| 1 | /*- |
| 2 | * Copyright (c) 1991 The Regents of the University of California. |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * This code is derived from software contributed to Berkeley by |
| 6 | * Kenneth Almquist. |
| 7 | * |
| 8 | * %sccs.include.redist.c% |
| 9 | */ |
| 10 | |
| 11 | #ifndef lint |
| 12 | static char sccsid[] = "@(#)parser.c 5.1 (Berkeley) %G%"; |
| 13 | #endif /* not lint */ |
| 14 | |
| 15 | #include "shell.h" |
| 16 | #include "parser.h" |
| 17 | #include "nodes.h" |
| 18 | #include "expand.h" /* defines rmescapes() */ |
| 19 | #include "redir.h" /* defines copyfd() */ |
| 20 | #include "syntax.h" |
| 21 | #include "options.h" |
| 22 | #include "input.h" |
| 23 | #include "output.h" |
| 24 | #include "var.h" |
| 25 | #include "error.h" |
| 26 | #include "memalloc.h" |
| 27 | #include "mystring.h" |
| 28 | |
| 29 | |
| 30 | /* |
| 31 | * Shell command parser. |
| 32 | */ |
| 33 | |
| 34 | #define EOFMARKLEN 79 |
| 35 | |
| 36 | /* values returned by readtoken */ |
| 37 | #include "token.def" |
| 38 | |
| 39 | |
| 40 | |
| 41 | struct heredoc { |
| 42 | struct heredoc *next; /* next here document in list */ |
| 43 | union node *here; /* redirection node */ |
| 44 | char *eofmark; /* string indicating end of input */ |
| 45 | int striptabs; /* if set, strip leading tabs */ |
| 46 | }; |
| 47 | |
| 48 | |
| 49 | |
| 50 | struct heredoc *heredoclist; /* list of here documents to read */ |
| 51 | int parsebackquote; /* nonzero if we are inside backquotes */ |
| 52 | int doprompt; /* if set, prompt the user */ |
| 53 | int needprompt; /* true if interactive and at start of line */ |
| 54 | int lasttoken; /* last token read */ |
| 55 | MKINIT int tokpushback; /* last token pushed back */ |
| 56 | char *wordtext; /* text of last word returned by readtoken */ |
| 57 | struct nodelist *backquotelist; |
| 58 | union node *redirnode; |
| 59 | struct heredoc *heredoc; |
| 60 | int quoteflag; /* set if (part of) last token was quoted */ |
| 61 | int startlinno; /* line # where last token started */ |
| 62 | |
| 63 | |
| 64 | #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ |
| 65 | #ifdef GDB_HACK |
| 66 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; |
| 67 | static const char types[] = "}-+?="; |
| 68 | #endif |
| 69 | |
| 70 | |
| 71 | #ifdef __STDC__ |
| 72 | STATIC union node *list(int); |
| 73 | STATIC union node *andor(void); |
| 74 | STATIC union node *pipeline(void); |
| 75 | STATIC union node *command(void); |
| 76 | STATIC union node *simplecmd(void); |
| 77 | STATIC void parsefname(void); |
| 78 | STATIC void parseheredoc(void); |
| 79 | STATIC void checkkwd(void); |
| 80 | STATIC int readtoken(void); |
| 81 | STATIC int readtoken1(int, char const *, char *, int); |
| 82 | STATIC void attyline(void); |
| 83 | STATIC int noexpand(char *); |
| 84 | STATIC void synexpect(int); |
| 85 | STATIC void synerror(char *); |
| 86 | #else |
| 87 | STATIC union node *list(); |
| 88 | STATIC union node *andor(); |
| 89 | STATIC union node *pipeline(); |
| 90 | STATIC union node *command(); |
| 91 | STATIC union node *simplecmd(); |
| 92 | STATIC void parsefname(); |
| 93 | STATIC void parseheredoc(); |
| 94 | STATIC void checkkwd(); |
| 95 | STATIC int readtoken(); |
| 96 | STATIC int readtoken1(); |
| 97 | STATIC void attyline(); |
| 98 | STATIC int noexpand(); |
| 99 | STATIC void synexpect(); |
| 100 | STATIC void synerror(); |
| 101 | #endif |
| 102 | |
| 103 | #if ATTY |
| 104 | #ifdef __STDC__ |
| 105 | STATIC void putprompt(char *); |
| 106 | #else |
| 107 | STATIC void putprompt(); |
| 108 | #endif |
| 109 | #else /* not ATTY */ |
| 110 | #define putprompt(s) out2str(s) |
| 111 | #endif |
| 112 | |
| 113 | |
| 114 | |
| 115 | |
| 116 | /* |
| 117 | * Read and parse a command. Returns NEOF on end of file. (NULL is a |
| 118 | * valid parse tree indicating a blank line.) |
| 119 | */ |
| 120 | |
| 121 | union node * |
| 122 | parsecmd(interact) { |
| 123 | int t; |
| 124 | |
| 125 | doprompt = interact; |
| 126 | if (doprompt) |
| 127 | putprompt(ps1val()); |
| 128 | needprompt = 0; |
| 129 | if ((t = readtoken()) == TEOF) |
| 130 | return NEOF; |
| 131 | if (t == TNL) |
| 132 | return NULL; |
| 133 | tokpushback++; |
| 134 | return list(1); |
| 135 | } |
| 136 | |
| 137 | |
| 138 | STATIC union node * |
| 139 | list(nlflag) { |
| 140 | union node *n1, *n2, *n3; |
| 141 | |
| 142 | checkkwd(); |
| 143 | if (nlflag == 0 && tokendlist[lasttoken]) |
| 144 | return NULL; |
| 145 | n1 = andor(); |
