| 1 | """executor.py."""
|
| 2 | from __future__ import print_function
|
| 3 |
|
| 4 | from errno import EINTR
|
| 5 |
|
| 6 | from _devbuild.gen.id_kind_asdl import Id
|
| 7 | from _devbuild.gen.option_asdl import builtin_i
|
| 8 | from _devbuild.gen.runtime_asdl import RedirValue, trace
|
| 9 | from _devbuild.gen.syntax_asdl import (
|
| 10 | command,
|
| 11 | command_e,
|
| 12 | CommandSub,
|
| 13 | CompoundWord,
|
| 14 | loc,
|
| 15 | loc_t,
|
| 16 | )
|
| 17 | from _devbuild.gen.value_asdl import value, value_e
|
| 18 | from builtin import hay_ysh
|
| 19 | from core import dev
|
| 20 | from core import error
|
| 21 | from core import process
|
| 22 | from core.error import e_die, e_die_status
|
| 23 | from core import pyos
|
| 24 | from core import pyutil
|
| 25 | from core import state
|
| 26 | from core import vm
|
| 27 | from display import ui
|
| 28 | from frontend import consts
|
| 29 | from frontend import lexer
|
| 30 | from mycpp import mylib
|
| 31 | from mycpp.mylib import log, print_stderr, tagswitch
|
| 32 | from pylib import os_path
|
| 33 | from pylib import path_stat
|
| 34 |
|
| 35 | import posix_ as posix
|
| 36 | from posix_ import X_OK # translated directly to C macro
|
| 37 |
|
| 38 | from typing import cast, Dict, List, Tuple, Optional, TYPE_CHECKING
|
| 39 | if TYPE_CHECKING:
|
| 40 | from _devbuild.gen.runtime_asdl import (cmd_value, CommandStatus,
|
| 41 | StatusArray)
|
| 42 | from _devbuild.gen.syntax_asdl import command_t
|
| 43 | from builtin import trap_osh
|
| 44 | from core import optview
|
| 45 | from core import state
|
| 46 |
|
| 47 | _ = log
|
| 48 |
|
| 49 |
|
| 50 | def LookupExecutable(name, path_dirs, exec_required=True):
|
| 51 | # type: (str, List[str], bool) -> Optional[str]
|
| 52 | """
|
| 53 | Returns either
|
| 54 | - the name if it's a relative path that exists
|
| 55 | - the executable name resolved against path_dirs
|
| 56 | - None if not found
|
| 57 | """
|
| 58 | if len(name) == 0: # special case for "$(true)"
|
| 59 | return None
|
| 60 |
|
| 61 | if '/' in name:
|
| 62 | return name if path_stat.exists(name) else None
|
| 63 |
|
| 64 | for path_dir in path_dirs:
|
| 65 | full_path = os_path.join(path_dir, name)
|
| 66 | if exec_required:
|
| 67 | found = posix.access(full_path, X_OK)
|
| 68 | else:
|
| 69 | found = path_stat.exists(full_path)
|
| 70 |
|
| 71 | if found:
|
| 72 | return full_path
|
| 73 |
|
| 74 | return None
|
| 75 |
|
| 76 |
|
| 77 | class SearchPath(object):
|
| 78 | """For looking up files in $PATH or ENV.PATH"""
|
| 79 |
|
| 80 | def __init__(self, mem, exec_opts):
|
| 81 | # type: (state.Mem, optview.Exec) -> None
|
| 82 | self.mem = mem
|
| 83 | # TODO: remove exec_opts
|
| 84 | self.cache = {} # type: Dict[str, str]
|
| 85 |
|
| 86 | def _GetPath(self):
|
| 87 | # type: () -> List[str]
|
| 88 |
|
| 89 | # In YSH, we read from ENV.PATH
|
| 90 | s = state.GetStringFromEnv(self.mem, 'PATH')
|
| 91 | if s is None:
|
| 92 | return [] # treat as empty path
|
| 93 |
|
| 94 | # TODO: Could cache this to avoid split() allocating all the time.
|
| 95 | return s.split(':')
|
| 96 |
|
| 97 | def LookupOne(self, name, exec_required=True):
|
| 98 | # type: (str, bool) -> Optional[str]
|
| 99 | """
|
| 100 | Returns the path itself (if relative path), the resolved path, or None.
|
| 101 | """
|
| 102 | return LookupExecutable(name,
|
| 103 | self._GetPath(),
|
| 104 | exec_required=exec_required)
|
| 105 |
|
| 106 | def LookupReflect(self, name, do_all):
|
| 107 | # type: (str, bool) -> List[str]
|
| 108 | """
|
| 109 | Like LookupOne(), with an option for 'type -a' to return all paths.
|
| 110 | """
|
| 111 | if len(name) == 0: # special case for "$(true)"
|
| 112 | return []
|
| 113 |
|
| 114 | if '/' in name:
|
| 115 | if path_stat.exists(name):
|
| 116 | return [name]
|
| 117 | else:
|
| 118 | return []
|
| 119 |
|
| 120 | results = [] # type: List[str]
|
| 121 | for path_dir in self._GetPath():
|
| 122 | full_path = os_path.join(path_dir, name)
|
| 123 | if path_stat.exists(full_path):
|
| 124 | results.append(full_path)
|
| 125 | if not do_all:
|
| 126 | return results
|
| 127 |
|
| 128 | return results
|
| 129 |
|
| 130 | def CachedLookup(self, name):
|
| 131 | # type: (str) -> Optional[str]
|
| 132 | #log('name %r', name)
|
| 133 | if name in self.cache:
|
| 134 | return self.cache[name]
|
| 135 |
|
| 136 | full_path = self.LookupOne(name)
|
| 137 | if full_path is not None:
|
| 138 | self.cache[name] = full_path
|
| 139 | return full_path
|
| 140 |
|
| 141 | def MaybeRemoveEntry(self, name):
|
| 142 | # type: (str) -> None
|
| 143 | """When the file system changes."""
