Commit | Line | Data |
---|---|---|
d787dbb5 KB |
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 | |
1545e6b2 | 12 | static char sccsid[] = "@(#)parser.c 5.4 (Berkeley) %G%"; |
d787dbb5 KB |
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 */ | |
21679e1e | 57 | int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ |
d787dbb5 KB |
58 | struct nodelist *backquotelist; |
59 | union node *redirnode; | |
60 | struct heredoc *heredoc; | |
61 | int quoteflag; /* set if (part of) last token was quoted */ | |
62 | int startlinno; /* line # where last token started */ | |
63 | ||
64 | ||
65 | #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ | |
66 | #ifdef GDB_HACK | |
67 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; | |
68 | static const char types[] = "}-+?="; | |
69 | #endif | |
70 | ||
71 | ||
21679e1e MT |
72 | STATIC union node *list __P((int)); |
73 | STATIC union node *andor __P((void)); | |
74 | STATIC union node *pipeline __P((void)); | |
75 | STATIC union node *command __P((void)); | |
76 | STATIC union node *simplecmd __P((void)); | |
77 | STATIC void parsefname __P((void)); | |
78 | STATIC void parseheredoc __P((void)); | |
79 | STATIC int readtoken __P((void)); | |
80 | STATIC int readtoken1 __P((int, char const *, char *, int)); | |
81 | STATIC void attyline __P((void)); | |
82 | STATIC int noexpand __P((char *)); | |
83 | STATIC void synexpect __P((int)); | |
84 | STATIC void synerror __P((char *)); | |
d787dbb5 KB |
85 | |
86 | #if ATTY | |
21679e1e | 87 | STATIC void putprompt __P((char *)); |
d787dbb5 KB |
88 | #else /* not ATTY */ |
89 | #define putprompt(s) out2str(s) | |
90 | #endif | |
91 | ||
92 | ||
93 | ||
94 | ||
95 | /* | |
96 | * Read and parse a command. Returns NEOF on end of file. (NULL is a | |
97 | * valid parse tree indicating a blank line.) | |
98 | */ | |
99 | ||
100 | union node * | |
101 | parsecmd(interact) { | |
102 | int t; | |
103 | ||
104 | doprompt = interact; | |
105 | if (doprompt) | |
106 | putprompt(ps1val()); | |
107 | needprompt = 0; | |
108 | if ((t = readtoken()) == TEOF) | |
109 | return NEOF; | |
110 | if (t == TNL) | |
111 | return NULL; | |
112 | tokpushback++; | |
113 | return list(1); | |
114 | } | |
115 | ||
116 | ||
117 | STATIC union node * | |
118 | list(nlflag) { | |
119 | union node *n1, *n2, *n3; | |
120 | ||
21679e1e MT |
121 | checkkwd = 2; |
122 | if (nlflag == 0 && tokendlist[peektoken()]) | |
d787dbb5 KB |
123 | return NULL; |
124 | n1 = andor(); | |
125 | for (;;) { | |
126 | switch (readtoken()) { | |
127 | case TBACKGND: | |
128 | if (n1->type == NCMD || n1->type == NPIPE) { | |
129 | n1->ncmd.backgnd = 1; | |
130 | } else if (n1->type == NREDIR) { | |
131 | n1->type = NBACKGND; | |
132 | } else { | |
133 | n3 = (union node *)stalloc(sizeof (struct nredir)); | |
134 | n3->type = NBACKGND; | |
135 | n3->nredir.n = n1; | |
136 | n3->nredir.redirect = NULL; | |
137 | n1 = n3; | |
138 | } | |
139 | goto tsemi; | |
140 | case TNL: | |
141 | tokpushback++; | |
142 | /* fall through */ | |
143 | tsemi: case TSEMI: | |
144 | if (readtoken() == TNL) { | |
145 | parseheredoc(); | |
146 | if (nlflag) | |
147 | return n1; | |
148 | } else { | |
149 | tokpushback++; | |
150 | } | |
21679e1e MT |
151 | checkkwd = 2; |
152 | if (tokendlist[peektoken()]) | |
d787dbb5 KB |
153 | return n1; |
154 | n2 = andor(); | |
155 | n3 = (union node *)stalloc(sizeof (struct nbinary)); | |
156 | n3->type = NSEMI; | |
157 | n3->nbinary.ch1 = n1; | |
158 | n3->nbinary.ch2 = n2; | |
159 | n1 = n3; | |
160 | break; | |
161 | case TEOF: | |
162 | if (heredoclist) | |
163 | parseheredoc(); | |
164 | else | |
165 | pungetc(); /* push back EOF on input */ | |
166 | return n1; | |
167 | default: | |
168 | if (nlflag) | |
169 | synexpect(-1); | |
170 | tokpushback++; | |
171 | return n1; | |
172 | } | |
173 | } | |
174 | } | |
175 | ||
176 | ||
177 | ||
178 | STATIC union node * | |
179 | andor() { | |
180 | union node *n1, *n2, *n3; | |
181 | int t; | |
182 | ||
183 | n1 = pipeline(); | |
184 | for (;;) { | |
185 | if ((t = readtoken()) == TAND) { | |
186 | t = NAND; | |
187 | } else if (t == TOR) { | |
188 | t = NOR; | |
189 | } else { | |
190 | tokpushback++; | |
191 | return n1; | |
192 | } | |
193 | n2 = pipeline(); | |
194 | n3 = (union node *)stalloc(sizeof (struct nbinary)); | |
195 | n3->type = t; | |
196 | n3->nbinary.ch1 = n1; | |
