Commit | Line | Data |
---|---|---|
af3e82f7 WJ |
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 | * Redistribution and use in source and binary forms, with or without | |
9 | * modification, are permitted provided that the following conditions | |
10 | * are met: | |
11 | * 1. Redistributions of source code must retain the above copyright | |
12 | * notice, this list of conditions and the following disclaimer. | |
13 | * 2. Redistributions in binary form must reproduce the above copyright | |
14 | * notice, this list of conditions and the following disclaimer in the | |
15 | * documentation and/or other materials provided with the distribution. | |
16 | * 3. All advertising materials mentioning features or use of this software | |
17 | * must display the following acknowledgement: | |
18 | * This product includes software developed by the University of | |
19 | * California, Berkeley and its contributors. | |
20 | * 4. Neither the name of the University nor the names of its contributors | |
21 | * may be used to endorse or promote products derived from this software | |
22 | * without specific prior written permission. | |
23 | * | |
24 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND | |
25 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | |
26 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE | |
27 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE | |
28 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | |
29 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS | |
30 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) | |
31 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT | |
32 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY | |
33 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF | |
34 | * SUCH DAMAGE. | |
35 | */ | |
36 | ||
37 | #ifndef lint | |
38 | static char sccsid[] = "@(#)parser.c 5.3 (Berkeley) 4/12/91"; | |
39 | #endif /* not lint */ | |
40 | ||
41 | #include "shell.h" | |
42 | #include "parser.h" | |
43 | #include "nodes.h" | |
44 | #include "expand.h" /* defines rmescapes() */ | |
45 | #include "redir.h" /* defines copyfd() */ | |
46 | #include "syntax.h" | |
47 | #include "options.h" | |
48 | #include "input.h" | |
49 | #include "output.h" | |
50 | #include "var.h" | |
51 | #include "error.h" | |
52 | #include "memalloc.h" | |
53 | #include "mystring.h" | |
54 | ||
55 | ||
56 | /* | |
57 | * Shell command parser. | |
58 | */ | |
59 | ||
60 | #define EOFMARKLEN 79 | |
61 | ||
62 | /* values returned by readtoken */ | |
63 | #include "token.def" | |
64 | ||
65 | ||
66 | ||
67 | struct heredoc { | |
68 | struct heredoc *next; /* next here document in list */ | |
69 | union node *here; /* redirection node */ | |
70 | char *eofmark; /* string indicating end of input */ | |
71 | int striptabs; /* if set, strip leading tabs */ | |
72 | }; | |
73 | ||
74 | ||
75 | ||
76 | struct heredoc *heredoclist; /* list of here documents to read */ | |
77 | int parsebackquote; /* nonzero if we are inside backquotes */ | |
78 | int doprompt; /* if set, prompt the user */ | |
79 | int needprompt; /* true if interactive and at start of line */ | |
80 | int lasttoken; /* last token read */ | |
81 | MKINIT int tokpushback; /* last token pushed back */ | |
82 | char *wordtext; /* text of last word returned by readtoken */ | |
83 | int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ | |
84 | struct nodelist *backquotelist; | |
85 | union node *redirnode; | |
86 | struct heredoc *heredoc; | |
87 | int quoteflag; /* set if (part of) last token was quoted */ | |
88 | int startlinno; /* line # where last token started */ | |
89 | ||
90 | ||
91 | #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ | |
92 | #ifdef GDB_HACK | |
93 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; | |
94 | static const char types[] = "}-+?="; | |
95 | #endif | |
96 | ||
97 | ||
98 | STATIC union node *list __P((int)); | |
99 | STATIC union node *andor __P((void)); | |
100 | STATIC union node *pipeline __P((void)); | |
101 | STATIC union node *command __P((void)); | |
102 | STATIC union node *simplecmd __P((void)); | |
103 | STATIC void parsefname __P((void)); | |
104 | STATIC void parseheredoc __P((void)); | |
105 | STATIC int readtoken __P((void)); | |
106 | STATIC int readtoken1 __P((int, char const *, char *, int)); | |
107 | STATIC void attyline __P((void)); | |
108 | STATIC int noexpand __P((char *)); | |
109 | STATIC void synexpect __P((int)); | |
110 | STATIC void synerror __P((char *)); | |
111 | ||
112 | #if ATTY | |
113 | STATIC void putprompt __P((char *)); | |
114 | #else /* not ATTY */ | |
115 | #define putprompt(s) out2str(s) | |