| 146 | for (;;) { |
| 147 | switch (readtoken()) { |
| 148 | case TBACKGND: |
| 149 | if (n1->type == NCMD || n1->type == NPIPE) { |
| 150 | n1->ncmd.backgnd = 1; |
| 151 | } else if (n1->type == NREDIR) { |
| 152 | n1->type = NBACKGND; |
| 153 | } else { |
| 154 | n3 = (union node *)stalloc(sizeof (struct nredir)); |
| 155 | n3->type = NBACKGND; |
| 156 | n3->nredir.n = n1; |
| 157 | n3->nredir.redirect = NULL; |
| 158 | n1 = n3; |
| 159 | } |
| 160 | goto tsemi; |
| 161 | case TNL: |
| 162 | tokpushback++; |
| 163 | /* fall through */ |
| 164 | tsemi: case TSEMI: |
| 165 | if (readtoken() == TNL) { |
| 166 | parseheredoc(); |
| 167 | if (nlflag) |
| 168 | return n1; |
| 169 | } else { |
| 170 | tokpushback++; |
| 171 | } |
| 172 | checkkwd(); |
| 173 | if (tokendlist[lasttoken]) |
| 174 | return n1; |
| 175 | n2 = andor(); |
| 176 | n3 = (union node *)stalloc(sizeof (struct nbinary)); |
| 177 | n3->type = NSEMI; |
| 178 | n3->nbinary.ch1 = n1; |
| 179 | n3->nbinary.ch2 = n2; |
| 180 | n1 = n3; |
| 181 | break; |
| 182 | case TEOF: |
| 183 | if (heredoclist) |
| 184 | parseheredoc(); |
| 185 | else |
| 186 | pungetc(); /* push back EOF on input */ |
| 187 | return n1; |
| 188 | default: |
| 189 | if (nlflag) |
| 190 | synexpect(-1); |
| 191 | tokpushback++; |
| 192 | return n1; |
| 193 | } |
| 194 | } |
| 195 | } |
| 196 | |
| 197 | |
| 198 | |
| 199 | STATIC union node * |
| 200 | andor() { |
| 201 | union node *n1, *n2, *n3; |
| 202 | int t; |
| 203 | |
| 204 | n1 = pipeline(); |
| 205 | for (;;) { |
| 206 | if ((t = readtoken()) == TAND) { |
| 207 | t = NAND; |
| 208 | } else if (t == TOR) { |
| 209 | t = NOR; |
| 210 | } else { |
| 211 | tokpushback++; |
| 212 | return n1; |
| 213 | } |
| 214 | n2 = pipeline(); |
| 215 | n3 = (union node *)stalloc(sizeof (struct nbinary)); |
| 216 | n3->type = t; |
| 217 | n3->nbinary.ch1 = n1; |
| 218 | n3->nbinary.ch2 = n2; |
| 219 | n1 = n3; |
| 220 | } |
| 221 | } |
| 222 | |
| 223 | |
| 224 | |
| 225 | STATIC union node * |
| 226 | pipeline() { |
| 227 | union node *n1, *pipenode; |
| 228 | struct nodelist *lp, *prev; |
| 229 | |
| 230 | n1 = command(); |
| 231 | if (readtoken() == TPIPE) { |
| 232 | pipenode = (union node *)stalloc(sizeof (struct npipe)); |
| 233 | pipenode->type = NPIPE; |
| 234 | pipenode->npipe.backgnd = 0; |
| 235 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); |
| 236 | pipenode->npipe.cmdlist = lp; |
| 237 | lp->n = n1; |
| 238 | do { |
| 239 | prev = lp; |
| 240 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); |
| 241 | lp->n = command(); |
| 242 | prev->next = lp; |
| 243 | } while (readtoken() == TPIPE); |
| 244 | lp->next = NULL; |
| 245 | n1 = pipenode; |
| 246 | } |
| 247 | tokpushback++; |
| 248 | return n1; |
| 249 | } |
| 250 | |
| 251 | |
| 252 | |
| 253 | STATIC union node * |
| 254 | command() { |
| 255 | union node *n1, *n2; |
| 256 | union node *ap, **app; |
| 257 | union node *cp, **cpp; |
| 258 | union node *redir, **rpp; |
| 259 | int t; |
| 260 | |
| 261 | checkkwd(); |
| 262 | switch (readtoken()) { |
| 263 | case TIF: |
| 264 | n1 = (union node *)stalloc(sizeof (struct nif)); |
| 265 | n1->type = NIF; |
| 266 | n1->nif.test = list(0); |
| 267 | if (readtoken() != TTHEN) |
| 268 | synexpect(TTHEN); |
| 269 | n1->nif.ifpart = list(0); |
| 270 | n2 = n1; |
| 271 | while (readtoken() == TELIF) { |
| 272 | n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); |
| 273 | n2 = n2->nif.elsepart; |
| 274 | n2->type = NIF; |
| 275 | n2->nif.test = list(0); |
| 276 | if (readtoken() != TTHEN) |
| 277 | synexpect(TTHEN); |
| 278 | n2->nif.ifpart = list(0); |
| 279 | } |
| 280 | if (lasttoken == TELSE) |
| 281 | n2->nif.elsepart = list(0); |
| 282 | else { |
| 283 | n2->nif.elsepart = NULL; |
| 284 | tokpushback++; |
| 285 | } |
| 286 | if (readtoken() != TFI) |
| 287 | synexpect(TFI); |
| 288 | break; |
| 289 | case TWHILE: |
| 290 | case TUNTIL: |
| 291 | n1 = (union node *)stalloc(sizeof (struct nbinary)); |
| 292 | n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; |
| 293 | n1->nbinary.ch1 = list(0); |
| 294 | if (readtoken() != TDO) |
| 295 | synexpect(TDO); |
| 296 | n1->nbinary.ch2 = list(0); |
| 297 | if (readtoken() != TDONE) |
| 298 | synexpect(TDONE); |
| 299 | break; |
| 300 | case TFOR: |
| 301 | if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) |
| 302 | synerror("Bad for loop variable"); |
| 303 | n1 = (union node *)stalloc(sizeof (struct nfor)); |
| 304 | n1->type = NFOR; |
| 305 | n1->nfor.var = wordtext; |
