| 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 builtin import hay_ysh
|
| 18 | from core import dev
|
| 19 | from core import error
|
| 20 | from core.error import e_die, e_die_status
|
| 21 | from core import process
|
| 22 | from core import pyos
|
| 23 | from core import state
|
| 24 | from core import vm
|
| 25 | from display import ui
|
| 26 | from frontend import consts
|
| 27 | from frontend import lexer
|
| 28 | from mycpp import mylib
|
| 29 | from mycpp.mylib import log, print_stderr
|
| 30 | from pylib import os_path
|
| 31 | from pylib import path_stat
|
| 32 |
|
| 33 | import posix_ as posix
|
| 34 | from posix_ import X_OK # translated directly to C macro
|
| 35 |
|
| 36 | from typing import cast, Dict, List, Tuple, Optional, TYPE_CHECKING
|
| 37 | if TYPE_CHECKING:
|
| 38 | from _devbuild.gen.runtime_asdl import (cmd_value, CommandStatus,
|
| 39 | StatusArray)
|
| 40 | from _devbuild.gen.syntax_asdl import command_t
|
| 41 | from builtin import trap_osh
|
| 42 | from core import optview
|
| 43 |
|
| 44 | _ = log
|
| 45 |
|
| 46 |
|
| 47 | def LookupExecutable(name, path_dirs, exec_required=True):
|
| 48 | # type: (str, List[str], bool) -> Optional[str]
|
| 49 | """
|
| 50 | Returns either
|
| 51 | - the name if it's a relative path that exists
|
| 52 | - the executable name resolved against path_dirs
|
| 53 | - None if not found
|
| 54 | """
|
| 55 | if len(name) == 0: # special case for "$(true)"
|
| 56 | return None
|
| 57 |
|
| 58 | if '/' in name:
|
| 59 | return name if path_stat.exists(name) else None
|
| 60 |
|
| 61 | for path_dir in path_dirs:
|
| 62 | full_path = os_path.join(path_dir, name)
|
| 63 | if exec_required:
|
| 64 | found = posix.access(full_path, X_OK)
|
| 65 | else:
|
| 66 | found = path_stat.exists(full_path)
|
| 67 |
|
| 68 | if found:
|
| 69 | return full_path
|
| 70 |
|
| 71 | return None
|
| 72 |
|
| 73 |
|
| 74 | class SearchPath(object):
|
| 75 | """For looking up files in $PATH or ENV.PATH"""
|
| 76 |
|
| 77 | def __init__(self, mem, exec_opts):
|
| 78 | # type: (state.Mem, optview.Exec) -> None
|
| 79 | self.mem = mem
|
| 80 | # TODO: remove exec_opts
|
| 81 | self.cache = {} # type: Dict[str, str]
|
| 82 |
|
| 83 | def _GetPath(self):
|
| 84 | # type: () -> List[str]
|
| 85 |
|
| 86 | # In YSH, we read from ENV.PATH
|
| 87 | s = self.mem.env_config.Get('PATH')
|
| 88 | if s is None:
|
| 89 | return [] # treat as empty path
|
| 90 |
|
| 91 | # TODO: Could cache this to avoid split() allocating all the time.
|
| 92 | return s.split(':')
|
| 93 |
|
| 94 | def LookupOne(self, name, exec_required=True):
|
| 95 | # type: (str, bool) -> Optional[str]
|
| 96 | """
|
| 97 | Returns the path itself (if relative path), the resolved path, or None.
|
| 98 | """
|
| 99 | return LookupExecutable(name,
|
| 100 | self._GetPath(),
|
| 101 | exec_required=exec_required)
|
| 102 |
|
| 103 | def LookupReflect(self, name, do_all):
|
| 104 | # type: (str, bool) -> List[str]
|
| 105 | """
|
| 106 | Like LookupOne(), with an option for 'type -a' to return all paths.
|
| 107 | """
|
| 108 | if len(name) == 0: # special case for "$(true)"
|
| 109 | return []
|
| 110 |
|
| 111 | if '/' in name:
|
| 112 | if path_stat.exists(name):
|
| 113 | return [name]
|
| 114 | else:
|
| 115 | return []
|
| 116 |
|
| 117 | results = [] # type: List[str]
|
| 118 | for path_dir in self._GetPath():
|
| 119 | full_path = os_path.join(path_dir, name)
|
| 120 | if path_stat.exists(full_path):
|
| 121 | results.append(full_path)
|
| 122 | if not do_all:
|
| 123 | return results
|
| 124 |
|
| 125 | return results
|
| 126 |
|
| 127 | def CachedLookup(self, name):
|
| 128 | # type: (str) -> Optional[str]
|
| 129 | #log('name %r', name)
|
| 130 | if name in self.cache:
|
| 131 | return self.cache[name]
|
| 132 |
|
| 133 | full_path = self.LookupOne(name)
|
| 134 | if full_path is not None:
|
| 135 | self.cache[name] = full_path
|
| 136 | return full_path
|
| 137 |
|
| 138 | def MaybeRemoveEntry(self, name):
|
| 139 | # type: (str) -> None
|
| 140 | """When the file system changes."""
|
| 141 | mylib.dict_erase(self.cache, name)
|
| 142 |
|
| 143 | def ClearCache(self):
|
| 144 | # type: () -> None
|
| 145 | """For hash -r."""
