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


a9dbce22a58db061900e7d4a6c9778e45880463d
[mandoc.git] / roff.c
1 /* $Id: roff.c,v 1.141 2011/05/24 21:18:06 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 /* Maximum number of nested if-else conditionals. */
32 #define RSTACK_MAX 128
33
34 enum rofft {
35 ROFF_ad,
36 ROFF_am,
37 ROFF_ami,
38 ROFF_am1,
39 ROFF_de,
40 ROFF_dei,
41 ROFF_de1,
42 ROFF_ds,
43 ROFF_el,
44 ROFF_hy,
45 ROFF_ie,
46 ROFF_if,
47 ROFF_ig,
48 ROFF_it,
49 ROFF_ne,
50 ROFF_nh,
51 ROFF_nr,
52 ROFF_ns,
53 ROFF_ps,
54 ROFF_rm,
55 ROFF_so,
56 ROFF_ta,
57 ROFF_tr,
58 ROFF_TS,
59 ROFF_TE,
60 ROFF_T_,
61 ROFF_EQ,
62 ROFF_EN,
63 ROFF_cblock,
64 ROFF_ccond,
65 ROFF_USERDEF,
66 ROFF_MAX
67 };
68
69 enum roffrule {
70 ROFFRULE_ALLOW,
71 ROFFRULE_DENY
72 };
73
74 struct roffstr {
75 char *name; /* key of symbol */
76 char *string; /* current value */
77 struct roffstr *next; /* next in list */
78 };
79
80 struct roff {
81 struct mparse *parse; /* parse point */
82 struct roffnode *last; /* leaf of stack */
83 enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
84 int rstackpos; /* position in rstack */
85 struct regset *regs; /* read/writable registers */
86 struct roffstr *first_string; /* user-defined strings & macros */
87 const char *current_string; /* value of last called user macro */
88 struct tbl_node *first_tbl; /* first table parsed */
89 struct tbl_node *last_tbl; /* last table parsed */
90 struct tbl_node *tbl; /* current table being parsed */
91 struct eqn_node *last_eqn; /* last equation parsed */
92 struct eqn_node *first_eqn; /* first equation parsed */
93 struct eqn_node *eqn; /* current equation being parsed */
94 };
95
96 struct roffnode {
97 enum rofft tok; /* type of node */
98 struct roffnode *parent; /* up one in stack */
99 int line; /* parse line */
100 int col; /* parse col */
101 char *name; /* node name, e.g. macro name */
102 char *end; /* end-rules: custom token */
103 int endspan; /* end-rules: next-line or infty */
104 enum roffrule rule; /* current evaluation rule */
105 };
106
107 #define ROFF_ARGS struct roff *r, /* parse ctx */ \
108 enum rofft tok, /* tok of macro */ \
109 char **bufp, /* input buffer */ \
110 size_t *szp, /* size of input buffer */ \
111 int ln, /* parse line */ \
112 int ppos, /* original pos in buffer */ \
113 int pos, /* current pos in buffer */ \
114 int *offs /* reset offset of buffer data */
115
116 typedef enum rofferr (*roffproc)(ROFF_ARGS);
117
118 struct roffmac {
119 const char *name; /* macro name */
120 roffproc proc; /* process new macro */
121 roffproc text; /* process as child text of macro */
122 roffproc sub; /* process as child of macro */
123 int flags;
124 #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
125 struct roffmac *next;
126 };
127
128 struct predef {
129 const char *name; /* predefined input name */
130 const char *str; /* replacement symbol */
131 };
132
133 #define PREDEF(__name, __str) \
134 { (__name), (__str) },
135
136 static enum rofferr roff_block(ROFF_ARGS);
137 static enum rofferr roff_block_text(ROFF_ARGS);
138 static enum rofferr roff_block_sub(ROFF_ARGS);
139 static enum rofferr roff_cblock(ROFF_ARGS);
140 static enum rofferr roff_ccond(ROFF_ARGS);
141 static enum rofferr roff_cond(ROFF_ARGS);
142 static enum rofferr roff_cond_text(ROFF_ARGS);
143 static enum rofferr roff_cond_sub(ROFF_ARGS);
144 static enum rofferr roff_ds(ROFF_ARGS);
145 static enum roffrule roff_evalcond(const char *, int *);
146 static void roff_freestr(struct roff *);
147 static char *roff_getname(struct roff *, char **, int, int);
148 static const char *roff_getstrn(const struct roff *,
149 const char *, size_t);
150 static enum rofferr roff_line_ignore(ROFF_ARGS);
151 static enum rofferr roff_nr(ROFF_ARGS);
152 static int roff_res(struct roff *,
153 char **, size_t *, int);
154 static enum rofferr roff_rm(ROFF_ARGS);
155 static void roff_setstr(struct roff *,
156 const char *, const char *, int);
157 static enum rofferr roff_so(ROFF_ARGS);
158 static enum rofferr roff_TE(ROFF_ARGS);
159 static enum rofferr roff_TS(ROFF_ARGS);
160 static enum rofferr roff_EQ(ROFF_ARGS);
161 static enum rofferr roff_EN(ROFF_ARGS);
162 static enum rofferr roff_T_(ROFF_ARGS);
