]> git.cameronkatri.com Git - mandoc.git/blob - roff.c
git.ckatri.com / mandoc.git / blob
? search:


35e0f883ef74eaa9ed41ab20c9eed25b2775b85f
[mandoc.git] / roff.c
1 /* $Id: roff.c,v 1.139 2011/05/24 14:00:39 kristaps Exp $ */
2 /*
3 * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 */
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdlib.h>
25 #include <string.h>
26
27 #include "mandoc.h"
28 #include "libroff.h"
29 #include "libmandoc.h"
30
31 #define RSTACK_MAX 128
32
33 enum rofft {
34 ROFF_ad,
35 ROFF_am,
36 ROFF_ami,
37 ROFF_am1,
38 ROFF_de,
39 ROFF_dei,
40 ROFF_de1,
41 ROFF_ds,
42 ROFF_el,
43 ROFF_hy,
44 ROFF_ie,
45 ROFF_if,
46 ROFF_ig,
47 ROFF_it,
48 ROFF_ne,
49 ROFF_nh,
50 ROFF_nr,
51 ROFF_ns,
52 ROFF_ps,
53 ROFF_rm,
54 ROFF_so,
55 ROFF_ta,
56 ROFF_tr,
57 ROFF_TS,
58 ROFF_TE,
59 ROFF_T_,
60 ROFF_EQ,
61 ROFF_EN,
62 ROFF_cblock,
63 ROFF_ccond, /* FIXME: remove this. */
64 ROFF_USERDEF,
65 ROFF_MAX
66 };
67
68 enum roffrule {
69 ROFFRULE_ALLOW,
70 ROFFRULE_DENY
71 };
72
73 struct roffstr {
74 char *name; /* key of symbol */
75 char *string; /* current value */
76 struct roffstr *next; /* next in list */
77 };
78
79 struct roff {
80 struct mparse *parse; /* parse point */
81 struct roffnode *last; /* leaf of stack */
82 enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
83 int rstackpos; /* position in rstack */
84 struct regset *regs; /* read/writable registers */
85 struct roffstr *first_string; /* user-defined strings & macros */
86 const char *current_string; /* value of last called user macro */
87 struct tbl_node *first_tbl; /* first table parsed */
88 struct tbl_node *last_tbl; /* last table parsed */
89 struct tbl_node *tbl; /* current table being parsed */
90 struct eqn_node *last_eqn; /* last equation parsed */
91 struct eqn_node *first_eqn; /* first equation parsed */
92 struct eqn_node *eqn; /* current equation being parsed */
93 };
94
95 struct roffnode {
96 enum rofft tok; /* type of node */
97 struct roffnode *parent; /* up one in stack */
98 int line; /* parse line */
99 int col; /* parse col */
100 char *name; /* node name, e.g. macro name */
101 char *end; /* end-rules: custom token */
102 int endspan; /* end-rules: next-line or infty */
103 enum roffrule rule; /* current evaluation rule */
104 };
105
106 #define ROFF_ARGS struct roff *r, /* parse ctx */ \
107 enum rofft tok, /* tok of macro */ \
108 char **bufp, /* input buffer */ \
109 size_t *szp, /* size of input buffer */ \
110 int ln, /* parse line */ \
111 int ppos, /* original pos in buffer */ \
112 int pos, /* current pos in buffer */ \
113 int *offs /* reset offset of buffer data */
114
115 typedef enum rofferr (*roffproc)(ROFF_ARGS);
116
117 struct roffmac {
118 const char *name; /* macro name */
119 roffproc proc; /* process new macro */
120 roffproc text; /* process as child text of macro */
121 roffproc sub; /* process as child of macro */
122 int flags;
123 #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
124 struct roffmac *next;
125 };
126
127 static enum rofferr roff_block(ROFF_ARGS);
128 static enum rofferr roff_block_text(ROFF_ARGS);
129 static enum rofferr roff_block_sub(ROFF_ARGS);
130 static enum rofferr roff_cblock(ROFF_ARGS);
131 static enum rofferr roff_ccond(ROFF_ARGS);
132 static enum rofferr roff_cond(ROFF_ARGS);
133 static enum rofferr roff_cond_text(ROFF_ARGS);
134 static enum rofferr roff_cond_sub(ROFF_ARGS);
135 static enum rofferr roff_ds(ROFF_ARGS);
136 static enum roffrule roff_evalcond(const char *, int *);
137 static void roff_freestr(struct roff *);
138 static char *roff_getname(struct roff *, char **, int, int);
139 static const char *roff_getstrn(const struct roff *,
140 const char *, size_t);
141 static enum rofferr roff_line_ignore(ROFF_ARGS);
142 static enum rofferr roff_nr(ROFF_ARGS);
143 static int roff_res(struct roff *,
144 char **, size_t *, int);
145 static enum rofferr roff_rm(ROFF_ARGS);
146 static void roff_setstr(struct roff *,
147 const char *, const char *, int);
148 static enum rofferr roff_so(ROFF_ARGS);
149 static enum rofferr roff_TE(ROFF_ARGS);
150 static enum rofferr roff_TS(ROFF_ARGS);
151 static enum rofferr roff_EQ(ROFF_ARGS);
