]> git.cameronkatri.com Git - mandoc.git/blob - roff.c
git.ckatri.com / mandoc.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Delete five standards that are:
[mandoc.git] / roff.c
1 /* $Id: roff.c,v 1.238 2014/11/01 06:03:13 schwarze Exp $ */
2 /*
3 * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2014 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 #include "config.h"
19
20 #include <sys/types.h>
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27
28 #include "mandoc.h"
29 #include "mandoc_aux.h"
30 #include "libmandoc.h"
31 #include "libroff.h"
32
33 /* Maximum number of nested if-else conditionals. */
34 #define RSTACK_MAX 128
35
36 /* Maximum number of string expansions per line, to break infinite loops. */
37 #define EXPAND_LIMIT 1000
38
39 enum rofft {
40 ROFF_ad,
41 ROFF_am,
42 ROFF_ami,
43 ROFF_am1,
44 ROFF_as,
45 ROFF_cc,
46 ROFF_ce,
47 ROFF_de,
48 ROFF_dei,
49 ROFF_de1,
50 ROFF_ds,
51 ROFF_el,
52 ROFF_fam,
53 ROFF_hw,
54 ROFF_hy,
55 ROFF_ie,
56 ROFF_if,
57 ROFF_ig,
58 ROFF_it,
59 ROFF_ne,
60 ROFF_nh,
61 ROFF_nr,
62 ROFF_ns,
63 ROFF_pl,
64 ROFF_ps,
65 ROFF_rm,
66 ROFF_rr,
67 ROFF_so,
68 ROFF_ta,
69 ROFF_tr,
70 ROFF_Dd,
71 ROFF_TH,
72 ROFF_TS,
73 ROFF_TE,
74 ROFF_T_,
75 ROFF_EQ,
76 ROFF_EN,
77 ROFF_cblock,
78 ROFF_USERDEF,
79 ROFF_MAX
80 };
81
82 /*
83 * An incredibly-simple string buffer.
84 */
85 struct roffstr {
86 char *p; /* nil-terminated buffer */
87 size_t sz; /* saved strlen(p) */
88 };
89
90 /*
91 * A key-value roffstr pair as part of a singly-linked list.
92 */
93 struct roffkv {
94 struct roffstr key;
95 struct roffstr val;
96 struct roffkv *next; /* next in list */
97 };
98
99 /*
100 * A single number register as part of a singly-linked list.
101 */
102 struct roffreg {
103 struct roffstr key;
104 int val;
105 struct roffreg *next;
106 };
107
108 struct roff {
109 struct mparse *parse; /* parse point */
110 const struct mchars *mchars; /* character table */
111 struct roffnode *last; /* leaf of stack */
112 int *rstack; /* stack of inverted `ie' values */
113 struct roffreg *regtab; /* number registers */
114 struct roffkv *strtab; /* user-defined strings & macros */
115 struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
116 struct roffstr *xtab; /* single-byte trans table (`tr') */
117 const char *current_string; /* value of last called user macro */
118 struct tbl_node *first_tbl; /* first table parsed */
119 struct tbl_node *last_tbl; /* last table parsed */
120 struct tbl_node *tbl; /* current table being parsed */
121 struct eqn_node *last_eqn; /* last equation parsed */
122 struct eqn_node *first_eqn; /* first equation parsed */
123 struct eqn_node *eqn; /* current equation being parsed */
124 int eqn_inline; /* current equation is inline */
125 int options; /* parse options */
126 int rstacksz; /* current size limit of rstack */
127 int rstackpos; /* position in rstack */
128 int format; /* current file in mdoc or man format */
129 char control; /* control character */
130 };
131
132 struct roffnode {
133 enum rofft tok; /* type of node */
134 struct roffnode *parent; /* up one in stack */
135 int line; /* parse line */
136 int col; /* parse col */
137 char *name; /* node name, e.g. macro name */
138 char *end; /* end-rules: custom token */
139 int endspan; /* end-rules: next-line or infty */
140 int rule; /* current evaluation rule */
141 };
142
143 #define ROFF_ARGS struct roff *r, /* parse ctx */ \
144 enum rofft tok, /* tok of macro */ \
145 struct buf *buf, /* input buffer */ \
146 int ln, /* parse line */ \
147 int ppos, /* original pos in buffer */ \
148 int pos, /* current pos in buffer */ \
149 int *offs /* reset offset of buffer data */
150
151 typedef enum rofferr (*roffproc)(ROFF_ARGS);
152
153 struct roffmac {
154 const char *name; /* macro name */
155 roffproc proc; /* process new macro */
156 roffproc text; /* process as child text of macro */
157 roffproc sub; /* process as child of macro */
158 int flags;
159 #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
160 struct roffmac *next;
161 };
162
163 struct predef {
164 const char *name; /* predefined input name */
165 const char *str; /* replacement symbol */
166 };
167
168 #define PREDEF(__name, __str) \
169 { (__name), (__str) },
170
171 static enum rofft roffhash_find(const char *, size_t);
172 static void roffhash_init(void);
173 static void roffnode_cleanscope(struct roff *);
174 static void roffnode_pop(struct roff *);
175 static void roffnode_push(struct roff *, enum rofft,
176 const char *, int, int);
177 static enum rofferr roff_block(ROFF_ARGS);
178 static enum rofferr roff_block_text(ROFF_ARGS);
179 static enum rofferr roff_block_sub(ROFF_ARGS);
180 static enum rofferr roff_cblock(ROFF_ARGS);
181 static enum rofferr roff_cc(ROFF_ARGS);
182 static void roff_ccond(struct roff *, int, int);
183 static enum rofferr roff_cond(ROFF_ARGS);
184 static enum rofferr roff_cond_text(ROFF_ARGS);
185 static enum rofferr roff_cond_sub(ROFF_ARGS);
186 static enum rofferr roff_ds(ROFF_ARGS);
187 static enum rofferr roff_eqndelim(struct roff *, struct buf *, int);
188 static int roff_evalcond(struct roff *r, int,
189 const char *, int *);
190 static int roff_evalnum(struct roff *, int,
191 const char *, int *, int *, int);
192 static int roff_evalpar(struct roff *, int,
193 const char *, int *, int *);
194 static int roff_evalstrcond(const char *, int *);
195 static void roff_free1(struct roff *);
196 static void roff_freereg(struct roffreg *);
197 static void roff_freestr(struct roffkv *);
198 static size_t roff_getname(struct roff *, char **, int, int);
199 static int roff_getnum(const char *, int *, int *);
200 static int roff_getop(const char *, int *, char *);
201 static int roff_getregn(const struct roff *,
202 const char *, size_t);
203 static int roff_getregro(const char *name);
204 static const char *roff_getstrn(const struct roff *,
205 const char *, size_t);
206 static enum rofferr roff_it(ROFF_ARGS);
207 static enum rofferr roff_line_ignore(ROFF_ARGS);
208 static enum rofferr roff_nr(ROFF_ARGS);
209 static enum rofft roff_parse(struct roff *, char *, int *,
210 int, int);
211 static enum rofferr roff_parsetext(struct buf *, int, int *);
