispatch_full_for_test(&mut self) { |
| 1829 | let (sender, receiver) = mpsc::sync_channel(0); |
| 1830 | self.observer_dispatcher.sender = Some(sender); |
| 1831 | // Keep the receiver connected but deliberately leave no worker waiting |
| 1832 | // on it. The production `try_send` path therefore returns `Full`. |
| 1833 | self.observer_dispatcher.held_receiver = Some(Arc::new(Mutex::new(receiver))); |
| 1834 | } |
| 1835 | |
| 1836 | #[cfg(test)] |
| 1837 | pub(crate) fn inject_observer_dispatch_disconnect_for_test(&mut self) { |
| 1838 | let (sender, receiver) = mpsc::sync_channel(1); |
| 1839 | drop(receiver); |
| 1840 | self.observer_dispatcher.sender = Some(sender); |
| 1841 | self.observer_dispatcher.held_receiver = None; |
| 1842 | } |
| 1843 | |
| 1844 | #[cfg(test)] |
| 1845 | fn inject_background_supervisor_full_for_test(&mut self) { |
| 1846 | let (sender, receiver) = mpsc::sync_channel(0); |
| 1847 | self.background_supervisor.sender = Some(sender); |
| 1848 | self.background_supervisor.held_receiver = Some(Arc::new(Mutex::new(receiver))); |
| 1849 | } |
| 1850 | |
| 1851 | /// Run every `ShellEnv` hook for this context and merge their stdout |
| 1852 | /// (`KEY=VALUE\n` lines) into a single env-var map. Used by the |
| 1853 | /// `exec_shell` tool to inject ephemeral credentials, per-skill PATH |
| 1854 | /// adjustments, etc. (#456). Failures don't abort the shell call — |
| 1855 | /// the hook simply contributes no vars and a `tracing::warn!` lands. |
| 1856 | /// |
| 1857 | /// Each successful hook's keys (NOT values) are written to the audit |
| 1858 | /// log so a session can be reconciled later without leaking the |
| 1859 | /// secret material itself. |
| 1860 | pub fn collect_shell_env(&self, context: &HookContext) -> HashMap<String, String> { |
| 1861 | let mut merged: HashMap<String, String> = HashMap::new(); |
| 1862 | if !self.config.enabled { |
| 1863 | return merged; |
| 1864 | } |
| 1865 | let hooks = self.config.hooks_for_event(HookEvent::ShellEnv); |
| 1866 | if hooks.is_empty() { |
| 1867 | return merged; |
| 1868 | } |
| 1869 | let env_vars = context.to_env_vars(); |
| 1870 | for hook in hooks { |
| 1871 | if !self.matches_condition(hook, context) { |
| 1872 | continue; |
| 1873 | } |
| 1874 | // ShellEnv hooks must be synchronous — their stdout is the contract. |
| 1875 | let result = self.execute_sync(hook, &env_vars); |
| 1876 | if !result.success { |
| 1877 | tracing::warn!( |
| 1878 | target: "hooks", |
| 1879 | hook = %sanitize_hook_label(result.name.as_deref()), |
| 1880 | event = "shell_env", |
| 1881 | exit_code = ?result.exit_code, |
| 1882 | detail = %generic_unavailable_detail(result.error.as_deref()), |
| 1883 | "shell_env hook failed; contributing no env vars" |
| 1884 | ); |
| 1885 | continue; |
| 1886 | } |
| 1887 | let parsed = parse_env_lines(&result.stdout); |
| 1888 | if parsed.is_empty() { |
| 1889 | continue; |
| 1890 | } |
| 1891 | // Audit-log the *keys* — never the values. |
| 1892 | crate::audit::log_sensitive_event( |
| 1893 | "shell_env_hook", |
| 1894 | serde_json::json!({ |
| 1895 | // Bounded and de-fanged like every other rendering of a |
| 1896 | // hook name: an audit record is read by a person, often |
| 1897 | // through `tail`, where a raw escape sequence still acts. |
| 1898 | "hook": sanitize_hook_label(result.name.as_deref()), |
| 1899 | "tool": context.tool_name, |
| 1900 | "keys": parsed.keys().cloned().collect::<Vec<_>>(), |
| 1901 | }), |
| 1902 | ); |
| 1903 | // Later hooks override earlier ones. Documented behavior. |
| 1904 | merged.extend(parsed); |
| 1905 | } |
| 1906 | merged |
| 1907 | } |
| 1908 | |
| 1909 | /// Execute all hooks for an event |
| 1910 | pub fn execute(&self, event: HookEvent, context: &HookContext) -> Vec<HookResult> { |
| 1911 | if !self.config.enabled { |
| 1912 | return Vec::new(); |
| 1913 | } |
| 1914 | |
| 1915 | let hooks = self.config.hooks_for_event(event); |
| 1916 | if hooks.is_empty() { |
| 1917 | // Fast path: no hooks for this event → skip the |
| 1918 | // `context.to_env_vars()` HashMap allocation. With |
| 1919 | // `tool_call_before` / `tool_call_after` firing per-tool |
| 1920 | // (#455) this allocation would otherwise happen on every |
| 1921 | // tool dispatch even for users with zero hooks configured. |
| 1922 | return Vec::new(); |
| 1923 | } |
| 1924 | let env_vars = context.to_env_vars(); |
| 1925 | let mut results = Vec::new(); |
| 1926 | |
| 1927 | for hook in hooks { |
| 1928 | if !self.matches_condition(hook, context) { |
| 1929 | continue; |
| 1930 | } |
| 1931 | |
| 1932 | let result = if hook.background { |
| 1933 | self.execute_background(hook, &env_vars) |
| 1934 | } else { |
| 1935 | self.execute_sync(hook, &env_vars) |
| 1936 | }; |
| 1937 | |
| 1938 | // Log failures via tracing so operators tailing |
| 1939 | // `deepseek` with `RUST_LOG=warn` can see hook errors |
| 1940 | // without instrumenting each call site. Successful runs |
| 1941 | // log nothing (would be too noisy on per-tool events). |
| 1942 | if !result.success { |
| 1943 | let label = sanitize_hook_label(result.name.as_deref()); |
| 1944 | tracing::warn!( |
| 1945 | target: "hooks", |
| 1946 | hook = %label, |
| 1947 | event = event.as_str(), |
| 1948 | exit_code = ?result.exit_code, |
| 1949 | duration_ms = result.duration.as_millis() as u64, |
| 1950 | detail = %generic_unavailable_detail(result.error.as_deref()), |
| 1951 | "hook failed" |
| 1952 | ); |
| 1953 | } |
| 1954 | |
| 1955 | let should_continue = result.success || hook.continue_on_error; |
| 1956 | results.push(result); |
| 1957 | |
| 1958 | if !should_continue { |
| 1959 | break; |
| 1960 | } |
| 1961 | } |
| 1962 | |
| 1963 | results |
| 1964 | } |
| 1965 | |
| 1966 | /// Execute observer hooks with a structured JSON stdin payload. |
| 1967 | /// |
| 1968 | /// Unlike `message_submit`, stdout is deliberately ignored by callers: |
| 1969 | /// these hooks are lifecycle observers and cannot mutate or block the |
| 1970 | /// underlying action. |
| 1971 | pub fn execute_json_observer( |
| 1972 | &self, |
| 1973 | event: HookEvent, |
| 1974 | context: &HookContext, |
| 1975 | payload: &serde_json::Value, |
| 1976 | ) -> Vec<HookResult> { |
| 1977 | if !self.config.enabled { |
| 1978 | return Vec::new(); |
| 1979 | } |
| 1980 | |
| 1981 | let hooks = self.config.hooks_for_event(event); |
| 1982 | if hooks.is_empty() { |
| 1983 | return Vec::new(); |
| 1984 | } |
| 1985 | |
| 1986 | let env_vars = context.to_env_vars(); |
| 1987 | let mut results = Vec::new(); |
| 1988 | for hook in hooks { |
| 1989 | if !self.matches_condition(hook, context) { |
| 1990 | continue; |
| 1991 | } |
| 1992 | |
| 1993 | let result = if hook.background { |
| 1994 | self.execute_background_with_stdin(hook, &env_vars, payload) |
| 1995 | } else { |
| 1996 | self.execute_sync_with_stdin(hook, &env_vars, payload) |
| 1997 | }; |
| 1998 | |
| 1999 | if !result.success { |
| 2000 | let label = sanitize_hook_label(result.name.as_deref()); |
| 2001 | tracing::warn!( |
| 2002 | target: "hooks", |
| 2003 | hook = %label, |
| 2004 | event = event.as_str(), |
| 2005 | exit_code = ?result.exit_code, |
| 2006 | duration_ms = result.duration.as_millis() as u64, |
| 2007 | detail = %generic_unavailable_detail(result.error.as_deref()), |
| 2008 | "observer hook failed" |
| 2009 | ); |
| 2010 | } |
| 2011 | |
| 2012 | results.push(result); |
| 2013 | } |
| 2014 | |
| 2015 | results |
| 2016 | } |
| 2017 | |
| 2018 | /// Submit an observer event without waiting on foreground child processes |
| 2019 | /// from the caller's thread. The outer worker is fallible and the failure |
| 2020 | /// is returned to the UI; silently dropping a configured observer is not a |
| 2021 | /// truthful fire-and-forget contract. |
| 2022 | pub fn submit_observer(&self, event: HookEvent, context: HookContext) -> Result<(), String> { |
| 2023 | if !self.has_hooks_for_event(event) { |
| 2024 | return Ok(()); |
| 2025 | } |
| 2026 | self.observer_dispatcher.submit( |
| 2027 | event, |
| 2028 | ObserverJob::Environment { |
| 2029 | hooks: self.clone(), |
| 2030 | event, |
| 2031 | context: context.bounded_for_observer(), |
| 2032 | }, |
| 2033 | ) |
| 2034 | } |
| 2035 | |
| 2036 | /// Structured-payload counterpart to [`Self::submit_observer`]. |
| 2037 | pub fn submit_json_observer( |
| 2038 | &self, |
| 2039 | event: HookEvent, |
| 2040 | context: HookContext, |
| 2041 | payload: serde_json::Value, |
| 2042 | ) -> Result<(), String> { |
| 2043 | if !self.has_hooks_for_event(event) { |
| 2044 | return Ok(()); |
| 2045 | } |
| 2046 | self.observer_dispatcher.submit( |
| 2047 | event, |
| 2048 | ObserverJob::Json { |
| 2049 | hooks: self.clone(), |
| 2050 | event, |
| 2051 | context: context.bounded_for_observer(), |
| 2052 | payload, |
| 2053 | }, |
| 2054 | ) |
| 2055 | } |
| 2056 | |
| 2057 | /// Check whether a tool name matches a condition pattern with `*` glob support. |
| 2058 | fn tool_name_matches_condition(tool_name: &str, pattern: &str) -> bool { |
| 2059 | if !pattern.contains('*') { |
| 2060 | return tool_name == pattern; |
| 2061 | } |
| 2062 | // #6208: the pattern is fixed by configuration while this runs once per |
| 2063 | // hook per tool-call/stop event, so compile it once and reuse it rather |
| 2064 | // than building a fresh `Regex` on every event. |
| 2065 | codewhale_execpolicy::matcher::compiled_glob(pattern) |
| 2066 | .is_some_and(|re| re.is_match(tool_name)) |
| 2067 | } |
| 2068 | |
| 2069 | /// Check if a hook's condition matches the context |
| 2070 | #[allow(clippy::only_used_in_recursion)] |
| 2071 | fn matches_condition(&self, hook: &Hook, context: &HookContext) -> bool { |
| 2072 | match &hook.condition { |
| 2073 | None | Some(HookCondition::Always) => true, |
| 2074 | Some(HookCondition::ToolName { name }) => { |
| 2075 | // #3026: Support `*` globs in tool_name conditions so |
| 2076 | // `mcp__*` matches all MCP tools. Exact names keep working. |
| 2077 | context |
| 2078 | .tool_name |
| 2079 | .as_ref() |
| 2080 | .is_some_and(|n| Self::tool_name_matches_condition(n, name)) |
| 2081 | } |
| 2082 | Some(HookCondition::ToolCategory { category }) => { |
| 2083 | let tool_category = context |
| 2084 | .tool_name |
| 2085 | .as_deref() |
| 2086 | .map(|name| tool_category_for(name, context.tool_args.as_deref())); |
| 2087 | tool_category.is_some_and(|c| c == category.as_str()) |
| 2088 | } |
| 2089 | Some(HookCondition::Mode { mode }) => context |
| 2090 | .mode |
| 2091 | .as_ref() |
| 2092 | .is_some_and(|m| m.eq_ignore_ascii_case(mode)), |
| 2093 | Some(HookCondition::ExitCode { code }) => context.tool_exit_code == Some(*code), |
| 2094 | Some(HookCondition::All { conditions }) => conditions.iter().all(|c| { |
| 2095 | self.matches_condition( |
| 2096 | &Hook { |
| 2097 | condition: Some(c.clone()), |
| 2098 | ..hook.clone() |
| 2099 | }, |
| 2100 | context, |
| 2101 | ) |
| 2102 | }), |
| 2103 | Some(HookCondition::Any { conditions }) => conditions.iter().any(|c| { |
| 2104 | self.matches_condition( |
| 2105 | &Hook { |
| 2106 | condition: Some(c.clone()), |
| 2107 | ..hook.clone() |
| 2108 | }, |
| 2109 | context, |
| 2110 | ) |
| 2111 | }), |
| 2112 | } |
| 2113 | } |
| 2114 | |
| 2115 | /// Execute a hook synchronously |
| 2116 | fn execute_sync(&self, hook: &Hook, env_vars: &HashMap<String, String>) -> HookResult { |
| 2117 | self.execute_sync_inner(hook, env_vars, None) |
| 2118 | } |
| 2119 | |
| 2120 | /// Execute a hook synchronously with a structured JSON stdin payload. |
| 2121 | /// |
| 2122 | /// Used by mutable `message_submit` hooks. Existing observer hooks keep the |
| 2123 | /// stdin-less [`Self::execute_sync`] path so their behavior is unchanged. |
| 2124 | fn execute_sync_with_stdin( |
| 2125 | &self, |
| 2126 | hook: &Hook, |
| 2127 | env_vars: &HashMap<String, String>, |
| 2128 | stdin_json: &serde_json::Value, |
| 2129 | ) -> HookResult { |
| 2130 | self.execute_sync_inner(hook, env_vars, Some(stdin_json)) |
| 2131 | } |
| 2132 | |
| 2133 | fn execute_sync_inner( |
| 2134 | &self, |
| 2135 | hook: &Hook, |
| 2136 | env_vars: &HashMap<String, String>, |
| 2137 | stdin_json: Option<&serde_json::Value>, |
| 2138 | ) -> HookResult { |
| 2139 | let started = Instant::now(); |
| 2140 | if let Err(reason) = super::authority::verify_hook_authorities( |
| 2141 | hook.plugin_authority.as_ref(), |
| 2142 | hook.project_authority.as_ref(), |
| 2143 | ) { |
| 2144 | return HookResult { |
| 2145 | name: hook.name.clone(), |
| 2146 | background: false, |
| 2147 | strict: !hook.continue_on_error, |
| 2148 | success: false, |
| 2149 | exit_code: None, |
| 2150 | stdout: String::new(), |
| 2151 | stderr: String::new(), |
| 2152 | duration: started.elapsed(), |
| 2153 | error: Some(format!("Hook authority was denied: {reason}")), |
| 2154 | }; |
| 2155 | } |
| 2156 | let working_dir = self |
| 2157 | .config |
| 2158 | .working_dir |
| 2159 | .clone() |
| 2160 | .unwrap_or_else(|| self.default_working_dir.clone()); |
| 2161 | |
| 2162 | let timeout_secs = self.effective_timeout_secs(hook); |
| 2163 | let timeout = Duration::from_secs(timeout_secs); |
| 2164 | // This path always runs the hook in the foreground and awaits it, so |
| 2165 | // `continue_on_error = false` is a live "do not proceed without my |
| 2166 | // answer" for whichever call this result belongs to. |
| 2167 | let strict = !hook.continue_on_error; |
| 2168 | |
| 2169 | let stdin_bytes = match stdin_json.map(serde_json::to_vec).transpose() { |
| 2170 | Ok(bytes) => bytes, |
| 2171 | Err(e) => { |
| 2172 | return HookResult { |
| 2173 | name: hook.name.clone(), |
| 2174 | background: false, |
| 2175 | strict, |
| 2176 | success: false, |
| 2177 | exit_code: None, |
| 2178 | stdout: String::new(), |
| 2179 | stderr: String::new(), |
| 2180 | duration: started.elapsed(), |
| 2181 | error: Some(format!("Failed to encode hook stdin: {e}")), |
| 2182 | }; |
| 2183 | } |
| 2184 | }; |
| 2185 | |
| 2186 | let mut command = Self::build_shell_command(&hook.command); |
| 2187 | command |
| 2188 | .current_dir(&working_dir) |
| 2189 | .envs(env_vars) |
| 2190 | .stdout(Stdio::piped()) |
| 2191 | .stderr(Stdio::piped()) |
| 2192 | // A closed pipe is a portable EOF signal through shell layers. |
| 2193 | // Windows cmd/PowerShell can reopen console input when handed |
| 2194 | // NUL, so Stdio::null() is not sufficient when the parent test or |
| 2195 | // terminal still owns a live stdin handle. |
| 2196 | .stdin(Stdio::piped()); |
| 2197 | |
| 2198 | let (mut child, process_tree) = match spawn_hook_child(&mut command) { |
| 2199 | Ok(child) => child, |
| 2200 | Err(e) => { |
| 2201 | // Generic on purpose: this string reaches the deny receipt and |
| 2202 | // the TUI, and a spawn error can otherwise echo the resolved |
| 2203 | // command line or interpreter path back to the transcript. |
| 2204 | tracing::warn!( |
| 2205 | target: "hooks", |
| 2206 | hook = %sanitize_hook_label(hook.name.as_deref()), |
| 2207 | error = %e, |
| 2208 | "failed to start hook process" |
| 2209 | ); |
| 2210 | return HookResult { |
| 2211 | name: hook.name.clone(), |
| 2212 | background: false, |
| 2213 | strict, |
| 2214 | success: false, |
| 2215 | exit_code: None, |
| 2216 | stdout: String::new(), |
| 2217 | stderr: String::new(), |
| 2218 | duration: started.elapsed(), |
| 2219 | error: Some(spawn_failure_message(&e)), |
| 2220 | }; |
| 2221 | } |
| 2222 | }; |
| 2223 | |
| 2224 | let stdout_reader = match child |
| 2225 | .stdout |
| 2226 | .take() |
| 2227 | .map(|pipe| spawn_pipe_reader(pipe, "hook-stdout-reader")) |
| 2228 | .transpose() |
| 2229 | { |
| 2230 | Ok(reader) => reader, |
| 2231 | Err(error) => { |
| 2232 | tracing::warn!( |
| 2233 | target: "hooks", |
| 2234 | hook = %sanitize_hook_label(hook.name.as_deref()), |
| 2235 | error_kind = ?error.kind(), |
| 2236 | "failed to start hook stdout reader" |
| 2237 | ); |
| 2238 | terminate_and_reap(hook.name.as_deref(), &mut child, process_tree); |
| 2239 | return HookResult { |
| 2240 | name: hook.name.clone(), |
| 2241 | background: false, |
| 2242 | strict, |
| 2243 | success: false, |
| 2244 | exit_code: None, |
| 2245 | stdout: String::new(), |
| 2246 | stderr: String::new(), |
| 2247 | duration: started.elapsed(), |
| 2248 | error: Some("hook stdout reader could not be started".to_string()), |
| 2249 | }; |
| 2250 | } |
| 2251 | }; |
| 2252 | let stderr_reader = match child |
| 2253 | .stderr |
| 2254 | .take() |
| 2255 | .map(|pipe| spawn_pipe_reader(pipe, "hook-stderr-reader")) |
| 2256 | .transpose() |
| 2257 | { |
| 2258 | Ok(reader) => reader, |
| 2259 | Err(error) => { |
| 2260 | tracing::warn!( |
| 2261 | target: "hooks", |
| 2262 | hook = %sanitize_hook_label(hook.name.as_deref()), |
| 2263 | error_kind = ?error.kind(), |
| 2264 | "failed to start hook stderr reader" |
| 2265 | ); |
| 2266 | terminate_and_reap(hook.name.as_deref(), &mut child, process_tree); |
| 2267 | let _ = collect_reader(stdout_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2268 | return HookResult { |
| 2269 | name: hook.name.clone(), |
| 2270 | background: false, |
| 2271 | strict, |
| 2272 | success: false, |
| 2273 | exit_code: None, |
| 2274 | stdout: String::new(), |
| 2275 | stderr: String::new(), |
| 2276 | duration: started.elapsed(), |
| 2277 | error: Some("hook stderr reader could not be started".to_string()), |
| 2278 | }; |
| 2279 | } |
| 2280 | }; |
| 2281 | let _stdin_writer = match (stdin_bytes, child.stdin.take()) { |
| 2282 | (Some(bytes), Some(stdin)) => match spawn_stdin_writer(stdin, bytes) { |
| 2283 | Ok(writer) => Some(writer), |
| 2284 | Err(error) => { |
| 2285 | tracing::warn!( |
| 2286 | target: "hooks", |
| 2287 | hook = %sanitize_hook_label(hook.name.as_deref()), |
| 2288 | error_kind = ?error.kind(), |
| 2289 | "failed to start hook stdin writer" |
| 2290 | ); |
| 2291 | terminate_and_reap(hook.name.as_deref(), &mut child, process_tree); |
| 2292 | let _ = collect_reader(stdout_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2293 | let _ = collect_reader(stderr_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2294 | return HookResult { |
| 2295 | name: hook.name.clone(), |
| 2296 | background: false, |
| 2297 | strict, |
| 2298 | success: false, |
| 2299 | exit_code: None, |
| 2300 | stdout: String::new(), |
| 2301 | stderr: String::new(), |
| 2302 | duration: started.elapsed(), |
| 2303 | error: Some("hook stdin writer could not be started".to_string()), |
| 2304 | }; |
| 2305 | } |
| 2306 | }, |
| 2307 | _ => None, |
| 2308 | }; |
| 2309 | |
| 2310 | match child.wait_timeout(timeout) { |
| 2311 | Ok(Some(status)) => { |
| 2312 | drop(process_tree); |
| 2313 | HookResult { |
| 2314 | name: hook.name.clone(), |
| 2315 | background: false, |
| 2316 | strict, |
| 2317 | success: status.success(), |
| 2318 | exit_code: status.code(), |
| 2319 | stdout: collect_reader(stdout_reader, HOOK_PIPE_DRAIN_TIMEOUT), |
| 2320 | stderr: collect_reader(stderr_reader, HOOK_PIPE_DRAIN_TIMEOUT), |
| 2321 | duration: started.elapsed(), |
| 2322 | error: None, |
| 2323 | } |
| 2324 | } |
| 2325 | Ok(None) => { |
| 2326 | let reaped = terminate_and_reap(hook.name.as_deref(), &mut child, process_tree); |
| 2327 | let _ = collect_reader(stdout_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2328 | let _ = collect_reader(stderr_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2329 | HookResult { |
| 2330 | name: hook.name.clone(), |
| 2331 | background: false, |
| 2332 | strict, |
| 2333 | success: false, |
| 2334 | exit_code: None, |
| 2335 | stdout: String::new(), |
| 2336 | stderr: String::new(), |
| 2337 | duration: started.elapsed(), |
| 2338 | error: Some(if reaped { |
| 2339 | format!("Hook timed out after {timeout_secs}s") |
| 2340 | } else { |
| 2341 | // The gate still did not answer, and now we also cannot |
| 2342 | // prove the process is gone. Say the weaker thing. |
| 2343 | "hook could not be reaped after its timeout".to_string() |
| 2344 | }), |
| 2345 | } |
| 2346 | } |
| 2347 | Err(e) => { |
| 2348 | tracing::warn!( |
| 2349 | target: "hooks", |
| 2350 | hook = %sanitize_hook_label(hook.name.as_deref()), |
| 2351 | error = %e, |
| 2352 | "failed to wait for hook process" |
| 2353 | ); |
| 2354 | terminate_and_reap(hook.name.as_deref(), &mut child, process_tree); |
| 2355 | let _ = collect_reader(stdout_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2356 | let _ = collect_reader(stderr_reader, HOOK_PIPE_SHUTDOWN_TIMEOUT); |
| 2357 | HookResult { |
| 2358 | name: hook.name.clone(), |
| 2359 | background: false, |
| 2360 | strict, |
| 2361 | success: false, |
| 2362 | exit_code: None, |
| 2363 | stdout: String::new(), |
| 2364 | stderr: String::new(), |
| 2365 | duration: started.elapsed(), |
| 2366 | // Generic on purpose, like the spawn path: an OS wait error |
