| 1 | //! Expand a shell command line into the set of commands a shell would run. |
| 2 | //! |
| 3 | //! Deny rules are the one gate that holds under `AskForApproval::Never`, so |
| 4 | //! they cannot be matched against the raw command string: the string a user |
| 5 | //! types and the set of commands the shell executes are different things. A |
| 6 | //! command substitution runs its body (`` `rm -rf /` ``, `$(rm -rf /)`), a |
| 7 | //! quoted argument executes with the quotes removed (`rm -rf "/"`), and a |
| 8 | //! wrapper hands its payload straight back to a shell (`bash -c '…'`, |
| 9 | //! `eval '…'`, `sudo …`). |
| 10 | //! |
| 11 | //! Matching one string pattern per metacharacter loses that race by |
| 12 | //! construction — every new quoting or wrapping form is another bypass. This |
| 13 | //! module instead tokenizes the command the way a POSIX shell word-splits it |
| 14 | //! and returns *every* command line that would actually be executed, so deny |
| 15 | //! rules can be matched against each one. |
| 16 | //! |
| 17 | //! Deliberately conservative in the deny direction: when a construct is |
| 18 | //! ambiguous the expander emits extra candidate command lines rather than |
| 19 | //! fewer. Over-emitting only makes deny matching stricter — `denied_prefix_matches` |
| 20 | //! stays anchored at the first positional token, so an extra candidate that no |
| 21 | //! rule names is inert. Under-emitting is a bypass. |
| 22 | //! |
| 23 | //! What it does *not* do is evaluate anything: `$VAR` is left as literal text, |
| 24 | //! and single-quoted text is never treated as code (`echo '` + "`" + `rm -rf /`" + |
| 25 | //! "`" + `'` really does just print). Fidelity to shell semantics is the point in |
| 26 | //! both directions. |
| 27 | //! |
| 28 | //! Because nothing is evaluated, some command words cannot be known without |
| 29 | //! running the shell: a parameter or command substitution (`$v`, `${v}`, |
| 30 | //! `$(…)`), a glob or brace expansion (`r[m]`, `{rm,-f,x}`), ANSI-C escapes, |
| 31 | //! and a shell that reads its script from a pipe, a here-string, or a process |
| 32 | //! substitution. [`Expansion::dynamic`] reports that case so the policy engine |
| 33 | //! can fail closed instead of matching deny rules against text the shell will |
| 34 | //! rewrite. [`Expansion::nested`] reports that some code runs inside another |
| 35 | //! command (a substitution, `eval`, a `-c` payload, `find -exec`), which is |
| 36 | //! never covered by an allow rule written for the outer command. |
| 37 | //! |
| 38 | //! Known limits: a script *file* (`bash ./x.sh`, `. ./env.sh`) is opaque |
| 39 | //! (a descriptor path such as `/dev/stdin` is not a file: it is dynamic), as |
| 40 | //! is any program that interprets its arguments as code (`python -c`, `ssh |
| 41 | //! host cmd`, `git -c alias.x=!cmd`). Aliases and functions defined in an |
| 42 | //! earlier call are not tracked. Arithmetic contexts (`(( ))`, `$(( ))`, |
| 43 | //! `let`, `[[ … -eq … ]]`) evaluate the *values* of variables they name, so a |
| 44 | //! value holding `a[$(cmd)]` runs `cmd` without it appearing in the text. |
| 45 | //! `cmd /c` and PowerShell `-Command` payloads are scanned with this POSIX |
| 46 | //! grammar, which finds their command words but not every cmd.exe or |
| 47 | //! PowerShell construct. |
| 48 | |
| 49 | use std::collections::HashSet; |
| 50 | |
| 51 | /// Maximum nesting depth followed through substitutions and `-c` payloads. |
| 52 | const MAX_DEPTH: usize = 8; |
| 53 | |
| 54 | /// Upper bound on emitted command lines, so a pathological input cannot turn |
| 55 | /// one policy check into unbounded work. |
| 56 | const MAX_COMMANDS: usize = 256; |
| 57 | |
| 58 | /// Upper bound on the parser states explored while locating the real command |
| 59 | /// word behind wrapper words. Running out marks the command unresolved. |
| 60 | const MAX_HEAD_STATES: usize = 512; |
| 61 | |
| 62 | /// A word that prefixes another command rather than being the command: the |
| 63 | /// real invocation is what follows. Stripping it keeps `sudo rm -rf /` |
| 64 | /// matchable by an `rm -rf /` rule. |
| 65 | /// |
| 66 | /// The option grammar matters: an option that takes a separate value |
| 67 | /// (`sudo -u root`, `timeout -s KILL`) and a required operand |
| 68 | /// (`chroot NEWROOT`, `timeout DURATION`) both sit between the wrapper and the |
| 69 | /// command it runs. An option missing from the table may or may not take a |
| 70 | /// value, so both readings are explored. |
| 71 | struct Wrapper { |
| 72 | name: &'static str, |
| 73 | /// Short options whose value is the rest of the word or the next word. |
| 74 | short_values: &'static str, |
| 75 | /// Short options known to take no value. Any other short option may or |
| 76 | /// may not take one, so both readings are explored. |
| 77 | short_switches: &'static str, |
| 78 | /// Long options whose value may be the next word. |
| 79 | long_values: &'static [&'static str], |
| 80 | /// Long options known to take no value. |
| 81 | long_switches: &'static [&'static str], |
| 82 | /// Required operands before the command (`chroot NEWROOT`). |
| 83 | operands: u8, |
| 84 | } |
| 85 | |
| 86 | static WRAPPERS: &[Wrapper] = &[ |
| 87 | Wrapper { |
| 88 | name: "sudo", |
| 89 | short_values: "ughpCDrtTRUac", |
| 90 | short_switches: "AbBEeHiKklNnPSsVv", |
| 91 | long_values: &[ |
| 92 | "user", |
| 93 | "group", |
| 94 | "host", |
| 95 | "prompt", |
| 96 | "close-from", |
| 97 | "chdir", |
| 98 | "role", |
| 99 | "type", |
| 100 | "command-timeout", |
| 101 | "chroot", |
| 102 | "other-user", |
| 103 | "auth-type", |
| 104 | "login-class", |
| 105 | ], |
| 106 | long_switches: &[ |
| 107 | "login", |
| 108 | "preserve-env", |
| 109 | "non-interactive", |
| 110 | "background", |
| 111 | "edit", |
| 112 | "shell", |
| 113 | "stdin", |
| 114 | "reset-timestamp", |
| 115 | "remove-timestamp", |
| 116 | "validate", |
| 117 | "list", |
| 118 | "set-home", |
| 119 | "bell", |
| 120 | "askpass", |
| 121 | "preserve-groups", |
| 122 | ], |
| 123 | operands: 0, |
| 124 | }, |
| 125 | Wrapper { |
| 126 | name: "doas", |
| 127 | short_values: "uCa", |
| 128 | short_switches: "Lns", |
| 129 | long_values: &[], |
| 130 | long_switches: &[], |
| 131 | operands: 0, |
| 132 | }, |
| 133 | Wrapper { |
| 134 | name: "env", |
| 135 | short_values: "uCSPa", |
| 136 | short_switches: "i0v", |
| 137 | long_values: &["unset", "chdir", "argv0"], |
| 138 | long_switches: &["ignore-environment", "null", "debug"], |
| 139 | operands: 0, |
| 140 | }, |
| 141 | Wrapper { |
| 142 | name: "nohup", |
| 143 | short_values: "", |
| 144 | short_switches: "", |
| 145 | long_values: &[], |
| 146 | long_switches: &[], |
| 147 | operands: 0, |
| 148 | }, |
| 149 | Wrapper { |
| 150 | name: "nice", |
| 151 | short_values: "n", |
| 152 | short_switches: "", |
| 153 | long_values: &["adjustment"], |
| 154 | long_switches: &[], |
| 155 | operands: 0, |
| 156 | }, |
| 157 | Wrapper { |
| 158 | name: "ionice", |
| 159 | short_values: "cnpPu", |
| 160 | short_switches: "t", |
| 161 | long_values: &["class", "classdata", "pid", "pgid", "uid"], |
| 162 | long_switches: &["ignore"], |
| 163 | operands: 0, |
| 164 | }, |
| 165 | Wrapper { |
| 166 | name: "time", |
| 167 | short_values: "fo", |
| 168 | short_switches: "aplqvh", |
| 169 | long_values: &["format", "output"], |
| 170 | long_switches: &["append", "portability", "verbose", "quiet"], |
| 171 | operands: 0, |
| 172 | }, |
| 173 | Wrapper { |
| 174 | name: "timeout", |
| 175 | short_values: "sk", |
| 176 | short_switches: "fpv", |
| 177 | long_values: &["signal", "kill-after"], |
| 178 | long_switches: &["preserve-status", "foreground", "verbose"], |
| 179 | operands: 1, |
| 180 | }, |
| 181 | Wrapper { |
| 182 | name: "stdbuf", |
| 183 | short_values: "ioe", |
| 184 | short_switches: "", |
| 185 | long_values: &["input", "output", "error"], |
| 186 | long_switches: &[], |
| 187 | operands: 0, |
| 188 | }, |
| 189 | Wrapper { |
| 190 | name: "setsid", |
| 191 | short_values: "", |
| 192 | short_switches: "cfw", |
| 193 | long_values: &[], |
| 194 | long_switches: &["ctty", "fork", "wait"], |
| 195 | operands: 0, |
| 196 | }, |
| 197 | Wrapper { |
| 198 | name: "command", |
| 199 | short_values: "", |
| 200 | short_switches: "pvV", |
| 201 | long_values: &[], |
| 202 | long_switches: &[], |
| 203 | operands: 0, |
| 204 | }, |
| 205 | Wrapper { |
| 206 | name: "builtin", |
| 207 | short_values: "", |
| 208 | short_switches: "", |
| 209 | long_values: &[], |
| 210 | long_switches: &[], |
| 211 | operands: 0, |
| 212 | }, |
| 213 | Wrapper { |
| 214 | name: "exec", |
| 215 | short_values: "a", |
| 216 | short_switches: "cl", |
| 217 | long_values: &[], |
| 218 | long_switches: &[], |
| 219 | operands: 0, |
| 220 | }, |
| 221 | Wrapper { |
| 222 | name: "xargs", |
| 223 | short_values: "adEILnPsJSR", |
| 224 | short_switches: "0eiloprtx", |
| 225 | long_values: &[ |
| 226 | "arg-file", |
| 227 | "delimiter", |
| 228 | "eof", |
| 229 | "replace", |
| 230 | "max-lines", |
| 231 | "max-args", |
| 232 | "max-procs", |
| 233 | "max-chars", |
| 234 | "process-slot-var", |
| 235 | ], |
| 236 | long_switches: &[ |
| 237 | "null", |
| 238 | "no-run-if-empty", |
| 239 | "interactive", |
| 240 | "verbose", |
| 241 | "exit", |
| 242 | "open-tty", |
| 243 | "show-limits", |
| 244 | ], |
| 245 | operands: 0, |
| 246 | }, |
| 247 | Wrapper { |
| 248 | name: "unbuffer", |
| 249 | short_values: "", |
| 250 | short_switches: "p", |
| 251 | long_values: &[], |
| 252 | long_switches: &[], |
| 253 | operands: 0, |
| 254 | }, |
| 255 | Wrapper { |
| 256 | name: "busybox", |
| 257 | short_values: "", |
| 258 | short_switches: "", |
| 259 | long_values: &[], |
| 260 | long_switches: &[], |
| 261 | operands: 0, |
| 262 | }, |
| 263 | Wrapper { |
| 264 | name: "chroot", |
| 265 | short_values: "ugG", |
| 266 | short_switches: "", |
| 267 | long_values: &["userspec", "groups"], |
| 268 | long_switches: &["skip-chdir"], |
| 269 | operands: 1, |
| 270 | }, |
| 271 | Wrapper { |
| 272 | name: "proot", |
| 273 | short_values: "rbmwqkRSvi", |
| 274 | short_switches: "0Vh", |
| 275 | long_values: &[ |
| 276 | "rootfs", |
| 277 | "bind", |
| 278 | "mount", |
| 279 | "cwd", |
| 280 | "pwd", |
| 281 | "qemu", |
| 282 | "kernel-release", |
| 283 | "verbose", |
| 284 | ], |
| 285 | long_switches: &[], |
| 286 | operands: 0, |
| 287 | }, |
| 288 | Wrapper { |
| 289 | name: "caffeinate", |
| 290 | short_values: "tw", |
| 291 | short_switches: "dimsu", |
| 292 | long_values: &[], |
| 293 | long_switches: &[], |
| 294 | operands: 0, |
| 295 | }, |
| 296 | Wrapper { |
| 297 | name: "arch", |
| 298 | short_values: "ed", |
| 299 | short_switches: "", |
| 300 | long_values: &[], |
| 301 | long_switches: &[], |