152 static enum rofferr roff_EN(ROFF_ARGS);
153 static enum rofferr roff_T_(ROFF_ARGS);
154 static enum rofferr roff_userdef(ROFF_ARGS);
155
156 /* See roff_hash_find() */
157
158 #define ASCII_HI 126
159 #define ASCII_LO 33
160 #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
161
162 static struct roffmac *hash[HASHWIDTH];
163
164 static struct roffmac roffs[ROFF_MAX] = {
165 { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
166 { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
167 { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
168 { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
169 { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
170 { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
171 { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
172 { "ds", roff_ds, NULL, NULL, 0, NULL },
173 { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
174 { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
175 { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
176 { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
177 { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
178 { "it", roff_line_ignore, NULL, NULL, 0, NULL },
179 { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
180 { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
181 { "nr", roff_nr, NULL, NULL, 0, NULL },
182 { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
183 { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
184 { "rm", roff_rm, NULL, NULL, 0, NULL },
185 { "so", roff_so, NULL, NULL, 0, NULL },
186 { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
187 { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
188 { "TS", roff_TS, NULL, NULL, 0, NULL },
189 { "TE", roff_TE, NULL, NULL, 0, NULL },
190 { "T&", roff_T_, NULL, NULL, 0, NULL },
191 { "EQ", roff_EQ, NULL, NULL, 0, NULL },
192 { "EN", roff_EN, NULL, NULL, 0, NULL },
193 { ".", roff_cblock, NULL, NULL, 0, NULL },
194 { "\\}", roff_ccond, NULL, NULL, 0, NULL },
195 { NULL, roff_userdef, NULL, NULL, 0, NULL },
196 };
197
198 static void roff_free1(struct roff *);
199 static enum rofft roff_hash_find(const char *, size_t);
200 static void roff_hash_init(void);
201 static void roffnode_cleanscope(struct roff *);
202 static void roffnode_push(struct roff *, enum rofft,
203 const char *, int, int);
204 static void roffnode_pop(struct roff *);
205 static enum rofft roff_parse(struct roff *, const char *, int *);
206
207 /* See roff_hash_find() */
208 #define ROFF_HASH(p) (p[0] - ASCII_LO)
209
210 static void
211 roff_hash_init(void)
212 {
213 struct roffmac *n;
214 int buc, i;
215
216 for (i = 0; i < (int)ROFF_USERDEF; i++) {
217 assert(roffs[i].name[0] >= ASCII_LO);
218 assert(roffs[i].name[0] <= ASCII_HI);
219
220 buc = ROFF_HASH(roffs[i].name);
221
222 if (NULL != (n = hash[buc])) {
223 for ( ; n->next; n = n->next)
224 /* Do nothing. */ ;
225 n->next = &roffs[i];
226 } else
227 hash[buc] = &roffs[i];
228 }
229 }
230
231
232 /*
233 * Look up a roff token by its name. Returns ROFF_MAX if no macro by
234 * the nil-terminated string name could be found.
235 */
236 static enum rofft
237 roff_hash_find(const char *p, size_t s)
238 {
239 int buc;
240 struct roffmac *n;
241
242 /*
243 * libroff has an extremely simple hashtable, for the time
244 * being, which simply keys on the first character, which must
245 * be printable, then walks a chain. It works well enough until
246 * optimised.
247 */
248
249 if (p[0] < ASCII_LO || p[0] > ASCII_HI)
250 return(ROFF_MAX);
251
252 buc = ROFF_HASH(p);
253
254 if (NULL == (n = hash[buc]))
255 return(ROFF_MAX);
256 for ( ; n; n = n->next)
257 if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
258 return((enum rofft)(n - roffs));
259
260 return(ROFF_MAX);
261 }
262
263
264 /*
265 * Pop the current node off of the stack of roff instructions currently
266 * pending.
267 */
268 static void
269 roffnode_pop(struct roff *r)
270 {
271 struct roffnode *p;
272
273 assert(r->last);
274 p = r->last;
275
276 r->last = r->last->parent;
277 free(p->name);
278 free(p->end);
279 free(p);
280 }
281
282
283 /*
284 * Push a roff node onto the instruction stack. This must later be
285 * removed with roffnode_pop().