| 2367 | // can name the child and reaches the deny receipt. |
| 2368 | error: Some("Failed to wait for hook".to_string()), |
| 2369 | } |
| 2370 | } |
| 2371 | } |
| 2372 | } |
| 2373 | |
| 2374 | /// Execute a hook in the background (non-blocking) |
| 2375 | fn execute_background(&self, hook: &Hook, env_vars: &HashMap<String, String>) -> HookResult { |
| 2376 | self.execute_background_inner(hook, env_vars, None) |
| 2377 | } |
| 2378 | |
| 2379 | fn execute_background_with_stdin( |
| 2380 | &self, |
| 2381 | hook: &Hook, |
| 2382 | env_vars: &HashMap<String, String>, |
| 2383 | stdin_json: &serde_json::Value, |
| 2384 | ) -> HookResult { |
| 2385 | self.execute_background_inner(hook, env_vars, Some(stdin_json)) |
| 2386 | } |
| 2387 | |
| 2388 | fn execute_background_inner( |
| 2389 | &self, |
| 2390 | hook: &Hook, |
| 2391 | env_vars: &HashMap<String, String>, |
| 2392 | stdin_json: Option<&serde_json::Value>, |
| 2393 | ) -> HookResult { |
| 2394 | let started = Instant::now(); |
| 2395 | if let Err(reason) = super::authority::verify_hook_authorities( |
| 2396 | hook.plugin_authority.as_ref(), |
| 2397 | hook.project_authority.as_ref(), |
| 2398 | ) { |
| 2399 | return HookResult { |
| 2400 | name: hook.name.clone(), |
| 2401 | background: true, |
| 2402 | strict: false, |
| 2403 | success: false, |
| 2404 | exit_code: None, |
| 2405 | stdout: String::new(), |
| 2406 | stderr: String::new(), |
| 2407 | duration: started.elapsed(), |
| 2408 | error: Some(format!("Hook authority was denied: {reason}")), |
| 2409 | }; |
| 2410 | } |
| 2411 | let working_dir = self |
| 2412 | .config |
| 2413 | .working_dir |
| 2414 | .clone() |
| 2415 | .unwrap_or_else(|| self.default_working_dir.clone()); |
| 2416 | |
| 2417 | let stdin_bytes = match stdin_json.map(serde_json::to_vec).transpose() { |
| 2418 | Ok(bytes) => bytes, |
| 2419 | Err(e) => { |
| 2420 | return HookResult { |
| 2421 | name: hook.name.clone(), |
| 2422 | background: true, |
| 2423 | strict: false, |
| 2424 | success: false, |
| 2425 | exit_code: None, |
| 2426 | stdout: String::new(), |
| 2427 | stderr: String::new(), |
| 2428 | duration: started.elapsed(), |
| 2429 | error: Some(format!("Failed to encode hook stdin: {e}")), |
| 2430 | }; |
| 2431 | } |
| 2432 | }; |
| 2433 | let submission = self.background_supervisor.submit(BackgroundHookJob { |
| 2434 | command: hook.command.clone(), |
| 2435 | env: env_vars.clone(), |
| 2436 | working_dir, |
| 2437 | stdin_bytes, |
| 2438 | label: sanitize_hook_label(hook.name.as_deref()), |
| 2439 | timeout: Duration::from_secs(self.effective_timeout_secs(hook)), |
| 2440 | plugin_authority: hook.plugin_authority.clone(), |
| 2441 | project_authority: hook.project_authority.clone(), |
| 2442 | }); |
| 2443 | |
| 2444 | // The result describes the bounded submission, not the run: no caller |
| 2445 | // can mistake "queued" for "exited 0". |
| 2446 | HookResult { |
| 2447 | name: hook.name.clone(), |
| 2448 | background: true, |
| 2449 | strict: false, |
| 2450 | success: submission.is_ok(), |
| 2451 | exit_code: None, |
| 2452 | stdout: String::new(), |
| 2453 | stderr: String::new(), |
| 2454 | duration: started.elapsed(), |
| 2455 | error: submission.err().map(|failure| match failure { |
| 2456 | BackgroundSupervisorFailure::Full => { |
| 2457 | "background hook supervisor queue is full".to_string() |
| 2458 | } |
| 2459 | BackgroundSupervisorFailure::Disconnected => { |
| 2460 | "background hook supervisor is unavailable".to_string() |
| 2461 | } |
| 2462 | }), |
| 2463 | } |
| 2464 | } |
| 2465 | |
| 2466 | /// The timeout actually applied to a hook, foreground or background. |
| 2467 | /// |
| 2468 | /// `[hooks].default_timeout_secs` *replaces* the per-hook value when set; |
| 2469 | /// that is the shipped behavior and is documented as such in |
| 2470 | /// `docs/HOOKS.md`. |
| 2471 | fn effective_timeout_secs(&self, hook: &Hook) -> u64 { |
| 2472 | self.config.effective_timeout_secs(hook) |
| 2473 | } |
| 2474 | } |
| 2475 | |
| 2476 | /// Classify a tool call for `condition = { type = "tool_category", … }`. |
| 2477 | /// |
| 2478 | /// Categories are `shell`, `file_write`, `safe`, and `other`, as documented in |
| 2479 | /// `docs/HOOKS.md`. This must be kept in step with the names the registry |
| 2480 | /// actually registers: before 2026-08-04 the map knew only the retired |
| 2481 | /// `exec_shell`/`write_file`/`read_file` spellings, so EVERY live call fell |
| 2482 | /// through to `other` and a `tool_category` **deny** hook silently never |
| 2483 | /// fired — the exact failure `docs/HOOKS.md` warns about ("a deny gate the |
| 2484 | /// operator believes is armed"). |
| 2485 | /// |
| 2486 | /// `File`, `Git`, and `Run` are multi-action, so the action decides the |
| 2487 | /// category: a `File` read is `safe` while a `File` write is `file_write`. |
| 2488 | /// An unparseable or absent argument blob is treated as the tool's most |
| 2489 | /// dangerous action, because a gate that cannot see the action must not |
| 2490 | /// assume the harmless one. |
| 2491 | fn tool_category_for(tool_name: &str, tool_args: Option<&str>) -> &'static str { |
| 2492 | let action = tool_args |
| 2493 | .and_then(|raw| serde_json::from_str::<serde_json::Value>(raw).ok()) |
| 2494 | .and_then(|value| { |
| 2495 | value |
| 2496 | .get("action") |
| 2497 | .and_then(serde_json::Value::as_str) |
| 2498 | .map(str::to_ascii_lowercase) |
| 2499 | }); |
| 2500 | |
| 2501 | match tool_name { |
| 2502 | // The shell surface. `exec_shell` is retired but kept here because |
| 2503 | // `shell.rs` still stamps it for the `shell_env` hook event. |
| 2504 | "bash" | "Bash" | "exec_shell" => "shell", |
| 2505 | // The lowercase primitives ship without an action envelope. |
| 2506 | "read" | "todo_write" => "safe", |
| 2507 | "write" | "edit" => "file_write", |
| 2508 | "File" | "file" => match action.as_deref() { |
| 2509 | Some("read" | "list" | "search_name" | "search_content") => "safe", |
| 2510 | // write/edit/patch, and the unknown-action case, are writes. |
| 2511 | _ => "file_write", |
| 2512 | }, |
| 2513 | "apply_patch" => "file_write", |
| 2514 | "Git" | "git" => match action.as_deref() { |
| 2515 | // Every shipped Git action is read-only today; classify by action |
| 2516 | // anyway so adding a mutating one cannot silently inherit `safe`. |
| 2517 | Some("status" | "diff" | "log" | "show" | "blame" | "commit_plan") => "safe", |
| 2518 | _ => "other", |
| 2519 | }, |
| 2520 | // `Run` executes test/verifier commands — closer to shell than safe. |
| 2521 | "Run" | "run" => "shell", |
| 2522 | _ => "other", |
| 2523 | } |
| 2524 | } |
| 2525 | |
| 2526 | const HOOK_PIPE_DRAIN_TIMEOUT: Duration = Duration::from_secs(5); |
| 2527 | const HOOK_PIPE_SHUTDOWN_TIMEOUT: Duration = Duration::from_millis(250); |
| 2528 | /// How long the timeout path waits for the killed child to be reaped. |
| 2529 | /// |
| 2530 | /// The wait after a kill is *bounded* rather than unbounded: `child.wait()` |
| 2531 | /// blocks forever if the kill did not take (a `SIGKILL`-immune uninterruptible |
| 2532 | /// state on Unix, a `TerminateJobObject` that a protected process survived on |
| 2533 | /// Windows), and that turned "this hook has a 30s budget" into a hung turn. |
| 2534 | const HOOK_REAP_TIMEOUT: Duration = Duration::from_secs(2); |
| 2535 | #[cfg(windows)] |
| 2536 | const WINDOWS_TASKKILL_TIMEOUT: Duration = Duration::from_secs(2); |
| 2537 | |
| 2538 | /// Kill the hook's process tree and wait, briefly, for the corpse. |
| 2539 | /// |
| 2540 | /// Termination is best-effort by nature — the OS owns whether a kill lands. |
| 2541 | /// What is guaranteed here is that *this* thread stops waiting: the |
| 2542 | /// containment guard is dropped first (which re-signals the Unix process group |
| 2543 | /// and closes the kill-on-close Windows Job Object), then the reap gets one |
| 2544 | /// bounded window. Returns `false` when the child could not be confirmed dead, |
| 2545 | /// so the caller can report the weaker claim instead of asserting cleanup. |
| 2546 | fn terminate_and_reap( |
| 2547 | hook_name: Option<&str>, |
| 2548 | child: &mut Child, |
| 2549 | process_tree: HookProcessTree, |
| 2550 | ) -> bool { |
| 2551 | process_tree.terminate(child); |
| 2552 | // Drop before the wait, not after: on Windows this closes the Job Object |
| 2553 | // and is itself a kill, and on Unix it re-signals the group. Waiting first |
| 2554 | // would delay the very thing meant to make the wait short. |
| 2555 | drop(process_tree); |
| 2556 | match child.wait_timeout(HOOK_REAP_TIMEOUT) { |
| 2557 | Ok(Some(_)) => true, |
| 2558 | Ok(None) => { |
| 2559 | tracing::warn!( |
| 2560 | target: "hooks", |
| 2561 | hook = %sanitize_hook_label(hook_name), |
| 2562 | reap_timeout_secs = HOOK_REAP_TIMEOUT.as_secs(), |
| 2563 | "hook process did not exit after its tree was killed; abandoning the reap" |
| 2564 | ); |
| 2565 | false |
| 2566 | } |
| 2567 | Err(error) => { |
| 2568 | tracing::warn!( |
| 2569 | target: "hooks", |
| 2570 | hook = %sanitize_hook_label(hook_name), |
| 2571 | %error, |
| 2572 | "failed to reap killed hook process" |
| 2573 | ); |
| 2574 | false |
| 2575 | } |
| 2576 | } |
| 2577 | } |
| 2578 | |
| 2579 | fn spawn_pipe_reader( |
| 2580 | mut pipe: impl Read + Send + 'static, |
| 2581 | worker_name: &str, |
| 2582 | ) -> std::io::Result<Receiver<String>> { |
| 2583 | let (tx, rx) = mpsc::channel(); |
| 2584 | std::thread::Builder::new() |
| 2585 | .name(worker_name.to_string()) |
| 2586 | .spawn(move || { |
| 2587 | let mut retained = Vec::with_capacity(HOOK_PIPE_CAPTURE_MAX_BYTES.min(8 * 1024)); |
| 2588 | let mut chunk = [0_u8; 8 * 1024]; |
| 2589 | let mut truncated = false; |
| 2590 | loop { |
| 2591 | match pipe.read(&mut chunk) { |
| 2592 | Ok(0) => break, |
| 2593 | Ok(read) => { |
| 2594 | let remaining = HOOK_PIPE_CAPTURE_MAX_BYTES.saturating_sub(retained.len()); |
| 2595 | let keep = remaining.min(read); |
| 2596 | retained.extend_from_slice(&chunk[..keep]); |
| 2597 | truncated |= keep < read; |
| 2598 | } |
| 2599 | Err(error) => { |
| 2600 | tracing::warn!(target: "hooks", %error, "failed while draining hook pipe"); |
| 2601 | break; |
| 2602 | } |
| 2603 | } |
| 2604 | } |
| 2605 | let mut output = String::from_utf8_lossy(&retained).into_owned(); |
| 2606 | if truncated { |
| 2607 | output.push_str("…[truncated]"); |
| 2608 | } |
| 2609 | let _ = tx.send(output); |
| 2610 | }) |
| 2611 | .map(|_| rx) |
| 2612 | } |
| 2613 | |
| 2614 | fn collect_reader(reader: Option<Receiver<String>>, timeout: Duration) -> String { |
| 2615 | let Some(reader) = reader else { |
| 2616 | return String::new(); |
| 2617 | }; |
| 2618 | match reader.recv_timeout(timeout) { |
| 2619 | Ok(output) => output, |
| 2620 | Err(RecvTimeoutError::Timeout) => { |
| 2621 | tracing::warn!( |
| 2622 | ?timeout, |
| 2623 | "hook pipe reader did not finish after process cleanup" |
| 2624 | ); |
| 2625 | String::new() |
| 2626 | } |
| 2627 | Err(RecvTimeoutError::Disconnected) => String::new(), |
| 2628 | } |
| 2629 | } |
| 2630 | |
| 2631 | fn spawn_stdin_writer( |
| 2632 | mut stdin: std::process::ChildStdin, |
| 2633 | mut bytes: Vec<u8>, |
| 2634 | ) -> std::io::Result<JoinHandle<()>> { |
| 2635 | std::thread::Builder::new() |
| 2636 | .name("hook-stdin-writer".to_string()) |
| 2637 | .spawn(move || { |
| 2638 | bytes.push(b'\n'); |
| 2639 | let _ = stdin.write_all(&bytes); |
| 2640 | let _ = stdin.flush(); |
| 2641 | }) |
| 2642 | } |
| 2643 | |
| 2644 | fn bounded_message_submit_metadata(value: Option<&str>, max_bytes: usize) -> Option<String> { |
| 2645 | value.map(|value| truncate_env_value(value, max_bytes)) |
| 2646 | } |
| 2647 | |
| 2648 | fn build_message_submit_payload( |
| 2649 | context: &HookContext, |
| 2650 | text: &str, |
| 2651 | original_bytes: usize, |
| 2652 | truncated: bool, |
| 2653 | metadata_max_bytes: Option<usize>, |
| 2654 | ) -> serde_json::Value { |
| 2655 | let mut payload = json!({ |
| 2656 | "event": HookEvent::MessageSubmit.as_str(), |
| 2657 | "text": text, |
| 2658 | "text_bytes": text.len(), |
| 2659 | "text_original_bytes": original_bytes, |
| 2660 | "text_truncated": truncated, |
| 2661 | }); |
| 2662 | if let Some(max_bytes) = metadata_max_bytes { |
| 2663 | let object = payload |
| 2664 | .as_object_mut() |
| 2665 | .expect("message_submit payload is an object"); |
| 2666 | object.insert( |
| 2667 | "session_id".to_string(), |
| 2668 | json!(bounded_message_submit_metadata( |
| 2669 | context.session_id.as_deref(), |
| 2670 | max_bytes |
| 2671 | )), |
| 2672 | ); |
| 2673 | object.insert( |
| 2674 | "workspace".to_string(), |
| 2675 | json!(bounded_message_submit_metadata( |
| 2676 | context.workspace.as_ref().and_then(|path| path.to_str()), |
| 2677 | max_bytes |
| 2678 | )), |
| 2679 | ); |
| 2680 | object.insert( |
| 2681 | "mode".to_string(), |
| 2682 | json!(bounded_message_submit_metadata( |
| 2683 | context.mode.as_deref(), |
| 2684 | max_bytes |
| 2685 | )), |
| 2686 | ); |
| 2687 | object.insert( |
| 2688 | "model".to_string(), |
| 2689 | json!(bounded_message_submit_metadata( |
| 2690 | context.model.as_deref(), |
| 2691 | max_bytes |
| 2692 | )), |
| 2693 | ); |
| 2694 | object.insert("total_tokens".to_string(), json!(context.total_tokens)); |
| 2695 | } |
| 2696 | payload |
| 2697 | } |
| 2698 | |
| 2699 | fn encoded_message_submit_payload_fits(payload: &serde_json::Value) -> bool { |
| 2700 | serde_json::to_vec(payload) |
| 2701 | .is_ok_and(|bytes| bytes.len() <= HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES) |
| 2702 | } |
| 2703 | |
| 2704 | fn finalize_message_submit_payload( |
| 2705 | payload: serde_json::Value, |
| 2706 | original_bytes: usize, |
| 2707 | ) -> serde_json::Value { |
| 2708 | if encoded_message_submit_payload_fits(&payload) { |
| 2709 | return payload; |
| 2710 | } |
| 2711 | |
| 2712 | // Serialization of a `Value` is infallible in practice, but the size |
| 2713 | // boundary is security-sensitive. If an invariant above ever regresses, |
| 2714 | // discard all user text and diagnostics rather than handing an oversized |
| 2715 | // document to a hook process. |
| 2716 | tracing::error!(target: "hooks", "message_submit payload fitter exceeded its hard byte cap"); |
| 2717 | let fail_closed = build_message_submit_payload( |
| 2718 | &HookContext::new(), |
| 2719 | "", |
| 2720 | original_bytes, |
| 2721 | original_bytes != 0, |
| 2722 | None, |
| 2723 | ); |
| 2724 | assert!( |
| 2725 | encoded_message_submit_payload_fits(&fail_closed), |
| 2726 | "minimal message_submit payload must fit the hard byte cap" |
| 2727 | ); |
| 2728 | fail_closed |
| 2729 | } |
| 2730 | |
| 2731 | /// Build the one canonical `message_submit` stdin document. |
| 2732 | /// |
| 2733 | /// Every producer — immediate input, restored queue entries, merged steers, |
| 2734 | /// and hook-to-hook replacements — crosses this serialization boundary. The |
| 2735 | /// largest UTF-8-safe text prefix that keeps the *serialized JSON* within the |
| 2736 | /// byte ceiling is retained, and explicit metadata tells the hook exactly |
| 2737 | /// what was clipped. |
| 2738 | pub(crate) fn message_submit_payload(context: &HookContext, text: &str) -> serde_json::Value { |
| 2739 | // Diagnostic metadata is useful but never allowed to crowd the actual |
| 2740 | // gate input out of the hard byte budget. Control-heavy strings can grow |
| 2741 | // sixfold when JSON-escaped, so try progressively smaller snapshots and |
| 2742 | // finally omit diagnostics altogether. |
| 2743 | let metadata_max_bytes = [ |
| 2744 | Some(HOOK_MESSAGE_SUBMIT_METADATA_MAX_BYTES), |
| 2745 | Some(1_024), |
| 2746 | Some(256), |
| 2747 | None, |
| 2748 | ] |
| 2749 | .into_iter() |
| 2750 | .find(|metadata_max_bytes| { |
| 2751 | encoded_message_submit_payload_fits(&build_message_submit_payload( |
| 2752 | context, |
| 2753 | "", |
| 2754 | text.len(), |
| 2755 | !text.is_empty(), |
| 2756 | *metadata_max_bytes, |
| 2757 | )) |
| 2758 | }) |
| 2759 | .unwrap_or(None); |
| 2760 | |
| 2761 | if text.len() <= HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES { |
| 2762 | let complete = |
| 2763 | build_message_submit_payload(context, text, text.len(), false, metadata_max_bytes); |
| 2764 | if encoded_message_submit_payload_fits(&complete) { |
| 2765 | return finalize_message_submit_payload(complete, text.len()); |
| 2766 | } |
| 2767 | } |
| 2768 | |
| 2769 | // No candidate can retain more raw bytes than the full JSON budget. Build |
| 2770 | // at most that many UTF-8 boundaries even if a restored queue entry is |
| 2771 | // unexpectedly enormous. |
| 2772 | let raw_prefix_cap = text.len().min(HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES); |
| 2773 | let mut utf8_ends = Vec::with_capacity(raw_prefix_cap.saturating_add(1)); |
| 2774 | utf8_ends.push(0); |
| 2775 | utf8_ends.extend( |
| 2776 | text.char_indices() |
| 2777 | .map(|(index, ch)| index + ch.len_utf8()) |
| 2778 | .take_while(|end| *end <= raw_prefix_cap), |
| 2779 | ); |
| 2780 | |
| 2781 | let mut lower = 0usize; |
| 2782 | let mut upper = utf8_ends.len(); |
| 2783 | while lower < upper { |
| 2784 | let middle = lower + (upper - lower) / 2; |
| 2785 | let end = utf8_ends[middle]; |
| 2786 | let candidate = build_message_submit_payload( |
| 2787 | context, |
| 2788 | &text[..end], |
| 2789 | text.len(), |
| 2790 | true, |
| 2791 | metadata_max_bytes, |
| 2792 | ); |
| 2793 | let fits = encoded_message_submit_payload_fits(&candidate); |
| 2794 | if fits { |
| 2795 | lower = middle + 1; |
| 2796 | } else { |
| 2797 | upper = middle; |
| 2798 | } |
| 2799 | } |
| 2800 | |
| 2801 | let retained_end = utf8_ends[lower.saturating_sub(1)]; |
| 2802 | finalize_message_submit_payload( |
| 2803 | build_message_submit_payload( |
| 2804 | context, |
| 2805 | &text[..retained_end], |
| 2806 | text.len(), |
| 2807 | true, |
| 2808 | metadata_max_bytes, |
| 2809 | ), |
| 2810 | text.len(), |
| 2811 | ) |
| 2812 | } |
| 2813 | |
| 2814 | pub fn turn_end_payload(input: TurnEndPayloadInput<'_>) -> serde_json::Value { |
| 2815 | let bounded_error = input |
| 2816 | .error |
| 2817 | .map(|error| sanitize_hook_text(error, HOOK_TURN_ERROR_MAX_CHARS)); |
| 2818 | json!({ |
| 2819 | "event": HookEvent::TurnEnd.as_str(), |
| 2820 | "session_id": input.context.session_id.as_deref(), |
| 2821 | "workspace": input.context.workspace.as_ref().map(|path| path.display().to_string()), |
| 2822 | "mode": input.context.mode.as_deref(), |
| 2823 | "created_at": input.created_at.to_rfc3339(), |
| 2824 | "model_backed": input.model_backed, |
| 2825 | "provider": input.provider, |
| 2826 | "billing_surface": input.billing_surface, |
| 2827 | "model": input.model.or(input.context.model.as_deref()), |
| 2828 | "turn_id": input.turn_id, |
| 2829 | "status": input.status, |
| 2830 | "error": bounded_error, |
| 2831 | "duration_ms": duration_ms_saturating(input.duration), |
| 2832 | "usage": { |
| 2833 | "input_tokens": input.usage.input_tokens, |
| 2834 | "output_tokens": input.usage.output_tokens, |
| 2835 | "prompt_cache_hit_tokens": input.usage.prompt_cache_hit_tokens, |
| 2836 | "prompt_cache_miss_tokens": input.usage.prompt_cache_miss_tokens, |
| 2837 | "prompt_cache_write_tokens": input.usage.prompt_cache_write_tokens, |
| 2838 | "reasoning_tokens": input.usage.reasoning_tokens, |
| 2839 | "reasoning_replay_tokens": input.usage.reasoning_replay_tokens, |
| 2840 | }, |
| 2841 | "totals": { |
| 2842 | "session_tokens": input.totals.session_tokens, |
| 2843 | "conversation_tokens": input.totals.conversation_tokens, |
| 2844 | "input_tokens": input.totals.input_tokens, |
| 2845 | "output_tokens": input.totals.output_tokens, |
| 2846 | }, |
| 2847 | "tool_count": input.tool_count, |
| 2848 | "queued_message_count": input.queued_message_count, |
| 2849 | "stop_hook_active": false, |
| 2850 | }) |
| 2851 | } |
| 2852 | |
| 2853 | fn duration_ms_saturating(duration: Duration) -> u64 { |
| 2854 | u64::try_from(duration.as_millis()).unwrap_or(u64::MAX) |
| 2855 | } |
| 2856 | |
| 2857 | fn parse_message_submit_stdout(stdout: &str) -> MessageSubmitStdout { |
| 2858 | let trimmed = stdout.trim(); |
| 2859 | if trimmed.is_empty() { |
| 2860 | return MessageSubmitStdout::Unchanged; |
| 2861 | } |
| 2862 | |
| 2863 | let value: serde_json::Value = match serde_json::from_str(trimmed) { |
| 2864 | Ok(value) => value, |
| 2865 | Err(e) => return MessageSubmitStdout::Invalid(format!("invalid JSON: {e}")), |
| 2866 | }; |
| 2867 | |
| 2868 | let Some(object) = value.as_object() else { |
| 2869 | return MessageSubmitStdout::Invalid("stdout JSON must be an object".to_string()); |
| 2870 | }; |
| 2871 | |
| 2872 | match object.get("text") { |
| 2873 | Some(serde_json::Value::String(text)) if !text.is_empty() => { |
