| 1 | //! External-binary dependency resolution for tools that shell out to |
| 2 | //! locally-installed programs (Python for `code_execution` / RLM REPL, |
| 3 | //! `pdftotext` for PDF reading in `read_file`, future tools as added). |
| 4 | //! |
| 5 | //! Before v0.8.31, tools that called external binaries hardcoded the |
| 6 | //! command name and failed at execution time when the binary wasn't on |
| 7 | //! `PATH`. The most-cited example was `code_execution`, which spawned |
| 8 | //! `python3` directly — Windows users (where the launcher is `py` or |
| 9 | //! `python`, not `python3`) saw `Failed to execute tool: program not |
| 10 | //! found` with no upstream hint of what was wrong. |
| 11 | //! |
| 12 | //! This module centralises the probe-then-decide pattern. The supported |
| 13 | //! callers today are: |
| 14 | //! |
| 15 | //! - Tool catalog construction (`core::engine::tool_catalog`): for |
| 16 | //! tools that should be advertised to the model only when the |
| 17 | //! required runtime is present. |
| 18 | //! - Doctor command (`run_doctor` in `main.rs`): for surfacing the |
| 19 | //! resolved state to the user so missing dependencies aren't an |
| 20 | //! invisible failure. |
| 21 | //! - Long-lived REPL runtime (`repl::runtime`): for RLM and inline `repl` |
| 22 | //! blocks that need to spawn Python on every supported platform. |
| 23 | //! |
| 24 | //! Results are cached for the process lifetime via [`std::sync::OnceLock`] |
| 25 | //! — probing a binary involves a `Command::output` per candidate and |
| 26 | //! we'd rather not pay that on every model turn. |
| 27 | |
| 28 | use std::path::{Path, PathBuf}; |
| 29 | use std::process::Command; |
| 30 | use std::sync::OnceLock; |
| 31 | |
| 32 | /// Candidate executable names for the Python interpreter, in the |
| 33 | /// order we try them. On Windows the launcher convention is `py -3`, |
| 34 | /// so we add it as a third option; the resolver splits on whitespace |
| 35 | /// at execution time so `py -3 /tmp/code.py` runs correctly. |
| 36 | /// |
| 37 | /// Order matters: `python3` first because it's the unambiguous v3 |
| 38 | /// binary on Unix and rules out Python 2 leftovers. `python` second |
| 39 | /// covers Windows installations that drop the version suffix and |
| 40 | /// modern macOS where Homebrew installs both. `py -3` last as a |
| 41 | /// Windows-launcher fallback. |
| 42 | pub const PYTHON_CANDIDATES: &[&str] = &["python3", "python", "py -3"]; |
| 43 | |
| 44 | /// Probe a single executable. Returns `true` when the candidate |
| 45 | /// responds to `--version` with a successful exit. Splits on |
| 46 | /// whitespace so `"py -3"` works as a candidate. |
| 47 | /// |
| 48 | /// We deliberately use `--version` rather than `which` so the probe |
| 49 | /// is portable across Unix, Windows (no `which` by default), and |
| 50 | /// containers. The downside is that we spawn a subprocess per |
| 51 | /// candidate; the resolver caches the result so this only fires |
| 52 | /// once per process. |
| 53 | #[must_use] |
| 54 | pub fn probe_executable(spec: &str) -> bool { |
| 55 | probe_executable_with_flag(spec, "--version") |
| 56 | } |
| 57 | |
| 58 | /// Probe a single executable using an explicit version/help flag. |
| 59 | /// |
| 60 | /// Most tools report their presence via `--version`, but some do not: |
| 61 | /// Poppler's `pdftotext` treats `--version` as an input *filename* and |
| 62 | /// exits non-zero ("I/O Error: Couldn't open file '--version'"), so the |
| 63 | /// default probe reports it missing even when it is installed (#1667). |
| 64 | /// Such tools pass their own flag (e.g. `-v`) here. |
| 65 | #[must_use] |
| 66 | pub fn probe_executable_with_flag(spec: &str, version_flag: &str) -> bool { |
| 67 | let mut parts = spec.split_whitespace(); |
| 68 | let Some(program) = parts.next() else { |
| 69 | return false; |
| 70 | }; |
| 71 | let mut cmd = Command::new(program); |
| 72 | crate::utils::suppress_console_window(&mut cmd); |
| 73 | for arg in parts { |
| 74 | cmd.arg(arg); |
| 75 | } |
| 76 | cmd.arg(version_flag); |
| 77 | |
| 78 | // Silence the subprocess's stdout/stderr — the version banner would |
| 79 | // otherwise print to our terminal during startup, which is |
| 80 | // confusing on the TUI's first frame. |
| 81 | cmd.stdout(std::process::Stdio::null()); |
| 82 | cmd.stderr(std::process::Stdio::null()); |
| 83 | |
| 84 | matches!(cmd.status(), Ok(status) if status.success()) |
| 85 | } |
| 86 | |
| 87 | /// Probe a single executable and capture its version banner in one spawn. |
| 88 | /// |
| 89 | /// Same contract as [`probe_executable`] (success = exit 0), but returns the |
| 90 | /// trimmed stdout so callers that want the banner don't need a second process |
| 91 | /// launch. Returns `None` when the probe fails or stdout is not valid UTF-8. |
| 92 | pub fn probe_executable_capturing(spec: &str, version_flag: &str) -> Option<String> { |
| 93 | let mut parts = spec.split_whitespace(); |
| 94 | let program = parts.next()?; |
| 95 | let mut cmd = Command::new(program); |
| 96 | crate::utils::suppress_console_window(&mut cmd); |
| 97 | for arg in parts { |
| 98 | cmd.arg(arg); |