212 static enum rofferr roff_res(struct roff *, struct buf *, int, int);
213 static enum rofferr roff_rm(ROFF_ARGS);
214 static enum rofferr roff_rr(ROFF_ARGS);
215 static void roff_setstr(struct roff *,
216 const char *, const char *, int);
217 static void roff_setstrn(struct roffkv **, const char *,
218 size_t, const char *, size_t, int);
219 static enum rofferr roff_so(ROFF_ARGS);
220 static enum rofferr roff_tr(ROFF_ARGS);
221 static enum rofferr roff_Dd(ROFF_ARGS);
222 static enum rofferr roff_TH(ROFF_ARGS);
223 static enum rofferr roff_TE(ROFF_ARGS);
224 static enum rofferr roff_TS(ROFF_ARGS);
225 static enum rofferr roff_EQ(ROFF_ARGS);
226 static enum rofferr roff_EN(ROFF_ARGS);
227 static enum rofferr roff_T_(ROFF_ARGS);
228 static enum rofferr roff_userdef(ROFF_ARGS);
229
230 /* See roffhash_find() */
231
232 #define ASCII_HI 126
233 #define ASCII_LO 33
234 #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
235
236 static struct roffmac *hash[HASHWIDTH];
237
238 static struct roffmac roffs[ROFF_MAX] = {
239 { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
240 { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
241 { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
242 { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
243 { "as", roff_ds, NULL, NULL, 0, NULL },
244 { "cc", roff_cc, NULL, NULL, 0, NULL },
245 { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
246 { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
247 { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
248 { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
249 { "ds", roff_ds, NULL, NULL, 0, NULL },
250 { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
251 { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
252 { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
253 { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
254 { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
255 { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
256 { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
257 { "it", roff_it, NULL, NULL, 0, NULL },
258 { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
259 { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
260 { "nr", roff_nr, NULL, NULL, 0, NULL },
261 { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
262 { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
263 { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
264 { "rm", roff_rm, NULL, NULL, 0, NULL },
265 { "rr", roff_rr, NULL, NULL, 0, NULL },
266 { "so", roff_so, NULL, NULL, 0, NULL },
267 { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
268 { "tr", roff_tr, NULL, NULL, 0, NULL },
269 { "Dd", roff_Dd, NULL, NULL, 0, NULL },
270 { "TH", roff_TH, NULL, NULL, 0, NULL },
271 { "TS", roff_TS, NULL, NULL, 0, NULL },
272 { "TE", roff_TE, NULL, NULL, 0, NULL },
273 { "T&", roff_T_, NULL, NULL, 0, NULL },
274 { "EQ", roff_EQ, NULL, NULL, 0, NULL },
275 { "EN", roff_EN, NULL, NULL, 0, NULL },
276 { ".", roff_cblock, NULL, NULL, 0, NULL },
277 { NULL, roff_userdef, NULL, NULL, 0, NULL },
278 };
279
280 /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
281 const char *const __mdoc_reserved[] = {
282 "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
283 "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
284 "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
285 "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
286 "Dt", "Dv", "Dx", "D1",
287 "Ec", "Ed", "Ef", "Ek", "El", "Em",
288 "En", "Eo", "Er", "Es", "Ev", "Ex",
289 "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
290 "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
291 "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
292 "Oc", "Oo", "Op", "Os", "Ot", "Ox",
293 "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
294 "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
295 "Sc", "Sh", "Sm", "So", "Sq",
296 "Ss", "St", "Sx", "Sy",
297 "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
298 "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
299 "%P", "%Q", "%R", "%T", "%U", "%V",
300 NULL
301 };
302
303 /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
304 const char *const __man_reserved[] = {
305 "AT", "B", "BI", "BR", "DT",
306 "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
307 "LP", "OP", "P", "PD", "PP",
308 "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
309 "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
310 NULL
311 };
312
313 /* Array of injected predefined strings. */
314 #define PREDEFS_MAX 38
315 static const struct predef predefs[PREDEFS_MAX] = {
316 #include "predefs.in"
317 };
318
319 /* See roffhash_find() */
320 #define ROFF_HASH(p) (p[0] - ASCII_LO)
321
322 static int roffit_lines; /* number of lines to delay */
323 static char *roffit_macro; /* nil-terminated macro line */
324
325
326 static void
327 roffhash_init(void)
328 {
329 struct roffmac *n;
330 int buc, i;
331
332 for (i = 0; i < (int)ROFF_USERDEF; i++) {
333 assert(roffs[i].name[0] >= ASCII_LO);
334 assert(roffs[i].name[0] <= ASCII_HI);
335
336 buc = ROFF_HASH(roffs[i].name);
337
338 if (NULL != (n = hash[buc])) {
339 for ( ; n->next; n = n->next)
340 /* Do nothing. */ ;
341 n->next = &roffs[i];
342 } else
343 hash[buc] = &roffs[i];
344 }
345 }
346
347 /*
348 * Look up a roff token by its name. Returns ROFF_MAX if no macro by
349 * the nil-terminated string name could be found.
350 */
351 static enum rofft
352 roffhash_find(const char *p, size_t s)
353 {
354 int buc;
355 struct roffmac *n;
356
357 /*
358 * libroff has an extremely simple hashtable, for the time
359 * being, which simply keys on the first character, which must
360 * be printable, then walks a chain. It works well enough until
361 * optimised.
362 */
363
364 if (p[0] < ASCII_LO || p[0] > ASCII_HI)
365 return(ROFF_MAX);
366
367 buc = ROFF_HASH(p);
368
369 if (NULL == (n = hash[buc]))
370 return(ROFF_MAX);
371 for ( ; n; n = n->next)
372 if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
373 return((enum rofft)(n - roffs));
374
375 return(ROFF_MAX);
376 }
377
378 /*
379 * Pop the current node off of the stack of roff instructions currently
380 * pending.
