| 1 | #!/usr/bin/env python2
|
| 2 | """
|
| 3 | location.py - Library to get source location info from nodes.
|
| 4 |
|
| 5 | This makes syntax errors nicer.
|
| 6 | """
|
| 7 | from __future__ import print_function
|
| 8 |
|
| 9 | from _devbuild.gen.syntax_asdl import (
|
| 10 | expr,
|
| 11 | expr_t,
|
| 12 | expr_e,
|
| 13 | loc,
|
| 14 | loc_t,
|
| 15 | loc_e,
|
| 16 | command,
|
| 17 | command_e,
|
| 18 | command_t,
|
| 19 | sh_lhs,
|
| 20 | sh_lhs_e,
|
| 21 | sh_lhs_t,
|
| 22 | word,
|
| 23 | word_e,
|
| 24 | word_t,
|
| 25 | word_part,
|
| 26 | word_part_e,
|
| 27 | word_part_t,
|
| 28 | CompoundWord,
|
| 29 | Token,
|
| 30 | SimpleVarSub,
|
| 31 | ShArrayLiteral,
|
| 32 | SingleQuoted,
|
| 33 | DoubleQuoted,
|
| 34 | CommandSub,
|
| 35 | BracedVarSub,
|
| 36 | BraceGroup,
|
| 37 | Subscript,
|
| 38 | Attribute,
|
| 39 | arith_expr,
|
| 40 | arith_expr_e,
|
| 41 | arith_expr_t,
|
| 42 | Eggex,
|
| 43 | )
|
| 44 | from _devbuild.gen.value_asdl import LeftName
|
| 45 | from mycpp.mylib import log
|
| 46 | from mycpp.mylib import tagswitch
|
| 47 |
|
| 48 | _ = log
|
| 49 |
|
| 50 | from typing import cast, Optional
|
| 51 |
|
| 52 |
|
| 53 | def LName(name):
|
| 54 | # type: (str) -> LeftName
|
| 55 | """Wrapper for LeftName() with location.
|
| 56 |
|
| 57 | TODO: add locations and remove this.
|
| 58 | """
|
| 59 | return LeftName(name, loc.Missing)
|
| 60 |
|
| 61 |
|
| 62 | def TokenFor(loc_):
|
| 63 | # type: (loc_t) -> Optional[Token]
|
| 64 | """Given a location, get a Token.
|
| 65 |
|
| 66 | This is useful because a Token points to a single line.
|
| 67 | """
|
| 68 | UP_location = loc_
|
| 69 | with tagswitch(loc_) as case:
|
| 70 | if case(loc_e.Missing):
|
| 71 | return None
|
| 72 |
|
| 73 | elif case(loc_e.Token):
|
| 74 | tok = cast(Token, UP_location)
|
| 75 | if tok:
|
| 76 | return tok
|
| 77 | else:
|
| 78 | return None
|
| 79 |
|
| 80 | elif case(loc_e.ArgWord):
|
| 81 | w = cast(CompoundWord, UP_location)
|
| 82 | return LeftTokenForWord(w)
|
| 83 |
|
| 84 | elif case(loc_e.WordPart):
|
| 85 | loc_ = cast(loc.WordPart, UP_location)
|
| 86 | if loc_.p:
|
| 87 | return LeftTokenForWordPart(loc_.p)
|
| 88 | else:
|
| 89 | return None
|
| 90 |
|
| 91 | elif case(loc_e.Word):
|
| 92 | loc_ = cast(loc.Word, UP_location)
|
| 93 | if loc_.w:
|
| 94 | return LeftTokenForWord(loc_.w)
|
| 95 | else:
|
| 96 | return None
|
| 97 |
|
| 98 | elif case(loc_e.Command):
|
| 99 | loc_ = cast(loc.Command, UP_location)
|
| 100 | if loc_.c:
|
| 101 | return TokenForCommand(loc_.c)
|
| 102 | else:
|
| 103 | return None
|
| 104 |
|
| 105 | elif case(loc_e.Arith):
|
| 106 | loc_ = cast(loc.Arith, UP_location)
|
| 107 | if loc_.a:
|
| 108 | return TokenForArith(loc_.a)
|
| 109 | else:
|
| 110 | return None
|
| 111 |
|
| 112 | else:
|
| 113 | raise AssertionError()
|
| 114 |
|
| 115 | raise AssertionError()
|
| 116 |
|
| 117 |
|
| 118 | def TokenForCommand(node):
|
| 119 | # type: (command_t) -> Optional[Token]