163 static enum rofferr roff_userdef(ROFF_ARGS);
164
165 /* See roff_hash_find() */
166
167 #define ASCII_HI 126
168 #define ASCII_LO 33
169 #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
170
171 static struct roffmac *hash[HASHWIDTH];
172
173 static struct roffmac roffs[ROFF_MAX] = {
174 { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
175 { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
176 { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
177 { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
178 { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
179 { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
180 { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
181 { "ds", roff_ds, NULL, NULL, 0, NULL },
182 { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
183 { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
184 { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
185 { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
186 { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
187 { "it", roff_line_ignore, NULL, NULL, 0, NULL },
188 { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
189 { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
190 { "nr", roff_nr, NULL, NULL, 0, NULL },
191 { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
192 { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
193 { "rm", roff_rm, NULL, NULL, 0, NULL },
194 { "so", roff_so, NULL, NULL, 0, NULL },
195 { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
196 { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
197 { "TS", roff_TS, NULL, NULL, 0, NULL },
198 { "TE", roff_TE, NULL, NULL, 0, NULL },
199 { "T&", roff_T_, NULL, NULL, 0, NULL },
200 { "EQ", roff_EQ, NULL, NULL, 0, NULL },
201 { "EN", roff_EN, NULL, NULL, 0, NULL },
202 { ".", roff_cblock, NULL, NULL, 0, NULL },
203 { "\\}", roff_ccond, NULL, NULL, 0, NULL },
204 { NULL, roff_userdef, NULL, NULL, 0, NULL },
205 };
206
207 /* Array of injected predefined strings. */
208 #define PREDEFS_MAX 38
209 static const struct predef predefs[PREDEFS_MAX] = {
210 #include "predefs.in"
211 };
212
213 static void roff_free1(struct roff *);
214 static enum rofft roff_hash_find(const char *, size_t);
215 static void roff_hash_init(void);
216 static void roffnode_cleanscope(struct roff *);
217 static void roffnode_push(struct roff *, enum rofft,
218 const char *, int, int);
219 static void roffnode_pop(struct roff *);
220 static enum rofft roff_parse(struct roff *, const char *, int *);
221
222 /* See roff_hash_find() */
223 #define ROFF_HASH(p) (p[0] - ASCII_LO)
224
225 static void
226 roff_hash_init(void)
227 {
228 struct roffmac *n;
229 int buc, i;
230
231 for (i = 0; i < (int)ROFF_USERDEF; i++) {
232 assert(roffs[i].name[0] >= ASCII_LO);
233 assert(roffs[i].name[0] <= ASCII_HI);
234
235 buc = ROFF_HASH(roffs[i].name);
236
237 if (NULL != (n = hash[buc])) {
238 for ( ; n->next; n = n->next)
239 /* Do nothing. */ ;
240 n->next = &roffs[i];
241 } else
242 hash[buc] = &roffs[i];
243 }
244 }
245
246 /*
247 * Look up a roff token by its name. Returns ROFF_MAX if no macro by
248 * the nil-terminated string name could be found.
249 */
250 static enum rofft
251 roff_hash_find(const char *p, size_t s)
252 {
253 int buc;
254 struct roffmac *n;
255
256 /*
257 * libroff has an extremely simple hashtable, for the time
258 * being, which simply keys on the first character, which must
259 * be printable, then walks a chain. It works well enough until
260 * optimised.
261 */
262
263 if (p[0] < ASCII_LO || p[0] > ASCII_HI)
264 return(ROFF_MAX);
265
266 buc = ROFF_HASH(p);
267
268 if (NULL == (n = hash[buc]))
269 return(ROFF_MAX);
270 for ( ; n; n = n->next)
271 if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
272 return((enum rofft)(n - roffs));
273
274 return(ROFF_MAX);
275 }
276
277
278 /*
279 * Pop the current node off of the stack of roff instructions currently
280 * pending.
281 */
282 static void
283 roffnode_pop(struct roff *r)
284 {
285 struct roffnode *p;
286
287 assert(r->last);
288 p = r->last;
289
290 r->last = r->last->parent;
291 free(p->name);
292 free(p->end);
293 free(p);
294 }
295
296
297 /*
298 * Push a roff node onto the instruction stack. This must later be
299 * removed with roffnode_pop().