197 | n3->nbinary.ch2 = n2; | |
198 | n1 = n3; | |
199 | } | |
200 | } | |
201 | ||
202 | ||
203 | ||
204 | STATIC union node * | |
205 | pipeline() { | |
1545e6b2 | 206 | union node *n1, *pipenode, *notnode; |
d787dbb5 | 207 | struct nodelist *lp, *prev; |
1545e6b2 | 208 | int negate = 0; |
d787dbb5 | 209 | |
1545e6b2 MT |
210 | TRACE(("pipeline: entered\n")); |
211 | while (readtoken() == TNOT) { | |
212 | TRACE(("pipeline: TNOT recognized\n")); | |
213 | negate = !negate; | |
214 | } | |
215 | tokpushback++; | |
d787dbb5 KB |
216 | n1 = command(); |
217 | if (readtoken() == TPIPE) { | |
218 | pipenode = (union node *)stalloc(sizeof (struct npipe)); | |
219 | pipenode->type = NPIPE; | |
220 | pipenode->npipe.backgnd = 0; | |
221 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
222 | pipenode->npipe.cmdlist = lp; | |
223 | lp->n = n1; | |
224 | do { | |
225 | prev = lp; | |
226 | lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
227 | lp->n = command(); | |
228 | prev->next = lp; | |
229 | } while (readtoken() == TPIPE); | |
230 | lp->next = NULL; | |
231 | n1 = pipenode; | |
232 | } | |
233 | tokpushback++; | |
1545e6b2 MT |
234 | if (negate) { |
235 | notnode = (union node *)stalloc(sizeof (struct nnot)); | |
236 | notnode->type = NNOT; | |
237 | notnode->nnot.com = n1; | |
238 | n1 = notnode; | |
239 | } | |
d787dbb5 KB |
240 | return n1; |
241 | } | |
242 | ||
243 | ||
244 | ||
245 | STATIC union node * | |
246 | command() { | |
247 | union node *n1, *n2; | |
248 | union node *ap, **app; | |
249 | union node *cp, **cpp; | |
250 | union node *redir, **rpp; | |
251 | int t; | |
252 | ||
21679e1e | 253 | checkkwd = 2; |
d787dbb5 KB |
254 | switch (readtoken()) { |
255 | case TIF: | |
256 | n1 = (union node *)stalloc(sizeof (struct nif)); | |
257 | n1->type = NIF; | |
258 | n1->nif.test = list(0); | |
259 | if (readtoken() != TTHEN) | |
260 | synexpect(TTHEN); | |
261 | n1->nif.ifpart = list(0); | |
262 | n2 = n1; | |
263 | while (readtoken() == TELIF) { | |
264 | n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); | |
265 | n2 = n2->nif.elsepart; | |
266 | n2->type = NIF; | |
267 | n2->nif.test = list(0); | |
268 | if (readtoken() != TTHEN) | |
269 | synexpect(TTHEN); | |
270 | n2->nif.ifpart = list(0); | |
271 | } | |
272 | if (lasttoken == TELSE) | |
273 | n2->nif.elsepart = list(0); | |
274 | else { | |
275 | n2->nif.elsepart = NULL; | |
276 | tokpushback++; | |
277 | } | |
278 | if (readtoken() != TFI) | |
279 | synexpect(TFI); | |
21679e1e | 280 | checkkwd = 1; |
d787dbb5 KB |
281 | break; |
282 | case TWHILE: | |
21679e1e MT |
283 | case TUNTIL: { |
284 | int got; | |
d787dbb5 KB |
285 | n1 = (union node *)stalloc(sizeof (struct nbinary)); |
286 | n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; | |
287 | n1->nbinary.ch1 = list(0); | |
21679e1e MT |
288 | if ((got=readtoken()) != TDO) { |
289 | TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); | |
d787dbb5 | 290 | synexpect(TDO); |
21679e1e | 291 | } |
d787dbb5 KB |
292 | n1->nbinary.ch2 = list(0); |
293 | if (readtoken() != TDONE) | |
294 | synexpect(TDONE); | |
21679e1e | 295 | checkkwd = 1; |
d787dbb5 | 296 | break; |
21679e1e | 297 | } |
d787dbb5 KB |
298 | case TFOR: |
299 | if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) | |
300 | synerror("Bad for loop variable"); | |
301 | n1 = (union node *)stalloc(sizeof (struct nfor)); | |
302 | n1->type = NFOR; | |
303 | n1->nfor.var = wordtext; | |
304 | if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { | |
305 | app = ≈ | |
306 | while (readtoken() == TWORD) { | |
307 | n2 = (union node *)stalloc(sizeof (struct narg)); | |
308 | n2->type = NARG; | |
309 | n2->narg.text = wordtext; | |
310 | n2->narg.backquote = backquotelist; | |
311 | *app = n2; | |
312 | app = &n2->narg.next; | |
313 | } | |
314 | *app = NULL; | |
315 | n1->nfor.args = ap; | |
316 | } else { | |
317 | #ifndef GDB_HACK | |
318 | static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, | |
319 | '@', '=', '\0'}; | |
320 | #endif | |
321 | n2 = (union node *)stalloc(sizeof (struct narg)); | |
322 | n2->type = NARG; | |
323 | n2->narg.text = (char *)argvars; | |
324 | n2->narg.backquote = NULL; | |
325 | n2->narg.next = NULL; | |
326 | n1->nfor.args = n2; | |
327 | } | |
328 | if (lasttoken != TNL && lasttoken != TSEMI) | |
329 | synexpect(-1); | |
21679e1e | 330 | checkkwd = 2; |