|
| 144 | mylib.dict_erase(self.cache, name)
|
| 145 |
|
| 146 | def ClearCache(self):
|
| 147 | # type: () -> None
|
| 148 | """For hash -r."""
|
| 149 | self.cache.clear()
|
| 150 |
|
| 151 | def CachedCommands(self):
|
| 152 | # type: () -> List[str]
|
| 153 | return self.cache.values()
|
| 154 |
|
| 155 |
|
| 156 | class _ProcessSubFrame(object):
|
| 157 | """To keep track of diff <(cat 1) <(cat 2) > >(tac)"""
|
| 158 |
|
| 159 | def __init__(self):
|
| 160 | # type: () -> None
|
| 161 |
|
| 162 | # These objects appear unconditionally in the main loop, and aren't
|
| 163 | # commonly used, so we manually optimize [] into None.
|
| 164 |
|
| 165 | self._to_wait = [] # type: List[process.Process]
|
| 166 | self._to_close = [] # type: List[int] # file descriptors
|
| 167 | self._locs = [] # type: List[loc_t]
|
| 168 | self._modified = False
|
| 169 |
|
| 170 | def WasModified(self):
|
| 171 | # type: () -> bool
|
| 172 | return self._modified
|
| 173 |
|
| 174 | def Append(self, p, fd, status_loc):
|
| 175 | # type: (process.Process, int, loc_t) -> None
|
| 176 | self._modified = True
|
| 177 |
|
| 178 | self._to_wait.append(p)
|
| 179 | self._to_close.append(fd)
|
| 180 | self._locs.append(status_loc)
|
| 181 |
|
| 182 | def MaybeWaitOnProcessSubs(self, waiter, status_array):
|
| 183 | # type: (process.Waiter, StatusArray) -> None
|
| 184 |
|
| 185 | # Wait in the same order that they were evaluated. That seems fine.
|
| 186 | for fd in self._to_close:
|
| 187 | posix.close(fd)
|
| 188 |
|
| 189 | codes = [] # type: List[int]
|
| 190 | locs = [] # type: List[loc_t]
|
| 191 | for i, p in enumerate(self._to_wait):
|
| 192 | #log('waiting for %s', p)
|
| 193 | st = p.Wait(waiter)
|
| 194 | codes.append(st)
|
| 195 | locs.append(self._locs[i])
|
| 196 |
|
| 197 | status_array.codes = codes
|
| 198 | status_array.locs = locs
|
| 199 |
|
| 200 |
|
| 201 | # Big flags for RunSimpleCommand
|
| 202 | IS_LAST_CMD = 1 << 1
|
| 203 | NO_CALL_PROCS = 1 << 2 # command ls suppresses function lookup
|
| 204 | USE_DEFAULT_PATH = 1 << 3 # for command -p ls changes the path
|
| 205 |
|
| 206 | # Copied from var.c in dash
|
| 207 | DEFAULT_PATH = [
|
| 208 | '/usr/local/sbin', '/usr/local/bin', '/usr/sbin', '/usr/bin', '/sbin',
|
| 209 | '/bin'
|
| 210 | ]
|
| 211 |
|
| 212 |
|
| 213 | class ShellExecutor(vm._Executor):
|
| 214 | """An executor combined with the OSH language evaluators in osh/ to create
|
| 215 | a shell interpreter."""
|
| 216 |
|
| 217 | def __init__(
|
| 218 | self,
|
| 219 | mem, # type: state.Mem
|
| 220 | exec_opts, # type: optview.Exec
|
| 221 | mutable_opts, # type: state.MutableOpts
|
| 222 | procs, # type: state.Procs
|
| 223 | hay_state, # type: hay_ysh.HayState
|
| 224 | builtins, # type: Dict[int, vm._Builtin]
|
| 225 | search_path, # type: SearchPath
|
| 226 | ext_prog, # type: process.ExternalProgram
|
| 227 | waiter, # type: process.Waiter
|
| 228 | tracer, # type: dev.Tracer
|
| 229 | job_control, # type: process.JobControl
|
| 230 | job_list, # type: process.JobList
|
| 231 | fd_state, # type: process.FdState
|
| 232 | trap_state, # type: trap_osh.TrapState
|
| 233 | errfmt # type: ui.ErrorFormatter
|
| 234 | ):
|
| 235 | # type: (...) -> None
|
| 236 | vm._Executor.__init__(self)
|
| 237 | self.mem = mem
|
| 238 | self.exec_opts = exec_opts
|
| 239 | self.mutable_opts = mutable_opts # for IsDisabled(), not mutating
|
| 240 | self.procs = procs
|
| 241 | self.hay_state = hay_state
|
| 242 | self.builtins = builtins
|
| 243 | self.search_path = search_path
|
| 244 | self.ext_prog = ext_prog
|
| 245 | self.waiter = waiter
|
| 246 | self.tracer = tracer
|
| 247 | self.multi_trace = tracer.multi_trace
|
| 248 | self.job_control = job_control
|
| 249 | # sleep 5 & puts a (PID, job#) entry here. And then "jobs" displays it.