116 | #endif | |
117 | ||
118 | ||
119 | ||
120 | ||
121 | /* | |
122 | * Read and parse a command. Returns NEOF on end of file. (NULL is a | |
123 | * valid parse tree indicating a blank line.) | |
124 | */ | |
125 | ||
126 | union node * | |
127 | parsecmd(interact) { | |
128 | int t; | |
129 | ||
130 | doprompt = interact; | |
131 | if (doprompt) | |
132 | putprompt(ps1val()); | |
133 | needprompt = 0; | |
134 | if ((t = readtoken()) == TEOF) | |
135 | return NEOF; | |
136 | if (t == TNL) | |
137 | return NULL; | |
138 | tokpushback++; | |
139 | return list(1); | |
140 | } | |
141 | ||
142 | ||
143 | STATIC union node * | |
144 | list(nlflag) { | |
145 | union node *n1, *n2, *n3; | |
146 | ||
147 | checkkwd = 2; | |
148 | if (nlflag == 0 && tokendlist[peektoken()]) | |
149 | return NULL; | |
150 | n1 = andor(); | |
151 | for (;;) { | |
152 | switch (readtoken()) { | |
153 | case TBACKGND: | |
154 | if (n1->type == NCMD || n1->type == NPIPE) { | |
155 | n1->ncmd.backgnd = 1; | |
156 | } else if (n1->type == NREDIR) { | |
157 | n1->type = NBACKGND; | |
158 | } else { | |
159 | n3 = (union node *)stalloc(sizeof (struct nredir)); | |
160 | n3->type = NBACKGND; | |
161 | n3->nredir.n = n1; | |
162 | n3->nredir.redirect = NULL; | |
163 | n1 = n3; | |
164 | } | |
165 | goto tsemi; | |
166 | case TNL: | |
167 | tokpushback++; | |
168 | /* fall through */ | |
169 | tsemi: case TSEMI: | |
170 | if (readtoken() == TNL) { | |
171 | parseheredoc(); | |
172 | if (nlflag) | |
173 | return n1; | |
174 | } else { | |
175 | tokpushback++; | |
176 | } | |
177 | checkkwd = 2; | |
178 | if (tokendlist[peektoken()]) | |
179 | return n1; | |
180 | n2 = andor(); | |
181 | n3 = (union node *)stalloc(sizeof (struct nbinary)); | |
182 | n3->type = NSEMI; | |
183 | n3->nbinary.ch1 = n1; | |
184 | n3->nbinary.ch2 = n2; | |
185 | n1 = n3; | |
186 | break; | |
187 | case TEOF: | |
188 | if (heredoclist) | |
189 | parseheredoc(); | |
190 | else | |
191 | pungetc(); /* push back EOF on input */ | |
192 | return n1; | |
193 | default: | |
194 | if (nlflag) | |
195 | synexpect(-1); | |
196 | tokpushback++; | |
197 | return n1; | |
198 | } | |
199 | } | |
200 | } | |
201 | ||
202 | ||
203 | ||
204 | STATIC union node * | |
205 | andor() { | |
206 | union node *n1, *n2, *n3; | |
207 | int t; | |
208 | ||
209 | n1 = pipeline(); | |
210 | for (;;) { | |
211 | if ((t = readtoken()) == TAND) { | |
212 | t = NAND; | |
213 | } else if (t == TOR) { | |
214 | t = NOR; | |
215 | } else { | |
216 | tokpushback++; | |
217 | return n1; | |
218 | } | |
219 | n2 = pipeline(); | |
220 | n3 = (union node *)stalloc(sizeof (struct nbinary)); | |
221 | n3->type = t; | |
222 | n3->nbinary.ch1 = n1; | |
223 | n3->nbinary.ch2 = n2; | |
224 | n1 = n3; | |
225 | } | |
226 | } | |
227 | ||
228 | ||
229 | ||
230 | STATIC union node * | |
231 | pipeline() { | |
232 | union node *n1, *pipenode; | |
233 | struct nodelist *lp, *prev; | |
234 | ||
235 | n1 = command(); | |
236 | if (readtoken() == TPIPE) { | |
237 | pipenode = (union node *)stalloc(sizeof (struct npipe)); | |
238 | pipenode->type = NPIPE; | |
239 | pipenode->npipe.backgnd = 0; | |
240 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
241 | pipenode->npipe.cmdlist = lp; | |
242 | lp->n = n1; | |
243 | do { | |
244 | prev = lp; | |
245 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
246 | lp->n = command(); | |
247 | prev->next = lp; | |
248 | } while (readtoken() == TPIPE); | |
249 | lp->next = NULL; | |
250 | n1 = pipenode; | |
251 | } | |
252 | tokpushback++; | |
253 | return n1; | |
254 | } | |
255 | ||
256 | ||
257 | ||
258 | STATIC union node * | |
259 | command() { | |
260 | union node *n1, *n2; | |
261 | union node *ap, **app; | |
262 | union node *cp, **cpp; | |
263 | union node *redir, **rpp; | |
264 | int t; | |
265 | ||
266 | checkkwd = 2; | |
267 | switch (readtoken()) { | |
268 | case TIF: | |
269 | n1 = (union node *)stalloc(sizeof (struct nif)); | |
270 | n1->type = NIF; | |
271 | n1->nif.test = list(0); | |
272 | if (readtoken() != TTHEN) | |
273 | synexpect(TTHEN); | |
274 | n1->nif.ifpart = list(0); | |
275 | n2 = n1; | |
276 | while (readtoken() == TELIF) { | |
277 | n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); | |
278 | n2 = n2->nif.elsepart; | |
279 | n2->type = NIF; | |
280 | n2->nif.test = list(0); | |
281 | if (readtoken() != TTHEN) | |
282 | synexpect(TTHEN); | |
283 | n2->nif.ifpart = list(0); | |
284 | } | |
285 | if (lasttoken == TELSE) | |
286 | n2->nif.elsepart = list(0); | |
287 | else { | |
288 | n2->nif.elsepart = NULL; | |
289 | tokpushback++; | |
290 | } | |
291 | if (readtoken() != TFI) | |