|
| 120 | """Used directly in _CheckStatus()"""
|
| 121 | UP_node = node # type: command_t
|
| 122 | tag = node.tag()
|
| 123 |
|
| 124 | if tag == command_e.Sentence:
|
| 125 | node = cast(command.Sentence, UP_node)
|
| 126 | #log("node.child %s", node.child)
|
| 127 | return node.terminator # & or ;
|
| 128 |
|
| 129 | if tag == command_e.Simple:
|
| 130 | node = cast(command.Simple, UP_node)
|
| 131 | return node.blame_tok
|
| 132 |
|
| 133 | if tag == command_e.ShAssignment:
|
| 134 | node = cast(command.ShAssignment, UP_node)
|
| 135 | return node.left
|
| 136 |
|
| 137 | if tag == command_e.Pipeline:
|
| 138 | node = cast(command.Pipeline, UP_node)
|
| 139 | if len(node.ops):
|
| 140 | return node.ops[0] # first | or |&
|
| 141 | else:
|
| 142 | assert node.negated is not None
|
| 143 | return node.negated # ! false
|
| 144 |
|
| 145 | if tag == command_e.AndOr:
|
| 146 | node = cast(command.AndOr, UP_node)
|
| 147 | return node.ops[0] # first && or ||
|
| 148 |
|
| 149 | if tag == command_e.DoGroup:
|
| 150 | node = cast(command.DoGroup, UP_node)
|
| 151 | return node.left # 'do' token
|
| 152 | if tag == command_e.BraceGroup:
|
| 153 | node = cast(BraceGroup, UP_node)
|
| 154 | return node.left # { token
|
| 155 | if tag == command_e.Subshell:
|
| 156 | node = cast(command.Subshell, UP_node)
|
| 157 | return node.left # ( token
|
| 158 |
|
| 159 | if tag == command_e.WhileUntil:
|
| 160 | node = cast(command.WhileUntil, UP_node)
|
| 161 | return node.keyword # while
|
| 162 | if tag == command_e.If:
|
| 163 | node = cast(command.If, UP_node)
|
| 164 | return node.if_kw
|
| 165 | if tag == command_e.Case:
|
| 166 | node = cast(command.Case, UP_node)
|
| 167 | return node.case_kw
|
| 168 | if tag == command_e.TimeBlock:
|
| 169 | node = cast(command.TimeBlock, UP_node)
|
| 170 | return node.keyword
|
| 171 |
|
| 172 | # We never have this case?
|
| 173 | #if node.tag == command_e.CommandList:
|
| 174 | # pass
|
| 175 |
|
| 176 | return None
|
| 177 |
|
| 178 |
|
| 179 | def TokenForArith(node):
|
| 180 | # type: (arith_expr_t) -> Optional[Token]
|
| 181 | UP_node = node
|
| 182 | with tagswitch(node) as case:
|
| 183 | if case(arith_expr_e.VarSub):
|
| 184 | vsub = cast(Token, UP_node)
|
| 185 | # $(( x ))
|
| 186 | return vsub
|
| 187 |
|
| 188 | elif case(arith_expr_e.Word):
|
| 189 | w = cast(CompoundWord, UP_node)
|
| 190 | return LeftTokenForWord(w)
|
| 191 |
|
| 192 | elif case(arith_expr_e.Unary):
|
| 193 | node = cast(arith_expr.Unary, UP_node)
|
| 194 | return TokenForArith(node.child)
|
| 195 |
|
| 196 | elif case(arith_expr_e.Binary):
|
| 197 | node = cast(arith_expr.Binary, UP_node)
|
| 198 | return TokenForArith(node.op)
|
| 199 |
|
| 200 | elif case(arith_expr_e.TernaryOp):
|
| 201 | node = cast(arith_expr.TernaryOp, UP_node)
|
| 202 |
|
| 203 | # TODO: should blame op
|
| 204 | # blaming cond is arbitrary, but better than nothing
|
| 205 | return TokenForArith(node.cond)
|