|
| 250 | self.job_list = job_list
|
| 251 | self.fd_state = fd_state
|
| 252 | self.trap_state = trap_state
|
| 253 | self.errfmt = errfmt
|
| 254 | self.process_sub_stack = [] # type: List[_ProcessSubFrame]
|
| 255 | self.clean_frame_pool = [] # type: List[_ProcessSubFrame]
|
| 256 |
|
| 257 | # When starting a pipeline in the foreground, we need to pass a handle to it
|
| 258 | # through the evaluation of the last node back to ourselves for execution.
|
| 259 | # We use this handle to make sure any processes forked for the last part of
|
| 260 | # the pipeline are placed into the same process group as the rest of the
|
| 261 | # pipeline. Since there is, by design, only ever one foreground pipeline and
|
| 262 | # any pipelines started within subshells run in their parent's process
|
| 263 | # group, we only need one pointer here, not some collection.
|
| 264 | self.fg_pipeline = None # type: Optional[process.Pipeline]
|
| 265 |
|
| 266 | def CheckCircularDeps(self):
|
| 267 | # type: () -> None
|
| 268 | assert self.cmd_ev is not None
|
| 269 |
|
| 270 | def _MakeProcess(self, node, inherit_errexit, inherit_errtrace):
|
| 271 | # type: (command_t, bool, bool) -> process.Process
|
| 272 | """Assume we will run the node in another process.
|
| 273 |
|
| 274 | Return a process.
|
| 275 | """
|
| 276 | UP_node = node
|
| 277 | if node.tag() == command_e.ControlFlow:
|
| 278 | node = cast(command.ControlFlow, UP_node)
|
| 279 | # Pipeline or subshells with control flow are invalid, e.g.:
|
| 280 | # - break | less
|
| 281 | # - continue | less
|
| 282 | # - ( return )
|
| 283 | # NOTE: This could be done at parse time too.
|
| 284 | if node.keyword.id != Id.ControlFlow_Exit:
|
| 285 | e_die(
|
| 286 | 'Invalid control flow %r in pipeline / subshell / background'
|
| 287 | % lexer.TokenVal(node.keyword), node.keyword)
|
| 288 |
|
| 289 | # NOTE: If ErrExit(), we could be verbose about subprogram errors? This
|
| 290 | # only really matters when executing 'exit 42', because the child shell
|
| 291 | # inherits errexit and will be verbose. Other notes:
|
| 292 | #
|
| 293 | # - We might want errors to fit on a single line so they don't get #
|
| 294 | # interleaved.
|
| 295 | # - We could turn the `exit` builtin into a error.FatalRuntime exception
|
| 296 | # and get this check for "free".
|
| 297 | thunk = process.SubProgramThunk(self.cmd_ev, node, self.trap_state,
|
| 298 | self.multi_trace, inherit_errexit,
|
| 299 | inherit_errtrace)
|
| 300 | p = process.Process(thunk, self.job_control, self.job_list,
|
| 301 | self.tracer)
|
| 302 | return p
|
| 303 |
|
| 304 | def RunBuiltin(self, builtin_id, cmd_val):
|
| 305 | # type: (int, cmd_value.Argv) -> int
|
| 306 | """Run a builtin.
|
| 307 |
|
| 308 | Also called by the 'builtin' builtin.
|
| 309 | """
|
| 310 | self.tracer.OnBuiltin(builtin_id, cmd_val.argv)
|
| 311 |
|
| 312 | builtin_proc = self.builtins[builtin_id]
|
| 313 |
|
| 314 | return self.RunBuiltinProc(builtin_proc, cmd_val)
|
| 315 |
|
| 316 | def RunBuiltinProc(self, builtin_proc, cmd_val):
|
| 317 | # type: (vm._Builtin, cmd_value.Argv) -> int
|
| 318 |
|
| 319 | io_errors = [] # type: List[error.IOError_OSError]
|
| 320 | with vm.ctx_FlushStdout(io_errors):
|
| 321 | # note: could be second word, like 'builtin read'
|
| 322 | with ui.ctx_Location(self.errfmt, cmd_val.arg_locs[0]):
|
| 323 | try:
|
| 324 | status = builtin_proc.Run(cmd_val)
|
| 325 | assert isinstance(status, int)
|
| 326 | except (IOError, OSError) as e:
|
| 327 | self.errfmt.PrintMessage(
|
| 328 | '%s builtin I/O error: %s' %
|
| 329 | (cmd_val.argv[0], pyutil.strerror(e)),
|
| 330 | cmd_val.arg_locs[0])
|
| 331 | return 1
|
| 332 | except error.Usage as e:
|
| 333 | arg0 = cmd_val.argv[0]
|
| 334 | # e.g. 'type' doesn't accept flag '-x'
|
| 335 | self.errfmt.PrefixPrint(e.msg, '%r ' % arg0, e.location)
|
| 336 | return 2 # consistent error code for usage error
|
| 337 |
|
| 338 | if len(io_errors): # e.g. disk full, ulimit
|
| 339 | self.errfmt.PrintMessage(
|
| 340 | '%s builtin I/O error: %s' %
|
| 341 | (cmd_val.argv[0], pyutil.strerror(io_errors[0])),
|
| 342 | cmd_val.arg_locs[0])
|
| 343 | return 1
|
| 344 |
|
| 345 | return status
|
| 346 |
|
| 347 | def RunSimpleCommand(self, cmd_val, cmd_st, run_flags):
|
| 348 | # type: (cmd_value.Argv, CommandStatus, int) -> int
|
| 349 | """Run builtins, functions, external commands.