|
| 146 | self.cache.clear()
|
| 147 |
|
| 148 | def CachedCommands(self):
|
| 149 | # type: () -> List[str]
|
| 150 | return self.cache.values()
|
| 151 |
|
| 152 |
|
| 153 | class _ProcessSubFrame(object):
|
| 154 | """To keep track of diff <(cat 1) <(cat 2) > >(tac)"""
|
| 155 |
|
| 156 | def __init__(self):
|
| 157 | # type: () -> None
|
| 158 |
|
| 159 | # These objects appear unconditionally in the main loop, and aren't
|
| 160 | # commonly used, so we manually optimize [] into None.
|
| 161 |
|
| 162 | self._to_wait = [] # type: List[process.Process]
|
| 163 | self._to_close = [] # type: List[int] # file descriptors
|
| 164 | self._locs = [] # type: List[loc_t]
|
| 165 | self._modified = False
|
| 166 |
|
| 167 | def WasModified(self):
|
| 168 | # type: () -> bool
|
| 169 | return self._modified
|
| 170 |
|
| 171 | def Append(self, p, fd, status_loc):
|
| 172 | # type: (process.Process, int, loc_t) -> None
|
| 173 | self._modified = True
|
| 174 |
|
| 175 | self._to_wait.append(p)
|
| 176 | self._to_close.append(fd)
|
| 177 | self._locs.append(status_loc)
|
| 178 |
|
| 179 | def MaybeWaitOnProcessSubs(self, waiter, status_array):
|
| 180 | # type: (process.Waiter, StatusArray) -> None
|
| 181 |
|
| 182 | # Wait in the same order that they were evaluated. That seems fine.
|
| 183 | for fd in self._to_close:
|
| 184 | posix.close(fd)
|
| 185 |
|
| 186 | codes = [] # type: List[int]
|
| 187 | locs = [] # type: List[loc_t]
|
| 188 | for i, p in enumerate(self._to_wait):
|
| 189 | #log('waiting for %s', p)
|
| 190 | st = p.Wait(waiter)
|
| 191 | codes.append(st)
|
| 192 | locs.append(self._locs[i])
|
| 193 |
|
| 194 | status_array.codes = codes
|
| 195 | status_array.locs = locs
|
| 196 |
|
| 197 |
|
| 198 | # Big flags for _RunSimpleCommand
|
| 199 | IS_LAST_CMD = 1 << 1
|
| 200 | NO_CALL_PROCS = 1 << 2 # command ls suppresses function lookup
|
| 201 | USE_DEFAULT_PATH = 1 << 3 # for command -p ls changes the path
|
| 202 |
|
| 203 | # Copied from var.c in dash
|
| 204 | DEFAULT_PATH = [
|
| 205 | '/usr/local/sbin', '/usr/local/bin', '/usr/sbin', '/usr/bin', '/sbin',
|
| 206 | '/bin'
|
| 207 | ]
|
| 208 |
|
| 209 | _PURITY_STATUS = 5
|
| 210 |
|
| 211 |
|
| 212 | class PureExecutor(vm._Executor):
|
| 213 |
|
| 214 | # mycpp needs this duplicate constructor
|
| 215 | def __init__(
|
| 216 | self,
|
| 217 | mem, # type: state.Mem
|
| 218 | exec_opts, # type: optview.Exec
|
| 219 | mutable_opts, # type: state.MutableOpts
|
| 220 | procs, # type: state.Procs
|
| 221 | hay_state, # type: hay_ysh.HayState
|
| 222 | builtins, # type: Dict[int, vm._Builtin]
|
| 223 | internals, # type: Dict[int, vm._Builtin]
|
| 224 | tracer, # type: dev.Tracer
|
| 225 | errfmt # type: ui.ErrorFormatter
|
| 226 | ):
|
| 227 | vm._Executor.__init__(self, mem, exec_opts, mutable_opts, procs,
|
| 228 | hay_state, builtins, internals, tracer, errfmt)
|
| 229 |
|
| 230 | def _RunSimpleCommand(self, arg0, arg0_loc, cmd_val, cmd_st, run_flags):
|
| 231 | # type: (str, loc_t, cmd_value.Argv, CommandStatus, int) -> int
|
| 232 |
|
| 233 | call_procs = not (run_flags & NO_CALL_PROCS)
|
| 234 | if call_procs:
|
| 235 | proc_val, self_obj = self.procs.GetInvokable(arg0)
|
| 236 | if proc_val is not None:
|
| 237 | return self._RunInvokable(proc_val, self_obj, arg0_loc,
|
| 238 | cmd_val)
|
| 239 |
|
| 240 | if self.hay_state.Resolve(arg0):
|
| 241 | return self.RunBuiltin(builtin_i.haynode, cmd_val)
|
| 242 |
|
| 243 | self.errfmt.Print_(
|
| 244 | 'Command %r not found in pure mode (OILS-ERR-102)' % arg0,
|
| 245 | arg0_loc)
|
| 246 | return 127
|
| 247 |
|
| 248 | def RunBackgroundJob(self, node):
|
| 249 | # type: (command_t) -> int
|
| 250 | raise error.Structured(
|
| 251 | _PURITY_STATUS,
|
| 252 | "Background jobs aren't allowed in pure mode (OILS-ERR-204)",
|
| 253 | loc.Command(node))
|
| 254 |
|
| 255 | def RunPipeline(self, node, status_out):
|
| 256 | # type: (command.Pipeline, CommandStatus) -> None
|
| 257 | raise error.Structured(
|
| 258 | _PURITY_STATUS,
|
| 259 | "Pipelines aren't allowed in pure mode (OILS-ERR-204)",
|
| 260 | loc.Command(node))
|
| 261 |
|
| 262 | def RunSubshell(self, node):
|
| 263 | # type: (command_t) -> int
|
| 264 | raise error.Structured(
|
| 265 | _PURITY_STATUS,
|
| 266 | "Subshells aren't allowed in pure mode (OILS-ERR-204)",
|
| 267 | loc.Command(node))
|