300 */
301 static void
302 roffnode_push(struct roff *r, enum rofft tok, const char *name,
303 int line, int col)
304 {
305 struct roffnode *p;
306
307 p = mandoc_calloc(1, sizeof(struct roffnode));
308 p->tok = tok;
309 if (name)
310 p->name = mandoc_strdup(name);
311 p->parent = r->last;
312 p->line = line;
313 p->col = col;
314 p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
315
316 r->last = p;
317 }
318
319
320 static void
321 roff_free1(struct roff *r)
322 {
323 struct tbl_node *t;
324 struct eqn_node *e;
325
326 while (NULL != (t = r->first_tbl)) {
327 r->first_tbl = t->next;
328 tbl_free(t);
329 }
330
331 r->first_tbl = r->last_tbl = r->tbl = NULL;
332
333 while (NULL != (e = r->first_eqn)) {
334 r->first_eqn = e->next;
335 eqn_free(e);
336 }
337
338 r->first_eqn = r->last_eqn = r->eqn = NULL;
339
340 while (r->last)
341 roffnode_pop(r);
342
343 roff_freestr(r);
344 }
345
346
347 void
348 roff_reset(struct roff *r)
349 {
350
351 roff_free1(r);
352 }
353
354
355 void
356 roff_free(struct roff *r)
357 {
358
359 roff_free1(r);
360 free(r);
361 }
362
363
364 struct roff *
365 roff_alloc(struct regset *regs, struct mparse *parse)
366 {
367 struct roff *r;
368 int i;
369
370 r = mandoc_calloc(1, sizeof(struct roff));
371 r->regs = regs;
372 r->parse = parse;
373 r->rstackpos = -1;
374
375 roff_hash_init();
376
377 for (i = 0; i < PREDEFS_MAX; i++)
378 roff_setstr(r, predefs[i].name, predefs[i].str, 0);
379
380 return(r);
381 }
382
383
384 /*
385 * Pre-filter each and every line for reserved words (one beginning with
386 * `\*', e.g., `\*(ab'). These must be handled before the actual line
387 * is processed.
388 */
389 static int
390 roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
391 {
392 const char *stesc; /* start of an escape sequence ('\\') */
393 const char *stnam; /* start of the name, after "[(*" */
394 const char *cp; /* end of the name, e.g. before ']' */
395 const char *res; /* the string to be substituted */
396 int i, maxl;
397 size_t nsz;
398 char *n;
399
400 /* Search for a leading backslash and save a pointer to it. */
401
402 cp = *bufp + pos;
403 while (NULL != (cp = strchr(cp, '\\'))) {
404 stesc = cp++;
405
406 /*
407 * The second character must be an asterisk.
408 * If it isn't, skip it anyway: It is escaped,
409 * so it can't start another escape sequence.
410 */
411
412 if ('\0' == *cp)
413 return(1);
414 if ('*' != *cp++)
415 continue;
416
417 /*
418 * The third character decides the length
419 * of the name of the string.
420 * Save a pointer to the name.
421 */
422
423 switch (*cp) {
424 case ('\0'):
425 return(1);
426 case ('('):
427 cp++;
428 maxl = 2;
429 break;
430 case ('['):
431 cp++;
432 maxl = 0;
433 break;
434 default:
435 maxl = 1;
436 break;
437 }
438 stnam = cp;
439
440 /* Advance to the end of the name. */
441
442 for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
443 if ('\0' == *cp)
444 return(1); /* Error. */
445 if (0 == maxl && ']' == *cp)
446 break;
447 }
448
449 /*
450 * Retrieve the replacement string; if it is
451 * undefined, resume searching for escapes.
452 */
453
454 res = roff_getstrn(r, stnam, (size_t)i);
455
456 if (NULL == res) {
457 cp -= maxl ? 1 : 0;
458 continue;
459 }
460
461 /* Replace the escape sequence by the string. */
462
463 nsz = *szp + strlen(res) + 1;
464 n = mandoc_malloc(nsz);
465
466 strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
467 strlcat(n, res, nsz);
468 strlcat(n, cp + (maxl ? 0 : 1), nsz);
469
470 free(*bufp);
471
472 *bufp = n;
473 *szp = nsz;
474 return(0);
475 }
476
477 return(1);
478 }
479
480
481 enum rofferr
482 roff_parseln(struct roff *r, int ln, char **bufp,
483 size_t *szp, int pos, int *offs)
484 {
485 enum rofft t;
486 enum rofferr e;
487 int ppos, ctl;
488
489 /*
490 * Run the reserved-word filter only if we have some reserved
491 * words to fill in.
492 */
493
494 if (r->first_string && ! roff_res(r, bufp, szp, pos))
495 return(ROFF_REPARSE);
496
497 ppos = pos;
498 ctl = mandoc_getcontrol(*bufp, &pos);
499
500 /*
501 * First, if a scope is open and we're not a macro, pass the
502 * text through the macro's filter. If a scope isn't open and
503 * we're not a macro, just let it through.
504 * Finally, if there's an equation scope open, divert it into it
505 * no matter our state.