292 | synexpect(TFI); | |
293 | checkkwd = 1; | |
294 | break; | |
295 | case TWHILE: | |
296 | case TUNTIL: { | |
297 | int got; | |
298 | n1 = (union node *)stalloc(sizeof (struct nbinary)); | |
299 | n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; | |
300 | n1->nbinary.ch1 = list(0); | |
301 | if ((got=readtoken()) != TDO) { | |
302 | TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); | |
303 | synexpect(TDO); | |
304 | } | |
305 | n1->nbinary.ch2 = list(0); | |
306 | if (readtoken() != TDONE) | |
307 | synexpect(TDONE); | |
308 | checkkwd = 1; | |
309 | break; | |
310 | } | |
311 | case TFOR: | |
312 | if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) | |
313 | synerror("Bad for loop variable"); | |
314 | n1 = (union node *)stalloc(sizeof (struct nfor)); | |
315 | n1->type = NFOR; | |
316 | n1->nfor.var = wordtext; | |
317 | if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { | |
318 | app = ≈ | |
319 | while (readtoken() == TWORD) { | |
320 | n2 = (union node *)stalloc(sizeof (struct narg)); | |
321 | n2->type = NARG; | |
322 | n2->narg.text = wordtext; | |
323 | n2->narg.backquote = backquotelist; | |
324 | *app = n2; | |
325 | app = &n2->narg.next; | |
326 | } | |
327 | *app = NULL; | |
328 | n1->nfor.args = ap; | |
329 | } else { | |
330 | #ifndef GDB_HACK | |
331 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, | |
332 | '@', '=', '\0'}; | |
333 | #endif | |
334 | n2 = (union node *)stalloc(sizeof (struct narg)); | |
335 | n2->type = NARG; | |
336 | n2->narg.text = (char *)argvars; | |
337 | n2->narg.backquote = NULL; | |
338 | n2->narg.next = NULL; | |
339 | n1->nfor.args = n2; | |
340 | } | |
341 | if (lasttoken != TNL && lasttoken != TSEMI) | |
342 | synexpect(-1); | |
343 | checkkwd = 2; | |
344 | if ((t = readtoken()) == TDO) | |
345 | t = TDONE; | |
346 | else if (t == TBEGIN) | |
347 | t = TEND; | |
348 | else | |
349 | synexpect(-1); | |
350 | n1->nfor.body = list(0); | |
351 | if (readtoken() != t) | |
352 | synexpect(t); | |
353 | checkkwd = 1; | |
354 | break; | |
355 | case TCASE: | |
356 | n1 = (union node *)stalloc(sizeof (struct ncase)); | |
357 | n1->type = NCASE; | |
358 | if (readtoken() != TWORD) | |
359 | synexpect(TWORD); | |
360 | n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); | |
361 | n2->type = NARG; | |
362 | n2->narg.text = wordtext; | |
363 | n2->narg.backquote = backquotelist; | |
364 | n2->narg.next = NULL; | |
365 | while (readtoken() == TNL); | |
366 | if (lasttoken != TWORD || ! equal(wordtext, "in")) | |
367 | synerror("expecting \"in\""); | |
368 | cpp = &n1->ncase.cases; | |
369 | while (checkkwd = 2, readtoken() == TWORD) { | |
370 | *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); | |
371 | cp->type = NCLIST; | |
372 | app = &cp->nclist.pattern; | |
373 | for (;;) { | |
374 | *app = ap = (union node *)stalloc(sizeof (struct narg)); | |
375 | ap->type = NARG; | |
376 | ap->narg.text = wordtext; | |
377 | ap->narg.backquote = backquotelist; | |
378 | if (readtoken() != TPIPE) | |
379 | break; | |
380 | app = &ap->narg.next; | |
381 | if (readtoken() != TWORD) | |
382 | synexpect(TWORD); | |
383 | } | |
384 | ap->narg.next = NULL; | |
385 | if (lasttoken != TRP) | |
386 | synexpect(TRP); | |
387 | cp->nclist.body = list(0); | |
388 | if ((t = readtoken()) == TESAC) | |
389 | tokpushback++; | |
390 | else if (t != TENDCASE) | |
391 | synexpect(TENDCASE); | |
392 | cpp = &cp->nclist.next; | |
393 | } | |
394 | *cpp = NULL; | |
395 | if (lasttoken != TESAC) | |
396 | synexpect(TESAC); | |
397 | checkkwd = 1; | |
398 | break; | |
399 | case TLP: | |
400 | n1 = (union node *)stalloc(sizeof (struct nredir)); | |
401 | n1->type = NSUBSHELL; | |
402 | n1->nredir.n = list(0); | |
403 | n1->nredir.redirect = NULL; | |
404 | if (readtoken() != TRP) | |
405 | synexpect(TRP); | |
406 | checkkwd = 1; | |
407 | break; | |
408 | case TBEGIN: | |
409 | n1 = list(0); | |
410 | if (readtoken() != TEND) | |
411 | synexpect(TEND); | |
412 | checkkwd = 1; | |
413 | break; | |
414 | case TWORD: | |
415 | case TREDIR: | |
416 | tokpushback++; | |
417 | return simplecmd(); | |
418 | default: | |
419 | synexpect(-1); | |
420 | } | |
421 | ||
422 | /* Now check for redirection which may follow command */ | |
423 | rpp = &redir; | |
424 | while (readtoken() == TREDIR) { | |
425 | *rpp = n2 = redirnode; | |
426 | rpp = &n2->nfile.next; | |
427 | parsefname(); | |
428 | } | |
429 | tokpushback++; | |
430 | *rpp = NULL; | |
431 | if (redir) { | |
432 | if (n1->type != NSUBSHELL) { | |
433 | n2 = (union node *)stalloc(sizeof (struct nredir)); | |
434 | n2->type = NREDIR; | |
435 | n2->nredir.n = n1; | |
436 | n1 = n2; | |