| 99 | } |
| 100 | cmd.arg(version_flag); |
| 101 | cmd.stderr(std::process::Stdio::null()); |
| 102 | |
| 103 | let output = cmd.output().ok()?; |
| 104 | if !output.status.success() { |
| 105 | return None; |
| 106 | } |
| 107 | String::from_utf8(output.stdout) |
| 108 | .ok() |
| 109 | .map(|s| s.trim().to_string()) |
| 110 | .filter(|s| !s.is_empty()) |
| 111 | } |
| 112 | |
| 113 | fn executable_path_candidates(program: &str) -> Vec<PathBuf> { |
| 114 | let program_path = Path::new(program); |
| 115 | if program_path.components().count() > 1 { |
| 116 | return vec![program_path.to_path_buf()]; |
| 117 | } |
| 118 | |
| 119 | let Some(path) = std::env::var_os("PATH") else { |
| 120 | return vec![PathBuf::from(program)]; |
| 121 | }; |
| 122 | |
| 123 | let mut candidates = Vec::new(); |
| 124 | for dir in std::env::split_paths(&path) { |
| 125 | let bare = dir.join(program); |
| 126 | candidates.push(bare.clone()); |
| 127 | |
| 128 | #[cfg(windows)] |
| 129 | if Path::new(program).extension().is_none() { |
| 130 | let pathext = |
| 131 | std::env::var_os("PATHEXT").unwrap_or_else(|| ".COM;.EXE;.BAT;.CMD".into()); |
| 132 | for ext in pathext.to_string_lossy().split(';') { |
| 133 | if ext.is_empty() { |
| 134 | continue; |
| 135 | } |
| 136 | candidates.push(bare.with_extension(ext.trim_start_matches('.'))); |
| 137 | } |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | candidates |
| 142 | } |
| 143 | |
| 144 | fn resolve_executable_path(spec: &str, version_flag: &str) -> Option<String> { |
| 145 | let mut parts = spec.split_whitespace(); |
| 146 | let program = parts.next()?; |
| 147 | let args: Vec<&str> = parts.collect(); |
| 148 | |
| 149 | for candidate in executable_path_candidates(program) { |
| 150 | if !candidate.is_file() { |
| 151 | continue; |
| 152 | } |
| 153 | |
| 154 | let mut cmd = Command::new(&candidate); |
| 155 | cmd.args(&args) |
| 156 | .arg(version_flag) |
| 157 | .stdout(std::process::Stdio::null()) |
| 158 | .stderr(std::process::Stdio::null()); |
| 159 | |
| 160 | if matches!(cmd.status(), Ok(status) if status.success()) { |
| 161 | return Some(candidate.to_string_lossy().into_owned()); |
| 162 | } |
| 163 | } |
| 164 | |
| 165 | None |
| 166 | } |
| 167 | |
| 168 | /// Resolve the Python interpreter once per process. Returns the |
| 169 | /// candidate spec (e.g. `"python3"` or `"py -3"`) that succeeded, |
| 170 | /// or `None` when every candidate failed. |
| 171 | /// |
| 172 | /// Callers that need to spawn the interpreter should split this |
| 173 | /// string on whitespace — see [`split_interpreter_spec`]. |
| 174 | pub fn resolve_python_interpreter() -> Option<String> { |
| 175 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 176 | CACHE |
| 177 | .get_or_init(|| { |
| 178 | for candidate in PYTHON_CANDIDATES { |
| 179 | if probe_executable(candidate) { |
| 180 | tracing::info!( |
| 181 | target: "tool_dependencies", |
| 182 | candidate = candidate, |
| 183 | "Resolved Python interpreter", |
| 184 | ); |
| 185 | return Some((*candidate).to_string()); |
| 186 | } |
| 187 | } |
| 188 | tracing::warn!( |
| 189 | target: "tool_dependencies", |
| 190 | tried = ?PYTHON_CANDIDATES, |
| 191 | "No Python interpreter found", |
| 192 | ); |
| 193 | None |
| 194 | }) |
| 195 | .clone() |
| 196 | } |
| 197 | |
| 198 | /// Resolve `pdftotext` (from Poppler) once per process. Used by |
| 199 | /// file and web PDF paths for truthful availability diagnostics. Unlike |
| 200 | /// the Python case, `read_file` itself still works for text files |
| 201 | /// when `pdftotext` is missing — this resolver exists so the doctor |
| 202 | /// command can surface the miss before a PDF read returns its typed |
| 203 | /// `binary_unavailable` result. |
| 204 | pub fn resolve_pdftotext() -> Option<String> { |
| 205 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 206 | CACHE |
| 207 | .get_or_init(|| { |
| 208 | // Poppler's `pdftotext` rejects `--version` (it is parsed as an |
| 209 | // input filename and exits non-zero), so probe with `-v`, which |
| 210 | // prints the version banner and exits 0 (#1667). |
| 211 | if probe_executable_with_flag("pdftotext", "-v") { |
| 212 | Some("pdftotext".to_string()) |
| 213 | } else { |
| 214 | None |
| 215 | } |
| 216 | }) |
| 217 | .clone() |
| 218 | } |
| 219 | |
| 220 | /// Resolve `tesseract` (OCR engine) once per process. Used by the |
| 221 | /// `image_ocr` tool on platforms that do not have a native OCR backend. |
| 222 | /// Tesseract is the de-facto open-source OCR engine and ships as a single |
| 223 | /// binary on every platform we support, so the candidate list is just |
| 224 | /// `tesseract`. |
| 225 | pub fn resolve_tesseract() -> Option<String> { |
| 226 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 227 | CACHE |
| 228 | .get_or_init(|| { |
| 229 | if probe_executable("tesseract") { |
| 230 | tracing::info!( |
| 231 | target: "tool_dependencies", |
| 232 | "Resolved tesseract binary for image_ocr", |
| 233 | ); |
| 234 | Some("tesseract".to_string()) |
| 235 | } else { |
| 236 | tracing::warn!( |
| 237 | target: "tool_dependencies", |
| 238 | "tesseract binary not found; image_ocr will rely on native OCR if available", |