| 302 | operands: 0, |
| 303 | }, |
| 304 | Wrapper { |
| 305 | name: "sandbox-exec", |
| 306 | short_values: "fnpD", |
| 307 | short_switches: "", |
| 308 | long_values: &[], |
| 309 | long_switches: &[], |
| 310 | operands: 0, |
| 311 | }, |
| 312 | Wrapper { |
| 313 | name: "noglob", |
| 314 | short_values: "", |
| 315 | short_switches: "", |
| 316 | long_values: &[], |
| 317 | long_switches: &[], |
| 318 | operands: 0, |
| 319 | }, |
| 320 | Wrapper { |
| 321 | name: "nocorrect", |
| 322 | short_values: "", |
| 323 | short_switches: "", |
| 324 | long_values: &[], |
| 325 | long_switches: &[], |
| 326 | operands: 0, |
| 327 | }, |
| 328 | Wrapper { |
| 329 | name: "-", |
| 330 | short_values: "", |
| 331 | short_switches: "", |
| 332 | long_values: &[], |
| 333 | long_switches: &[], |
| 334 | operands: 0, |
| 335 | }, |
| 336 | Wrapper { |
| 337 | name: "nsenter", |
| 338 | short_values: "tSG", |
| 339 | short_switches: "amuinpUCTrwFZcy", |
| 340 | long_values: &["target", "setuid", "setgid"], |
| 341 | long_switches: &["all", "preserve-credentials", "no-fork"], |
| 342 | operands: 0, |
| 343 | }, |
| 344 | Wrapper { |
| 345 | name: "unshare", |
| 346 | short_values: "SGRw", |
| 347 | short_switches: "mupinUCTrcfk", |
| 348 | long_values: &["root", "wd", "setuid", "setgid", "propagation", "setgroups"], |
| 349 | long_switches: &[ |
| 350 | "fork", |
| 351 | "mount-proc", |
| 352 | "map-root-user", |
| 353 | "map-current-user", |
| 354 | "kill-child", |
| 355 | ], |
| 356 | operands: 0, |
| 357 | }, |
| 358 | Wrapper { |
| 359 | name: "runuser", |
| 360 | short_values: "ugGcswf", |
| 361 | short_switches: "lmpP", |
| 362 | long_values: &[ |
| 363 | "user", |
| 364 | "group", |
| 365 | "supp-group", |
| 366 | "command", |
| 367 | "shell", |
| 368 | "whitelist-environment", |
| 369 | "session-command", |
| 370 | ], |
| 371 | long_switches: &["login", "preserve-environment", "pty"], |
| 372 | operands: 0, |
| 373 | }, |
| 374 | Wrapper { |
| 375 | name: "flock", |
| 376 | short_values: "wEc", |
| 377 | short_switches: "sxunoFh", |
| 378 | long_values: &["timeout", "wait", "conflict-exit-code", "command"], |
| 379 | long_switches: &[ |
| 380 | "shared", |
| 381 | "exclusive", |
| 382 | "unlock", |
| 383 | "nonblock", |
| 384 | "nb", |
| 385 | "close", |
| 386 | "no-fork", |
| 387 | "verbose", |
| 388 | ], |
| 389 | operands: 1, |
| 390 | }, |
| 391 | Wrapper { |
| 392 | name: "watch", |
| 393 | short_values: "nq", |
| 394 | short_switches: "bcdeghprtwx", |
| 395 | long_values: &["interval", "equexit"], |
| 396 | long_switches: &[ |
| 397 | "beep", |
| 398 | "color", |
| 399 | "no-color", |
| 400 | "differences", |
| 401 | "errexit", |
| 402 | "chgexit", |
| 403 | "precise", |
| 404 | "no-title", |
| 405 | "no-wrap", |
| 406 | "exec", |
| 407 | ], |
| 408 | operands: 0, |
| 409 | }, |
| 410 | Wrapper { |
| 411 | name: "wsl", |
| 412 | short_values: "du", |
| 413 | short_switches: "e", |
| 414 | long_values: &["distribution", "user", "cd", "shell-type"], |
| 415 | long_switches: &["exec"], |
| 416 | operands: 0, |
| 417 | }, // Privilege, sandbox, scheduling and tracing launchers: each runs its |
| 418 | // operands as a command. |
| 419 | Wrapper { |
| 420 | name: "pkexec", |
| 421 | short_values: "", |
| 422 | short_switches: "", |
| 423 | long_values: &["user"], |
| 424 | long_switches: &["disable-internal-agent", "keep-cwd"], |
| 425 | operands: 0, |
| 426 | }, |
| 427 | Wrapper { |
| 428 | name: "run0", |
| 429 | short_values: "ugD", |
| 430 | short_switches: "", |
| 431 | long_values: &[ |
| 432 | "user", |
| 433 | "group", |
| 434 | "nice", |
| 435 | "chdir", |
| 436 | "setenv", |
| 437 | "unit", |
| 438 | "property", |
| 439 | "description", |
| 440 | "slice", |
| 441 | "machine", |
| 442 | ], |
| 443 | long_switches: &["no-ask-password", "background", "pty", "pipe"], |
| 444 | operands: 0, |
| 445 | }, |
| 446 | Wrapper { |
| 447 | name: "fakeroot", |
| 448 | short_values: "lsib", |
| 449 | short_switches: "uhv", |
| 450 | long_values: &["lib", "faked"], |
| 451 | long_switches: &["unknown-is-real"], |
| 452 | operands: 0, |
| 453 | }, |
| 454 | Wrapper { |
| 455 | name: "taskset", |
| 456 | short_values: "", |
| 457 | short_switches: "acpV", |
| 458 | long_values: &[], |
| 459 | long_switches: &["all-tasks", "cpu-list", "pid"], |
| 460 | operands: 1, |
| 461 | }, |
| 462 | Wrapper { |
| 463 | name: "chrt", |
| 464 | short_values: "TPD", |
| 465 | short_switches: "abdefimoprRv", |
| 466 | long_values: &["sched-runtime", "sched-period", "sched-deadline"], |
| 467 | long_switches: &[ |
| 468 | "all-tasks", |
| 469 | "batch", |
| 470 | "deadline", |
| 471 | "fifo", |
| 472 | "idle", |
| 473 | "other", |
| 474 | "pid", |
| 475 | "rr", |
| 476 | "reset-on-fork", |
| 477 | "verbose", |
| 478 | "max", |
| 479 | ], |
| 480 | operands: 1, |
| 481 | }, |
| 482 | Wrapper { |
| 483 | name: "prlimit", |
| 484 | short_values: "po", |
| 485 | short_switches: "", |
| 486 | long_values: &["pid", "output"], |
| 487 | long_switches: &["noheadings", "raw", "verbose"], |
| 488 | operands: 0, |
| 489 | }, |
| 490 | Wrapper { |
| 491 | name: "strace", |
| 492 | short_values: "abeIoOpPsSuEX", |
| 493 | short_switches: "cCdDfFhikNqrtTvVwxyYzZ", |
| 494 | long_values: &[ |
| 495 | "output", |
| 496 | "expr", |
| 497 | "trace", |
| 498 | "signal", |
| 499 | "status", |
| 500 | "attach", |
| 501 | "user", |
| 502 | "env", |
| 503 | "string-limit", |
| 504 | "trace-path", |
| 505 | ], |
| 506 | long_switches: &[ |
| 507 | "follow-forks", |
| 508 | "output-separately", |
| 509 | "summary-only", |
| 510 | "summary", |
| 511 | "no-abbrev", |
| 512 | "verbose", |
| 513 | ], |
| 514 | operands: 0, |
| 515 | }, |
| 516 | Wrapper { |
| 517 | name: "ltrace", |
| 518 | short_values: "aAeDFlnopsuxwX", |
| 519 | short_switches: "bcCfhiLrStTV", |
| 520 | long_values: &["output", "library", "indent", "align"], |
| 521 | long_switches: &["demangle", "help", "version"], |
| 522 | operands: 0, |
| 523 | }, |
| 524 | Wrapper { |
| 525 | name: "systemd-run", |
| 526 | short_values: "puEHM", |
| 527 | short_switches: "rtqGdPS", |
| 528 | long_values: &[ |
| 529 | "property", |
| 530 | "unit", |
| 531 | "description", |
| 532 | "slice", |
| 533 | "setenv", |
| 534 | "uid", |
| 535 | "gid", |
| 536 | "nice", |
| 537 | "working-directory", |
| 538 | "host", |
| 539 | "machine", |
| 540 | "service-type", |
| 541 | "on-active", |
| 542 | "on-boot", |
| 543 | "on-startup", |
| 544 | "on-unit-active", |
| 545 | "on-unit-inactive", |
| 546 | "on-calendar", |
| 547 | "timer-property", |
| 548 | "path-property", |
| 549 | "socket-property", |
| 550 | ], |
| 551 | long_switches: &[ |
| 552 | "user", |
| 553 | "system", |
| 554 | "scope", |
| 555 | "pty", |
| 556 | "pipe", |
| 557 | "wait", |
| 558 | "collect", |
| 559 | "quiet", |
| 560 | "no-ask-password", |
| 561 | "remain-after-exit", |
| 562 | "same-dir", |
| 563 | "no-block", |
| 564 | "send-sighup", |
| 565 | ], |
| 566 | operands: 0, |
| 567 | }, |
| 568 | Wrapper { |
| 569 | name: "numactl", |
| 570 | short_values: "NmCipPw", |
| 571 | short_switches: "laHsS", |
| 572 | long_values: &[ |
| 573 | "cpunodebind", |
| 574 | "membind", |
| 575 | "physcpubind", |
| 576 | "interleave", |
| 577 | "preferred", |
| 578 | "preferred-many", |
| 579 | "weighted-interleave", |
| 580 | "huge", |
| 581 | "offset", |
| 582 | "length", |
| 583 | "mode", |
| 584 | "strict", |
| 585 | "shmmode", |
| 586 | "shmid", |
| 587 | "shm", |
| 588 | "file", |
| 589 | ], |
| 590 | long_switches: &["localalloc", "all", "hardware", "show", "touch"], |
| 591 | operands: 0, |
| 592 | }, |
| 593 | Wrapper { |
| 594 | name: "firejail", |
| 595 | short_values: "", |
| 596 | short_switches: "", |
| 597 | long_values: &[], |
| 598 | long_switches: &[ |
| 599 | "noprofile", |
| 600 | "quiet", |
| 601 | "private", |
| 602 | "private-dev", |
| 603 | "private-tmp", |
| 604 | "nonewprivs", |
| 605 | "noroot", |
| 606 | "seccomp", |
| 607 | "x11", |
| 608 | "appimage", |
| 609 | "allusers", |
| 610 | ], |
| 611 | operands: 0, |
| 612 | }, |
| 613 | Wrapper { |
| 614 | name: "xvfb-run", |
| 615 | short_values: "efnpsw", |
| 616 | short_switches: "alh", |
| 617 | long_values: &[ |
| 618 | "error-file", |
| 619 | "auth-file", |
| 620 | "server-num", |
| 621 | "xauth-protocol", |
| 622 | "server-args", |
| 623 | "wait", |
| 624 | ], |
| 625 | long_switches: &["auto-servernum", "listen-tcp", "help"], |
| 626 | operands: 0, |
| 627 | }, |
| 628 | Wrapper { |
| 629 | name: "dbus-launch", |
| 630 | short_values: "", |
| 631 | short_switches: "", |
| 632 | long_values: &[], |
| 633 | long_switches: &[ |
| 634 | "sh-syntax", |
| 635 | "csh-syntax", |
| 636 | "auto-syntax", |
| 637 | "binary-syntax", |
| 638 | "close-stderr", |
| 639 | "exit-with-session", |
| 640 | "exit-with-x11", |
| 641 | "version", |
| 642 | ], |
| 643 | operands: 0, |
| 644 | }, |
| 645 | Wrapper { |
| 646 | name: "dbus-run-session", |
| 647 | short_values: "", |
| 648 | short_switches: "", |
| 649 | long_values: &["config-file", "dbus-daemon"], |
| 650 | long_switches: &["session", "version", "help"], |
| 651 | operands: 0, |
| 652 | }, |
| 653 | Wrapper { |
| 654 | name: "sg", |
| 655 | short_values: "", |
| 656 | short_switches: "c", |
| 657 | long_values: &[], |
| 658 | long_switches: &[], |
| 659 | operands: 1, |
| 660 | }, |
| 661 | Wrapper { |
| 662 | name: "proxychains", |
| 663 | short_values: "f", |
| 664 | short_switches: "q", |
| 665 | long_values: &[], |
| 666 | long_switches: &[], |
| 667 | operands: 0, |
| 668 | }, |
| 669 | Wrapper { |
| 670 | name: "proxychains4", |
| 671 | short_values: "f", |
| 672 | short_switches: "q", |
| 673 | long_values: &[], |
| 674 | long_switches: &[], |
| 675 | operands: 0, |
| 676 | }, |
| 677 | Wrapper { |
| 678 | name: "torsocks", |
| 679 | short_values: "uapP", |
| 680 | short_switches: "idqh", |
| 681 | long_values: &["user", "pass", "address", "port"], |
| 682 | long_switches: &["isolate", "debug", "quiet", "shell", "help", "version"], |
| 683 | operands: 0, |
| 684 | }, |
| 685 | Wrapper { |
| 686 | name: "tsocks", |
| 687 | short_values: "", |
| 688 | short_switches: "", |
| 689 | long_values: &[], |
| 690 | long_switches: &[], |
| 691 | operands: 0, |
| 692 | }, |
| 693 | Wrapper { |
| 694 | name: "eatmydata", |
| 695 | short_values: "", |
| 696 | short_switches: "", |
| 697 | long_values: &[], |
| 698 | long_switches: &[], |
| 699 | operands: 0, |
| 700 | }, |
| 701 | Wrapper { |
| 702 | name: "cpulimit", |
| 703 | short_values: "lpe", |
| 704 | short_switches: "mzikvh", |
| 705 | long_values: &["limit", "pid", "exe", "cpu"], |
| 706 | long_switches: &[ |
| 707 | "monitor-forks", |
| 708 | "lazy", |
| 709 | "include-children", |
| 710 | "kill", |
| 711 | "verbose", |
| 712 | "help", |
| 713 | ], |
| 714 | operands: 0, |
| 715 | }, |