506 */
507
508 if (r->last && ! ctl) {
509 t = r->last->tok;
510 assert(roffs[t].text);
511 e = (*roffs[t].text)
512 (r, t, bufp, szp, ln, pos, pos, offs);
513 assert(ROFF_IGN == e || ROFF_CONT == e);
514 if (ROFF_CONT != e)
515 return(e);
516 if (r->eqn)
517 return(eqn_read(&r->eqn, ln, *bufp, pos));
518 if (r->tbl)
519 return(tbl_read(r->tbl, ln, *bufp, pos));
520 return(ROFF_CONT);
521 } else if ( ! ctl) {
522 if (r->eqn)
523 return(eqn_read(&r->eqn, ln, *bufp, pos));
524 if (r->tbl)
525 return(tbl_read(r->tbl, ln, *bufp, pos));
526 return(ROFF_CONT);
527 } else if (r->eqn)
528 return(eqn_read(&r->eqn, ln, *bufp, ppos));
529
530 /*
531 * If a scope is open, go to the child handler for that macro,
532 * as it may want to preprocess before doing anything with it.
533 * Don't do so if an equation is open.
534 */
535
536 if (r->last) {
537 t = r->last->tok;
538 assert(roffs[t].sub);
539 return((*roffs[t].sub)
540 (r, t, bufp, szp,
541 ln, ppos, pos, offs));
542 }
543
544 /*
545 * Lastly, as we've no scope open, try to look up and execute
546 * the new macro. If no macro is found, simply return and let
547 * the compilers handle it.
548 */
549
550 if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
551 return(ROFF_CONT);
552
553 assert(roffs[t].proc);
554 return((*roffs[t].proc)
555 (r, t, bufp, szp,
556 ln, ppos, pos, offs));
557 }
558
559
560 void
561 roff_endparse(struct roff *r)
562 {
563
564 if (r->last)
565 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
566 r->last->line, r->last->col, NULL);
567
568 if (r->eqn) {
569 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
570 r->eqn->eqn.line, r->eqn->eqn.pos, NULL);
571 eqn_end(r->eqn);
572 r->eqn = NULL;
573 }
574
575 if (r->tbl) {
576 mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
577 r->tbl->line, r->tbl->pos, NULL);
578 tbl_end(r->tbl);
579 r->tbl = NULL;
580 }
581 }
582
583 /*
584 * Parse a roff node's type from the input buffer. This must be in the
585 * form of ".foo xxx" in the usual way.
586 */
587 static enum rofft
588 roff_parse(struct roff *r, const char *buf, int *pos)
589 {
590 const char *mac;
591 size_t maclen;
592 enum rofft t;
593
594 if ('\0' == buf[*pos] || '"' == buf[*pos])
595 return(ROFF_MAX);
596
597 mac = buf + *pos;
598 maclen = strcspn(mac, " \\\t\0");
599
600 t = (r->current_string = roff_getstrn(r, mac, maclen))
601 ? ROFF_USERDEF : roff_hash_find(mac, maclen);
602
603 *pos += (int)maclen;
604
605 while (buf[*pos] && ' ' == buf[*pos])
606 (*pos)++;
607
608 return(t);
609 }
610
611 /* ARGSUSED */
612 static enum rofferr
613 roff_cblock(ROFF_ARGS)
614 {
615
616 /*
617 * A block-close `..' should only be invoked as a child of an
618 * ignore macro, otherwise raise a warning and just ignore it.
619 */
620
621 if (NULL == r->last) {
622 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
623 return(ROFF_IGN);
624 }
625
626 switch (r->last->tok) {
627 case (ROFF_am):
628 /* FALLTHROUGH */
629 case (ROFF_ami):
630 /* FALLTHROUGH */
631 case (ROFF_am1):
632 /* FALLTHROUGH */
633 case (ROFF_de):
634 /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
635 /* FALLTHROUGH */
636 case (ROFF_dei):
637 /* FALLTHROUGH */
638 case (ROFF_ig):
639 break;
640 default:
641 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
642 return(ROFF_IGN);
643 }
644
645 if ((*bufp)[pos])
646 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
647
648 roffnode_pop(r);
649 roffnode_cleanscope(r);
650 return(ROFF_IGN);
651
652 }
653
654
655 static void
656 roffnode_cleanscope(struct roff *r)
657 {
658
659 while (r->last) {
660 if (--r->last->endspan < 0)
661 break;
662 roffnode_pop(r);
663 }
664 }
665
666
667 /* ARGSUSED */
668 static enum rofferr
669 roff_ccond(ROFF_ARGS)
670 {
671
672 if (NULL == r->last) {
673 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
674 return(ROFF_IGN);
675 }
676
677 switch (r->last->tok) {
678 case (ROFF_el):
679 /* FALLTHROUGH */
680 case (ROFF_ie):
681 /* FALLTHROUGH */
682 case (ROFF_if):
683 break;
684 default:
685 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
686 return(ROFF_IGN);
687 }
688
689 if (r->last->endspan > -1) {
690 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
691 return(ROFF_IGN);
692 }
693
694 if ((*bufp)[pos])
695 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
696
697 roffnode_pop(r);
698 roffnode_cleanscope(r);
699 return(ROFF_IGN);
700 }
701
702
703 /* ARGSUSED */
704 static enum rofferr
705 roff_block(ROFF_ARGS)
706 {
707 int sv;
708 size_t sz;
709 char *name;
710
711 name = NULL;
712
713 if (ROFF_ig != tok) {
714 if ('\0' == (*bufp)[pos]) {
715 mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
716 return(ROFF_IGN);
717 }
718
719 /*
720 * Re-write `de1', since we don't really care about
721 * groff's strange compatibility mode, into `de'.