d787dbb5 KB |
331 | if ((t = readtoken()) == TDO) |
332 | t = TDONE; | |
333 | else if (t == TBEGIN) | |
334 | t = TEND; | |
335 | else | |
336 | synexpect(-1); | |
337 | n1->nfor.body = list(0); | |
338 | if (readtoken() != t) | |
339 | synexpect(t); | |
21679e1e | 340 | checkkwd = 1; |
d787dbb5 KB |
341 | break; |
342 | case TCASE: | |
343 | n1 = (union node *)stalloc(sizeof (struct ncase)); | |
344 | n1->type = NCASE; | |
345 | if (readtoken() != TWORD) | |
346 | synexpect(TWORD); | |
347 | n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); | |
348 | n2->type = NARG; | |
349 | n2->narg.text = wordtext; | |
350 | n2->narg.backquote = backquotelist; | |
351 | n2->narg.next = NULL; | |
352 | while (readtoken() == TNL); | |
353 | if (lasttoken != TWORD || ! equal(wordtext, "in")) | |
354 | synerror("expecting \"in\""); | |
355 | cpp = &n1->ncase.cases; | |
21679e1e | 356 | while (checkkwd = 2, readtoken() == TWORD) { |
d787dbb5 KB |
357 | *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); |
358 | cp->type = NCLIST; | |
359 | app = &cp->nclist.pattern; | |
360 | for (;;) { | |
361 | *app = ap = (union node *)stalloc(sizeof (struct narg)); | |
362 | ap->type = NARG; | |
363 | ap->narg.text = wordtext; | |
364 | ap->narg.backquote = backquotelist; | |
365 | if (readtoken() != TPIPE) | |
366 | break; | |
367 | app = &ap->narg.next; | |
368 | if (readtoken() != TWORD) | |
369 | synexpect(TWORD); | |
370 | } | |
371 | ap->narg.next = NULL; | |
372 | if (lasttoken != TRP) | |
373 | synexpect(TRP); | |
374 | cp->nclist.body = list(0); | |
375 | if ((t = readtoken()) == TESAC) | |
376 | tokpushback++; | |
377 | else if (t != TENDCASE) | |
378 | synexpect(TENDCASE); | |
379 | cpp = &cp->nclist.next; | |
380 | } | |
381 | *cpp = NULL; | |
382 | if (lasttoken != TESAC) | |
383 | synexpect(TESAC); | |
21679e1e | 384 | checkkwd = 1; |
d787dbb5 KB |
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); | |
21679e1e | 393 | checkkwd = 1; |
d787dbb5 KB |
394 | break; |
395 | case TBEGIN: | |
396 | n1 = list(0); | |
397 | if (readtoken() != TEND) | |
398 | synexpect(TEND); | |
21679e1e | 399 | checkkwd = 1; |
d787dbb5 KB |
400 | break; |
401 | case TWORD: | |
402 | case TREDIR: | |
403 | tokpushback++; | |
404 | return simplecmd(); | |
405 | default: | |
406 | synexpect(-1); | |
407 | } | |
408 | ||
409 | /* Now check for redirection which may follow command */ | |
410 | rpp = &redir; | |
411 | while (readtoken() == TREDIR) { | |
412 | *rpp = n2 = redirnode; | |
413 | rpp = &n2->nfile.next; | |
414 | parsefname(); | |
415 | } | |
416 | tokpushback++; | |
417 | *rpp = NULL; | |
418 | if (redir) { | |
419 | if (n1->type != NSUBSHELL) { | |
420 | n2 = (union node *)stalloc(sizeof (struct nredir)); | |
421 | n2->type = NREDIR; | |
422 | n2->nredir.n = n1; | |
423 | n1 = n2; | |
424 | } | |
425 | n1->nredir.redirect = redir; | |
426 | } | |
427 | return n1; | |
428 | } | |
429 | ||
430 | ||
431 | STATIC union node * | |
432 | simplecmd() { | |
433 | union node *args, **app; | |
434 | union node *redir, **rpp; | |
435 | union node *n; | |
436 | ||
437 | args = NULL; | |
438 | app = &args; | |
439 | rpp = &redir; | |
440 | for (;;) { | |
441 | if (readtoken() == TWORD) { | |
442 | n = (union node *)stalloc(sizeof (struct narg)); | |
443 | n->type = NARG; | |
444 | n->narg.text = wordtext; | |
445 | n->narg.backquote = backquotelist; | |
446 | *app = n; | |
447 | app = &n->narg.next; | |
448 | } else if (lasttoken == TREDIR) { | |
449 | *rpp = n = redirnode; | |
450 | rpp = &n->nfile.next; | |
451 | parsefname(); /* read name of redirection file */ | |
452 | } else if (lasttoken == TLP && app == &args->narg.next | |
453 | && rpp == &redir) { | |
454 | /* We have a function */ | |
455 | if (readtoken() != TRP) | |
456 | synexpect(TRP); | |
ddba57cd | 457 | #ifdef notdef |
d787dbb5 KB |
458 | if (! goodname(n->narg.text)) |
459 | synerror("Bad function name"); | |
ddba57cd | 460 | #endif |
d787dbb5 KB |
461 | n->type = NDEFUN; |
462 | n->narg.next = command(); | |
463 | return n; | |
464 | } else { | |
465 | tokpushback++; | |
466 | break; | |
467 | } | |
468 | } | |
469 | *app = NULL; | |
470 | *rpp = NULL; | |
471 | n = (union node *)stalloc(sizeof (struct ncmd)); | |
472 | n->type = NCMD; | |
473 | n->ncmd.backgnd = 0; | |
474 | n->ncmd.args = args; | |
475 | n->ncmd.redirect = redir; | |
476 | return n; | |
477 | } | |