| 268 |
|
| 269 | def CaptureStdout(self, node):
|
| 270 | # type: (command_t) -> Tuple[int, str]
|
| 271 | """
|
| 272 | Used by io->captureStdout() method, and called by command sub
|
| 273 | """
|
| 274 | return 0, ''
|
| 275 |
|
| 276 | def Capture3(self, node):
|
| 277 | # type: (command_t) -> Tuple[int, str, str]
|
| 278 | """
|
| 279 | Used by io->captureAll() method, and called by command sub
|
| 280 | """
|
| 281 | return 0, '', ''
|
| 282 |
|
| 283 | def RunCommandSub(self, cs_part):
|
| 284 | # type: (CommandSub) -> str
|
| 285 | raise error.Structured(
|
| 286 | _PURITY_STATUS,
|
| 287 | "Command subs aren't allowed in pure mode (OILS-ERR-204)",
|
| 288 | loc.WordPart(cs_part))
|
| 289 |
|
| 290 | def RunProcessSub(self, cs_part):
|
| 291 | # type: (CommandSub) -> str
|
| 292 | raise error.Structured(
|
| 293 | _PURITY_STATUS,
|
| 294 | "Process subs aren't allowed in pure mode (OILS-ERR-204)",
|
| 295 | loc.WordPart(cs_part))
|
| 296 |
|
| 297 | def PushRedirects(self, redirects, err_out):
|
| 298 | # type: (List[RedirValue], List[error.IOError_OSError]) -> None
|
| 299 | pass
|
| 300 |
|
| 301 | def PopRedirects(self, num_redirects, err_out):
|
| 302 | # type: (int, List[error.IOError_OSError]) -> None
|
| 303 | pass
|
| 304 |
|
| 305 | def PushProcessSub(self):
|
| 306 | # type: () -> None
|
| 307 | pass
|
| 308 |
|
| 309 | def PopProcessSub(self, compound_st):
|
| 310 | # type: (StatusArray) -> None
|
| 311 | pass
|
| 312 |
|
| 313 |
|
| 314 | class ShellExecutor(vm._Executor):
|
| 315 | """An executor combined with the OSH language evaluators in osh/ to create
|
| 316 | a shell interpreter."""
|
| 317 |
|
| 318 | def __init__(
|
| 319 | self,
|
| 320 | mem, # type: state.Mem
|
| 321 | exec_opts, # type: optview.Exec
|
| 322 | mutable_opts, # type: state.MutableOpts
|
| 323 | procs, # type: state.Procs
|
| 324 | hay_state, # type: hay_ysh.HayState
|
| 325 | builtins, # type: Dict[int, vm._Builtin]
|
| 326 | internals, # type: Dict[int, vm._Builtin]
|
| 327 | tracer, # type: dev.Tracer
|
| 328 | errfmt, # type: ui.ErrorFormatter
|
| 329 | search_path, # type: SearchPath
|
| 330 | ext_prog, # type: process.ExternalProgram
|
| 331 | waiter, # type: process.Waiter
|
| 332 | job_control, # type: process.JobControl
|
| 333 | job_list, # type: process.JobList
|
| 334 | fd_state, # type: process.FdState
|
| 335 | trap_state, # type: trap_osh.TrapState
|
| 336 | ):
|
| 337 | # type: (...) -> None
|
| 338 | vm._Executor.__init__(self, mem, exec_opts, mutable_opts, procs,
|
| 339 | hay_state, builtins, internals, tracer, errfmt)
|
| 340 | self.search_path = search_path
|
| 341 | self.ext_prog = ext_prog
|
| 342 | self.waiter = waiter
|
| 343 | self.multi_trace = tracer.multi_trace
|
| 344 | self.job_control = job_control
|
| 345 | # sleep 5 & puts a (PID, job#) entry here. And then "jobs" displays it.
|
| 346 | self.job_list = job_list
|
| 347 | self.fd_state = fd_state
|
| 348 | self.trap_state = trap_state
|
| 349 | self.process_sub_stack = [] # type: List[_ProcessSubFrame]
|
| 350 | self.clean_frame_pool = [] # type: List[_ProcessSubFrame]
|
| 351 |
|
| 352 | # When starting a pipeline in the foreground, we need to pass a handle to it
|
| 353 | # through the evaluation of the last node back to ourselves for execution.
|
| 354 | # We use this handle to make sure any processes forked for the last part of
|
| 355 | # the pipeline are placed into the same process group as the rest of the
|
| 356 | # pipeline. Since there is, by design, only ever one foreground pipeline and
|
| 357 | # any pipelines started within subshells run in their parent's process
|
| 358 | # group, we only need one pointer here, not some collection.
|
| 359 | self.fg_pipeline = None # type: Optional[process.Pipeline]
|
| 360 |
|
| 361 | def _MakeProcess(self, node, inherit_errexit, inherit_errtrace):
|
| 362 | # type: (command_t, bool, bool) -> process.Process
|
| 363 | """Assume we will run the node in another process.
|
| 364 |
|
| 365 | Return a process.
|
| 366 | """
|
| 367 | UP_node = node
|
| 368 | if node.tag() == command_e.ControlFlow:
|
| 369 | node = cast(command.ControlFlow, UP_node)
|
| 370 | # Pipeline or subshells with control flow are invalid, e.g.:
|
| 371 | # - break | less
|
| 372 | # - continue | less
|
| 373 | # - ( return )
|
| 374 | # NOTE: This could be done at parse time too.