| 2874 | if text.chars().count() > HOOK_MESSAGE_REPLACEMENT_MAX_CHARS { |
| 2875 | MessageSubmitStdout::Invalid(format!( |
| 2876 | "stdout `text` field exceeds {HOOK_MESSAGE_REPLACEMENT_MAX_CHARS} characters" |
| 2877 | )) |
| 2878 | } else { |
| 2879 | MessageSubmitStdout::Replaced(text.clone()) |
| 2880 | } |
| 2881 | } |
| 2882 | Some(serde_json::Value::String(_)) => { |
| 2883 | MessageSubmitStdout::Invalid("stdout `text` field must not be empty".to_string()) |
| 2884 | } |
| 2885 | Some(_) => MessageSubmitStdout::Invalid("stdout `text` field must be a string".to_string()), |
| 2886 | None => MessageSubmitStdout::Unchanged, |
| 2887 | } |
| 2888 | } |
| 2889 | |
| 2890 | fn message_submit_continue_warning(result: &HookResult) -> Option<String> { |
| 2891 | message_submit_stdout_reason(&result.stdout) |
| 2892 | .or_else(|| { |
| 2893 | Some(generic_unavailable_detail(result.error.as_deref())) |
| 2894 | .filter(|detail| detail != "hook returned no verdict") |
| 2895 | }) |
| 2896 | .or_else(|| { |
| 2897 | result |
| 2898 | .observed_exit_code() |
| 2899 | .map(|code| format!("message_submit hook exited with code {code}")) |
| 2900 | }) |
| 2901 | } |
| 2902 | |
| 2903 | fn message_submit_block_reason(result: &HookResult, fallback: &str) -> String { |
| 2904 | if let Some(reason) = message_submit_stdout_reason(&result.stdout) { |
| 2905 | return reason; |
| 2906 | } |
| 2907 | let detail = generic_unavailable_detail(result.error.as_deref()); |
| 2908 | if detail != "hook returned no verdict" { |
| 2909 | return detail; |
| 2910 | } |
| 2911 | fallback.to_string() |
| 2912 | } |
| 2913 | |
| 2914 | fn message_submit_stdout_reason(stdout: &str) -> Option<String> { |
| 2915 | let value: serde_json::Value = serde_json::from_str(stdout.trim()).ok()?; |
| 2916 | value |
| 2917 | .get("reason") |
| 2918 | .and_then(serde_json::Value::as_str) |
| 2919 | .map(sanitize_hook_denial_reason) |
| 2920 | } |
| 2921 | |
| 2922 | /// Largest single `shell_env` value that is accepted, in bytes. |
| 2923 | const SHELL_ENV_VALUE_MAX_BYTES: usize = 32 * 1024; |
| 2924 | /// Largest total `shell_env` contribution from one hook, in bytes of |
| 2925 | /// `KEY` + `VALUE`. Past this, later entries from that hook are dropped. |
| 2926 | const SHELL_ENV_TOTAL_MAX_BYTES: usize = 256 * 1024; |
| 2927 | |
| 2928 | /// Whether a parsed name is usable as an environment variable name. |
| 2929 | /// |
| 2930 | /// `Command::env` **panics** on a key containing a NUL byte or `=`, and an |
| 2931 | /// empty key is meaningless, so an entry that fails this check is dropped |
| 2932 | /// rather than carried into `exec_shell`'s environment. A `shell_env` hook is |
| 2933 | /// a normal process whose stdout can contain anything — including a NUL |
| 2934 | /// straight out of a binary — and "the hook printed something odd" must never |
| 2935 | /// become "Codewhale aborted the tool call". |
| 2936 | fn is_valid_env_key(key: &str) -> bool { |
| 2937 | !key.is_empty() |
| 2938 | && !key.contains('=') |
| 2939 | && !key.chars().any(|c| c == '\0' || c.is_control() || c == ' ') |
| 2940 | } |
| 2941 | |
| 2942 | /// Parse `KEY=VALUE\n` lines from a `shell_env` hook's stdout into a map. |
| 2943 | /// |
| 2944 | /// Tolerated: blank lines, leading whitespace, `#` comment lines (ignored), |
| 2945 | /// `export KEY=VALUE` (the `export ` prefix is dropped), surrounding quotes |
| 2946 | /// on the value. Lines without `=` are silently dropped — easier than |
| 2947 | /// failing the whole hook for one stray line of human-friendly output. |
| 2948 | /// Values are otherwise taken verbatim; we don't run them through a shell |
| 2949 | /// for variable expansion to avoid surprises. |
| 2950 | /// |
| 2951 | /// Rejected: entries whose key is unusable ([`is_valid_env_key`]), values |
| 2952 | /// containing a NUL byte, values over [`SHELL_ENV_VALUE_MAX_BYTES`], and |
| 2953 | /// anything past [`SHELL_ENV_TOTAL_MAX_BYTES`] of accumulated output. Each |
| 2954 | /// drop is logged by key name only — never by value. |
| 2955 | fn parse_env_lines(stdout: &str) -> HashMap<String, String> { |
| 2956 | let mut out: HashMap<String, String> = HashMap::new(); |
| 2957 | let mut total_bytes = 0usize; |
| 2958 | for raw in stdout.lines() { |
| 2959 | let line = raw.trim(); |
| 2960 | if line.is_empty() || line.starts_with('#') { |
| 2961 | continue; |
| 2962 | } |
| 2963 | let line = line.strip_prefix("export ").unwrap_or(line); |
| 2964 | let Some((key, value)) = line.split_once('=') else { |
| 2965 | continue; |
| 2966 | }; |
| 2967 | let key = key.trim(); |
| 2968 | if !is_valid_env_key(key) { |
| 2969 | tracing::warn!( |
| 2970 | target: "hooks", |
| 2971 | "shell_env hook produced an unusable variable name; dropping the entry" |
| 2972 | ); |
| 2973 | continue; |
| 2974 | } |
| 2975 | let value = value.trim(); |
| 2976 | let stripped = value |
| 2977 | .strip_prefix('"') |
| 2978 | .and_then(|v| v.strip_suffix('"')) |
| 2979 | .or_else(|| value.strip_prefix('\'').and_then(|v| v.strip_suffix('\''))) |
| 2980 | .unwrap_or(value); |
| 2981 | if stripped.contains('\0') { |
| 2982 | tracing::warn!( |
| 2983 | target: "hooks", |
| 2984 | key, |
| 2985 | "shell_env value contains a NUL byte; dropping the entry" |
| 2986 | ); |
| 2987 | continue; |
| 2988 | } |
| 2989 | if stripped.len() > SHELL_ENV_VALUE_MAX_BYTES { |
| 2990 | tracing::warn!( |
| 2991 | target: "hooks", |
| 2992 | key, |
| 2993 | limit = SHELL_ENV_VALUE_MAX_BYTES, |
| 2994 | "shell_env value exceeds the per-value limit; dropping the entry" |
| 2995 | ); |
| 2996 | continue; |
| 2997 | } |
| 2998 | let entry_bytes = key.len() + stripped.len(); |
| 2999 | if total_bytes.saturating_add(entry_bytes) > SHELL_ENV_TOTAL_MAX_BYTES { |
| 3000 | tracing::warn!( |
| 3001 | target: "hooks", |
| 3002 | key, |
| 3003 | limit = SHELL_ENV_TOTAL_MAX_BYTES, |
| 3004 | "shell_env output exceeds the total limit; dropping the remaining entries" |
| 3005 | ); |
| 3006 | break; |
| 3007 | } |
| 3008 | total_bytes += entry_bytes; |
| 3009 | out.insert(key.to_string(), stripped.to_string()); |
| 3010 | } |
| 3011 | out |
| 3012 | } |
| 3013 | |
| 3014 | // === Unit Tests === |
| 3015 | |
| 3016 | #[cfg(test)] |
| 3017 | mod tests { |
| 3018 | use super::*; |
| 3019 | use crate::test_support::{EnvVarGuard, lock_test_env}; |
| 3020 | use std::collections::HashMap; |
| 3021 | use std::path::{Path, PathBuf}; |
| 3022 | |
| 3023 | fn trust_workspace_for_project_hooks(workspace: &Path, config_path: &Path) -> EnvVarGuard { |
| 3024 | let guard = EnvVarGuard::set("CODEWHALE_CONFIG_PATH", config_path); |
| 3025 | crate::config::save_workspace_trust(workspace).expect("save workspace trust"); |
| 3026 | guard |
| 3027 | } |
| 3028 | |
| 3029 | #[test] |
| 3030 | fn config_types_are_available_from_config_module() { |
| 3031 | let hook = crate::hooks::config::Hook::new( |
| 3032 | crate::hooks::config::HookEvent::SessionStart, |
| 3033 | "echo ready", |
| 3034 | ); |
| 3035 | let config = crate::hooks::config::HooksConfig { |
| 3036 | enabled: true, |
| 3037 | hooks: vec![hook], |
| 3038 | ..Default::default() |
| 3039 | }; |
| 3040 | |
| 3041 | let hooks = config.hooks_for_event(crate::hooks::config::HookEvent::SessionStart); |
| 3042 | |
| 3043 | assert_eq!(hooks.len(), 1); |
| 3044 | } |
| 3045 | |
| 3046 | #[cfg(unix)] |
| 3047 | #[test] |
| 3048 | fn plugin_hook_runs_after_restart_and_process_spawn_rechecks_revocation() { |
| 3049 | let _lock = lock_test_env(); |
| 3050 | let fixture = crate::plugins::test_fixture::DeclarativePluginFixture::new(); |
| 3051 | let config = HooksConfig::load_with_project_and_plugins( |
| 3052 | HooksConfig { |
| 3053 | enabled: true, |
| 3054 | ..HooksConfig::default() |
| 3055 | }, |
| 3056 | &fixture.workspace, |
| 3057 | Some(&fixture.registry), |
| 3058 | ); |
| 3059 | assert!(config.problems.is_empty(), "{:?}", config.problems); |
| 3060 | assert_eq!(config.hooks.len(), 1); |
| 3061 | assert!(config.hooks[0].plugin_authority.is_some()); |
| 3062 | let executor = HookExecutor::new(config, fixture.workspace.clone()); |
| 3063 | let context = HookContext::new().with_workspace(fixture.workspace.clone()); |
| 3064 | |
| 3065 | let ran = executor.execute(HookEvent::SessionStart, &context); |
| 3066 | assert_eq!(ran.len(), 1); |
| 3067 | assert!(ran[0].success, "{:?}", ran[0].error); |
| 3068 | assert_eq!( |
| 3069 | std::fs::read_to_string(&fixture.marker).expect("plugin hook marker"), |
| 3070 | "plugin-hook-ran" |
| 3071 | ); |
| 3072 | std::fs::remove_file(&fixture.marker).expect("clear marker"); |
| 3073 | |
| 3074 | let inactive = fixture.revoke_from_fresh_registry(); |
| 3075 | let denied = executor.execute(HookEvent::SessionStart, &context); |
| 3076 | assert_eq!(denied.len(), 1); |
| 3077 | assert!(!denied[0].success); |
| 3078 | assert!( |
| 3079 | denied[0] |
| 3080 | .error |
| 3081 | .as_deref() |
| 3082 | .is_some_and(|error| error.contains("authority was denied")), |
| 3083 | "{:?}", |
| 3084 | denied[0].error |
| 3085 | ); |
| 3086 | assert!( |
| 3087 | !fixture.marker.exists(), |
| 3088 | "revoked hook must be denied before process spawn" |
| 3089 | ); |
| 3090 | |
| 3091 | let reloaded = HooksConfig::load_with_project_and_plugins( |
| 3092 | HooksConfig { |
| 3093 | enabled: true, |
| 3094 | ..HooksConfig::default() |
| 3095 | }, |
| 3096 | &fixture.workspace, |
| 3097 | Some(&inactive), |
| 3098 | ); |
| 3099 | assert!( |
| 3100 | reloaded.hooks.is_empty(), |
| 3101 | "reload removes the revoked plugin Hook" |
| 3102 | ); |
| 3103 | } |
| 3104 | |
| 3105 | #[cfg(unix)] |
| 3106 | #[test] |
| 3107 | fn queued_plugin_hook_rechecks_revocation_at_dequeue() { |
| 3108 | let _lock = lock_test_env(); |
| 3109 | let fixture = crate::plugins::test_fixture::DeclarativePluginFixture::new(); |
| 3110 | let blocker_one = Hook::new(HookEvent::SessionStart, "sleep 1").background(); |
| 3111 | let blocker_two = Hook::new(HookEvent::SessionStart, "sleep 1").background(); |
| 3112 | let mut config = HooksConfig::load_with_project_and_plugins( |
| 3113 | HooksConfig { |
| 3114 | enabled: true, |
| 3115 | hooks: vec![blocker_one, blocker_two], |
| 3116 | ..HooksConfig::default() |
| 3117 | }, |
| 3118 | &fixture.workspace, |
| 3119 | Some(&fixture.registry), |
| 3120 | ); |
| 3121 | assert_eq!(config.hooks.len(), 3); |
| 3122 | config.hooks[2].background = true; |
| 3123 | let executor = HookExecutor::new(config, fixture.workspace.clone()); |
| 3124 | |
| 3125 | let submitted = executor.execute( |
| 3126 | HookEvent::SessionStart, |
| 3127 | &HookContext::new().with_workspace(fixture.workspace.clone()), |
| 3128 | ); |
| 3129 | assert_eq!(submitted.len(), 3); |
| 3130 | assert!(submitted.iter().all(|result| result.background)); |
| 3131 | fixture.revoke_from_fresh_registry(); |
| 3132 | |
| 3133 | std::thread::sleep(Duration::from_millis(1_500)); |
| 3134 | assert!( |
| 3135 | !fixture.marker.exists(), |
| 3136 | "queued hook must recheck authority after the preceding job finishes" |
| 3137 | ); |
| 3138 | } |
| 3139 | |
| 3140 | #[test] |
| 3141 | fn executor_type_is_available_from_executor_module() { |
| 3142 | let executor = crate::hooks::executor::HookExecutor::disabled(); |
| 3143 | |
| 3144 | assert!(!executor.is_enabled()); |
| 3145 | } |
| 3146 | |
| 3147 | /// #456 — `parse_env_lines` covers the formats users actually emit from |
| 3148 | /// shell hooks: bare `KEY=VAL`, `export KEY=VAL`, quoted values, comments, |
| 3149 | /// blank lines. Lines without `=` are dropped; values are taken verbatim |
| 3150 | /// (no shell expansion). |
| 3151 | #[test] |
| 3152 | fn parse_env_lines_handles_realistic_hook_output() { |
| 3153 | let stdout = r#" |
| 3154 | # Aux comment line, ignored |
| 3155 | AWS_ACCESS_KEY_ID=AKIAEXAMPLE |
| 3156 | export GITHUB_TOKEN=ghp_examplevalue |
| 3157 | QUOTED="value with spaces" |
| 3158 | SINGLE='also valid' |
| 3159 | |
| 3160 | = empty key dropped |
| 3161 | NOEQUAL line dropped |
| 3162 | "#; |
| 3163 | let parsed = super::parse_env_lines(stdout); |
| 3164 | assert_eq!( |
| 3165 | parsed.get("AWS_ACCESS_KEY_ID"), |
| 3166 | Some(&"AKIAEXAMPLE".to_string()) |
| 3167 | ); |
| 3168 | assert_eq!( |
| 3169 | parsed.get("GITHUB_TOKEN"), |
| 3170 | Some(&"ghp_examplevalue".to_string()) |
| 3171 | ); |
| 3172 | assert_eq!(parsed.get("QUOTED"), Some(&"value with spaces".to_string())); |
| 3173 | assert_eq!(parsed.get("SINGLE"), Some(&"also valid".to_string())); |
| 3174 | assert!(!parsed.contains_key("")); |
| 3175 | assert!(!parsed.contains_key("NOEQUAL line dropped")); |
| 3176 | // 4 valid entries above; nothing else. |
| 3177 | assert_eq!(parsed.len(), 4); |
| 3178 | } |
| 3179 | |
| 3180 | /// #456 — empty stdout (or only blank/comments) yields an empty map. |
| 3181 | #[test] |
| 3182 | fn parse_env_lines_empty_when_no_assignments() { |
| 3183 | let parsed = super::parse_env_lines("# nothing\n\n \n"); |
| 3184 | assert!(parsed.is_empty()); |
| 3185 | } |
| 3186 | |
| 3187 | #[test] |
| 3188 | fn parse_message_submit_stdout_replaces_text() { |
| 3189 | assert_eq!( |
| 3190 | super::parse_message_submit_stdout(r#"{"text":"changed"}"#), |
| 3191 | MessageSubmitStdout::Replaced("changed".to_string()) |
| 3192 | ); |
| 3193 | } |
| 3194 | |
| 3195 | #[test] |
| 3196 | fn parse_message_submit_stdout_empty_is_unchanged() { |
| 3197 | assert_eq!( |
| 3198 | super::parse_message_submit_stdout(" \n\t "), |
| 3199 | MessageSubmitStdout::Unchanged |
| 3200 | ); |
| 3201 | } |
| 3202 | |
| 3203 | #[test] |
| 3204 | fn parse_message_submit_stdout_without_text_is_unchanged() { |
| 3205 | assert_eq!( |
| 3206 | super::parse_message_submit_stdout(r#"{"reason":"only used for blocks"}"#), |
| 3207 | MessageSubmitStdout::Unchanged |
| 3208 | ); |
| 3209 | } |
| 3210 | |
| 3211 | #[test] |
| 3212 | fn message_submit_payload_is_byte_bounded_after_json_escaping() { |
| 3213 | let original = "用户\"\\\n".repeat(20_000); |
| 3214 | let payload = super::message_submit_payload( |
| 3215 | &HookContext::new() |
| 3216 | .with_session_id("sess_test") |
| 3217 | .with_model("model"), |
| 3218 | &original, |
| 3219 | ); |
| 3220 | let encoded = serde_json::to_vec(&payload).expect("serialize bounded payload"); |
| 3221 | |
| 3222 | assert!( |
| 3223 | encoded.len() <= super::HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES, |
| 3224 | "serialized payload was {} bytes", |
| 3225 | encoded.len() |
| 3226 | ); |
| 3227 | assert_eq!(payload["text_truncated"], true); |
| 3228 | assert_eq!(payload["text_original_bytes"], original.len()); |
| 3229 | let retained = payload["text"].as_str().expect("text string"); |
| 3230 | assert_eq!(payload["text_bytes"], retained.len()); |
| 3231 | assert!(original.starts_with(retained)); |
| 3232 | assert!(std::str::from_utf8(retained.as_bytes()).is_ok()); |
| 3233 | } |
| 3234 | |
| 3235 | #[test] |
| 3236 | fn message_submit_payload_omits_hostile_diagnostics_before_exceeding_cap() { |
| 3237 | let hostile = "\u{0}\u{1}\u{1f}\"\\".repeat(8_000); |
| 3238 | let original = "\u{0}\"\\用户".repeat(20_000); |
| 3239 | let context = HookContext::new() |
| 3240 | .with_session_id(&hostile) |
| 3241 | .with_workspace(PathBuf::from(&hostile)) |
| 3242 | .with_mode(&hostile) |
| 3243 | .with_model(&hostile); |
| 3244 | let payload = super::message_submit_payload(&context, &original); |
| 3245 | let encoded = serde_json::to_vec(&payload).expect("serialize hostile payload"); |
| 3246 | |
| 3247 | assert!( |
| 3248 | encoded.len() <= super::HOOK_MESSAGE_SUBMIT_PAYLOAD_MAX_BYTES, |
| 3249 | "serialized payload was {} bytes", |
| 3250 | encoded.len() |
| 3251 | ); |
| 3252 | assert_eq!(payload["text_truncated"], true); |
| 3253 | assert_eq!(payload["text_original_bytes"], original.len()); |
| 3254 | assert!(original.starts_with(payload["text"].as_str().expect("text"))); |
| 3255 | |
| 3256 | for key in ["session_id", "workspace", "mode", "model"] { |
| 3257 | if let Some(value) = payload.get(key).and_then(serde_json::Value::as_str) { |
| 3258 | assert!( |
| 3259 | value.len() <= super::HOOK_MESSAGE_SUBMIT_METADATA_MAX_BYTES + 16, |
| 3260 | "{key} was not bounded before serialization" |
| 3261 | ); |
| 3262 | } |
| 3263 | } |
| 3264 | } |
| 3265 | |
| 3266 | #[test] |
| 3267 | fn short_message_submit_payload_carries_explicit_untruncated_metadata() { |
| 3268 | let payload = super::message_submit_payload(&HookContext::new(), "hello 用户"); |
| 3269 | assert_eq!(payload["text"], "hello 用户"); |
| 3270 | assert_eq!(payload["text_bytes"], "hello 用户".len()); |
| 3271 | assert_eq!(payload["text_original_bytes"], "hello 用户".len()); |
| 3272 | assert_eq!(payload["text_truncated"], false); |
| 3273 | } |
| 3274 | |
| 3275 | #[test] |
| 3276 | fn parse_message_submit_stdout_rejects_malformed_json() { |
| 3277 | assert!(matches!( |
| 3278 | super::parse_message_submit_stdout("not json"), |
| 3279 | MessageSubmitStdout::Invalid(_) |
| 3280 | )); |
| 3281 | } |
| 3282 | |
| 3283 | #[test] |
| 3284 | fn parse_message_submit_stdout_rejects_non_string_text() { |
| 3285 | assert!(matches!( |
| 3286 | super::parse_message_submit_stdout(r#"{"text":123}"#), |
| 3287 | MessageSubmitStdout::Invalid(_) |
| 3288 | )); |
| 3289 | } |
| 3290 | |
| 3291 | #[test] |
| 3292 | fn parse_message_submit_stdout_rejects_empty_text() { |
| 3293 | assert_eq!( |
| 3294 | super::parse_message_submit_stdout(r#"{"text":""}"#), |
| 3295 | MessageSubmitStdout::Invalid("stdout `text` field must not be empty".to_string()) |
| 3296 | ); |
| 3297 | } |
| 3298 | |
| 3299 | #[test] |
| 3300 | fn parse_message_submit_stdout_rejects_non_object_json() { |
| 3301 | assert!(matches!( |
| 3302 | super::parse_message_submit_stdout(r#"["not", "an", "object"]"#), |
| 3303 | MessageSubmitStdout::Invalid(_) |
| 3304 | )); |
| 3305 | assert!(matches!( |
| 3306 | super::parse_message_submit_stdout(r#""not an object""#), |
| 3307 | MessageSubmitStdout::Invalid(_) |
| 3308 | )); |
| 3309 | } |
| 3310 | |
| 3311 | #[test] |
| 3312 | fn test_hook_event_as_str() { |
| 3313 | assert_eq!(HookEvent::SessionStart.as_str(), "session_start"); |
| 3314 | assert_eq!(HookEvent::ToolCallAfter.as_str(), "tool_call_after"); |
| 3315 | assert_eq!(HookEvent::ModeChange.as_str(), "mode_change"); |
| 3316 | assert_eq!(HookEvent::TurnEnd.as_str(), "turn_end"); |
| 3317 | assert_eq!(HookEvent::SubagentSpawn.as_str(), "subagent_spawn"); |
| 3318 | assert_eq!(HookEvent::SubagentComplete.as_str(), "subagent_complete"); |
| 3319 | } |
| 3320 | |
| 3321 | #[test] |
| 3322 | fn turn_end_payload_contains_post_turn_observer_fields() { |
| 3323 | let context = HookContext::new() |
| 3324 | .with_session_id("sess_test") |
| 3325 | .with_workspace(PathBuf::from("/tmp/codewhale")) |
| 3326 | .with_mode("agent") |
| 3327 | .with_model("deepseek-v4") |
| 3328 | .with_tokens(125); |
| 3329 | let usage = codewhale_models::Usage { |
| 3330 | input_tokens: 40, |
| 3331 | output_tokens: 9, |
| 3332 | prompt_cache_hit_tokens: Some(10), |
| 3333 | prompt_cache_miss_tokens: Some(30), |
| 3334 | prompt_cache_write_tokens: None, |
| 3335 | reasoning_tokens: Some(4), |
| 3336 | reasoning_replay_tokens: Some(2), |
| 3337 | server_tool_use: None, |
| 3338 | }; |
| 3339 | |
| 3340 | let payload = super::turn_end_payload(TurnEndPayloadInput { |
| 3341 | context: &context, |
| 3342 | created_at: "2026-07-12T10:30:00Z".parse().expect("timestamp"), |
| 3343 | model_backed: true, |
| 3344 | provider: Some("deepseek"), |
| 3345 | billing_surface: Some("test-payg"), |
| 3346 | model: Some("deepseek-v4-pro"), |
| 3347 | turn_id: "turn_123", |
| 3348 | status: "completed", |
| 3349 | error: None, |
| 3350 | duration: Duration::from_millis(321), |
| 3351 | usage: &usage, |
| 3352 | totals: TurnEndTotals { |
| 3353 | session_tokens: 125, |
| 3354 | conversation_tokens: 100, |
| 3355 | input_tokens: 100, |
| 3356 | output_tokens: 25, |
| 3357 | }, |
| 3358 | tool_count: 2, |
| 3359 | queued_message_count: 1, |
| 3360 | }); |
| 3361 | |
| 3362 | assert_eq!(payload["event"], "turn_end"); |
| 3363 | assert_eq!(payload["session_id"], "sess_test"); |
| 3364 | assert_eq!(payload["workspace"], "/tmp/codewhale"); |
| 3365 | assert_eq!(payload["mode"], "agent"); |
| 3366 | assert_eq!(payload["created_at"], "2026-07-12T10:30:00+00:00"); |
| 3367 | assert_eq!(payload["model_backed"], true); |
| 3368 | assert_eq!(payload["provider"], "deepseek"); |
| 3369 | assert_eq!(payload["billing_surface"], "test-payg"); |
| 3370 | assert!(payload.get("base_url").is_none()); |
| 3371 | assert_eq!(payload["model"], "deepseek-v4-pro"); |
| 3372 | assert_eq!(payload["turn_id"], "turn_123"); |
| 3373 | assert_eq!(payload["status"], "completed"); |
| 3374 | assert_eq!(payload["error"], serde_json::Value::Null); |
| 3375 | assert_eq!(payload["duration_ms"], 321); |
| 3376 | assert_eq!(payload["usage"]["input_tokens"], 40); |