722 */
723
724 if (ROFF_de1 == tok)
725 tok = ROFF_de;
726 if (ROFF_de == tok)
727 name = *bufp + pos;
728 else
729 mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
730 roffs[tok].name);
731
732 while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
733 pos++;
734
735 while (isspace((unsigned char)(*bufp)[pos]))
736 (*bufp)[pos++] = '\0';
737 }
738
739 roffnode_push(r, tok, name, ln, ppos);
740
741 /*
742 * At the beginning of a `de' macro, clear the existing string
743 * with the same name, if there is one. New content will be
744 * added from roff_block_text() in multiline mode.
745 */
746
747 if (ROFF_de == tok)
748 roff_setstr(r, name, "", 0);
749
750 if ('\0' == (*bufp)[pos])
751 return(ROFF_IGN);
752
753 /* If present, process the custom end-of-line marker. */
754
755 sv = pos;
756 while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
757 pos++;
758
759 /*
760 * Note: groff does NOT like escape characters in the input.
761 * Instead of detecting this, we're just going to let it fly and
762 * to hell with it.
763 */
764
765 assert(pos > sv);
766 sz = (size_t)(pos - sv);
767
768 if (1 == sz && '.' == (*bufp)[sv])
769 return(ROFF_IGN);
770
771 r->last->end = mandoc_malloc(sz + 1);
772
773 memcpy(r->last->end, *bufp + sv, sz);
774 r->last->end[(int)sz] = '\0';
775
776 if ((*bufp)[pos])
777 mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
778
779 return(ROFF_IGN);
780 }
781
782
783 /* ARGSUSED */
784 static enum rofferr
785 roff_block_sub(ROFF_ARGS)
786 {
787 enum rofft t;
788 int i, j;
789
790 /*
791 * First check whether a custom macro exists at this level. If
792 * it does, then check against it. This is some of groff's
793 * stranger behaviours. If we encountered a custom end-scope
794 * tag and that tag also happens to be a "real" macro, then we
795 * need to try interpreting it again as a real macro. If it's
796 * not, then return ignore. Else continue.
797 */
798
799 if (r->last->end) {
800 for (i = pos, j = 0; r->last->end[j]; j++, i++)
801 if ((*bufp)[i] != r->last->end[j])
802 break;
803
804 if ('\0' == r->last->end[j] &&
805 ('\0' == (*bufp)[i] ||
806 ' ' == (*bufp)[i] ||
807 '\t' == (*bufp)[i])) {
808 roffnode_pop(r);
809 roffnode_cleanscope(r);
810
811 while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
812 i++;
813
814 pos = i;
815 if (ROFF_MAX != roff_parse(r, *bufp, &pos))
816 return(ROFF_RERUN);
817 return(ROFF_IGN);
818 }
819 }
820
821 /*
822 * If we have no custom end-query or lookup failed, then try
823 * pulling it out of the hashtable.
824 */
825
826 t = roff_parse(r, *bufp, &pos);
827
828 /*
829 * Macros other than block-end are only significant
830 * in `de' blocks; elsewhere, simply throw them away.
831 */
832 if (ROFF_cblock != t) {
833 if (ROFF_de == tok)
834 roff_setstr(r, r->last->name, *bufp + ppos, 1);
835 return(ROFF_IGN);
836 }
837
838 assert(roffs[t].proc);
839 return((*roffs[t].proc)(r, t, bufp, szp,
840 ln, ppos, pos, offs));
841 }
842
843
844 /* ARGSUSED */
845 static enum rofferr
846 roff_block_text(ROFF_ARGS)
847 {
848
849 if (ROFF_de == tok)
850 roff_setstr(r, r->last->name, *bufp + pos, 1);
851
852 return(ROFF_IGN);
853 }
854
855
856 /* ARGSUSED */
857 static enum rofferr
858 roff_cond_sub(ROFF_ARGS)
859 {
860 enum rofft t;
861 enum roffrule rr;
862 char *ep;
863
864 rr = r->last->rule;
865 roffnode_cleanscope(r);
866
867 /*
868 * If the macro is unknown, first check if it contains a closing
869 * delimiter `\}'. If it does, close out our scope and return
870 * the currently-scoped rule (ignore or continue). Else, drop
871 * into the currently-scoped rule.