|
| 350 |
|
| 351 | Possible variations:
|
| 352 | - YSH might have different, simpler rules. No special builtins, etc.
|
| 353 | - YSH might have OILS_PATH = :| /bin /usr/bin | or something.
|
| 354 | - Interpreters might want to define all their own builtins.
|
| 355 | """
|
| 356 | argv = cmd_val.argv
|
| 357 | if len(cmd_val.arg_locs):
|
| 358 | arg0_loc = cmd_val.arg_locs[0] # type: loc_t
|
| 359 | else:
|
| 360 | arg0_loc = loc.Missing
|
| 361 |
|
| 362 | # This happens when you write "$@" but have no arguments.
|
| 363 | if len(argv) == 0:
|
| 364 | if self.exec_opts.strict_argv():
|
| 365 | e_die("Command evaluated to an empty argv array", arg0_loc)
|
| 366 | else:
|
| 367 | return 0 # status 0, or skip it?
|
| 368 |
|
| 369 | arg0 = argv[0]
|
| 370 |
|
| 371 | builtin_id = consts.LookupAssignBuiltin(arg0)
|
| 372 | if builtin_id != consts.NO_INDEX:
|
| 373 | # command readonly is disallowed, for technical reasons. Could relax it
|
| 374 | # later.
|
| 375 | self.errfmt.Print_("Can't run assignment builtin recursively",
|
| 376 | arg0_loc)
|
| 377 | return 1
|
| 378 |
|
| 379 | builtin_id = consts.LookupSpecialBuiltin(arg0)
|
| 380 | if builtin_id != consts.NO_INDEX:
|
| 381 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 382 | status = self.RunBuiltin(builtin_id, cmd_val)
|
| 383 | # TODO: Enable this and fix spec test failures.
|
| 384 | # Also update _SPECIAL_BUILTINS in osh/builtin.py.
|
| 385 | #if status != 0:
|
| 386 | # e_die_status(status, 'special builtin failed')
|
| 387 | return status
|
| 388 |
|
| 389 | # Builtins like 'true' can be redefined as functions.
|
| 390 | call_procs = not (run_flags & NO_CALL_PROCS)
|
| 391 | if call_procs:
|
| 392 | proc_val, self_obj = self.procs.GetInvokable(arg0)
|
| 393 | cmd_val.self_obj = self_obj # MAYBE bind self
|
| 394 |
|
| 395 | if proc_val is not None:
|
| 396 | if self.exec_opts.strict_errexit():
|
| 397 | disabled_tok = self.mutable_opts.ErrExitDisabledToken()
|
| 398 | if disabled_tok:
|
| 399 | self.errfmt.Print_(
|
| 400 | 'errexit was disabled for this construct',
|
| 401 | disabled_tok)
|
| 402 | self.errfmt.StderrLine('')
|
| 403 | e_die(
|
| 404 | "Can't run a proc while errexit is disabled. "
|
| 405 | "Use 'try' or wrap it in a process with $0 myproc",
|
| 406 | arg0_loc)
|
| 407 |
|
| 408 | with tagswitch(proc_val) as case:
|
| 409 | if case(value_e.BuiltinProc):
|
| 410 | # Handle the special case of the BUILTIN proc
|
| 411 | # module_ysh.ModuleInvoke, which is returned on the Obj
|
| 412 | # created by 'use util.ysh'
|
| 413 | builtin_proc = cast(value.BuiltinProc, proc_val)
|
| 414 | b = cast(vm._Builtin, builtin_proc.builtin)
|
| 415 | status = self.RunBuiltinProc(b, cmd_val)
|
| 416 |
|
| 417 | elif case(value_e.Proc):
|
| 418 | proc = cast(value.Proc, proc_val)
|
| 419 | with dev.ctx_Tracer(self.tracer, 'proc', argv):
|
| 420 | # NOTE: Functions could call 'exit 42' directly, etc.
|
| 421 | status = self.cmd_ev.RunProc(proc, cmd_val)
|
| 422 |
|
| 423 | else:
|
| 424 | # GetInvokable() should only return 1 of 2 things
|
| 425 | raise AssertionError()
|
| 426 |
|
| 427 | return status
|
| 428 |
|
| 429 | # Notes:
|
| 430 | # - procs shadow hay names
|
| 431 | # - hay names shadow normal builtins? Should we limit to CAPS or no?
|
| 432 | if self.hay_state.Resolve(arg0):
|
| 433 | return self.RunBuiltin(builtin_i.haynode, cmd_val)
|
| 434 |
|
| 435 | builtin_id = consts.LookupNormalBuiltin(arg0)
|
| 436 |
|
| 437 | if self.exec_opts._running_hay():
|
| 438 | # Hay: limit the builtins that can be run
|
| 439 | # - declare 'use dialect'
|
| 440 | # - echo and write for debugging
|
| 441 | # - no JSON?