381 */
382 static void
383 roffnode_pop(struct roff *r)
384 {
385 struct roffnode *p;
386
387 assert(r->last);
388 p = r->last;
389
390 r->last = r->last->parent;
391 free(p->name);
392 free(p->end);
393 free(p);
394 }
395
396 /*
397 * Push a roff node onto the instruction stack. This must later be
398 * removed with roffnode_pop().
399 */
400 static void
401 roffnode_push(struct roff *r, enum rofft tok, const char *name,
402 int line, int col)
403 {
404 struct roffnode *p;
405
406 p = mandoc_calloc(1, sizeof(struct roffnode));
407 p->tok = tok;
408 if (name)
409 p->name = mandoc_strdup(name);
410 p->parent = r->last;
411 p->line = line;
412 p->col = col;
413 p->rule = p->parent ? p->parent->rule : 0;
414
415 r->last = p;
416 }
417
418 static void
419 roff_free1(struct roff *r)
420 {
421 struct tbl_node *tbl;
422 struct eqn_node *e;
423 int i;
424
425 while (NULL != (tbl = r->first_tbl)) {
426 r->first_tbl = tbl->next;
427 tbl_free(tbl);
428 }
429 r->first_tbl = r->last_tbl = r->tbl = NULL;
430
431 while (NULL != (e = r->first_eqn)) {
432 r->first_eqn = e->next;
433 eqn_free(e);
434 }
435 r->first_eqn = r->last_eqn = r->eqn = NULL;
436
437 while (r->last)
438 roffnode_pop(r);
439
440 free (r->rstack);
441 r->rstack = NULL;
442 r->rstacksz = 0;
443 r->rstackpos = -1;
444
445 roff_freereg(r->regtab);
446 r->regtab = NULL;
447
448 roff_freestr(r->strtab);
449 roff_freestr(r->xmbtab);
450 r->strtab = r->xmbtab = NULL;
451
452 if (r->xtab)
453 for (i = 0; i < 128; i++)
454 free(r->xtab[i].p);
455 free(r->xtab);
456 r->xtab = NULL;
457 }
458
459 void
460 roff_reset(struct roff *r)
461 {
462
463 roff_free1(r);
464 r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
465 r->control = 0;
466 }
467
468 void
469 roff_free(struct roff *r)
470 {
471
472 roff_free1(r);
473 free(r);
474 }
475
476 struct roff *
477 roff_alloc(struct mparse *parse, const struct mchars *mchars, int options)
478 {
479 struct roff *r;
480
481 r = mandoc_calloc(1, sizeof(struct roff));
482 r->parse = parse;
483 r->mchars = mchars;
484 r->options = options;
485 r->format = options & (MPARSE_MDOC | MPARSE_MAN);
486 r->rstackpos = -1;
487
488 roffhash_init();
489
490 return(r);
491 }
492
493 /*
494 * In the current line, expand escape sequences that tend to get
495 * used in numerical expressions and conditional requests.
496 * Also check the syntax of the remaining escape sequences.
497 */
498 static enum rofferr
499 roff_res(struct roff *r, struct buf *buf, int ln, int pos)
500 {
501 char ubuf[24]; /* buffer to print the number */
502 const char *start; /* start of the string to process */
503 char *stesc; /* start of an escape sequence ('\\') */
504 const char *stnam; /* start of the name, after "[(*" */
505 const char *cp; /* end of the name, e.g. before ']' */
506 const char *res; /* the string to be substituted */
507 char *nbuf; /* new buffer to copy buf->buf to */
508 size_t maxl; /* expected length of the escape name */
509 size_t naml; /* actual length of the escape name */
510 enum mandoc_esc esc; /* type of the escape sequence */
511 int inaml; /* length returned from mandoc_escape() */
512 int expand_count; /* to avoid infinite loops */
513 int npos; /* position in numeric expression */
514 int arg_complete; /* argument not interrupted by eol */
515 char term; /* character terminating the escape */
516
517 expand_count = 0;
518 start = buf->buf + pos;
519 stesc = strchr(start, '\0') - 1;
520 while (stesc-- > start) {
521
522 /* Search backwards for the next backslash. */
523
524 if (*stesc != '\\')
525 continue;
526
527 /* If it is escaped, skip it. */
528
529 for (cp = stesc - 1; cp >= start; cp--)
530 if (*cp != '\\')
531 break;
532
533 if ((stesc - cp) % 2 == 0) {
534 stesc = (char *)cp;
535 continue;
536 }
537
538 /* Decide whether to expand or to check only. */
539
540 term = '\0';
541 cp = stesc + 1;
542 switch (*cp) {
543 case '*':
544 res = NULL;
545 break;
546 case 'B':
547 /* FALLTHROUGH */
548 case 'w':
549 term = cp[1];
550 /* FALLTHROUGH */
551 case 'n':
552 res = ubuf;
553 break;
554 default:
555 esc = mandoc_escape(&cp, &stnam, &inaml);
556 if (esc == ESCAPE_ERROR ||
557 (esc == ESCAPE_SPECIAL &&
558 mchars_spec2cp(r->mchars, stnam, inaml) < 0))
559 mandoc_vmsg(MANDOCERR_ESC_BAD,
560 r->parse, ln, (int)(stesc - buf->buf),
561 "%.*s", (int)(cp - stesc), stesc);
562 continue;
563 }
564
565 if (EXPAND_LIMIT < ++expand_count) {
566 mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
567 ln, (int)(stesc - buf->buf), NULL);
568 return(ROFF_IGN);
569 }
570
571 /*
572 * The third character decides the length
573 * of the name of the string or register.
574 * Save a pointer to the name.
575 */
576
577 if (term == '\0') {
578 switch (*++cp) {
579 case '\0':
580 maxl = 0;
581 break;
582 case '(':
583 cp++;
584 maxl = 2;
585 break;
586 case '[':
587 cp++;
588 term = ']';
589 maxl = 0;
590 break;
591 default:
592 maxl = 1;
593 break;
594 }
595 } else {
596 cp += 2;
597 maxl = 0;
598 }
599 stnam = cp;
600
601 /* Advance to the end of the name. */
602
603 arg_complete = 1;
604 for (naml = 0; maxl == 0 || naml < maxl; naml++, cp++) {
605 if (*cp == '\0') {
606 mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
607 ln, (int)(stesc - buf->buf), stesc);
608 arg_complete = 0;
609 break;
610 }
611 if (maxl == 0 && *cp == term) {
612 cp++;
613 break;
614 }
615 }
616
617 /*
618 * Retrieve the replacement string; if it is
619 * undefined, resume searching for escapes.