| 239 | ); |
| 240 | None |
| 241 | } |
| 242 | }) |
| 243 | .clone() |
| 244 | } |
| 245 | |
| 246 | /// Resolve `pandoc` (universal document converter) once per |
| 247 | /// process. Used by the `pandoc_convert` tool to decide whether |
| 248 | /// to register itself with the model. Pandoc is a single-binary |
| 249 | /// install, so the candidate list is just `pandoc` — no platform |
| 250 | /// fallback path. |
| 251 | pub fn resolve_pandoc() -> Option<String> { |
| 252 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 253 | CACHE |
| 254 | .get_or_init(|| { |
| 255 | if let Some(path) = resolve_executable_path("pandoc", "--version") { |
| 256 | tracing::info!( |
| 257 | target: "tool_dependencies", |
| 258 | "Resolved pandoc binary for pandoc_convert", |
| 259 | ); |
| 260 | Some(path) |
| 261 | } else { |
| 262 | tracing::warn!( |
| 263 | target: "tool_dependencies", |
| 264 | "pandoc binary not found; pandoc_convert tool will not be registered", |
| 265 | ); |
| 266 | None |
| 267 | } |
| 268 | }) |
| 269 | .clone() |
| 270 | } |
| 271 | |
| 272 | /// Resolve the Node.js runtime once per process. Used by the |
| 273 | /// `js_execution` tool to decide whether to advertise itself in |
| 274 | /// the catalog. Unlike Python, the executable name `node` is the |
| 275 | /// same across every platform we ship to — there's no `node3` or |
| 276 | /// `node.exe` variant to fall through to — so this is a single |
| 277 | /// probe rather than a candidate ladder. |
| 278 | pub fn resolve_node() -> Option<String> { |
| 279 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 280 | CACHE |
| 281 | .get_or_init(|| { |
| 282 | if probe_executable("node") { |
| 283 | tracing::info!( |
| 284 | target: "tool_dependencies", |
| 285 | "Resolved Node.js runtime for js_execution", |
| 286 | ); |
| 287 | Some("node".to_string()) |
| 288 | } else { |
| 289 | tracing::warn!( |
| 290 | target: "tool_dependencies", |
| 291 | "Node.js runtime not found; js_execution tool will not be advertised", |
| 292 | ); |
| 293 | None |
| 294 | } |
| 295 | }) |
| 296 | .clone() |
| 297 | } |
| 298 | |
| 299 | // --------------------------------------------------------------------------- |
| 300 | // ExternalTool trait — unified subprocess interface |
| 301 | // --------------------------------------------------------------------------- |
| 302 | |
| 303 | /// A tool that DeepSeek-TUI shells out to. Instead of scattering |
| 304 | /// `Command::new("git")` / `Command::new("gh")` across the codebase, |
| 305 | /// each external dependency implements this trait once in this module. |
| 306 | /// Callers ask the tool for a pre-populated [`Command`] and chain their |
| 307 | /// own args, working directory, and spawn method. |
| 308 | /// |
| 309 | /// # Example |
| 310 | /// |
| 311 | /// ```ignore |
| 312 | /// let output = Git::command() |
| 313 | /// .expect("git not found") |
| 314 | /// .args(["diff", "--stat"]) |
| 315 | /// .current_dir(&workspace) |
| 316 | /// .output()?; |
| 317 | /// ``` |
| 318 | pub trait ExternalTool { |
| 319 | /// Candidate binary names, tried in order until one responds to |
| 320 | /// `--version`. For single-binary tools (git, gh, node) this is a |
| 321 | /// one-element slice. |
| 322 | fn candidates() -> &'static [&'static str]; |
| 323 | |
| 324 | /// Resolve the best candidate once per process (cached). Returns |
| 325 | /// the spec string (e.g. `"python3"` or `"py -3"`). |
| 326 | fn resolve() -> Option<String>; |
| 327 | |
| 328 | /// Quick availability check — true when the tool was found on PATH. |
| 329 | fn available() -> bool { |
| 330 | Self::resolve().is_some() |
| 331 | } |
| 332 | |
| 333 | /// Build a `std::process::Command` pre-populated with the resolved |
| 334 | /// binary (and any fixed arguments from a multi-word candidate like |
| 335 | /// `"py -3"`). Returns `None` when the tool isn't installed. |
| 336 | /// |
| 337 | /// Callers should chain `.args(...)`, `.current_dir(...)`, and then |
| 338 | /// call `.output()`, `.status()`, or `.spawn()`. |
| 339 | fn command() -> Option<Command> { |
| 340 | let spec = Self::resolve()?; |
| 341 | let (program, fixed_args) = split_interpreter_spec(&spec); |
| 342 | let mut cmd = Command::new(&program); |
| 343 | crate::utils::suppress_console_window(&mut cmd); |
| 344 | for arg in &fixed_args { |
| 345 | cmd.arg(arg); |
| 346 | } |
| 347 | Some(cmd) |
| 348 | } |
| 349 | |
| 350 | /// Convenience: run the tool with arguments in a working directory |
| 351 | /// and return the captured output. |
| 352 | fn output(args: &[&str], cwd: &std::path::Path) -> std::io::Result<std::process::Output> { |
| 353 | let mut cmd = Self::command().ok_or_else(|| { |
| 354 | std::io::Error::new( |
| 355 | std::io::ErrorKind::NotFound, |
| 356 | format!("{} not found on PATH", std::any::type_name::<Self>()), |
| 357 | ) |
| 358 | })?; |
| 359 | cmd.args(args).current_dir(cwd).output() |
| 360 | } |
| 361 | |
| 362 | /// Convenience: run the tool with arguments and return only the |
| 363 | /// exit status (discards stdout/stderr). |
| 364 | #[cfg_attr(not(test), expect(dead_code))] |
| 365 | fn status(args: &[&str], cwd: &std::path::Path) -> std::io::Result<std::process::ExitStatus> { |