286 */
287 static void
288 roffnode_push(struct roff *r, enum rofft tok, const char *name,
289 int line, int col)
290 {
291 struct roffnode *p;
292
293 p = mandoc_calloc(1, sizeof(struct roffnode));
294 p->tok = tok;
295 if (name)
296 p->name = mandoc_strdup(name);
297 p->parent = r->last;
298 p->line = line;
299 p->col = col;
300 p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
301
302 r->last = p;
303 }
304
305
306 static void
307 roff_free1(struct roff *r)
308 {
309 struct tbl_node *t;
310 struct eqn_node *e;
311
312 while (NULL != (t = r->first_tbl)) {
313 r->first_tbl = t->next;
314 tbl_free(t);
315 }
316
317 r->first_tbl = r->last_tbl = r->tbl = NULL;
318
319 while (NULL != (e = r->first_eqn)) {
320 r->first_eqn = e->next;
321 eqn_free(e);
322 }
323
324 r->first_eqn = r->last_eqn = r->eqn = NULL;
325
326 while (r->last)
327 roffnode_pop(r);
328
329 roff_freestr(r);
330 }
331
332
333 void
334 roff_reset(struct roff *r)
335 {
336
337 roff_free1(r);
338 }
339
340
341 void
342 roff_free(struct roff *r)
343 {
344
345 roff_free1(r);
346 free(r);
347 }
348
349
350 struct roff *
351 roff_alloc(struct regset *regs, struct mparse *parse)
352 {
353 struct roff *r;
354
355 r = mandoc_calloc(1, sizeof(struct roff));
356 r->regs = regs;
357 r->parse = parse;
358 r->rstackpos = -1;
359
360 roff_hash_init();
361 return(r);
362 }
363
364
365 /*
366 * Pre-filter each and every line for reserved words (one beginning with
367 * `\*', e.g., `\*(ab'). These must be handled before the actual line
368 * is processed.
369 */
370 static int
371 roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
372 {
373 const char *stesc; /* start of an escape sequence ('\\') */
374 const char *stnam; /* start of the name, after "[(*" */
375 const char *cp; /* end of the name, e.g. before ']' */
376 const char *res; /* the string to be substituted */
377 int i, maxl;
378 size_t nsz;
379 char *n;
380
381 /* Search for a leading backslash and save a pointer to it. */
382
383 cp = *bufp + pos;
384 while (NULL != (cp = strchr(cp, '\\'))) {
385 stesc = cp++;
386
387 /*
388 * The second character must be an asterisk.
389 * If it isn't, skip it anyway: It is escaped,
390 * so it can't start another escape sequence.
391 */
392
393 if ('\0' == *cp)
394 return(1);
395 if ('*' != *cp++)
396 continue;
397
398 /*
399 * The third character decides the length
400 * of the name of the string.
401 * Save a pointer to the name.
402 */
403
404 switch (*cp) {
405 case ('\0'):
406 return(1);
407 case ('('):
408 cp++;
409 maxl = 2;
410 break;
411 case ('['):
412 cp++;
413 maxl = 0;
414 break;
415 default:
416 maxl = 1;
417 break;
418 }
419 stnam = cp;
420
421 /* Advance to the end of the name. */
422
423 for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
424 if ('\0' == *cp)
425 return(1); /* Error. */
426 if (0 == maxl && ']' == *cp)
427 break;
428 }
429
430 /*
431 * Retrieve the replacement string; if it is
432 * undefined, resume searching for escapes.
433 */
434
435 res = roff_getstrn(r, stnam, (size_t)i);
436
437 if (NULL == res) {
438 cp -= maxl ? 1 : 0;
439 continue;
440 }
441
442 /* Replace the escape sequence by the string. */
443
444 nsz = *szp + strlen(res) + 1;
445 n = mandoc_malloc(nsz);
446
447 strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
448 strlcat(n, res, nsz);
449 strlcat(n, cp + (maxl ? 0 : 1), nsz);
450
451 free(*bufp);
452
453 *bufp = n;
454 *szp = nsz;
455 return(0);
456 }
457
458 return(1);
459 }
460
461
462 enum rofferr
463 roff_parseln(struct roff *r, int ln, char **bufp,
464 size_t *szp, int pos, int *offs)
465 {
466 enum rofft t;
467 enum rofferr e;
468 int ppos, ctl;
469
470 /*
471 * Run the reserved-word filter only if we have some reserved
472 * words to fill in.
473 */
474
475 if (r->first_string && ! roff_res(r, bufp, szp, pos))
476 return(ROFF_REPARSE);
477
478 ppos = pos;
479 ctl = mandoc_getcontrol(*bufp, &pos);
480
481 /*
482 * First, if a scope is open and we're not a macro, pass the
483 * text through the macro's filter. If a scope isn't open and
484 * we're not a macro, just let it through.
485 * Finally, if there's an equation scope open, divert it into it
486 * no matter our state.
487 */
488
489 if (r->last && ! ctl) {
490 t = r->last->tok;
491 assert(roffs[t].text);
492 e = (*roffs[t].text)
493 (r, t, bufp, szp, ln, pos, pos, offs);
494 assert(ROFF_IGN == e || ROFF_CONT == e);
495 if (ROFF_CONT != e)
496 return(e);
497 if (r->eqn)
498 return(eqn_read(&r->eqn, ln, *bufp, pos));
499 if (r->tbl)
500 return(tbl_read(r->tbl, ln, *bufp, pos));
501 return(ROFF_CONT);
502 } else if ( ! ctl) {
503 if (r->eqn)
504 return(eqn_read(&r->eqn, ln, *bufp, pos));
505 if (r->tbl)
506 return(tbl_read(r->tbl, ln, *bufp, pos));
507 return(ROFF_CONT);
508 } else if (r->eqn)
509 return(eqn_read(&r->eqn, ln, *bufp, ppos));
510
511 /*
512 * If a scope is open, go to the child handler for that macro,
513 * as it may want to preprocess before doing anything with it.