| 3377 | assert_eq!(payload["usage"]["output_tokens"], 9); |
| 3378 | assert_eq!(payload["usage"]["prompt_cache_hit_tokens"], 10); |
| 3379 | assert_eq!(payload["usage"]["prompt_cache_miss_tokens"], 30); |
| 3380 | assert_eq!(payload["usage"]["reasoning_tokens"], 4); |
| 3381 | assert_eq!(payload["usage"]["reasoning_replay_tokens"], 2); |
| 3382 | assert_eq!(payload["totals"]["session_tokens"], 125); |
| 3383 | assert_eq!(payload["totals"]["conversation_tokens"], 100); |
| 3384 | assert_eq!(payload["totals"]["input_tokens"], 100); |
| 3385 | assert_eq!(payload["totals"]["output_tokens"], 25); |
| 3386 | assert_eq!(payload["tool_count"], 2); |
| 3387 | assert_eq!(payload["queued_message_count"], 1); |
| 3388 | assert_eq!(payload["stop_hook_active"], false); |
| 3389 | } |
| 3390 | |
| 3391 | #[test] |
| 3392 | fn test_hook_context_to_env_vars() { |
| 3393 | let ctx = HookContext::new() |
| 3394 | .with_tool_name("exec_shell") |
| 3395 | .with_mode("agent") |
| 3396 | .with_workspace(PathBuf::from("/tmp")); |
| 3397 | |
| 3398 | let env = ctx.to_env_vars(); |
| 3399 | |
| 3400 | assert_eq!( |
| 3401 | env.get("DEEPSEEK_TOOL_NAME"), |
| 3402 | Some(&"exec_shell".to_string()) |
| 3403 | ); |
| 3404 | assert_eq!(env.get("DEEPSEEK_MODE"), Some(&"agent".to_string())); |
| 3405 | assert_eq!(env.get("DEEPSEEK_WORKSPACE"), Some(&"/tmp".to_string())); |
| 3406 | } |
| 3407 | |
| 3408 | #[test] |
| 3409 | fn test_hook_condition_always() { |
| 3410 | let hook = Hook::new(HookEvent::SessionStart, "echo test"); |
| 3411 | let executor = HookExecutor::disabled(); |
| 3412 | let context = HookContext::new(); |
| 3413 | |
| 3414 | assert!(executor.matches_condition(&hook, &context)); |
| 3415 | } |
| 3416 | |
| 3417 | #[test] |
| 3418 | fn test_hook_condition_tool_name() { |
| 3419 | let hook = Hook::new(HookEvent::ToolCallBefore, "echo test").with_condition( |
| 3420 | HookCondition::ToolName { |
| 3421 | name: "exec_shell".to_string(), |
| 3422 | }, |
| 3423 | ); |
| 3424 | |
| 3425 | let executor = HookExecutor::disabled(); |
| 3426 | |
| 3427 | let context_match = HookContext::new().with_tool_name("exec_shell"); |
| 3428 | let context_no_match = HookContext::new().with_tool_name("write_file"); |
| 3429 | |
| 3430 | assert!(executor.matches_condition(&hook, &context_match)); |
| 3431 | assert!(!executor.matches_condition(&hook, &context_no_match)); |
| 3432 | } |
| 3433 | |
| 3434 | #[test] |
| 3435 | fn test_hook_condition_mode() { |
| 3436 | let hook = |
| 3437 | Hook::new(HookEvent::ModeChange, "echo test").with_condition(HookCondition::Mode { |
| 3438 | mode: "agent".to_string(), |
| 3439 | }); |
| 3440 | |
| 3441 | let executor = HookExecutor::disabled(); |
| 3442 | |
| 3443 | let context_match = HookContext::new().with_mode("AGENT"); // Case insensitive |
| 3444 | let context_no_match = HookContext::new().with_mode("normal"); |
| 3445 | |
| 3446 | assert!(executor.matches_condition(&hook, &context_match)); |
| 3447 | assert!(!executor.matches_condition(&hook, &context_no_match)); |
| 3448 | } |
| 3449 | |
| 3450 | #[test] |
| 3451 | fn test_hooks_config_for_event() { |
| 3452 | let config = HooksConfig { |
| 3453 | enabled: true, |
| 3454 | hooks: vec![ |
| 3455 | Hook::new(HookEvent::SessionStart, "echo start"), |
| 3456 | Hook::new(HookEvent::SessionEnd, "echo end"), |
| 3457 | Hook::new(HookEvent::SessionStart, "echo start2"), |
| 3458 | ], |
| 3459 | ..Default::default() |
| 3460 | }; |
| 3461 | |
| 3462 | let start_hooks = config.hooks_for_event(HookEvent::SessionStart); |
| 3463 | assert_eq!(start_hooks.len(), 2); |
| 3464 | |
| 3465 | let end_hooks = config.hooks_for_event(HookEvent::SessionEnd); |
| 3466 | assert_eq!(end_hooks.len(), 1); |
| 3467 | } |
| 3468 | |
| 3469 | #[test] |
| 3470 | fn test_hooks_config_disabled() { |
| 3471 | let config = HooksConfig { |
| 3472 | enabled: false, |
| 3473 | hooks: vec![Hook::new(HookEvent::SessionStart, "echo start")], |
| 3474 | ..Default::default() |
| 3475 | }; |
| 3476 | |
| 3477 | let hooks = config.hooks_for_event(HookEvent::SessionStart); |
| 3478 | assert!(hooks.is_empty()); |
| 3479 | } |
| 3480 | |
| 3481 | #[test] |
| 3482 | fn test_hook_builder() { |
| 3483 | let hook = Hook::new(HookEvent::ToolCallAfter, "notify.sh") |
| 3484 | .with_name("notify_tool") |
| 3485 | .with_timeout(60) |
| 3486 | .background() |
| 3487 | .with_condition(HookCondition::ToolCategory { |
| 3488 | category: "shell".to_string(), |
| 3489 | }); |
| 3490 | |
| 3491 | assert_eq!(hook.name, Some("notify_tool".to_string())); |
| 3492 | assert_eq!(hook.timeout_secs, 60); |
| 3493 | assert!(hook.background); |
| 3494 | assert!(matches!( |
| 3495 | hook.condition, |
| 3496 | Some(HookCondition::ToolCategory { .. }) |
| 3497 | )); |
| 3498 | } |
| 3499 | |
| 3500 | #[test] |
| 3501 | fn test_hook_timeout_enforced() { |
| 3502 | let command = if cfg!(windows) { |
| 3503 | "ping -n 3 127.0.0.1 > nul" |
| 3504 | } else { |
| 3505 | "sleep 2" |
| 3506 | }; |
| 3507 | let hook = Hook::new(HookEvent::SessionStart, command).with_timeout(1); |
| 3508 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 3509 | let env_vars = HashMap::new(); |
| 3510 | |
| 3511 | let result = executor.execute_sync(&hook, &env_vars); |
| 3512 | assert!(!result.success); |
| 3513 | assert!( |
| 3514 | result |
| 3515 | .error |
| 3516 | .as_ref() |
| 3517 | .is_some_and(|e| e.contains("timed out")) |
| 3518 | ); |
| 3519 | } |
| 3520 | |
| 3521 | #[test] |
| 3522 | fn observer_hook_receives_eof_instead_of_inheriting_terminal_stdin() { |
| 3523 | const INNER_ENV: &str = "CODEWHALE_TEST_HOOK_EOF_INNER"; |
| 3524 | const TEST_NAME: &str = |
| 3525 | "hooks::tests::observer_hook_receives_eof_instead_of_inheriting_terminal_stdin"; |
| 3526 | |
| 3527 | if std::env::var_os(INNER_ENV).is_some() { |
| 3528 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3529 | #[cfg(not(windows))] |
| 3530 | let command = write_hook_script( |
| 3531 | &dir, |
| 3532 | "read_to_eof.sh", |
| 3533 | r#"#!/bin/sh |
| 3534 | payload=$(cat) |
| 3535 | printf 'stdin-bytes=%s\n' "${#payload}" |
| 3536 | "#, |
| 3537 | ); |
| 3538 | #[cfg(windows)] |
| 3539 | let command = "powershell -NoProfile -Command \"$value = [Console]::In.ReadToEnd(); [Console]::Out.WriteLine(('stdin-bytes=' + $value.Length))\"".to_string(); |
| 3540 | // A cold PowerShell process can take several seconds to start on a |
| 3541 | // contended Windows CI runner. Keep the hook timeout finite so the |
| 3542 | // regression still detects an inherited live stdin pipe, while |
| 3543 | // allowing enough startup time for the EOF assertion itself. |
| 3544 | let hook_timeout_secs = if cfg!(windows) { 10 } else { 2 }; |
| 3545 | let hook = |
| 3546 | Hook::new(HookEvent::ToolCallBefore, &command).with_timeout(hook_timeout_secs); |
| 3547 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 3548 | |
| 3549 | let result = executor.execute_sync(&hook, &HashMap::new()); |
| 3550 | assert!(result.success, "stdin-less hook should finish: {result:?}"); |
| 3551 | assert_eq!(result.stdout.trim(), "stdin-bytes=0"); |
| 3552 | return; |
| 3553 | } |
| 3554 | |
| 3555 | // Keep this subprocess's stdin pipe deliberately open. Before #4489, |
| 3556 | // the hook inherited that live pipe and blocked instead of receiving |
| 3557 | // EOF. The inner test can only finish when HookExecutor closes the |
| 3558 | // child's stdin write end. |
| 3559 | let mut child = Command::new(std::env::current_exe().expect("current test binary")) |
| 3560 | .args(["--exact", TEST_NAME, "--nocapture", "--test-threads=1"]) |
| 3561 | .env(INNER_ENV, "1") |
| 3562 | .stdin(Stdio::piped()) |
| 3563 | .spawn() |
| 3564 | .expect("spawn isolated hook EOF test"); |
| 3565 | let held_open_stdin = child.stdin.take().expect("piped child stdin"); |
| 3566 | // Leave headroom around the inner hook timeout so a cold Windows test |
| 3567 | // process can start, without weakening the held-open-pipe regression. |
| 3568 | let isolated_timeout_secs = if cfg!(windows) { 25 } else { 10 }; |
| 3569 | let status = match child |
| 3570 | .wait_timeout(Duration::from_secs(isolated_timeout_secs)) |
| 3571 | .expect("wait for isolated hook EOF test") |
| 3572 | { |
| 3573 | Some(status) => status, |
| 3574 | None => { |
| 3575 | let _ = child.kill(); |
| 3576 | let _ = child.wait(); |
| 3577 | panic!("isolated hook EOF test hung with parent stdin open"); |
| 3578 | } |
| 3579 | }; |
| 3580 | drop(held_open_stdin); |
| 3581 | assert!(status.success(), "isolated hook EOF test failed: {status}"); |
| 3582 | } |
| 3583 | |
| 3584 | #[cfg(not(windows))] |
| 3585 | #[test] |
| 3586 | fn timed_out_hook_kills_descendant_process_group() { |
| 3587 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3588 | let marker = dir.path().join("descendant-survived"); |
| 3589 | let command = write_hook_script( |
| 3590 | &dir, |
| 3591 | "spawn_descendant.sh", |
| 3592 | &format!( |
| 3593 | "#!/bin/sh\n(sleep 2; printf leaked > '{}') &\nsleep 5\n", |
| 3594 | marker.display() |
| 3595 | ), |
| 3596 | ); |
| 3597 | let hook = Hook::new(HookEvent::ToolCallBefore, &command).with_timeout(1); |
| 3598 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 3599 | |
| 3600 | let result = executor.execute_sync(&hook, &HashMap::new()); |
| 3601 | assert!( |
| 3602 | result |
| 3603 | .error |
| 3604 | .as_ref() |
| 3605 | .is_some_and(|error| error.contains("timed out")), |
| 3606 | "hook should time out: {result:?}" |
| 3607 | ); |
| 3608 | std::thread::sleep(Duration::from_millis(1_500)); |
| 3609 | assert!( |
| 3610 | !marker.exists(), |
| 3611 | "the timed-out hook's descendant escaped its process group" |
| 3612 | ); |
| 3613 | } |
| 3614 | |
| 3615 | #[cfg(windows)] |
| 3616 | #[test] |
| 3617 | fn timed_out_hook_kills_windows_descendant_job() { |
| 3618 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3619 | let started = dir.path().join("descendant-started.txt"); |
| 3620 | let survived = dir.path().join("descendant-survived.txt"); |
| 3621 | let descendant = dir.path().join("descendant.cmd"); |
| 3622 | std::fs::write( |
| 3623 | &descendant, |
| 3624 | "@echo off\r\necho started>descendant-started.txt\r\nping -n 5 127.0.0.1 > nul\r\necho survived>descendant-survived.txt\r\n", |
| 3625 | ) |
| 3626 | .expect("write descendant script"); |
| 3627 | let parent = dir.path().join("parent.cmd"); |
| 3628 | std::fs::write( |
| 3629 | &parent, |
| 3630 | "@echo off\r\nstart \"\" /b cmd.exe /d /c descendant.cmd\r\n:wait_for_child\r\nif exist descendant-started.txt goto child_started\r\nping -n 2 127.0.0.1 > nul\r\ngoto wait_for_child\r\n:child_started\r\nping -n 10 127.0.0.1 > nul\r\n", |
| 3631 | ) |
| 3632 | .expect("write parent script"); |
| 3633 | let hook = Hook::new(HookEvent::ToolCallBefore, "call parent.cmd").with_timeout(3); |
| 3634 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 3635 | |
| 3636 | let result = executor.execute_sync(&hook, &HashMap::new()); |
| 3637 | assert!( |
| 3638 | result |
| 3639 | .error |
| 3640 | .as_ref() |
| 3641 | .is_some_and(|error| error.contains("timed out")), |
| 3642 | "hook should time out: {result:?}" |
| 3643 | ); |
| 3644 | assert!( |
| 3645 | started.exists(), |
| 3646 | "descendant never reached its start handshake" |
| 3647 | ); |
| 3648 | std::thread::sleep(Duration::from_secs(5)); |
| 3649 | assert!( |
| 3650 | !survived.exists(), |
| 3651 | "the timed-out hook's descendant escaped its Job Object" |
| 3652 | ); |
| 3653 | } |
| 3654 | |
| 3655 | #[cfg(not(windows))] |
| 3656 | #[test] |
| 3657 | fn message_submit_stdin_write_does_not_deadlock_when_hook_writes_first() { |
| 3658 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3659 | let command = write_hook_script( |
| 3660 | &dir, |
| 3661 | "write_before_read.sh", |
| 3662 | r#"#!/bin/sh |
| 3663 | dd if=/dev/zero bs=1024 count=256 2>/dev/null | tr '\000' x |
| 3664 | dd if=/dev/zero bs=1024 count=256 2>/dev/null | tr '\000' e >&2 |
| 3665 | payload=$(cat) |
| 3666 | printf '\ndone:%s\n' "${#payload}" |
| 3667 | "#, |
| 3668 | ); |
| 3669 | let hook = Hook::new(HookEvent::MessageSubmit, &command).with_timeout(5); |
| 3670 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 3671 | let env_vars = HashMap::new(); |
| 3672 | let payload = json!({ |
| 3673 | "event": "message_submit", |
| 3674 | "text": "x".repeat(256 * 1024), |
| 3675 | }); |
| 3676 | |
| 3677 | let result = executor.execute_sync_with_stdin(&hook, &env_vars, &payload); |
| 3678 | |
| 3679 | assert!(result.success, "hook should complete: {result:?}"); |
| 3680 | assert!(result.stdout.ends_with("…[truncated]")); |
| 3681 | assert!(result.stderr.ends_with("…[truncated]")); |
| 3682 | assert!(result.stdout.len() <= HOOK_PIPE_CAPTURE_MAX_BYTES + 16); |
| 3683 | assert!(result.stderr.len() <= HOOK_PIPE_CAPTURE_MAX_BYTES + 16); |
| 3684 | } |
| 3685 | |
| 3686 | #[test] |
| 3687 | fn test_executor_session_id() { |
| 3688 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 3689 | |
| 3690 | assert!(executor.session_id().starts_with("sess_")); |
| 3691 | assert_eq!(executor.session_id().len(), 13); // "sess_" + 8 chars |
| 3692 | } |
| 3693 | |
| 3694 | #[cfg(not(windows))] |
| 3695 | fn write_hook_script(dir: &tempfile::TempDir, name: &str, content: &str) -> String { |
| 3696 | let path = dir.path().join(name); |
| 3697 | std::fs::write(&path, content).expect("write hook script"); |
| 3698 | format!("sh {}", path.display()) |
| 3699 | } |
| 3700 | |
| 3701 | #[cfg(not(windows))] |
| 3702 | fn submit_context(dir: &tempfile::TempDir) -> HookContext { |
| 3703 | HookContext::new() |
| 3704 | .with_session_id("sess_test") |
| 3705 | .with_workspace(dir.path().to_path_buf()) |
| 3706 | .with_mode("agent") |
| 3707 | .with_model("deepseek-test") |
| 3708 | .with_tokens(42) |
| 3709 | } |
| 3710 | |
| 3711 | #[cfg(not(windows))] |
| 3712 | #[test] |
| 3713 | fn json_observer_hook_receives_structured_stdin() { |
| 3714 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3715 | let out = dir.path().join("payload.json"); |
| 3716 | let command = write_hook_script( |
| 3717 | &dir, |
| 3718 | "capture_observer.sh", |
| 3719 | &format!( |
| 3720 | r#"#!/bin/sh |
| 3721 | cat > "{}" |
| 3722 | "#, |
| 3723 | out.display() |
| 3724 | ), |
| 3725 | ); |
| 3726 | let executor = HookExecutor::new( |
| 3727 | HooksConfig { |
| 3728 | enabled: true, |
| 3729 | hooks: vec![Hook::new(HookEvent::SubagentSpawn, &command)], |
| 3730 | ..Default::default() |
| 3731 | }, |
| 3732 | dir.path().to_path_buf(), |
| 3733 | ); |
| 3734 | let payload = json!({ |
| 3735 | "event": "subagent_spawn", |
| 3736 | "agent_id": "agent_123", |
| 3737 | "prompt_preview": "inspect this", |
| 3738 | "prompt_truncated": false, |
| 3739 | }); |
| 3740 | |
| 3741 | let results = executor.execute_json_observer( |
| 3742 | HookEvent::SubagentSpawn, |
| 3743 | &submit_context(&dir), |
| 3744 | &payload, |
| 3745 | ); |
| 3746 | |
| 3747 | assert_eq!(results.len(), 1); |
| 3748 | assert!(results[0].success); |
| 3749 | let captured: serde_json::Value = |
| 3750 | serde_json::from_str(&std::fs::read_to_string(out).expect("payload written")) |
| 3751 | .expect("valid JSON payload"); |
| 3752 | assert_eq!(captured["event"], "subagent_spawn"); |
| 3753 | assert_eq!(captured["agent_id"], "agent_123"); |
| 3754 | assert_eq!(captured["prompt_preview"], "inspect this"); |
| 3755 | assert_eq!(captured["prompt_truncated"], false); |
| 3756 | } |
| 3757 | |
| 3758 | #[cfg(not(windows))] |
| 3759 | #[test] |
| 3760 | fn turn_end_observer_hook_receives_stdin_json_and_ignores_stdout_contract() { |
| 3761 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3762 | let out = dir.path().join("turn_end.json"); |
| 3763 | let command = write_hook_script( |
| 3764 | &dir, |
| 3765 | "capture_turn_end.sh", |
| 3766 | &format!( |
| 3767 | r#"#!/bin/sh |
| 3768 | cat > "{}" |
| 3769 | printf '%s\n' '{{"text":"stdout is not a mutation contract"}}' |
| 3770 | "#, |
| 3771 | out.display() |
| 3772 | ), |
| 3773 | ); |
| 3774 | let executor = HookExecutor::new( |
| 3775 | HooksConfig { |
| 3776 | enabled: true, |
| 3777 | hooks: vec![Hook::new(HookEvent::TurnEnd, &command)], |
| 3778 | ..Default::default() |
| 3779 | }, |
| 3780 | dir.path().to_path_buf(), |
| 3781 | ); |
| 3782 | let usage = codewhale_models::Usage { |
| 3783 | input_tokens: 12, |
| 3784 | output_tokens: 3, |
| 3785 | prompt_cache_hit_tokens: None, |
| 3786 | prompt_cache_miss_tokens: None, |
| 3787 | prompt_cache_write_tokens: None, |
| 3788 | reasoning_tokens: None, |
| 3789 | reasoning_replay_tokens: None, |
| 3790 | server_tool_use: None, |
| 3791 | }; |
| 3792 | let context = submit_context(&dir).with_tokens(15); |
| 3793 | let payload = super::turn_end_payload(TurnEndPayloadInput { |
| 3794 | context: &context, |
| 3795 | created_at: "2026-07-12T10:30:00Z".parse().expect("timestamp"), |
| 3796 | model_backed: true, |
| 3797 | provider: Some("openai"), |
| 3798 | billing_surface: None, |
| 3799 | model: Some("gpt-5.5"), |
| 3800 | turn_id: "turn_observed", |
| 3801 | status: "completed", |
| 3802 | error: None, |
| 3803 | duration: Duration::from_millis(7), |
| 3804 | usage: &usage, |
| 3805 | totals: TurnEndTotals { |
| 3806 | session_tokens: 15, |
| 3807 | conversation_tokens: 15, |
| 3808 | input_tokens: 12, |
| 3809 | output_tokens: 3, |
| 3810 | }, |
| 3811 | tool_count: 0, |
| 3812 | queued_message_count: 0, |
| 3813 | }); |
| 3814 | |
| 3815 | let results = executor.execute_json_observer(HookEvent::TurnEnd, &context, &payload); |
| 3816 | |
| 3817 | assert_eq!(results.len(), 1); |
| 3818 | assert!(results[0].success); |
| 3819 | assert!( |
| 3820 | results[0] |
| 3821 | .stdout |
| 3822 | .contains("stdout is not a mutation contract"), |
| 3823 | "stdout is still captured for diagnostics" |
| 3824 | ); |
| 3825 | let captured: serde_json::Value = |
| 3826 | serde_json::from_str(&std::fs::read_to_string(out).expect("payload written")) |
| 3827 | .expect("valid JSON payload"); |
| 3828 | assert_eq!(captured["event"], "turn_end"); |
| 3829 | assert_eq!(captured["created_at"], "2026-07-12T10:30:00+00:00"); |
| 3830 | assert_eq!(captured["provider"], "openai"); |
| 3831 | assert_eq!(captured["model"], "gpt-5.5"); |
| 3832 | assert_eq!(captured["turn_id"], "turn_observed"); |
| 3833 | assert_eq!(captured["totals"]["input_tokens"], 12); |
| 3834 | assert_eq!(captured["totals"]["output_tokens"], 3); |
| 3835 | } |
| 3836 | |
| 3837 | #[cfg(not(windows))] |
| 3838 | #[test] |
| 3839 | fn json_observer_hook_failure_does_not_stop_later_hooks() { |
| 3840 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3841 | let marker = dir.path().join("later-ran"); |
| 3842 | let failing = write_hook_script( |
| 3843 | &dir, |
| 3844 | "failing_observer.sh", |
| 3845 | r#"#!/bin/sh |
| 3846 | echo boom >&2 |
| 3847 | exit 1 |
| 3848 | "#, |
| 3849 | ); |
| 3850 | let later = write_hook_script( |
| 3851 | &dir, |
| 3852 | "later_observer.sh", |
| 3853 | &format!( |
| 3854 | r#"#!/bin/sh |
| 3855 | cat > "{}" |
| 3856 | "#, |
| 3857 | marker.display() |
| 3858 | ), |
| 3859 | ); |
| 3860 | let mut first = Hook::new(HookEvent::SubagentComplete, &failing); |
| 3861 | first.continue_on_error = false; |
| 3862 | let executor = HookExecutor::new( |
| 3863 | HooksConfig { |
| 3864 | enabled: true, |
| 3865 | hooks: vec![first, Hook::new(HookEvent::SubagentComplete, &later)], |
| 3866 | ..Default::default() |
| 3867 | }, |
| 3868 | dir.path().to_path_buf(), |
| 3869 | ); |
| 3870 | let payload = json!({ |
| 3871 | "event": "subagent_complete", |
| 3872 | "agent_id": "agent_456", |
| 3873 | "status": "completed", |
| 3874 | }); |
| 3875 | |
| 3876 | let results = executor.execute_json_observer( |
| 3877 | HookEvent::SubagentComplete, |
| 3878 | &submit_context(&dir), |
| 3879 | &payload, |
| 3880 | ); |
| 3881 | |
| 3882 | assert_eq!(results.len(), 2); |
| 3883 | assert!(!results[0].success); |
| 3884 | assert!(results[1].success); |
| 3885 | assert!( |
| 3886 | marker.exists(), |
| 3887 | "observer failures must be warn-only and non-blocking" |
| 3888 | ); |
| 3889 | } |
| 3890 | |
| 3891 | #[cfg(not(windows))] |
| 3892 | #[test] |
| 3893 | fn message_submit_transform_applies_hooks_in_order() { |
| 3894 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3895 | let first = write_hook_script( |
| 3896 | &dir, |
| 3897 | "first.sh", |
| 3898 | r#"#!/bin/sh |
| 3899 | printf '%s\n' '{"text":"first"}' |
| 3900 | "#, |
| 3901 | ); |
| 3902 | let second = write_hook_script( |
| 3903 | &dir, |
| 3904 | "second.sh", |
| 3905 | r#"#!/bin/sh |
| 3906 | payload=$(cat) |
| 3907 | case "$payload" in |
| 3908 | *'"text":"first"'*) printf '%s\n' '{"text":"first second"}' ;; |
| 3909 | *) printf '%s\n' '{"text":"wrong"}' ;; |
| 3910 | esac |
| 3911 | "#, |
| 3912 | ); |
| 3913 | let config = HooksConfig { |