| 716 | Wrapper { |
| 717 | name: "setpriv", |
| 718 | short_values: "", |
| 719 | short_switches: "", |
| 720 | long_values: &[ |
| 721 | "reuid", |
| 722 | "regid", |
| 723 | "groups", |
| 724 | "inh-caps", |
| 725 | "ambient-caps", |
| 726 | "bounding-set", |
| 727 | "securebits", |
| 728 | "pdeathsig", |
| 729 | "selinux-label", |
| 730 | "apparmor-profile", |
| 731 | "landlock-access", |
| 732 | "landlock-rule", |
| 733 | ], |
| 734 | long_switches: &[ |
| 735 | "clear-groups", |
| 736 | "keep-groups", |
| 737 | "init-groups", |
| 738 | "nnp", |
| 739 | "no-new-privs", |
| 740 | "reset-env", |
| 741 | "dump", |
| 742 | "list-caps", |
| 743 | ], |
| 744 | operands: 0, |
| 745 | }, |
| 746 | Wrapper { |
| 747 | name: "trickle", |
| 748 | short_values: "udwtlnL", |
| 749 | short_switches: "svh", |
| 750 | long_values: &[], |
| 751 | long_switches: &[], |
| 752 | operands: 0, |
| 753 | }, |
| 754 | // GNU coreutils as installed on macOS (`brew install coreutils`). |
| 755 | Wrapper { |
| 756 | name: "gtimeout", |
| 757 | short_values: "sk", |
| 758 | short_switches: "fpv", |
| 759 | long_values: &["signal", "kill-after"], |
| 760 | long_switches: &["preserve-status", "foreground", "verbose"], |
| 761 | operands: 1, |
| 762 | }, |
| 763 | Wrapper { |
| 764 | name: "gnice", |
| 765 | short_values: "n", |
| 766 | short_switches: "", |
| 767 | long_values: &["adjustment"], |
| 768 | long_switches: &[], |
| 769 | operands: 0, |
| 770 | }, |
| 771 | Wrapper { |
| 772 | name: "gnohup", |
| 773 | short_values: "", |
| 774 | short_switches: "", |
| 775 | long_values: &[], |
| 776 | long_switches: &[], |
| 777 | operands: 0, |
| 778 | }, |
| 779 | Wrapper { |
| 780 | name: "gstdbuf", |
| 781 | short_values: "ioe", |
| 782 | short_switches: "", |
| 783 | long_values: &["input", "output", "error"], |
| 784 | long_switches: &[], |
| 785 | operands: 0, |
| 786 | }, |
| 787 | Wrapper { |
| 788 | name: "gchroot", |
| 789 | short_values: "ugG", |
| 790 | short_switches: "", |
| 791 | long_values: &["userspec", "groups"], |
| 792 | long_switches: &["skip-chdir"], |
| 793 | operands: 1, |
| 794 | }, |
| 795 | ]; |
| 796 | |
| 797 | /// Shell reserved words that can lead a simple command without being it: |
| 798 | /// `if rm x; then …`, `while rm x; do …`, `! rm x`. |
| 799 | const RESERVED_PREFIXES: &[&str] = &[ |
| 800 | "if", "then", "elif", "else", "do", "while", "until", "!", "coproc", |
| 801 | ]; |
| 802 | |
| 803 | /// Shells whose `-c` argument is a command line to be parsed, not an operand. |
| 804 | const SHELL_NAMES: &[&str] = &[ |
| 805 | "sh", "bash", "zsh", "dash", "ksh", "ksh93", "mksh", "ash", "fish", "csh", "tcsh", "rbash", |
| 806 | "yash", |
| 807 | ]; |
| 808 | |
| 809 | /// The commands a shell line would run, plus what could not be resolved. |
| 810 | #[derive(Debug, Clone, Default, PartialEq, Eq)] |
| 811 | pub struct Expansion { |
| 812 | /// Every command line found, as [`expanded_commands`] returns them. |
| 813 | pub commands: Vec<String>, |
| 814 | /// A command word (or a shell's script) is only known at run time, so no |
| 815 | /// deny rule can be matched against it reliably. |
| 816 | pub dynamic: bool, |
| 817 | /// Some code runs nested inside another command: a command or process |
| 818 | /// substitution, `eval`, a shell `-c` payload or stdin script, or a |
| 819 | /// `find -exec` payload. |
| 820 | pub nested: bool, |
| 821 | /// An argument's value is resolved only at execution time. A prefix grant |
| 822 | /// cannot check whether that value introduces a write/execute option. |
| 823 | /// Unlike `dynamic`, this does not make the command head unknowable. |
| 824 | pub arguments_dynamic: bool, |
| 825 | /// Unquoted control operators or grouping. Recorded before command |
| 826 | /// deduplication: `echo x; echo x` still contains a command list. |
| 827 | pub control: bool, |
| 828 | /// Unquoted redirection syntax, including descriptor duplication and |
| 829 | /// heredocs. Quoted operator characters and heredoc body data are excluded. |
| 830 | pub redirects: bool, |
| 831 | } |
| 832 | |
| 833 | /// Returns every command line the shell would execute for `command`. |
| 834 | /// |
| 835 | /// Results contain word-split commands, substitutions and wrapper payloads. |
| 836 | /// Literal data, including quoted heredoc bodies, is not treated as code. |
| 837 | /// Windows scans retain native path separators as well as POSIX candidates. |
| 838 | pub fn expanded_commands(command: &str) -> Vec<String> { |
| 839 | expand_command(command).commands |
| 840 | } |
| 841 | |
| 842 | /// [`expanded_commands`] plus the flags a policy needs to fail closed. |
| 843 | pub fn expand_command(command: &str) -> Expansion { |
| 844 | expand_for_platform(command, cfg!(windows)) |
| 845 | } |
| 846 | |
| 847 | #[cfg(test)] |
| 848 | fn expanded_commands_for_platform(command: &str, windows: bool) -> Vec<String> { |
| 849 | expand_for_platform(command, windows).commands |
| 850 | } |
| 851 | |
| 852 | fn expand_for_platform(command: &str, windows: bool) -> Expansion { |
| 853 | let mut expander = Expander { |
| 854 | out: Vec::new(), |
| 855 | seen: HashSet::new(), |
| 856 | literal_backslashes: windows, |
| 857 | dynamic: false, |
| 858 | nested: false, |
| 859 | arguments_dynamic: false, |
| 860 | control: false, |
| 861 | redirects: false, |
| 862 | }; |
| 863 | // Native Windows shells preserve path separators. Also retain the POSIX |
| 864 | // interpretation for Bash/WSL commands. Both passes use the same bounded, |
| 865 | // heredoc-aware parser and only contribute deny targets, never grants. |
| 866 | expander.expand(command, 0); |
| 867 | if windows { |
| 868 | expander.literal_backslashes = false; |
| 869 | expander.expand(command, 0); |
| 870 | } |
| 871 | Expansion { |
| 872 | commands: expander.out, |
| 873 | dynamic: expander.dynamic, |
| 874 | nested: expander.nested, |
| 875 | arguments_dynamic: expander.arguments_dynamic, |
| 876 | control: expander.control, |
| 877 | redirects: expander.redirects, |
| 878 | } |
| 879 | } |
| 880 | |
| 881 | struct Expander { |
| 882 | out: Vec<String>, |
| 883 | seen: HashSet<String>, |
| 884 | literal_backslashes: bool, |
| 885 | dynamic: bool, |
| 886 | nested: bool, |
| 887 | arguments_dynamic: bool, |
| 888 | control: bool, |
| 889 | redirects: bool, |
| 890 | } |
| 891 | |
| 892 | /// The word being read, with what the parser learned about it. |
| 893 | #[derive(Default)] |
| 894 | struct WordBuf { |
| 895 | text: String, |
| 896 | started: bool, |
| 897 | quoted: bool, |
| 898 | redirect_operand: bool, |
| 899 | /// The word's final text depends on an expansion the parser does not run. |
| 900 | dynamic: bool, |
| 901 | /// An unquoted `{` appeared, so brace expansion may rewrite the word. |
| 902 | brace: bool, |
| 903 | /// An unquoted `[` appeared, so the word may be a bracket glob. |
| 904 | bracket: bool, |
| 905 | } |
| 906 | |
| 907 | impl WordBuf { |
| 908 | fn flush(&mut self, line: &mut Line) { |
| 909 | if self.started || !self.text.is_empty() || self.dynamic { |
| 910 | if self.redirect_operand { |
| 911 | self.text.clear(); |
| 912 | self.redirect_operand = false; |
| 913 | } else { |
| 914 | // `[` alone is the test builtin and `{}` is a find |
| 915 | // placeholder; only a closed bracket or a brace list expands. |
| 916 | let dynamic = self.dynamic |
| 917 | || (self.bracket && self.text.len() > 1 && self.text.contains(']')) |
| 918 | || (self.brace && (self.text.contains(',') || self.text.contains(".."))); |
| 919 | line.words.push(std::mem::take(&mut self.text)); |
| 920 | line.dynamic.push(dynamic); |
| 921 | } |
| 922 | self.started = false; |
| 923 | } |
| 924 | self.quoted = false; |
| 925 | self.dynamic = false; |
| 926 | self.brace = false; |
| 927 | self.bracket = false; |
| 928 | } |
| 929 | } |
| 930 | |
| 931 | /// One simple command: its words, which of them are dynamic, and whether a |
| 932 | /// heredoc feeds its stdin. |
| 933 | #[derive(Default, Clone)] |
| 934 | struct Line { |
| 935 | words: Vec<String>, |
| 936 | dynamic: Vec<bool>, |
| 937 | heredoc: bool, |
| 938 | } |
| 939 | |
| 940 | /// How a shell invocation receives its script. |
| 941 | enum ShellInput { |
| 942 | /// `-c` / `--command`: the word at each offset may be the command line. |
| 943 | Command(Vec<usize>), |
| 944 | /// A script file operand at this offset. |
| 945 | Script(usize), |
| 946 | /// The script is read from stdin. |
| 947 | Stdin, |
| 948 | } |
| 949 | |
| 950 | impl Expander { |
| 951 | fn emit(&mut self, tokens: &[String]) { |
| 952 | let joined = tokens |
| 953 | .iter() |
| 954 | .filter(|token| !token.is_empty()) |
| 955 | .cloned() |
| 956 | .collect::<Vec<_>>() |
| 957 | .join(" "); |
| 958 | if joined.is_empty() || self.seen.contains(&joined) { |
| 959 | return; |
| 960 | } |
| 961 | if self.out.len() >= MAX_COMMANDS { |
| 962 | // A dropped command line was never checked: unresolved. |
| 963 | self.dynamic = true; |
| 964 | return; |
| 965 | } |
| 966 | self.seen.insert(joined.clone()); |
| 967 | self.out.push(joined); |
| 968 | } |
| 969 | |
| 970 | /// Word-split `input` into command lines and record each one, recursing |
| 971 | /// into every nested command text found along the way. |
| 972 | fn expand(&mut self, input: &str, depth: usize) { |
| 973 | if depth > MAX_DEPTH || self.out.len() >= MAX_COMMANDS { |
| 974 | // Unexamined code is unresolved code. |
| 975 | self.dynamic = true; |
| 976 | return; |
| 977 | } |
| 978 | if depth > 0 && !input.trim().is_empty() { |
| 979 | self.nested = true; |
| 980 | } |
| 981 | // shlex does not implement ANSI-C/localized quoting or shell CR |
| 982 | // semantics. Keep the conservative scan for those forms rather than |
| 983 | // let an unrecognized heredoc delimiter hide following commands. |
| 984 | if input.contains('\r') || input.contains("$'") || input.contains("$\"") { |
| 985 | for segment in super::command_segments(input) { |
| 986 | self.emit(&[segment]); |
| 987 | } |
| 988 | } |
| 989 | let chars: Vec<char> = input.chars().collect(); |
| 990 | let n = chars.len(); |
| 991 | let mut i = 0usize; |
| 992 | let mut commands: Vec<Line> = Vec::new(); |
| 993 | let mut line = Line::default(); |
| 994 | let mut word = WordBuf::default(); |
| 995 | let mut nested: Vec<String> = Vec::new(); |
| 996 | let mut heredocs = Vec::new(); |
| 997 | // Set when the text does not parse cleanly (an unterminated quote, |
| 998 | // substitution or heredoc, a `case` inside a substitution): where |
| 999 | // this parser ends a region may not be where the shell ends it. |
| 1000 | let mut uncertain = false; |
| 1001 | |
| 1002 | while i < n { |
| 1003 | let c = chars[i]; |
| 1004 | match c { |
| 1005 | '#' if !word.started && word.text.is_empty() => { |
| 1006 | while i < n && chars[i] != '\n' { |