514 * Don't do so if an equation is open.
515 */
516
517 if (r->last) {
518 t = r->last->tok;
519 assert(roffs[t].sub);
520 return((*roffs[t].sub)
521 (r, t, bufp, szp,
522 ln, ppos, pos, offs));
523 }
524
525 /*
526 * Lastly, as we've no scope open, try to look up and execute
527 * the new macro. If no macro is found, simply return and let
528 * the compilers handle it.
529 */
530
531 if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
532 return(ROFF_CONT);
533
534 assert(roffs[t].proc);
535 return((*roffs[t].proc)
536 (r, t, bufp, szp,
537 ln, ppos, pos, offs));
538 }
539
540
541 void
542 roff_endparse(struct roff *r)
543 {
544
545 if (r->last)
546 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
547 r->last->line, r->last->col, NULL);
548
549 if (r->eqn) {
550 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
551 r->eqn->eqn.line, r->eqn->eqn.pos, NULL);
552 eqn_end(r->eqn);
553 r->eqn = NULL;
554 }
555
556 if (r->tbl) {
557 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
558 r->tbl->line, r->tbl->pos, NULL);
559 tbl_end(r->tbl);
560 r->tbl = NULL;
561 }
562 }
563
564 /*
565 * Parse a roff node's type from the input buffer. This must be in the
566 * form of ".foo xxx" in the usual way.
567 */
568 static enum rofft
569 roff_parse(struct roff *r, const char *buf, int *pos)
570 {
571 const char *mac;
572 size_t maclen;
573 enum rofft t;
574
575 if ('\0' == buf[*pos] || '"' == buf[*pos])
576 return(ROFF_MAX);
577
578 mac = buf + *pos;
579 maclen = strcspn(mac, " \\\t\0");
580
581 t = (r->current_string = roff_getstrn(r, mac, maclen))
582 ? ROFF_USERDEF : roff_hash_find(mac, maclen);
583
584 *pos += (int)maclen;
585
586 while (buf[*pos] && ' ' == buf[*pos])
587 (*pos)++;
588
589 return(t);
590 }
591
592 /* ARGSUSED */
593 static enum rofferr
594 roff_cblock(ROFF_ARGS)
595 {
596
597 /*
598 * A block-close `..' should only be invoked as a child of an
599 * ignore macro, otherwise raise a warning and just ignore it.
600 */
601
602 if (NULL == r->last) {
603 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
604 return(ROFF_IGN);
605 }
606
607 switch (r->last->tok) {
608 case (ROFF_am):
609 /* FALLTHROUGH */
610 case (ROFF_ami):
611 /* FALLTHROUGH */
612 case (ROFF_am1):
613 /* FALLTHROUGH */
614 case (ROFF_de):
615 /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
616 /* FALLTHROUGH */
617 case (ROFF_dei):
618 /* FALLTHROUGH */
619 case (ROFF_ig):
620 break;
621 default:
622 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
623 return(ROFF_IGN);
624 }
625
626 if ((*bufp)[pos])
627 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
628
629 roffnode_pop(r);
630 roffnode_cleanscope(r);
631 return(ROFF_IGN);
632
633 }
634
635
636 static void
637 roffnode_cleanscope(struct roff *r)
638 {
639
640 while (r->last) {
641 if (--r->last->endspan < 0)
642 break;
643 roffnode_pop(r);
644 }
645 }
646
647
648 /* ARGSUSED */
649 static enum rofferr
650 roff_ccond(ROFF_ARGS)
651 {
652
653 if (NULL == r->last) {
654 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
655 return(ROFF_IGN);
656 }
657
658 switch (r->last->tok) {
659 case (ROFF_el):
660 /* FALLTHROUGH */
661 case (ROFF_ie):
662 /* FALLTHROUGH */
663 case (ROFF_if):
664 break;
665 default:
666 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
667 return(ROFF_IGN);
668 }
669
670 if (r->last->endspan > -1) {
671 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
672 return(ROFF_IGN);
673 }
674
675 if ((*bufp)[pos])
676 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
677
678 roffnode_pop(r);
679 roffnode_cleanscope(r);
680 return(ROFF_IGN);
681 }
682
683
684 /* ARGSUSED */
685 static enum rofferr
686 roff_block(ROFF_ARGS)
687 {
688 int sv;
689 size_t sz;
690 char *name;
691
692 name = NULL;
693
694 if (ROFF_ig != tok) {
695 if ('\0' == (*bufp)[pos]) {
696 mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
697 return(ROFF_IGN);
698 }
699
700 /*
701 * Re-write `de1', since we don't really care about
702 * groff's strange compatibility mode, into `de'.