872 */
873
874 if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
875 ep = &(*bufp)[pos];
876 for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
877 ep++;
878 if ('}' != *ep)
879 continue;
880 *ep = '&';
881 roff_ccond(r, ROFF_ccond, bufp, szp,
882 ln, pos, pos + 2, offs);
883 break;
884 }
885 return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
886 }
887
888 /*
889 * A denied conditional must evaluate its children if and only
890 * if they're either structurally required (such as loops and
891 * conditionals) or a closing macro.
892 */
893
894 if (ROFFRULE_DENY == rr)
895 if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
896 if (ROFF_ccond != t)
897 return(ROFF_IGN);
898
899 assert(roffs[t].proc);
900 return((*roffs[t].proc)(r, t, bufp, szp,
901 ln, ppos, pos, offs));
902 }
903
904 /* ARGSUSED */
905 static enum rofferr
906 roff_cond_text(ROFF_ARGS)
907 {
908 char *ep;
909 enum roffrule rr;
910
911 rr = r->last->rule;
912 roffnode_cleanscope(r);
913
914 ep = &(*bufp)[pos];
915 for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
916 ep++;
917 if ('}' != *ep)
918 continue;
919 *ep = '&';
920 roff_ccond(r, ROFF_ccond, bufp, szp,
921 ln, pos, pos + 2, offs);
922 }
923 return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
924 }
925
926 static enum roffrule
927 roff_evalcond(const char *v, int *pos)
928 {
929
930 switch (v[*pos]) {
931 case ('n'):
932 (*pos)++;
933 return(ROFFRULE_ALLOW);
934 case ('e'):
935 /* FALLTHROUGH */
936 case ('o'):
937 /* FALLTHROUGH */
938 case ('t'):
939 (*pos)++;
940 return(ROFFRULE_DENY);
941 default:
942 break;
943 }
944
945 while (v[*pos] && ' ' != v[*pos])
946 (*pos)++;
947 return(ROFFRULE_DENY);
948 }
949
950 /* ARGSUSED */
951 static enum rofferr
952 roff_line_ignore(ROFF_ARGS)
953 {
954
955 if (ROFF_it == tok)
956 mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
957
958 return(ROFF_IGN);
959 }
960
961 /* ARGSUSED */
962 static enum rofferr
963 roff_cond(ROFF_ARGS)
964 {
965 int sv;
966 enum roffrule rule;
967
968 /*
969 * An `.el' has no conditional body: it will consume the value
970 * of the current rstack entry set in prior `ie' calls or
971 * defaults to DENY.
972 *
973 * If we're not an `el', however, then evaluate the conditional.
974 */
975
976 rule = ROFF_el == tok ?
977 (r->rstackpos < 0 ?
978 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
979 roff_evalcond(*bufp, &pos);
980
981 sv = pos;
982 while (' ' == (*bufp)[pos])
983 pos++;
984
985 /*
986 * Roff is weird. If we have just white-space after the
987 * conditional, it's considered the BODY and we exit without
988 * really doing anything. Warn about this. It's probably
989 * wrong.
990 */
991
992 if ('\0' == (*bufp)[pos] && sv != pos) {
993 mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
994 return(ROFF_IGN);
995 }
996
997 roffnode_push(r, tok, NULL, ln, ppos);
998
999 r->last->rule = rule;
1000
1001 /*
1002 * An if-else will put the NEGATION of the current evaluated
1003 * conditional into the stack of rules.
1004 */
1005
1006 if (ROFF_ie == tok) {
1007 if (r->rstackpos == RSTACK_MAX - 1) {
1008 mandoc_msg(MANDOCERR_MEM,
1009 r->parse, ln, ppos, NULL);
1010 return(ROFF_ERR);
1011 }
1012 r->rstack[++r->rstackpos] =
1013 ROFFRULE_DENY == r->last->rule ?
1014 ROFFRULE_ALLOW : ROFFRULE_DENY;
1015 }
1016
1017 /* If the parent has false as its rule, then so do we. */
1018
1019 if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1020 r->last->rule = ROFFRULE_DENY;
1021
1022 /*
1023 * Determine scope. If we're invoked with "\{" trailing the
1024 * conditional, then we're in a multiline scope. Else our scope
1025 * expires on the next line.
1026 */
1027
1028 r->last->endspan = 1;
1029
1030 if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1031 r->last->endspan = -1;
1032 pos += 2;
1033 }
1034
1035 /*
1036 * If there are no arguments on the line, the next-line scope is
1037 * assumed.
1038 */
1039
1040 if ('\0' == (*bufp)[pos])
1041 return(ROFF_IGN);
1042
1043 /* Otherwise re-run the roff parser after recalculating. */
1044
1045 *offs = pos;
1046 return(ROFF_RERUN);
1047 }
1048
1049
1050 /* ARGSUSED */
1051 static enum rofferr
1052 roff_ds(ROFF_ARGS)
1053 {
1054 char *name, *string;
1055
1056 /*
1057 * A symbol is named by the first word following the macro
1058 * invocation up to a space. Its value is anything after the
1059 * name's trailing whitespace and optional double-quote. Thus,
1060 *
1061 * [.ds foo "bar " ]
1062 *
1063 * will have `bar " ' as its value.