|
| 375 | if node.keyword.id != Id.ControlFlow_Exit:
|
| 376 | e_die(
|
| 377 | 'Invalid control flow %r in pipeline / subshell / background'
|
| 378 | % lexer.TokenVal(node.keyword), node.keyword)
|
| 379 |
|
| 380 | # NOTE: If ErrExit(), we could be verbose about subprogram errors? This
|
| 381 | # only really matters when executing 'exit 42', because the child shell
|
| 382 | # inherits errexit and will be verbose. Other notes:
|
| 383 | #
|
| 384 | # - We might want errors to fit on a single line so they don't get
|
| 385 | # interleaved.
|
| 386 | # - We could turn the `exit` builtin into a error.FatalRuntime exception
|
| 387 | # and get this check for "free".
|
| 388 | thunk = process.SubProgramThunk(self.cmd_ev, node, self.trap_state,
|
| 389 | self.multi_trace, inherit_errexit,
|
| 390 | inherit_errtrace)
|
| 391 | p = process.Process(thunk, self.job_control, self.job_list,
|
| 392 | self.tracer)
|
| 393 | return p
|
| 394 |
|
| 395 | def _RunSimpleCommand(self, arg0, arg0_loc, cmd_val, cmd_st, run_flags):
|
| 396 | # type: (str, loc_t, cmd_value.Argv, CommandStatus, int) -> int
|
| 397 | """Run builtins, functions, external commands.
|
| 398 |
|
| 399 | Possible variations:
|
| 400 | - YSH might have different, simpler rules. No special builtins, etc.
|
| 401 | - YSH might have OILS_PATH = :| /bin /usr/bin | or something.
|
| 402 | - Interpreters might want to define all their own builtins.
|
| 403 | """
|
| 404 |
|
| 405 | builtin_id = consts.LookupAssignBuiltin(arg0)
|
| 406 | if builtin_id != consts.NO_INDEX:
|
| 407 | # command readonly is disallowed, for technical reasons. Could relax it
|
| 408 | # later.
|
| 409 | self.errfmt.Print_("Simple command can't run assignment builtin",
|
| 410 | arg0_loc)
|
| 411 | return 1
|
| 412 |
|
| 413 | builtin_id = consts.LookupSpecialBuiltin(arg0)
|
| 414 | if builtin_id != consts.NO_INDEX:
|
| 415 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 416 | status = self.RunBuiltin(builtin_id, cmd_val)
|
| 417 | # TODO: Enable this and fix spec test failures.
|
| 418 | # Also update _SPECIAL_BUILTINS in osh/builtin.py.
|
| 419 | #if status != 0:
|
| 420 | # e_die_status(status, 'special builtin failed')
|
| 421 | return status
|
| 422 |
|
| 423 | # Call procs first. Builtins like 'true' can be redefined.
|
| 424 | call_procs = not (run_flags & NO_CALL_PROCS)
|
| 425 | if call_procs:
|
| 426 | proc_val, self_obj = self.procs.GetInvokable(arg0)
|
| 427 | if proc_val is not None:
|
| 428 | return self._RunInvokable(proc_val, self_obj, arg0_loc,
|
| 429 | cmd_val)
|
| 430 |
|
| 431 | # Notes:
|
| 432 | # - procs shadow hay names
|
| 433 | # - hay names shadow normal builtins? Should we limit to CAPS or no?
|
| 434 | if self.hay_state.Resolve(arg0):
|
| 435 | return self.RunBuiltin(builtin_i.haynode, cmd_val)
|
| 436 |
|
| 437 | builtin_id = consts.LookupNormalBuiltin(arg0)
|
| 438 |
|
| 439 | if self.exec_opts._running_hay():
|
| 440 | # Hay: limit the builtins that can be run
|
| 441 | # - declare 'use dialect'
|
| 442 | # - echo and write for debugging
|
| 443 | # - no JSON?
|
| 444 | if builtin_id in (builtin_i.haynode, builtin_i.use, builtin_i.echo,
|
| 445 | builtin_i.write):
|
| 446 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 447 | return self.RunBuiltin(builtin_id, cmd_val)
|
| 448 |
|
| 449 | self.errfmt.Print_('Unknown command %r while running hay' % arg0,
|
| 450 | arg0_loc)
|
| 451 | return 127
|
| 452 |
|
| 453 | if builtin_id != consts.NO_INDEX:
|
| 454 | cmd_st.show_code = True # this is a "leaf" for errors
|
| 455 | return self.RunBuiltin(builtin_id, cmd_val)
|
| 456 |
|
| 457 | environ = self.mem.GetEnv() # Include temporary variables
|
| 458 |
|
| 459 | if cmd_val.proc_args:
|
| 460 | e_die(
|
| 461 | '%r appears to be external. External commands don\'t accept typed args (OILS-ERR-200)'
|
| 462 | % arg0, cmd_val.proc_args.typed_args.left)
|
| 463 |
|
| 464 | # Resolve argv[0] BEFORE forking.