703 */
704
705 if (ROFF_de1 == tok)
706 tok = ROFF_de;
707 if (ROFF_de == tok)
708 name = *bufp + pos;
709 else
710 mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
711 roffs[tok].name);
712
713 while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
714 pos++;
715
716 while (isspace((unsigned char)(*bufp)[pos]))
717 (*bufp)[pos++] = '\0';
718 }
719
720 roffnode_push(r, tok, name, ln, ppos);
721
722 /*
723 * At the beginning of a `de' macro, clear the existing string
724 * with the same name, if there is one. New content will be
725 * added from roff_block_text() in multiline mode.
726 */
727
728 if (ROFF_de == tok)
729 roff_setstr(r, name, "", 0);
730
731 if ('\0' == (*bufp)[pos])
732 return(ROFF_IGN);
733
734 /* If present, process the custom end-of-line marker. */
735
736 sv = pos;
737 while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
738 pos++;
739
740 /*
741 * Note: groff does NOT like escape characters in the input.
742 * Instead of detecting this, we're just going to let it fly and
743 * to hell with it.
744 */
745
746 assert(pos > sv);
747 sz = (size_t)(pos - sv);
748
749 if (1 == sz && '.' == (*bufp)[sv])
750 return(ROFF_IGN);
751
752 r->last->end = mandoc_malloc(sz + 1);
753
754 memcpy(r->last->end, *bufp + sv, sz);
755 r->last->end[(int)sz] = '\0';
756
757 if ((*bufp)[pos])
758 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
759
760 return(ROFF_IGN);
761 }
762
763
764 /* ARGSUSED */
765 static enum rofferr
766 roff_block_sub(ROFF_ARGS)
767 {
768 enum rofft t;
769 int i, j;
770
771 /*
772 * First check whether a custom macro exists at this level. If
773 * it does, then check against it. This is some of groff's
774 * stranger behaviours. If we encountered a custom end-scope
775 * tag and that tag also happens to be a "real" macro, then we
776 * need to try interpreting it again as a real macro. If it's
777 * not, then return ignore. Else continue.
778 */
779
780 if (r->last->end) {
781 for (i = pos, j = 0; r->last->end[j]; j++, i++)
782 if ((*bufp)[i] != r->last->end[j])
783 break;
784
785 if ('\0' == r->last->end[j] &&
786 ('\0' == (*bufp)[i] ||
787 ' ' == (*bufp)[i] ||
788 '\t' == (*bufp)[i])) {
789 roffnode_pop(r);
790 roffnode_cleanscope(r);
791
792 while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
793 i++;
794
795 pos = i;
796 if (ROFF_MAX != roff_parse(r, *bufp, &pos))
797 return(ROFF_RERUN);
798 return(ROFF_IGN);
799 }
800 }
801
802 /*
803 * If we have no custom end-query or lookup failed, then try
804 * pulling it out of the hashtable.
805 */
806
807 t = roff_parse(r, *bufp, &pos);
808
809 /*
810 * Macros other than block-end are only significant
811 * in `de' blocks; elsewhere, simply throw them away.
812 */
813 if (ROFF_cblock != t) {
814 if (ROFF_de == tok)
815 roff_setstr(r, r->last->name, *bufp + ppos, 1);
816 return(ROFF_IGN);
817 }
818
819 assert(roffs[t].proc);
820 return((*roffs[t].proc)(r, t, bufp, szp,
821 ln, ppos, pos, offs));
822 }
823
824
825 /* ARGSUSED */
826 static enum rofferr
827 roff_block_text(ROFF_ARGS)
828 {
829
830 if (ROFF_de == tok)
831 roff_setstr(r, r->last->name, *bufp + pos, 1);
832
833 return(ROFF_IGN);
834 }
835
836
837 /* ARGSUSED */
838 static enum rofferr
839 roff_cond_sub(ROFF_ARGS)
840 {
841 enum rofft t;
842 enum roffrule rr;
843 char *ep;
844
845 rr = r->last->rule;
846 roffnode_cleanscope(r);
847
848 /*
849 * If the macro is unknown, first check if it contains a closing
850 * delimiter `\}'. If it does, close out our scope and return
851 * the currently-scoped rule (ignore or continue). Else, drop
852 * into the currently-scoped rule.
853 */
854
855 if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
856 /*
857 * Jump through hoops to detect a \}, because it could
858 * be (say) \\}, which is something completely
859 * different.
860 */
861 ep = &(*bufp)[pos];
862 for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
863 ep++;
864 if ('}' != *ep)
865 continue;
866 *--ep = '\0';
867 roff_ccond(r, ROFF_ccond, bufp, szp,
868 ln, pos, pos + 2, offs);
869 break;
870 }
871 return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
872 }
873
874 /*
875 * A denied conditional must evaluate its children if and only
876 * if they're either structurally required (such as loops and
877 * conditionals) or a closing macro.