478 | ||
479 | ||
480 | STATIC void | |
481 | parsefname() { | |
482 | union node *n = redirnode; | |
483 | ||
484 | if (readtoken() != TWORD) | |
485 | synexpect(-1); | |
486 | if (n->type == NHERE) { | |
487 | struct heredoc *here = heredoc; | |
488 | struct heredoc *p; | |
489 | int i; | |
490 | ||
491 | if (quoteflag == 0) | |
492 | n->type = NXHERE; | |
493 | TRACE(("Here document %d\n", n->type)); | |
494 | if (here->striptabs) { | |
495 | while (*wordtext == '\t') | |
496 | wordtext++; | |
497 | } | |
498 | if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) | |
499 | synerror("Illegal eof marker for << redirection"); | |
500 | rmescapes(wordtext); | |
501 | here->eofmark = wordtext; | |
502 | here->next = NULL; | |
503 | if (heredoclist == NULL) | |
504 | heredoclist = here; | |
505 | else { | |
506 | for (p = heredoclist ; p->next ; p = p->next); | |
507 | p->next = here; | |
508 | } | |
509 | } else if (n->type == NTOFD || n->type == NFROMFD) { | |
510 | if (is_digit(wordtext[0])) | |
511 | n->ndup.dupfd = digit_val(wordtext[0]); | |
512 | else if (wordtext[0] == '-') | |
513 | n->ndup.dupfd = -1; | |
514 | else | |
515 | goto bad; | |
516 | if (wordtext[1] != '\0') { | |
517 | bad: | |
518 | synerror("Bad fd number"); | |
519 | } | |
520 | } else { | |
521 | n->nfile.fname = (union node *)stalloc(sizeof (struct narg)); | |
522 | n = n->nfile.fname; | |
523 | n->type = NARG; | |
524 | n->narg.next = NULL; | |
525 | n->narg.text = wordtext; | |
526 | n->narg.backquote = backquotelist; | |
527 | } | |
528 | } | |
529 | ||
530 | ||
531 | /* | |
532 | * Input any here documents. | |
533 | */ | |
534 | ||
535 | STATIC void | |
536 | parseheredoc() { | |
537 | struct heredoc *here; | |
538 | union node *n; | |
539 | ||
540 | while (heredoclist) { | |
541 | here = heredoclist; | |
542 | heredoclist = here->next; | |
543 | if (needprompt) { | |
544 | putprompt(ps2val()); | |
545 | needprompt = 0; | |
546 | } | |
547 | readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, | |
548 | here->eofmark, here->striptabs); | |
549 | n = (union node *)stalloc(sizeof (struct narg)); | |
550 | n->narg.type = NARG; | |
551 | n->narg.next = NULL; | |
552 | n->narg.text = wordtext; | |
553 | n->narg.backquote = backquotelist; | |
554 | here->here->nhere.doc = n; | |
555 | } | |
556 | } | |
557 | ||
21679e1e MT |
558 | STATIC int |
559 | peektoken() { | |
d787dbb5 KB |
560 | int t; |
561 | ||
21679e1e | 562 | t = readtoken(); |
d787dbb5 | 563 | tokpushback++; |
21679e1e | 564 | return (t); |
d787dbb5 KB |
565 | } |
566 | ||
d787dbb5 KB |
567 | STATIC int xxreadtoken(); |
568 | ||
569 | STATIC int | |
570 | readtoken() { | |
571 | int t; | |
21679e1e MT |
572 | #ifdef DEBUG |
573 | int alreadyseen = tokpushback; | |
574 | #endif | |
575 | ||
576 | t = xxreadtoken(); | |
577 | ||
578 | if (checkkwd) { | |
579 | /* | |
580 | * eat newlines | |
581 | */ | |
582 | if (checkkwd == 2) { | |
583 | checkkwd = 0; | |
584 | while (t == TNL) { | |
585 | parseheredoc(); | |
586 | t = xxreadtoken(); | |
587 | } | |
588 | } else | |
589 | checkkwd = 0; | |
590 | /* | |
591 | * check for keywords | |
592 | */ | |
593 | if (t == TWORD && !quoteflag) { | |
594 | register char **pp; | |
595 | ||
596 | for (pp = parsekwd; *pp; pp++) { | |
597 | if (**pp == *wordtext && equal(*pp, wordtext)) { | |
598 | lasttoken = t = pp - parsekwd + KWDOFFSET; | |
599 | TRACE(("keyword %s recognized\n", tokname[t])); | |
600 | break; | |
601 | } | |
602 | } | |
603 | } | |
d787dbb5 | 604 | } |
21679e1e MT |
605 | #ifdef DEBUG |
606 | if (!alreadyseen) | |
607 | TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); | |
608 | else | |
609 | TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); | |
610 | #endif | |
611 | return (t); | |
d787dbb5 KB |
612 | } |
613 | ||
614 | ||
615 | /* | |
616 | * Read the next input token. | |
617 | * If the token is a word, we set backquotelist to the list of cmds in | |
618 | * backquotes. We set quoteflag to true if any part of the word was | |
619 | * quoted. | |
620 | * If the token is TREDIR, then we set redirnode to a structure containing | |
621 | * the redirection. | |
622 | * In all cases, the variable startlinno is set to the number of the line | |
623 | * on which the token starts. | |
624 | * | |
625 | * [Change comment: here documents and internal procedures] | |