620 */
621
622 switch (stesc[1]) {
623 case '*':
624 if (arg_complete)
625 res = roff_getstrn(r, stnam, naml);
626 break;
627 case 'B':
628 npos = 0;
629 ubuf[0] = arg_complete &&
630 roff_evalnum(r, ln, stnam, &npos, NULL, 0) &&
631 stnam + npos + 1 == cp ? '1' : '0';
632 ubuf[1] = '\0';
633 break;
634 case 'n':
635 if (arg_complete)
636 (void)snprintf(ubuf, sizeof(ubuf), "%d",
637 roff_getregn(r, stnam, naml));
638 else
639 ubuf[0] = '\0';
640 break;
641 case 'w':
642 /* use even incomplete args */
643 (void)snprintf(ubuf, sizeof(ubuf), "%d",
644 24 * (int)naml);
645 break;
646 }
647
648 if (res == NULL) {
649 mandoc_vmsg(MANDOCERR_STR_UNDEF,
650 r->parse, ln, (int)(stesc - buf->buf),
651 "%.*s", (int)naml, stnam);
652 res = "";
653 }
654
655 /* Replace the escape sequence by the string. */
656
657 *stesc = '\0';
658 buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
659 buf->buf, res, cp) + 1;
660
661 /* Prepare for the next replacement. */
662
663 start = nbuf + pos;
664 stesc = nbuf + (stesc - buf->buf) + strlen(res);
665 free(buf->buf);
666 buf->buf = nbuf;
667 }
668 return(ROFF_CONT);
669 }
670
671 /*
672 * Process text streams:
673 * Convert all breakable hyphens into ASCII_HYPH.
674 * Decrement and spring input line trap.
675 */
676 static enum rofferr
677 roff_parsetext(struct buf *buf, int pos, int *offs)
678 {
679 size_t sz;
680 const char *start;
681 char *p;
682 int isz;
683 enum mandoc_esc esc;
684
685 start = p = buf->buf + pos;
686
687 while (*p != '\0') {
688 sz = strcspn(p, "-\\");
689 p += sz;
690
691 if (*p == '\0')
692 break;
693
694 if (*p == '\\') {
695 /* Skip over escapes. */
696 p++;
697 esc = mandoc_escape((const char **)&p, NULL, NULL);
698 if (esc == ESCAPE_ERROR)
699 break;
700 continue;
701 } else if (p == start) {
702 p++;
703 continue;
704 }
705
706 if (isalpha((unsigned char)p[-1]) &&
707 isalpha((unsigned char)p[1]))
708 *p = ASCII_HYPH;
709 p++;
710 }
711
712 /* Spring the input line trap. */
713 if (roffit_lines == 1) {
714 isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
715 free(buf->buf);
716 buf->buf = p;
717 buf->sz = isz + 1;
718 *offs = 0;
719 free(roffit_macro);
720 roffit_lines = 0;
721 return(ROFF_REPARSE);
722 } else if (roffit_lines > 1)
723 --roffit_lines;
724 return(ROFF_CONT);
725 }
726
727 enum rofferr
728 roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
729 {
730 enum rofft t;
731 enum rofferr e;
732 int pos; /* parse point */
733 int ppos; /* original offset in buf->buf */
734 int ctl; /* macro line (boolean) */
735
736 ppos = pos = *offs;
737
738 /* Handle in-line equation delimiters. */
739
740 if (r->tbl == NULL &&
741 r->last_eqn != NULL && r->last_eqn->delim &&
742 (r->eqn == NULL || r->eqn_inline)) {
743 e = roff_eqndelim(r, buf, pos);
744 if (e == ROFF_REPARSE)
745 return(e);
746 assert(e == ROFF_CONT);
747 }
748
749 /* Expand some escape sequences. */
750
751 e = roff_res(r, buf, ln, pos);
752 if (e == ROFF_IGN)
753 return(e);
754 assert(e == ROFF_CONT);
755
756 ctl = roff_getcontrol(r, buf->buf, &pos);
757
758 /*
759 * First, if a scope is open and we're not a macro, pass the
760 * text through the macro's filter. If a scope isn't open and
761 * we're not a macro, just let it through.
762 * Finally, if there's an equation scope open, divert it into it
763 * no matter our state.
764 */
765
766 if (r->last && ! ctl) {
767 t = r->last->tok;
768 assert(roffs[t].text);
769 e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
770 assert(e == ROFF_IGN || e == ROFF_CONT);
771 if (e != ROFF_CONT)
772 return(e);
773 }
774 if (r->eqn)
775 return(eqn_read(&r->eqn, ln, buf->buf, ppos, offs));
776 if ( ! ctl) {
777 if (r->tbl)
778 return(tbl_read(r->tbl, ln, buf->buf, pos));
779 return(roff_parsetext(buf, pos, offs));
780 }
781
782 /* Skip empty request lines. */
783
784 if (buf->buf[pos] == '"') {
785 mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
786 ln, pos, NULL);
787 return(ROFF_IGN);
788 } else if (buf->buf[pos] == '\0')
789 return(ROFF_IGN);
790
791 /*
792 * If a scope is open, go to the child handler for that macro,
793 * as it may want to preprocess before doing anything with it.
794 * Don't do so if an equation is open.
795 */
796
797 if (r->last) {
798 t = r->last->tok;
799 assert(roffs[t].sub);
800 return((*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs));
801 }
802
803 /*
804 * Lastly, as we've no scope open, try to look up and execute
805 * the new macro. If no macro is found, simply return and let
806 * the compilers handle it.
807 */
808
809 if ((t = roff_parse(r, buf->buf, &pos, ln, ppos)) == ROFF_MAX)
810 return(ROFF_CONT);
811
812 assert(roffs[t].proc);
813 return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
814 }
815
816 void
817 roff_endparse(struct roff *r)
818 {
819
820 if (r->last)
821 mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
822 r->last->line, r->last->col,
823 roffs[r->last->tok].name);
824
825 if (r->eqn) {
826 mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
827 r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
828 eqn_end(&r->eqn);
829 }
830
831 if (r->tbl) {
832 mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
833 r->tbl->line, r->tbl->pos, "TS");
834 tbl_end(&r->tbl);
835 }
836 }
837
838 /*
839 * Parse a roff node's type from the input buffer. This must be in the
840 * form of ".foo xxx" in the usual way.