|
| 442 | if builtin_id in (builtin_i.haynode, builtin_i.use, builtin_i.echo,
|
| 443 | builtin_i.write):
|
| 444 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 445 | return self.RunBuiltin(builtin_id, cmd_val)
|
| 446 |
|
| 447 | self.errfmt.Print_('Unknown command %r while running hay' % arg0,
|
| 448 | arg0_loc)
|
| 449 | return 127
|
| 450 |
|
| 451 | if builtin_id != consts.NO_INDEX:
|
| 452 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 453 | return self.RunBuiltin(builtin_id, cmd_val)
|
| 454 |
|
| 455 | environ = self.mem.GetEnv() # Include temporary variables
|
| 456 |
|
| 457 | if cmd_val.proc_args:
|
| 458 | e_die(
|
| 459 | '%r appears to be external. External commands don\'t accept typed args (OILS-ERR-200)'
|
| 460 | % arg0, cmd_val.proc_args.typed_args.left)
|
| 461 |
|
| 462 | # Resolve argv[0] BEFORE forking.
|
| 463 | if run_flags & USE_DEFAULT_PATH:
|
| 464 | argv0_path = LookupExecutable(arg0, DEFAULT_PATH)
|
| 465 | else:
|
| 466 | argv0_path = self.search_path.CachedLookup(arg0)
|
| 467 | if argv0_path is None:
|
| 468 | self.errfmt.Print_('%r not found (OILS-ERR-100)' % arg0, arg0_loc)
|
| 469 | return 127
|
| 470 |
|
| 471 | if self.trap_state.ThisProcessHasTraps():
|
| 472 | do_fork = True
|
| 473 | else:
|
| 474 | do_fork = not cmd_val.is_last_cmd
|
| 475 |
|
| 476 | # Normal case: ls /
|
| 477 | if do_fork:
|
| 478 | thunk = process.ExternalThunk(self.ext_prog, argv0_path, cmd_val,
|
| 479 | environ)
|
| 480 | p = process.Process(thunk, self.job_control, self.job_list,
|
| 481 | self.tracer)
|
| 482 |
|
| 483 | if self.job_control.Enabled():
|
| 484 | if self.fg_pipeline is not None:
|
| 485 | pgid = self.fg_pipeline.ProcessGroupId()
|
| 486 | # If job control is enabled, this should be true
|
| 487 | assert pgid != process.INVALID_PGID
|
| 488 |
|
| 489 | change = process.SetPgid(pgid, self.tracer)
|
| 490 | self.fg_pipeline = None # clear to avoid confusion in subshells
|
| 491 | else:
|
| 492 | change = process.SetPgid(process.OWN_LEADER, self.tracer)
|
| 493 | p.AddStateChange(change)
|
| 494 |
|
| 495 | status = p.RunProcess(self.waiter, trace.External(cmd_val.argv))
|
| 496 |
|
| 497 | # this is close to a "leaf" for errors
|
| 498 | # problem: permission denied EACCESS prints duplicate messages
|
| 499 | # TODO: add message command 'ls' failed
|
| 500 | cmd_st.show_code = True
|
| 501 |
|
| 502 | return status
|
| 503 |
|
| 504 | self.tracer.OnExec(cmd_val.argv)
|
| 505 |
|
| 506 | # Already forked for pipeline: ls / | wc -l
|
| 507 | self.ext_prog.Exec(argv0_path, cmd_val, environ) # NEVER RETURNS
|
| 508 |
|
| 509 | raise AssertionError('for -Wreturn-type in C++')
|
| 510 |
|
| 511 | def RunBackgroundJob(self, node):
|
| 512 | # type: (command_t) -> int
|
| 513 | """For & etc."""
|
| 514 | # Special case for pipeline. There is some evidence here:
|
| 515 | # https://www.gnu.org/software/libc/manual/html_node/Launching-Jobs.html#Launching-Jobs
|
| 516 | #
|
| 517 | # "You can either make all the processes in the process group be children
|
| 518 | # of the shell process, or you can make one process in group be the
|
| 519 | # ancestor of all the other processes in that group. The sample shell
|
| 520 | # program presented in this chapter uses the first approach because it
|
| 521 | # makes bookkeeping somewhat simpler."
|
| 522 | UP_node = node
|
| 523 |
|
| 524 | if UP_node.tag() == command_e.Pipeline:
|
| 525 | node = cast(command.Pipeline, UP_node)
|
| 526 | pi = process.Pipeline(self.exec_opts.sigpipe_status_ok(),
|
| 527 | self.job_control, self.job_list, self.tracer)
|
| 528 | for child in node.children:
|
| 529 | p = self._MakeProcess(child, True, self.exec_opts.errtrace())
|
| 530 | p.Init_ParentPipeline(pi)
|
| 531 | pi.Add(p)
|
| 532 |
|
| 533 | pi.StartPipeline(self.waiter)
|
| 534 | pi.SetBackground()
|
| 535 | last_pid = pi.LastPid()
|
| 536 | self.mem.last_bg_pid = last_pid # for $!