437 | } | |
438 | n1->nredir.redirect = redir; | |
439 | } | |
440 | return n1; | |
441 | } | |
442 | ||
443 | ||
444 | STATIC union node * | |
445 | simplecmd() { | |
446 | union node *args, **app; | |
447 | union node *redir, **rpp; | |
448 | union node *n; | |
449 | ||
450 | args = NULL; | |
451 | app = &args; | |
452 | rpp = &redir; | |
453 | for (;;) { | |
454 | if (readtoken() == TWORD) { | |
455 | n = (union node *)stalloc(sizeof (struct narg)); | |
456 | n->type = NARG; | |
457 | n->narg.text = wordtext; | |
458 | n->narg.backquote = backquotelist; | |
459 | *app = n; | |
460 | app = &n->narg.next; | |
461 | } else if (lasttoken == TREDIR) { | |
462 | *rpp = n = redirnode; | |
463 | rpp = &n->nfile.next; | |
464 | parsefname(); /* read name of redirection file */ | |
465 | } else if (lasttoken == TLP && app == &args->narg.next | |
466 | && rpp == &redir) { | |
467 | /* We have a function */ | |
468 | if (readtoken() != TRP) | |
469 | synexpect(TRP); | |
470 | #ifdef notdef | |
471 | if (! goodname(n->narg.text)) | |
472 | synerror("Bad function name"); | |
473 | #endif | |
474 | n->type = NDEFUN; | |
475 | n->narg.next = command(); | |
476 | return n; | |
477 | } else { | |
478 | tokpushback++; | |
479 | break; | |
480 | } | |
481 | } | |
482 | *app = NULL; | |
483 | *rpp = NULL; | |
484 | n = (union node *)stalloc(sizeof (struct ncmd)); | |
485 | n->type = NCMD; | |
486 | n->ncmd.backgnd = 0; | |
487 | n->ncmd.args = args; | |
488 | n->ncmd.redirect = redir; | |
489 | return n; | |
490 | } | |
491 | ||
492 | ||
493 | STATIC void | |
494 | parsefname() { | |
495 | union node *n = redirnode; | |
496 | ||
497 | if (readtoken() != TWORD) | |
498 | synexpect(-1); | |
499 | if (n->type == NHERE) { | |
500 | struct heredoc *here = heredoc; | |
501 | struct heredoc *p; | |
502 | int i; | |
503 | ||
504 | if (quoteflag == 0) | |
505 | n->type = NXHERE; | |
506 | TRACE(("Here document %d\n", n->type)); | |
507 | if (here->striptabs) { | |
508 | while (*wordtext == '\t') | |
509 | wordtext++; | |
510 | } | |
511 | if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) | |
512 | synerror("Illegal eof marker for << redirection"); | |
513 | rmescapes(wordtext); | |
514 | here->eofmark = wordtext; | |
515 | here->next = NULL; | |
516 | if (heredoclist == NULL) | |
517 | heredoclist = here; | |
518 | else { | |
519 | for (p = heredoclist ; p->next ; p = p->next); | |
520 | p->next = here; | |
521 | } | |
522 | } else if (n->type == NTOFD || n->type == NFROMFD) { | |
523 | if (is_digit(wordtext[0])) | |
524 | n->ndup.dupfd = digit_val(wordtext[0]); | |
525 | else if (wordtext[0] == '-') | |
526 | n->ndup.dupfd = -1; | |
527 | else | |
528 | goto bad; | |
529 | if (wordtext[1] != '\0') { | |
530 | bad: | |
531 | synerror("Bad fd number"); | |
532 | } | |
533 | } else { | |
534 | n->nfile.fname = (union node *)stalloc(sizeof (struct narg)); | |
535 | n = n->nfile.fname; | |
536 | n->type = NARG; | |
537 | n->narg.next = NULL; | |
538 | n->narg.text = wordtext; | |
539 | n->narg.backquote = backquotelist; | |
540 | } | |
541 | } | |
542 | ||
543 | ||
544 | /* | |
545 | * Input any here documents. | |
546 | */ | |
547 | ||
548 | STATIC void | |
549 | parseheredoc() { | |
550 | struct heredoc *here; | |
551 | union node *n; | |
552 | ||
553 | while (heredoclist) { | |
554 | here = heredoclist; | |
555 | heredoclist = here->next; | |
556 | if (needprompt) { | |
557 | putprompt(ps2val()); | |
558 | needprompt = 0; | |
559 | } | |
560 | readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, | |
561 | here->eofmark, here->striptabs); | |
562 | n = (union node *)stalloc(sizeof (struct narg)); | |
563 | n->narg.type = NARG; | |
564 | n->narg.next = NULL; | |
565 | n->narg.text = wordtext; | |
566 | n->narg.backquote = backquotelist; | |
567 | here->here->nhere.doc = n; | |
568 | } | |
569 | } | |
570 | ||
571 | STATIC int | |
572 | peektoken() { | |
573 | int t; | |
574 | ||
575 | t = readtoken(); | |
576 | tokpushback++; | |
577 | return (t); | |
578 | } | |
579 | ||
580 | STATIC int xxreadtoken(); | |
581 | ||
582 | STATIC int | |
583 | readtoken() { | |
584 | int t; | |
585 | #ifdef DEBUG | |
586 | int alreadyseen = tokpushback; | |
587 | #endif | |
588 | ||
589 | t = xxreadtoken(); | |
590 | ||
591 | if (checkkwd) { | |
592 | /* | |
593 | * eat newlines | |
594 | */ | |
595 | if (checkkwd == 2) { | |
596 | checkkwd = 0; | |
597 | while (t == TNL) { | |
598 | parseheredoc(); | |
599 | t = xxreadtoken(); | |
600 | } | |
601 | } else | |
602 | checkkwd = 0; | |
603 | /* | |
604 | * check for keywords | |
605 | */ | |
606 | if (t == TWORD && !quoteflag) { | |
607 | register char **pp; | |