841 */
842 static enum rofft
843 roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
844 {
845 char *cp;
846 const char *mac;
847 size_t maclen;
848 enum rofft t;
849
850 cp = buf + *pos;
851
852 if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
853 return(ROFF_MAX);
854
855 mac = cp;
856 maclen = roff_getname(r, &cp, ln, ppos);
857
858 t = (r->current_string = roff_getstrn(r, mac, maclen))
859 ? ROFF_USERDEF : roffhash_find(mac, maclen);
860
861 if (ROFF_MAX != t)
862 *pos = cp - buf;
863
864 return(t);
865 }
866
867 static enum rofferr
868 roff_cblock(ROFF_ARGS)
869 {
870
871 /*
872 * A block-close `..' should only be invoked as a child of an
873 * ignore macro, otherwise raise a warning and just ignore it.
874 */
875
876 if (r->last == NULL) {
877 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
878 ln, ppos, "..");
879 return(ROFF_IGN);
880 }
881
882 switch (r->last->tok) {
883 case ROFF_am:
884 /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
885 /* FALLTHROUGH */
886 case ROFF_ami:
887 /* FALLTHROUGH */
888 case ROFF_de:
889 /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
890 /* FALLTHROUGH */
891 case ROFF_dei:
892 /* FALLTHROUGH */
893 case ROFF_ig:
894 break;
895 default:
896 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
897 ln, ppos, "..");
898 return(ROFF_IGN);
899 }
900
901 if (buf->buf[pos] != '\0')
902 mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
903 ".. %s", buf->buf + pos);
904
905 roffnode_pop(r);
906 roffnode_cleanscope(r);
907 return(ROFF_IGN);
908
909 }
910
911 static void
912 roffnode_cleanscope(struct roff *r)
913 {
914
915 while (r->last) {
916 if (--r->last->endspan != 0)
917 break;
918 roffnode_pop(r);
919 }
920 }
921
922 static void
923 roff_ccond(struct roff *r, int ln, int ppos)
924 {
925
926 if (NULL == r->last) {
927 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
928 ln, ppos, "\\}");
929 return;
930 }
931
932 switch (r->last->tok) {
933 case ROFF_el:
934 /* FALLTHROUGH */
935 case ROFF_ie:
936 /* FALLTHROUGH */
937 case ROFF_if:
938 break;
939 default:
940 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
941 ln, ppos, "\\}");
942 return;
943 }
944
945 if (r->last->endspan > -1) {
946 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
947 ln, ppos, "\\}");
948 return;
949 }
950
951 roffnode_pop(r);
952 roffnode_cleanscope(r);
953 return;
954 }
955
956 static enum rofferr
957 roff_block(ROFF_ARGS)
958 {
959 const char *name;
960 char *iname, *cp;
961 size_t namesz;
962
963 /* Ignore groff compatibility mode for now. */
964
965 if (tok == ROFF_de1)
966 tok = ROFF_de;
967 else if (tok == ROFF_am1)
968 tok = ROFF_am;
969
970 /* Parse the macro name argument. */
971
972 cp = buf->buf + pos;
973 if (tok == ROFF_ig) {
974 iname = NULL;
975 namesz = 0;
976 } else {
977 iname = cp;
978 namesz = roff_getname(r, &cp, ln, ppos);
979 iname[namesz] = '\0';
980 }
981
982 /* Resolve the macro name argument if it is indirect. */
983
984 if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
985 if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
986 mandoc_vmsg(MANDOCERR_STR_UNDEF,
987 r->parse, ln, (int)(iname - buf->buf),
988 "%.*s", (int)namesz, iname);
989 namesz = 0;
990 } else
991 namesz = strlen(name);
992 } else
993 name = iname;
994
995 if (namesz == 0 && tok != ROFF_ig) {
996 mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
997 ln, ppos, roffs[tok].name);
998 return(ROFF_IGN);
999 }
1000
1001 roffnode_push(r, tok, name, ln, ppos);
1002
1003 /*
1004 * At the beginning of a `de' macro, clear the existing string
1005 * with the same name, if there is one. New content will be
1006 * appended from roff_block_text() in multiline mode.
1007 */
1008
1009 if (tok == ROFF_de || tok == ROFF_dei)
1010 roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1011
1012 if (*cp == '\0')
1013 return(ROFF_IGN);
1014
1015 /* Get the custom end marker. */
1016
1017 iname = cp;
1018 namesz = roff_getname(r, &cp, ln, ppos);
1019
1020 /* Resolve the end marker if it is indirect. */
1021
1022 if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1023 if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
1024 mandoc_vmsg(MANDOCERR_STR_UNDEF,
1025 r->parse, ln, (int)(iname - buf->buf),
1026 "%.*s", (int)namesz, iname);
1027 namesz = 0;
1028 } else
1029 namesz = strlen(name);
1030 } else
1031 name = iname;
1032
1033 if (namesz)
1034 r->last->end = mandoc_strndup(name, namesz);
1035
1036 if (*cp != '\0')
1037 mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1038 ln, pos, ".%s ... %s", roffs[tok].name, cp);
1039
1040 return(ROFF_IGN);
1041 }
1042
1043 static enum rofferr
1044 roff_block_sub(ROFF_ARGS)
1045 {
1046 enum rofft t;
1047 int i, j;
1048
1049 /*
1050 * First check whether a custom macro exists at this level. If
1051 * it does, then check against it. This is some of groff's
1052 * stranger behaviours. If we encountered a custom end-scope
1053 * tag and that tag also happens to be a "real" macro, then we
1054 * need to try interpreting it again as a real macro. If it's
1055 * not, then return ignore. Else continue.
1056 */
1057
1058 if (r->last->end) {
1059 for (i = pos, j = 0; r->last->end[j]; j++, i++)
1060 if (buf->buf[i] != r->last->end[j])
1061 break;
1062
1063 if (r->last->end[j] == '\0' &&
1064 (buf->buf[i] == '\0' ||
1065 buf->buf[i] == ' ' ||
1066 buf->buf[i] == '\t')) {
1067 roffnode_pop(r);
1068 roffnode_cleanscope(r);
1069
1070 while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1071 i++;
1072
1073 pos = i;
1074 if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1075 ROFF_MAX)
1076 return(ROFF_RERUN);
1077 return(ROFF_IGN);
1078 }
1079 }
1080
1081 /*
1082 * If we have no custom end-query or lookup failed, then try
1083 * pulling it out of the hashtable.