| 366 | let mut cmd = Self::command().ok_or_else(|| { |
| 367 | std::io::Error::new( |
| 368 | std::io::ErrorKind::NotFound, |
| 369 | format!("{} not found on PATH", std::any::type_name::<Self>()), |
| 370 | ) |
| 371 | })?; |
| 372 | cmd.args(args).current_dir(cwd).status() |
| 373 | } |
| 374 | |
| 375 | /// Build a `tokio::process::Command` pre-populated with the resolved |
| 376 | /// binary (and any fixed arguments from a multi-word candidate like |
| 377 | /// `"py -3"`). Returns `None` when the tool isn't installed. |
| 378 | /// |
| 379 | /// Async callers (`code_execution`, `js_execution`) use this instead |
| 380 | /// of [`ExternalTool::command`] so they can `.await` the child. |
| 381 | fn tokio_command() -> Option<tokio::process::Command> { |
| 382 | let spec = Self::resolve()?; |
| 383 | let (program, fixed_args) = split_interpreter_spec(&spec); |
| 384 | let mut cmd = tokio::process::Command::new(&program); |
| 385 | crate::utils::suppress_tokio_console_window(&mut cmd); |
| 386 | for arg in &fixed_args { |
| 387 | cmd.arg(arg); |
| 388 | } |
| 389 | Some(cmd) |
| 390 | } |
| 391 | } |
| 392 | |
| 393 | // --------------------------------------------------------------------------- |
| 394 | // Concrete tool implementations |
| 395 | // --------------------------------------------------------------------------- |
| 396 | |
| 397 | /// Git version control. |
| 398 | pub struct Git; |
| 399 | |
| 400 | impl Git { |
| 401 | /// Construct a read-only review command with content conversion disabled. |
| 402 | /// Review callers also pass `--no-ext-diff` and `--no-textconv` for diffs. |
| 403 | /// Configured filters otherwise execute even when those flags are present. |
| 404 | pub(crate) fn review_command(workspace: &Path) -> anyhow::Result<Command> { |
| 405 | use anyhow::{Context, bail}; |
| 406 | |
| 407 | let base = || -> anyhow::Result<Command> { |
| 408 | let mut command = Self::command().context("git not found on PATH")?; |
| 409 | command |
| 410 | .current_dir(workspace) |
| 411 | .stdin(std::process::Stdio::null()) |
| 412 | // GIT_CONFIG redirects only `git config`, not `git diff`. |
| 413 | // Both phases must observe the same effective repository config. |
| 414 | .env_remove("GIT_CONFIG") |
| 415 | .env_remove("GIT_CONFIG_PARAMETERS") |
| 416 | .env("GIT_CONFIG_COUNT", "2") |
| 417 | .env("GIT_CONFIG_KEY_0", "core.fsmonitor") |
| 418 | .env("GIT_CONFIG_VALUE_0", "false") |
| 419 | .env("GIT_CONFIG_KEY_1", "core.hooksPath") |
| 420 | .env( |
| 421 | "GIT_CONFIG_VALUE_1", |
| 422 | if cfg!(windows) { "NUL" } else { "/dev/null" }, |
| 423 | ) |
| 424 | .env("GIT_NO_LAZY_FETCH", "1") |
| 425 | .env("GIT_NO_REPLACE_OBJECTS", "1") |
| 426 | .env("GIT_TERMINAL_PROMPT", "0") |
| 427 | .env("GIT_PAGER", ""); |
| 428 | Ok(command) |
| 429 | }; |
| 430 | let output = base()? |
| 431 | .args([ |
| 432 | "config", |
| 433 | "--null", |
| 434 | "--name-only", |
| 435 | "--get-regexp", |
| 436 | r"^filter\..*\.(clean|process|required)$", |
| 437 | ]) |
| 438 | .output() |
| 439 | .context("Failed to inspect Git review filters")?; |
| 440 | let no_filters = |
| 441 | output.status.code() == Some(1) && output.stdout.is_empty() && output.stderr.is_empty(); |
| 442 | if (!output.status.success() && !no_filters) |
| 443 | || (!output.stdout.is_empty() && !output.stdout.ends_with(&[0])) |
| 444 | { |
| 445 | bail!("Cannot safely inspect Git review configuration"); |
| 446 | } |
| 447 | let mut filters = std::collections::BTreeSet::new(); |
| 448 | for key in output |
| 449 | .stdout |
| 450 | .split(|byte| *byte == 0) |
| 451 | .filter(|key| !key.is_empty()) |
| 452 | { |
| 453 | let key = |
| 454 | std::str::from_utf8(key).context("Git review filter name is not valid UTF-8")?; |
| 455 | let (driver, _) = key |
| 456 | .rsplit_once('.') |
| 457 | .context("Invalid Git review filter key")?; |
| 458 | filters.insert(driver); |
| 459 | } |
| 460 | let mut command = base()?; |
| 461 | let mut count = 2; |
| 462 | for driver in filters { |
| 463 | for (suffix, value) in [("clean", ""), ("process", ""), ("required", "false")] { |
| 464 | // A subsection may contain '='; `-c key=value` would then |
| 465 | // override a different key. Separate env fields preserve it. |
| 466 | command.env( |
| 467 | format!("GIT_CONFIG_KEY_{count}"), |
| 468 | format!("{driver}.{suffix}"), |
| 469 | ); |
| 470 | command.env(format!("GIT_CONFIG_VALUE_{count}"), value); |
| 471 | count += 1; |
| 472 | } |
| 473 | } |
| 474 | command.env("GIT_CONFIG_COUNT", count.to_string()); |
| 475 | Ok(command) |
| 476 | } |
| 477 | } |
| 478 | |
| 479 | impl ExternalTool for Git { |
| 480 | fn candidates() -> &'static [&'static str] { |
| 481 | &["git"] |
| 482 | } |
| 483 | |
| 484 | /// Every `Git` invocation in the product is issued against a repository |
| 485 | /// the user also works in by hand. `git status` and `git diff` |
| 486 | /// opportunistically refresh the index, and that refresh takes |
| 487 | /// `.git/index.lock` — which is why a user's own `git commit` could fail |
| 488 | /// with "Unable to create '.../.git/index.lock': File exists" while |