| 206 |
|
| 207 | return None
|
| 208 |
|
| 209 |
|
| 210 | def LeftTokenForWordPart(part):
|
| 211 | # type: (word_part_t) -> Optional[Token]
|
| 212 | UP_part = part
|
| 213 | with tagswitch(part) as case:
|
| 214 | if case(word_part_e.ShArrayLiteral):
|
| 215 | part = cast(ShArrayLiteral, UP_part)
|
| 216 | return part.left
|
| 217 |
|
| 218 | elif case(word_part_e.BashAssocLiteral):
|
| 219 | part = cast(word_part.BashAssocLiteral, UP_part)
|
| 220 | return part.left
|
| 221 |
|
| 222 | elif case(word_part_e.Literal):
|
| 223 | tok = cast(Token, UP_part)
|
| 224 | return tok
|
| 225 |
|
| 226 | elif case(word_part_e.EscapedLiteral):
|
| 227 | part = cast(word_part.EscapedLiteral, UP_part)
|
| 228 | return part.token
|
| 229 |
|
| 230 | elif case(word_part_e.SingleQuoted):
|
| 231 | part = cast(SingleQuoted, UP_part)
|
| 232 | return part.left
|
| 233 |
|
| 234 | elif case(word_part_e.DoubleQuoted):
|
| 235 | part = cast(DoubleQuoted, UP_part)
|
| 236 | return part.left
|
| 237 |
|
| 238 | elif case(word_part_e.SimpleVarSub):
|
| 239 | part = cast(SimpleVarSub, UP_part)
|
| 240 | return part.tok
|
| 241 |
|
| 242 | elif case(word_part_e.BracedVarSub):
|
| 243 | part = cast(BracedVarSub, UP_part)
|
| 244 | return part.left
|
| 245 |
|
| 246 | elif case(word_part_e.CommandSub):
|
| 247 | part = cast(CommandSub, UP_part)
|
| 248 | return part.left_token
|
| 249 |
|
| 250 | elif case(word_part_e.TildeSub):
|
| 251 | part = cast(word_part.TildeSub, UP_part)
|
| 252 | return part.left
|
| 253 |
|
| 254 | elif case(word_part_e.ArithSub):
|
| 255 | part = cast(word_part.ArithSub, UP_part)
|
| 256 | return part.left
|
| 257 |
|
| 258 | elif case(word_part_e.ExtGlob):
|
| 259 | part = cast(word_part.ExtGlob, UP_part)
|
| 260 | return part.op
|
| 261 |
|
| 262 | elif case(word_part_e.BracedRange):
|
| 263 | part = cast(word_part.BracedRange, UP_part)
|
| 264 | return part.blame_tok
|
| 265 |
|
| 266 | elif case(word_part_e.BracedTuple):
|
| 267 | part = cast(word_part.BracedTuple, UP_part)
|
| 268 | # TODO: Derive token from part.words[0]
|
| 269 | return None
|
| 270 |
|
| 271 | elif case(word_part_e.Splice):
|
| 272 | part = cast(word_part.Splice, UP_part)
|
| 273 | return part.blame_tok
|
| 274 |
|
| 275 | elif case(word_part_e.ExprSub):
|
| 276 | part = cast(word_part.ExprSub, UP_part)
|
| 277 | return part.left # $[
|
| 278 |
|
| 279 | else:
|
| 280 | raise AssertionError(part.tag())
|
| 281 |
|
| 282 |
|
| 283 | def _RightTokenForWordPart(part):
|
| 284 | # type: (word_part_t) -> Token
|
| 285 | UP_part = part
|
| 286 | with tagswitch(part) as case:
|
| 287 | if case(word_part_e.ShArrayLiteral):
|
| 288 | part = cast(ShArrayLiteral, UP_part)
|
| 289 | return part.right
|
| 290 |
|
| 291 | elif case(word_part_e.BashAssocLiteral):
|
| 292 | part = cast(word_part.BashAssocLiteral, UP_part)
|
| 293 | return part.right
|
| 294 |
|
| 295 | elif case(word_part_e.Literal):
|
| 296 | tok = cast(Token, UP_part)
|