| 1007 | i += 1; |
| 1008 | } |
| 1009 | } |
| 1010 | // A backslash outside quotes escapes exactly one character, |
| 1011 | // including an operator: `echo a\;b` is one word, not two |
| 1012 | // commands. A backslash-newline is a line continuation. |
| 1013 | '\\' if self.literal_backslashes => { |
| 1014 | word.text.push('\\'); |
| 1015 | word.started = true; |
| 1016 | i += 1; |
| 1017 | } |
| 1018 | '\\' => { |
| 1019 | if i + 1 < n { |
| 1020 | if chars[i + 1] != '\n' { |
| 1021 | word.text.push(chars[i + 1]); |
| 1022 | word.started = true; |
| 1023 | word.quoted = true; |
| 1024 | } |
| 1025 | i += 2; |
| 1026 | } else { |
| 1027 | i += 1; |
| 1028 | } |
| 1029 | } |
| 1030 | // Single quotes are fully literal: no substitution, no escapes. |
| 1031 | '\'' => { |
| 1032 | word.started = true; |
| 1033 | word.quoted = true; |
| 1034 | i += 1; |
| 1035 | while i < n && chars[i] != '\'' { |
| 1036 | word.text.push(chars[i]); |
| 1037 | i += 1; |
| 1038 | } |
| 1039 | uncertain |= i >= n; |
| 1040 | i = (i + 1).min(n); |
| 1041 | } |
| 1042 | // Double quotes suppress word splitting but NOT substitution. |
| 1043 | '"' => { |
| 1044 | word.started = true; |
| 1045 | word.quoted = true; |
| 1046 | i += 1; |
| 1047 | while i < n && chars[i] != '"' { |
| 1048 | match chars[i] { |
| 1049 | '\\' if self.literal_backslashes => { |
| 1050 | word.text.push('\\'); |
| 1051 | i += 1; |
| 1052 | } |
| 1053 | '\\' if i + 1 < n => { |
| 1054 | word.text.push(chars[i + 1]); |
| 1055 | i += 2; |
| 1056 | } |
| 1057 | '`' => { |
| 1058 | word.dynamic = true; |
| 1059 | let (inner, next, certain) = read_backtick(&chars, i); |
| 1060 | uncertain |= !certain; |
| 1061 | nested.push(inner); |
| 1062 | i = next; |
| 1063 | } |
| 1064 | '$' if i + 1 < n && matches!(chars[i + 1], '(' | '{') => { |
| 1065 | word.dynamic = true; |
| 1066 | let (inner, next, certain) = read_substitution(&chars, i + 1); |
| 1067 | uncertain |= !certain; |
| 1068 | nested.push(inner); |
| 1069 | i = next; |
| 1070 | } |
| 1071 | ch => { |
| 1072 | if ch == '$' && starts_parameter(chars.get(i + 1)) { |
| 1073 | word.dynamic = true; |
| 1074 | } |
| 1075 | word.text.push(ch); |
| 1076 | i += 1; |
| 1077 | } |
| 1078 | } |
| 1079 | } |
| 1080 | uncertain |= i >= n; |
| 1081 | i = (i + 1).min(n); |
| 1082 | } |
| 1083 | // `$'…'` (ANSI-C quoting) is literal text with C escapes. The |
| 1084 | // escapes are not decoded here, so a word that uses them is |
| 1085 | // not known statically (`$'\x72m'` is `rm`). |
| 1086 | '$' if i + 1 < n && chars[i + 1] == '\'' => { |
| 1087 | word.started = true; |
| 1088 | word.quoted = true; |
| 1089 | i += 2; |
| 1090 | while i < n && chars[i] != '\'' { |
| 1091 | if chars[i] == '\\' && i + 1 < n { |
| 1092 | word.dynamic = true; |
| 1093 | word.text.push(chars[i + 1]); |
| 1094 | i += 2; |
| 1095 | } else { |
| 1096 | word.text.push(chars[i]); |
| 1097 | i += 1; |
| 1098 | } |
| 1099 | } |
| 1100 | uncertain |= i >= n; |
| 1101 | i = (i + 1).min(n); |
| 1102 | } |
| 1103 | // Command substitution, both spellings. The body is a command |
| 1104 | // line in its own right; the substitution contributes no text |
| 1105 | // to the enclosing word (we do not evaluate output). |
| 1106 | '`' => { |
| 1107 | word.started |= word.redirect_operand; |
| 1108 | word.dynamic = true; |
| 1109 | let (inner, next, certain) = read_backtick(&chars, i); |
| 1110 | uncertain |= !certain; |
| 1111 | nested.push(inner); |
| 1112 | i = next; |
| 1113 | } |
| 1114 | // `$(…)`, and `${…}`: an expansion, not a command — but it can |
| 1115 | // *contain* one (`${x:-$(rm -rf /)}`), so the body is rescanned. |
| 1116 | // Process substitution `<(…)` / `>(…)` also runs its body. |
| 1117 | '$' | '<' | '>' |
| 1118 | if i + 1 < n && (chars[i + 1] == '(' || (c == '$' && chars[i + 1] == '{')) => |
| 1119 | { |
| 1120 | word.started |= word.redirect_operand; |
| 1121 | word.dynamic = true; |
| 1122 | let (inner, next, certain) = read_substitution(&chars, i + 1); |
| 1123 | uncertain |= !certain; |
| 1124 | nested.push(inner); |
| 1125 | i = next; |
| 1126 | } |
| 1127 | '<' if chars.get(i + 1) == Some(&'<') && chars.get(i + 2) != Some(&'<') => { |
| 1128 | self.redirects = true; |
| 1129 | if !word.quoted && is_redirect_descriptor(&word.text) { |
| 1130 | word.text.clear(); |
| 1131 | word.started = false; |
| 1132 | } |
| 1133 | word.flush(&mut line); |
| 1134 | line.heredoc = true; |
| 1135 | i += 2; |
| 1136 | let strip_tabs = chars.get(i) == Some(&'-'); |
| 1137 | if strip_tabs { |
| 1138 | i += 1; |
| 1139 | } |
| 1140 | while i < n && matches!(chars[i], ' ' | '\t') { |
| 1141 | i += 1; |
| 1142 | } |
| 1143 | let start = i; |
| 1144 | let mut quote = None; |
| 1145 | let mut literal = false; |
| 1146 | while i < n { |
| 1147 | let ch = chars[i]; |
| 1148 | if quote.is_none() |
| 1149 | && matches!(ch, ' ' | '\t' | '\n' | ';' | '|' | '&' | '<' | '>') |
| 1150 | { |
| 1151 | break; |
| 1152 | } |
| 1153 | if ch == '\\' && quote != Some('\'') { |
| 1154 | literal = true; |
| 1155 | i = (i + 2).min(n); |
| 1156 | continue; |
| 1157 | } |
| 1158 | if matches!(ch, '\'' | '"') { |
| 1159 | literal = true; |
| 1160 | if quote == Some(ch) { |
| 1161 | quote = None; |
| 1162 | } else if quote.is_none() { |
| 1163 | quote = Some(ch); |
| 1164 | } |
| 1165 | } |
| 1166 | i += 1; |
| 1167 | } |
| 1168 | let raw: String = chars[start..i].iter().collect(); |
| 1169 | if let Some(delimiter) = shlex::split(&raw) |
| 1170 | .and_then(|mut words| (words.len() == 1).then(|| words.remove(0))) |
| 1171 | { |
| 1172 | heredocs.push((delimiter, literal, strip_tabs)); |
| 1173 | } |
| 1174 | } |
| 1175 | // Unquoted redirections are syntax, even without whitespace. |
| 1176 | // Keep the command words on both sides together, but omit the |
| 1177 | // descriptor and next operand. Parse that operand normally so |
| 1178 | // nested substitutions are still checked as commands. |
| 1179 | '<' | '>' | '&' if redirection_len(&chars[i..]) > 0 => { |
| 1180 | self.redirects = true; |
| 1181 | if !word.quoted && is_redirect_descriptor(&word.text) { |
| 1182 | word.text.clear(); |
| 1183 | word.started = false; |
| 1184 | } |
| 1185 | word.flush(&mut line); |
| 1186 | word.redirect_operand = true; |
| 1187 | i += redirection_len(&chars[i..]); |
| 1188 | } |
| 1189 | ' ' | '\t' => { |
| 1190 | word.flush(&mut line); |
| 1191 | i += 1; |
| 1192 | } |
| 1193 | // A subshell boundary. `$(`, `<(` and `>(` were consumed by the |
| 1194 | // arms above, so a bare paren here is grouping: the body is a |
| 1195 | // command list of its own, not part of the surrounding word. |
| 1196 | '(' | ')' => { |
| 1197 | self.control = true; |
| 1198 | word.flush(&mut line); |
| 1199 | end_command(&mut commands, &mut line); |
| 1200 | word.redirect_operand = false; |
| 1201 | if c == '(' && chars.get(i + 1) == Some(&'(') { |
| 1202 | // `(( … ))` is arithmetic, where `<<` is a shift and |
| 1203 | // not a heredoc. Its substitutions still run, and |
| 1204 | // `((cmd) )` is two subshells, so the body is scanned |
| 1205 | // as code on its own; a heredoc it appears to open |
| 1206 | // cannot swallow the lines after it. |
| 1207 | let (inner, next, certain) = read_substitution(&chars, i); |
| 1208 | uncertain |= !certain; |
| 1209 | nested.push(inner); |
| 1210 | i = next; |
| 1211 | } else { |
| 1212 | i += 1; |
| 1213 | } |
| 1214 | } |
| 1215 | // Control operators end the current command line. `&&`, `||`, |
| 1216 | // `;;`, `|&` and runs of newlines collapse into one break. |
| 1217 | '\n' | '\r' | ';' | '&' | '|' => { |
| 1218 | self.control = true; |
| 1219 | word.flush(&mut line); |
| 1220 | end_command(&mut commands, &mut line); |
| 1221 | word.redirect_operand = false; |
| 1222 | i += 1; |
| 1223 | if c == '\n' && !heredocs.is_empty() { |
| 1224 | let shell_stdin = commands.iter().any(|line| { |
| 1225 | command_heads(&line.words).heads.iter().any(|&head| { |
| 1226 | let name = command_name(&line.words[head]); |
| 1227 | SHELL_NAMES.contains(&name.as_str()) |
| 1228 | || matches!(name.as_str(), "source" | ".") |
| 1229 | }) |
| 1230 | }); |
| 1231 | for (delimiter, literal, strip_tabs) in heredocs.drain(..) { |
| 1232 | let mut body = String::new(); |
| 1233 | // A body that runs to the end of the input may be |
| 1234 | // a misread `<<`; what it swallowed is unexamined. |
| 1235 | let mut terminated = false; |
| 1236 | while i < n { |
| 1237 | let start = i; |
| 1238 | while i < n && chars[i] != '\n' { |
| 1239 | i += 1; |
| 1240 | } |
| 1241 | let mut text: String = chars[start..i].iter().collect(); |
| 1242 | if i < n { |
| 1243 | i += 1; |
| 1244 | } |
| 1245 | // An unquoted heredoc joins escaped newlines |
| 1246 | // before checking its delimiter (E\ + OF can |
| 1247 | // terminate EOF). Do not swallow later code. |
| 1248 | while !literal |
| 1249 | && text.chars().rev().take_while(|c| *c == '\\').count() % 2 |
| 1250 | == 1 |
| 1251 | && i < n |
| 1252 | { |
| 1253 | text.pop(); |
| 1254 | let start = i; |
| 1255 | while i < n && chars[i] != '\n' { |
| 1256 | i += 1; |
| 1257 | } |
| 1258 | text.extend(chars[start..i].iter()); |
| 1259 | if i < n { |
| 1260 | i += 1; |
| 1261 | } |
| 1262 | } |
| 1263 | let text = if strip_tabs { |
| 1264 | text.trim_start_matches('\t') |
| 1265 | } else { |
| 1266 | &text |
| 1267 | }; |
| 1268 | if text == delimiter { |
| 1269 | terminated = true; |
| 1270 | break; |
| 1271 | } |
| 1272 | body.push_str(text); |
| 1273 | body.push('\n'); |
| 1274 | } |
| 1275 | uncertain |= !terminated; |
| 1276 | if shell_stdin { |
| 1277 | nested.push(body); |
| 1278 | } else if !literal { |
| 1279 | let (bodies, certain) = heredoc_substitutions(&body); |
| 1280 | uncertain |= !certain; |
| 1281 | nested.extend(bodies); |
| 1282 | } |
| 1283 | } |
| 1284 | } |
| 1285 | } |
| 1286 | _ => { |
| 1287 | if c == '$' && starts_parameter(chars.get(i + 1)) { |
| 1288 | word.dynamic = true; |
| 1289 | } |
| 1290 | match c { |
| 1291 | '*' | '?' => word.dynamic = true, |
| 1292 | '[' => word.bracket = true, |
| 1293 | '{' => word.brace = true, |
| 1294 | _ => {} |
| 1295 | } |
| 1296 | word.text.push(c); |
| 1297 | word.started = true; |
| 1298 | i += 1; |
| 1299 | } |
| 1300 | } |
| 1301 | } |
| 1302 | word.flush(&mut line); |
| 1303 | end_command(&mut commands, &mut line); |
| 1304 | self.dynamic |= uncertain; |
| 1305 | |
| 1306 | for line in &commands { |
| 1307 | self.record(line, depth); |
| 1308 | } |
| 1309 | for inner in nested { |
| 1310 | self.expand(&inner, depth + 1); |
| 1311 | } |
| 1312 | } |
| 1313 | |
| 1314 | /// Record one word-split command line, plus the invocation hiding inside it |
| 1315 | /// when the head is a wrapper, and flag what cannot be resolved. |
| 1316 | fn record(&mut self, line: &Line, depth: usize) { |
| 1317 | let tokens = &line.words; |