| 3914 | enabled: true, |
| 3915 | hooks: vec![ |
| 3916 | Hook::new(HookEvent::MessageSubmit, &first), |
| 3917 | Hook::new(HookEvent::MessageSubmit, &second), |
| 3918 | ], |
| 3919 | working_dir: Some(dir.path().to_path_buf()), |
| 3920 | ..HooksConfig::default() |
| 3921 | }; |
| 3922 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 3923 | |
| 3924 | assert_eq!( |
| 3925 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 3926 | MessageSubmitOutcome::replaced("first second".to_string()) |
| 3927 | ); |
| 3928 | } |
| 3929 | |
| 3930 | #[cfg(not(windows))] |
| 3931 | #[test] |
| 3932 | fn message_submit_transform_exit_two_blocks_submission() { |
| 3933 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3934 | let command = write_hook_script( |
| 3935 | &dir, |
| 3936 | "block.sh", |
| 3937 | r#"#!/bin/sh |
| 3938 | printf '%s\n' '{"reason":"policy blocked this prompt"}' |
| 3939 | exit 2 |
| 3940 | "#, |
| 3941 | ); |
| 3942 | let config = HooksConfig { |
| 3943 | enabled: true, |
| 3944 | hooks: vec![Hook::new(HookEvent::MessageSubmit, &command)], |
| 3945 | working_dir: Some(dir.path().to_path_buf()), |
| 3946 | ..HooksConfig::default() |
| 3947 | }; |
| 3948 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 3949 | |
| 3950 | assert_eq!( |
| 3951 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 3952 | MessageSubmitOutcome::Blocked { |
| 3953 | reason: "policy blocked this prompt".to_string() |
| 3954 | } |
| 3955 | ); |
| 3956 | } |
| 3957 | |
| 3958 | #[cfg(not(windows))] |
| 3959 | #[test] |
| 3960 | fn background_message_submit_hook_is_observer_only() { |
| 3961 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3962 | let command = write_hook_script( |
| 3963 | &dir, |
| 3964 | "background.sh", |
| 3965 | r#"#!/bin/sh |
| 3966 | printf '%s\n' '{"text":"ignored"}' |
| 3967 | "#, |
| 3968 | ); |
| 3969 | let config = HooksConfig { |
| 3970 | enabled: true, |
| 3971 | hooks: vec![Hook::new(HookEvent::MessageSubmit, &command).background()], |
| 3972 | working_dir: Some(dir.path().to_path_buf()), |
| 3973 | ..HooksConfig::default() |
| 3974 | }; |
| 3975 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 3976 | |
| 3977 | assert_eq!( |
| 3978 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 3979 | MessageSubmitOutcome::unchanged() |
| 3980 | ); |
| 3981 | } |
| 3982 | |
| 3983 | #[test] |
| 3984 | fn message_submit_transform_without_configured_hooks_is_unchanged() { |
| 3985 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 3986 | |
| 3987 | assert_eq!( |
| 3988 | executor.execute_message_submit_transform(&HookContext::new(), "original"), |
| 3989 | MessageSubmitOutcome::unchanged() |
| 3990 | ); |
| 3991 | } |
| 3992 | |
| 3993 | #[cfg(not(windows))] |
| 3994 | #[test] |
| 3995 | fn message_submit_transform_skips_non_matching_condition() { |
| 3996 | let dir = tempfile::tempdir().expect("tempdir"); |
| 3997 | let command = write_hook_script( |
| 3998 | &dir, |
| 3999 | "replace.sh", |
| 4000 | r#"#!/bin/sh |
| 4001 | printf '%s\n' '{"text":"should not apply"}' |
| 4002 | "#, |
| 4003 | ); |
| 4004 | let hook = |
| 4005 | Hook::new(HookEvent::MessageSubmit, &command).with_condition(HookCondition::Mode { |
| 4006 | mode: "plan".into(), |
| 4007 | }); |
| 4008 | let config = HooksConfig { |
| 4009 | enabled: true, |
| 4010 | hooks: vec![hook], |
| 4011 | working_dir: Some(dir.path().to_path_buf()), |
| 4012 | ..HooksConfig::default() |
| 4013 | }; |
| 4014 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4015 | |
| 4016 | assert_eq!( |
| 4017 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 4018 | MessageSubmitOutcome::unchanged() |
| 4019 | ); |
| 4020 | } |
| 4021 | |
| 4022 | #[cfg(not(windows))] |
| 4023 | #[test] |
| 4024 | fn message_submit_continue_on_error_true_keeps_text_and_runs_later_hooks() { |
| 4025 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4026 | let failing = write_hook_script( |
| 4027 | &dir, |
| 4028 | "fail_continue.sh", |
| 4029 | r#"#!/bin/sh |
| 4030 | printf '%s\n' 'soft failure' >&2 |
| 4031 | exit 9 |
| 4032 | "#, |
| 4033 | ); |
| 4034 | let replacing = write_hook_script( |
| 4035 | &dir, |
| 4036 | "replace_after_failure.sh", |
| 4037 | r#"#!/bin/sh |
| 4038 | printf '%s\n' '{"text":"recovered"}' |
| 4039 | "#, |
| 4040 | ); |
| 4041 | let config = HooksConfig { |
| 4042 | enabled: true, |
| 4043 | hooks: vec![ |
| 4044 | Hook::new(HookEvent::MessageSubmit, &failing), |
| 4045 | Hook::new(HookEvent::MessageSubmit, &replacing), |
| 4046 | ], |
| 4047 | working_dir: Some(dir.path().to_path_buf()), |
| 4048 | ..HooksConfig::default() |
| 4049 | }; |
| 4050 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4051 | |
| 4052 | assert_eq!( |
| 4053 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 4054 | MessageSubmitOutcome::replaced("recovered".to_string()) |
| 4055 | .with_warning(Some("message_submit hook exited with code 9".to_string())) |
| 4056 | ); |
| 4057 | } |
| 4058 | |
| 4059 | #[cfg(not(windows))] |
| 4060 | #[test] |
| 4061 | fn message_submit_timeout_continue_surfaces_warning_and_runs_later_hooks() { |
| 4062 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4063 | let slow = write_hook_script( |
| 4064 | &dir, |
| 4065 | "slow_continue.sh", |
| 4066 | r#"#!/bin/sh |
| 4067 | sleep 2 |
| 4068 | "#, |
| 4069 | ); |
| 4070 | let replacing = write_hook_script( |
| 4071 | &dir, |
| 4072 | "replace_after_timeout.sh", |
| 4073 | r#"#!/bin/sh |
| 4074 | printf '%s\n' '{"text":"after timeout"}' |
| 4075 | "#, |
| 4076 | ); |
| 4077 | let mut slow_hook = Hook::new(HookEvent::MessageSubmit, &slow).with_timeout(1); |
| 4078 | slow_hook.continue_on_error = true; |
| 4079 | let config = HooksConfig { |
| 4080 | enabled: true, |
| 4081 | hooks: vec![slow_hook, Hook::new(HookEvent::MessageSubmit, &replacing)], |
| 4082 | working_dir: Some(dir.path().to_path_buf()), |
| 4083 | ..HooksConfig::default() |
| 4084 | }; |
| 4085 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4086 | |
| 4087 | assert_eq!( |
| 4088 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 4089 | MessageSubmitOutcome::replaced("after timeout".to_string()) |
| 4090 | .with_warning(Some("hook timed out after 1s".to_string())) |
| 4091 | ); |
| 4092 | } |
| 4093 | |
| 4094 | #[cfg(not(windows))] |
| 4095 | #[test] |
| 4096 | fn message_submit_invalid_stdout_keeps_text_and_runs_later_hooks() { |
| 4097 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4098 | let invalid = write_hook_script( |
| 4099 | &dir, |
| 4100 | "invalid_stdout.sh", |
| 4101 | r#"#!/bin/sh |
| 4102 | printf '%s\n' 'not json' |
| 4103 | "#, |
| 4104 | ); |
| 4105 | let replacing = write_hook_script( |
| 4106 | &dir, |
| 4107 | "replace_after_invalid.sh", |
| 4108 | r#"#!/bin/sh |
| 4109 | printf '%s\n' '{"text":"valid later"}' |
| 4110 | "#, |
| 4111 | ); |
| 4112 | let config = HooksConfig { |
| 4113 | enabled: true, |
| 4114 | hooks: vec![ |
| 4115 | Hook::new(HookEvent::MessageSubmit, &invalid), |
| 4116 | Hook::new(HookEvent::MessageSubmit, &replacing), |
| 4117 | ], |
| 4118 | working_dir: Some(dir.path().to_path_buf()), |
| 4119 | ..HooksConfig::default() |
| 4120 | }; |
| 4121 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4122 | |
| 4123 | assert_eq!( |
| 4124 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 4125 | MessageSubmitOutcome::replaced("valid later".to_string()) |
| 4126 | ); |
| 4127 | } |
| 4128 | |
| 4129 | #[cfg(not(windows))] |
| 4130 | #[test] |
| 4131 | fn message_submit_continue_on_error_false_blocks_on_failure() { |
| 4132 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4133 | let command = write_hook_script( |
| 4134 | &dir, |
| 4135 | "fail.sh", |
| 4136 | r#"#!/bin/sh |
| 4137 | printf '%s\n' 'hard failure' >&2 |
| 4138 | exit 7 |
| 4139 | "#, |
| 4140 | ); |
| 4141 | let mut hook = Hook::new(HookEvent::MessageSubmit, &command); |
| 4142 | hook.continue_on_error = false; |
| 4143 | let config = HooksConfig { |
| 4144 | enabled: true, |
| 4145 | hooks: vec![hook], |
| 4146 | working_dir: Some(dir.path().to_path_buf()), |
| 4147 | ..HooksConfig::default() |
| 4148 | }; |
| 4149 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4150 | |
| 4151 | assert_eq!( |
| 4152 | executor.execute_message_submit_transform(&submit_context(&dir), "original"), |
| 4153 | MessageSubmitOutcome::Blocked { |
| 4154 | reason: "message_submit hook failed and blocked submission".to_string() |
| 4155 | } |
| 4156 | ); |
| 4157 | } |
| 4158 | |
| 4159 | #[test] |
| 4160 | fn has_hooks_for_event_fast_path_returns_false_for_empty_config() { |
| 4161 | let executor = HookExecutor::disabled(); |
| 4162 | // No hooks configured AT ALL — every event is a fast skip. |
| 4163 | for event in [ |
| 4164 | HookEvent::SessionStart, |
| 4165 | HookEvent::SessionEnd, |
| 4166 | HookEvent::MessageSubmit, |
| 4167 | HookEvent::ToolCallBefore, |
| 4168 | HookEvent::ToolCallAfter, |
| 4169 | HookEvent::ModeChange, |
| 4170 | HookEvent::OnError, |
| 4171 | HookEvent::TurnEnd, |
| 4172 | HookEvent::SubagentSpawn, |
| 4173 | HookEvent::SubagentComplete, |
| 4174 | ] { |
| 4175 | assert!( |
| 4176 | !executor.has_hooks_for_event(event), |
| 4177 | "empty config must short-circuit for {event:?}" |
| 4178 | ); |
| 4179 | } |
| 4180 | } |
| 4181 | |
| 4182 | #[test] |
| 4183 | fn has_hooks_for_event_returns_false_when_globally_disabled() { |
| 4184 | let config = HooksConfig { |
| 4185 | enabled: false, |
| 4186 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo blocked")], |
| 4187 | ..HooksConfig::default() |
| 4188 | }; |
| 4189 | let executor = HookExecutor::new(config, PathBuf::from(".")); |
| 4190 | assert!( |
| 4191 | !executor.has_hooks_for_event(HookEvent::ToolCallBefore), |
| 4192 | "globally-disabled hooks must report no fires even when one is configured" |
| 4193 | ); |
| 4194 | } |
| 4195 | |
| 4196 | #[test] |
| 4197 | fn has_hooks_for_event_distinguishes_event_types() { |
| 4198 | let config = HooksConfig { |
| 4199 | enabled: true, |
| 4200 | hooks: vec![ |
| 4201 | Hook::new(HookEvent::SessionStart, "echo start"), |
| 4202 | Hook::new(HookEvent::ToolCallBefore, "echo before"), |
| 4203 | ], |
| 4204 | ..HooksConfig::default() |
| 4205 | }; |
| 4206 | let executor = HookExecutor::new(config, PathBuf::from(".")); |
| 4207 | // Configured events return true. |
| 4208 | assert!(executor.has_hooks_for_event(HookEvent::SessionStart)); |
| 4209 | assert!(executor.has_hooks_for_event(HookEvent::ToolCallBefore)); |
| 4210 | // Unconfigured events return false even when other events are present. |
| 4211 | assert!(!executor.has_hooks_for_event(HookEvent::ToolCallAfter)); |
| 4212 | assert!(!executor.has_hooks_for_event(HookEvent::OnError)); |
| 4213 | assert!(!executor.has_hooks_for_event(HookEvent::ModeChange)); |
| 4214 | } |
| 4215 | |
| 4216 | // ── #3026: tool_call_before stdout decision contract ────────────────── |
| 4217 | |
| 4218 | #[test] |
| 4219 | fn tool_call_before_stdout_parses_deny_with_reason() { |
| 4220 | let parsed = |
| 4221 | parse_tool_call_before_stdout(r#"{"decision":"deny","reason":"blocked by policy"}"#); |
| 4222 | assert_eq!(parsed.decision, Some(ToolCallDecision::Deny)); |
| 4223 | assert_eq!(parsed.reason.as_deref(), Some("blocked by policy")); |
| 4224 | assert!(parsed.updated_input.is_none()); |
| 4225 | assert!(parsed.additional_context.is_none()); |
| 4226 | } |
| 4227 | |
| 4228 | #[test] |
| 4229 | fn tool_call_before_stdout_parses_ask_and_allow() { |
| 4230 | let ask = parse_tool_call_before_stdout(r#"{"decision":"ask"}"#); |
| 4231 | assert_eq!(ask.decision, Some(ToolCallDecision::Ask)); |
| 4232 | |
| 4233 | let allow = parse_tool_call_before_stdout(r#"{"decision":"allow"}"#); |
| 4234 | assert_eq!(allow.decision, Some(ToolCallDecision::Allow)); |
| 4235 | } |
| 4236 | |
| 4237 | #[test] |
| 4238 | fn tool_call_before_stdout_parses_updated_input_object() { |
| 4239 | let parsed = |
| 4240 | parse_tool_call_before_stdout(r#"{"updatedInput":{"command":"ls -la","timeout":5}}"#); |
| 4241 | assert!(parsed.decision.is_none()); |
| 4242 | assert_eq!( |
| 4243 | parsed.updated_input, |
| 4244 | Some(serde_json::json!({"command":"ls -la","timeout":5})) |
| 4245 | ); |
| 4246 | } |
| 4247 | |
| 4248 | #[test] |
| 4249 | fn tool_call_before_stdout_rejects_non_object_updated_input() { |
| 4250 | let parsed = parse_tool_call_before_stdout(r#"{"updatedInput":"rm -rf /"}"#); |
| 4251 | assert!( |
| 4252 | parsed.updated_input.is_none(), |
| 4253 | "updatedInput must be a JSON object" |
| 4254 | ); |
| 4255 | let parsed = parse_tool_call_before_stdout(r#"{"updatedInput":[1,2]}"#); |
| 4256 | assert!(parsed.updated_input.is_none()); |
| 4257 | } |
| 4258 | |
| 4259 | #[test] |
| 4260 | fn tool_call_before_stdout_parses_additional_context() { |
| 4261 | let parsed = |
| 4262 | parse_tool_call_before_stdout(r#"{"additionalContext":"remember the style guide"}"#); |
| 4263 | assert_eq!( |
| 4264 | parsed.additional_context.as_deref(), |
| 4265 | Some("remember the style guide") |
| 4266 | ); |
| 4267 | } |
| 4268 | |
| 4269 | #[test] |
| 4270 | fn tool_call_before_stdout_empty_and_non_json_are_passthrough() { |
| 4271 | for stdout in ["", " \n ", "ok, proceeding", "exit code zero"] { |
| 4272 | let parsed = parse_tool_call_before_stdout(stdout); |
| 4273 | assert!(parsed.decision.is_none(), "stdout {stdout:?}"); |
| 4274 | assert!(parsed.reason.is_none()); |
| 4275 | assert!(parsed.updated_input.is_none()); |
| 4276 | assert!(parsed.additional_context.is_none()); |
| 4277 | } |
| 4278 | } |
| 4279 | |
| 4280 | #[test] |
| 4281 | fn tool_call_before_stdout_json_without_decision_is_passthrough() { |
| 4282 | let parsed = parse_tool_call_before_stdout(r#"{"status":"fine"}"#); |
| 4283 | assert!(parsed.decision.is_none()); |
| 4284 | } |
| 4285 | |
| 4286 | #[test] |
| 4287 | fn tool_call_before_stdout_non_object_json_is_passthrough() { |
| 4288 | for stdout in [r#""deny""#, "[1,2,3]", "42", "true"] { |
| 4289 | let parsed = parse_tool_call_before_stdout(stdout); |
| 4290 | assert!(parsed.decision.is_none(), "stdout {stdout:?}"); |
| 4291 | } |
| 4292 | } |
| 4293 | |
| 4294 | #[test] |
| 4295 | fn tool_call_before_stdout_unknown_decision_treated_as_allow() { |
| 4296 | let parsed = parse_tool_call_before_stdout(r#"{"decision":"block"}"#); |
| 4297 | assert!(parsed.decision.is_none()); |
| 4298 | } |
| 4299 | |
| 4300 | // ── #3026: glob matchers for tool_name conditions ────────────────────── |
| 4301 | |
| 4302 | #[test] |
| 4303 | fn tool_name_glob_matches_mcp_prefix() { |
| 4304 | assert!(HookExecutor::tool_name_matches_condition( |
| 4305 | "mcp__github__create_issue", |
| 4306 | "mcp__*" |
| 4307 | )); |
| 4308 | assert!(!HookExecutor::tool_name_matches_condition( |
| 4309 | "read_file", |
| 4310 | "mcp__*" |
| 4311 | )); |
| 4312 | } |
| 4313 | |
| 4314 | #[test] |
| 4315 | fn tool_name_exact_match_still_works() { |
| 4316 | assert!(HookExecutor::tool_name_matches_condition( |
| 4317 | "read_file", |
| 4318 | "read_file" |
| 4319 | )); |
| 4320 | assert!(!HookExecutor::tool_name_matches_condition( |
| 4321 | "read_files", |
| 4322 | "read_file" |
| 4323 | )); |
| 4324 | } |
| 4325 | |
| 4326 | #[test] |
| 4327 | fn tool_name_glob_escapes_regex_metacharacters() { |
| 4328 | // Without escaping, `.` would match any character. |
| 4329 | assert!(!HookExecutor::tool_name_matches_condition( |
| 4330 | "mcpXgithub", |
| 4331 | "mcp.git*" |
| 4332 | )); |
| 4333 | assert!(HookExecutor::tool_name_matches_condition( |
| 4334 | "mcp.github", |
| 4335 | "mcp.git*" |
| 4336 | )); |
| 4337 | // `+` and parens must be literal too. |
| 4338 | assert!(HookExecutor::tool_name_matches_condition( |
| 4339 | "weird+tool(name)", |
| 4340 | "weird+tool(*)" |
| 4341 | )); |
| 4342 | } |
| 4343 | |
| 4344 | #[test] |
| 4345 | fn tool_name_glob_supports_infix_and_suffix_positions() { |
| 4346 | assert!(HookExecutor::tool_name_matches_condition( |
| 4347 | "mcp__github__create_issue", |
| 4348 | "mcp__*__create_issue" |
| 4349 | )); |
| 4350 | assert!(HookExecutor::tool_name_matches_condition( |
| 4351 | "task_shell_start", |
| 4352 | "*_shell_start" |
| 4353 | )); |
| 4354 | assert!(!HookExecutor::tool_name_matches_condition( |
| 4355 | "task_shell_wait", |
| 4356 | "*_shell_start" |
| 4357 | )); |
| 4358 | } |
| 4359 | |
| 4360 | // ── #3026: project-local hooks ───────────────────────────────────────── |
| 4361 | |
| 4362 | #[test] |
| 4363 | fn load_with_project_missing_file_keeps_global() { |
| 4364 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4365 | let global = HooksConfig { |
| 4366 | enabled: true, |
| 4367 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo global")], |
| 4368 | ..HooksConfig::default() |
| 4369 | }; |
| 4370 | |
| 4371 | let merged = HooksConfig::load_with_project(global.clone(), dir.path()); |
| 4372 | assert_eq!(merged.hooks.len(), 1); |
| 4373 | assert_eq!(merged.hooks[0].command, "echo global"); |
| 4374 | } |
| 4375 | |
| 4376 | #[test] |
| 4377 | fn load_with_project_appends_project_hooks_after_global() { |
| 4378 | let _lock = lock_test_env(); |
| 4379 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4380 | let config_path = dir.path().join("user-config.toml"); |
| 4381 | let _config = trust_workspace_for_project_hooks(dir.path(), &config_path); |
| 4382 | let _legacy_config = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4383 | let project_dir = dir.path().join(".codewhale"); |
| 4384 | std::fs::create_dir_all(&project_dir).expect("mkdir .codewhale"); |
| 4385 | std::fs::write( |
| 4386 | project_dir.join("hooks.toml"), |
| 4387 | r#" |
| 4388 | [[hooks]] |
| 4389 | event = "tool_call_before" |
| 4390 | command = "echo project" |
| 4391 | "#, |
| 4392 | ) |
| 4393 | .expect("write hooks.toml"); |
| 4394 | |
| 4395 | let global = HooksConfig { |
| 4396 | enabled: true, |
| 4397 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo global")], |
| 4398 | ..HooksConfig::default() |
| 4399 | }; |
| 4400 | |
| 4401 | let (authority, _) = super::super::authority::review_project_hooks(dir.path()).unwrap(); |
| 4402 | super::super::authority::approve_project_hooks(dir.path(), &authority.digest).unwrap(); |
| 4403 | let merged = HooksConfig::load_with_project(global, dir.path()); |
| 4404 | assert_eq!(merged.hooks.len(), 2); |
| 4405 | assert_eq!( |
| 4406 | merged.hooks[0].command, "echo global", |
| 4407 | "global hooks run first" |
| 4408 | ); |
| 4409 | assert_eq!( |
| 4410 | merged.hooks[1].command, "echo project", |
| 4411 | "project hooks are appended after global" |
| 4412 | ); |
| 4413 | } |
| 4414 | |
| 4415 | #[test] |
| 4416 | fn load_with_project_ignores_project_hooks_until_workspace_trusted() { |
| 4417 | let _lock = lock_test_env(); |
| 4418 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4419 | let _config = EnvVarGuard::set("CODEWHALE_CONFIG_PATH", dir.path().join("config.toml")); |
| 4420 | let _legacy_config = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4421 | let project_dir = dir.path().join(".codewhale"); |
| 4422 | std::fs::create_dir_all(&project_dir).expect("mkdir .codewhale"); |
| 4423 | std::fs::write( |
| 4424 | project_dir.join("hooks.toml"), |
| 4425 | r#" |
| 4426 | [[hooks]] |
| 4427 | event = "tool_call_before" |
| 4428 | command = "echo project" |
| 4429 | "#, |
| 4430 | ) |
| 4431 | .expect("write hooks.toml"); |
| 4432 | |
| 4433 | let global = HooksConfig { |
| 4434 | enabled: true, |
| 4435 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo global")], |
| 4436 | ..HooksConfig::default() |
| 4437 | }; |
| 4438 | |
| 4439 | let merged = HooksConfig::load_with_project(global, dir.path()); |
| 4440 | assert_eq!(merged.hooks.len(), 1); |
| 4441 | assert_eq!(merged.hooks[0].command, "echo global"); |
| 4442 | } |
| 4443 | |
| 4444 | #[test] |
| 4445 | fn load_with_project_ignores_project_local_legacy_trust_marker() { |
| 4446 | let _lock = lock_test_env(); |
| 4447 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4448 | let _config = EnvVarGuard::set("CODEWHALE_CONFIG_PATH", dir.path().join("config.toml")); |
| 4449 | let _legacy_config = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4450 | let project_dir = dir.path().join(".codewhale"); |
| 4451 | let legacy_trust_dir = dir.path().join(".deepseek"); |
| 4452 | std::fs::create_dir_all(&project_dir).expect("mkdir .codewhale"); |
| 4453 | std::fs::create_dir_all(&legacy_trust_dir).expect("mkdir .deepseek"); |
| 4454 | std::fs::write(legacy_trust_dir.join("trusted"), "").expect("write legacy trust marker"); |
| 4455 | std::fs::write( |
| 4456 | project_dir.join("hooks.toml"), |
| 4457 | r#" |
| 4458 | [[hooks]] |
| 4459 | event = "tool_call_before" |
| 4460 | command = "echo project" |
| 4461 | "#, |
| 4462 | ) |
| 4463 | .expect("write hooks.toml"); |
| 4464 | |
| 4465 | let global = HooksConfig { |
| 4466 | enabled: true, |