878 */
879
880 if (ROFFRULE_DENY == rr)
881 if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
882 if (ROFF_ccond != t)
883 return(ROFF_IGN);
884
885 assert(roffs[t].proc);
886 return((*roffs[t].proc)(r, t, bufp, szp,
887 ln, ppos, pos, offs));
888 }
889
890
891 /* ARGSUSED */
892 static enum rofferr
893 roff_cond_text(ROFF_ARGS)
894 {
895 char *ep, *st;
896 enum roffrule rr;
897
898 rr = r->last->rule;
899
900 /*
901 * We display the value of the text if out current evaluation
902 * scope permits us to do so.
903 */
904
905 /* FIXME: use roff_ccond? */
906
907 st = &(*bufp)[pos];
908 if (NULL == (ep = strstr(st, "\\}"))) {
909 roffnode_cleanscope(r);
910 return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
911 }
912
913 if (ep == st || (ep > st && '\\' != *(ep - 1)))
914 roffnode_pop(r);
915
916 roffnode_cleanscope(r);
917 return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
918 }
919
920
921 static enum roffrule
922 roff_evalcond(const char *v, int *pos)
923 {
924
925 switch (v[*pos]) {
926 case ('n'):
927 (*pos)++;
928 return(ROFFRULE_ALLOW);
929 case ('e'):
930 /* FALLTHROUGH */
931 case ('o'):
932 /* FALLTHROUGH */
933 case ('t'):
934 (*pos)++;
935 return(ROFFRULE_DENY);
936 default:
937 break;
938 }
939
940 while (v[*pos] && ' ' != v[*pos])
941 (*pos)++;
942 return(ROFFRULE_DENY);
943 }
944
945 /* ARGSUSED */
946 static enum rofferr
947 roff_line_ignore(ROFF_ARGS)
948 {
949
950 if (ROFF_it == tok)
951 mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
952
953 return(ROFF_IGN);
954 }
955
956 /* ARGSUSED */
957 static enum rofferr
958 roff_cond(ROFF_ARGS)
959 {
960 int sv;
961 enum roffrule rule;
962
963 /*
964 * An `.el' has no conditional body: it will consume the value
965 * of the current rstack entry set in prior `ie' calls or
966 * defaults to DENY.
967 *
968 * If we're not an `el', however, then evaluate the conditional.
969 */
970
971 rule = ROFF_el == tok ?
972 (r->rstackpos < 0 ?
973 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
974 roff_evalcond(*bufp, &pos);
975
976 sv = pos;
977 while (' ' == (*bufp)[pos])
978 pos++;
979
980 /*
981 * Roff is weird. If we have just white-space after the
982 * conditional, it's considered the BODY and we exit without
983 * really doing anything. Warn about this. It's probably
984 * wrong.
985 */
986
987 if ('\0' == (*bufp)[pos] && sv != pos) {
988 mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
989 return(ROFF_IGN);
990 }
991
992 roffnode_push(r, tok, NULL, ln, ppos);
993
994 r->last->rule = rule;
995
996 /*
997 * An if-else will put the NEGATION of the current evaluated
998 * conditional into the stack of rules.
999 */
1000
1001 if (ROFF_ie == tok) {
1002 if (r->rstackpos == RSTACK_MAX - 1) {
1003 mandoc_msg(MANDOCERR_MEM,
1004 r->parse, ln, ppos, NULL);
1005 return(ROFF_ERR);
1006 }
1007 r->rstack[++r->rstackpos] =
1008 ROFFRULE_DENY == r->last->rule ?
1009 ROFFRULE_ALLOW : ROFFRULE_DENY;
1010 }
1011
1012 /* If the parent has false as its rule, then so do we. */
1013
1014 if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1015 r->last->rule = ROFFRULE_DENY;
1016
1017 /*
1018 * Determine scope. If we're invoked with "\{" trailing the
1019 * conditional, then we're in a multiline scope. Else our scope
1020 * expires on the next line.
1021 */
1022
1023 r->last->endspan = 1;
1024
1025 if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1026 r->last->endspan = -1;
1027 pos += 2;
1028 }
1029
1030 /*
1031 * If there are no arguments on the line, the next-line scope is
1032 * assumed.
1033 */
1034
1035 if ('\0' == (*bufp)[pos])
1036 return(ROFF_IGN);
1037
1038 /* Otherwise re-run the roff parser after recalculating. */
1039
1040 *offs = pos;
1041 return(ROFF_RERUN);
1042 }
1043
1044
1045 /* ARGSUSED */
1046 static enum rofferr
1047 roff_ds(ROFF_ARGS)
1048 {
1049 char *name, *string;
1050
1051 /*
1052 * A symbol is named by the first word following the macro
1053 * invocation up to a space. Its value is anything after the
1054 * name's trailing whitespace and optional double-quote. Thus,
1055 *
1056 * [.ds foo "bar " ]
1057 *
1058 * will have `bar " ' as its value.