626 | * [Readtoken shouldn't have any arguments. Perhaps we should make the | |
627 | * word parsing code into a separate routine. In this case, readtoken | |
628 | * doesn't need to have any internal procedures, but parseword does. | |
629 | * We could also make parseoperator in essence the main routine, and | |
630 | * have parseword (readtoken1?) handle both words and redirection.] | |
631 | */ | |
632 | ||
633 | #define RETURN(token) return lasttoken = token | |
634 | ||
635 | STATIC int | |
636 | xxreadtoken() { | |
637 | register c; | |
638 | ||
639 | if (tokpushback) { | |
640 | tokpushback = 0; | |
641 | return lasttoken; | |
642 | } | |
643 | if (needprompt) { | |
644 | putprompt(ps2val()); | |
645 | needprompt = 0; | |
646 | } | |
647 | startlinno = plinno; | |
648 | for (;;) { /* until token or start of word found */ | |
649 | c = pgetc_macro(); | |
650 | if (c == ' ' || c == '\t') | |
651 | continue; /* quick check for white space first */ | |
652 | switch (c) { | |
653 | case ' ': case '\t': | |
654 | continue; | |
655 | case '#': | |
656 | while ((c = pgetc()) != '\n' && c != PEOF); | |
657 | pungetc(); | |
658 | continue; | |
659 | case '\\': | |
660 | if (pgetc() == '\n') { | |
661 | startlinno = ++plinno; | |
662 | if (doprompt) | |
663 | putprompt(ps2val()); | |
664 | continue; | |
665 | } | |
666 | pungetc(); | |
667 | goto breakloop; | |
668 | case '\n': | |
669 | plinno++; | |
670 | needprompt = doprompt; | |
671 | RETURN(TNL); | |
672 | case PEOF: | |
673 | RETURN(TEOF); | |
674 | case '&': | |
675 | if (pgetc() == '&') | |
676 | RETURN(TAND); | |
677 | pungetc(); | |
678 | RETURN(TBACKGND); | |
679 | case '|': | |
680 | if (pgetc() == '|') | |
681 | RETURN(TOR); | |
682 | pungetc(); | |
683 | RETURN(TPIPE); | |
684 | case ';': | |
685 | if (pgetc() == ';') | |
686 | RETURN(TENDCASE); | |
687 | pungetc(); | |
688 | RETURN(TSEMI); | |
689 | case '(': | |
690 | RETURN(TLP); | |
691 | case ')': | |
692 | RETURN(TRP); | |
693 | default: | |
694 | goto breakloop; | |
695 | } | |
696 | } | |
697 | breakloop: | |
698 | return readtoken1(c, BASESYNTAX, (char *)NULL, 0); | |
699 | #undef RETURN | |
700 | } | |
701 | ||
702 | ||
703 | ||
704 | /* | |
705 | * If eofmark is NULL, read a word or a redirection symbol. If eofmark | |
706 | * is not NULL, read a here document. In the latter case, eofmark is the | |
707 | * word which marks the end of the document and striptabs is true if | |
708 | * leading tabs should be stripped from the document. The argument firstc | |
709 | * is the first character of the input token or document. | |
710 | * | |
711 | * Because C does not have internal subroutines, I have simulated them | |
712 | * using goto's to implement the subroutine linkage. The following macros | |
713 | * will run code that appears at the end of readtoken1. | |
714 | */ | |
715 | ||
716 | #define CHECKEND() {goto checkend; checkend_return:;} | |
717 | #define PARSEREDIR() {goto parseredir; parseredir_return:;} | |
718 | #define PARSESUB() {goto parsesub; parsesub_return:;} | |
719 | #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} | |
720 | #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} | |
721 | ||
722 | STATIC int | |
723 | readtoken1(firstc, syntax, eofmark, striptabs) | |
724 | int firstc; | |
725 | char const *syntax; | |
726 | char *eofmark; | |
727 | int striptabs; | |
728 | { | |
729 | register c = firstc; | |
730 | register char *out; | |
731 | int len; | |
732 | char line[EOFMARKLEN + 1]; | |
733 | struct nodelist *bqlist; | |
734 | int quotef; | |
735 | int dblquote; | |
736 | int varnest; | |
737 | int oldstyle; | |
738 | ||
739 | startlinno = plinno; | |
740 | dblquote = 0; | |
741 | if (syntax == DQSYNTAX) | |
742 | dblquote = 1; | |
743 | quotef = 0; | |
744 | bqlist = NULL; | |
745 | varnest = 0; | |
746 | STARTSTACKSTR(out); | |
747 | loop: { /* for each line, until end of word */ | |
748 | #if ATTY | |
749 | if (c == '\034' && doprompt | |
750 | && attyset() && ! equal(termval(), "emacs")) { | |
751 | attyline(); | |
752 | if (syntax == BASESYNTAX) | |
753 | return readtoken(); | |
754 | c = pgetc(); | |
755 | goto loop; | |
756 | } | |
757 | #endif | |
758 | CHECKEND(); /* set c to PEOF if at end of here document */ | |
759 | for (;;) { /* until end of line or end of word */ | |
760 | CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ | |
761 | switch(syntax[c]) { | |