| 306 | if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { |
| 307 | app = ≈ |
| 308 | while (readtoken() == TWORD) { |
| 309 | n2 = (union node *)stalloc(sizeof (struct narg)); |
| 310 | n2->type = NARG; |
| 311 | n2->narg.text = wordtext; |
| 312 | n2->narg.backquote = backquotelist; |
| 313 | *app = n2; |
| 314 | app = &n2->narg.next; |
| 315 | } |
| 316 | *app = NULL; |
| 317 | n1->nfor.args = ap; |
| 318 | } else { |
| 319 | #ifndef GDB_HACK |
| 320 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, |
| 321 | '@', '=', '\0'}; |
| 322 | #endif |
| 323 | n2 = (union node *)stalloc(sizeof (struct narg)); |
| 324 | n2->type = NARG; |
| 325 | n2->narg.text = (char *)argvars; |
| 326 | n2->narg.backquote = NULL; |
| 327 | n2->narg.next = NULL; |
| 328 | n1->nfor.args = n2; |
| 329 | } |
| 330 | if (lasttoken != TNL && lasttoken != TSEMI) |
| 331 | synexpect(-1); |
| 332 | checkkwd(); |
| 333 | if ((t = readtoken()) == TDO) |
| 334 | t = TDONE; |
| 335 | else if (t == TBEGIN) |
| 336 | t = TEND; |
| 337 | else |
| 338 | synexpect(-1); |
| 339 | n1->nfor.body = list(0); |
| 340 | if (readtoken() != t) |
| 341 | synexpect(t); |
| 342 | break; |
| 343 | case TCASE: |
| 344 | n1 = (union node *)stalloc(sizeof (struct ncase)); |
| 345 | n1->type = NCASE; |
| 346 | if (readtoken() != TWORD) |
| 347 | synexpect(TWORD); |
| 348 | n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); |
| 349 | n2->type = NARG; |
| 350 | n2->narg.text = wordtext; |
| 351 | n2->narg.backquote = backquotelist; |
| 352 | n2->narg.next = NULL; |
| 353 | while (readtoken() == TNL); |
| 354 | if (lasttoken != TWORD || ! equal(wordtext, "in")) |
| 355 | synerror("expecting \"in\""); |
| 356 | cpp = &n1->ncase.cases; |
| 357 | while (checkkwd(), readtoken() == TWORD) { |
| 358 | *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); |
| 359 | cp->type = NCLIST; |
| 360 | app = &cp->nclist.pattern; |
| 361 | for (;;) { |
| 362 | *app = ap = (union node *)stalloc(sizeof (struct narg)); |
| 363 | ap->type = NARG; |
| 364 | ap->narg.text = wordtext; |
| 365 | ap->narg.backquote = backquotelist; |
| 366 | if (readtoken() != TPIPE) |
| 367 | break; |
| 368 | app = &ap->narg.next; |
| 369 | if (readtoken() != TWORD) |
| 370 | synexpect(TWORD); |
| 371 | } |
| 372 | ap->narg.next = NULL; |
| 373 | if (lasttoken != TRP) |
| 374 | synexpect(TRP); |
| 375 | cp->nclist.body = list(0); |
| 376 | if ((t = readtoken()) == TESAC) |
| 377 | tokpushback++; |
| 378 | else if (t != TENDCASE) |
| 379 | synexpect(TENDCASE); |
| 380 | cpp = &cp->nclist.next; |
| 381 | } |
| 382 | *cpp = NULL; |
| 383 | if (lasttoken != TESAC) |
| 384 | synexpect(TESAC); |
| 385 | break; |
| 386 | case TLP: |
| 387 | n1 = (union node *)stalloc(sizeof (struct nredir)); |
| 388 | n1->type = NSUBSHELL; |
| 389 | n1->nredir.n = list(0); |
| 390 | n1->nredir.redirect = NULL; |
| 391 | if (readtoken() != TRP) |
| 392 | synexpect(TRP); |
| 393 | break; |
| 394 | case TBEGIN: |
| 395 | n1 = list(0); |
| 396 | if (readtoken() != TEND) |
| 397 | synexpect(TEND); |
| 398 | break; |
| 399 | case TWORD: |
| 400 | case TREDIR: |
| 401 | tokpushback++; |
| 402 | return simplecmd(); |
| 403 | default: |
| 404 | synexpect(-1); |
| 405 | } |
| 406 | |
| 407 | /* Now check for redirection which may follow command */ |
| 408 | rpp = &redir; |
| 409 | while (readtoken() == TREDIR) { |
| 410 | *rpp = n2 = redirnode; |
| 411 | rpp = &n2->nfile.next; |
| 412 | parsefname(); |
| 413 | } |
| 414 | tokpushback++; |
| 415 | *rpp = NULL; |
| 416 | if (redir) { |
| 417 | if (n1->type != NSUBSHELL) { |
| 418 | n2 = (union node *)stalloc(sizeof (struct nredir)); |
| 419 | n2->type = NREDIR; |
| 420 | n2->nredir.n = n1; |
| 421 | n1 = n2; |
| 422 | } |
| 423 | n1->nredir.redirect = redir; |
| 424 | } |
| 425 | return n1; |
| 426 | } |
| 427 | |
| 428 | |
| 429 | STATIC union node * |
| 430 | simplecmd() { |
| 431 | union node *args, **app; |
| 432 | union node *redir, **rpp; |
| 433 | union node *n; |
| 434 | |
| 435 | args = NULL; |
| 436 | app = &args; |
| 437 | rpp = &redir; |
| 438 | for (;;) { |
| 439 | if (readtoken() == TWORD) { |
| 440 | n = (union node *)stalloc(sizeof (struct narg)); |
| 441 | n->type = NARG; |
| 442 | n->narg.text = wordtext; |
| 443 | n->narg.backquote = backquotelist; |
| 444 | *app = n; |
| 445 | app = &n->narg.next; |
| 446 | } else if (lasttoken == TREDIR) { |
| 447 | *rpp = n = redirnode; |
| 448 | rpp = &n->nfile.next; |
| 449 | parsefname(); /* read name of redirection file */ |