1059 */
1060
1061 string = *bufp + pos;
1062 name = roff_getname(r, &string, ln, pos);
1063 if ('\0' == *name)
1064 return(ROFF_IGN);
1065
1066 /* Read past initial double-quote. */
1067 if ('"' == *string)
1068 string++;
1069
1070 /* The rest is the value. */
1071 roff_setstr(r, name, string, 0);
1072 return(ROFF_IGN);
1073 }
1074
1075
1076 /* ARGSUSED */
1077 static enum rofferr
1078 roff_nr(ROFF_ARGS)
1079 {
1080 const char *key;
1081 char *val;
1082 int iv;
1083 struct reg *rg;
1084
1085 val = *bufp + pos;
1086 key = roff_getname(r, &val, ln, pos);
1087 rg = r->regs->regs;
1088
1089 if (0 == strcmp(key, "nS")) {
1090 rg[(int)REG_nS].set = 1;
1091 if ((iv = mandoc_strntou(val, strlen(val), 10)) >= 0)
1092 rg[REG_nS].v.u = (unsigned)iv;
1093 else
1094 rg[(int)REG_nS].v.u = 0u;
1095 }
1096
1097 return(ROFF_IGN);
1098 }
1099
1100 /* ARGSUSED */
1101 static enum rofferr
1102 roff_rm(ROFF_ARGS)
1103 {
1104 const char *name;
1105 char *cp;
1106
1107 cp = *bufp + pos;
1108 while ('\0' != *cp) {
1109 name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1110 if ('\0' != *name)
1111 roff_setstr(r, name, NULL, 0);
1112 }
1113 return(ROFF_IGN);
1114 }
1115
1116 /* ARGSUSED */
1117 static enum rofferr
1118 roff_TE(ROFF_ARGS)
1119 {
1120
1121 if (NULL == r->tbl)
1122 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1123 else
1124 tbl_end(r->tbl);
1125
1126 r->tbl = NULL;
1127 return(ROFF_IGN);
1128 }
1129
1130 /* ARGSUSED */
1131 static enum rofferr
1132 roff_T_(ROFF_ARGS)
1133 {
1134
1135 if (NULL == r->tbl)
1136 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1137 else
1138 tbl_restart(ppos, ln, r->tbl);
1139
1140 return(ROFF_IGN);
1141 }
1142
1143 /* ARGSUSED */
1144 static enum rofferr
1145 roff_EQ(ROFF_ARGS)
1146 {
1147 struct eqn_node *e;
1148
1149 assert(NULL == r->eqn);
1150 e = eqn_alloc(ppos, ln);
1151
1152 if (r->last_eqn)
1153 r->last_eqn->next = e;
1154 else
1155 r->first_eqn = r->last_eqn = e;
1156
1157 r->eqn = r->last_eqn = e;
1158 return(ROFF_IGN);
1159 }
1160
1161 /* ARGSUSED */
1162 static enum rofferr
1163 roff_EN(ROFF_ARGS)
1164 {
1165
1166 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1167 return(ROFF_IGN);
1168 }
1169
1170 /* ARGSUSED */
1171 static enum rofferr
1172 roff_TS(ROFF_ARGS)
1173 {
1174 struct tbl_node *t;
1175
1176 if (r->tbl) {
1177 mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1178 tbl_end(r->tbl);
1179 }
1180
1181 t = tbl_alloc(ppos, ln, r->parse);
1182
1183 if (r->last_tbl)
1184 r->last_tbl->next = t;
1185 else
1186 r->first_tbl = r->last_tbl = t;
1187
1188 r->tbl = r->last_tbl = t;
1189 return(ROFF_IGN);
1190 }
1191
1192 /* ARGSUSED */
1193 static enum rofferr
1194 roff_so(ROFF_ARGS)
1195 {
1196 char *name;
1197
1198 mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1199
1200 /*
1201 * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1202 * opening anything that's not in our cwd or anything beneath
1203 * it. Thus, explicitly disallow traversing up the file-system
1204 * or using absolute paths.
1205 */
1206
1207 name = *bufp + pos;
1208 if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1209 mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1210 return(ROFF_ERR);
1211 }
1212
1213 *offs = pos;
1214 return(ROFF_SO);
1215 }
1216
1217 /* ARGSUSED */
1218 static enum rofferr
1219 roff_userdef(ROFF_ARGS)
1220 {
1221 const char *arg[9];
1222 char *cp, *n1, *n2;
1223 int i;
1224
1225 /*
1226 * Collect pointers to macro argument strings
1227 * and null-terminate them.
1228 */
1229 cp = *bufp + pos;
1230 for (i = 0; i < 9; i++)
1231 arg[i] = '\0' == *cp ? "" :
1232 mandoc_getarg(r->parse, &cp, ln, &pos);
1233
1234 /*
1235 * Expand macro arguments.