| 1318 | if tokens.is_empty() { |
| 1319 | return; |
| 1320 | } |
| 1321 | if depth > MAX_DEPTH { |
| 1322 | self.dynamic = true; |
| 1323 | return; |
| 1324 | } |
| 1325 | self.emit(tokens); |
| 1326 | |
| 1327 | let heads = command_heads(tokens); |
| 1328 | self.dynamic |= heads.truncated; |
| 1329 | for payload in &heads.split_payloads { |
| 1330 | // `env -S 'cmd args'` splits its value into a command line. |
| 1331 | self.expand(payload, depth + 1); |
| 1332 | } |
| 1333 | let mut dynamic = line.dynamic.clone(); |
| 1334 | mark_replacement_operands(tokens, &heads.heads, &mut dynamic); |
| 1335 | for &head in &heads.heads { |
| 1336 | // `sudo rm -rf /` is an `rm -rf /`: emit the command the wrappers |
| 1337 | // run, with their options and operands removed. |
| 1338 | if head > 0 { |
| 1339 | self.emit(&tokens[head..]); |
| 1340 | } |
| 1341 | // Also keep a wrapper's options in front of the command it runs, |
| 1342 | // so a deny rule can still skip them as flags if the option table |
| 1343 | // misreads one. (`command -v NAME` runs nothing: no head follows.) |
| 1344 | if wrapper_index(&tokens[head]).is_some() |
| 1345 | && heads.heads.iter().any(|&inner| inner > head) |
| 1346 | { |
| 1347 | self.emit(&tokens[head + 1..]); |
| 1348 | } |
| 1349 | // A command word (or a wrapper word before it) that the shell |
| 1350 | // rewrites at run time cannot be matched against any rule. |
| 1351 | if tokens[..=head] |
| 1352 | .iter() |
| 1353 | .zip(&dynamic) |
| 1354 | .any(|(token, dynamic)| *dynamic && !is_env_assignment(token)) |
| 1355 | { |
| 1356 | self.dynamic = true; |
| 1357 | } |
| 1358 | let name = command_name(&tokens[head]); |
| 1359 | let args_dynamic = dynamic[head + 1..].iter().any(|dynamic| *dynamic); |
| 1360 | self.arguments_dynamic |= args_dynamic; |
| 1361 | match name.as_str() { |
| 1362 | // `eval …` takes a *command line* as data. Parse it. |
| 1363 | "eval" => self.code_payload(&tokens[head + 1..].join(" "), args_dynamic, depth), |
| 1364 | "source" | "." => match tokens.get(head + 1) { |
| 1365 | None if !line.heredoc => self.dynamic = true, |
| 1366 | Some(_) if dynamic[head + 1] => self.dynamic = true, |
| 1367 | Some(script) => self.dynamic |= script_read_at_run_time(script, line), |
| 1368 | _ => {} |
| 1369 | }, |
| 1370 | "find" => self.record_find_exec(line, head, depth), |
| 1371 | // `trap 'code' SIGNAL…` runs its first operand later in this |
| 1372 | // shell. |
| 1373 | "trap" => { |
| 1374 | if let Some(at) = |
| 1375 | (head + 1..tokens.len()).find(|&at| !tokens[at].starts_with('-')) |
| 1376 | { |
| 1377 | self.code_payload(&tokens[at], dynamic[at], depth); |
| 1378 | } |
| 1379 | } |
| 1380 | // These hand the value of `-c` / `--command` to a shell. |
| 1381 | "su" | "runuser" | "flock" | "script" => { |
| 1382 | for (payload, payload_dynamic) in |
| 1383 | command_option_payloads(tokens, head, &dynamic) |
| 1384 | { |
| 1385 | self.code_payload(&payload, payload_dynamic, depth); |
| 1386 | } |
| 1387 | } |
| 1388 | // `watch` and `sg GROUP [-c]` join their operands and hand |
| 1389 | // them to `sh -c`. |
| 1390 | "watch" | "sg" => { |
| 1391 | for &inner in heads.heads.iter().filter(|&&inner| inner > head) { |
| 1392 | let inner_dynamic = dynamic[inner..].iter().any(|dynamic| *dynamic); |
| 1393 | self.code_payload(&tokens[inner..].join(" "), inner_dynamic, depth); |
| 1394 | } |
| 1395 | } |
| 1396 | // `cmd /c …` parses the rest as a cmd.exe command line, which |
| 1397 | // expands `%VAR%` and `!VAR!` and strips `^` escapes itself. |
| 1398 | "cmd" => { |
| 1399 | if let Some(at) = (head + 1..tokens.len()).find(|&at| { |
| 1400 | matches!(tokens[at].to_ascii_lowercase().as_str(), "/c" | "/k" | "/r") |
| 1401 | }) { |
| 1402 | let payload = tokens[at + 1..].join(" "); |
| 1403 | let payload_dynamic = dynamic[at + 1..].iter().any(|dynamic| *dynamic) |
| 1404 | || payload.contains(['%', '!', '^']); |
| 1405 | self.code_payload(&payload, payload_dynamic, depth); |
| 1406 | } |
| 1407 | } |
| 1408 | "powershell" | "pwsh" => self.record_powershell(tokens, head, &dynamic, depth), |
| 1409 | _ if SHELL_NAMES.contains(&name.as_str()) => { |
| 1410 | match shell_input(&tokens[head..]) { |
| 1411 | ShellInput::Command(offsets) => { |
| 1412 | for offset in offsets { |
| 1413 | self.code_payload( |
| 1414 | &tokens[head + offset], |
| 1415 | dynamic[head + offset], |
| 1416 | depth, |
| 1417 | ); |
| 1418 | } |
| 1419 | } |
| 1420 | ShellInput::Script(offset) => { |
| 1421 | self.dynamic |= dynamic[head + offset] |
| 1422 | || script_read_at_run_time(&tokens[head + offset], line); |
| 1423 | } |
| 1424 | // A heredoc body was already expanded as the script; |
| 1425 | // any other stdin is only known at run time. |
| 1426 | ShellInput::Stdin => self.dynamic |= !line.heredoc, |
| 1427 | } |
| 1428 | } |
| 1429 | _ => {} |
| 1430 | } |
| 1431 | } |
| 1432 | } |
| 1433 | |
| 1434 | /// Code handed to another command as a string: parse it as a command line. |
| 1435 | fn code_payload(&mut self, payload: &str, dynamic: bool, depth: usize) { |
| 1436 | self.dynamic |= dynamic; |
| 1437 | self.expand(payload, depth + 1); |
| 1438 | } |
| 1439 | |
| 1440 | /// `powershell` / `pwsh`: `-Command` (or any prefix of it) takes the rest |
| 1441 | /// of the line as code, `-EncodedCommand` cannot be read statically, and |
| 1442 | /// a bare operand may start code too (Windows PowerShell's default). |
| 1443 | /// PowerShell syntax is only approximated by the POSIX parser, which is |
| 1444 | /// enough to find the command words a deny rule names. |
| 1445 | fn record_powershell( |
| 1446 | &mut self, |
| 1447 | tokens: &[String], |
| 1448 | head: usize, |
| 1449 | dynamic: &[bool], |
| 1450 | depth: usize, |
| 1451 | ) { |
| 1452 | let rest = |at: usize| { |
| 1453 | ( |
| 1454 | tokens[at..].join(" "), |
| 1455 | dynamic[at..].iter().any(|dynamic| *dynamic), |
| 1456 | ) |
| 1457 | }; |
| 1458 | let mut positional_seen = false; |
| 1459 | for at in head + 1..tokens.len() { |
| 1460 | let Some(option) = tokens[at].strip_prefix(['-', '/']) else { |
| 1461 | if !positional_seen { |
| 1462 | positional_seen = true; |
| 1463 | let (payload, payload_dynamic) = rest(at); |
| 1464 | self.code_payload(&payload, payload_dynamic, depth); |
| 1465 | } |
| 1466 | continue; |
| 1467 | }; |
| 1468 | let option = option.to_ascii_lowercase(); |
| 1469 | if option.is_empty() { |
| 1470 | continue; |
| 1471 | } |
| 1472 | if "command".starts_with(option.as_str()) { |
| 1473 | if at + 1 >= tokens.len() || tokens[at + 1] == "-" { |
| 1474 | // The code arrives on stdin. |
| 1475 | self.dynamic = true; |
| 1476 | } else { |
| 1477 | let (payload, payload_dynamic) = rest(at + 1); |
| 1478 | self.code_payload(&payload, payload_dynamic, depth); |
| 1479 | } |
| 1480 | return; |
| 1481 | } |
| 1482 | if option == "ec" || "encodedcommand".starts_with(option.as_str()) { |
| 1483 | self.dynamic = true; |
| 1484 | return; |
| 1485 | } |
| 1486 | if "file".starts_with(option.as_str()) { |
| 1487 | // A script file is opaque, as it is for every other shell. |
| 1488 | return; |
| 1489 | } |
| 1490 | } |
| 1491 | } |
| 1492 | |
| 1493 | /// `find … -exec CMD … ;` runs `CMD` for every match. |
| 1494 | fn record_find_exec(&mut self, line: &Line, head: usize, depth: usize) { |
| 1495 | let tokens = &line.words; |
| 1496 | let mut index = head + 1; |
| 1497 | while index < tokens.len() { |
| 1498 | if matches!( |
| 1499 | tokens[index].as_str(), |
| 1500 | "-exec" | "-execdir" | "-ok" | "-okdir" |
| 1501 | ) { |
| 1502 | let start = index + 1; |
| 1503 | let mut end = start; |
| 1504 | while end < tokens.len() && !matches!(tokens[end].as_str(), ";" | "+") { |
| 1505 | end += 1; |
| 1506 | } |
| 1507 | if start < end { |
| 1508 | self.nested = true; |
| 1509 | // `{}` is replaced by each file name found, so a word |
| 1510 | // holding it is only known at run time. |
| 1511 | let payload = Line { |
| 1512 | words: tokens[start..end].to_vec(), |
| 1513 | dynamic: (start..end) |
| 1514 | .map(|at| line.dynamic[at] || tokens[at].contains("{}")) |
| 1515 | .collect(), |
| 1516 | heredoc: false, |
| 1517 | }; |
| 1518 | self.record(&payload, depth + 1); |
| 1519 | } |
| 1520 | index = end; |
| 1521 | } |
| 1522 | index += 1; |
| 1523 | } |
| 1524 | } |
| 1525 | } |
| 1526 | |
| 1527 | /// Values of `-c CMD`, `-cCMD`, `--command CMD`, `--command=CMD` and |
| 1528 | /// `--session-command` after `tokens[head]`, with whether each is dynamic: |
| 1529 | /// the code `su`, `runuser`, `flock` and `script` hand to a shell. |
| 1530 | fn command_option_payloads( |
| 1531 | tokens: &[String], |
| 1532 | head: usize, |
| 1533 | dynamic: &[bool], |
| 1534 | ) -> Vec<(String, bool)> { |
| 1535 | let mut payloads = Vec::new(); |
| 1536 | for (at, token) in tokens.iter().enumerate().skip(head + 1) { |
| 1537 | let next = || { |
| 1538 | tokens |
| 1539 | .get(at + 1) |
| 1540 | .map(|value| (value.clone(), dynamic[at + 1])) |
| 1541 | }; |
| 1542 | if let Some(long) = token.strip_prefix("--") { |
| 1543 | let (name, inline) = match long.split_once('=') { |
| 1544 | Some((name, value)) => (name, Some(value)), |
| 1545 | None => (long, None), |
| 1546 | }; |
| 1547 | if matches!(name, "command" | "session-command") { |
| 1548 | payloads.extend(match inline { |
| 1549 | Some(value) => Some((value.to_string(), dynamic[at])), |
| 1550 | None => next(), |
| 1551 | }); |
| 1552 | } |
| 1553 | } else if let Some(flags) = token.strip_prefix('-') |
| 1554 | && let Some(offset) = flags.find('c') |
| 1555 | { |
| 1556 | let value = &flags[offset + 1..]; |
| 1557 | payloads.extend(if value.is_empty() { |
| 1558 | next() |
| 1559 | } else { |
| 1560 | Some((value.to_string(), dynamic[at])) |
| 1561 | }); |
| 1562 | } |
| 1563 | } |
| 1564 | payloads |
| 1565 | } |
| 1566 | |
| 1567 | /// Mark the operands of an `xargs` that contain its replacement string |
| 1568 | /// (`-I R`, `-J R`, `-i[R]`, `--replace[=R]`, `{}` by default) as known only |
| 1569 | /// at run time: `xargs -I{} sh -c {}` runs whatever arrives on stdin. |
| 1570 | fn mark_replacement_operands(tokens: &[String], heads: &[usize], dynamic: &mut [bool]) { |
| 1571 | let Some(xargs) = WRAPPERS.iter().find(|wrapper| wrapper.name == "xargs") else { |
| 1572 | return; |
| 1573 | }; |
| 1574 | for &head in heads { |
| 1575 | if command_name(&tokens[head]) != "xargs" { |
| 1576 | continue; |
| 1577 | } |
| 1578 | let mut replacements: Vec<String> = Vec::new(); |
| 1579 | let mut values = HashSet::new(); |
| 1580 | let mut index = head + 1; |
| 1581 | while index < tokens.len() && tokens[index].starts_with('-') && tokens[index] != "--" { |
| 1582 | let token = tokens[index].as_str(); |
| 1583 | if let Some(long) = token.strip_prefix("--") { |
| 1584 | match long.split_once('=') { |