| 450 | } else if (lasttoken == TLP && app == &args->narg.next |
| 451 | && rpp == &redir) { |
| 452 | /* We have a function */ |
| 453 | if (readtoken() != TRP) |
| 454 | synexpect(TRP); |
| 455 | if (! goodname(n->narg.text)) |
| 456 | synerror("Bad function name"); |
| 457 | n->type = NDEFUN; |
| 458 | n->narg.next = command(); |
| 459 | return n; |
| 460 | } else { |
| 461 | tokpushback++; |
| 462 | break; |
| 463 | } |
| 464 | } |
| 465 | *app = NULL; |
| 466 | *rpp = NULL; |
| 467 | n = (union node *)stalloc(sizeof (struct ncmd)); |
| 468 | n->type = NCMD; |
| 469 | n->ncmd.backgnd = 0; |
| 470 | n->ncmd.args = args; |
| 471 | n->ncmd.redirect = redir; |
| 472 | return n; |
| 473 | } |
| 474 | |
| 475 | |
| 476 | STATIC void |
| 477 | parsefname() { |
| 478 | union node *n = redirnode; |
| 479 | |
| 480 | if (readtoken() != TWORD) |
| 481 | synexpect(-1); |
| 482 | if (n->type == NHERE) { |
| 483 | struct heredoc *here = heredoc; |
| 484 | struct heredoc *p; |
| 485 | int i; |
| 486 | |
| 487 | if (quoteflag == 0) |
| 488 | n->type = NXHERE; |
| 489 | TRACE(("Here document %d\n", n->type)); |
| 490 | if (here->striptabs) { |
| 491 | while (*wordtext == '\t') |
| 492 | wordtext++; |
| 493 | } |
| 494 | if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) |
| 495 | synerror("Illegal eof marker for << redirection"); |
| 496 | rmescapes(wordtext); |
| 497 | here->eofmark = wordtext; |
| 498 | here->next = NULL; |
| 499 | if (heredoclist == NULL) |
| 500 | heredoclist = here; |
| 501 | else { |
| 502 | for (p = heredoclist ; p->next ; p = p->next); |
| 503 | p->next = here; |
| 504 | } |
| 505 | } else if (n->type == NTOFD || n->type == NFROMFD) { |
| 506 | if (is_digit(wordtext[0])) |
| 507 | n->ndup.dupfd = digit_val(wordtext[0]); |
| 508 | else if (wordtext[0] == '-') |
| 509 | n->ndup.dupfd = -1; |
| 510 | else |
| 511 | goto bad; |
| 512 | if (wordtext[1] != '\0') { |
| 513 | bad: |
| 514 | synerror("Bad fd number"); |
| 515 | } |
| 516 | } else { |
| 517 | n->nfile.fname = (union node *)stalloc(sizeof (struct narg)); |
| 518 | n = n->nfile.fname; |
| 519 | n->type = NARG; |
| 520 | n->narg.next = NULL; |
| 521 | n->narg.text = wordtext; |
| 522 | n->narg.backquote = backquotelist; |
| 523 | } |
| 524 | } |
| 525 | |
| 526 | |
| 527 | /* |
| 528 | * Input any here documents. |
| 529 | */ |
| 530 | |
| 531 | STATIC void |
| 532 | parseheredoc() { |
| 533 | struct heredoc *here; |
| 534 | union node *n; |
| 535 | |
| 536 | while (heredoclist) { |
| 537 | here = heredoclist; |
| 538 | heredoclist = here->next; |
| 539 | if (needprompt) { |
| 540 | putprompt(ps2val()); |
| 541 | needprompt = 0; |
| 542 | } |
| 543 | readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, |
| 544 | here->eofmark, here->striptabs); |
| 545 | n = (union node *)stalloc(sizeof (struct narg)); |
| 546 | n->narg.type = NARG; |
| 547 | n->narg.next = NULL; |
| 548 | n->narg.text = wordtext; |
| 549 | n->narg.backquote = backquotelist; |
| 550 | here->here->nhere.doc = n; |
| 551 | } |
| 552 | } |
| 553 | |
| 554 | |
| 555 | |
| 556 | /* |
| 557 | * This routine is called to tell readtoken that we are at the beginning |
| 558 | * of a command, so newlines should be ignored and keywords should be |
| 559 | * checked for. We munge things here rather than setting a flag for |
| 560 | * readtoken. |
| 561 | */ |
| 562 | |
| 563 | STATIC void |
| 564 | checkkwd() { |
| 565 | register char *const *pp; |
| 566 | int t; |
| 567 | |
| 568 | while ((t = readtoken()) == TNL); |
| 569 | if (t == TWORD && quoteflag == 0) { |
| 570 | for (pp = parsekwd ; *pp ; pp++) { |
| 571 | if (**pp == *wordtext && equal(*pp, wordtext)) { |
| 572 | lasttoken = pp - parsekwd + KWDOFFSET; |
| 573 | break; |
| 574 | } |
| 575 | } |
| 576 | } |
| 577 | tokpushback++; |
| 578 | } |
| 579 | |
| 580 | |
| 581 | |
| 582 | STATIC int xxreadtoken(); |
| 583 | |
| 584 | STATIC int |
| 585 | readtoken() { |
| 586 | int t; |
| 587 | |
| 588 | if (tokpushback) { |
| 589 | return xxreadtoken(); |
| 590 | } else { |
| 591 | t = xxreadtoken(); |
| 592 | TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); |
| 593 | if (vflag) |
| 594 | outfmt(out2, "%s %s\n", tokname[t], |
| 595 | t == TWORD ? wordtext : ""); |
| 596 | return t; |
| 597 | } |
| 598 | } |
| 599 | |
| 600 | |
| 601 | /* |
| 602 | * Read the next input token. |
| 603 | * If the token is a word, we set backquotelist to the list of cmds in |
| 604 | * backquotes. We set quoteflag to true if any part of the word was |