1236 */
1237 *szp = 0;
1238 n1 = cp = mandoc_strdup(r->current_string);
1239 while (NULL != (cp = strstr(cp, "\\$"))) {
1240 i = cp[2] - '1';
1241 if (0 > i || 8 < i) {
1242 /* Not an argument invocation. */
1243 cp += 2;
1244 continue;
1245 }
1246
1247 *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1248 n2 = mandoc_malloc(*szp);
1249
1250 strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1251 strlcat(n2, arg[i], *szp);
1252 strlcat(n2, cp + 3, *szp);
1253
1254 cp = n2 + (cp - n1);
1255 free(n1);
1256 n1 = n2;
1257 }
1258
1259 /*
1260 * Replace the macro invocation
1261 * by the expanded macro.
1262 */
1263 free(*bufp);
1264 *bufp = n1;
1265 if (0 == *szp)
1266 *szp = strlen(*bufp) + 1;
1267
1268 return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1269 ROFF_REPARSE : ROFF_APPEND);
1270 }
1271
1272 static char *
1273 roff_getname(struct roff *r, char **cpp, int ln, int pos)
1274 {
1275 char *name, *cp;
1276
1277 name = *cpp;
1278 if ('\0' == *name)
1279 return(name);
1280
1281 /* Read until end of name. */
1282 for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1283 if ('\\' != *cp)
1284 continue;
1285 cp++;
1286 if ('\\' == *cp)
1287 continue;
1288 mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1289 *cp = '\0';
1290 name = cp;
1291 }
1292
1293 /* Nil-terminate name. */
1294 if ('\0' != *cp)
1295 *(cp++) = '\0';
1296
1297 /* Read past spaces. */
1298 while (' ' == *cp)
1299 cp++;
1300
1301 *cpp = cp;
1302 return(name);
1303 }
1304
1305 /*
1306 * Store *string into the user-defined string called *name.
1307 * In multiline mode, append to an existing entry and append '\n';
1308 * else replace the existing entry, if there is one.
1309 * To clear an existing entry, call with (*r, *name, NULL, 0).
1310 */
1311 static void
1312 roff_setstr(struct roff *r, const char *name, const char *string,
1313 int multiline)
1314 {
1315 struct roffstr *n;
1316 char *c;
1317 size_t oldch, newch;
1318
1319 /* Search for an existing string with the same name. */
1320 n = r->first_string;
1321 while (n && strcmp(name, n->name))
1322 n = n->next;
1323
1324 if (NULL == n) {
1325 /* Create a new string table entry. */
1326 n = mandoc_malloc(sizeof(struct roffstr));
1327 n->name = mandoc_strdup(name);
1328 n->string = NULL;
1329 n->next = r->first_string;
1330 r->first_string = n;
1331 } else if (0 == multiline) {
1332 /* In multiline mode, append; else replace. */
1333 free(n->string);
1334 n->string = NULL;
1335 }
1336
1337 if (NULL == string)
1338 return;
1339
1340 /*
1341 * One additional byte for the '\n' in multiline mode,
1342 * and one for the terminating '\0'.
1343 */
1344 newch = strlen(string) + (multiline ? 2u : 1u);
1345 if (NULL == n->string) {
1346 n->string = mandoc_malloc(newch);
1347 *n->string = '\0';
1348 oldch = 0;
1349 } else {
1350 oldch = strlen(n->string);
1351 n->string = mandoc_realloc(n->string, oldch + newch);
1352 }
1353
1354 /* Skip existing content in the destination buffer. */
1355 c = n->string + (int)oldch;
1356
1357 /* Append new content to the destination buffer. */
1358 while (*string) {
1359 /*
1360 * Rudimentary roff copy mode:
1361 * Handle escaped backslashes.
1362 */
1363 if ('\\' == *string && '\\' == *(string + 1))
1364 string++;
1365 *c++ = *string++;
1366 }
1367
1368 /* Append terminating bytes. */
1369 if (multiline)
1370 *c++ = '\n';
1371 *c = '\0';
1372 }
1373
1374 static const char *
1375 roff_getstrn(const struct roff *r, const char *name, size_t len)
1376 {
1377 const struct roffstr *n;
1378
1379 n = r->first_string;
1380 while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1381 n = n->next;
1382
1383 return(n ? n->string : NULL);
1384 }
1385
1386 static void
1387 roff_freestr(struct roff *r)
1388 {
1389 struct roffstr *n, *nn;
1390
1391 for (n = r->first_string; n; n = nn) {
1392 free(n->name);
1393 free(n->string);
1394 nn = n->next;
1395 free(n);
1396 }
1397
1398 r->first_string = NULL;
1399 }
1400
1401 const struct tbl_span *
1402 roff_span(const struct roff *r)
1403 {
1404
1405 return(r->tbl ? tbl_span(r->tbl) : NULL);
1406 }
1407
1408 const struct eqn *
1409 roff_eqn(const struct roff *r)
1410 {
1411
1412 return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1413 }
A git conversion mirror of mandoc.bsd.lv