| 297 | # Just use the token
|
| 298 | return tok
|
| 299 |
|
| 300 | elif case(word_part_e.EscapedLiteral):
|
| 301 | part = cast(word_part.EscapedLiteral, UP_part)
|
| 302 | return part.token
|
| 303 |
|
| 304 | elif case(word_part_e.SingleQuoted):
|
| 305 | part = cast(SingleQuoted, UP_part)
|
| 306 | return part.right # right '
|
| 307 |
|
| 308 | elif case(word_part_e.DoubleQuoted):
|
| 309 | part = cast(DoubleQuoted, UP_part)
|
| 310 | return part.right # right "
|
| 311 |
|
| 312 | elif case(word_part_e.SimpleVarSub):
|
| 313 | part = cast(SimpleVarSub, UP_part)
|
| 314 | # left and right are the same for $myvar
|
| 315 | return part.tok
|
| 316 |
|
| 317 | elif case(word_part_e.BracedVarSub):
|
| 318 | part = cast(BracedVarSub, UP_part)
|
| 319 | return part.right
|
| 320 |
|
| 321 | elif case(word_part_e.CommandSub):
|
| 322 | part = cast(CommandSub, UP_part)
|
| 323 | return part.right
|
| 324 |
|
| 325 | elif case(word_part_e.TildeSub):
|
| 326 | part = cast(word_part.TildeSub, UP_part)
|
| 327 | if part.name is not None:
|
| 328 | return part.name # ~bob/
|
| 329 | else:
|
| 330 | return part.left # ~/
|
| 331 |
|
| 332 | elif case(word_part_e.ArithSub):
|
| 333 | part = cast(word_part.ArithSub, UP_part)
|
| 334 | return part.right
|
| 335 |
|
| 336 | elif case(word_part_e.ExtGlob):
|
| 337 | part = cast(word_part.ExtGlob, UP_part)
|
| 338 | return part.right
|
| 339 |
|
| 340 | elif case(word_part_e.BracedRange):
|
| 341 | part = cast(word_part.BracedRange, UP_part)
|
| 342 | return part.blame_tok
|
| 343 |
|
| 344 | elif case(word_part_e.BracedTuple):
|
| 345 | part = cast(word_part.BracedTuple, UP_part)
|
| 346 | # TODO: Derive token from part.words[0]
|
| 347 | return None
|
| 348 |
|
| 349 | elif case(word_part_e.Splice):
|
| 350 | part = cast(word_part.Splice, UP_part)
|
| 351 | return part.blame_tok
|
| 352 |
|
| 353 | elif case(word_part_e.ExprSub):
|
| 354 | part = cast(word_part.ExprSub, UP_part)
|
| 355 | return part.right
|
| 356 |
|
| 357 | else:
|
| 358 | raise AssertionError(part.tag())
|
| 359 |
|
| 360 |
|
| 361 | def LeftTokenForCompoundWord(w):
|
| 362 | # type: (CompoundWord) -> Optional[Token]
|
| 363 | if len(w.parts):
|
| 364 | return LeftTokenForWordPart(w.parts[0])
|
| 365 | else:
|
| 366 | # This is possible for empty brace sub alternative {a,b,}
|
| 367 | return None
|
| 368 |
|
| 369 |
|
| 370 | def LeftTokenForWord(w):
|
| 371 | # type: (word_t) -> Optional[Token]
|
| 372 | if w is None:
|
| 373 | return None # e.g. builtin_bracket word.String() EOF
|
| 374 |
|
| 375 | UP_w = w
|
| 376 | with tagswitch(w) as case:
|
| 377 | if case(word_e.Compound):
|
| 378 | w = cast(CompoundWord, UP_w)
|
| 379 | return LeftTokenForCompoundWord(w)
|
| 380 |
|
| 381 | elif case(word_e.Operator):
|
| 382 | tok = cast(Token, UP_w)
|
| 383 | return tok
|
| 384 |
|
| 385 | elif case(word_e.BracedTree):
|
| 386 | w = cast(word.BracedTree, UP_w)
|
| 387 | # This should always have one part?