| 605 | * quoted. |
| 606 | * If the token is TREDIR, then we set redirnode to a structure containing |
| 607 | * the redirection. |
| 608 | * In all cases, the variable startlinno is set to the number of the line |
| 609 | * on which the token starts. |
| 610 | * |
| 611 | * [Change comment: here documents and internal procedures] |
| 612 | * [Readtoken shouldn't have any arguments. Perhaps we should make the |
| 613 | * word parsing code into a separate routine. In this case, readtoken |
| 614 | * doesn't need to have any internal procedures, but parseword does. |
| 615 | * We could also make parseoperator in essence the main routine, and |
| 616 | * have parseword (readtoken1?) handle both words and redirection.] |
| 617 | */ |
| 618 | |
| 619 | #define RETURN(token) return lasttoken = token |
| 620 | |
| 621 | STATIC int |
| 622 | xxreadtoken() { |
| 623 | register c; |
| 624 | |
| 625 | if (tokpushback) { |
| 626 | tokpushback = 0; |
| 627 | return lasttoken; |
| 628 | } |
| 629 | if (needprompt) { |
| 630 | putprompt(ps2val()); |
| 631 | needprompt = 0; |
| 632 | } |
| 633 | startlinno = plinno; |
| 634 | for (;;) { /* until token or start of word found */ |
| 635 | c = pgetc_macro(); |
| 636 | if (c == ' ' || c == '\t') |
| 637 | continue; /* quick check for white space first */ |
| 638 | switch (c) { |
| 639 | case ' ': case '\t': |
| 640 | continue; |
| 641 | case '#': |
| 642 | while ((c = pgetc()) != '\n' && c != PEOF); |
| 643 | pungetc(); |
| 644 | continue; |
| 645 | case '\\': |
| 646 | if (pgetc() == '\n') { |
| 647 | startlinno = ++plinno; |
| 648 | if (doprompt) |
| 649 | putprompt(ps2val()); |
| 650 | continue; |
| 651 | } |
| 652 | pungetc(); |
| 653 | goto breakloop; |
| 654 | case '\n': |
| 655 | plinno++; |
| 656 | needprompt = doprompt; |
| 657 | RETURN(TNL); |
| 658 | case PEOF: |
| 659 | RETURN(TEOF); |
| 660 | case '&': |
| 661 | if (pgetc() == '&') |
| 662 | RETURN(TAND); |
| 663 | pungetc(); |
| 664 | RETURN(TBACKGND); |
| 665 | case '|': |
| 666 | if (pgetc() == '|') |
| 667 | RETURN(TOR); |
| 668 | pungetc(); |
| 669 | RETURN(TPIPE); |
| 670 | case ';': |
| 671 | if (pgetc() == ';') |
| 672 | RETURN(TENDCASE); |
| 673 | pungetc(); |
| 674 | RETURN(TSEMI); |
| 675 | case '(': |
| 676 | RETURN(TLP); |
| 677 | case ')': |
| 678 | RETURN(TRP); |
| 679 | default: |
| 680 | goto breakloop; |
| 681 | } |
| 682 | } |
| 683 | breakloop: |
| 684 | return readtoken1(c, BASESYNTAX, (char *)NULL, 0); |
| 685 | #undef RETURN |
| 686 | } |
| 687 | |
| 688 | |
| 689 | |
| 690 | /* |
| 691 | * If eofmark is NULL, read a word or a redirection symbol. If eofmark |
| 692 | * is not NULL, read a here document. In the latter case, eofmark is the |
| 693 | * word which marks the end of the document and striptabs is true if |
| 694 | * leading tabs should be stripped from the document. The argument firstc |
| 695 | * is the first character of the input token or document. |
| 696 | * |
| 697 | * Because C does not have internal subroutines, I have simulated them |
| 698 | * using goto's to implement the subroutine linkage. The following macros |
| 699 | * will run code that appears at the end of readtoken1. |
| 700 | */ |
| 701 | |
| 702 | #define CHECKEND() {goto checkend; checkend_return:;} |
| 703 | #define PARSEREDIR() {goto parseredir; parseredir_return:;} |
| 704 | #define PARSESUB() {goto parsesub; parsesub_return:;} |
| 705 | #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} |
| 706 | #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} |
| 707 | |
| 708 | STATIC int |
| 709 | readtoken1(firstc, syntax, eofmark, striptabs) |
| 710 | int firstc; |
| 711 | char const *syntax; |
| 712 | char *eofmark; |
| 713 | int striptabs; |
| 714 | { |
| 715 | register c = firstc; |
| 716 | register char *out; |
| 717 | int len; |
| 718 | char line[EOFMARKLEN + 1]; |
| 719 | struct nodelist *bqlist; |
| 720 | int quotef; |
| 721 | int dblquote; |
| 722 | int varnest; |
| 723 | int oldstyle; |
| 724 | |
| 725 | startlinno = plinno; |
| 726 | dblquote = 0; |
| 727 | if (syntax == DQSYNTAX) |
| 728 | dblquote = 1; |
| 729 | quotef = 0; |
| 730 | bqlist = NULL; |
| 731 | varnest = 0; |
| 732 | STARTSTACKSTR(out); |
| 733 | loop: { /* for each line, until end of word */ |
| 734 | #if ATTY |
| 735 | if (c == '\034' && doprompt |
| 736 | && attyset() && ! equal(termval(), "emacs")) { |
| 737 | attyline(); |
| 738 | if (syntax == BASESYNTAX) |
| 739 | return readtoken(); |
| 740 | c = pgetc(); |
| 741 | goto loop; |
| 742 | } |