|
| 537 |
|
| 538 | job_id = self.job_list.AddJob(pi) # show in 'jobs' list
|
| 539 |
|
| 540 | else:
|
| 541 | # Problem: to get the 'set -b' behavior of immediate notifications, we
|
| 542 | # have to register SIGCHLD. But then that introduces race conditions.
|
| 543 | # If we haven't called Register yet, then we won't know who to notify.
|
| 544 |
|
| 545 | p = self._MakeProcess(node, True, self.exec_opts.errtrace())
|
| 546 | if self.job_control.Enabled():
|
| 547 | p.AddStateChange(
|
| 548 | process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 549 |
|
| 550 | p.SetBackground()
|
| 551 | pid = p.StartProcess(trace.Fork)
|
| 552 | self.mem.last_bg_pid = pid # for $!
|
| 553 | job_id = self.job_list.AddJob(p) # show in 'jobs' list
|
| 554 |
|
| 555 | if self.exec_opts.interactive():
|
| 556 | # Print it like %1 to show it's a job
|
| 557 | print_stderr('[%%%d] PID %d Started' %
|
| 558 | (job_id, self.mem.last_bg_pid))
|
| 559 |
|
| 560 | return 0
|
| 561 |
|
| 562 | def RunPipeline(self, node, status_out):
|
| 563 | # type: (command.Pipeline, CommandStatus) -> None
|
| 564 |
|
| 565 | pi = process.Pipeline(self.exec_opts.sigpipe_status_ok(),
|
| 566 | self.job_control, self.job_list, self.tracer)
|
| 567 |
|
| 568 | # initialized with CommandStatus.CreateNull()
|
| 569 | pipe_locs = [] # type: List[loc_t]
|
| 570 |
|
| 571 | # First n-1 processes (which is empty when n == 1)
|
| 572 | n = len(node.children)
|
| 573 | for i in xrange(n - 1):
|
| 574 | child = node.children[i]
|
| 575 |
|
| 576 | # TODO: determine these locations at parse time?
|
| 577 | pipe_locs.append(loc.Command(child))
|
| 578 |
|
| 579 | p = self._MakeProcess(child, True, self.exec_opts.errtrace())
|
| 580 | p.Init_ParentPipeline(pi)
|
| 581 | pi.Add(p)
|
| 582 |
|
| 583 | last_child = node.children[n - 1]
|
| 584 | # Last piece of code is in THIS PROCESS. 'echo foo | read line; echo $line'
|
| 585 | pi.AddLast((self.cmd_ev, last_child))
|
| 586 | pipe_locs.append(loc.Command(last_child))
|
| 587 |
|
| 588 | with dev.ctx_Tracer(self.tracer, 'pipeline', None):
|
| 589 | pi.StartPipeline(self.waiter)
|
| 590 | self.fg_pipeline = pi
|
| 591 | status_out.pipe_status = pi.RunLastPart(self.waiter, self.fd_state)
|
| 592 | self.fg_pipeline = None # clear in case we didn't end up forking
|
| 593 |
|
| 594 | status_out.pipe_locs = pipe_locs
|
| 595 |
|
| 596 | def RunSubshell(self, node):
|
| 597 | # type: (command_t) -> int
|
| 598 | p = self._MakeProcess(node, True, self.exec_opts.errtrace())
|
| 599 | if self.job_control.Enabled():
|
| 600 | p.AddStateChange(process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 601 |
|
| 602 | return p.RunProcess(self.waiter, trace.ForkWait)
|
| 603 |
|
| 604 | def CaptureStdout(self, node):
|
| 605 | # type: (command_t) -> Tuple[int, str]
|
| 606 |
|
| 607 | p = self._MakeProcess(node, self.exec_opts.inherit_errexit(),
|
| 608 | self.exec_opts.errtrace())
|
| 609 | # Shell quirk: Command subs remain part of the shell's process group, so we
|
| 610 | # don't use p.AddStateChange(process.SetPgid(...))
|
| 611 |
|
| 612 | r, w = posix.pipe()
|
| 613 | p.AddStateChange(process.StdoutToPipe(r, w))
|
| 614 |
|
| 615 | p.StartProcess(trace.CommandSub)
|
| 616 | #log('Command sub started %d', pid)
|
| 617 |
|
| 618 | chunks = [] # type: List[str]
|
| 619 | posix.close(w) # not going to write
|
| 620 | while True:
|
| 621 | n, err_num = pyos.Read(r, 4096, chunks)
|
| 622 |
|
| 623 | if n < 0:
|
| 624 | if err_num == EINTR:
|
| 625 | pass # retry
|
| 626 | else:
|
| 627 | # Like the top level IOError handler
|
| 628 | e_die_status(
|
| 629 | 2,
|
| 630 | 'Oils I/O error (read): %s' % posix.strerror(err_num))
|
| 631 |
|
| 632 | elif n == 0: # EOF
|
| 633 | break
|
| 634 | posix.close(r)
|
| 635 |
|
| 636 | status = p.Wait(self.waiter)
|
| 637 | stdout_str = ''.join(chunks).rstrip('\n')
|
| 638 |
|
| 639 | return status, stdout_str
|
| 640 |
|
| 641 | def RunCommandSub(self, cs_part):
|
| 642 | # type: (CommandSub) -> str
|
| 643 |
|
| 644 | if not self.exec_opts._allow_command_sub():
|
| 645 | # _allow_command_sub is used in two places. Only one of them turns
|
| 646 | # off _allow_process_sub
|
| 647 | if not self.exec_opts._allow_process_sub():
|
| 648 | why = "status wouldn't be checked (strict_errexit)"
|
| 649 | else:
|
| 650 | why = 'eval_unsafe_arith is off'
|
| 651 |
|
| 652 | e_die("Command subs not allowed here because %s" % why,
|
| 653 | loc.WordPart(cs_part))
|
| 654 |
|
| 655 | node = cs_part.child
|
| 656 |
|
| 657 | # Hack for weird $(<file) construct
|
| 658 | if node.tag() == command_e.Redirect:
|
| 659 | redir_node = cast(command.Redirect, node)
|
| 660 | # Detect '< file'
|
| 661 | if (len(redir_node.redirects) == 1 and
|
| 662 | redir_node.redirects[0].op.id == Id.Redir_Less and
|
| 663 | redir_node.child.tag() == command_e.NoOp):
|
| 664 |
|
| 665 | # Change it to __cat < file.