1084 */
1085
1086 t = roff_parse(r, buf->buf, &pos, ln, ppos);
1087
1088 if (t != ROFF_cblock) {
1089 if (tok != ROFF_ig)
1090 roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1091 return(ROFF_IGN);
1092 }
1093
1094 assert(roffs[t].proc);
1095 return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1096 }
1097
1098 static enum rofferr
1099 roff_block_text(ROFF_ARGS)
1100 {
1101
1102 if (tok != ROFF_ig)
1103 roff_setstr(r, r->last->name, buf->buf + pos, 2);
1104
1105 return(ROFF_IGN);
1106 }
1107
1108 static enum rofferr
1109 roff_cond_sub(ROFF_ARGS)
1110 {
1111 enum rofft t;
1112 char *ep;
1113 int rr;
1114
1115 rr = r->last->rule;
1116 roffnode_cleanscope(r);
1117 t = roff_parse(r, buf->buf, &pos, ln, ppos);
1118
1119 /*
1120 * Fully handle known macros when they are structurally
1121 * required or when the conditional evaluated to true.
1122 */
1123
1124 if ((t != ROFF_MAX) &&
1125 (rr || roffs[t].flags & ROFFMAC_STRUCT)) {
1126 assert(roffs[t].proc);
1127 return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1128 }
1129
1130 /*
1131 * If `\}' occurs on a macro line without a preceding macro,
1132 * drop the line completely.
1133 */
1134
1135 ep = buf->buf + pos;
1136 if (ep[0] == '\\' && ep[1] == '}')
1137 rr = 0;
1138
1139 /* Always check for the closing delimiter `\}'. */
1140
1141 while ((ep = strchr(ep, '\\')) != NULL) {
1142 if (*(++ep) == '}') {
1143 *ep = '&';
1144 roff_ccond(r, ln, ep - buf->buf - 1);
1145 }
1146 ++ep;
1147 }
1148 return(rr ? ROFF_CONT : ROFF_IGN);
1149 }
1150
1151 static enum rofferr
1152 roff_cond_text(ROFF_ARGS)
1153 {
1154 char *ep;
1155 int rr;
1156
1157 rr = r->last->rule;
1158 roffnode_cleanscope(r);
1159
1160 ep = buf->buf + pos;
1161 while ((ep = strchr(ep, '\\')) != NULL) {
1162 if (*(++ep) == '}') {
1163 *ep = '&';
1164 roff_ccond(r, ln, ep - buf->buf - 1);
1165 }
1166 ++ep;
1167 }
1168 return(rr ? ROFF_CONT : ROFF_IGN);
1169 }
1170
1171 /*
1172 * Parse a single signed integer number. Stop at the first non-digit.
1173 * If there is at least one digit, return success and advance the
1174 * parse point, else return failure and let the parse point unchanged.
1175 * Ignore overflows, treat them just like the C language.
1176 */
1177 static int
1178 roff_getnum(const char *v, int *pos, int *res)
1179 {
1180 int myres, n, p;
1181
1182 if (NULL == res)
1183 res = &myres;
1184
1185 p = *pos;
1186 n = v[p] == '-';
1187 if (n)
1188 p++;
1189
1190 for (*res = 0; isdigit((unsigned char)v[p]); p++)
1191 *res = 10 * *res + v[p] - '0';
1192 if (p == *pos + n)
1193 return 0;
1194
1195 if (n)
1196 *res = -*res;
1197
1198 *pos = p;
1199 return 1;
1200 }
1201
1202 /*
1203 * Evaluate a string comparison condition.
1204 * The first character is the delimiter.
1205 * Succeed if the string up to its second occurrence
1206 * matches the string up to its third occurence.
1207 * Advance the cursor after the third occurrence
1208 * or lacking that, to the end of the line.
1209 */
1210 static int
1211 roff_evalstrcond(const char *v, int *pos)
1212 {
1213 const char *s1, *s2, *s3;
1214 int match;
1215
1216 match = 0;
1217 s1 = v + *pos; /* initial delimiter */
1218 s2 = s1 + 1; /* for scanning the first string */
1219 s3 = strchr(s2, *s1); /* for scanning the second string */
1220
1221 if (NULL == s3) /* found no middle delimiter */
1222 goto out;
1223
1224 while ('\0' != *++s3) {
1225 if (*s2 != *s3) { /* mismatch */
1226 s3 = strchr(s3, *s1);
1227 break;
1228 }
1229 if (*s3 == *s1) { /* found the final delimiter */
1230 match = 1;
1231 break;
1232 }
1233 s2++;
1234 }
1235
1236 out:
1237 if (NULL == s3)
1238 s3 = strchr(s2, '\0');
1239 else
1240 s3++;
1241 *pos = s3 - v;
1242 return(match);
1243 }
1244
1245 /*
1246 * Evaluate an optionally negated single character, numerical,
1247 * or string condition.
1248 */
1249 static int
1250 roff_evalcond(struct roff *r, int ln, const char *v, int *pos)
1251 {
1252 int wanttrue, number;
1253
1254 if ('!' == v[*pos]) {
1255 wanttrue = 0;
1256 (*pos)++;
1257 } else
1258 wanttrue = 1;
1259
1260 switch (v[*pos]) {
1261 case 'n':
1262 /* FALLTHROUGH */
1263 case 'o':
1264 (*pos)++;
1265 return(wanttrue);
1266 case 'c':
1267 /* FALLTHROUGH */
1268 case 'd':
1269 /* FALLTHROUGH */
1270 case 'e':
1271 /* FALLTHROUGH */
1272 case 'r':
1273 /* FALLTHROUGH */
1274 case 't':
1275 (*pos)++;
1276 return(!wanttrue);
1277 default:
1278 break;
1279 }
1280
1281 if (roff_evalnum(r, ln, v, pos, &number, 0))
1282 return((number > 0) == wanttrue);
1283 else
1284 return(roff_evalstrcond(v, pos) == wanttrue);
1285 }
1286
1287 static enum rofferr
1288 roff_line_ignore(ROFF_ARGS)
1289 {
1290
1291 return(ROFF_IGN);
1292 }
1293
1294 static enum rofferr
1295 roff_cond(ROFF_ARGS)
1296 {
1297
1298 roffnode_push(r, tok, NULL, ln, ppos);
1299
1300 /*
1301 * An `.el' has no conditional body: it will consume the value
1302 * of the current rstack entry set in prior `ie' calls or
1303 * defaults to DENY.