608 | ||
609 | for (pp = parsekwd; *pp; pp++) { | |
610 | if (**pp == *wordtext && equal(*pp, wordtext)) { | |
611 | lasttoken = t = pp - parsekwd + KWDOFFSET; | |
612 | TRACE(("keyword %s recognized\n", tokname[t])); | |
613 | break; | |
614 | } | |
615 | } | |
616 | } | |
617 | } | |
618 | #ifdef DEBUG | |
619 | if (!alreadyseen) | |
620 | TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); | |
621 | else | |
622 | TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); | |
623 | #endif | |
624 | return (t); | |
625 | } | |
626 | ||
627 | ||
628 | /* | |
629 | * Read the next input token. | |
630 | * If the token is a word, we set backquotelist to the list of cmds in | |
631 | * backquotes. We set quoteflag to true if any part of the word was | |
632 | * quoted. | |
633 | * If the token is TREDIR, then we set redirnode to a structure containing | |
634 | * the redirection. | |
635 | * In all cases, the variable startlinno is set to the number of the line | |
636 | * on which the token starts. | |
637 | * | |
638 | * [Change comment: here documents and internal procedures] | |
639 | * [Readtoken shouldn't have any arguments. Perhaps we should make the | |
640 | * word parsing code into a separate routine. In this case, readtoken | |
641 | * doesn't need to have any internal procedures, but parseword does. | |
642 | * We could also make parseoperator in essence the main routine, and | |
643 | * have parseword (readtoken1?) handle both words and redirection.] | |
644 | */ | |
645 | ||
646 | #define RETURN(token) return lasttoken = token | |
647 | ||
648 | STATIC int | |
649 | xxreadtoken() { | |
650 | register c; | |
651 | ||
652 | if (tokpushback) { | |
653 | tokpushback = 0; | |
654 | return lasttoken; | |
655 | } | |
656 | if (needprompt) { | |
657 | putprompt(ps2val()); | |
658 | needprompt = 0; | |
659 | } | |
660 | startlinno = plinno; | |
661 | for (;;) { /* until token or start of word found */ | |
662 | c = pgetc_macro(); | |
663 | if (c == ' ' || c == '\t') | |
664 | continue; /* quick check for white space first */ | |
665 | switch (c) { | |
666 | case ' ': case '\t': | |
667 | continue; | |
668 | case '#': | |
669 | while ((c = pgetc()) != '\n' && c != PEOF); | |
670 | pungetc(); | |
671 | continue; | |
672 | case '\\': | |
673 | if (pgetc() == '\n') { | |
674 | startlinno = ++plinno; | |
675 | if (doprompt) | |
676 | putprompt(ps2val()); | |
677 | continue; | |
678 | } | |
679 | pungetc(); | |
680 | goto breakloop; | |
681 | case '\n': | |
682 | plinno++; | |
683 | needprompt = doprompt; | |
684 | RETURN(TNL); | |
685 | case PEOF: | |
686 | RETURN(TEOF); | |
687 | case '&': | |
688 | if (pgetc() == '&') | |
689 | RETURN(TAND); | |
690 | pungetc(); | |
691 | RETURN(TBACKGND); | |
692 | case '|': | |
693 | if (pgetc() == '|') | |
694 | RETURN(TOR); | |
695 | pungetc(); | |
696 | RETURN(TPIPE); | |
697 | case ';': | |
698 | if (pgetc() == ';') | |
699 | RETURN(TENDCASE); | |
700 | pungetc(); | |
701 | RETURN(TSEMI); | |
702 | case '(': | |
703 | RETURN(TLP); | |
704 | case ')': | |
705 | RETURN(TRP); | |
706 | default: | |
707 | goto breakloop; | |
708 | } | |
709 | } | |
710 | breakloop: | |
711 | return readtoken1(c, BASESYNTAX, (char *)NULL, 0); | |
712 | #undef RETURN | |
713 | } | |
714 | ||
715 | ||
716 | ||
717 | /* | |
718 | * If eofmark is NULL, read a word or a redirection symbol. If eofmark | |
719 | * is not NULL, read a here document. In the latter case, eofmark is the | |
720 | * word which marks the end of the document and striptabs is true if | |
721 | * leading tabs should be stripped from the document. The argument firstc | |
722 | * is the first character of the input token or document. | |
723 | * | |
724 | * Because C does not have internal subroutines, I have simulated them | |
725 | * using goto's to implement the subroutine linkage. The following macros | |
726 | * will run code that appears at the end of readtoken1. | |
727 | */ | |
728 | ||
729 | #define CHECKEND() {goto checkend; checkend_return:;} | |
730 | #define PARSEREDIR() {goto parseredir; parseredir_return:;} | |
731 | #define PARSESUB() {goto parsesub; parsesub_return:;} | |
732 | #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} | |
733 | #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} | |
734 | ||
735 | STATIC int | |
736 | readtoken1(firstc, syntax, eofmark, striptabs) | |
737 | int firstc; | |
738 | char const *syntax; | |
739 | char *eofmark; | |
740 | int striptabs; | |
741 | { | |
742 | register c = firstc; | |
743 | register char *out; | |
744 | int len; | |
745 | char line[EOFMARKLEN + 1]; | |
746 | struct nodelist *bqlist; | |