|
| 388 | return LeftTokenForWordPart(w.parts[0])
|
| 389 |
|
| 390 | elif case(word_e.String):
|
| 391 | w = cast(word.String, UP_w)
|
| 392 | # See _StringWordEmitter in osh/builtin_bracket.py
|
| 393 | return LeftTokenForWord(w.blame_loc)
|
| 394 |
|
| 395 | else:
|
| 396 | raise AssertionError(w.tag())
|
| 397 |
|
| 398 | raise AssertionError('for -Wreturn-type in C++')
|
| 399 |
|
| 400 |
|
| 401 | def RightTokenForWord(w):
|
| 402 | # type: (word_t) -> Token
|
| 403 | """Used for alias expansion and history substitution.
|
| 404 |
|
| 405 | and here doc delimiters?
|
| 406 | """
|
| 407 | UP_w = w
|
| 408 | with tagswitch(w) as case:
|
| 409 | if case(word_e.Compound):
|
| 410 | w = cast(CompoundWord, UP_w)
|
| 411 | if len(w.parts):
|
| 412 | end = w.parts[-1]
|
| 413 | return _RightTokenForWordPart(end)
|
| 414 | else:
|
| 415 | # This is possible for empty brace sub alternative {a,b,}
|
| 416 | return None
|
| 417 |
|
| 418 | elif case(word_e.Operator):
|
| 419 | tok = cast(Token, UP_w)
|
| 420 | return tok
|
| 421 |
|
| 422 | elif case(word_e.BracedTree):
|
| 423 | w = cast(word.BracedTree, UP_w)
|
| 424 | # Note: this case may be unused
|
| 425 | return _RightTokenForWordPart(w.parts[-1])
|
| 426 |
|
| 427 | elif case(word_e.String):
|
| 428 | w = cast(word.String, UP_w)
|
| 429 | # Note: this case may be unused
|
| 430 | return RightTokenForWord(w.blame_loc)
|
| 431 |
|
| 432 | else:
|
| 433 | raise AssertionError(w.tag())
|
| 434 |
|
| 435 | raise AssertionError('for -Wreturn-type in C++')
|
| 436 |
|
| 437 |
|
| 438 | def TokenForLhsExpr(node):
|
| 439 | # type: (sh_lhs_t) -> Token
|
| 440 | """Currently unused?
|
| 441 |
|
| 442 | Will be useful for translating YSH assignment
|
| 443 | """
|
| 444 | # This switch is annoying but we don't have inheritance from the sum type
|
| 445 | # (because of diamond issue). We might change the schema later, which maeks
|
| 446 | # it moot. See the comment in frontend/syntax.asdl.
|
| 447 | UP_node = node
|
| 448 | with tagswitch(node) as case:
|
| 449 | if case(sh_lhs_e.Name):
|
| 450 | node = cast(sh_lhs.Name, UP_node)
|
| 451 | return node.left
|
| 452 | elif case(sh_lhs_e.IndexedName):
|
| 453 | node = cast(sh_lhs.IndexedName, UP_node)
|
| 454 | return node.left
|
| 455 | else:
|
| 456 | # Should not see UnparsedIndex
|
| 457 | raise AssertionError()
|
| 458 |
|
| 459 | raise AssertionError()
|
| 460 |
|
| 461 |
|
| 462 | # TODO: Token instead of loc_t once all cases are implemented
|
| 463 | def TokenForExpr(node):
|
| 464 | # type: (expr_t) -> loc_t
|
| 465 | """Returns the token associated with the given expression."""