762 | case CNL: /* '\n' */ | |
763 | if (syntax == BASESYNTAX) | |
764 | goto endword; /* exit outer loop */ | |
765 | USTPUTC(c, out); | |
766 | plinno++; | |
767 | if (doprompt) { | |
768 | putprompt(ps2val()); | |
769 | } | |
770 | c = pgetc(); | |
771 | goto loop; /* continue outer loop */ | |
772 | case CWORD: | |
773 | USTPUTC(c, out); | |
774 | break; | |
775 | case CCTL: | |
776 | if (eofmark == NULL || dblquote) | |
777 | USTPUTC(CTLESC, out); | |
778 | USTPUTC(c, out); | |
779 | break; | |
780 | case CBACK: /* backslash */ | |
781 | c = pgetc(); | |
782 | if (c == PEOF) { | |
783 | USTPUTC('\\', out); | |
784 | pungetc(); | |
785 | } else if (c == '\n') { | |
786 | if (doprompt) | |
787 | putprompt(ps2val()); | |
788 | } else { | |
789 | if (dblquote && c != '\\' && c != '`' && c != '$' | |
790 | && (c != '"' || eofmark != NULL)) | |
791 | USTPUTC('\\', out); | |
792 | if (SQSYNTAX[c] == CCTL) | |
793 | USTPUTC(CTLESC, out); | |
794 | USTPUTC(c, out); | |
795 | quotef++; | |
796 | } | |
797 | break; | |
798 | case CSQUOTE: | |
799 | syntax = SQSYNTAX; | |
800 | break; | |
801 | case CDQUOTE: | |
802 | syntax = DQSYNTAX; | |
803 | dblquote = 1; | |
804 | break; | |
805 | case CENDQUOTE: | |
806 | if (eofmark) { | |
807 | USTPUTC(c, out); | |
808 | } else { | |
809 | syntax = BASESYNTAX; | |
810 | quotef++; | |
811 | dblquote = 0; | |
812 | } | |
813 | break; | |
814 | case CVAR: /* '$' */ | |
815 | PARSESUB(); /* parse substitution */ | |
816 | break; | |
817 | case CENDVAR: /* '}' */ | |
818 | if (varnest > 0) { | |
819 | varnest--; | |
820 | USTPUTC(CTLENDVAR, out); | |
821 | } else { | |
822 | USTPUTC(c, out); | |
823 | } | |
824 | break; | |
825 | case CBQUOTE: /* '`' */ | |
826 | if (parsebackquote && syntax == BASESYNTAX) { | |
827 | if (out == stackblock()) | |
828 | return lasttoken = TENDBQUOTE; | |
829 | else | |
830 | goto endword; /* exit outer loop */ | |
831 | } | |
832 | PARSEBACKQOLD(); | |
833 | break; | |
834 | case CEOF: | |
835 | goto endword; /* exit outer loop */ | |
836 | default: | |
837 | if (varnest == 0) | |
838 | goto endword; /* exit outer loop */ | |
839 | USTPUTC(c, out); | |
840 | } | |
841 | c = pgetc_macro(); | |
842 | } | |
843 | } | |
844 | endword: | |
845 | if (syntax != BASESYNTAX && eofmark == NULL) | |
846 | synerror("Unterminated quoted string"); | |
847 | if (varnest != 0) { | |
848 | startlinno = plinno; | |
849 | synerror("Missing '}'"); | |
850 | } | |
851 | USTPUTC('\0', out); | |
852 | len = out - stackblock(); | |
853 | out = stackblock(); | |
854 | if (eofmark == NULL) { | |
855 | if ((c == '>' || c == '<') | |
856 | && quotef == 0 | |
857 | && len <= 2 | |
858 | && (*out == '\0' || is_digit(*out))) { | |
859 | PARSEREDIR(); | |
860 | return lasttoken = TREDIR; | |
861 | } else { | |
862 | pungetc(); | |
863 | } | |
864 | } | |
865 | quoteflag = quotef; | |
866 | backquotelist = bqlist; | |
867 | grabstackblock(len); | |
868 | wordtext = out; | |
869 | return lasttoken = TWORD; | |
870 | /* end of readtoken routine */ | |
871 | ||
872 | ||
873 | ||
874 | /* | |
875 | * Check to see whether we are at the end of the here document. When this | |
876 | * is called, c is set to the first character of the next input line. If | |
877 | * we are at the end of the here document, this routine sets the c to PEOF. | |
878 | */ | |
879 | ||
880 | checkend: { | |
881 | if (eofmark) { | |
882 | if (striptabs) { | |
883 | while (c == '\t') | |
884 | c = pgetc(); | |
885 | } | |
886 | if (c == *eofmark) { | |
887 | if (pfgets(line, sizeof line) != NULL) { | |
888 | register char *p, *q; | |
889 | ||
890 | p = line; | |
891 | for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); | |
892 | if (*p == '\n' && *q == '\0') { | |
893 | c = PEOF; | |
894 | plinno++; | |
895 | needprompt = doprompt; | |
896 | } else { | |
897 | ppushback(line, strlen(line)); | |
898 | } | |
899 | } | |
900 | } | |
901 | } | |
902 | goto checkend_return; | |
903 | } | |
904 | ||
905 | ||
906 | /* | |
907 | * Parse a redirection operator. The variable "out" points to a string | |
908 | * specifying the fd to be redirected. The variable "c" contains the | |
909 | * first character of the redirection operator. | |
910 | */ | |
911 | ||
912 | parseredir: { | |
913 | char fd = *out; | |
914 | union node *np; | |
915 | ||
916 | np = (union node *)stalloc(sizeof (struct nfile)); | |