1064 */
1065
1066 string = *bufp + pos;
1067 name = roff_getname(r, &string, ln, pos);
1068 if ('\0' == *name)
1069 return(ROFF_IGN);
1070
1071 /* Read past initial double-quote. */
1072 if ('"' == *string)
1073 string++;
1074
1075 /* The rest is the value. */
1076 roff_setstr(r, name, string, 0);
1077 return(ROFF_IGN);
1078 }
1079
1080
1081 /* ARGSUSED */
1082 static enum rofferr
1083 roff_nr(ROFF_ARGS)
1084 {
1085 const char *key;
1086 char *val;
1087 int iv;
1088 struct reg *rg;
1089
1090 val = *bufp + pos;
1091 key = roff_getname(r, &val, ln, pos);
1092 rg = r->regs->regs;
1093
1094 if (0 == strcmp(key, "nS")) {
1095 rg[(int)REG_nS].set = 1;
1096 if ((iv = mandoc_strntou(val, strlen(val), 10)) >= 0)
1097 rg[REG_nS].v.u = (unsigned)iv;
1098 else
1099 rg[(int)REG_nS].v.u = 0u;
1100 }
1101
1102 return(ROFF_IGN);
1103 }
1104
1105 /* ARGSUSED */
1106 static enum rofferr
1107 roff_rm(ROFF_ARGS)
1108 {
1109 const char *name;
1110 char *cp;
1111
1112 cp = *bufp + pos;
1113 while ('\0' != *cp) {
1114 name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1115 if ('\0' != *name)
1116 roff_setstr(r, name, NULL, 0);
1117 }
1118 return(ROFF_IGN);
1119 }
1120
1121 /* ARGSUSED */
1122 static enum rofferr
1123 roff_TE(ROFF_ARGS)
1124 {
1125
1126 if (NULL == r->tbl)
1127 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1128 else
1129 tbl_end(r->tbl);
1130
1131 r->tbl = NULL;
1132 return(ROFF_IGN);
1133 }
1134
1135 /* ARGSUSED */
1136 static enum rofferr
1137 roff_T_(ROFF_ARGS)
1138 {
1139
1140 if (NULL == r->tbl)
1141 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1142 else
1143 tbl_restart(ppos, ln, r->tbl);
1144
1145 return(ROFF_IGN);
1146 }
1147
1148 /* ARGSUSED */
1149 static enum rofferr
1150 roff_EQ(ROFF_ARGS)
1151 {
1152 struct eqn_node *e;
1153
1154 assert(NULL == r->eqn);
1155 e = eqn_alloc(ppos, ln);
1156
1157 if (r->last_eqn)
1158 r->last_eqn->next = e;
1159 else
1160 r->first_eqn = r->last_eqn = e;
1161
1162 r->eqn = r->last_eqn = e;
1163 return(ROFF_IGN);
1164 }
1165
1166 /* ARGSUSED */
1167 static enum rofferr
1168 roff_EN(ROFF_ARGS)
1169 {
1170
1171 mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1172 return(ROFF_IGN);
1173 }
1174
1175 /* ARGSUSED */
1176 static enum rofferr
1177 roff_TS(ROFF_ARGS)
1178 {
1179 struct tbl_node *t;
1180
1181 if (r->tbl) {
1182 mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1183 tbl_end(r->tbl);
1184 }
1185
1186 t = tbl_alloc(ppos, ln, r->parse);
1187
1188 if (r->last_tbl)
1189 r->last_tbl->next = t;
1190 else
1191 r->first_tbl = r->last_tbl = t;
1192
1193 r->tbl = r->last_tbl = t;
1194 return(ROFF_IGN);
1195 }
1196
1197 /* ARGSUSED */
1198 static enum rofferr
1199 roff_so(ROFF_ARGS)
1200 {
1201 char *name;
1202
1203 mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1204
1205 /*
1206 * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1207 * opening anything that's not in our cwd or anything beneath
1208 * it. Thus, explicitly disallow traversing up the file-system
1209 * or using absolute paths.
1210 */
1211
1212 name = *bufp + pos;
1213 if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1214 mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1215 return(ROFF_ERR);
1216 }
1217
1218 *offs = pos;
1219 return(ROFF_SO);
1220 }
1221
1222 /* ARGSUSED */
1223 static enum rofferr
1224 roff_userdef(ROFF_ARGS)
1225 {
1226 const char *arg[9];
1227 char *cp, *n1, *n2;
1228 int i;
1229
1230 /*
1231 * Collect pointers to macro argument strings
1232 * and null-terminate them.
1233 */
1234 cp = *bufp + pos;
1235 for (i = 0; i < 9; i++)
1236 arg[i] = '\0' == *cp ? "" :
1237 mandoc_getarg(r->parse, &cp, ln, &pos);
1238
1239 /*
1240 * Expand macro arguments.