| 4467 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo global")], |
| 4468 | ..HooksConfig::default() |
| 4469 | }; |
| 4470 | |
| 4471 | let merged = HooksConfig::load_with_project(global, dir.path()); |
| 4472 | assert_eq!(merged.hooks.len(), 1); |
| 4473 | assert_eq!(merged.hooks[0].command, "echo global"); |
| 4474 | } |
| 4475 | |
| 4476 | #[test] |
| 4477 | fn load_with_project_malformed_file_falls_back_to_global() { |
| 4478 | let _lock = lock_test_env(); |
| 4479 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4480 | let config_path = dir.path().join("user-config.toml"); |
| 4481 | let _config = trust_workspace_for_project_hooks(dir.path(), &config_path); |
| 4482 | let _legacy_config = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4483 | let project_dir = dir.path().join(".codewhale"); |
| 4484 | std::fs::create_dir_all(&project_dir).expect("mkdir .codewhale"); |
| 4485 | std::fs::write(project_dir.join("hooks.toml"), "this is [ not toml") |
| 4486 | .expect("write hooks.toml"); |
| 4487 | |
| 4488 | let global = HooksConfig { |
| 4489 | enabled: true, |
| 4490 | hooks: vec![Hook::new(HookEvent::ToolCallBefore, "echo global")], |
| 4491 | ..HooksConfig::default() |
| 4492 | }; |
| 4493 | |
| 4494 | let merged = HooksConfig::load_with_project(global, dir.path()); |
| 4495 | assert_eq!(merged.hooks.len(), 1, "malformed project file is ignored"); |
| 4496 | assert_eq!(merged.hooks[0].command, "echo global"); |
| 4497 | } |
| 4498 | |
| 4499 | #[cfg(unix)] |
| 4500 | #[test] |
| 4501 | fn project_hooks_require_exact_review_at_load_and_every_spawn() { |
| 4502 | let _lock = lock_test_env(); |
| 4503 | let dir = tempfile::tempdir().unwrap(); |
| 4504 | let _config = trust_workspace_for_project_hooks(dir.path(), &dir.path().join("user.toml")); |
| 4505 | let _legacy = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4506 | std::fs::create_dir(dir.path().join(".codewhale")).unwrap(); |
| 4507 | let hook_path = dir.path().join(".codewhale/hooks.toml"); |
| 4508 | let contents = "[[hooks]]\nevent = \"session_start\"\ncommand = \"touch hook-ran\"\n"; |
| 4509 | std::fs::write(&hook_path, contents).unwrap(); |
| 4510 | let load = || { |
| 4511 | HooksConfig::load_with_project( |
| 4512 | HooksConfig { |
| 4513 | enabled: true, |
| 4514 | ..Default::default() |
| 4515 | }, |
| 4516 | dir.path(), |
| 4517 | ) |
| 4518 | }; |
| 4519 | assert!(load().hooks.is_empty(), "folder trust is not hook approval"); |
| 4520 | let (authority, _) = super::super::authority::review_project_hooks(dir.path()).unwrap(); |
| 4521 | assert!(super::super::authority::approve_project_hooks(dir.path(), "bad-digest").is_err()); |
| 4522 | super::super::authority::approve_project_hooks(dir.path(), &authority.digest).unwrap(); |
| 4523 | let config = load(); |
| 4524 | let hook = config.hooks[0].clone(); |
| 4525 | let executor = HookExecutor::new(config, dir.path().to_path_buf()); |
| 4526 | let good = executor.execute_sync(&hook, &HashMap::new()); |
| 4527 | assert!(good.success, "{good:?}"); |
| 4528 | std::fs::remove_file(dir.path().join("hook-ran")).unwrap(); |
| 4529 | std::fs::write(&hook_path, format!("{contents}# changed\n")).unwrap(); |
| 4530 | assert!(load().hooks.is_empty()); |
| 4531 | assert!(!executor.execute_sync(&hook, &HashMap::new()).success); |
| 4532 | assert!( |
| 4533 | !executor |
| 4534 | .execute_background_inner(&hook, &HashMap::new(), None) |
| 4535 | .success |
| 4536 | ); |
| 4537 | let queued = BackgroundHookJob { |
| 4538 | command: hook.command.clone(), |
| 4539 | env: HashMap::new(), |
| 4540 | working_dir: dir.path().to_path_buf(), |
| 4541 | stdin_bytes: None, |
| 4542 | label: "project".into(), |
| 4543 | timeout: Duration::from_secs(2), |
| 4544 | plugin_authority: None, |
| 4545 | project_authority: hook.project_authority.clone(), |
| 4546 | }; |
| 4547 | queued.run(); |
| 4548 | assert!( |
| 4549 | !dir.path().join("hook-ran").exists(), |
| 4550 | "queued work must revalidate" |
| 4551 | ); |
| 4552 | std::fs::write(&hook_path, contents).unwrap(); |
| 4553 | crate::config::save_workspace_hook_receipt(dir.path(), "").unwrap(); |
| 4554 | assert!(!executor.execute_sync(&hook, &HashMap::new()).success); |
| 4555 | assert!(!dir.path().join("hook-ran").exists()); |
| 4556 | } |
| 4557 | |
| 4558 | #[cfg(unix)] |
| 4559 | #[test] |
| 4560 | fn project_hook_approval_rejects_symlinks_and_repository_receipts() { |
| 4561 | let _lock = lock_test_env(); |
| 4562 | let dir = tempfile::tempdir().unwrap(); |
| 4563 | let _config = trust_workspace_for_project_hooks(dir.path(), &dir.path().join("user.toml")); |
| 4564 | let _legacy = EnvVarGuard::remove("DEEPSEEK_CONFIG_PATH"); |
| 4565 | std::fs::create_dir(dir.path().join(".codewhale")).unwrap(); |
| 4566 | let target = dir.path().join("hook-source.toml"); |
| 4567 | std::fs::write( |
| 4568 | &target, |
| 4569 | "[[hooks]]\nevent = \"session_start\"\ncommand = \"true\"\n", |
| 4570 | ) |
| 4571 | .unwrap(); |
| 4572 | let hook_path = dir.path().join(".codewhale/hooks.toml"); |
| 4573 | std::os::unix::fs::symlink(&target, &hook_path).unwrap(); |
| 4574 | assert!(super::super::authority::review_project_hooks(dir.path()).is_err()); |
| 4575 | std::fs::remove_file(&hook_path).unwrap(); |
| 4576 | std::fs::copy(&target, &hook_path).unwrap(); |
| 4577 | let (authority, _) = super::super::authority::review_project_hooks(dir.path()).unwrap(); |
| 4578 | std::fs::write( |
| 4579 | dir.path().join(".codewhale/config.toml"), |
| 4580 | format!("hooks_sha256 = \"{}\"", authority.digest), |
| 4581 | ) |
| 4582 | .unwrap(); |
| 4583 | assert!( |
| 4584 | HooksConfig::load_with_project(HooksConfig::default(), dir.path()) |
| 4585 | .hooks |
| 4586 | .is_empty() |
| 4587 | ); |
| 4588 | } |
| 4589 | |
| 4590 | // === v0.9.2 hooks contract regression tests =============================== |
| 4591 | // |
| 4592 | // Each of these pins a claim that `docs/HOOKS.md` makes, so the docs cannot |
| 4593 | // drift ahead of the runtime again. All of them are provider-free: they |
| 4594 | // spawn `sh`, never a model. |
| 4595 | |
| 4596 | #[test] |
| 4597 | fn background_result_is_a_submission_not_an_observed_exit_code() { |
| 4598 | // `background` is the flag that keeps "queued" from reading as |
| 4599 | // "exited 0". Steering paths gate on `observed_exit_code`. |
| 4600 | let submitted = HookResult { |
| 4601 | background: true, |
| 4602 | success: true, |
| 4603 | exit_code: None, |
| 4604 | ..HookResult::default() |
| 4605 | }; |
| 4606 | assert!(submitted.background); |
| 4607 | assert_eq!(submitted.observed_exit_code(), None); |
| 4608 | |
| 4609 | // A foreground deny still reads through. |
| 4610 | let denied = HookResult { |
| 4611 | background: false, |
| 4612 | success: false, |
| 4613 | exit_code: Some(2), |
| 4614 | ..HookResult::default() |
| 4615 | }; |
| 4616 | assert_eq!(denied.observed_exit_code(), Some(2)); |
| 4617 | |
| 4618 | // A foreground timeout has no exit code either, but it is *not* a |
| 4619 | // background submission — callers must be able to tell them apart. |
| 4620 | let timed_out = HookResult { |
| 4621 | background: false, |
| 4622 | success: false, |
| 4623 | exit_code: None, |
| 4624 | error: Some("Hook timed out after 1s".to_string()), |
| 4625 | ..HookResult::default() |
| 4626 | }; |
| 4627 | assert!(!timed_out.background); |
| 4628 | assert_eq!(timed_out.observed_exit_code(), None); |
| 4629 | } |
| 4630 | |
| 4631 | #[test] |
| 4632 | fn default_timeout_secs_replaces_per_hook_timeout() { |
| 4633 | // Documented as-implemented: the global value overrides, it does not |
| 4634 | // merely fill in for hooks that omit one. |
| 4635 | let hook = Hook::new(HookEvent::SessionStart, "true").with_timeout(90); |
| 4636 | let overridden = HookExecutor::new( |
| 4637 | HooksConfig { |
| 4638 | default_timeout_secs: Some(5), |
| 4639 | ..HooksConfig::default() |
| 4640 | }, |
| 4641 | PathBuf::from("."), |
| 4642 | ); |
| 4643 | assert_eq!(overridden.effective_timeout_secs(&hook), 5); |
| 4644 | |
| 4645 | let per_hook = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 4646 | assert_eq!(per_hook.effective_timeout_secs(&hook), 90); |
| 4647 | } |
| 4648 | |
| 4649 | #[test] |
| 4650 | fn foreground_timeout_result_is_bounded_and_carries_no_payload() { |
| 4651 | // The timeout result must not leak the stdin payload, the environment, |
| 4652 | // or partial output back to the caller. |
| 4653 | let command = if cfg!(windows) { |
| 4654 | "ping -n 4 127.0.0.1 > nul" |
| 4655 | } else { |
| 4656 | "echo secret-stdout; sleep 5" |
| 4657 | }; |
| 4658 | let hook = Hook::new(HookEvent::MessageSubmit, command).with_timeout(1); |
| 4659 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 4660 | let payload = serde_json::json!({ "text": "super secret user text" }); |
| 4661 | |
| 4662 | let result = executor.execute_sync_with_stdin(&hook, &HashMap::new(), &payload); |
| 4663 | |
| 4664 | assert!(!result.success); |
| 4665 | assert!(!result.background); |
| 4666 | assert_eq!(result.exit_code, None); |
| 4667 | assert!(result.stdout.is_empty(), "stdout leaked: {}", result.stdout); |
| 4668 | assert!(result.stderr.is_empty(), "stderr leaked: {}", result.stderr); |
| 4669 | let error = result.error.unwrap_or_default(); |
| 4670 | assert!(error.contains("timed out"), "{error}"); |
| 4671 | assert!(!error.contains("super secret"), "{error}"); |
| 4672 | } |
| 4673 | |
| 4674 | #[cfg(unix)] |
| 4675 | #[test] |
| 4676 | fn background_hook_timeout_kills_and_reaps_its_process_tree() { |
| 4677 | // The claim under test: "There is no path on which a timed-out hook |
| 4678 | // keeps running." A background hook used to be waited on with an |
| 4679 | // unbounded `child.wait()`, so a runaway command outlived the session. |
| 4680 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4681 | let marker = dir.path().join("survived.txt"); |
| 4682 | // The inner `sh -c ... &` is a grandchild: killing only the immediate |
| 4683 | // shell would leave it alive, so this also covers process-group kill. |
| 4684 | let command = format!( |
| 4685 | "sh -c 'sleep 4; echo survived > {}' & wait", |
| 4686 | marker.display() |
| 4687 | ); |
| 4688 | let hook = Hook::new(HookEvent::SessionStart, &command).with_timeout(1); |
| 4689 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 4690 | |
| 4691 | let result = executor.execute_background(&hook, &HashMap::new()); |
| 4692 | assert!(result.background, "background submission must be flagged"); |
| 4693 | assert_eq!(result.observed_exit_code(), None); |
| 4694 | |
| 4695 | // Well past the hook's 1s budget, and past the 4s the command wanted. |
| 4696 | std::thread::sleep(Duration::from_secs(6)); |
| 4697 | assert!( |
| 4698 | !marker.exists(), |
| 4699 | "background hook outlived its timeout and kept running" |
| 4700 | ); |
| 4701 | } |
| 4702 | |
| 4703 | #[cfg(unix)] |
| 4704 | #[test] |
| 4705 | fn background_hook_receives_the_same_stdin_payload_as_foreground() { |
| 4706 | // Background changes scheduling, not the payload contract. |
| 4707 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4708 | let out = dir.path().join("bg-stdin.json"); |
| 4709 | let command = write_hook_script( |
| 4710 | &dir, |
| 4711 | "capture_bg_stdin.sh", |
| 4712 | &format!("#!/bin/sh\ncat > {}\n", out.display()), |
| 4713 | ); |
| 4714 | let hook = Hook::new(HookEvent::MessageSubmit, &command) |
| 4715 | .with_name("bg") |
| 4716 | .background(); |
| 4717 | let executor = HookExecutor::new( |
| 4718 | HooksConfig { |
| 4719 | enabled: true, |
| 4720 | hooks: vec![hook], |
| 4721 | ..HooksConfig::default() |
| 4722 | }, |
| 4723 | dir.path().to_path_buf(), |
| 4724 | ); |
| 4725 | |
| 4726 | let context = submit_context(&dir); |
| 4727 | let outcome = executor.execute_message_submit_transform(&context, "hello world"); |
| 4728 | // Background hooks cannot steer. |
| 4729 | assert_eq!(outcome, MessageSubmitOutcome::unchanged()); |
| 4730 | |
| 4731 | let raw = wait_for_captured_output(&out); |
| 4732 | let payload: serde_json::Value = serde_json::from_str(raw.trim()).expect("valid JSON"); |
| 4733 | assert_eq!(payload["event"], "message_submit"); |
| 4734 | assert_eq!(payload["text"], "hello world"); |
| 4735 | assert_eq!(payload["session_id"], "sess_test"); |
| 4736 | assert_eq!(payload["mode"], "agent"); |
| 4737 | assert_eq!(payload["model"], "deepseek-test"); |
| 4738 | assert_eq!(payload["total_tokens"], 42); |
| 4739 | } |
| 4740 | |
| 4741 | #[cfg(unix)] |
| 4742 | #[test] |
| 4743 | fn background_hook_receives_the_documented_environment() { |
| 4744 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4745 | let out = dir.path().join("bg-env.txt"); |
| 4746 | let command = write_hook_script( |
| 4747 | &dir, |
| 4748 | "capture_bg_env.sh", |
| 4749 | &format!( |
| 4750 | "#!/bin/sh\nprintf '%s|%s|%s\\n' \"$DEEPSEEK_SESSION_ID\" \"$DEEPSEEK_MODE\" \ |
| 4751 | \"$DEEPSEEK_TOOL_NAME\" > {}\n", |
| 4752 | out.display() |
| 4753 | ), |
| 4754 | ); |
| 4755 | let hook = Hook::new(HookEvent::ToolCallAfter, &command) |
| 4756 | .with_name("bg-env") |
| 4757 | .background(); |
| 4758 | let executor = HookExecutor::new( |
| 4759 | HooksConfig { |
| 4760 | enabled: true, |
| 4761 | hooks: vec![hook], |
| 4762 | ..HooksConfig::default() |
| 4763 | }, |
| 4764 | dir.path().to_path_buf(), |
| 4765 | ); |
| 4766 | |
| 4767 | let context = submit_context(&dir).with_tool_name("exec_shell"); |
| 4768 | let results = executor.execute(HookEvent::ToolCallAfter, &context); |
| 4769 | assert_eq!(results.len(), 1); |
| 4770 | assert!(results[0].background); |
| 4771 | |
| 4772 | let captured = wait_for_captured_output(&out); |
| 4773 | assert_eq!(captured.trim(), "sess_test|agent|exec_shell"); |
| 4774 | } |
| 4775 | |
| 4776 | /// Wait for a background hook's capture file to hold real bytes. |
| 4777 | /// |
| 4778 | /// The capture scripts redirect with `> out`, so the shell creates the |
| 4779 | /// file — empty — before `cat`/`printf` writes the payload. Polling for |
| 4780 | /// existence alone can win that race under load and read an empty capture, |
| 4781 | /// which surfaced in CI as `valid JSON: EOF while parsing a value` (#5929). |
| 4782 | /// Both captures are single small writes, so waiting for non-empty bytes |
| 4783 | /// means the write has landed without weakening what the tests assert. |
| 4784 | #[cfg(unix)] |
| 4785 | fn wait_for_captured_output(path: &std::path::Path) -> String { |
| 4786 | let deadline = std::time::Instant::now() + Duration::from_secs(10); |
| 4787 | loop { |
| 4788 | if let Ok(raw) = std::fs::read_to_string(path) |
| 4789 | && !raw.trim().is_empty() |
| 4790 | { |
| 4791 | return raw; |
| 4792 | } |
| 4793 | assert!( |
| 4794 | std::time::Instant::now() < deadline, |
| 4795 | "background hook wrote no output to {}", |
| 4796 | path.display() |
| 4797 | ); |
| 4798 | std::thread::sleep(Duration::from_millis(50)); |
| 4799 | } |
| 4800 | } |
| 4801 | |
| 4802 | #[test] |
| 4803 | fn session_id_is_stable_across_every_event_and_survives_a_rebind() { |
| 4804 | // One TUI session, one `CODEWHALE_SESSION_ID`. The legacy |
| 4805 | // `DEEPSEEK_SESSION_ID` alias carries the same value so existing hook |
| 4806 | // records stay correlatable; assert both names over every event. |
| 4807 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 4808 | let session_id = executor.session_id().to_string(); |
| 4809 | assert!( |
| 4810 | session_id.starts_with("sess_"), |
| 4811 | "unexpected session id shape: {session_id}" |
| 4812 | ); |
| 4813 | |
| 4814 | for event in crate::hooks::ALL_HOOK_EVENTS { |
| 4815 | let context = HookContext::new() |
| 4816 | .with_session_id(executor.session_id()) |
| 4817 | .with_tool_name(event.as_str()); |
| 4818 | let env = context.to_env_vars(); |
| 4819 | assert_eq!( |
| 4820 | env.get("CODEWHALE_SESSION_ID"), |
| 4821 | Some(&session_id), |
| 4822 | "event `{}` reported a different Codewhale session id", |
| 4823 | event.as_str() |
| 4824 | ); |
| 4825 | assert_eq!( |
| 4826 | env.get("DEEPSEEK_SESSION_ID"), |
| 4827 | Some(&session_id), |
| 4828 | "event `{}` reported a different legacy session id", |
| 4829 | event.as_str() |
| 4830 | ); |
| 4831 | } |
| 4832 | |
| 4833 | // A workspace switch or trust decision reloads the hook set. It must |
| 4834 | // not mint a new identity. |
| 4835 | let rebound = executor.rebind( |
| 4836 | HooksConfig { |
| 4837 | enabled: true, |
| 4838 | hooks: vec![Hook::new(HookEvent::SessionStart, "true")], |
| 4839 | ..HooksConfig::default() |
| 4840 | }, |
| 4841 | PathBuf::from("/tmp"), |
| 4842 | ); |
| 4843 | assert_eq!(rebound.session_id(), session_id); |
| 4844 | assert_eq!(rebound.config().hooks.len(), 1); |
| 4845 | |
| 4846 | // A genuinely new executor is a genuinely new session. |
| 4847 | let fresh = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 4848 | assert_ne!(fresh.session_id(), session_id); |
| 4849 | } |
| 4850 | |
| 4851 | #[test] |
| 4852 | fn exit_code_condition_matches_only_a_real_exit_code() { |
| 4853 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 4854 | let hook = Hook::new(HookEvent::ToolCallAfter, "true") |
| 4855 | .with_condition(HookCondition::ExitCode { code: 1 }); |
| 4856 | |
| 4857 | // No exit code reported at all: must not match. Notably it must not be |
| 4858 | // satisfied by the failure flag either. |
| 4859 | let no_code = HookContext::new() |
| 4860 | .with_tool_name("read_file") |
| 4861 | .with_tool_result("boom", false, None); |
| 4862 | assert!(!executor.matches_condition(&hook, &no_code)); |
| 4863 | |
| 4864 | // A different exit code: no match. |
| 4865 | let other_code = HookContext::new() |
| 4866 | .with_tool_name("exec_shell") |
| 4867 | .with_tool_result("boom", false, Some(127)); |
| 4868 | assert!(!executor.matches_condition(&hook, &other_code)); |
| 4869 | |
| 4870 | // The real thing. |
| 4871 | let exact = HookContext::new() |
| 4872 | .with_tool_name("exec_shell") |
| 4873 | .with_tool_result("boom", false, Some(1)); |
| 4874 | assert!(executor.matches_condition(&hook, &exact)); |
| 4875 | |
| 4876 | // Exit code 0 on a successful call is a real code and matches a |
| 4877 | // `code = 0` predicate. |
| 4878 | let zero_hook = Hook::new(HookEvent::ToolCallAfter, "true") |
| 4879 | .with_condition(HookCondition::ExitCode { code: 0 }); |
| 4880 | let zero = HookContext::new() |
| 4881 | .with_tool_name("exec_shell") |
| 4882 | .with_tool_result("ok", true, Some(0)); |
| 4883 | assert!(executor.matches_condition(&zero_hook, &zero)); |
| 4884 | assert!(!executor.matches_condition(&zero_hook, &no_code)); |
| 4885 | } |
| 4886 | |
| 4887 | #[test] |
| 4888 | fn tool_call_id_is_exported_for_correlation() { |
| 4889 | let env = HookContext::new() |
| 4890 | .with_tool_name("exec_shell") |
| 4891 | .with_tool_call_id("call_abc123") |
| 4892 | .to_env_vars(); |
| 4893 | assert_eq!( |
| 4894 | env.get("CODEWHALE_TOOL_CALL_ID"), |
| 4895 | Some(&"call_abc123".to_string()) |
| 4896 | ); |
| 4897 | assert_eq!( |
| 4898 | env.get("DEEPSEEK_TOOL_CALL_ID"), |
| 4899 | Some(&"call_abc123".to_string()) |
| 4900 | ); |
| 4901 | |
| 4902 | // Absent when unknown — never synthesized. |
| 4903 | let without = HookContext::new() |
| 4904 | .with_tool_name("exec_shell") |
| 4905 | .to_env_vars(); |
| 4906 | assert!(!without.contains_key("CODEWHALE_TOOL_CALL_ID")); |
| 4907 | assert!(!without.contains_key("DEEPSEEK_TOOL_CALL_ID")); |
| 4908 | } |
| 4909 | |
| 4910 | #[test] |
| 4911 | fn tool_exit_code_env_var_is_absent_when_the_tool_reported_none() { |
| 4912 | let with_code = HookContext::new() |
| 4913 | .with_tool_result("out", false, Some(3)) |
| 4914 | .to_env_vars(); |
| 4915 | assert_eq!( |
| 4916 | with_code.get("DEEPSEEK_TOOL_EXIT_CODE"), |
| 4917 | Some(&"3".to_string()) |
| 4918 | ); |
| 4919 | assert_eq!( |
| 4920 | with_code.get("DEEPSEEK_TOOL_SUCCESS"), |
| 4921 | Some(&"false".to_string()) |
| 4922 | ); |
| 4923 | |
| 4924 | let without_code = HookContext::new() |
| 4925 | .with_tool_result("out", false, None) |
| 4926 | .to_env_vars(); |
| 4927 | assert!(!without_code.contains_key("DEEPSEEK_TOOL_EXIT_CODE")); |
| 4928 | assert_eq!( |
| 4929 | without_code.get("DEEPSEEK_TOOL_SUCCESS"), |
| 4930 | Some(&"false".to_string()) |
| 4931 | ); |
| 4932 | } |
| 4933 | |
| 4934 | #[test] |
| 4935 | fn payload_env_vars_are_bounded() { |
| 4936 | // Errors used to be the one unbounded field; a failed `exec_shell` |
| 4937 | // could push its whole output into `DEEPSEEK_ERROR`. |
| 4938 | let long = "x".repeat(20_000); |
| 4939 | let env = HookContext::new() |
| 4940 | .with_error(&long) |
| 4941 | .with_message(&long) |
| 4942 | .with_tool_result(&long, false, None) |
| 4943 | .to_env_vars(); |
| 4944 | |
| 4945 | for key in ["DEEPSEEK_ERROR", "DEEPSEEK_MESSAGE", "DEEPSEEK_TOOL_RESULT"] { |