|
| 465 | if run_flags & USE_DEFAULT_PATH:
|
| 466 | argv0_path = LookupExecutable(arg0, DEFAULT_PATH)
|
| 467 | else:
|
| 468 | argv0_path = self.search_path.CachedLookup(arg0)
|
| 469 | if argv0_path is None:
|
| 470 | self.errfmt.Print_('Command %r not found (OILS-ERR-100)' % arg0,
|
| 471 | arg0_loc)
|
| 472 | return 127
|
| 473 |
|
| 474 | if self.trap_state.ThisProcessHasTraps():
|
| 475 | do_fork = True
|
| 476 | else:
|
| 477 | do_fork = not cmd_val.is_last_cmd
|
| 478 |
|
| 479 | # Normal case: ls /
|
| 480 | if do_fork:
|
| 481 | thunk = process.ExternalThunk(self.ext_prog, argv0_path, cmd_val,
|
| 482 | environ)
|
| 483 | p = process.Process(thunk, self.job_control, self.job_list,
|
| 484 | self.tracer)
|
| 485 |
|
| 486 | if self.job_control.Enabled():
|
| 487 | if self.fg_pipeline is not None:
|
| 488 | pgid = self.fg_pipeline.ProcessGroupId()
|
| 489 | # If job control is enabled, this should be true
|
| 490 | assert pgid != process.INVALID_PGID
|
| 491 |
|
| 492 | change = process.SetPgid(pgid, self.tracer)
|
| 493 | self.fg_pipeline = None # clear to avoid confusion in subshells
|
| 494 | else:
|
| 495 | change = process.SetPgid(process.OWN_LEADER, self.tracer)
|
| 496 | p.AddStateChange(change)
|
| 497 |
|
| 498 | status = p.RunProcess(self.waiter, trace.External(cmd_val.argv))
|
| 499 |
|
| 500 | # this is close to a "leaf" for errors
|
| 501 | # problem: permission denied EACCESS prints duplicate messages
|
| 502 | # TODO: add message command 'ls' failed
|
| 503 | cmd_st.show_code = True
|
| 504 |
|
| 505 | return status
|
| 506 |
|
| 507 | self.tracer.OnExec(cmd_val.argv)
|
| 508 |
|
| 509 | # Already forked for pipeline: ls / | wc -l
|
| 510 | self.ext_prog.Exec(argv0_path, cmd_val, environ) # NEVER RETURNS
|
| 511 |
|
| 512 | raise AssertionError('for -Wreturn-type in C++')
|
| 513 |
|
| 514 | def RunBackgroundJob(self, node):
|
| 515 | # type: (command_t) -> int
|
| 516 | """For & etc."""
|
| 517 | # Special case for pipeline. There is some evidence here:
|
| 518 | # https://www.gnu.org/software/libc/manual/html_node/Launching-Jobs.html#Launching-Jobs
|
| 519 | #
|
| 520 | # "You can either make all the processes in the process group be children
|
| 521 | # of the shell process, or you can make one process in group be the
|
| 522 | # ancestor of all the other processes in that group. The sample shell
|
| 523 | # program presented in this chapter uses the first approach because it
|
| 524 | # makes bookkeeping somewhat simpler."
|
| 525 | UP_node = node
|
| 526 |
|
| 527 | if UP_node.tag() == command_e.Pipeline:
|
| 528 | node = cast(command.Pipeline, UP_node)
|
| 529 | pi = process.Pipeline(self.exec_opts.sigpipe_status_ok(),
|
| 530 | self.job_control, self.job_list, self.tracer)
|
| 531 | for child in node.children:
|
| 532 | p = self._MakeProcess(child, True, self.exec_opts.errtrace())
|
| 533 | p.Init_ParentPipeline(pi)
|
| 534 | pi.Add(p)
|
| 535 |
|
| 536 | pi.StartPipeline(self.waiter)
|
| 537 | pi.SetBackground()
|
| 538 | self.mem.last_bg_pid = pi.PidForWait() # for $!
|
| 539 | job_id = self.job_list.RegisterJob(pi) # show in 'jobs' list
|
| 540 |
|
| 541 | else:
|
| 542 | # Problem: to get the 'set -b' behavior of immediate notifications, we
|
| 543 | # have to register SIGCHLD. But then that introduces race conditions.
|
| 544 | # If we haven't called Register yet, then we won't know who to notify.
|
| 545 |
|
| 546 | p = self._MakeProcess(node, True, self.exec_opts.errtrace())
|
| 547 | if self.job_control.Enabled():
|
| 548 | p.AddStateChange(
|
| 549 | process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 550 |
|
| 551 | p.SetBackground()
|
| 552 | pid = p.StartProcess(trace.Fork)
|
| 553 | self.mem.last_bg_pid = p.PidForWait() # for $!
|
| 554 | job_id = self.job_list.RegisterJob(p) # show in 'jobs' list
|
| 555 |
|
| 556 | if self.exec_opts.interactive():
|
| 557 | # Print it like %1 to show it's a job
|
| 558 | print_stderr('[%%%d] PID %d Started' %
|
| 559 | (job_id, self.mem.last_bg_pid))
|
| 560 |
|
| 561 | return 0
|
| 562 |
|
| 563 | def RunPipeline(self, node, status_out):
|
| 564 | # type: (command.Pipeline, CommandStatus) -> None
|
| 565 |
|
| 566 | pi = process.Pipeline(self.exec_opts.sigpipe_status_ok(),
|
| 567 | self.job_control, self.job_list, self.tracer)
|
| 568 |
|
| 569 | # initialized with CommandStatus.CreateNull()
|
| 570 | pipe_locs = [] # type: List[loc_t]
|
| 571 |
|
| 572 | # First n-1 processes (which is empty when n == 1)
|
| 573 | n = len(node.children)
|
| 574 | for i in xrange(n - 1):
|
| 575 | child = node.children[i]
|
| 576 |
|
| 577 | # TODO: determine these locations at parse time?