917 | if (c == '>') { | |
918 | np->nfile.fd = 1; | |
919 | c = pgetc(); | |
920 | if (c == '>') | |
921 | np->type = NAPPEND; | |
922 | else if (c == '&') | |
923 | np->type = NTOFD; | |
924 | else { | |
925 | np->type = NTO; | |
926 | pungetc(); | |
927 | } | |
928 | } else { /* c == '<' */ | |
929 | np->nfile.fd = 0; | |
930 | c = pgetc(); | |
931 | if (c == '<') { | |
932 | if (sizeof (struct nfile) != sizeof (struct nhere)) { | |
933 | np = (union node *)stalloc(sizeof (struct nhere)); | |
934 | np->nfile.fd = 0; | |
935 | } | |
936 | np->type = NHERE; | |
937 | heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); | |
938 | heredoc->here = np; | |
939 | if ((c = pgetc()) == '-') { | |
940 | heredoc->striptabs = 1; | |
941 | } else { | |
942 | heredoc->striptabs = 0; | |
943 | pungetc(); | |
944 | } | |
945 | } else if (c == '&') | |
946 | np->type = NFROMFD; | |
947 | else { | |
948 | np->type = NFROM; | |
949 | pungetc(); | |
950 | } | |
951 | } | |
952 | if (fd != '\0') | |
953 | np->nfile.fd = digit_val(fd); | |
954 | redirnode = np; | |
955 | goto parseredir_return; | |
956 | } | |
957 | ||
958 | ||
959 | /* | |
960 | * Parse a substitution. At this point, we have read the dollar sign | |
961 | * and nothing else. | |
962 | */ | |
963 | ||
964 | parsesub: { | |
965 | int subtype; | |
966 | int typeloc; | |
967 | int flags; | |
968 | char *p; | |
969 | #ifndef GDB_HACK | |
970 | static const char types[] = "}-+?="; | |
971 | #endif | |
972 | ||
973 | c = pgetc(); | |
ddba57cd | 974 | if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) { |
d787dbb5 KB |
975 | USTPUTC('$', out); |
976 | pungetc(); | |
977 | } else if (c == '(') { /* $(command) */ | |
978 | PARSEBACKQNEW(); | |
979 | } else { | |
980 | USTPUTC(CTLVAR, out); | |
981 | typeloc = out - stackblock(); | |
982 | USTPUTC(VSNORMAL, out); | |
983 | subtype = VSNORMAL; | |
984 | if (c == '{') { | |
985 | c = pgetc(); | |
986 | subtype = 0; | |
987 | } | |
988 | if (is_name(c)) { | |
989 | do { | |
990 | STPUTC(c, out); | |
991 | c = pgetc(); | |
992 | } while (is_in_name(c)); | |
993 | } else { | |
994 | if (! is_special(c)) | |
ddba57cd | 995 | badsub: synerror("Bad substitution"); |
d787dbb5 KB |
996 | USTPUTC(c, out); |
997 | c = pgetc(); | |
998 | } | |
999 | STPUTC('=', out); | |
1000 | flags = 0; | |
1001 | if (subtype == 0) { | |
1002 | if (c == ':') { | |
1003 | flags = VSNUL; | |
1004 | c = pgetc(); | |
1005 | } | |
1006 | p = strchr(types, c); | |
1007 | if (p == NULL) | |
1008 | goto badsub; | |
1009 | subtype = p - types + VSNORMAL; | |
1010 | } else { | |
1011 | pungetc(); | |
1012 | } | |
1013 | if (dblquote) | |
1014 | flags |= VSQUOTE; | |
1015 | *(stackblock() + typeloc) = subtype | flags; | |
1016 | if (subtype != VSNORMAL) | |
1017 | varnest++; | |
1018 | } | |
1019 | goto parsesub_return; | |
1020 | } | |
1021 | ||
1022 | ||
1023 | /* | |
1024 | * Called to parse command substitutions. Newstyle is set if the command | |
1025 | * is enclosed inside $(...); nlpp is a pointer to the head of the linked | |
1026 | * list of commands (passed by reference), and savelen is the number of | |
1027 | * characters on the top of the stack which must be preserved. | |
1028 | */ | |
1029 | ||
1030 | parsebackq: { | |
1031 | struct nodelist **nlpp; | |
1032 | int savepbq; | |
1033 | union node *n; | |
1034 | char *volatile str; | |
1035 | struct jmploc jmploc; | |
1036 | struct jmploc *volatile savehandler; | |
1037 | int savelen; | |
1038 | int t; | |
1039 | ||
1040 | savepbq = parsebackquote; | |
1041 | if (setjmp(jmploc.loc)) { | |
1042 | if (str) | |
1043 | ckfree(str); | |
1044 | parsebackquote = 0; | |
1045 | handler = savehandler; | |
1046 | longjmp(handler, 1); | |
1047 | } | |
1048 | INTOFF; | |
1049 | str = NULL; | |
1050 | savelen = out - stackblock(); | |
1051 | if (savelen > 0) { | |
1052 | str = ckmalloc(savelen); | |
1053 | bcopy(stackblock(), str, savelen); | |
1054 | } | |
1055 | savehandler = handler; | |
1056 | handler = &jmploc; | |
1057 | INTON; | |
1058 | nlpp = &bqlist; | |
1059 | while (*nlpp) | |
1060 | nlpp = &(*nlpp)->next; | |
1061 | *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); | |
1062 | (*nlpp)->next = NULL; | |
1063 | parsebackquote = oldstyle; | |
1064 | n = list(0); | |
1065 | t = oldstyle? TENDBQUOTE : TRP; | |
1066 | if (readtoken() != t) | |
1067 | synexpect(t); | |
1068 | (*nlpp)->n = n; | |