| 4946 | let value = env.get(key).unwrap_or_else(|| panic!("{key} missing")); |
| 4947 | assert!(value.len() < 20_000, "{key} was not truncated"); |
| 4948 | assert!(value.ends_with("...[truncated]"), "{key} lost its marker"); |
| 4949 | } |
| 4950 | } |
| 4951 | |
| 4952 | #[test] |
| 4953 | fn truncate_env_value_respects_utf8_boundaries() { |
| 4954 | // 4-byte characters straddling the cap must not panic or split. |
| 4955 | let value = "🐋".repeat(100); |
| 4956 | let truncated = super::truncate_env_value(&value, 10); |
| 4957 | assert!(truncated.ends_with("...[truncated]")); |
| 4958 | let head = truncated.trim_end_matches("...[truncated]"); |
| 4959 | assert!(head.chars().all(|c| c == '🐋')); |
| 4960 | assert!(head.len() <= 12); |
| 4961 | } |
| 4962 | |
| 4963 | #[cfg(unix)] |
| 4964 | #[test] |
| 4965 | fn collect_shell_env_merges_later_hooks_over_earlier_ones() { |
| 4966 | // The documented merge: parsed verbatim, later hooks win, failures |
| 4967 | // contribute nothing and do not abort. |
| 4968 | let dir = tempfile::tempdir().expect("tempdir"); |
| 4969 | let first = write_hook_script( |
| 4970 | &dir, |
| 4971 | "env_first.sh", |
| 4972 | "#!/bin/sh\necho SHARED=first\necho ONLY_FIRST=1\n", |
| 4973 | ); |
| 4974 | let second = write_hook_script( |
| 4975 | &dir, |
| 4976 | "env_second.sh", |
| 4977 | "#!/bin/sh\necho SHARED=second\necho QUOTED=\"has spaces\"\n", |
| 4978 | ); |
| 4979 | let failing = write_hook_script(&dir, "env_fail.sh", "#!/bin/sh\necho NEVER=1\nexit 1\n"); |
| 4980 | |
| 4981 | let executor = HookExecutor::new( |
| 4982 | HooksConfig { |
| 4983 | enabled: true, |
| 4984 | hooks: vec![ |
| 4985 | Hook::new(HookEvent::ShellEnv, &first).with_name("first"), |
| 4986 | Hook::new(HookEvent::ShellEnv, &second).with_name("second"), |
| 4987 | Hook::new(HookEvent::ShellEnv, &failing).with_name("failing"), |
| 4988 | ], |
| 4989 | ..HooksConfig::default() |
| 4990 | }, |
| 4991 | dir.path().to_path_buf(), |
| 4992 | ); |
| 4993 | |
| 4994 | let context = HookContext::new().with_tool_name("exec_shell"); |
| 4995 | let merged = executor.collect_shell_env(&context); |
| 4996 | |
| 4997 | assert_eq!(merged.get("SHARED"), Some(&"second".to_string())); |
| 4998 | assert_eq!(merged.get("ONLY_FIRST"), Some(&"1".to_string())); |
| 4999 | assert_eq!(merged.get("QUOTED"), Some(&"has spaces".to_string())); |
| 5000 | assert!( |
| 5001 | !merged.contains_key("NEVER"), |
| 5002 | "a failing shell_env hook must contribute nothing" |
| 5003 | ); |
| 5004 | } |
| 5005 | |
| 5006 | #[cfg(unix)] |
| 5007 | #[test] |
| 5008 | fn shell_env_ignores_the_background_flag_and_still_collects_stdout() { |
| 5009 | // `background` is not honored here: the stdout IS the contract, so the |
| 5010 | // hook runs in the foreground regardless of how it is configured. |
| 5011 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5012 | let script = write_hook_script(&dir, "env_bg.sh", "#!/bin/sh\necho FROM_BG=yes\n"); |
| 5013 | let executor = HookExecutor::new( |
| 5014 | HooksConfig { |
| 5015 | enabled: true, |
| 5016 | hooks: vec![ |
| 5017 | Hook::new(HookEvent::ShellEnv, &script) |
| 5018 | .with_name("bg-shell-env") |
| 5019 | .background(), |
| 5020 | ], |
| 5021 | ..HooksConfig::default() |
| 5022 | }, |
| 5023 | dir.path().to_path_buf(), |
| 5024 | ); |
| 5025 | |
| 5026 | let merged = executor.collect_shell_env(&HookContext::new().with_tool_name("exec_shell")); |
| 5027 | assert_eq!(merged.get("FROM_BG"), Some(&"yes".to_string())); |
| 5028 | } |
| 5029 | |
| 5030 | #[cfg(unix)] |
| 5031 | #[test] |
| 5032 | fn shell_env_hook_receives_only_the_narrow_documented_context() { |
| 5033 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5034 | let out = dir.path().join("shell-env-context.txt"); |
| 5035 | let script = write_hook_script( |
| 5036 | &dir, |
| 5037 | "env_context.sh", |
| 5038 | &format!( |
| 5039 | "#!/bin/sh\nprintf 'name=%s args=%s session=%s mode=%s\\n' \ |
| 5040 | \"$DEEPSEEK_TOOL_NAME\" \"$DEEPSEEK_TOOL_ARGS\" \"$DEEPSEEK_SESSION_ID\" \ |
| 5041 | \"$DEEPSEEK_MODE\" > {}\n", |
| 5042 | out.display() |
| 5043 | ), |
| 5044 | ); |
| 5045 | let executor = HookExecutor::new( |
| 5046 | HooksConfig { |
| 5047 | enabled: true, |
| 5048 | hooks: vec![Hook::new(HookEvent::ShellEnv, &script)], |
| 5049 | ..HooksConfig::default() |
| 5050 | }, |
| 5051 | dir.path().to_path_buf(), |
| 5052 | ); |
| 5053 | |
| 5054 | let context = HookContext::new() |
| 5055 | .with_tool_name("exec_shell") |
| 5056 | .with_tool_args(&serde_json::json!({ "command": "ls" })); |
| 5057 | let _ = executor.collect_shell_env(&context); |
| 5058 | |
| 5059 | let captured = std::fs::read_to_string(&out).expect("shell_env hook wrote nothing"); |
| 5060 | assert!(captured.contains("name=exec_shell"), "{captured}"); |
| 5061 | assert!(captured.contains(r#""command":"ls""#), "{captured}"); |
| 5062 | // No session id or mode is supplied for this event, which is why a |
| 5063 | // `mode` condition on `shell_env` is rejected at load. |
| 5064 | assert!(captured.contains("session= "), "{captured}"); |
| 5065 | assert!(captured.trim_end().ends_with("mode="), "{captured}"); |
| 5066 | } |
| 5067 | |
| 5068 | /// Strictness has to travel on the result, because only the results tell |
| 5069 | /// you which hooks actually *matched* this call. |
| 5070 | #[cfg(unix)] |
| 5071 | #[test] |
| 5072 | fn results_carry_the_strictness_of_the_hook_that_produced_them() { |
| 5073 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5074 | let mut strict = Hook::new(HookEvent::ToolCallBefore, "true") |
| 5075 | .with_name("strict") |
| 5076 | .with_condition(HookCondition::ToolName { |
| 5077 | name: "write_file".to_string(), |
| 5078 | }); |
| 5079 | strict.continue_on_error = false; |
| 5080 | let lenient = Hook::new(HookEvent::ToolCallBefore, "true") |
| 5081 | .with_name("lenient") |
| 5082 | .with_condition(HookCondition::ToolName { |
| 5083 | name: "exec_shell".to_string(), |
| 5084 | }); |
| 5085 | |
| 5086 | let executor = HookExecutor::new( |
| 5087 | HooksConfig { |
| 5088 | enabled: true, |
| 5089 | hooks: vec![strict, lenient], |
| 5090 | ..HooksConfig::default() |
| 5091 | }, |
| 5092 | dir.path().to_path_buf(), |
| 5093 | ); |
| 5094 | |
| 5095 | // Only the lenient hook matches an `exec_shell` call, so nothing about |
| 5096 | // this call is strict — even though a strict hook exists in config. |
| 5097 | let shell = executor.execute( |
| 5098 | HookEvent::ToolCallBefore, |
| 5099 | &HookContext::new().with_tool_name("exec_shell"), |
| 5100 | ); |
| 5101 | assert_eq!(shell.len(), 1); |
| 5102 | assert_eq!(shell[0].name.as_deref(), Some("lenient")); |
| 5103 | assert!(!shell[0].strict); |
| 5104 | |
| 5105 | // The `write_file` call is the one the strict gate guards. |
| 5106 | let write = executor.execute( |
| 5107 | HookEvent::ToolCallBefore, |
| 5108 | &HookContext::new().with_tool_name("write_file"), |
| 5109 | ); |
| 5110 | assert_eq!(write.len(), 1); |
| 5111 | assert_eq!(write[0].name.as_deref(), Some("strict")); |
| 5112 | assert!(write[0].strict); |
| 5113 | } |
| 5114 | |
| 5115 | #[cfg(unix)] |
| 5116 | #[test] |
| 5117 | fn background_results_are_never_strict() { |
| 5118 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5119 | let mut hook = Hook::new(HookEvent::ToolCallBefore, "true") |
| 5120 | .with_name("bg-strict") |
| 5121 | .background(); |
| 5122 | hook.continue_on_error = false; |
| 5123 | let executor = HookExecutor::new(HooksConfig::default(), dir.path().to_path_buf()); |
| 5124 | |
| 5125 | let result = executor.execute_background(&hook, &HashMap::new()); |
| 5126 | assert!(result.background); |
| 5127 | assert!( |
| 5128 | !result.strict, |
| 5129 | "nothing is awaited, so there is no answer to withhold" |
| 5130 | ); |
| 5131 | } |
| 5132 | |
| 5133 | /// A background hook that never reads stdin used to hang the supervising |
| 5134 | /// thread forever: the payload was written synchronously *before* |
| 5135 | /// `wait_timeout`, so a payload larger than the pipe buffer blocked, and |
| 5136 | /// the timeout / kill / reap below it were never reached. |
| 5137 | #[cfg(unix)] |
| 5138 | #[test] |
| 5139 | fn oversized_background_stdin_still_times_out_and_kills_the_tree() { |
| 5140 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5141 | let marker = dir.path().join("survived.txt"); |
| 5142 | // Never reads stdin, and spawns a grandchild so this also covers the |
| 5143 | // process-group kill that the blocked write used to prevent. |
| 5144 | let command = write_hook_script( |
| 5145 | &dir, |
| 5146 | "ignores_stdin.sh", |
| 5147 | &format!( |
| 5148 | "#!/bin/sh\nsh -c 'sleep 6; echo survived > {}' &\nsleep 6\n", |
| 5149 | marker.display() |
| 5150 | ), |
| 5151 | ); |
| 5152 | let hook = Hook::new(HookEvent::TurnEnd, &command) |
| 5153 | .with_name("deaf") |
| 5154 | .background() |
| 5155 | .with_timeout(1); |
| 5156 | let executor = HookExecutor::new( |
| 5157 | HooksConfig { |
| 5158 | enabled: true, |
| 5159 | hooks: vec![hook.clone()], |
| 5160 | ..HooksConfig::default() |
| 5161 | }, |
| 5162 | dir.path().to_path_buf(), |
| 5163 | ); |
| 5164 | |
| 5165 | // Far beyond any pipe buffer (64 KiB on Linux, 8–64 KiB on macOS). |
| 5166 | let payload = json!({ "event": "turn_end", "blob": "x".repeat(4 * 1024 * 1024) }); |
| 5167 | |
| 5168 | let submitted = Instant::now(); |
| 5169 | let result = executor.execute_background_with_stdin(&hook, &HashMap::new(), &payload); |
| 5170 | assert!( |
| 5171 | submitted.elapsed() < Duration::from_secs(2), |
| 5172 | "submission blocked on the stdin write: {:?}", |
| 5173 | submitted.elapsed() |
| 5174 | ); |
| 5175 | assert!(result.background); |
| 5176 | assert!(result.success, "submission failed: {result:?}"); |
| 5177 | |
| 5178 | // Past the hook's 1s budget and past the 6s the command wanted. |
| 5179 | std::thread::sleep(Duration::from_secs(8)); |
| 5180 | assert!( |
| 5181 | !marker.exists(), |
| 5182 | "background hook with an unread oversized stdin outlived its timeout" |
| 5183 | ); |
| 5184 | } |
| 5185 | |
| 5186 | #[test] |
| 5187 | fn spawn_failure_messages_carry_no_command_or_path() { |
| 5188 | let error = std::io::Error::new( |
| 5189 | std::io::ErrorKind::NotFound, |
| 5190 | "'C:\\Users\\dev\\secret hooks\\gate.cmd' is not recognized", |
| 5191 | ); |
| 5192 | let message = super::spawn_failure_message(&error); |
| 5193 | assert!(message.contains("NotFound"), "{message}"); |
| 5194 | assert!(!message.contains("gate.cmd"), "{message}"); |
| 5195 | assert!(!message.contains("C:\\"), "{message}"); |
| 5196 | assert!(!message.contains("secret"), "{message}"); |
| 5197 | } |
| 5198 | |
| 5199 | #[test] |
| 5200 | fn parse_env_lines_drops_nul_bearing_entries() { |
| 5201 | // `Command::env` panics on a NUL in a key or value, so a hook that |
| 5202 | // prints binary garbage must contribute nothing rather than take the |
| 5203 | // tool call down with it. |
| 5204 | let parsed = super::parse_env_lines("GOOD=fine\nBAD=tok\0en\nBA\0D2=x\nALSO_GOOD=2\n"); |
| 5205 | assert_eq!(parsed.get("GOOD"), Some(&"fine".to_string())); |
| 5206 | assert_eq!(parsed.get("ALSO_GOOD"), Some(&"2".to_string())); |
| 5207 | assert!(!parsed.contains_key("BAD"), "{parsed:?}"); |
| 5208 | assert_eq!(parsed.len(), 2, "{parsed:?}"); |
| 5209 | for (key, value) in &parsed { |
| 5210 | assert!(!key.contains('\0')); |
| 5211 | assert!(!value.contains('\0')); |
| 5212 | // The invariants `Command::env` asserts on. |
| 5213 | assert!(!key.is_empty() && !key.contains('=')); |
| 5214 | } |
| 5215 | } |
| 5216 | |
| 5217 | #[test] |
| 5218 | fn parse_env_lines_bounds_values_and_the_aggregate() { |
| 5219 | let huge = "x".repeat(super::SHELL_ENV_VALUE_MAX_BYTES + 1); |
| 5220 | let parsed = super::parse_env_lines(&format!("OK=1\nHUGE={huge}\n")); |
| 5221 | assert_eq!(parsed.get("OK"), Some(&"1".to_string())); |
| 5222 | assert!(!parsed.contains_key("HUGE"), "over-long value was kept"); |
| 5223 | |
| 5224 | // Many individually-legal values still cannot add up to an unbounded |
| 5225 | // environment. |
| 5226 | let chunk = "y".repeat(16 * 1024); |
| 5227 | let mut stdout = String::new(); |
| 5228 | for i in 0..64 { |
| 5229 | stdout.push_str(&format!("K{i}={chunk}\n")); |
| 5230 | } |
| 5231 | let bulk = super::parse_env_lines(&stdout); |
| 5232 | let total: usize = bulk.iter().map(|(k, v)| k.len() + v.len()).sum(); |
| 5233 | assert!(total <= super::SHELL_ENV_TOTAL_MAX_BYTES, "{total} bytes"); |
| 5234 | assert!(!bulk.is_empty(), "the bound must not drop everything"); |
| 5235 | } |
| 5236 | |
| 5237 | #[cfg(unix)] |
| 5238 | #[test] |
| 5239 | fn shell_env_hook_printing_nul_contributes_nothing_and_does_not_panic() { |
| 5240 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5241 | let script = write_hook_script( |
| 5242 | &dir, |
| 5243 | "env_nul.sh", |
| 5244 | "#!/bin/sh\nprintf 'TOKEN=abc\\000def\\n'\nprintf 'SAFE=ok\\n'\n", |
| 5245 | ); |
| 5246 | let executor = HookExecutor::new( |
| 5247 | HooksConfig { |
| 5248 | enabled: true, |
| 5249 | hooks: vec![Hook::new(HookEvent::ShellEnv, &script).with_name("nul")], |
| 5250 | ..HooksConfig::default() |
| 5251 | }, |
| 5252 | dir.path().to_path_buf(), |
| 5253 | ); |
| 5254 | |
| 5255 | let merged = executor.collect_shell_env(&HookContext::new().with_tool_name("exec_shell")); |
| 5256 | assert!(!merged.contains_key("TOKEN"), "{merged:?}"); |
| 5257 | assert_eq!(merged.get("SAFE"), Some(&"ok".to_string())); |
| 5258 | // What `Command::env` would be handed must be panic-free. |
| 5259 | for (key, value) in &merged { |
| 5260 | assert!(!key.is_empty()); |
| 5261 | assert!(!key.contains('=') && !key.contains('\0')); |
| 5262 | assert!(!value.contains('\0')); |
| 5263 | } |
| 5264 | } |
| 5265 | |
| 5266 | #[test] |
| 5267 | fn tool_call_before_text_fields_are_sanitized_and_bounded() { |
| 5268 | let long = "z".repeat(super::HOOK_TEXT_FIELD_MAX_CHARS * 3); |
| 5269 | let stdout = serde_json::json!({ |
| 5270 | "decision": "deny", |
| 5271 | "reason": format!("blocked\u{1b}[31m {long}"), |
| 5272 | "additionalContext": format!("ctx\u{0}\r\nline {long}"), |
| 5273 | }) |
| 5274 | .to_string(); |
| 5275 | |
| 5276 | let parsed = super::parse_tool_call_before_stdout(&stdout); |
| 5277 | |
| 5278 | let reason = parsed.reason.expect("reason kept"); |
| 5279 | assert!(reason.chars().count() <= super::HOOK_TEXT_FIELD_MAX_CHARS + 16); |
| 5280 | assert!(reason.ends_with("…[truncated]"), "{reason}"); |
| 5281 | assert!(!reason.contains('\u{1b}'), "escape sequence survived"); |
| 5282 | |
| 5283 | let context = parsed.additional_context.expect("context kept"); |
| 5284 | assert!(context.chars().count() <= super::HOOK_TEXT_FIELD_MAX_CHARS + 16); |
| 5285 | assert!(!context.contains('\u{0}')); |
| 5286 | assert!(!context.contains('\r')); |
| 5287 | // Legitimate multi-line context still survives. |
| 5288 | assert!( |
| 5289 | context.contains('\n'), |
| 5290 | "{}", |
| 5291 | &context[..40.min(context.len())] |
| 5292 | ); |
| 5293 | } |
| 5294 | |
| 5295 | #[test] |
| 5296 | fn hook_context_bounds_tool_args_environment_value() { |
| 5297 | let env = HookContext::new() |
| 5298 | .with_tool_args(&serde_json::json!({ |
| 5299 | "command": "x".repeat(super::HOOK_TOOL_ARGS_ENV_MAX_BYTES * 3) |
| 5300 | })) |
| 5301 | .to_env_vars(); |
| 5302 | let args = env.get("DEEPSEEK_TOOL_ARGS").expect("tool args env"); |
| 5303 | assert!( |
| 5304 | args.len() <= super::HOOK_TOOL_ARGS_ENV_MAX_BYTES + "...[truncated]".len(), |
| 5305 | "{} bytes", |
| 5306 | args.len() |
| 5307 | ); |
| 5308 | assert!(args.ends_with("...[truncated]")); |
| 5309 | } |
| 5310 | |
| 5311 | #[test] |
| 5312 | fn steering_objects_and_replacement_messages_have_independent_caps() { |
| 5313 | let oversized_input = serde_json::json!({ |
| 5314 | "updatedInput": { "command": "x".repeat(super::HOOK_UPDATED_INPUT_MAX_BYTES * 2) } |
| 5315 | }) |
| 5316 | .to_string(); |
| 5317 | assert!( |
| 5318 | parse_tool_call_before_stdout(&oversized_input) |
| 5319 | .updated_input |
| 5320 | .is_none() |
| 5321 | ); |
| 5322 | |
| 5323 | let oversized_message = serde_json::json!({ |
| 5324 | "text": "x".repeat(super::HOOK_MESSAGE_REPLACEMENT_MAX_CHARS + 1) |
| 5325 | }) |
| 5326 | .to_string(); |
| 5327 | assert!(matches!( |
| 5328 | super::parse_message_submit_stdout(&oversized_message), |
| 5329 | super::MessageSubmitStdout::Invalid(reason) |
| 5330 | if reason.contains("exceeds") |
| 5331 | )); |
| 5332 | } |
| 5333 | |
| 5334 | #[test] |
| 5335 | fn turn_end_error_is_sanitized_and_bounded() { |
| 5336 | let context = HookContext::new(); |
| 5337 | let usage = codewhale_models::Usage::default(); |
| 5338 | let error = format!( |
| 5339 | "boom\u{1b}[2J{}", |
| 5340 | "x".repeat(super::HOOK_TURN_ERROR_MAX_CHARS * 2) |
| 5341 | ); |
| 5342 | let payload = super::turn_end_payload(TurnEndPayloadInput { |
| 5343 | context: &context, |
| 5344 | created_at: chrono::Utc::now(), |
| 5345 | model_backed: true, |
| 5346 | provider: Some("test"), |
| 5347 | billing_surface: None, |
| 5348 | model: Some("test-model"), |
| 5349 | turn_id: "turn_test", |
| 5350 | status: "failed", |
| 5351 | error: Some(&error), |
| 5352 | duration: Duration::from_millis(1), |
| 5353 | usage: &usage, |
| 5354 | totals: TurnEndTotals { |
| 5355 | session_tokens: 0, |
| 5356 | conversation_tokens: 0, |
| 5357 | input_tokens: 0, |
| 5358 | output_tokens: 0, |
| 5359 | }, |
| 5360 | tool_count: 0, |
| 5361 | queued_message_count: 0, |
| 5362 | }); |
| 5363 | let rendered = payload["error"].as_str().expect("bounded error"); |
| 5364 | assert!(!rendered.contains('\u{1b}')); |
| 5365 | assert!(rendered.ends_with("…[truncated]")); |
| 5366 | assert!( |
| 5367 | rendered.chars().count() <= super::HOOK_TURN_ERROR_MAX_CHARS + 16, |
| 5368 | "{} chars", |
| 5369 | rendered.chars().count() |
| 5370 | ); |
| 5371 | } |
| 5372 | |
| 5373 | #[test] |
| 5374 | fn denial_reason_redacts_paths_arguments_and_secret_assignments() { |
| 5375 | let rendered = super::sanitize_hook_denial_reason( |
| 5376 | "denied /Users/alice/private --command token=SUPERSECRET safe", |
| 5377 | ); |
| 5378 | assert_eq!(rendered, "denied [path] [argument] [secret] safe"); |
| 5379 | assert!(!rendered.contains("alice")); |
| 5380 | assert!(!rendered.contains("SUPERSECRET")); |
| 5381 | assert!(!rendered.contains("--command")); |
| 5382 | |
| 5383 | let command = super::sanitize_hook_denial_reason("blocked command rm bearer abc123"); |
| 5384 | assert_eq!(command, "blocked [command] [command] [secret] [secret]"); |
| 5385 | assert!(!command.contains("rm")); |
| 5386 | assert!(!command.contains("abc123")); |
| 5387 | } |
| 5388 | |
| 5389 | #[test] |
| 5390 | fn denial_reason_redacts_adversarial_header_path_and_command_forms() { |
| 5391 | for reason in [ |
| 5392 | r#"Denied Authorization: Bearer TOPSECRET path="/Users/alice/private key" command='rm -rf /tmp/private' safe"#, |
| 5393 | r#"Denied authorization:"Bearer TOPSECRET" path=../private command="curl --header secret" safe"#, |
| 5394 | r#"Denied (Authorization: Bearer TOPSECRET), path = C:\private command = "powershell -enc SECRET" safe"#, |
| 5395 | ] { |
| 5396 | let rendered = super::sanitize_hook_denial_reason(reason); |
| 5397 | for secret in [ |
| 5398 | "TOPSECRET", |
| 5399 | "alice", |
| 5400 | "private key", |
| 5401 | "../private", |
| 5402 | "C:\\private", |
| 5403 | "curl", |
| 5404 | "powershell", |
| 5405 | "SECRET", |
| 5406 | ] { |
| 5407 | assert!(!rendered.contains(secret), "leaked {secret}: {rendered}"); |
| 5408 | } |
| 5409 | assert!(rendered.contains("[secret]"), "{rendered}"); |
| 5410 | assert!(rendered.contains("[path]"), "{rendered}"); |
| 5411 | assert!(rendered.contains("[command]"), "{rendered}"); |
| 5412 | } |
| 5413 | } |
| 5414 | |
| 5415 | #[test] |
| 5416 | fn denial_reason_redacts_auth_schemes_normalized_secrets_and_relative_paths() { |
| 5417 | for reason in [ |
| 5418 | "Denied Authorization: Basic dXNlcjpwYXNz src/private/config.toml", |
| 5419 | "Denied Authorization=Digest deadbeef service.API-KEY=topsecret", |
| 5420 | "Denied authorization Negotiate kerberos AWS_SESSION_TOKEN=abc123", |
| 5421 | "Denied authorization NTLM credential internal_secret=hunter2", |
| 5422 | "Denied authorization Proprietary-Scheme opaque-credential src/private/key.txt", |
| 5423 | ] { |
| 5424 | let rendered = super::sanitize_hook_denial_reason(reason); |
| 5425 | for sensitive in [ |
| 5426 | "dXNlcjpwYXNz", |
| 5427 | "deadbeef", |
| 5428 | "kerberos", |
| 5429 | "credential", |
| 5430 | "topsecret", |
| 5431 | "abc123", |
| 5432 | "hunter2", |
| 5433 | "src/private/config.toml", |
| 5434 | "opaque-credential", |
| 5435 | "src/private/key.txt", |
| 5436 | ] { |
| 5437 | assert!( |
| 5438 | !rendered.contains(sensitive), |
| 5439 | "leaked {sensitive}: {rendered}" |
| 5440 | ); |
| 5441 | } |