|
| 578 | pipe_locs.append(loc.Command(child))
|
| 579 |
|
| 580 | p = self._MakeProcess(child, True, self.exec_opts.errtrace())
|
| 581 | p.Init_ParentPipeline(pi)
|
| 582 | pi.Add(p)
|
| 583 |
|
| 584 | last_child = node.children[n - 1]
|
| 585 | # Last piece of code is in THIS PROCESS. 'echo foo | read line; echo $line'
|
| 586 | pi.AddLast((self.cmd_ev, last_child))
|
| 587 | pipe_locs.append(loc.Command(last_child))
|
| 588 |
|
| 589 | with dev.ctx_Tracer(self.tracer, 'pipeline', None):
|
| 590 | pi.StartPipeline(self.waiter)
|
| 591 | self.fg_pipeline = pi
|
| 592 | status_out.pipe_status = pi.RunLastPart(self.waiter, self.fd_state)
|
| 593 | self.fg_pipeline = None # clear in case we didn't end up forking
|
| 594 |
|
| 595 | status_out.pipe_locs = pipe_locs
|
| 596 |
|
| 597 | def RunSubshell(self, node):
|
| 598 | # type: (command_t) -> int
|
| 599 | p = self._MakeProcess(node, True, self.exec_opts.errtrace())
|
| 600 | if self.job_control.Enabled():
|
| 601 | p.AddStateChange(process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 602 |
|
| 603 | return p.RunProcess(self.waiter, trace.ForkWait)
|
| 604 |
|
| 605 | def CaptureStdout(self, node):
|
| 606 | # type: (command_t) -> Tuple[int, str]
|
| 607 |
|
| 608 | p = self._MakeProcess(node, self.exec_opts.inherit_errexit(),
|
| 609 | self.exec_opts.errtrace())
|
| 610 | # Shell quirk: Command subs remain part of the shell's process group, so we
|
| 611 | # don't use p.AddStateChange(process.SetPgid(...))
|
| 612 |
|
| 613 | r, w = posix.pipe()
|
| 614 | p.AddStateChange(process.StdoutToPipe(r, w))
|
| 615 |
|
| 616 | p.StartProcess(trace.CommandSub)
|
| 617 | #log('Command sub started %d', pid)
|
| 618 |
|
| 619 | chunks = [] # type: List[str]
|
| 620 | posix.close(w) # not going to write
|
| 621 | while True:
|
| 622 | n, err_num = pyos.Read(r, 4096, chunks)
|
| 623 |
|
| 624 | if n == 0: # EOF
|
| 625 | break
|
| 626 |
|
| 627 | elif n > 0:
|
| 628 | # common shell behavior: remove NUL from stdout
|
| 629 | chunks[-1] = chunks[-1].replace('\0', '')
|
| 630 |
|
| 631 | else: # n < 0
|
| 632 | if err_num == EINTR:
|
| 633 | pass # retry
|
| 634 | else:
|
| 635 | # Like the top level IOError handler
|
| 636 | e_die_status(
|
| 637 | 2,
|
| 638 | 'Oils I/O error (read): %s' % posix.strerror(err_num))
|
| 639 |
|
| 640 | posix.close(r)
|
| 641 |
|
| 642 | status = p.Wait(self.waiter)
|
| 643 | stdout_str = ''.join(chunks).rstrip('\n')
|
| 644 |
|
| 645 | return status, stdout_str
|
| 646 |
|
| 647 | def Capture3(self, node):
|
| 648 | # type: (command_t) -> Tuple[int, str, str]
|
| 649 |
|
| 650 | p = self._MakeProcess(node, self.exec_opts.inherit_errexit(),
|
| 651 | self.exec_opts.errtrace())
|
| 652 | # Shell quirk: Command subs remain part of the shell's process group, so we
|
| 653 | # don't use p.AddStateChange(process.SetPgid(...))
|
| 654 |
|
| 655 | stdout_fd, w = posix.pipe()
|
| 656 | stderr_fd, w2 = posix.pipe()
|
| 657 | p.AddStateChange(process.StdoutToPipe(stdout_fd, w))
|
| 658 | p.AddStateChange(process.StderrToPipe(stderr_fd, w2))
|
| 659 |
|
| 660 | p.StartProcess(trace.CommandSub)
|
| 661 | #log('Command sub started %d', pid)
|
| 662 |
|
| 663 | stdout_chunks = [] # type: List[str]
|
| 664 | stderr_chunks = [] # type: List[str]
|
| 665 | posix.close(w) # not going to write
|
| 666 | posix.close(w2) # not going to write
|
| 667 | open_fds = [stdout_fd, stderr_fd]
|
| 668 | while True:
|
| 669 | fds = pyos.WaitForReading(open_fds)
|
| 670 |
|
| 671 | # zero outputs mean something went wrong
|
| 672 | if len(fds) == 0:
|
| 673 | break
|
| 674 |
|
| 675 | for fd in fds:
|
| 676 | if fd == stdout_fd:
|
| 677 | n, err_num = pyos.Read(fd, 4096, stdout_chunks)
|
| 678 | else:
|
| 679 | n, err_num = pyos.Read(fd, 4096, stderr_chunks)
|
| 680 | if n < 0:
|
| 681 | if err_num == EINTR:
|
| 682 | pass # retry
|
| 683 | else:
|
| 684 | # Like the top level IOError handler
|
| 685 | e_die_status(
|
| 686 | 2, 'Oils I/O error (read): %s' %
|
| 687 | posix.strerror(err_num))
|
| 688 | elif n == 0: # EOF
|
| 689 | open_fds.remove(fd)
|
| 690 |
|
| 691 | if len(open_fds) == 0:
|
| 692 | break
|
| 693 |
|
| 694 | posix.close(stdout_fd)
|
| 695 | posix.close(stderr_fd)
|
| 696 |
|
| 697 | status = p.Wait(self.waiter)
|
| 698 | stdout_str = ''.join(stdout_chunks)
|
| 699 | stderr_str = ''.join(stderr_chunks)
|
| 700 |
|
| 701 | return status, stdout_str, stderr_str
|
| 702 |
|
| 703 | def RunCommandSub(self, cs_part):
|
| 704 | # type: (CommandSub) -> str
|
| 705 |
|
| 706 | if not self.exec_opts._allow_command_sub():
|
| 707 | # _allow_command_sub is used in two places. Only one of them turns
|
| 708 | # off _allow_process_sub
|
| 709 | if not self.exec_opts._allow_process_sub():
|
| 710 | why = "status wouldn't be checked (strict_errexit)"
|
| 711 | else:
|
| 712 | why = 'eval_unsafe_arith is off'
|
| 713 |
|
| 714 | e_die("Command subs not allowed here because %s" % why,
|
| 715 | loc.WordPart(cs_part))
|
| 716 |
|
| 717 | node = cs_part.child
|
| 718 |
|
| 719 | # Hack for weird $(<file) construct
|
| 720 | if node.tag() == command_e.Redirect:
|
| 721 | redir_node = cast(command.Redirect, node)
|
| 722 | # Detect '< file'
|
| 723 | if (len(redir_node.redirects) == 1 and
|
| 724 | redir_node.redirects[0].op.id == Id.Redir_Less and
|
| 725 | redir_node.child.tag() == command_e.NoOp):
|
| 726 |
|
| 727 | # Change it to __cat < file.