1304 *
1305 * If we're not an `el', however, then evaluate the conditional.
1306 */
1307
1308 r->last->rule = tok == ROFF_el ?
1309 (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1310 roff_evalcond(r, ln, buf->buf, &pos);
1311
1312 /*
1313 * An if-else will put the NEGATION of the current evaluated
1314 * conditional into the stack of rules.
1315 */
1316
1317 if (tok == ROFF_ie) {
1318 if (r->rstackpos + 1 == r->rstacksz) {
1319 r->rstacksz += 16;
1320 r->rstack = mandoc_reallocarray(r->rstack,
1321 r->rstacksz, sizeof(int));
1322 }
1323 r->rstack[++r->rstackpos] = !r->last->rule;
1324 }
1325
1326 /* If the parent has false as its rule, then so do we. */
1327
1328 if (r->last->parent && !r->last->parent->rule)
1329 r->last->rule = 0;
1330
1331 /*
1332 * Determine scope.
1333 * If there is nothing on the line after the conditional,
1334 * not even whitespace, use next-line scope.
1335 */
1336
1337 if (buf->buf[pos] == '\0') {
1338 r->last->endspan = 2;
1339 goto out;
1340 }
1341
1342 while (buf->buf[pos] == ' ')
1343 pos++;
1344
1345 /* An opening brace requests multiline scope. */
1346
1347 if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1348 r->last->endspan = -1;
1349 pos += 2;
1350 goto out;
1351 }
1352
1353 /*
1354 * Anything else following the conditional causes
1355 * single-line scope. Warn if the scope contains
1356 * nothing but trailing whitespace.
1357 */
1358
1359 if (buf->buf[pos] == '\0')
1360 mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1361 ln, ppos, roffs[tok].name);
1362
1363 r->last->endspan = 1;
1364
1365 out:
1366 *offs = pos;
1367 return(ROFF_RERUN);
1368 }
1369
1370 static enum rofferr
1371 roff_ds(ROFF_ARGS)
1372 {
1373 char *string;
1374 const char *name;
1375 size_t namesz;
1376
1377 /*
1378 * The first word is the name of the string.
1379 * If it is empty or terminated by an escape sequence,
1380 * abort the `ds' request without defining anything.
1381 */
1382
1383 name = string = buf->buf + pos;
1384 if (*name == '\0')
1385 return(ROFF_IGN);
1386
1387 namesz = roff_getname(r, &string, ln, pos);
1388 if (name[namesz] == '\\')
1389 return(ROFF_IGN);
1390
1391 /* Read past the initial double-quote, if any. */
1392 if (*string == '"')
1393 string++;
1394
1395 /* The rest is the value. */
1396 roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1397 ROFF_as == tok);
1398 return(ROFF_IGN);
1399 }
1400
1401 /*
1402 * Parse a single operator, one or two characters long.
1403 * If the operator is recognized, return success and advance the
1404 * parse point, else return failure and let the parse point unchanged.
1405 */
1406 static int
1407 roff_getop(const char *v, int *pos, char *res)
1408 {
1409
1410 *res = v[*pos];
1411
1412 switch (*res) {
1413 case '+':
1414 /* FALLTHROUGH */
1415 case '-':
1416 /* FALLTHROUGH */
1417 case '*':
1418 /* FALLTHROUGH */
1419 case '/':
1420 /* FALLTHROUGH */
1421 case '%':
1422 /* FALLTHROUGH */
1423 case '&':
1424 /* FALLTHROUGH */
1425 case ':':
1426 break;
1427 case '<':
1428 switch (v[*pos + 1]) {
1429 case '=':
1430 *res = 'l';
1431 (*pos)++;
1432 break;
1433 case '>':
1434 *res = '!';
1435 (*pos)++;
1436 break;
1437 case '?':
1438 *res = 'i';
1439 (*pos)++;
1440 break;
1441 default:
1442 break;
1443 }
1444 break;
1445 case '>':
1446 switch (v[*pos + 1]) {
1447 case '=':
1448 *res = 'g';
1449 (*pos)++;
1450 break;
1451 case '?':
1452 *res = 'a';
1453 (*pos)++;
1454 break;
1455 default:
1456 break;
1457 }
1458 break;
1459 case '=':
1460 if ('=' == v[*pos + 1])
1461 (*pos)++;
1462 break;
1463 default:
1464 return(0);
1465 }
1466 (*pos)++;
1467
1468 return(*res);
1469 }
1470
1471 /*
1472 * Evaluate either a parenthesized numeric expression
1473 * or a single signed integer number.
1474 */
1475 static int
1476 roff_evalpar(struct roff *r, int ln,
1477 const char *v, int *pos, int *res)
1478 {
1479
1480 if ('(' != v[*pos])
1481 return(roff_getnum(v, pos, res));
1482
1483 (*pos)++;
1484 if ( ! roff_evalnum(r, ln, v, pos, res, 1))
1485 return(0);
1486
1487 /*
1488 * Omission of the closing parenthesis
1489 * is an error in validation mode,
1490 * but ignored in evaluation mode.
1491 */
1492
1493 if (')' == v[*pos])
1494 (*pos)++;
1495 else if (NULL == res)
1496 return(0);
1497
1498 return(1);
1499 }
1500
1501 /*
1502 * Evaluate a complete numeric expression.
1503 * Proceed left to right, there is no concept of precedence.