| 1585 | Some(("replace", value)) => replacements.push(value.to_string()), |
| 1586 | None if long == "replace" => replacements.push("{}".to_string()), |
| 1587 | None if xargs.long_values.contains(&long) => { |
| 1588 | values.insert(index + 1); |
| 1589 | index += 1; |
| 1590 | } |
| 1591 | _ => {} |
| 1592 | } |
| 1593 | } else { |
| 1594 | for (at, flag) in token[1..].char_indices() { |
| 1595 | let rest = &token[1 + at + flag.len_utf8()..]; |
| 1596 | if flag == 'i' { |
| 1597 | replacements.push(if rest.is_empty() { "{}" } else { rest }.to_string()); |
| 1598 | break; |
| 1599 | } |
| 1600 | if xargs.short_values.contains(flag) { |
| 1601 | let value = if rest.is_empty() { |
| 1602 | values.insert(index + 1); |
| 1603 | index += 1; |
| 1604 | tokens.get(index).cloned() |
| 1605 | } else { |
| 1606 | Some(rest.to_string()) |
| 1607 | }; |
| 1608 | if matches!(flag, 'I' | 'J') { |
| 1609 | replacements.extend(value); |
| 1610 | } |
| 1611 | break; |
| 1612 | } |
| 1613 | } |
| 1614 | } |
| 1615 | index += 1; |
| 1616 | } |
| 1617 | for (at, token) in tokens.iter().enumerate().skip(head + 1) { |
| 1618 | if !values.contains(&at) |
| 1619 | && !token.starts_with('-') |
| 1620 | && replacements.iter().any(|replacement| { |
| 1621 | !replacement.is_empty() && token.contains(replacement.as_str()) |
| 1622 | }) |
| 1623 | { |
| 1624 | dynamic[at] = true; |
| 1625 | } |
| 1626 | } |
| 1627 | } |
| 1628 | } |
| 1629 | |
| 1630 | /// True when `$` followed by `next` starts a parameter expansion (`$v`, `$1`, |
| 1631 | /// `$@`, `$"…"`), as opposed to a literal dollar sign. |
| 1632 | fn starts_parameter(next: Option<&char>) -> bool { |
| 1633 | next.is_some_and(|c| { |
| 1634 | c.is_ascii_alphanumeric() |
| 1635 | || matches!(c, '_' | '@' | '*' | '#' | '?' | '$' | '!' | '-' | '"') |
| 1636 | }) |
| 1637 | } |
| 1638 | |
| 1639 | /// Unquoted heredocs expand substitutions, but quotes and ordinary lines are |
| 1640 | /// data. Also reports whether every substitution was read with certainty. |
| 1641 | fn heredoc_substitutions(body: &str) -> (Vec<String>, bool) { |
| 1642 | let chars: Vec<char> = body.chars().collect(); |
| 1643 | let mut result = Vec::new(); |
| 1644 | let mut all_certain = true; |
| 1645 | let mut i = 0; |
| 1646 | while i < chars.len() { |
| 1647 | match chars[i] { |
| 1648 | '\\' if chars |
| 1649 | .get(i + 1) |
| 1650 | .is_some_and(|c| matches!(c, '$' | '`' | '\\' | '\n')) => |
| 1651 | { |
| 1652 | i += 2 |
| 1653 | } |
| 1654 | '`' => { |
| 1655 | let (inner, next, certain) = read_backtick(&chars, i); |
| 1656 | all_certain &= certain; |
| 1657 | result.push(inner); |
| 1658 | i = next; |
| 1659 | } |
| 1660 | '$' if chars.get(i + 1) == Some(&'(') => { |
| 1661 | let (inner, next, certain) = read_substitution(&chars, i + 1); |
| 1662 | all_certain &= certain; |
| 1663 | result.push(inner); |
| 1664 | i = next; |
| 1665 | } |
| 1666 | _ => i += 1, |
| 1667 | } |
| 1668 | } |
| 1669 | (result, all_certain) |
| 1670 | } |
| 1671 | |
| 1672 | fn redirection_len(chars: &[char]) -> usize { |
| 1673 | match chars { |
| 1674 | ['&', '>', '>', ..] | ['<', '<', '<' | '-', ..] => 3, |
| 1675 | ['&', '>', ..] | ['<', '<' | '>' | '&', ..] | ['>', '>' | '&' | '|', ..] => 2, |
| 1676 | ['<' | '>', ..] => 1, |
| 1677 | _ => 0, |
| 1678 | } |
| 1679 | } |
| 1680 | |
| 1681 | fn is_redirect_descriptor(word: &str) -> bool { |
| 1682 | if !word.is_empty() && word.bytes().all(|b| b.is_ascii_digit()) { |
| 1683 | return true; |
| 1684 | } |
| 1685 | // Bash also accepts an unquoted `{name}` in place of an IO number. |
| 1686 | word.strip_prefix('{') |
| 1687 | .and_then(|word| word.strip_suffix('}')) |
| 1688 | .is_some_and(|name| { |
| 1689 | let mut chars = name.chars(); |
| 1690 | chars |
| 1691 | .next() |
| 1692 | .is_some_and(|c| c == '_' || c.is_ascii_alphabetic()) |
| 1693 | && chars.all(|c| c == '_' || c.is_ascii_alphanumeric()) |
| 1694 | }) |
| 1695 | } |
| 1696 | |
| 1697 | /// Close the current simple command. |
| 1698 | /// |
| 1699 | /// `{` and `}` stand alone as reserved words in `{ cmd; }` — they group a |
| 1700 | /// command rather than being part of one, so they split the line |
| 1701 | /// (`function f { rm x; }` runs `rm x`). Reserved words that lead a command |
| 1702 | /// (`if`, `then`, `while`, `do`, `!`, …) are dropped so the word after them is |
| 1703 | /// read as the command word. Every downstream consumer (wrapper detection, |
| 1704 | /// emission) then looks at real command words only. |
| 1705 | fn end_command(commands: &mut Vec<Line>, line: &mut Line) { |
| 1706 | let line = std::mem::take(line); |
| 1707 | let mut piece = Line { |
| 1708 | heredoc: line.heredoc, |
| 1709 | ..Line::default() |
| 1710 | }; |
| 1711 | for (word, dynamic) in line.words.into_iter().zip(line.dynamic) { |
| 1712 | if matches!(word.as_str(), "{" | "}") { |
| 1713 | push_command(commands, &mut piece); |
| 1714 | continue; |
| 1715 | } |
| 1716 | if piece.words.is_empty() && RESERVED_PREFIXES.contains(&word.as_str()) { |
| 1717 | continue; |
| 1718 | } |
| 1719 | piece.words.push(word); |
| 1720 | piece.dynamic.push(dynamic); |
| 1721 | } |
| 1722 | push_command(commands, &mut piece); |
| 1723 | } |
| 1724 | |
| 1725 | fn push_command(commands: &mut Vec<Line>, piece: &mut Line) { |
| 1726 | if !piece.words.is_empty() { |
| 1727 | let heredoc = piece.heredoc; |
| 1728 | commands.push(std::mem::take(piece)); |
| 1729 | piece.heredoc = heredoc; |
| 1730 | } |
| 1731 | } |
| 1732 | |
| 1733 | /// Read a backtick substitution. `start` indexes the opening backtick; returns |
| 1734 | /// the body, the index just past the closing backtick, and whether a closing |
| 1735 | /// backtick was found. |
| 1736 | fn read_backtick(chars: &[char], start: usize) -> (String, usize, bool) { |
| 1737 | let mut i = start + 1; |
| 1738 | let mut inner = String::new(); |
| 1739 | while i < chars.len() { |
| 1740 | match chars[i] { |
| 1741 | '\\' if i + 1 < chars.len() => { |
| 1742 | inner.push(chars[i]); |
| 1743 | inner.push(chars[i + 1]); |
| 1744 | i += 2; |
| 1745 | } |
| 1746 | '`' => return (inner, i + 1, true), |
| 1747 | c => { |
| 1748 | inner.push(c); |
| 1749 | i += 1; |
| 1750 | } |
| 1751 | } |
| 1752 | } |
| 1753 | (inner, i, false) |
| 1754 | } |
| 1755 | |
| 1756 | /// Read a `( … )` or `{ … }` region. `open_at` indexes the opening delimiter; |
| 1757 | /// returns the body, the index just past the matching close, and whether the |
| 1758 | /// region was read with certainty. |
| 1759 | /// |
| 1760 | /// Quoted text, escapes, backticks and nested `$(`/`${` are skipped the way |
| 1761 | /// the shell skips them, so `$(echo ")"; rm x)` closes at the last paren. A |
| 1762 | /// `case` inside a `$( … )` is flagged uncertain rather than parsed: its |
| 1763 | /// `pattern)` closes no paren, so the shell may end the region elsewhere. An |
| 1764 | /// unterminated region is uncertain too. |
| 1765 | fn read_substitution(chars: &[char], open_at: usize) -> (String, usize, bool) { |
| 1766 | let n = chars.len(); |
| 1767 | let (open, close) = if chars[open_at] == '{' { |
| 1768 | ('{', '}') |
| 1769 | } else { |
| 1770 | ('(', ')') |
| 1771 | }; |
| 1772 | let start = open_at + 1; |
| 1773 | let mut depth = 1usize; |
| 1774 | let mut certain = true; |
| 1775 | let mut i = start; |
| 1776 | while i < n { |
| 1777 | let c = chars[i]; |
| 1778 | match c { |
| 1779 | '\\' => i += 2, |
| 1780 | '\'' => { |
| 1781 | i += 1; |
| 1782 | while i < n && chars[i] != '\'' { |
| 1783 | i += 1; |
| 1784 | } |
| 1785 | certain &= i < n; |
| 1786 | i += 1; |
| 1787 | } |
| 1788 | '"' => { |
| 1789 | i += 1; |
| 1790 | while i < n && chars[i] != '"' { |
| 1791 | match chars[i] { |
| 1792 | '\\' => i += 2, |
| 1793 | '`' => { |
| 1794 | let (_, next, inner_certain) = read_backtick(chars, i); |
| 1795 | certain &= inner_certain; |
| 1796 | i = next; |
| 1797 | } |
| 1798 | '$' if matches!(chars.get(i + 1), Some('(' | '{')) => { |
| 1799 | let (_, next, inner_certain) = read_substitution(chars, i + 1); |
| 1800 | certain &= inner_certain; |
| 1801 | i = next; |
| 1802 | } |
| 1803 | _ => i += 1, |
| 1804 | } |
| 1805 | } |
| 1806 | certain &= i < n; |
| 1807 | i += 1; |
| 1808 | } |
| 1809 | '`' => { |
| 1810 | let (_, next, inner_certain) = read_backtick(chars, i); |
| 1811 | certain &= inner_certain; |
| 1812 | i = next; |
| 1813 | } |
| 1814 | '$' if matches!(chars.get(i + 1), Some('(' | '{')) => { |
| 1815 | let (_, next, inner_certain) = read_substitution(chars, i + 1); |
| 1816 | certain &= inner_certain; |
| 1817 | i = next; |
| 1818 | } |
| 1819 | _ if c == close => { |
| 1820 | depth -= 1; |
| 1821 | if depth == 0 { |
| 1822 | return (chars[start..i].iter().collect(), i + 1, certain); |
| 1823 | } |
| 1824 | i += 1; |
| 1825 | } |
| 1826 | _ => { |
| 1827 | if c == open { |
| 1828 | depth += 1; |
| 1829 | } else if open == '(' |
| 1830 | && c == 'c' |
| 1831 | && chars[i..].starts_with(&['c', 'a', 's', 'e']) |
| 1832 | && chars.get(i + 4).is_none_or(|next| next.is_whitespace()) |
| 1833 | && (i == start |
| 1834 | || matches!( |
| 1835 | chars[i - 1], |
| 1836 | ' ' | '\t' | '\n' | ';' | '&' | '|' | '(' | '!' |
| 1837 | )) |
| 1838 | { |
| 1839 | certain = false; |
| 1840 | } |
| 1841 | i += 1; |
| 1842 | } |
| 1843 | } |
| 1844 | } |
| 1845 | (chars[start..n.max(start)].iter().collect(), n, false) |
| 1846 | } |
| 1847 | |
| 1848 | /// The final path component, so `/usr/bin/sudo` reads as `sudo`. |
| 1849 | fn basename(token: &str) -> &str { |
| 1850 | token |
| 1851 | .rsplit(['/', '\\']) |
| 1852 | .next() |
| 1853 | .filter(|part| !part.is_empty()) |
| 1854 | .unwrap_or(token) |
| 1855 | } |
| 1856 | |
| 1857 | fn is_env_assignment(token: &str) -> bool { |
| 1858 | match token.split_once('=') { |
| 1859 | Some((name, _)) => { |
| 1860 | !name.is_empty() |
| 1861 | && !name.starts_with('-') |
| 1862 | && name |
| 1863 | .chars() |
| 1864 | .all(|ch| ch.is_ascii_alphanumeric() || ch == '_') |
| 1865 | } |
| 1866 | None => false, |
| 1867 | } |
| 1868 | } |
| 1869 | |
| 1870 | /// True for a bare scalar operand that may belong to a wrapper word rather |
| 1871 | /// than start a command — `timeout 5`, `nice -n 10`, `timeout 1.5s`. |
| 1872 | fn is_scalar_operand(token: &str) -> bool { |
| 1873 | let body = token.trim_end_matches(['s', 'm', 'h', 'd']); |
| 1874 | !body.is_empty() && body.chars().all(|ch| ch.is_ascii_digit() || ch == '.') |
| 1875 | } |
| 1876 | |
| 1877 | /// The command a word names, for comparing against a table of names: the |
| 1878 | /// lowercased basename with any `.exe` suffix removed, so `/usr/bin/sudo`, |
| 1879 | /// `bash.exe` and `C:\Git\bin\bash.EXE` read as `sudo` and `bash`. |
| 1880 | fn command_name(token: &str) -> String { |
| 1881 | let name = basename(token).to_ascii_lowercase(); |