| 743 | #endif |
| 744 | CHECKEND(); /* set c to PEOF if at end of here document */ |
| 745 | for (;;) { /* until end of line or end of word */ |
| 746 | CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ |
| 747 | switch(syntax[c]) { |
| 748 | case CNL: /* '\n' */ |
| 749 | if (syntax == BASESYNTAX) |
| 750 | goto endword; /* exit outer loop */ |
| 751 | USTPUTC(c, out); |
| 752 | plinno++; |
| 753 | if (doprompt) { |
| 754 | putprompt(ps2val()); |
| 755 | } |
| 756 | c = pgetc(); |
| 757 | goto loop; /* continue outer loop */ |
| 758 | case CWORD: |
| 759 | USTPUTC(c, out); |
| 760 | break; |
| 761 | case CCTL: |
| 762 | if (eofmark == NULL || dblquote) |
| 763 | USTPUTC(CTLESC, out); |
| 764 | USTPUTC(c, out); |
| 765 | break; |
| 766 | case CBACK: /* backslash */ |
| 767 | c = pgetc(); |
| 768 | if (c == PEOF) { |
| 769 | USTPUTC('\\', out); |
| 770 | pungetc(); |
| 771 | } else if (c == '\n') { |
| 772 | if (doprompt) |
| 773 | putprompt(ps2val()); |
| 774 | } else { |
| 775 | if (dblquote && c != '\\' && c != '`' && c != '$' |
| 776 | && (c != '"' || eofmark != NULL)) |
| 777 | USTPUTC('\\', out); |
| 778 | if (SQSYNTAX[c] == CCTL) |
| 779 | USTPUTC(CTLESC, out); |
| 780 | USTPUTC(c, out); |
| 781 | quotef++; |
| 782 | } |
| 783 | break; |
| 784 | case CSQUOTE: |
| 785 | syntax = SQSYNTAX; |
| 786 | break; |
| 787 | case CDQUOTE: |
| 788 | syntax = DQSYNTAX; |
| 789 | dblquote = 1; |
| 790 | break; |
| 791 | case CENDQUOTE: |
| 792 | if (eofmark) { |
| 793 | USTPUTC(c, out); |
| 794 | } else { |
| 795 | syntax = BASESYNTAX; |
| 796 | quotef++; |
| 797 | dblquote = 0; |
| 798 | } |
| 799 | break; |
| 800 | case CVAR: /* '$' */ |
| 801 | PARSESUB(); /* parse substitution */ |
| 802 | break; |
| 803 | case CENDVAR: /* '}' */ |
| 804 | if (varnest > 0) { |
| 805 | varnest--; |
| 806 | USTPUTC(CTLENDVAR, out); |
| 807 | } else { |
| 808 | USTPUTC(c, out); |
| 809 | } |
| 810 | break; |
| 811 | case CBQUOTE: /* '`' */ |
| 812 | if (parsebackquote && syntax == BASESYNTAX) { |
| 813 | if (out == stackblock()) |
| 814 | return lasttoken = TENDBQUOTE; |
| 815 | else |
| 816 | goto endword; /* exit outer loop */ |
| 817 | } |
| 818 | PARSEBACKQOLD(); |
| 819 | break; |
| 820 | case CEOF: |
| 821 | goto endword; /* exit outer loop */ |
| 822 | default: |
| 823 | if (varnest == 0) |
| 824 | goto endword; /* exit outer loop */ |
| 825 | USTPUTC(c, out); |
| 826 | } |
| 827 | c = pgetc_macro(); |
| 828 | } |
| 829 | } |
| 830 | endword: |
| 831 | if (syntax != BASESYNTAX && eofmark == NULL) |
| 832 | synerror("Unterminated quoted string"); |
| 833 | if (varnest != 0) { |
| 834 | startlinno = plinno; |
| 835 | synerror("Missing '}'"); |
| 836 | } |
| 837 | USTPUTC('\0', out); |
| 838 | len = out - stackblock(); |
| 839 | out = stackblock(); |
| 840 | if (eofmark == NULL) { |
| 841 | if ((c == '>' || c == '<') |
| 842 | && quotef == 0 |
| 843 | && len <= 2 |
| 844 | && (*out == '\0' || is_digit(*out))) { |
| 845 | PARSEREDIR(); |
| 846 | return lasttoken = TREDIR; |
| 847 | } else { |
| 848 | pungetc(); |
| 849 | } |
| 850 | } |
| 851 | quoteflag = quotef; |
| 852 | backquotelist = bqlist; |
| 853 | grabstackblock(len); |
| 854 | wordtext = out; |
| 855 | return lasttoken = TWORD; |
| 856 | /* end of readtoken routine */ |
| 857 | |
| 858 | |
| 859 | |
| 860 | /* |
| 861 | * Check to see whether we are at the end of the here document. When this |
| 862 | * is called, c is set to the first character of the next input line. If |
| 863 | * we are at the end of the here document, this routine sets the c to PEOF. |
| 864 | */ |
| 865 | |
| 866 | checkend: { |
| 867 | if (eofmark) { |
| 868 | if (striptabs) { |
| 869 | while (c == '\t') |
| 870 | c = pgetc(); |
| 871 | } |
| 872 | if (c == *eofmark) { |
| 873 | if (pfgets(line, sizeof line) != NULL) { |
| 874 | register char *p, *q; |
| 875 | |
| 876 | p = line; |
| 877 | for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); |
| 878 | if (*p == '\n' && *q == '\0') { |
| 879 | c = PEOF; |
| 880 | plinno++; |
| 881 | needprompt = doprompt; |
| 882 | } else { |
| 883 | ppushback(line, strlen(line)); |
| 884 | } |
| 885 | } |
| 886 | } |
| 887 | } |
| 888 | goto checkend_return; |
| 889 | } |
| 890 | |
| 891 | |
| 892 | /* |
| 893 | * Parse a redirection operator. The variable "out" points to a string |
| 894 | * specifying the fd to be redirected. The variable "c" contains the |
| 895 | * first character of the redirection operator. |
| 896 | */ |
| 897 | |
| 898 | parseredir: { |