1241 */
1242 *szp = 0;
1243 n1 = cp = mandoc_strdup(r->current_string);
1244 while (NULL != (cp = strstr(cp, "\\$"))) {
1245 i = cp[2] - '1';
1246 if (0 > i || 8 < i) {
1247 /* Not an argument invocation. */
1248 cp += 2;
1249 continue;
1250 }
1251
1252 *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1253 n2 = mandoc_malloc(*szp);
1254
1255 strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1256 strlcat(n2, arg[i], *szp);
1257 strlcat(n2, cp + 3, *szp);
1258
1259 cp = n2 + (cp - n1);
1260 free(n1);
1261 n1 = n2;
1262 }
1263
1264 /*
1265 * Replace the macro invocation
1266 * by the expanded macro.
1267 */
1268 free(*bufp);
1269 *bufp = n1;
1270 if (0 == *szp)
1271 *szp = strlen(*bufp) + 1;
1272
1273 return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1274 ROFF_REPARSE : ROFF_APPEND);
1275 }
1276
1277 static char *
1278 roff_getname(struct roff *r, char **cpp, int ln, int pos)
1279 {
1280 char *name, *cp;
1281
1282 name = *cpp;
1283 if ('\0' == *name)
1284 return(name);
1285
1286 /* Read until end of name. */
1287 for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1288 if ('\\' != *cp)
1289 continue;
1290 cp++;
1291 if ('\\' == *cp)
1292 continue;
1293 mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1294 *cp = '\0';
1295 name = cp;
1296 }
1297
1298 /* Nil-terminate name. */
1299 if ('\0' != *cp)
1300 *(cp++) = '\0';
1301
1302 /* Read past spaces. */
1303 while (' ' == *cp)
1304 cp++;
1305
1306 *cpp = cp;
1307 return(name);
1308 }
1309
1310 /*
1311 * Store *string into the user-defined string called *name.
1312 * In multiline mode, append to an existing entry and append '\n';
1313 * else replace the existing entry, if there is one.
1314 * To clear an existing entry, call with (*r, *name, NULL, 0).
1315 */
1316 static void
1317 roff_setstr(struct roff *r, const char *name, const char *string,
1318 int multiline)
1319 {
1320 struct roffstr *n;
1321 char *c;
1322 size_t oldch, newch;
1323
1324 /* Search for an existing string with the same name. */
1325 n = r->first_string;
1326 while (n && strcmp(name, n->name))
1327 n = n->next;
1328
1329 if (NULL == n) {
1330 /* Create a new string table entry. */
1331 n = mandoc_malloc(sizeof(struct roffstr));
1332 n->name = mandoc_strdup(name);
1333 n->string = NULL;
1334 n->next = r->first_string;
1335 r->first_string = n;
1336 } else if (0 == multiline) {
1337 /* In multiline mode, append; else replace. */
1338 free(n->string);
1339 n->string = NULL;
1340 }
1341
1342 if (NULL == string)
1343 return;
1344
1345 /*
1346 * One additional byte for the '\n' in multiline mode,
1347 * and one for the terminating '\0'.
1348 */
1349 newch = strlen(string) + (multiline ? 2u : 1u);
1350 if (NULL == n->string) {
1351 n->string = mandoc_malloc(newch);
1352 *n->string = '\0';
1353 oldch = 0;
1354 } else {
1355 oldch = strlen(n->string);
1356 n->string = mandoc_realloc(n->string, oldch + newch);
1357 }
1358
1359 /* Skip existing content in the destination buffer. */
1360 c = n->string + (int)oldch;
1361
1362 /* Append new content to the destination buffer. */
1363 while (*string) {
1364 /*
1365 * Rudimentary roff copy mode:
1366 * Handle escaped backslashes.
1367 */
1368 if ('\\' == *string && '\\' == *(string + 1))
1369 string++;
1370 *c++ = *string++;
1371 }
1372
1373 /* Append terminating bytes. */
1374 if (multiline)
1375 *c++ = '\n';
1376 *c = '\0';
1377 }
1378
1379 static const char *
1380 roff_getstrn(const struct roff *r, const char *name, size_t len)
1381 {
1382 const struct roffstr *n;
1383
1384 n = r->first_string;
1385 while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1386 n = n->next;
1387
1388 return(n ? n->string : NULL);
1389 }
1390
1391 static void
1392 roff_freestr(struct roff *r)
1393 {
1394 struct roffstr *n, *nn;
1395
1396 for (n = r->first_string; n; n = nn) {
1397 free(n->name);
1398 free(n->string);
1399 nn = n->next;
1400 free(n);
1401 }
1402
1403 r->first_string = NULL;
1404 }
1405
1406 const struct tbl_span *
1407 roff_span(const struct roff *r)
1408 {
1409
1410 return(r->tbl ? tbl_span(r->tbl) : NULL);
1411 }
1412
1413 const struct eqn *
1414 roff_eqn(const struct roff *r)
1415 {
1416
1417 return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1418 }
A git conversion mirror of mandoc.bsd.lv