| 1882 | match name.strip_suffix(".exe") { |
| 1883 | Some(stem) if !stem.is_empty() => stem.to_string(), |
| 1884 | _ => name, |
| 1885 | } |
| 1886 | } |
| 1887 | |
| 1888 | /// Index into [`WRAPPERS`] when `token` names a wrapper word. |
| 1889 | fn wrapper_index(token: &str) -> Option<usize> { |
| 1890 | let name = command_name(token); |
| 1891 | WRAPPERS.iter().position(|wrapper| wrapper.name == name) |
| 1892 | } |
| 1893 | |
| 1894 | /// What [`command_heads`] found. |
| 1895 | struct Heads { |
| 1896 | /// Every index that may be a command word, ascending. |
| 1897 | heads: Vec<usize>, |
| 1898 | /// `env -S` values, which are command lines. |
| 1899 | split_payloads: Vec<String>, |
| 1900 | /// The walk ran out of states before every reading was explored. |
| 1901 | truncated: bool, |
| 1902 | } |
| 1903 | |
| 1904 | /// Every index in `tokens` that may be a command word, walking leading |
| 1905 | /// environment assignments and wrapper words together with the options and |
| 1906 | /// operands those wrappers take. Wrapper words are themselves command words |
| 1907 | /// and are included. Also returns `env -S` payloads, which are command lines. |
| 1908 | /// |
| 1909 | /// Where the grammar is ambiguous (an option missing from the table that may |
| 1910 | /// take a value, a bare number after a wrapper) both readings are kept, |
| 1911 | /// because an extra candidate only makes deny matching stricter while a |
| 1912 | /// missing one is a bypass. |
| 1913 | fn command_heads(tokens: &[String]) -> Heads { |
| 1914 | #[derive(Clone, Copy, PartialEq, Eq, Hash)] |
| 1915 | enum State { |
| 1916 | /// At a command word position. |
| 1917 | Command, |
| 1918 | /// Inside the arguments of `WRAPPERS[wrapper]`. |
| 1919 | Wrapper { |
| 1920 | wrapper: usize, |
| 1921 | operands: u8, |
| 1922 | options_done: bool, |
| 1923 | }, |
| 1924 | } |
| 1925 | let mut heads = Vec::new(); |
| 1926 | let mut payloads = Vec::new(); |
| 1927 | let mut truncated = false; |
| 1928 | let mut seen = HashSet::new(); |
| 1929 | let mut stack = vec![(0usize, State::Command)]; |
| 1930 | while let Some((index, state)) = stack.pop() { |
| 1931 | if index >= tokens.len() || seen.contains(&(index, state)) { |
| 1932 | continue; |
| 1933 | } |
| 1934 | if seen.len() >= MAX_HEAD_STATES { |
| 1935 | truncated = true; |
| 1936 | break; |
| 1937 | } |
| 1938 | seen.insert((index, state)); |
| 1939 | let token = tokens[index].as_str(); |
| 1940 | match state { |
| 1941 | State::Command => { |
| 1942 | if is_env_assignment(token) { |
| 1943 | stack.push((index + 1, State::Command)); |
| 1944 | continue; |
| 1945 | } |
| 1946 | if !heads.contains(&index) { |
| 1947 | heads.push(index); |
| 1948 | } |
| 1949 | if let Some(position) = wrapper_index(token) { |
| 1950 | stack.push(( |
| 1951 | index + 1, |
| 1952 | State::Wrapper { |
| 1953 | wrapper: position, |
| 1954 | operands: WRAPPERS[position].operands, |
| 1955 | options_done: false, |
| 1956 | }, |
| 1957 | )); |
| 1958 | } |
| 1959 | } |
| 1960 | State::Wrapper { |
| 1961 | wrapper, |
| 1962 | operands, |
| 1963 | options_done, |
| 1964 | } => { |
| 1965 | let spec = &WRAPPERS[wrapper]; |
| 1966 | let next = |operands, options_done| State::Wrapper { |
| 1967 | wrapper, |
| 1968 | operands, |
| 1969 | options_done, |
| 1970 | }; |
| 1971 | if !options_done && token == "--" { |
| 1972 | stack.push((index + 1, next(operands, true))); |
| 1973 | } else if !options_done && token.len() > 1 && token.starts_with('-') { |
| 1974 | let same = next(operands, false); |
| 1975 | if let Some(long) = token.strip_prefix("--") { |
| 1976 | let (name, inline) = match long.split_once('=') { |
| 1977 | Some((name, value)) => (name, Some(value)), |
| 1978 | None => (long, None), |
| 1979 | }; |
| 1980 | if spec.name == "env" && name == "split-string" { |
| 1981 | match inline { |
| 1982 | Some(value) => payloads.push(value.to_string()), |
| 1983 | None => payloads.extend(tokens.get(index + 1).cloned()), |
| 1984 | } |
| 1985 | } |
| 1986 | if inline.is_some() || spec.long_switches.contains(&name) { |
| 1987 | stack.push((index + 1, same)); |
| 1988 | } else if spec.long_values.contains(&name) { |
| 1989 | stack.push((index + 2, same)); |
| 1990 | } else { |
| 1991 | // Unknown long option: it may or may not take |
| 1992 | // the next word as its value. |
| 1993 | stack.push((index + 1, same)); |
| 1994 | stack.push((index + 2, same)); |
| 1995 | } |
| 1996 | } else { |
| 1997 | let flags = &token[1..]; |
| 1998 | // `command -v NAME` / `-V` only looks NAME up. |
| 1999 | if spec.name == "command" && flags.contains(['v', 'V']) { |
| 2000 | continue; |
| 2001 | } |
| 2002 | if spec.name == "env" |
| 2003 | && let Some(at) = flags.find('S') |
| 2004 | { |
| 2005 | let value = &flags[at + 1..]; |
| 2006 | if value.is_empty() { |
| 2007 | payloads.extend(tokens.get(index + 1).cloned()); |
| 2008 | } else { |
| 2009 | payloads.push(value.to_string()); |
| 2010 | } |
| 2011 | } |
| 2012 | // Walk the cluster: a switch continues it, a value |
| 2013 | // option takes the rest of the word or the next word, |
| 2014 | // and an option missing from the table is read both |
| 2015 | // ways (`env -P DIR cmd`, `xargs -J % cmd %`). |
| 2016 | let mut step = Some(1); |
| 2017 | for (at, flag) in flags.char_indices() { |
| 2018 | if spec.short_values.contains(flag) { |
| 2019 | if at + flag.len_utf8() == flags.len() { |
| 2020 | step = Some(2); |
| 2021 | } |
| 2022 | break; |
| 2023 | } |
| 2024 | if !flag.is_ascii_digit() && !spec.short_switches.contains(flag) { |
| 2025 | step = None; |
| 2026 | break; |
| 2027 | } |
| 2028 | } |
| 2029 | match step { |
| 2030 | Some(step) => stack.push((index + step, same)), |
| 2031 | None => { |
| 2032 | stack.push((index + 1, same)); |
| 2033 | stack.push((index + 2, same)); |
| 2034 | } |
| 2035 | } |
| 2036 | } |
| 2037 | } else if operands > 0 { |
| 2038 | stack.push((index + 1, next(operands - 1, options_done))); |
| 2039 | } else { |
| 2040 | if is_scalar_operand(token) { |
| 2041 | stack.push((index + 1, next(0, options_done))); |
| 2042 | } |
| 2043 | stack.push((index, State::Command)); |
| 2044 | } |
| 2045 | } |
| 2046 | } |
| 2047 | } |
| 2048 | heads.sort_unstable(); |
| 2049 | Heads { |
| 2050 | heads, |
| 2051 | split_payloads: payloads, |
| 2052 | truncated, |
| 2053 | } |
| 2054 | } |
| 2055 | |
| 2056 | /// A script operand that names a file descriptor rather than a file: |
| 2057 | /// `/dev/stdin`, `/dev/fd/N` or `/proc/<pid>/fd/N`. Its text arrives through |
| 2058 | /// a pipe, a here-string or a redirect, so it is only known at run time — |
| 2059 | /// unless it is stdin and a heredoc body (already expanded as the script) |
| 2060 | /// feeds it. |
| 2061 | fn script_read_at_run_time(script: &str, line: &Line) -> bool { |
| 2062 | let mut parts: Vec<&str> = Vec::new(); |
| 2063 | for part in script.split('/') { |
| 2064 | match part { |
| 2065 | "" | "." => {} |
| 2066 | ".." => { |
| 2067 | parts.pop(); |
| 2068 | } |
| 2069 | part => parts.push(part), |
| 2070 | } |
| 2071 | } |
| 2072 | if !script.starts_with('/') { |
| 2073 | return false; |
| 2074 | } |
| 2075 | let descriptor = match parts.as_slice() { |
| 2076 | ["dev", "stdin"] => "0", |
| 2077 | ["dev", "fd", fd] => fd, |
| 2078 | ["proc", process, "fd", fd] |
| 2079 | if *process == "self" |
| 2080 | || *process == "thread-self" |
| 2081 | || process.chars().all(|ch| ch.is_ascii_digit()) => |
| 2082 | { |
| 2083 | fd |
| 2084 | } |
| 2085 | _ => return false, |
| 2086 | }; |
| 2087 | descriptor != "0" || !line.heredoc |
| 2088 | } |
| 2089 | |
| 2090 | /// How the shell invocation `tokens` (with `tokens[0]` the shell) gets its |
| 2091 | /// script. |
| 2092 | /// |
| 2093 | /// Combined short flags count (`bash -lc '…'`). The scan deliberately does |
| 2094 | /// NOT stop at the first non-flag operand: an earlier version did, and |
| 2095 | /// `bash -o vi -c 'payload'` walked straight past the deny expander because |
| 2096 | /// `vi` (the argument of `-o`) ended the scan before `-c` was seen |
| 2097 | /// (2026-08-04 review). Continuing the scan can over-read a `-c` that is |
| 2098 | /// really an argument to a script (`bash script.sh -c x`), but this |
| 2099 | /// expander's contract is explicit that over-emitting targets is safe and |
| 2100 | /// under-emitting is a bypass. |
| 2101 | /// |
| 2102 | /// `-c` only switches the shell into command mode: options may still follow |
| 2103 | /// it, and the command string is the first *operand* after option parsing |
| 2104 | /// (`bash -c -e 'cmd'`, `sh -c -- 'cmd'`, `bash -c -o pipefail 'cmd'` all run |
| 2105 | /// `cmd`). Shells that read `-c`'s value as the very next word (fish) run |
| 2106 | /// that word instead, so both candidates are reported when they differ. |
| 2107 | fn shell_input(tokens: &[String]) -> ShellInput { |
| 2108 | let mut script = None; |
| 2109 | let mut from_stdin = false; |
| 2110 | let mut options_done = false; |
| 2111 | let mut command = Vec::new(); |
| 2112 | let mut index = 1usize; |
| 2113 | while index < tokens.len() { |
| 2114 | let token = tokens[index].as_str(); |
| 2115 | if !options_done && token == "--" { |
| 2116 | options_done = true; |
| 2117 | } else if !options_done && token == "-" { |
| 2118 | // `bash -` reads the script from stdin; after `-c` it only ends |
| 2119 | // the options. |
| 2120 | from_stdin |= command.is_empty(); |
| 2121 | options_done = true; |
| 2122 | } else if let Some(long) = token.strip_prefix("--").filter(|_| !options_done) { |
| 2123 | if long.eq_ignore_ascii_case("command") { |
| 2124 | return match tokens.get(index + 1) { |
| 2125 | Some(_) => ShellInput::Command(vec![index + 1]), |
| 2126 | None => ShellInput::Stdin, |
| 2127 | }; |
| 2128 | } |
| 2129 | if matches!(long, "rcfile" | "init-file") { |
| 2130 | index += 1; |
| 2131 | } |
| 2132 | } else if !options_done |
| 2133 | && token.len() > 1 |
| 2134 | && (token.starts_with('-') || token.starts_with('+')) |
| 2135 | { |
| 2136 | let flags = &token[1..]; |
| 2137 | if token.starts_with('-') && flags.contains('c') && command.is_empty() { |
| 2138 | command.push(index + 1); |
| 2139 | } |
| 2140 | from_stdin |= token.starts_with('-') && flags.contains('s'); |
| 2141 | if flags.contains(['o', 'O']) { |
| 2142 | index += 1; |
| 2143 | } |
| 2144 | } else if !command.is_empty() { |
| 2145 | if command[0] != index { |
| 2146 | command.push(index); |
| 2147 | } |
| 2148 | return ShellInput::Command(command); |
| 2149 | } else if script.is_none() { |
| 2150 | script = Some(index); |
| 2151 | } |
| 2152 | index += 1; |
| 2153 | } |
| 2154 | command.retain(|&at| at < tokens.len()); |
| 2155 | match script { |
| 2156 | _ if !command.is_empty() => ShellInput::Command(command), |
| 2157 | Some(index) if !from_stdin => ShellInput::Script(index), |