| 899 | char fd = *out; |
| 900 | union node *np; |
| 901 | |
| 902 | np = (union node *)stalloc(sizeof (struct nfile)); |
| 903 | if (c == '>') { |
| 904 | np->nfile.fd = 1; |
| 905 | c = pgetc(); |
| 906 | if (c == '>') |
| 907 | np->type = NAPPEND; |
| 908 | else if (c == '&') |
| 909 | np->type = NTOFD; |
| 910 | else { |
| 911 | np->type = NTO; |
| 912 | pungetc(); |
| 913 | } |
| 914 | } else { /* c == '<' */ |
| 915 | np->nfile.fd = 0; |
| 916 | c = pgetc(); |
| 917 | if (c == '<') { |
| 918 | if (sizeof (struct nfile) != sizeof (struct nhere)) { |
| 919 | np = (union node *)stalloc(sizeof (struct nhere)); |
| 920 | np->nfile.fd = 0; |
| 921 | } |
| 922 | np->type = NHERE; |
| 923 | heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); |
| 924 | heredoc->here = np; |
| 925 | if ((c = pgetc()) == '-') { |
| 926 | heredoc->striptabs = 1; |
| 927 | } else { |
| 928 | heredoc->striptabs = 0; |
| 929 | pungetc(); |
| 930 | } |
| 931 | } else if (c == '&') |
| 932 | np->type = NFROMFD; |
| 933 | else { |
| 934 | np->type = NFROM; |
| 935 | pungetc(); |
| 936 | } |
| 937 | } |
| 938 | if (fd != '\0') |
| 939 | np->nfile.fd = digit_val(fd); |
| 940 | redirnode = np; |
| 941 | goto parseredir_return; |
| 942 | } |
| 943 | |
| 944 | |
| 945 | /* |
| 946 | * Parse a substitution. At this point, we have read the dollar sign |
| 947 | * and nothing else. |
| 948 | */ |
| 949 | |
| 950 | parsesub: { |
| 951 | int subtype; |
| 952 | int typeloc; |
| 953 | int flags; |
| 954 | char *p; |
| 955 | #ifndef GDB_HACK |
| 956 | static const char types[] = "}-+?="; |
| 957 | #endif |
| 958 | |
| 959 | c = pgetc(); |
| 960 | if (c == ' ' || c == '\t' || c == '\n' || c == PEOF |
| 961 | #ifndef STRICT_VARCHECKING /* make this an option? */ |
| 962 | || (c != '(' && c != '{' && !is_name(c) && !is_special(c)) |
| 963 | #endif |
| 964 | ) { |
| 965 | USTPUTC('$', out); |
| 966 | pungetc(); |
| 967 | } else if (c == '(') { /* $(command) */ |
| 968 | PARSEBACKQNEW(); |
| 969 | } else { |
| 970 | USTPUTC(CTLVAR, out); |
| 971 | typeloc = out - stackblock(); |
| 972 | USTPUTC(VSNORMAL, out); |
| 973 | subtype = VSNORMAL; |
| 974 | if (c == '{') { |
| 975 | c = pgetc(); |
| 976 | subtype = 0; |
| 977 | } |
| 978 | if (is_name(c)) { |
| 979 | do { |
| 980 | STPUTC(c, out); |
| 981 | c = pgetc(); |
| 982 | } while (is_in_name(c)); |
| 983 | } else { |
| 984 | if (! is_special(c)) |
| 985 | badsub: synerror("Bad substitution"); |
| 986 | USTPUTC(c, out); |
| 987 | c = pgetc(); |
| 988 | } |
| 989 | STPUTC('=', out); |
| 990 | flags = 0; |
| 991 | if (subtype == 0) { |
| 992 | if (c == ':') { |
| 993 | flags = VSNUL; |
| 994 | c = pgetc(); |
| 995 | } |
| 996 | p = strchr(types, c); |
| 997 | if (p == NULL) |
| 998 | goto badsub; |
| 999 | subtype = p - types + VSNORMAL; |
| 1000 | } else { |
| 1001 | pungetc(); |
| 1002 | } |
| 1003 | if (dblquote) |
| 1004 | flags |= VSQUOTE; |
| 1005 | *(stackblock() + typeloc) = subtype | flags; |
| 1006 | if (subtype != VSNORMAL) |
| 1007 | varnest++; |
| 1008 | } |
| 1009 | goto parsesub_return; |
| 1010 | } |
| 1011 | |
| 1012 | |
| 1013 | /* |
| 1014 | * Called to parse command substitutions. Newstyle is set if the command |
| 1015 | * is enclosed inside $(...); nlpp is a pointer to the head of the linked |
| 1016 | * list of commands (passed by reference), and savelen is the number of |
| 1017 | * characters on the top of the stack which must be preserved. |
| 1018 | */ |
| 1019 | |
| 1020 | parsebackq: { |
| 1021 | struct nodelist **nlpp; |
| 1022 | int savepbq; |
| 1023 | union node *n; |
| 1024 | char *volatile str; |
| 1025 | struct jmploc jmploc; |
| 1026 | struct jmploc *volatile savehandler; |
| 1027 | int savelen; |
| 1028 | int t; |
| 1029 | |
| 1030 | savepbq = parsebackquote; |
| 1031 | if (setjmp(jmploc.loc)) { |
| 1032 | if (str) |
| 1033 | ckfree(str); |
| 1034 | parsebackquote = 0; |
| 1035 | handler = savehandler; |
| 1036 | longjmp(handler, 1); |
| 1037 | } |
| 1038 | INTOFF; |
| 1039 | str = NULL; |
| 1040 | savelen = out - stackblock(); |
| 1041 | if (savelen > 0) { |
| 1042 | str = ckmalloc(savelen); |
| 1043 | bcopy(stackblock(), str, savelen); |
| 1044 | } |
| 1045 | savehandler = handler; |
| 1046 | handler = &jmploc; |
| 1047 | INTON; |
| 1048 | nlpp = &bqlist; |
| 1049 | while (*nlpp) |
| 1050 | nlpp = &(*nlpp)->next; |
| 1051 | *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); |
| 1052 | (*nlpp)->next = NULL; |
| 1053 | parsebackquote = oldstyle; |
| 1054 | n = list(0); |