| 489 | /// codewhale was merely idling in the same repo (#5617, reported by |
| 490 | /// @LmeSzinc). |
| 491 | /// |
| 492 | /// `GIT_OPTIONAL_LOCKS=0` suppresses only *optional* lock-taking, so |
| 493 | /// reads stop touching the index while genuine writes (`add`, `commit`, |
| 494 | /// `stash`, `update-ref`) are unaffected — including the snapshot |
| 495 | /// side-repo runner, which writes to its own git dir. `git diff --quiet` |
| 496 | /// exit-code semantics are preserved, which `snapshot::repo` relies on |
| 497 | /// for `/undo` cursoring. |
| 498 | /// |
| 499 | /// Set here rather than on the `ExternalTool::command` default so it does |
| 500 | /// not leak onto `Gh`, `Cargo`, `Node`, `Python`, or `RustC`. Prefer the |
| 501 | /// environment variable over the `--no-optional-locks` flag: the flag is |
| 502 | /// top-level (it must precede the subcommand, awkward for the several |
| 503 | /// call sites that build argument vectors), it would change the |
| 504 | /// agent-visible command string rendered by `tools::git::format_command`, |
| 505 | /// and an unknown flag hard-fails on old git while an unknown environment |
| 506 | /// variable is silently ignored. |
| 507 | fn command() -> Option<Command> { |
| 508 | let spec = Self::resolve()?; |
| 509 | let (program, fixed_args) = split_interpreter_spec(&spec); |
| 510 | let mut cmd = Command::new(&program); |
| 511 | crate::utils::suppress_console_window(&mut cmd); |
| 512 | for arg in &fixed_args { |
| 513 | cmd.arg(arg); |
| 514 | } |
| 515 | cmd.env("GIT_OPTIONAL_LOCKS", "0"); |
| 516 | Some(cmd) |
| 517 | } |
| 518 | |
| 519 | fn resolve() -> Option<String> { |
| 520 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 521 | CACHE |
| 522 | .get_or_init(|| { |
| 523 | for candidate in Self::candidates() { |
| 524 | if probe_executable(candidate) { |
| 525 | tracing::info!(target: "tool_dependencies", "Resolved git binary"); |
| 526 | return Some((*candidate).to_string()); |
| 527 | } |
| 528 | } |
| 529 | None |
| 530 | }) |
| 531 | .clone() |
| 532 | } |
| 533 | } |
| 534 | |
| 535 | /// GitHub CLI. |
| 536 | pub struct Gh; |
| 537 | |
| 538 | impl ExternalTool for Gh { |
| 539 | fn candidates() -> &'static [&'static str] { |
| 540 | &["gh"] |
| 541 | } |
| 542 | |
| 543 | fn resolve() -> Option<String> { |
| 544 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 545 | CACHE |
| 546 | .get_or_init(|| { |
| 547 | for candidate in Self::candidates() { |
| 548 | if probe_executable(candidate) { |
| 549 | tracing::info!(target: "tool_dependencies", "Resolved gh binary"); |
| 550 | return Some((*candidate).to_string()); |
| 551 | } |
| 552 | } |
| 553 | None |
| 554 | }) |
| 555 | .clone() |
| 556 | } |
| 557 | } |
| 558 | |
| 559 | /// Rust compiler — used for version reporting in diagnostics. |
| 560 | pub struct RustC; |
| 561 | |
| 562 | impl ExternalTool for RustC { |
| 563 | fn candidates() -> &'static [&'static str] { |
| 564 | &["rustc"] |
| 565 | } |
| 566 | |
| 567 | fn resolve() -> Option<String> { |
| 568 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 569 | CACHE |
| 570 | .get_or_init(|| { |
| 571 | // Probe with capture so the `--version` banner observed during |
| 572 | // resolution is reused by [`rustc_version_banner`] instead of |
| 573 | // paying a second rustc process launch (each launch loads |
| 574 | // libLLVM, which dominated diagnostic-command init profiles). |
| 575 | for candidate in Self::candidates() { |
| 576 | if let Some(banner) = probe_executable_capturing(candidate, "--version") { |
| 577 | tracing::info!(target: "tool_dependencies", "Resolved rustc binary"); |
| 578 | let _ = RUSTC_VERSION_BANNER.set(Some(banner)); |
| 579 | return Some((*candidate).to_string()); |
| 580 | } |
| 581 | } |
| 582 | None |
| 583 | }) |
| 584 | .clone() |
| 585 | } |
| 586 | } |
| 587 | |
| 588 | /// Captured `--version` banner from the [`RustC`] resolution probe. |
| 589 | /// |
| 590 | /// `None` until `RustC::resolve()`/`available()`/`command()` first runs, or |
| 591 | /// when rustc is absent/failing. Reading this after an `available()` check |
| 592 | /// yields the same string the tool would print, without a second process. |
| 593 | static RUSTC_VERSION_BANNER: OnceLock<Option<String>> = OnceLock::new(); |
| 594 | |
| 595 | /// The rustc `--version` banner, if rustc resolved successfully. |
| 596 | /// |
| 597 | /// Populated as a side effect of resolving [`RustC`]; this reads no fresh |
| 598 | /// process state. Callers wanting the value should touch `RustC::available()` |
| 599 | /// first (as the diagnostics path does). |
| 600 | #[must_use] |
| 601 | pub fn rustc_version_banner() -> Option<String> { |
| 602 | RUSTC_VERSION_BANNER.get().cloned().flatten() |
| 603 | } |
| 604 | |
| 605 | /// Rust build tool — used by the `run_tests` tool. |
| 606 | pub struct Cargo; |
| 607 | |
| 608 | impl ExternalTool for Cargo { |
| 609 | fn candidates() -> &'static [&'static str] { |
| 610 | &["cargo"] |
| 611 | } |
| 612 | |
| 613 | fn resolve() -> Option<String> { |
| 614 | static CACHE: OnceLock<Option<String>> = OnceLock::new(); |
| 615 | CACHE |
| 616 | .get_or_init(|| { |
| 617 | for candidate in Self::candidates() { |