|
| 666 | # TODO: could be 'internal cat' (issue #1013)
|
| 667 | tok = lexer.DummyToken(Id.Lit_Chars, '__cat')
|
| 668 | cat_word = CompoundWord([tok])
|
| 669 |
|
| 670 | # Blame < because __cat has no location
|
| 671 | blame_tok = redir_node.redirects[0].op
|
| 672 | simple = command.Simple(blame_tok, [], [cat_word], None, None,
|
| 673 | False)
|
| 674 |
|
| 675 | # MUTATE redir node so it's like $(<file _cat)
|
| 676 | redir_node.child = simple
|
| 677 |
|
| 678 | status, stdout_str = self.CaptureStdout(node)
|
| 679 |
|
| 680 | # OSH has the concept of aborting in the middle of a WORD. We're not
|
| 681 | # waiting until the command is over!
|
| 682 | if self.exec_opts.command_sub_errexit():
|
| 683 | if status != 0:
|
| 684 | msg = 'Command Sub exited with status %d' % status
|
| 685 | raise error.ErrExit(status, msg, loc.WordPart(cs_part))
|
| 686 |
|
| 687 | else:
|
| 688 | # Set a flag so we check errexit at the same time as bash. Example:
|
| 689 | #
|
| 690 | # a=$(false)
|
| 691 | # echo foo # no matter what comes here, the flag is reset
|
| 692 | #
|
| 693 | # Set ONLY until this command node has finished executing.
|
| 694 |
|
| 695 | # HACK: move this
|
| 696 | self.cmd_ev.check_command_sub_status = True
|
| 697 | self.mem.SetLastStatus(status)
|
| 698 |
|
| 699 | # Runtime errors test case: # $("echo foo > $@")
|
| 700 | # Why rstrip()?
|
| 701 | # https://unix.stackexchange.com/questions/17747/why-does-shell-command-substitution-gobble-up-a-trailing-newline-char
|
| 702 | return stdout_str
|
| 703 |
|
| 704 | def RunProcessSub(self, cs_part):
|
| 705 | # type: (CommandSub) -> str
|
| 706 | """Process sub creates a forks a process connected to a pipe.
|
| 707 |
|
| 708 | The pipe is typically passed to another process via a /dev/fd/$FD path.
|
| 709 |
|
| 710 | Life cycle of a process substitution:
|
| 711 |
|
| 712 | 1. Start with this code
|
| 713 |
|
| 714 | diff <(seq 3) <(seq 4)
|
| 715 |
|
| 716 | 2. To evaluate the command line, we evaluate every word. The
|
| 717 | NormalWordEvaluator this method, RunProcessSub(), which does 3 things:
|
| 718 |
|
| 719 | a. Create a pipe(), getting r and w
|
| 720 | b. Starts the seq process, which inherits r and w
|
| 721 | It has a StdoutToPipe() redirect, which means that it dup2(w, 1)
|
| 722 | and close(r)
|
| 723 | c. Close the w FD, because neither the shell or 'diff' will write to it.
|
| 724 | However we must retain 'r', because 'diff' hasn't opened /dev/fd yet!
|
| 725 | d. We evaluate <(seq 3) to /dev/fd/$r, so "diff" can read from it
|
| 726 |
|
| 727 | 3. Now we're done evaluating every word, so we know the command line of
|
| 728 | diff, which looks like
|
| 729 |
|
| 730 | diff /dev/fd/64 /dev/fd/65
|
| 731 |
|
| 732 | Those are the FDs for the read ends of the pipes we created.
|
| 733 |
|
| 734 | 4. diff inherits a copy of the read end of bot pipes. But it actually
|
| 735 | calls open() both files passed as argv. (I think this is fine.)
|
| 736 |
|
| 737 | 5. wait() for the diff process.
|
| 738 |
|
| 739 | 6. The shell closes both the read ends of both pipes. Neither us or
|
| 740 | 'diffd' will read again.
|
| 741 |
|
| 742 | 7. The shell waits for both 'seq' processes.