| 1055 | t = oldstyle? TENDBQUOTE : TRP; |
| 1056 | if (readtoken() != t) |
| 1057 | synexpect(t); |
| 1058 | (*nlpp)->n = n; |
| 1059 | while (stackblocksize() <= savelen) |
| 1060 | growstackblock(); |
| 1061 | STARTSTACKSTR(out); |
| 1062 | if (str) { |
| 1063 | bcopy(str, out, savelen); |
| 1064 | STADJUST(savelen, out); |
| 1065 | INTOFF; |
| 1066 | ckfree(str); |
| 1067 | str = NULL; |
| 1068 | INTON; |
| 1069 | } |
| 1070 | parsebackquote = savepbq; |
| 1071 | handler = savehandler; |
| 1072 | USTPUTC(CTLBACKQ + dblquote, out); |
| 1073 | if (oldstyle) |
| 1074 | goto parsebackq_oldreturn; |
| 1075 | else |
| 1076 | goto parsebackq_newreturn; |
| 1077 | } |
| 1078 | |
| 1079 | } /* end of readtoken */ |
| 1080 | |
| 1081 | |
| 1082 | |
| 1083 | #ifdef mkinit |
| 1084 | RESET { |
| 1085 | tokpushback = 0; |
| 1086 | } |
| 1087 | #endif |
| 1088 | |
| 1089 | |
| 1090 | #if ATTY |
| 1091 | /* |
| 1092 | * Called to process a command generated by atty. We execute the line, |
| 1093 | * and catch any errors that occur so they don't propagate outside of |
| 1094 | * this routine. |
| 1095 | */ |
| 1096 | |
| 1097 | STATIC void |
| 1098 | attyline() { |
| 1099 | char line[256]; |
| 1100 | struct stackmark smark; |
| 1101 | struct jmploc jmploc; |
| 1102 | struct jmploc *volatile savehandler; |
| 1103 | |
| 1104 | if (pfgets(line, sizeof line) == NULL) |
| 1105 | return; /* "can't happen" */ |
| 1106 | if (setjmp(jmploc.loc)) { |
| 1107 | if (exception == EXERROR) |
| 1108 | out2str("\033]D\n"); |
| 1109 | handler = savehandler; |
| 1110 | longjmp(handler, 1); |
| 1111 | } |
| 1112 | savehandler = handler; |
| 1113 | handler = &jmploc; |
| 1114 | setstackmark(&smark); |
| 1115 | evalstring(line); |
| 1116 | popstackmark(&smark); |
| 1117 | handler = savehandler; |
| 1118 | doprompt = 1; |
| 1119 | } |
| 1120 | |
| 1121 | |
| 1122 | /* |
| 1123 | * Output a prompt for atty. We output the prompt as part of the |
| 1124 | * appropriate escape sequence. |
| 1125 | */ |
| 1126 | |
| 1127 | STATIC void |
| 1128 | putprompt(s) |
| 1129 | char *s; |
| 1130 | { |
| 1131 | register char *p; |
| 1132 | |
| 1133 | if (attyset() && ! equal(termval(), "emacs")) { |
| 1134 | if (strchr(s, '\7')) |
| 1135 | out2c('\7'); |
| 1136 | out2str("\033]P1;"); |
| 1137 | for (p = s ; *p ; p++) { |
| 1138 | if ((unsigned)(*p - ' ') <= '~' - ' ') |
| 1139 | out2c(*p); |
| 1140 | } |
| 1141 | out2c('\n'); |
| 1142 | } else { |
| 1143 | out2str(s); |
| 1144 | } |
| 1145 | } |
| 1146 | #endif |
| 1147 | |
| 1148 | |
| 1149 | |
| 1150 | /* |
| 1151 | * Returns true if the text contains nothing to expand (no dollar signs |
| 1152 | * or backquotes). |
| 1153 | */ |
| 1154 | |
| 1155 | STATIC int |
| 1156 | noexpand(text) |
| 1157 | char *text; |
| 1158 | { |
| 1159 | register char *p; |
| 1160 | register char c; |
| 1161 | |
| 1162 | p = text; |
| 1163 | while ((c = *p++) != '\0') { |
| 1164 | if (c == CTLESC) |
| 1165 | p++; |
| 1166 | else if (BASESYNTAX[c] == CCTL) |
| 1167 | return 0; |
| 1168 | } |
| 1169 | return 1; |
| 1170 | } |
| 1171 | |
| 1172 | |
| 1173 | /* |
| 1174 | * Return true if the argument is a legal variable name (a letter or |
| 1175 | * underscore followed by zero or more letters, underscores, and digits). |
| 1176 | */ |
| 1177 | |
| 1178 | int |
| 1179 | goodname(name) |
| 1180 | char *name; |
| 1181 | { |
| 1182 | register char *p; |
| 1183 | |
| 1184 | p = name; |
| 1185 | if (! is_name(*p)) |
| 1186 | return 0; |
| 1187 | while (*++p) { |
| 1188 | if (! is_in_name(*p)) |
| 1189 | return 0; |
| 1190 | } |
| 1191 | return 1; |
| 1192 | } |
| 1193 | |
| 1194 | |
| 1195 | /* |
| 1196 | * Called when an unexpected token is read during the parse. The argument |
| 1197 | * is the token that is expected, or -1 if more than one type of token can |
| 1198 | * occur at this point. |
| 1199 | */ |
| 1200 | |
| 1201 | STATIC void |
| 1202 | synexpect(token) { |
| 1203 | char msg[64]; |
| 1204 | |
| 1205 | if (token >= 0) { |
| 1206 | fmtstr(msg, 64, "%s unexpected (expecting %s)", |
| 1207 | tokname[lasttoken], tokname[token]); |
| 1208 | } else { |
| 1209 | fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); |
| 1210 | } |
| 1211 | synerror(msg); |
| 1212 | } |
| 1213 | |
| 1214 | |
| 1215 | STATIC void |
| 1216 | synerror(msg) |
| 1217 | char *msg; |
| 1218 | { |
| 1219 | if (commandname) |
| 1220 | outfmt(&errout, "%s: %d: ", commandname, startlinno); |
| 1221 | outfmt(&errout, "Syntax error: %s\n", msg); |
| 1222 | error((char *)NULL); |
| 1223 | } |