|
| 728 | # TODO: could be 'internal cat' (issue #1013)
|
| 729 | tok = lexer.DummyToken(Id.Lit_Chars, '__cat')
|
| 730 | cat_word = CompoundWord([tok])
|
| 731 |
|
| 732 | # Blame < because __cat has no location
|
| 733 | blame_tok = redir_node.redirects[0].op
|
| 734 | simple = command.Simple(blame_tok, [], [cat_word], None, None,
|
| 735 | False)
|
| 736 |
|
| 737 | # MUTATE redir node so it's like $(<file _cat)
|
| 738 | redir_node.child = simple
|
| 739 |
|
| 740 | status, stdout_str = self.CaptureStdout(node)
|
| 741 |
|
| 742 | # OSH has the concept of aborting in the middle of a WORD. We're not
|
| 743 | # waiting until the command is over!
|
| 744 | if self.exec_opts.command_sub_errexit():
|
| 745 | if status != 0:
|
| 746 | msg = 'Command Sub exited with status %d' % status
|
| 747 | raise error.ErrExit(status, msg, loc.WordPart(cs_part))
|
| 748 |
|
| 749 | else:
|
| 750 | # Set a flag so we check errexit at the same time as bash. Example:
|
| 751 | #
|
| 752 | # a=$(false)
|
| 753 | # echo foo # no matter what comes here, the flag is reset
|
| 754 | #
|
| 755 | # Set ONLY until this command node has finished executing.
|
| 756 |
|
| 757 | # HACK: move this
|
| 758 | self.cmd_ev.check_command_sub_status = True
|
| 759 | self.mem.SetLastStatus(status)
|
| 760 |
|
| 761 | # Runtime errors test case: # $("echo foo > $@")
|
| 762 | # Why rstrip()?
|
| 763 | # https://unix.stackexchange.com/questions/17747/why-does-shell-command-substitution-gobble-up-a-trailing-newline-char
|
| 764 | return stdout_str
|
| 765 |
|
| 766 | def RunProcessSub(self, cs_part):
|
| 767 | # type: (CommandSub) -> str
|
| 768 | """Process sub creates a forks a process connected to a pipe.
|
| 769 |
|
| 770 | The pipe is typically passed to another process via a /dev/fd/$FD path.
|
| 771 |
|
| 772 | Life cycle of a process substitution:
|
| 773 |
|
| 774 | 1. Start with this code
|
| 775 |
|
| 776 | diff <(seq 3) <(seq 4)
|
| 777 |
|
| 778 | 2. To evaluate the command line, we evaluate every word. The
|
| 779 | NormalWordEvaluator this method, RunProcessSub(), which does 3 things:
|
| 780 |
|
| 781 | a. Create a pipe(), getting r and w
|
| 782 | b. Starts the seq process, which inherits r and w
|
| 783 | It has a StdoutToPipe() redirect, which means that it dup2(w, 1)
|
| 784 | and close(r)
|
| 785 | c. Close the w FD, because neither the shell or 'diff' will write to it.
|
| 786 | However we must retain 'r', because 'diff' hasn't opened /dev/fd yet!
|
| 787 | d. We evaluate <(seq 3) to /dev/fd/$r, so "diff" can read from it
|
| 788 |
|
| 789 | 3. Now we're done evaluating every word, so we know the command line of
|
| 790 | diff, which looks like
|
| 791 |
|
| 792 | diff /dev/fd/64 /dev/fd/65
|
| 793 |
|
| 794 | Those are the FDs for the read ends of the pipes we created.
|
| 795 |
|
| 796 | 4. diff inherits a copy of the read end of bot pipes. But it actually
|
| 797 | calls open() both files passed as argv. (I think this is fine.)
|
| 798 |
|
| 799 | 5. wait() for the diff process.
|
| 800 |
|
| 801 | 6. The shell closes both the read ends of both pipes. Neither us or
|
| 802 | 'diffd' will read again.
|
| 803 |
|
| 804 | 7. The shell waits for both 'seq' processes.