747 | int quotef; | |
748 | int dblquote; | |
749 | int varnest; | |
750 | int oldstyle; | |
751 | ||
752 | startlinno = plinno; | |
753 | dblquote = 0; | |
754 | if (syntax == DQSYNTAX) | |
755 | dblquote = 1; | |
756 | quotef = 0; | |
757 | bqlist = NULL; | |
758 | varnest = 0; | |
759 | STARTSTACKSTR(out); | |
760 | loop: { /* for each line, until end of word */ | |
761 | #if ATTY | |
762 | if (c == '\034' && doprompt | |
763 | && attyset() && ! equal(termval(), "emacs")) { | |
764 | attyline(); | |
765 | if (syntax == BASESYNTAX) | |
766 | return readtoken(); | |
767 | c = pgetc(); | |
768 | goto loop; | |
769 | } | |
770 | #endif | |
771 | CHECKEND(); /* set c to PEOF if at end of here document */ | |
772 | for (;;) { /* until end of line or end of word */ | |
773 | CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ | |
774 | switch(syntax[c]) { | |
775 | case CNL: /* '\n' */ | |
776 | if (syntax == BASESYNTAX) | |
777 | goto endword; /* exit outer loop */ | |
778 | USTPUTC(c, out); | |
779 | plinno++; | |
780 | if (doprompt) { | |
781 | putprompt(ps2val()); | |
782 | } | |
783 | c = pgetc(); | |
784 | goto loop; /* continue outer loop */ | |
785 | case CWORD: | |
786 | USTPUTC(c, out); | |
787 | break; | |
788 | case CCTL: | |
789 | if (eofmark == NULL || dblquote) | |
790 | USTPUTC(CTLESC, out); | |
791 | USTPUTC(c, out); | |
792 | break; | |
793 | case CBACK: /* backslash */ | |
794 | c = pgetc(); | |
795 | if (c == PEOF) { | |
796 | USTPUTC('\\', out); | |
797 | pungetc(); | |
798 | } else if (c == '\n') { | |
799 | if (doprompt) | |
800 | putprompt(ps2val()); | |
801 | } else { | |
802 | if (dblquote && c != '\\' && c != '`' && c != '$' | |
803 | && (c != '"' || eofmark != NULL)) | |
804 | USTPUTC('\\', out); | |
805 | if (SQSYNTAX[c] == CCTL) | |
806 | USTPUTC(CTLESC, out); | |
807 | USTPUTC(c, out); | |
808 | quotef++; | |
809 | } | |
810 | break; | |
811 | case CSQUOTE: | |
812 | syntax = SQSYNTAX; | |
813 | break; | |
814 | case CDQUOTE: | |
815 | syntax = DQSYNTAX; | |
816 | dblquote = 1; | |
817 | break; | |
818 | case CENDQUOTE: | |
819 | if (eofmark) { | |
820 | USTPUTC(c, out); | |
821 | } else { | |
822 | syntax = BASESYNTAX; | |
823 | quotef++; | |
824 | dblquote = 0; | |
825 | } | |
826 | break; | |
827 | case CVAR: /* '$' */ | |
828 | PARSESUB(); /* parse substitution */ | |
829 | break; | |
830 | case CENDVAR: /* '}' */ | |
831 | if (varnest > 0) { | |
832 | varnest--; | |
833 | USTPUTC(CTLENDVAR, out); | |
834 | } else { | |
835 | USTPUTC(c, out); | |
836 | } | |
837 | break; | |
838 | case CBQUOTE: /* '`' */ | |
839 | if (parsebackquote && syntax == BASESYNTAX) { | |
840 | if (out == stackblock()) | |
841 | return lasttoken = TENDBQUOTE; | |
842 | else | |
843 | goto endword; /* exit outer loop */ | |
844 | } | |
845 | PARSEBACKQOLD(); | |
846 | break; | |
847 | case CEOF: | |
848 | goto endword; /* exit outer loop */ | |
849 | default: | |
850 | if (varnest == 0) | |
851 | goto endword; /* exit outer loop */ | |
852 | USTPUTC(c, out); | |
853 | } | |
854 | c = pgetc_macro(); | |
855 | } | |
856 | } | |
857 | endword: | |
858 | if (syntax != BASESYNTAX && eofmark == NULL) | |
859 | synerror("Unterminated quoted string"); | |
860 | if (varnest != 0) { | |
861 | startlinno = plinno; | |
862 | synerror("Missing '}'"); | |
863 | } | |
864 | USTPUTC('\0', out); | |
865 | len = out - stackblock(); | |
866 | out = stackblock(); | |
867 | if (eofmark == NULL) { | |
868 | if ((c == '>' || c == '<') | |
869 | && quotef == 0 | |
870 | && len <= 2 | |
871 | && (*out == '\0' || is_digit(*out))) { | |
872 | PARSEREDIR(); | |
873 | return lasttoken = TREDIR; | |
874 | } else { | |
875 | pungetc(); | |
876 | } | |
877 | } | |
878 | quoteflag = quotef; | |
879 | backquotelist = bqlist; | |
880 | grabstackblock(len); | |
881 | wordtext = out; | |
882 | return lasttoken = TWORD; | |
883 | /* end of readtoken routine */ | |
884 | ||
885 | ||
886 | ||
887 | /* | |
888 | * Check to see whether we are at the end of the here document. When this | |
889 | * is called, c is set to the first character of the next input line. If | |
890 | * we are at the end of the here document, this routine sets the c to PEOF. | |
891 | */ | |
892 | ||
893 | checkend: { | |
894 | if (eofmark) { | |
895 | if (striptabs) { | |
896 | while (c == '\t') | |
897 | c = pgetc(); | |
898 | } | |
899 | if (c == *eofmark) { | |
900 | if (pfgets(line, sizeof line) != NULL) { | |
901 | register char *p, *q; | |
902 | ||
903 | p = line; | |
904 | for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); | |
905 | if (*p == '\n' && *q == '\0') { | |
906 | c = PEOF; | |
907 | plinno++; | |