| 618 | if probe_executable(candidate) { |
| 619 | tracing::info!(target: "tool_dependencies", "Resolved cargo binary"); |
| 620 | return Some((*candidate).to_string()); |
| 621 | } |
| 622 | } |
| 623 | None |
| 624 | }) |
| 625 | .clone() |
| 626 | } |
| 627 | } |
| 628 | |
| 629 | /// Python interpreter — used by `code_execution` tool and RLM REPL. |
| 630 | /// Delegates to the existing [`resolve_python_interpreter`] so the |
| 631 | /// multi-candidate ladder (`python3` → `python` → `py -3`) is |
| 632 | /// shared with legacy callers until they migrate to the trait. |
| 633 | pub struct Python; |
| 634 | |
| 635 | impl ExternalTool for Python { |
| 636 | fn candidates() -> &'static [&'static str] { |
| 637 | PYTHON_CANDIDATES |
| 638 | } |
| 639 | |
| 640 | fn resolve() -> Option<String> { |
| 641 | resolve_python_interpreter() |
| 642 | } |
| 643 | } |
| 644 | |
| 645 | /// Node.js runtime — used by the `js_execution` tool. |
| 646 | /// The binary name `node` is the same on every platform we support, |
| 647 | /// so this is a single probe rather than a candidate ladder. |
| 648 | pub struct Node; |
| 649 | |
| 650 | impl ExternalTool for Node { |
| 651 | fn candidates() -> &'static [&'static str] { |
| 652 | &["node"] |
| 653 | } |
| 654 | |
| 655 | fn resolve() -> Option<String> { |
| 656 | resolve_node() |
| 657 | } |
| 658 | } |
| 659 | |
| 660 | // --------------------------------------------------------------------------- |
| 661 | // Legacy interpreter helpers (kept for existing callers until migrated) |
| 662 | // --------------------------------------------------------------------------- |
| 663 | |
| 664 | /// Split an interpreter spec like `"py -3"` into the program name |
| 665 | /// and any initial arguments. Returns `("py", vec!["-3"])` for the |
| 666 | /// example; returns `("python3", vec![])` for a bare name. |
| 667 | /// |
| 668 | /// Callers spawn `Command::new(program).args(args).arg(script_path)`. |
| 669 | #[must_use] |
| 670 | pub fn split_interpreter_spec(spec: &str) -> (String, Vec<String>) { |
| 671 | let mut parts = spec.split_whitespace(); |
| 672 | let program = parts.next().unwrap_or("").to_string(); |
| 673 | let args = parts.map(str::to_string).collect(); |
| 674 | (program, args) |
| 675 | } |
| 676 | |
| 677 | #[cfg(test)] |
| 678 | mod tests { |
| 679 | use super::*; |
| 680 | |
| 681 | #[test] |
| 682 | fn probe_executable_returns_false_for_unknown_binary() { |
| 683 | // Pick a name we're confident isn't on any developer's PATH. |
| 684 | // If this ever starts failing locally, rename it. |
| 685 | assert!(!probe_executable("codewhale-tui-imaginary-binary-xyz123")); |
| 686 | } |
| 687 | |
| 688 | #[test] |
| 689 | fn probe_executable_handles_multi_word_specs() { |
| 690 | // `py -3` should split correctly. The probe will fail on |
| 691 | // most non-Windows machines (no `py` launcher), which is |
| 692 | // fine — we're checking that the *split* doesn't crash. |
| 693 | let _ = probe_executable("py -3"); |
| 694 | } |
| 695 | |
| 696 | #[test] |
| 697 | fn probe_executable_with_flag_returns_false_for_unknown_binary() { |
| 698 | assert!(!probe_executable_with_flag( |
| 699 | "codewhale-tui-imaginary-binary-xyz123", |
| 700 | "-v" |
| 701 | )); |
| 702 | } |
| 703 | |
| 704 | #[test] |
| 705 | fn probe_executable_delegates_to_double_dash_version() { |
| 706 | // `probe_executable` must remain exactly |
| 707 | // `probe_executable_with_flag(.., "--version")`. |
| 708 | let spec = "codewhale-tui-imaginary-binary-xyz123"; |
| 709 | assert_eq!( |
| 710 | probe_executable(spec), |
| 711 | probe_executable_with_flag(spec, "--version") |
| 712 | ); |
| 713 | } |
| 714 | |
| 715 | #[test] |
| 716 | fn pdftotext_resolver_detects_installed_poppler_via_dash_v() { |
| 717 | // Regression for #1667: Poppler's `pdftotext` rejects `--version` |
| 718 | // (it is parsed as an input filename and exits non-zero), so the |
| 719 | // generic `--version` probe reports it missing even when installed. |
| 720 | // The resolver must probe with `-v`. Gated on pdftotext actually |
| 721 | // being installed so CI without Poppler stays green. |
| 722 | if probe_executable_with_flag("pdftotext", "-v") { |
| 723 | assert!( |
| 724 | resolve_pdftotext().is_some(), |
| 725 | "an installed pdftotext must be detected via -v (#1667)" |
| 726 | ); |
| 727 | } |
| 728 | } |
| 729 | |
| 730 | #[test] |
| 731 | fn split_interpreter_spec_strips_args() { |
| 732 | assert_eq!( |
| 733 | split_interpreter_spec("python3"), |
| 734 | ("python3".to_string(), Vec::<String>::new()) |
| 735 | ); |
| 736 | assert_eq!( |
| 737 | split_interpreter_spec("py -3"), |
| 738 | ("py".to_string(), vec!["-3".to_string()]) |
| 739 | ); |
| 740 | assert_eq!( |
| 741 | split_interpreter_spec(" python3 "), |
| 742 | ("python3".to_string(), Vec::<String>::new()), |
| 743 | "leading/trailing whitespace must be tolerated" |
| 744 | ); |
| 745 | } |
| 746 | |
| 747 | #[test] |
| 748 | fn split_interpreter_spec_handles_empty_string() { |
| 749 | assert_eq!( |
| 750 | split_interpreter_spec(""), |
| 751 | (String::new(), Vec::<String>::new()) |
| 752 | ); |
| 753 | } |
| 754 | |
| 755 | #[test] |
| 756 | fn python_resolver_is_cached_across_calls() { |