|
| 805 |
|
| 806 | Related:
|
| 807 | shopt -s process_sub_fail
|
| 808 | _process_sub_status
|
| 809 | """
|
| 810 | cs_loc = loc.WordPart(cs_part)
|
| 811 |
|
| 812 | if not self.exec_opts._allow_process_sub():
|
| 813 | e_die(
|
| 814 | "Process subs not allowed here because status wouldn't be checked (strict_errexit)",
|
| 815 | cs_loc)
|
| 816 |
|
| 817 | p = self._MakeProcess(cs_part.child, True, self.exec_opts.errtrace())
|
| 818 |
|
| 819 | r, w = posix.pipe()
|
| 820 | #log('pipe = %d, %d', r, w)
|
| 821 |
|
| 822 | op_id = cs_part.left_token.id
|
| 823 | if op_id == Id.Left_ProcSubIn:
|
| 824 | # Example: cat < <(head foo.txt)
|
| 825 | #
|
| 826 | # The head process should write its stdout to a pipe.
|
| 827 | redir = process.StdoutToPipe(r,
|
| 828 | w) # type: process.ChildStateChange
|
| 829 |
|
| 830 | elif op_id == Id.Left_ProcSubOut:
|
| 831 | # Example: head foo.txt > >(tac)
|
| 832 | #
|
| 833 | # The tac process should read its stdin from a pipe.
|
| 834 |
|
| 835 | # Note: this example sometimes requires you to hit "enter" in bash and
|
| 836 | # zsh. WHy?
|
| 837 | redir = process.StdinFromPipe(r, w)
|
| 838 |
|
| 839 | else:
|
| 840 | raise AssertionError()
|
| 841 |
|
| 842 | p.AddStateChange(redir)
|
| 843 |
|
| 844 | if self.job_control.Enabled():
|
| 845 | p.AddStateChange(process.SetPgid(process.OWN_LEADER, self.tracer))
|
| 846 |
|
| 847 | # Fork, letting the child inherit the pipe file descriptors.
|
| 848 | p.StartProcess(trace.ProcessSub)
|
| 849 |
|
| 850 | ps_frame = self.process_sub_stack[-1]
|
| 851 |
|
| 852 | # Note: bash never waits() on the process, but zsh does. The calling
|
| 853 | # program needs to read() before we can wait, e.g.
|
| 854 | # diff <(sort left.txt) <(sort right.txt)
|
| 855 |
|
| 856 | # After forking, close the end of the pipe we're not using.
|
| 857 | if op_id == Id.Left_ProcSubIn:
|
| 858 | posix.close(w) # cat < <(head foo.txt)
|
| 859 | ps_frame.Append(p, r, cs_loc) # close later
|
| 860 | elif op_id == Id.Left_ProcSubOut:
|
| 861 | posix.close(r)
|
| 862 | #log('Left_ProcSubOut closed %d', r)
|
| 863 | ps_frame.Append(p, w, cs_loc) # close later
|
| 864 | else:
|
| 865 | raise AssertionError()
|
| 866 |
|
| 867 | # Is /dev Linux-specific?
|
| 868 | if op_id == Id.Left_ProcSubIn:
|
| 869 | return '/dev/fd/%d' % r
|
| 870 |
|
| 871 | elif op_id == Id.Left_ProcSubOut:
|
| 872 | return '/dev/fd/%d' % w
|
| 873 |
|
| 874 | else:
|
| 875 | raise AssertionError()
|
| 876 |
|
| 877 | def PushRedirects(self, redirects, err_out):
|
| 878 | # type: (List[RedirValue], List[error.IOError_OSError]) -> None
|
| 879 | if len(redirects) == 0: # Optimized to avoid allocs
|
| 880 | return
|
| 881 | self.fd_state.Push(redirects, err_out)
|
| 882 |
|
| 883 | def PopRedirects(self, num_redirects, err_out):
|
| 884 | # type: (int, List[error.IOError_OSError]) -> None
|
| 885 | if num_redirects == 0: # Optimized to avoid allocs
|
| 886 | return
|
| 887 | self.fd_state.Pop(err_out)
|
| 888 |
|
| 889 | def PushProcessSub(self):
|
| 890 | # type: () -> None
|
| 891 | if len(self.clean_frame_pool):
|
| 892 | # Optimized to avoid allocs
|
| 893 | new_frame = self.clean_frame_pool.pop()
|
| 894 | else:
|
| 895 | new_frame = _ProcessSubFrame()
|
| 896 | self.process_sub_stack.append(new_frame)
|
| 897 |
|
| 898 | def PopProcessSub(self, compound_st):
|
| 899 | # type: (StatusArray) -> None
|
| 900 | """This method is called by a context manager, which means we always
|
| 901 | wait() on the way out, which I think is the right thing.
|
| 902 |
|
| 903 | We don't always set _process_sub_status, e.g. if some fatal
|
| 904 | error occurs first, but we always wait.
|
| 905 | """
|
| 906 | frame = self.process_sub_stack.pop()
|
| 907 | if frame.WasModified():
|
| 908 | frame.MaybeWaitOnProcessSubs(self.waiter, compound_st)
|
| 909 | else:
|
| 910 | # Optimized to avoid allocs
|
| 911 | self.clean_frame_pool.append(frame)
|
| 912 |
|
| 913 | # Note: the 3 lists in _ProcessSubFrame are hot in our profiles. It would
|
| 914 | # be nice to somehow "destroy" them here, rather than letting them become
|
| 915 | # garbage that needs to be traced.
|
| 916 |
|
| 917 | # The CommandEvaluator could have a ProcessSubStack, which supports Push(),
|
| 918 | # Pop(), and Top() of VALUES rather than GC objects?
|
| 919 |
|
| 920 |
|
| 921 | # vim: sw=4
|