|
| 743 |
|
| 744 | Related:
|
| 745 | shopt -s process_sub_fail
|
| 746 | _process_sub_status
|
| 747 | """
|
| 748 | cs_loc = loc.WordPart(cs_part)
|
| 749 |
|
| 750 | if not self.exec_opts._allow_process_sub():
|
| 751 | e_die(
|
| 752 | "Process subs not allowed here because status wouldn't be checked (strict_errexit)",
|
| 753 | cs_loc)
|
| 754 |
|
| 755 | p = self._MakeProcess(cs_part.child, True, self.exec_opts.errtrace())
|
| 756 |
|
| 757 | r, w = posix.pipe()
|
| 758 | #log('pipe = %d, %d', r, w)
|
| 759 |
|
| 760 | op_id = cs_part.left_token.id
|
| 761 | if op_id == Id.Left_ProcSubIn:
|
| 762 | # Example: cat < <(head foo.txt)
|
| 763 | #
|
| 764 | # The head process should write its stdout to a pipe.
|
| 765 | redir = process.StdoutToPipe(r,
|
| 766 | w) # type: process.ChildStateChange
|
| 767 |
|
| 768 | elif op_id == Id.Left_ProcSubOut:
|
| 769 | # Example: head foo.txt > >(tac)
|
| 770 | #
|
| 771 | # The tac process should read its stdin from a pipe.
|
| 772 |
|
| 773 | # Note: this example sometimes requires you to hit "enter" in bash and
|
| 774 | # zsh. WHy?
|
| 775 | redir = process.StdinFromPipe(r, w)
|
| 776 |
|
| 777 | else:
|
| 778 | raise AssertionError()
|
| 779 |
|
| 780 | p.AddStateChange(redir)
|
| 781 |
|
| 782 | if self.job_control.Enabled():
|
| 783 | p.AddStateChange(process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 784 |
|
| 785 | # Fork, letting the child inherit the pipe file descriptors.
|
| 786 | p.StartProcess(trace.ProcessSub)
|
| 787 |
|
| 788 | ps_frame = self.process_sub_stack[-1]
|
| 789 |
|
| 790 | # Note: bash never waits() on the process, but zsh does. The calling
|
| 791 | # program needs to read() before we can wait, e.g.
|
| 792 | # diff <(sort left.txt) <(sort right.txt)
|
| 793 |
|
| 794 | # After forking, close the end of the pipe we're not using.
|
| 795 | if op_id == Id.Left_ProcSubIn:
|
| 796 | posix.close(w) # cat < <(head foo.txt)
|
| 797 | ps_frame.Append(p, r, cs_loc) # close later
|
| 798 | elif op_id == Id.Left_ProcSubOut:
|
| 799 | posix.close(r)
|
| 800 | #log('Left_ProcSubOut closed %d', r)
|
| 801 | ps_frame.Append(p, w, cs_loc) # close later
|
| 802 | else:
|
| 803 | raise AssertionError()
|
| 804 |
|
| 805 | # Is /dev Linux-specific?
|
| 806 | if op_id == Id.Left_ProcSubIn:
|
| 807 | return '/dev/fd/%d' % r
|
| 808 |
|
| 809 | elif op_id == Id.Left_ProcSubOut:
|
| 810 | return '/dev/fd/%d' % w
|
| 811 |
|
| 812 | else:
|
| 813 | raise AssertionError()
|
| 814 |
|
| 815 | def PushRedirects(self, redirects, err_out):
|
| 816 | # type: (List[RedirValue], List[error.IOError_OSError]) -> None
|
| 817 | if len(redirects) == 0: # Optimized to avoid allocs
|
| 818 | return
|
| 819 | self.fd_state.Push(redirects, err_out)
|
| 820 |
|
| 821 | def PopRedirects(self, num_redirects, err_out):
|
| 822 | # type: (int, List[error.IOError_OSError]) -> None
|
| 823 | if num_redirects == 0: # Optimized to avoid allocs
|
| 824 | return
|
| 825 | self.fd_state.Pop(err_out)
|
| 826 |
|
| 827 | def PushProcessSub(self):
|
| 828 | # type: () -> None
|
| 829 | if len(self.clean_frame_pool):
|
| 830 | # Optimized to avoid allocs
|
| 831 | new_frame = self.clean_frame_pool.pop()
|
| 832 | else:
|
| 833 | new_frame = _ProcessSubFrame()
|
| 834 | self.process_sub_stack.append(new_frame)
|
| 835 |
|
| 836 | def PopProcessSub(self, compound_st):
|
| 837 | # type: (StatusArray) -> None
|
| 838 | """This method is called by a context manager, which means we always
|
| 839 | wait() on the way out, which I think is the right thing.
|
| 840 |
|
| 841 | We don't always set _process_sub_status, e.g. if some fatal
|
| 842 | error occurs first, but we always wait.
|
| 843 | """
|
| 844 | frame = self.process_sub_stack.pop()
|
| 845 | if frame.WasModified():
|
| 846 | frame.MaybeWaitOnProcessSubs(self.waiter, compound_st)
|
| 847 | else:
|
| 848 | # Optimized to avoid allocs
|
| 849 | self.clean_frame_pool.append(frame)
|
| 850 |
|
| 851 | # Note: the 3 lists in _ProcessSubFrame are hot in our profiles. It would
|
| 852 | # be nice to somehow "destroy" them here, rather than letting them become
|
| 853 | # garbage that needs to be traced.
|
| 854 |
|
| 855 | # The CommandEvaluator could have a ProcessSubStack, which supports Push(),
|
| 856 | # Pop(), and Top() of VALUES rather than GC objects?
|
| 857 |
|
| 858 |
|
| 859 | # vim: sw=4
|