| 757 | // Whatever the first call returns, subsequent calls return |
| 758 | // the same value (cached). If this test ever flakes, the |
| 759 | // OnceLock semantics changed and we need to rethink the |
| 760 | // resolver. |
| 761 | let first = resolve_python_interpreter(); |
| 762 | let second = resolve_python_interpreter(); |
| 763 | assert_eq!(first, second); |
| 764 | } |
| 765 | |
| 766 | #[test] |
| 767 | fn python_resolver_returns_some_on_developer_machines() { |
| 768 | // CI hosts have Python; developer machines have Python. |
| 769 | // The one environment where this returns None is bare-bones |
| 770 | // Windows / minimal CI containers — fine, those just don't |
| 771 | // get code_execution registered, which is the whole point. |
| 772 | // We don't assert Some() because we don't want this test |
| 773 | // to fail in those environments. Instead we just confirm |
| 774 | // the resolver doesn't panic and returns a stable value. |
| 775 | let resolved = resolve_python_interpreter(); |
| 776 | if let Some(name) = resolved { |
| 777 | assert!( |
| 778 | !name.is_empty(), |
| 779 | "resolved interpreter name must be non-empty" |
| 780 | ); |
| 781 | // The resolved name must be one of our candidates. |
| 782 | assert!( |
| 783 | PYTHON_CANDIDATES.contains(&name.as_str()), |
| 784 | "resolved {name:?} is not in PYTHON_CANDIDATES {PYTHON_CANDIDATES:?}" |
| 785 | ); |
| 786 | } |
| 787 | } |
| 788 | |
| 789 | // =================================================================== |
| 790 | // ExternalTool trait tests |
| 791 | // =================================================================== |
| 792 | |
| 793 | #[test] |
| 794 | fn python_candidates_matches_const() { |
| 795 | assert_eq!(Python::candidates(), PYTHON_CANDIDATES); |
| 796 | } |
| 797 | |
| 798 | #[test] |
| 799 | fn node_candidates_is_node_only() { |
| 800 | assert_eq!(Node::candidates(), &["node"]); |
| 801 | } |
| 802 | |
| 803 | #[test] |
| 804 | fn git_candidates_is_git_only() { |
| 805 | assert_eq!(Git::candidates(), &["git"]); |
| 806 | } |
| 807 | |
| 808 | #[test] |
| 809 | fn gh_candidates_is_gh_only() { |
| 810 | assert_eq!(Gh::candidates(), &["gh"]); |
| 811 | } |
| 812 | |
| 813 | #[test] |
| 814 | fn rustc_candidates_is_rustc_only() { |
| 815 | assert_eq!(RustC::candidates(), &["rustc"]); |
| 816 | } |
| 817 | |
| 818 | #[test] |
| 819 | fn cargo_candidates_is_cargo_only() { |
| 820 | assert_eq!(Cargo::candidates(), &["cargo"]); |
| 821 | } |
| 822 | |
| 823 | #[test] |
| 824 | fn concrete_resolvers_do_not_cross_contaminate_when_available() { |
| 825 | let values = [ |
| 826 | Git::resolve().map(|v| ("git", v)), |
| 827 | Gh::resolve().map(|v| ("gh", v)), |
| 828 | RustC::resolve().map(|v| ("rustc", v)), |
| 829 | Cargo::resolve().map(|v| ("cargo", v)), |
| 830 | Node::resolve().map(|v| ("node", v)), |
| 831 | ]; |
| 832 | let resolved: Vec<(&str, String)> = values.into_iter().flatten().collect(); |
| 833 | |
| 834 | for i in 0..resolved.len() { |
| 835 | for j in (i + 1)..resolved.len() { |
| 836 | assert_ne!( |
| 837 | resolved[i].1, resolved[j].1, |
| 838 | "{} and {} unexpectedly resolved to the same binary", |
| 839 | resolved[i].0, resolved[j].0 |
| 840 | ); |
| 841 | } |
| 842 | } |
| 843 | } |
| 844 | |
| 845 | #[test] |
| 846 | fn git_resolve_is_cached() { |
| 847 | let first = Git::resolve(); |
| 848 | let second = Git::resolve(); |
| 849 | assert_eq!(first, second); |
| 850 | } |
| 851 | |
| 852 | #[test] |
| 853 | fn gh_resolve_is_cached() { |
| 854 | let first = Gh::resolve(); |
| 855 | let second = Gh::resolve(); |
| 856 | assert_eq!(first, second); |
| 857 | } |
| 858 | |
| 859 | #[test] |
| 860 | fn python_trait_resolve_is_cached() { |
| 861 | let first = Python::resolve(); |
| 862 | let second = Python::resolve(); |
| 863 | assert_eq!(first, second); |
| 864 | } |
| 865 | |
| 866 | #[test] |
| 867 | fn node_resolve_is_cached() { |
| 868 | let first = Node::resolve(); |
| 869 | let second = Node::resolve(); |
| 870 | assert_eq!(first, second); |
| 871 | } |
| 872 | |
| 873 | #[test] |
| 874 | fn rustc_resolve_is_cached() { |
| 875 | let first = RustC::resolve(); |
| 876 | let second = RustC::resolve(); |
| 877 | assert_eq!(first, second); |
| 878 | } |
| 879 | |
| 880 | #[test] |
| 881 | fn cargo_resolve_is_cached() { |
| 882 | let first = Cargo::resolve(); |
| 883 | let second = Cargo::resolve(); |
| 884 | assert_eq!(first, second); |
| 885 | } |
| 886 | |
| 887 | #[test] |
| 888 | fn git_available_matches_resolve() { |
| 889 | assert_eq!(Git::available(), Git::resolve().is_some()); |
| 890 | } |
| 891 | |
| 892 | #[test] |
| 893 | fn python_available_matches_resolve() { |
| 894 | assert_eq!(Python::available(), Python::resolve().is_some()); |
| 895 | } |
| 896 | |
| 897 | #[test] |
| 898 | fn node_available_matches_resolve() { |
| 899 | assert_eq!(Node::available(), Node::resolve().is_some()); |
| 900 | } |
| 901 | |
| 902 | #[test] |
| 903 | fn rustc_available_matches_resolve() { |
| 904 | assert_eq!(RustC::available(), RustC::resolve().is_some()); |
| 905 | } |
| 906 | |
| 907 | #[test] |
| 908 | fn cargo_available_matches_resolve() { |