|
| 466 |
|
| 467 | UP_node = node # type: expr_t
|
| 468 | with tagswitch(node) as case:
|
| 469 | if case(expr_e.Const):
|
| 470 | node = cast(expr.Const, UP_node)
|
| 471 | return node.c
|
| 472 |
|
| 473 | elif case(expr_e.Var):
|
| 474 | node = cast(expr.Var, UP_node)
|
| 475 | return node.left
|
| 476 |
|
| 477 | elif case(expr_e.Place):
|
| 478 | node = cast(expr.Place, UP_node)
|
| 479 | return node.blame_tok
|
| 480 |
|
| 481 | elif case(expr_e.CommandSub):
|
| 482 | node = cast(CommandSub, UP_node)
|
| 483 | return node.left_token
|
| 484 |
|
| 485 | elif case(expr_e.ShArrayLiteral):
|
| 486 | node = cast(ShArrayLiteral, UP_node)
|
| 487 | return node.left
|
| 488 |
|
| 489 | elif case(expr_e.DoubleQuoted):
|
| 490 | node = cast(DoubleQuoted, UP_node)
|
| 491 | return node.left
|
| 492 |
|
| 493 | elif case(expr_e.SingleQuoted):
|
| 494 | node = cast(SingleQuoted, UP_node)
|
| 495 | return node.left
|
| 496 |
|
| 497 | elif case(expr_e.BracedVarSub):
|
| 498 | node = cast(BracedVarSub, UP_node)
|
| 499 | return node.left
|
| 500 |
|
| 501 | elif case(expr_e.SimpleVarSub):
|
| 502 | node = cast(SimpleVarSub, UP_node)
|
| 503 | return node.tok
|
| 504 |
|
| 505 | elif case(expr_e.Unary):
|
| 506 | node = cast(expr.Unary, UP_node)
|
| 507 | return node.op
|
| 508 |
|
| 509 | elif case(expr_e.Binary):
|
| 510 | node = cast(expr.Binary, UP_node)
|
| 511 | return node.op
|
| 512 |
|
| 513 | elif case(expr_e.Slice):
|
| 514 | node = cast(expr.Slice, UP_node)
|
| 515 | return node.op
|
| 516 |
|
| 517 | elif case(expr_e.Range):
|
| 518 | node = cast(expr.Range, UP_node)
|
| 519 | return node.op
|
| 520 |
|
| 521 | elif case(expr_e.Compare):
|
| 522 | node = cast(expr.Compare, UP_node)
|
| 523 | # TODO: use operator instead?
|
| 524 | return TokenForExpr(node.left)
|
| 525 |
|
| 526 | elif case(expr_e.IfExp):
|
| 527 | # TODO
|
| 528 | return loc.Missing
|
| 529 |
|
| 530 | elif case(expr_e.List):
|
| 531 | node = cast(expr.List, UP_node)
|
| 532 | return node.left
|
| 533 |
|
| 534 | elif case(expr_e.Tuple):
|
| 535 | node = cast(expr.Tuple, UP_node)
|
| 536 | return node.left
|
| 537 |
|
| 538 | elif case(expr_e.Dict):
|
| 539 | node = cast(expr.Dict, UP_node)
|
| 540 | return node.left
|
| 541 |
|
| 542 | elif case(expr_e.ListComp):
|
| 543 | node = cast(expr.ListComp, UP_node)
|
| 544 | return node.left
|
| 545 |
|
| 546 | elif case(expr_e.GeneratorExp):
|
| 547 | # TODO
|
| 548 | return loc.Missing
|
| 549 |
|
| 550 | elif case(expr_e.Lambda):
|
| 551 | # TODO
|
| 552 | return loc.Missing
|
| 553 |
|
| 554 | elif case(expr_e.FuncCall):
|
| 555 | node = cast(expr.FuncCall, UP_node)
|
| 556 | return node.args.left
|
| 557 |
|
| 558 | elif case(expr_e.Subscript):
|
| 559 | node = cast(Subscript, UP_node)
|
| 560 | return node.left
|
| 561 |
|
| 562 | elif case(expr_e.Attribute):
|
| 563 | node = cast(Attribute, UP_node)
|
| 564 | return node.op
|
| 565 |
|
| 566 | elif case(expr_e.Eggex):
|
| 567 | node = cast(Eggex, UP_node)
|
| 568 | return node.left
|
| 569 |
|
| 570 | else:
|
| 571 | raise AssertionError(node.__class__.__name__)
|