| 5442 | assert!(rendered.contains("[secret]"), "{rendered}"); |
| 5443 | } |
| 5444 | } |
| 5445 | |
| 5446 | #[test] |
| 5447 | fn observer_dispatch_failures_are_event_specific_and_fixed() { |
| 5448 | let config = HooksConfig { |
| 5449 | enabled: true, |
| 5450 | hooks: vec![Hook::new(HookEvent::TurnEnd, "true")], |
| 5451 | ..HooksConfig::default() |
| 5452 | }; |
| 5453 | let mut full = HookExecutor::new(config.clone(), PathBuf::from(".")); |
| 5454 | full.inject_observer_dispatch_full_for_test(); |
| 5455 | let error = full |
| 5456 | .submit_observer(HookEvent::TurnEnd, HookContext::new()) |
| 5457 | .expect_err("full queue must be visible"); |
| 5458 | assert_eq!( |
| 5459 | error, |
| 5460 | "turn_end observer hook queue is full; event was not submitted" |
| 5461 | ); |
| 5462 | |
| 5463 | let mut disconnected = HookExecutor::new(config, PathBuf::from(".")); |
| 5464 | disconnected.inject_observer_dispatch_disconnect_for_test(); |
| 5465 | let error = disconnected |
| 5466 | .submit_observer(HookEvent::TurnEnd, HookContext::new()) |
| 5467 | .expect_err("disconnected dispatcher must be visible"); |
| 5468 | assert_eq!( |
| 5469 | error, |
| 5470 | "turn_end observer hook dispatcher is unavailable; event was not submitted" |
| 5471 | ); |
| 5472 | } |
| 5473 | |
| 5474 | #[test] |
| 5475 | fn observer_context_is_bounded_before_enqueue() { |
| 5476 | let huge = "用户".repeat(20_000); |
| 5477 | let bounded = HookContext { |
| 5478 | tool_args: Some(huge.clone()), |
| 5479 | tool_result: Some(huge.clone()), |
| 5480 | error_message: Some(huge.clone()), |
| 5481 | message: Some(huge.clone()), |
| 5482 | model: Some(huge), |
| 5483 | ..HookContext::new() |
| 5484 | } |
| 5485 | .bounded_for_observer(); |
| 5486 | |
| 5487 | assert!(bounded.tool_args.expect("args").len() <= super::HOOK_TOOL_ARGS_ENV_MAX_BYTES + 16); |
| 5488 | assert!( |
| 5489 | bounded.tool_result.expect("result").len() |
| 5490 | <= super::HOOK_TOOL_RESULT_CONTEXT_MAX_BYTES + 16 |
| 5491 | ); |
| 5492 | assert!( |
| 5493 | bounded.error_message.expect("error").len() <= super::HOOK_ERROR_CONTEXT_MAX_BYTES + 16 |
| 5494 | ); |
| 5495 | assert!( |
| 5496 | bounded.message.expect("message").len() <= super::HOOK_MESSAGE_CONTEXT_MAX_BYTES + 16 |
| 5497 | ); |
| 5498 | assert!( |
| 5499 | bounded.model.expect("model").len() <= super::HOOK_OBSERVER_METADATA_MAX_BYTES + 16 |
| 5500 | ); |
| 5501 | } |
| 5502 | |
| 5503 | #[test] |
| 5504 | fn background_supervisor_saturation_is_a_failed_submission() { |
| 5505 | let hook = Hook::new(HookEvent::TurnEnd, "true").background(); |
| 5506 | let mut executor = HookExecutor::new( |
| 5507 | HooksConfig { |
| 5508 | enabled: true, |
| 5509 | hooks: vec![hook], |
| 5510 | ..HooksConfig::default() |
| 5511 | }, |
| 5512 | PathBuf::from("."), |
| 5513 | ); |
| 5514 | executor.inject_background_supervisor_full_for_test(); |
| 5515 | |
| 5516 | let results = executor.execute(HookEvent::TurnEnd, &HookContext::new()); |
| 5517 | assert_eq!(results.len(), 1); |
| 5518 | assert!(results[0].background); |
| 5519 | assert!(!results[0].success); |
| 5520 | assert_eq!( |
| 5521 | results[0].error.as_deref(), |
| 5522 | Some("background hook supervisor queue is full") |
| 5523 | ); |
| 5524 | } |
| 5525 | |
| 5526 | #[cfg(not(windows))] |
| 5527 | #[test] |
| 5528 | fn bounded_observer_dispatcher_executes_a_submitted_event() { |
| 5529 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5530 | let receipt = dir.path().join("observer-receipt.json"); |
| 5531 | let command = write_hook_script( |
| 5532 | &dir, |
| 5533 | "persistent_observer.sh", |
| 5534 | &format!("#!/bin/sh\ncat > '{}'\n", receipt.display()), |
| 5535 | ); |
| 5536 | let executor = HookExecutor::new( |
| 5537 | HooksConfig { |
| 5538 | enabled: true, |
| 5539 | hooks: vec![Hook::new(HookEvent::TurnEnd, &command)], |
| 5540 | ..HooksConfig::default() |
| 5541 | }, |
| 5542 | dir.path().to_path_buf(), |
| 5543 | ); |
| 5544 | executor |
| 5545 | .submit_json_observer( |
| 5546 | HookEvent::TurnEnd, |
| 5547 | HookContext::new(), |
| 5548 | serde_json::json!({"event": "turn_end", "turn_id": "turn_test"}), |
| 5549 | ) |
| 5550 | .expect("bounded submission"); |
| 5551 | |
| 5552 | let deadline = Instant::now() + Duration::from_secs(2); |
| 5553 | let payload = loop { |
| 5554 | if let Ok(raw) = std::fs::read_to_string(&receipt) |
| 5555 | && let Ok(payload) = serde_json::from_str::<serde_json::Value>(&raw) |
| 5556 | { |
| 5557 | break payload; |
| 5558 | } |
| 5559 | assert!( |
| 5560 | Instant::now() < deadline, |
| 5561 | "persistent worker did not finish a valid receipt" |
| 5562 | ); |
| 5563 | std::thread::sleep(Duration::from_millis(10)); |
| 5564 | }; |
| 5565 | assert_eq!(payload["turn_id"], "turn_test"); |
| 5566 | } |
| 5567 | |
| 5568 | #[cfg(unix)] |
| 5569 | #[test] |
| 5570 | fn explicit_message_denial_never_copies_raw_process_diagnostics() { |
| 5571 | let dir = tempfile::tempdir().expect("tempdir"); |
| 5572 | let command = r#"printf '%s\n' '{"reason":"blocked /Users/alice/private --run token=SUPERSECRET"}'; printf '%s\n' 'stderr-secret /tmp/private' >&2; exit 2"#; |
| 5573 | let executor = HookExecutor::new( |
| 5574 | HooksConfig { |
| 5575 | enabled: true, |
| 5576 | hooks: vec![Hook::new(HookEvent::MessageSubmit, command)], |
| 5577 | ..HooksConfig::default() |
| 5578 | }, |
| 5579 | dir.path().to_path_buf(), |
| 5580 | ); |
| 5581 | let outcome = executor.execute_message_submit_transform(&HookContext::new(), "hello"); |
| 5582 | let MessageSubmitOutcome::Blocked { reason } = outcome else { |
| 5583 | panic!("expected explicit block"); |
| 5584 | }; |
| 5585 | assert_eq!(reason, "blocked [path] [argument] [secret]"); |
| 5586 | for secret in ["alice", "SUPERSECRET", "stderr-secret", "/tmp/private"] { |
| 5587 | assert!(!reason.contains(secret), "leaked {secret}: {reason}"); |
| 5588 | } |
| 5589 | } |
| 5590 | |
| 5591 | #[cfg(unix)] |
| 5592 | #[test] |
| 5593 | fn foreground_pipe_capture_is_bounded_while_verbose_child_is_drained() { |
| 5594 | let hook = Hook::new( |
| 5595 | HookEvent::SessionStart, |
| 5596 | "head -c 200000 /dev/zero | tr '\\0' o; head -c 200000 /dev/zero | tr '\\0' e >&2", |
| 5597 | ) |
| 5598 | .with_timeout(5); |
| 5599 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 5600 | let result = executor.execute_sync(&hook, &HashMap::new()); |
| 5601 | assert!(result.success, "{:?}", result.error); |
| 5602 | for output in [&result.stdout, &result.stderr] { |
| 5603 | assert!(output.ends_with("…[truncated]")); |
| 5604 | assert!( |
| 5605 | output.len() <= super::HOOK_PIPE_CAPTURE_MAX_BYTES + "…[truncated]".len(), |
| 5606 | "{} bytes", |
| 5607 | output.len() |
| 5608 | ); |
| 5609 | } |
| 5610 | } |
| 5611 | |
| 5612 | #[cfg(unix)] |
| 5613 | #[test] |
| 5614 | fn helper_wait_and_uncontained_reap_paths_are_bounded() { |
| 5615 | // These helpers reap one immediate child. A shell can fork `sleep` |
| 5616 | // and leave that descendant holding the test's output pipes. |
| 5617 | let mut helper = Command::new("sleep") |
| 5618 | .arg("30") |
| 5619 | .spawn() |
| 5620 | .expect("spawn helper"); |
| 5621 | let started = Instant::now(); |
| 5622 | let error = super::wait_for_helper_status(&mut helper, Duration::from_millis(20)) |
| 5623 | .expect_err("slow helper must time out"); |
| 5624 | assert_eq!(error.kind(), std::io::ErrorKind::TimedOut); |
| 5625 | assert!(started.elapsed() < super::HOOK_REAP_TIMEOUT + Duration::from_secs(1)); |
| 5626 | assert!(matches!(helper.try_wait(), Ok(Some(_)))); |
| 5627 | |
| 5628 | let mut uncontained = Command::new("sleep") |
| 5629 | .arg("30") |
| 5630 | .spawn() |
| 5631 | .expect("spawn uncontained child"); |
| 5632 | assert!(super::kill_and_reap_immediate_child( |
| 5633 | &mut uncontained, |
| 5634 | Duration::from_secs(1) |
| 5635 | )); |
| 5636 | assert!(matches!(uncontained.try_wait(), Ok(Some(_)))); |
| 5637 | } |
| 5638 | |
| 5639 | #[test] |
| 5640 | fn sanitize_hook_text_keeps_short_text_verbatim() { |
| 5641 | assert_eq!( |
| 5642 | super::sanitize_hook_text("plain reason", 100), |
| 5643 | "plain reason" |
| 5644 | ); |
| 5645 | assert_eq!(super::sanitize_hook_text("a\tb\nc", 100), "a\tb\nc"); |
| 5646 | assert_eq!(super::sanitize_hook_text("", 100), ""); |
| 5647 | } |
| 5648 | |
| 5649 | #[test] |
| 5650 | fn sanitize_hook_line_flattens_structure_characters() { |
| 5651 | assert_eq!(super::sanitize_hook_line("a\tb\nc", 100), "a b c"); |
| 5652 | assert_eq!(super::sanitize_hook_line("a\u{1b}[2Jb\r", 100), "a [2Jb"); |
| 5653 | } |
| 5654 | |
| 5655 | #[test] |
| 5656 | fn sanitize_hook_label_bounds_and_defangs_operator_names() { |
| 5657 | let noisy = format!("\u{1b}[2Jgate\twith\nnoise{}", "x".repeat(1_000)); |
| 5658 | let label = super::sanitize_hook_label(Some(&noisy)); |
| 5659 | assert!(!label.contains('\u{1b}'), "{label}"); |
| 5660 | assert!(!label.contains('\n') && !label.contains('\t'), "{label}"); |
| 5661 | assert!(label.contains("gate"), "{label}"); |
| 5662 | assert!( |
| 5663 | label.chars().count() <= super::HOOK_LABEL_MAX_CHARS + 16, |
| 5664 | "{} chars", |
| 5665 | label.chars().count() |
| 5666 | ); |
| 5667 | |
| 5668 | assert_eq!(super::sanitize_hook_label(None), "(unnamed)"); |
| 5669 | assert_eq!(super::sanitize_hook_label(Some("")), "(unnamed)"); |
| 5670 | assert_eq!(super::sanitize_hook_label(Some(" \t ")), "(unnamed)"); |
| 5671 | assert_eq!(super::sanitize_hook_label(Some(" gate ")), "gate"); |
| 5672 | } |
| 5673 | |
| 5674 | /// The point of the boundary: recognized failures are re-rendered from |
| 5675 | /// parts, and anything else — including a string a future producer forgot |
| 5676 | /// to genericize — collapses instead of passing through. |
| 5677 | #[test] |
| 5678 | fn generic_unavailable_detail_is_an_allowlist_not_a_passthrough() { |
| 5679 | use super::generic_unavailable_detail as detail; |
| 5680 | |
| 5681 | assert_eq!( |
| 5682 | detail(Some("Hook timed out after 30s")), |
| 5683 | "hook timed out after 30s" |
| 5684 | ); |
| 5685 | assert_eq!( |
| 5686 | detail(Some("hook process could not be started (NotFound)")), |
| 5687 | "hook process could not be started (NotFound)" |
| 5688 | ); |
| 5689 | assert_eq!( |
| 5690 | detail(Some("Failed to wait for hook: os error 10")), |
| 5691 | "hook did not complete cleanly" |
| 5692 | ); |
| 5693 | assert_eq!( |
| 5694 | detail(Some("hook could not be reaped after its timeout")), |
| 5695 | "hook did not complete cleanly" |
| 5696 | ); |
| 5697 | assert_eq!( |
| 5698 | detail(Some("Failed to submit background hook: os error 11")), |
| 5699 | "hook could not be submitted" |
| 5700 | ); |
| 5701 | assert_eq!( |
| 5702 | detail(Some("hook executor did not run")), |
| 5703 | "hook executor did not run" |
| 5704 | ); |
| 5705 | assert_eq!(detail(None), "hook returned no verdict"); |
| 5706 | |
| 5707 | // A hypothetical future producer that leaks. |
| 5708 | let leaky = "spawn failed: /Users/someone/.aws/credentials --token=SECRET"; |
| 5709 | let rendered = detail(Some(leaky)); |
| 5710 | assert_eq!(rendered, "hook returned no verdict"); |
| 5711 | assert!(!rendered.contains("SECRET")); |
| 5712 | assert!(!rendered.contains('/')); |
| 5713 | |
| 5714 | // And a recognized prefix cannot be used to smuggle a tail along. |
| 5715 | let smuggled = detail(Some( |
| 5716 | "Hook timed out after 30s while running /usr/bin/leak --token=SECRET", |
| 5717 | )); |
| 5718 | assert_eq!(smuggled, "hook timed out after 30s"); |
| 5719 | let smuggled = detail(Some( |
| 5720 | "hook process could not be started (NotFound) /usr/bin/leak", |
| 5721 | )); |
| 5722 | assert_eq!(smuggled, "hook process could not be started (NotFound)"); |
| 5723 | } |
| 5724 | |
| 5725 | /// The gate set the caller has to fail closed on if the executor is lost. |
| 5726 | #[cfg(unix)] |
| 5727 | #[test] |
| 5728 | fn matched_strict_gate_labels_names_only_gates_that_would_run() { |
| 5729 | use crate::hooks::{Hook, HookCondition, HookEvent, HooksConfig}; |
| 5730 | |
| 5731 | let strict_shell = { |
| 5732 | let mut hook = Hook::new(HookEvent::ToolCallBefore, "true") |
| 5733 | .with_name("shell-gate") |
| 5734 | .with_condition(HookCondition::ToolName { |
| 5735 | name: "exec_shell".into(), |
| 5736 | }); |
| 5737 | hook.continue_on_error = false; |
| 5738 | hook |
| 5739 | }; |
| 5740 | let strict_write = { |
| 5741 | let mut hook = Hook::new(HookEvent::ToolCallBefore, "true") |
| 5742 | .with_name("write-gate") |
| 5743 | .with_condition(HookCondition::ToolName { |
| 5744 | name: "write_file".into(), |
| 5745 | }); |
| 5746 | hook.continue_on_error = false; |
| 5747 | hook |
| 5748 | }; |
| 5749 | let lenient_shell = Hook::new(HookEvent::ToolCallBefore, "true").with_name("lenient"); |
| 5750 | let background_strict = { |
| 5751 | let mut hook = Hook::new(HookEvent::ToolCallBefore, "true").with_name("bg-gate"); |
| 5752 | hook.continue_on_error = false; |
| 5753 | hook.background = true; |
| 5754 | hook |
| 5755 | }; |
| 5756 | let other_event = { |
| 5757 | let mut hook = Hook::new(HookEvent::ToolCallAfter, "true").with_name("after-gate"); |
| 5758 | hook.continue_on_error = false; |
| 5759 | hook |
| 5760 | }; |
| 5761 | |
| 5762 | let executor = HookExecutor::new( |
| 5763 | HooksConfig { |
| 5764 | enabled: true, |
| 5765 | hooks: vec![ |
| 5766 | strict_shell, |
| 5767 | strict_write, |
| 5768 | lenient_shell, |
| 5769 | background_strict, |
| 5770 | other_event, |
| 5771 | ], |
| 5772 | ..HooksConfig::default() |
| 5773 | }, |
| 5774 | std::env::temp_dir(), |
| 5775 | ); |
| 5776 | |
| 5777 | let labels = executor.matched_strict_gate_labels( |
| 5778 | HookEvent::ToolCallBefore, |
| 5779 | &HookContext::new().with_tool_name("exec_shell"), |
| 5780 | ); |
| 5781 | assert_eq!(labels, vec!["shell-gate".to_string()], "{labels:?}"); |
| 5782 | |
| 5783 | // Globally disabled hooks are not gates either. |
| 5784 | let disabled = HookExecutor::disabled(); |
| 5785 | assert!( |
| 5786 | disabled |
| 5787 | .matched_strict_gate_labels( |
| 5788 | HookEvent::ToolCallBefore, |
| 5789 | &HookContext::new().with_tool_name("exec_shell"), |
| 5790 | ) |
| 5791 | .is_empty() |
| 5792 | ); |
| 5793 | } |
| 5794 | |
| 5795 | /// The reap after a kill is bounded. This asserts the ordinary case is |
| 5796 | /// still confirmed dead and, more importantly, that the call returns — |
| 5797 | /// the regression it guards is a hang, not a wrong value. |
| 5798 | #[cfg(unix)] |
| 5799 | #[test] |
| 5800 | fn timed_out_hook_is_killed_and_reaped_within_the_bound() { |
| 5801 | use crate::hooks::{Hook, HookEvent, HooksConfig}; |
| 5802 | |
| 5803 | let hook = Hook::new(HookEvent::SessionStart, "sleep 30") |
| 5804 | .with_name("slow") |
| 5805 | .with_timeout(1); |
| 5806 | let executor = HookExecutor::new( |
| 5807 | HooksConfig { |
| 5808 | enabled: true, |
| 5809 | hooks: vec![hook], |
| 5810 | ..HooksConfig::default() |
| 5811 | }, |
| 5812 | std::env::temp_dir(), |
| 5813 | ); |
| 5814 | |
| 5815 | let started = Instant::now(); |
| 5816 | let results = executor.execute(HookEvent::SessionStart, &HookContext::new()); |
| 5817 | let elapsed = started.elapsed(); |
| 5818 | |
| 5819 | assert_eq!(results.len(), 1); |
| 5820 | assert_eq!( |
| 5821 | results[0].error.as_deref(), |
| 5822 | Some("Hook timed out after 1s"), |
| 5823 | "the child was reaped, so the stronger claim is the honest one" |
| 5824 | ); |
| 5825 | assert!( |
| 5826 | elapsed < Duration::from_secs(1) + super::HOOK_REAP_TIMEOUT + Duration::from_secs(5), |
| 5827 | "timeout path took {elapsed:?}" |
| 5828 | ); |
| 5829 | } |
| 5830 | |
| 5831 | #[test] |
| 5832 | fn tool_exit_code_env_var_survives_a_windows_crash_code() { |
| 5833 | // 0xC0000005 (access violation) does not fit in an `i32`. It used to |
| 5834 | // be dropped on the floor before the hook ever saw it. |
| 5835 | let env = HookContext::new() |
| 5836 | .with_tool_result("crashed", false, Some(3_221_225_477)) |
| 5837 | .to_env_vars(); |
| 5838 | assert_eq!( |
| 5839 | env.get("DEEPSEEK_TOOL_EXIT_CODE"), |
| 5840 | Some(&"3221225477".to_string()) |
| 5841 | ); |
| 5842 | |
| 5843 | let executor = HookExecutor::new(HooksConfig::default(), PathBuf::from(".")); |
| 5844 | let hook = |
| 5845 | Hook::new(HookEvent::ToolCallAfter, "true").with_condition(HookCondition::ExitCode { |
| 5846 | code: 3_221_225_477, |
| 5847 | }); |
| 5848 | let context = HookContext::new() |
| 5849 | .with_tool_name("exec_shell") |
| 5850 | .with_tool_result("crashed", false, Some(3_221_225_477)); |
| 5851 | assert!(executor.matches_condition(&hook, &context)); |
| 5852 | } |
| 5853 | |
| 5854 | /// 2026-08-04: the category map knew only retired tool names, so every |
| 5855 | /// live call fell through to `other` and a `tool_category` deny hook — |
| 5856 | /// the security control `docs/HOOKS.md` documents — silently never fired. |
| 5857 | #[test] |
| 5858 | fn tool_category_classifies_the_names_the_registry_actually_registers() { |
| 5859 | use super::tool_category_for; |
| 5860 | |
| 5861 | // Anchor to the real catalog. Everything below this pins hardcoded |
| 5862 | // names, which would stay green through a tool rename while the gate |
| 5863 | // quietly reclassified the renamed tool. `DEFAULT_ACTIVE_NATIVE_TOOLS` |
| 5864 | // is the list the engine actually puts on the wire, so if a name here |
| 5865 | // stops being a name the product ships, this fails first. |
| 5866 | // |
| 5867 | // Note the fallback is "other", not "safe" — asserting against "safe" |
| 5868 | // here would never fire. This table is checked in both directions, so |
| 5869 | // a rename fails on the missing entry and a classifier change fails on |
| 5870 | // the mismatched category. |
| 5871 | const EXPECTED: &[(&str, &str)] = &[ |
| 5872 | ("read", "safe"), |
| 5873 | ("write", "file_write"), |
| 5874 | ("edit", "file_write"), |
| 5875 | ("bash", "shell"), |
| 5876 | // The router itself touches nothing a hook needs to gate. |
| 5877 | ("agent", "other"), |
| 5878 | ("workflow", "other"), |
| 5879 | ("todo_write", "safe"), |
| 5880 | // Goal controls retain their existing hook classification when |
| 5881 | // promoted from deferred discovery to the eager catalog. |
| 5882 | ("create_goal", "other"), |
| 5883 | ("get_goal", "other"), |
| 5884 | ("update_goal", "other"), |
| 5885 | ]; |
| 5886 | for name in crate::core::engine::tool_catalog::DEFAULT_ACTIVE_NATIVE_TOOLS { |
| 5887 | let expected = EXPECTED.iter().find(|(n, _)| n == name).map(|(_, c)| *c); |
| 5888 | assert_eq!( |
| 5889 | Some(tool_category_for(name, None)), |
| 5890 | expected, |
| 5891 | "default-active tool {name:?} is not covered by this test's \ |
| 5892 | table. It was renamed or added without updating the hook \ |
| 5893 | gate's classifier, so the gate now sees a shipped tool it \ |
| 5894 | does not recognise." |
| 5895 | ); |
| 5896 | } |
| 5897 | for (name, _) in EXPECTED { |
| 5898 | assert!( |
| 5899 | crate::core::engine::tool_catalog::DEFAULT_ACTIVE_NATIVE_TOOLS.contains(name), |
| 5900 | "{name:?} is pinned here but is no longer default-active; drop \ |
| 5901 | it so this table keeps describing what actually ships." |
| 5902 | ); |
| 5903 | } |
| 5904 | |
| 5905 | // The shell surface. |
| 5906 | assert_eq!(tool_category_for("Bash", None), "shell"); |
| 5907 | // Retained: shell.rs stamps this for the shell_env event. |
| 5908 | assert_eq!(tool_category_for("exec_shell", None), "shell"); |
| 5909 | // Run executes commands, so it gates with shell rather than safe. |
| 5910 | assert_eq!(tool_category_for("Run", None), "shell"); |
| 5911 | |
| 5912 | // File is multi-action: the action decides. |
| 5913 | let read = r#"{"action":"read","path":"a.rs"}"#; |
| 5914 | let write = r#"{"action":"write","path":"a.rs","content":"x"}"#; |
| 5915 | assert_eq!(tool_category_for("File", Some(read)), "safe"); |
| 5916 | assert_eq!(tool_category_for("File", Some(write)), "file_write"); |
| 5917 | assert_eq!( |
| 5918 | tool_category_for("File", Some(r#"{"action":"edit"}"#)), |
| 5919 | "file_write" |
| 5920 | ); |
| 5921 | assert_eq!( |
| 5922 | tool_category_for("File", Some(r#"{"action":"search_content"}"#)), |
| 5923 | "safe" |
| 5924 | ); |
| 5925 | |
| 5926 | // A gate that cannot see the action must assume the dangerous one. |
| 5927 | assert_eq!(tool_category_for("File", None), "file_write"); |
| 5928 | assert_eq!(tool_category_for("File", Some("not json")), "file_write"); |
| 5929 | |
| 5930 | assert_eq!(tool_category_for("apply_patch", None), "file_write"); |
| 5931 | assert_eq!( |
| 5932 | tool_category_for("Git", Some(r#"{"action":"log"}"#)), |
| 5933 | "safe" |
| 5934 | ); |
| 5935 | assert_eq!( |
| 5936 | tool_category_for("Git", Some(r#"{"action":"commit_plan"}"#)), |
| 5937 | "safe" |
| 5938 | ); |
| 5939 | assert_eq!(tool_category_for("web.run", None), "other"); |
| 5940 | } |
| 5941 | } |
| 5942 |