908 | needprompt = doprompt; | |
909 | } else { | |
910 | ppushback(line, strlen(line)); | |
911 | } | |
912 | } | |
913 | } | |
914 | } | |
915 | goto checkend_return; | |
916 | } | |
917 | ||
918 | ||
919 | /* | |
920 | * Parse a redirection operator. The variable "out" points to a string | |
921 | * specifying the fd to be redirected. The variable "c" contains the | |
922 | * first character of the redirection operator. | |
923 | */ | |
924 | ||
925 | parseredir: { | |
926 | char fd = *out; | |
927 | union node *np; | |
928 | ||
929 | np = (union node *)stalloc(sizeof (struct nfile)); | |
930 | if (c == '>') { | |
931 | np->nfile.fd = 1; | |
932 | c = pgetc(); | |
933 | if (c == '>') | |
934 | np->type = NAPPEND; | |
935 | else if (c == '&') | |
936 | np->type = NTOFD; | |
937 | else { | |
938 | np->type = NTO; | |
939 | pungetc(); | |
940 | } | |
941 | } else { /* c == '<' */ | |
942 | np->nfile.fd = 0; | |
943 | c = pgetc(); | |
944 | if (c == '<') { | |
945 | if (sizeof (struct nfile) != sizeof (struct nhere)) { | |
946 | np = (union node *)stalloc(sizeof (struct nhere)); | |
947 | np->nfile.fd = 0; | |
948 | } | |
949 | np->type = NHERE; | |
950 | heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); | |
951 | heredoc->here = np; | |
952 | if ((c = pgetc()) == '-') { | |
953 | heredoc->striptabs = 1; | |
954 | } else { | |
955 | heredoc->striptabs = 0; | |
956 | pungetc(); | |
957 | } | |
958 | } else if (c == '&') | |
959 | np->type = NFROMFD; | |
960 | else { | |
961 | np->type = NFROM; | |
962 | pungetc(); | |
963 | } | |
964 | } | |
965 | if (fd != '\0') | |
966 | np->nfile.fd = digit_val(fd); | |
967 | redirnode = np; | |
968 | goto parseredir_return; | |
969 | } | |
970 | ||
971 | ||
972 | /* | |
973 | * Parse a substitution. At this point, we have read the dollar sign | |
974 | * and nothing else. | |
975 | */ | |
976 | ||
977 | parsesub: { | |
978 | int subtype; | |
979 | int typeloc; | |
980 | int flags; | |
981 | char *p; | |
982 | #ifndef GDB_HACK | |
983 | static const char types[] = "}-+?="; | |
984 | #endif | |
985 | ||
986 | c = pgetc(); | |
987 | if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) { | |
988 | USTPUTC('$', out); | |
989 | pungetc(); | |
990 | } else if (c == '(') { /* $(command) */ | |
991 | PARSEBACKQNEW(); | |
992 | } else { | |
993 | USTPUTC(CTLVAR, out); | |
994 | typeloc = out - stackblock(); | |
995 | USTPUTC(VSNORMAL, out); | |
996 | subtype = VSNORMAL; | |
997 | if (c == '{') { | |
998 | c = pgetc(); | |
999 | subtype = 0; | |
1000 | } | |
1001 | if (is_name(c)) { | |
1002 | do { | |
1003 | STPUTC(c, out); | |
1004 | c = pgetc(); | |
1005 | } while (is_in_name(c)); | |
1006 | } else { | |
1007 | if (! is_special(c)) | |
1008 | badsub: synerror("Bad substitution"); | |
1009 | USTPUTC(c, out); | |
1010 | c = pgetc(); | |
1011 | } | |
1012 | STPUTC('=', out); | |
1013 | flags = 0; | |
1014 | if (subtype == 0) { | |
1015 | if (c == ':') { | |
1016 | flags = VSNUL; | |
1017 | c = pgetc(); | |
1018 | } | |
1019 | p = strchr(types, c); | |
1020 | if (p == NULL) | |
1021 | goto badsub; | |
1022 | subtype = p - types + VSNORMAL; | |
1023 | } else { | |
1024 | pungetc(); | |
1025 | } | |
1026 | if (dblquote) | |
1027 | flags |= VSQUOTE; | |
1028 | *(stackblock() + typeloc) = subtype | flags; | |
1029 | if (subtype != VSNORMAL) | |
1030 | varnest++; | |
1031 | } | |
1032 | goto parsesub_return; | |
1033 | } | |
1034 | ||
1035 | ||
1036 | /* | |
1037 | * Called to parse command substitutions. Newstyle is set if the command | |
1038 | * is enclosed inside $(...); nlpp is a pointer to the head of the linked | |
1039 | * list of commands (passed by reference), and savelen is the number of | |
1040 | * characters on the top of the stack which must be preserved. | |
1041 | */ | |
1042 | ||
1043 | parsebackq: { | |
1044 | struct nodelist **nlpp; | |
1045 | int savepbq; | |
1046 | union node *n; | |
1047 | char *volatile str; | |
1048 | struct jmploc jmploc; | |
1049 | struct jmploc *volatile savehandler; | |
1050 | int savelen; | |
1051 | int t; | |
1052 | ||
1053 | savepbq = parsebackquote; | |
1054 | if (setjmp(jmploc.loc)) { | |
1055 | if (str) | |
1056 | ckfree(str); | |
1057 | parsebackquote = 0; | |
1058 | handler = savehandler; | |
1059 | longjmp(handler, 1); | |
1060 | } | |
1061 | INTOFF; | |
1062 | str = NULL; | |
1063 | savelen = out - stackblock(); | |
1064 | if (savelen > 0) { | |
1065 | str = ckmalloc(savelen); | |
1066 | bcopy(stackblock(), str, savelen); | |
1067 | } | |
1068 | savehandler = handler; | |
1069 | handler = &jmploc; | |
1070 | INTON; | |
1071 | nlpp = &bqlist; | |
1072 | while (*nlpp) | |
1073 | nlpp = &(*nlpp)->next; | |