| 2158 | _ => ShellInput::Stdin, |
| 2159 | } |
| 2160 | } |
| 2161 | |
| 2162 | #[cfg(test)] |
| 2163 | mod tests { |
| 2164 | use super::*; |
| 2165 | |
| 2166 | fn expand(command: &str) -> Vec<String> { |
| 2167 | // Exercise the POSIX grammar consistently on every test host. |
| 2168 | expanded_commands_for_platform(command, false) |
| 2169 | } |
| 2170 | |
| 2171 | #[test] |
| 2172 | fn windows_scan_retains_native_paths_and_posix_deny_candidates() { |
| 2173 | for (command, expected) in [ |
| 2174 | ( |
| 2175 | r"C:\Windows\System32\cat.exe ~/.ssh/id_rsa", |
| 2176 | r"C:\Windows\System32\cat.exe ~/.ssh/id_rsa", |
| 2177 | ), |
| 2178 | (r"del /f c:\users\x\file", r"del /f c:\users\x\file"), |
| 2179 | ( |
| 2180 | r"echo safe & xcopy /e /y c:\src d:\dst", |
| 2181 | r"xcopy /e /y c:\src d:\dst", |
| 2182 | ), |
| 2183 | ( |
| 2184 | r#""C:\Program Files\cat.exe" "c:\path with spaces\file""#, |
| 2185 | r"C:\Program Files\cat.exe c:\path with spaces\file", |
| 2186 | ), |
| 2187 | (r"del relative\file", r"del relative\file"), |
| 2188 | ( |
| 2189 | r"\\server\share\cat.exe file", |
| 2190 | r"\\server\share\cat.exe file", |
| 2191 | ), |
| 2192 | (r"bash -c 'rm -rf \/'", "rm -rf /"), |
| 2193 | ] { |
| 2194 | let targets = expanded_commands_for_platform(command, true); |
| 2195 | assert!( |
| 2196 | targets.iter().any(|target| target == expected), |
| 2197 | "missing {expected:?} from {targets:?}" |
| 2198 | ); |
| 2199 | assert!(targets.len() <= MAX_COMMANDS); |
| 2200 | } |
| 2201 | let targets = |
| 2202 | expanded_commands_for_platform("cat <<'EOF'\ndel c:\\users\\x\\file\nEOF", true); |
| 2203 | assert!( |
| 2204 | !targets.iter().any(|target| target.starts_with("del ")), |
| 2205 | "literal heredoc data must stay inert: {targets:?}" |
| 2206 | ); |
| 2207 | } |
| 2208 | |
| 2209 | fn contains(command: &str, expected: &str) -> bool { |
| 2210 | expand(command).iter().any(|target| target == expected) |
| 2211 | } |
| 2212 | |
| 2213 | #[test] |
| 2214 | fn backtick_body_is_a_command() { |
| 2215 | assert!(contains("`rm -rf /`", "rm -rf /")); |
| 2216 | assert!(contains("echo `rm -rf /`", "rm -rf /")); |
| 2217 | assert!(contains("echo `rm -rf /`", "echo")); |
| 2218 | } |
| 2219 | |
| 2220 | #[test] |
| 2221 | fn dollar_paren_body_is_a_command() { |
| 2222 | assert!(contains("echo $(rm -rf /)", "rm -rf /")); |
| 2223 | assert!(contains("x=$(rm -rf /)", "rm -rf /")); |
| 2224 | assert!(contains("echo \"$(rm -rf /)\"", "rm -rf /")); |
| 2225 | } |
| 2226 | |
| 2227 | #[test] |
| 2228 | fn nested_substitution_is_followed() { |
| 2229 | assert!(contains("echo $(echo `rm -rf /`)", "rm -rf /")); |
| 2230 | } |
| 2231 | |
| 2232 | #[test] |
| 2233 | fn quotes_are_removed_from_operands() { |
| 2234 | assert!(contains("rm -rf \"/\"", "rm -rf /")); |
| 2235 | assert!(contains("rm -rf '/'", "rm -rf /")); |
| 2236 | assert!(contains("\"rm\" -rf /", "rm -rf /")); |
| 2237 | assert!(contains("rm -r\"f\" /", "rm -rf /")); |
| 2238 | } |
| 2239 | |
| 2240 | #[test] |
| 2241 | fn single_quoted_text_is_not_a_command() { |
| 2242 | // A literal backtick inside single quotes is printed, not executed. |
| 2243 | let targets = expand("echo '`rm -rf /`'"); |
| 2244 | assert!( |
| 2245 | !targets.iter().any(|t| t == "rm -rf /"), |
| 2246 | "single-quoted text must not become a command: {targets:?}" |
| 2247 | ); |
| 2248 | } |
| 2249 | |
| 2250 | #[test] |
| 2251 | fn escaped_operators_do_not_split() { |
| 2252 | let targets = expand("echo a\\;b"); |
| 2253 | assert_eq!(targets, vec!["echo a;b".to_string()]); |
| 2254 | assert!(targets.contains(&"echo a;b".to_string()), "{targets:?}"); |
| 2255 | } |
| 2256 | |
| 2257 | #[test] |
| 2258 | fn control_operators_split_commands() { |
| 2259 | for command in [ |
| 2260 | "ls && rm -rf /", |
| 2261 | "ls || rm -rf /", |
| 2262 | "ls ; rm -rf /", |
| 2263 | "ls | rm -rf /", |
| 2264 | "ls & rm -rf /", |
| 2265 | "ls\nrm -rf /", |
| 2266 | ] { |
| 2267 | assert!(contains(command, "rm -rf /"), "{command}"); |
| 2268 | } |
| 2269 | } |
| 2270 | |
| 2271 | #[test] |
| 2272 | fn wrappers_and_payloads_are_unwrapped() { |
| 2273 | for command in [ |
| 2274 | "sudo rm -rf /", |
| 2275 | "env rm -rf /", |
| 2276 | "timeout 5 rm -rf /", |
| 2277 | "nohup rm -rf /", |
| 2278 | "xargs rm -rf /", |
| 2279 | "/usr/bin/sudo rm -rf /", |
| 2280 | "eval 'rm -rf /'", |
| 2281 | "bash -c 'rm -rf /'", |
| 2282 | "sh -lc \"rm -rf /\"", |
| 2283 | "sudo -u root bash -c 'rm -rf /'", |
| 2284 | // 2026-08-04: `-o vi` used to end the flag scan before `-c` was |
| 2285 | // seen, so the payload skipped deny expansion entirely. |
| 2286 | "bash -o vi -c 'rm -rf /'", |
| 2287 | "zsh --norcs -c 'rm -rf /'", |
| 2288 | ] { |
| 2289 | assert!( |
| 2290 | contains(command, "rm -rf /"), |
| 2291 | "{command}: {:?}", |
| 2292 | expand(command) |
| 2293 | ); |
| 2294 | } |
| 2295 | } |
| 2296 | |
| 2297 | #[test] |
| 2298 | fn options_after_dash_c_do_not_hide_the_command_string() { |
| 2299 | // The command string is the first operand once options end, so an |
| 2300 | // option between `-c` and it does not change what runs. |
| 2301 | for command in [ |
| 2302 | "bash -c -e 'rm -rf /'", |
| 2303 | "bash -c -l 'rm -rf /'", |
| 2304 | "sh -c -x 'rm -rf /'", |
| 2305 | "bash -c -- 'rm -rf /'", |
| 2306 | "sh -c - 'rm -rf /'", |
| 2307 | "bash -c +e 'rm -rf /'", |
| 2308 | "bash -lc -e 'rm -rf /'", |
| 2309 | "bash -c -o pipefail 'rm -rf /'", |
| 2310 | "bash -c -O extglob -e 'rm -rf /'", |
| 2311 | "bash script.sh -c 'rm -rf /'", |
| 2312 | ] { |
| 2313 | assert!( |
| 2314 | contains(command, "rm -rf /"), |
| 2315 | "{command}: {:?}", |
| 2316 | expand(command) |
| 2317 | ); |
| 2318 | } |
| 2319 | // Arguments after the command string are positional parameters. |
| 2320 | assert!(!contains("bash -c 'echo $0' 'rm -rf /'", "rm -rf /")); |
| 2321 | } |
| 2322 | |
| 2323 | #[test] |
| 2324 | fn script_operand_naming_a_descriptor_is_read_at_run_time() { |
| 2325 | for command in [ |
| 2326 | "echo 'rm -rf /' | bash /dev/stdin", |
| 2327 | "bash /dev/stdin <<< 'rm -rf /'", |
| 2328 | "sh /dev/fd/0 <<< 'rm -rf /'", |
| 2329 | "sh /proc/self/fd/0 <<< 'rm -rf /'", |
| 2330 | "bash //dev/./stdin <<< 'rm -rf /'", |
| 2331 | "bash /dev/fd/3 3< script", |
| 2332 | ". /dev/stdin <<< 'rm -rf /'", |
| 2333 | "source /proc/self/fd/0 <<< 'rm -rf /'", |
| 2334 | ] { |
| 2335 | assert!(expand_command(command).dynamic, "{command}"); |
| 2336 | } |
| 2337 | // A heredoc on stdin was expanded as the script itself. |
| 2338 | let heredoc = expand_for_platform("bash /dev/stdin <<EOF\nrm -rf /\nEOF", false); |
| 2339 | assert!(!heredoc.dynamic); |
| 2340 | assert!(heredoc.commands.iter().any(|target| target == "rm -rf /")); |
| 2341 | // An ordinary script file is still opaque, not unresolved. |
| 2342 | assert!(!expand_command("bash ./dev/stdin").dynamic); |
| 2343 | assert!(!expand_command(". ./env.sh").dynamic); |
| 2344 | } |
| 2345 | |
| 2346 | #[test] |
| 2347 | fn wrapper_head_scan_stops_at_a_real_command() { |
| 2348 | // `echo` prints its arguments; nothing here is executed as a shell. |
| 2349 | let targets = expand("echo bash -c 'rm -rf /'"); |
| 2350 | assert!( |
| 2351 | !targets.iter().any(|t| t == "rm -rf /"), |
| 2352 | "arguments of a printing command must not be parsed as code: {targets:?}" |
| 2353 | ); |
| 2354 | } |
| 2355 | |
| 2356 | #[test] |
| 2357 | fn process_and_parameter_substitution_bodies_are_commands() { |
| 2358 | assert!(contains("diff <(rm -rf /) b", "rm -rf /")); |
| 2359 | assert!(contains("echo ${x:-$(rm -rf /)}", "rm -rf /")); |
| 2360 | } |
| 2361 | |
| 2362 | #[test] |
| 2363 | fn expansion_is_bounded() { |
| 2364 | let deep = "$(".repeat(64) + "rm -rf /" + &")".repeat(64); |
| 2365 | let targets = expand(&deep); |
| 2366 | assert!(targets.len() <= MAX_COMMANDS); |
| 2367 | } |
| 2368 | |
| 2369 | #[test] |
| 2370 | fn grouping_is_a_command_boundary() { |
| 2371 | assert!(contains("(rm -rf /)", "rm -rf /")); |
| 2372 | assert!(contains("{ rm -rf /; }", "rm -rf /")); |
| 2373 | assert!(contains("(cd /tmp && rm -rf /)", "rm -rf /")); |
| 2374 | // Escaped and quoted parens are operands, not grouping. |
| 2375 | assert!(contains( |
| 2376 | "find . \\( -name a \\) -print", |
| 2377 | "find . ( -name a ) -print" |
| 2378 | )); |
| 2379 | } |
| 2380 | |
| 2381 | #[test] |
| 2382 | fn syntax_metadata_is_independent_of_candidates_and_literal_data() { |
| 2383 | for command in [ |
| 2384 | "printf x; printf x", |
| 2385 | "git log | git log", |
| 2386 | "cargo test && cargo test", |
| 2387 | "(git log)", |
| 2388 | ] { |
| 2389 | let expansion = expand_for_platform(command, false); |
| 2390 | assert!(expansion.control, "{command}: {expansion:?}"); |
| 2391 | } |
| 2392 | assert_eq!( |
| 2393 | expand_for_platform("printf x; printf x", false) |
| 2394 | .commands |
| 2395 | .len(), |
| 2396 | 1 |
| 2397 | ); |
| 2398 | assert!(expand_for_platform("sudo git log", false).commands.len() > 1); |
| 2399 | assert!(!expand_for_platform("sudo git log", false).control); |
| 2400 | for command in [ |
| 2401 | r#"grep -E "a|b" src"#, |
| 2402 | r#"git commit -m "fix: a & b; c""#, |
| 2403 | r"echo a\;b\&c\|d\>e", |
| 2404 | "echo '> < ; | &'", |
| 2405 | "echo ok # ; | > ignored", |
| 2406 | ] { |
| 2407 | let expansion = expand_for_platform(command, false); |
| 2408 | assert!( |
| 2409 | !expansion.control && !expansion.redirects, |
| 2410 | "{command}: {expansion:?}" |
| 2411 | ); |
| 2412 | } |
| 2413 | for command in [ |
| 2414 | "cargo test 2>&1", |
| 2415 | "cargo test &>result.log", |
| 2416 | "cat a<>out", |
| 2417 | "cat a>|out", |
| 2418 | "cat 3<&0", |
| 2419 | "cat<<<literal", |
| 2420 | ] { |
| 2421 | for windows in [false, true] { |
| 2422 | let expansion = expand_for_platform(command, windows); |
| 2423 | assert!( |
| 2424 | expansion.redirects && !expansion.control, |
| 2425 | "{command}: {expansion:?}" |
| 2426 | ); |
| 2427 | } |
| 2428 | } |
| 2429 | for command in ["git log $FLAGS", r#"git log "$FLAGS""#] { |
| 2430 | let expansion = expand_for_platform(command, false); |
| 2431 | assert!( |
| 2432 | expansion.arguments_dynamic && !expansion.dynamic, |
| 2433 | "{command}: {expansion:?}" |
| 2434 | ); |
| 2435 | } |
| 2436 | assert!(!expand_for_platform("git log '$FLAGS'", false).arguments_dynamic); |
| 2437 | assert!(!expand_for_platform(r"git log \$FLAGS", false).arguments_dynamic); |
| 2438 | let heredoc = expand_for_platform("cat <<'EOF'\n; | > $(not-code)\nEOF", false); |
| 2439 | assert!(heredoc.redirects); |
| 2440 | assert!(!heredoc.nested); |
| 2441 | assert_eq!(heredoc.commands, vec!["cat"]); |
| 2442 | } |
| 2443 | |
| 2444 | #[test] |
| 2445 | fn plain_command_expands_to_itself() { |
| 2446 | assert_eq!(expand("git status -s"), vec!["git status -s".to_string()]); |
| 2447 | } |
| 2448 | } |
| 2449 |