1069 | while (stackblocksize() <= savelen) | |
1070 | growstackblock(); | |
1071 | STARTSTACKSTR(out); | |
1072 | if (str) { | |
1073 | bcopy(str, out, savelen); | |
1074 | STADJUST(savelen, out); | |
1075 | INTOFF; | |
1076 | ckfree(str); | |
1077 | str = NULL; | |
1078 | INTON; | |
1079 | } | |
1080 | parsebackquote = savepbq; | |
1081 | handler = savehandler; | |
1082 | USTPUTC(CTLBACKQ + dblquote, out); | |
1083 | if (oldstyle) | |
1084 | goto parsebackq_oldreturn; | |
1085 | else | |
1086 | goto parsebackq_newreturn; | |
1087 | } | |
1088 | ||
1089 | } /* end of readtoken */ | |
1090 | ||
1091 | ||
1092 | ||
1093 | #ifdef mkinit | |
1094 | RESET { | |
1095 | tokpushback = 0; | |
1096 | } | |
1097 | #endif | |
1098 | ||
1099 | ||
1100 | #if ATTY | |
1101 | /* | |
1102 | * Called to process a command generated by atty. We execute the line, | |
1103 | * and catch any errors that occur so they don't propagate outside of | |
1104 | * this routine. | |
1105 | */ | |
1106 | ||
1107 | STATIC void | |
1108 | attyline() { | |
1109 | char line[256]; | |
1110 | struct stackmark smark; | |
1111 | struct jmploc jmploc; | |
1112 | struct jmploc *volatile savehandler; | |
1113 | ||
1114 | if (pfgets(line, sizeof line) == NULL) | |
1115 | return; /* "can't happen" */ | |
1116 | if (setjmp(jmploc.loc)) { | |
1117 | if (exception == EXERROR) | |
1118 | out2str("\033]D\n"); | |
1119 | handler = savehandler; | |
1120 | longjmp(handler, 1); | |
1121 | } | |
1122 | savehandler = handler; | |
1123 | handler = &jmploc; | |
1124 | setstackmark(&smark); | |
1125 | evalstring(line); | |
1126 | popstackmark(&smark); | |
1127 | handler = savehandler; | |
1128 | doprompt = 1; | |
1129 | } | |
1130 | ||
1131 | ||
1132 | /* | |
1133 | * Output a prompt for atty. We output the prompt as part of the | |
1134 | * appropriate escape sequence. | |
1135 | */ | |
1136 | ||
1137 | STATIC void | |
1138 | putprompt(s) | |
1139 | char *s; | |
1140 | { | |
1141 | register char *p; | |
1142 | ||
1143 | if (attyset() && ! equal(termval(), "emacs")) { | |
1144 | if (strchr(s, '\7')) | |
1145 | out2c('\7'); | |
1146 | out2str("\033]P1;"); | |
1147 | for (p = s ; *p ; p++) { | |
1148 | if ((unsigned)(*p - ' ') <= '~' - ' ') | |
1149 | out2c(*p); | |
1150 | } | |
1151 | out2c('\n'); | |
1152 | } else { | |
1153 | out2str(s); | |
1154 | } | |
1155 | } | |
1156 | #endif | |
1157 | ||
1158 | ||
1159 | ||
1160 | /* | |
1161 | * Returns true if the text contains nothing to expand (no dollar signs | |
1162 | * or backquotes). | |
1163 | */ | |
1164 | ||
1165 | STATIC int | |
1166 | noexpand(text) | |
1167 | char *text; | |
1168 | { | |
1169 | register char *p; | |
1170 | register char c; | |
1171 | ||
1172 | p = text; | |
1173 | while ((c = *p++) != '\0') { | |
1174 | if (c == CTLESC) | |
1175 | p++; | |
1176 | else if (BASESYNTAX[c] == CCTL) | |
1177 | return 0; | |
1178 | } | |
1179 | return 1; | |
1180 | } | |
1181 | ||
1182 | ||
1183 | /* | |
1184 | * Return true if the argument is a legal variable name (a letter or | |
1185 | * underscore followed by zero or more letters, underscores, and digits). | |
1186 | */ | |
1187 | ||
1188 | int | |
1189 | goodname(name) | |
1190 | char *name; | |
1191 | { | |
1192 | register char *p; | |
1193 | ||
1194 | p = name; | |
1195 | if (! is_name(*p)) | |
1196 | return 0; | |
1197 | while (*++p) { | |
1198 | if (! is_in_name(*p)) | |
1199 | return 0; | |
1200 | } | |
1201 | return 1; | |
1202 | } | |
1203 | ||
1204 | ||
1205 | /* | |
1206 | * Called when an unexpected token is read during the parse. The argument | |
1207 | * is the token that is expected, or -1 if more than one type of token can | |
1208 | * occur at this point. | |
1209 | */ | |
1210 | ||
1211 | STATIC void | |
1212 | synexpect(token) { | |
1213 | char msg[64]; | |
1214 | ||
1215 | if (token >= 0) { | |
1216 | fmtstr(msg, 64, "%s unexpected (expecting %s)", | |
1217 | tokname[lasttoken], tokname[token]); | |
1218 | } else { | |
1219 | fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); | |
1220 | } | |
1221 | synerror(msg); | |
1222 | } | |
1223 | ||
1224 | ||
1225 | STATIC void | |
1226 | synerror(msg) | |
1227 | char *msg; | |
1228 | { | |
1229 | if (commandname) | |
1230 | outfmt(&errout, "%s: %d: ", commandname, startlinno); | |
1231 | outfmt(&errout, "Syntax error: %s\n", msg); | |
1232 | error((char *)NULL); | |
1233 | } |