1074 | *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
1075 | (*nlpp)->next = NULL; | |
1076 | parsebackquote = oldstyle; | |
1077 | n = list(0); | |
1078 | t = oldstyle? TENDBQUOTE : TRP; | |
1079 | if (readtoken() != t) | |
1080 | synexpect(t); | |
1081 | (*nlpp)->n = n; | |
1082 | while (stackblocksize() <= savelen) | |
1083 | growstackblock(); | |
1084 | STARTSTACKSTR(out); | |
1085 | if (str) { | |
1086 | bcopy(str, out, savelen); | |
1087 | STADJUST(savelen, out); | |
1088 | INTOFF; | |
1089 | ckfree(str); | |
1090 | str = NULL; | |
1091 | INTON; | |
1092 | } | |
1093 | parsebackquote = savepbq; | |
1094 | handler = savehandler; | |
1095 | USTPUTC(CTLBACKQ + dblquote, out); | |
1096 | if (oldstyle) | |
1097 | goto parsebackq_oldreturn; | |
1098 | else | |
1099 | goto parsebackq_newreturn; | |
1100 | } | |
1101 | ||
1102 | } /* end of readtoken */ | |
1103 | ||
1104 | ||
1105 | ||
1106 | #ifdef mkinit | |
1107 | RESET { | |
1108 | tokpushback = 0; | |
1109 | } | |
1110 | #endif | |
1111 | ||
1112 | ||
1113 | #if ATTY | |
1114 | /* | |
1115 | * Called to process a command generated by atty. We execute the line, | |
1116 | * and catch any errors that occur so they don't propagate outside of | |
1117 | * this routine. | |
1118 | */ | |
1119 | ||
1120 | STATIC void | |
1121 | attyline() { | |
1122 | char line[256]; | |
1123 | struct stackmark smark; | |
1124 | struct jmploc jmploc; | |
1125 | struct jmploc *volatile savehandler; | |
1126 | ||
1127 | if (pfgets(line, sizeof line) == NULL) | |
1128 | return; /* "can't happen" */ | |
1129 | if (setjmp(jmploc.loc)) { | |
1130 | if (exception == EXERROR) | |
1131 | out2str("\033]D\n"); | |
1132 | handler = savehandler; | |
1133 | longjmp(handler, 1); | |
1134 | } | |
1135 | savehandler = handler; | |
1136 | handler = &jmploc; | |
1137 | setstackmark(&smark); | |
1138 | evalstring(line); | |
1139 | popstackmark(&smark); | |
1140 | handler = savehandler; | |
1141 | doprompt = 1; | |
1142 | } | |
1143 | ||
1144 | ||
1145 | /* | |
1146 | * Output a prompt for atty. We output the prompt as part of the | |
1147 | * appropriate escape sequence. | |
1148 | */ | |
1149 | ||
1150 | STATIC void | |
1151 | putprompt(s) | |
1152 | char *s; | |
1153 | { | |
1154 | register char *p; | |
1155 | ||
1156 | if (attyset() && ! equal(termval(), "emacs")) { | |
1157 | if (strchr(s, '\7')) | |
1158 | out2c('\7'); | |
1159 | out2str("\033]P1;"); | |
1160 | for (p = s ; *p ; p++) { | |
1161 | if ((unsigned)(*p - ' ') <= '~' - ' ') | |
1162 | out2c(*p); | |
1163 | } | |
1164 | out2c('\n'); | |
1165 | } else { | |
1166 | out2str(s); | |
1167 | } | |
1168 | } | |
1169 | #endif | |
1170 | ||
1171 | ||
1172 | ||
1173 | /* | |
1174 | * Returns true if the text contains nothing to expand (no dollar signs | |
1175 | * or backquotes). | |
1176 | */ | |
1177 | ||
1178 | STATIC int | |
1179 | noexpand(text) | |
1180 | char *text; | |
1181 | { | |
1182 | register char *p; | |
1183 | register char c; | |
1184 | ||
1185 | p = text; | |
1186 | while ((c = *p++) != '\0') { | |
1187 | if (c == CTLESC) | |
1188 | p++; | |
1189 | else if (BASESYNTAX[c] == CCTL) | |
1190 | return 0; | |
1191 | } | |
1192 | return 1; | |
1193 | } | |
1194 | ||
1195 | ||
1196 | /* | |
1197 | * Return true if the argument is a legal variable name (a letter or | |
1198 | * underscore followed by zero or more letters, underscores, and digits). | |
1199 | */ | |
1200 | ||
1201 | int | |
1202 | goodname(name) | |
1203 | char *name; | |
1204 | { | |
1205 | register char *p; | |
1206 | ||
1207 | p = name; | |
1208 | if (! is_name(*p)) | |
1209 | return 0; | |
1210 | while (*++p) { | |
1211 | if (! is_in_name(*p)) | |
1212 | return 0; | |
1213 | } | |
1214 | return 1; | |
1215 | } | |
1216 | ||
1217 | ||
1218 | /* | |
1219 | * Called when an unexpected token is read during the parse. The argument | |
1220 | * is the token that is expected, or -1 if more than one type of token can | |
1221 | * occur at this point. | |
1222 | */ | |
1223 | ||
1224 | STATIC void | |
1225 | synexpect(token) { | |
1226 | char msg[64]; | |
1227 | ||
1228 | if (token >= 0) { | |
1229 | fmtstr(msg, 64, "%s unexpected (expecting %s)", | |
1230 | tokname[lasttoken], tokname[token]); | |
1231 | } else { | |
1232 | fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); | |
1233 | } | |
1234 | synerror(msg); | |
1235 | } | |
1236 | ||
1237 | ||
1238 | STATIC void | |
1239 | synerror(msg) | |
1240 | char *msg; | |
1241 | { | |
1242 | if (commandname) | |
1243 | outfmt(&errout, "%s: %d: ", commandname, startlinno); | |
1244 | outfmt(&errout, "Syntax error: %s\n", msg); | |
1245 | error((char *)NULL); | |
1246 | } |