| 909 | assert_eq!(Cargo::available(), Cargo::resolve().is_some()); |
| 910 | } |
| 911 | |
| 912 | #[test] |
| 913 | fn git_command_returns_some_when_available() { |
| 914 | if Git::available() { |
| 915 | assert!(Git::command().is_some()); |
| 916 | } |
| 917 | } |
| 918 | |
| 919 | /// Every git command we build must be lock-free (#5617). Without this, |
| 920 | /// a read-only probe can take `.git/index.lock` in the user's own |
| 921 | /// repository and break a `git commit` they run by hand. |
| 922 | #[test] |
| 923 | fn git_command_never_takes_optional_locks() { |
| 924 | if !Git::available() { |
| 925 | return; |
| 926 | } |
| 927 | let cmd = Git::command().expect("git resolves when available"); |
| 928 | let value = cmd |
| 929 | .get_envs() |
| 930 | .find(|(key, _)| *key == std::ffi::OsStr::new("GIT_OPTIONAL_LOCKS")) |
| 931 | .and_then(|(_, value)| value) |
| 932 | .expect("GIT_OPTIONAL_LOCKS must be set on every git command"); |
| 933 | assert_eq!(value, std::ffi::OsStr::new("0")); |
| 934 | } |
| 935 | |
| 936 | /// The suppression is deliberately scoped to git. Other external tools |
| 937 | /// have no index to protect and must not inherit a git-specific variable. |
| 938 | #[test] |
| 939 | fn optional_lock_suppression_does_not_leak_to_other_tools() { |
| 940 | for cmd in [Gh::command(), Cargo::command(), Node::command()] |
| 941 | .into_iter() |
| 942 | .flatten() |
| 943 | { |
| 944 | assert!( |
| 945 | !cmd.get_envs() |
| 946 | .any(|(key, _)| key == std::ffi::OsStr::new("GIT_OPTIONAL_LOCKS")), |
| 947 | "only Git may set GIT_OPTIONAL_LOCKS" |
| 948 | ); |
| 949 | } |
| 950 | } |
| 951 | |
| 952 | #[test] |
| 953 | fn python_command_returns_some_when_available() { |
| 954 | if Python::available() { |
| 955 | assert!(Python::command().is_some()); |
| 956 | } |
| 957 | } |
| 958 | |
| 959 | #[test] |
| 960 | fn python_tokio_command_returns_some_when_available() { |
| 961 | if Python::available() { |
| 962 | assert!(Python::tokio_command().is_some()); |
| 963 | } |
| 964 | } |
| 965 | |
| 966 | #[test] |
| 967 | fn node_tokio_command_returns_some_when_available() { |
| 968 | if Node::available() { |
| 969 | assert!(Node::tokio_command().is_some()); |
| 970 | } |
| 971 | } |
| 972 | |
| 973 | #[test] |
| 974 | fn git_output_version_succeeds() { |
| 975 | // Only run when git is actually installed. |
| 976 | if !Git::available() { |
| 977 | return; |
| 978 | } |
| 979 | let tmp = std::env::temp_dir(); |
| 980 | let out = Git::output(&["--version"], &tmp); |
| 981 | assert!( |
| 982 | out.is_ok(), |
| 983 | "git --version must succeed when git is available" |
| 984 | ); |
| 985 | let out = out.unwrap(); |
| 986 | assert!(out.status.success(), "git --version must exit 0"); |
| 987 | let stdout = String::from_utf8_lossy(&out.stdout); |
| 988 | assert!( |
| 989 | stdout.contains("git version"), |
| 990 | "git --version stdout must contain 'git version', got: {}", |
| 991 | stdout.trim() |
| 992 | ); |
| 993 | } |
| 994 | |
| 995 | #[test] |
| 996 | fn python_output_version_succeeds() { |
| 997 | if !Python::available() { |
| 998 | return; |
| 999 | } |
| 1000 | let tmp = std::env::temp_dir(); |
| 1001 | let out = Python::output(&["--version"], &tmp); |
| 1002 | assert!(out.is_ok(), "python --version must spawn"); |
| 1003 | let out = out.unwrap(); |
| 1004 | // Python --version writes to stdout on 3.x, so just check |
| 1005 | // that it succeeded (exit 0). |
| 1006 | assert!(out.status.success(), "python --version must exit 0"); |
| 1007 | } |
| 1008 | |
| 1009 | #[test] |
| 1010 | fn node_output_version_succeeds() { |
| 1011 | if !Node::available() { |
| 1012 | return; |
| 1013 | } |
| 1014 | let tmp = std::env::temp_dir(); |
| 1015 | let out = Node::output(&["--version"], &tmp); |
| 1016 | assert!(out.is_ok(), "node --version must spawn"); |
| 1017 | let out = out.unwrap(); |
| 1018 | assert!(out.status.success(), "node --version must exit 0"); |
| 1019 | } |
| 1020 | |
| 1021 | #[test] |
| 1022 | fn cargo_output_version_succeeds() { |
| 1023 | if !Cargo::available() { |
| 1024 | return; |
| 1025 | } |
| 1026 | let tmp = std::env::temp_dir(); |
| 1027 | let out = Cargo::output(&["--version"], &tmp); |
| 1028 | assert!(out.is_ok(), "cargo --version must spawn"); |
| 1029 | let out = out.unwrap(); |
| 1030 | assert!(out.status.success(), "cargo --version must exit 0"); |
| 1031 | } |
| 1032 | |
| 1033 | #[test] |
| 1034 | fn external_tool_output_respects_cwd() { |
| 1035 | // Verify that `output()` runs in the requested directory. |
| 1036 | if !Git::available() { |
| 1037 | return; |
| 1038 | } |
| 1039 | let tmp = std::env::temp_dir(); |
| 1040 | let out = Git::output(&["rev-parse", "--show-toplevel"], &tmp); |
| 1041 | assert!(out.is_ok(), "git rev-parse must spawn"); |
| 1042 | let out = out.unwrap(); |
| 1043 | // rev-parse --show-toplevel in a non-git dir should fail |
| 1044 | // because temp_dir is not a git repo. That's expected. |
| 1045 | // The key assertion: the command executed without IO errors. |
| 1046 | // We don't assert success because temp_dir might or might not |
| 1047 | // be inside a git worktree. |
| 1048 | let _ = out; // just checking it didn't panic/IO-error |
| 1049 | } |
| 1050 | } |
| 1051 |