1504 */
1505 static int
1506 roff_evalnum(struct roff *r, int ln, const char *v,
1507 int *pos, int *res, int skipwhite)
1508 {
1509 int mypos, operand2;
1510 char operator;
1511
1512 if (NULL == pos) {
1513 mypos = 0;
1514 pos = &mypos;
1515 }
1516
1517 if (skipwhite)
1518 while (isspace((unsigned char)v[*pos]))
1519 (*pos)++;
1520
1521 if ( ! roff_evalpar(r, ln, v, pos, res))
1522 return(0);
1523
1524 while (1) {
1525 if (skipwhite)
1526 while (isspace((unsigned char)v[*pos]))
1527 (*pos)++;
1528
1529 if ( ! roff_getop(v, pos, &operator))
1530 break;
1531
1532 if (skipwhite)
1533 while (isspace((unsigned char)v[*pos]))
1534 (*pos)++;
1535
1536 if ( ! roff_evalpar(r, ln, v, pos, &operand2))
1537 return(0);
1538
1539 if (skipwhite)
1540 while (isspace((unsigned char)v[*pos]))
1541 (*pos)++;
1542
1543 if (NULL == res)
1544 continue;
1545
1546 switch (operator) {
1547 case '+':
1548 *res += operand2;
1549 break;
1550 case '-':
1551 *res -= operand2;
1552 break;
1553 case '*':
1554 *res *= operand2;
1555 break;
1556 case '/':
1557 if (0 == operand2) {
1558 mandoc_msg(MANDOCERR_DIVZERO,
1559 r->parse, ln, *pos, v);
1560 *res = 0;
1561 break;
1562 }
1563 *res /= operand2;
1564 break;
1565 case '%':
1566 *res %= operand2;
1567 break;
1568 case '<':
1569 *res = *res < operand2;
1570 break;
1571 case '>':
1572 *res = *res > operand2;
1573 break;
1574 case 'l':
1575 *res = *res <= operand2;
1576 break;
1577 case 'g':
1578 *res = *res >= operand2;
1579 break;
1580 case '=':
1581 *res = *res == operand2;
1582 break;
1583 case '!':
1584 *res = *res != operand2;
1585 break;
1586 case '&':
1587 *res = *res && operand2;
1588 break;
1589 case ':':
1590 *res = *res || operand2;
1591 break;
1592 case 'i':
1593 if (operand2 < *res)
1594 *res = operand2;
1595 break;
1596 case 'a':
1597 if (operand2 > *res)
1598 *res = operand2;
1599 break;
1600 default:
1601 abort();
1602 }
1603 }
1604 return(1);
1605 }
1606
1607 void
1608 roff_setreg(struct roff *r, const char *name, int val, char sign)
1609 {
1610 struct roffreg *reg;
1611
1612 /* Search for an existing register with the same name. */
1613 reg = r->regtab;
1614
1615 while (reg && strcmp(name, reg->key.p))
1616 reg = reg->next;
1617
1618 if (NULL == reg) {
1619 /* Create a new register. */
1620 reg = mandoc_malloc(sizeof(struct roffreg));
1621 reg->key.p = mandoc_strdup(name);
1622 reg->key.sz = strlen(name);
1623 reg->val = 0;
1624 reg->next = r->regtab;
1625 r->regtab = reg;
1626 }
1627
1628 if ('+' == sign)
1629 reg->val += val;
1630 else if ('-' == sign)
1631 reg->val -= val;
1632 else
1633 reg->val = val;
1634 }
1635
1636 /*
1637 * Handle some predefined read-only number registers.
1638 * For now, return -1 if the requested register is not predefined;
1639 * in case a predefined read-only register having the value -1
1640 * were to turn up, another special value would have to be chosen.
1641 */
1642 static int
1643 roff_getregro(const char *name)
1644 {
1645
1646 switch (*name) {
1647 case 'A': /* ASCII approximation mode is always off. */
1648 return(0);
1649 case 'g': /* Groff compatibility mode is always on. */
1650 return(1);
1651 case 'H': /* Fixed horizontal resolution. */
1652 return (24);
1653 case 'j': /* Always adjust left margin only. */
1654 return(0);
1655 case 'T': /* Some output device is always defined. */
1656 return(1);
1657 case 'V': /* Fixed vertical resolution. */
1658 return (40);
1659 default:
1660 return (-1);
1661 }
1662 }
1663
1664 int
1665 roff_getreg(const struct roff *r, const char *name)
1666 {
1667 struct roffreg *reg;
1668 int val;
1669
1670 if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1671 val = roff_getregro(name + 1);
1672 if (-1 != val)
1673 return (val);
1674 }
1675
1676 for (reg = r->regtab; reg; reg = reg->next)
1677 if (0 == strcmp(name, reg->key.p))
1678 return(reg->val);
1679
1680 return(0);
1681 }
1682
1683 static int
1684 roff_getregn(const struct roff *r, const char *name, size_t len)
1685 {
1686 struct roffreg *reg;
1687 int val;
1688
1689 if ('.' == name[0] && 2 == len) {
1690 val = roff_getregro(name + 1);
1691 if (-1 != val)
1692 return (val);
1693 }
1694
1695 for (reg = r->regtab; reg; reg = reg->next)
1696 if (len == reg->key.sz &&
1697 0 == strncmp(name, reg->key.p, len))
1698 return(reg->val);
1699
1700 return(0);
1701 }
1702
1703 static void
1704 roff_freereg(struct roffreg *reg)
1705 {
1706 struct roffreg *old_reg;
1707
1708 while (NULL != reg) {
1709 free(reg->key.p);
1710 old_reg = reg;
1711 reg = reg->next;
1712 free(old_reg);
1713 }
1714 }
1715
1716 static enum rofferr
1717 roff_nr(ROFF_ARGS)
1718 {
1719 char *key, *val;
1720 size_t keysz;
1721 int iv;
1722 char sign;
1723
1724 key = val = buf->buf + pos;
1725 if (*key == '\0')
1726 return(ROFF_IGN);
1727
1728 keysz = roff_getname(r, &val, ln, pos);
1729 if (key[keysz] == '\\')
1730 return(ROFF_IGN);
1731 key[keysz] = '\0';
1732
1733 sign = *val;
1734 if (sign == '+' || sign == '-')
1735 val++;
1736
1737 if (roff_evalnum(r, ln, val, NULL, &iv, 0))
1738 roff_setreg(r, key, iv, sign);
1739
1740 return(ROFF_IGN);
1741 }
1742
1743 static enum rofferr
1744 roff_rr(ROFF_ARGS)
1745 {
1746 struct roffreg *reg, **prev;
1747 char *name, *cp;
1748 size_t namesz;
1749
1750 name = cp = buf->buf + pos;
1751 if (*name == '\0')
1752 return(ROFF_IGN);
1753 namesz = roff_getname(r, &cp, ln, pos);
1754 name[namesz] = '\0';
1755
1756 prev = &r->regtab;
1757 while (1) {
1758 reg = *prev;
1759 if (reg == NULL || !strcmp(name, reg->key.p))
1760 break;
1761 prev = &reg->next;
1762 }
1763 if (reg != NULL) {
1764 *prev = reg->next;
1765 free(reg->key.p);
1766 free(reg);
1767 }
1768 return(ROFF_IGN);
1769 }
1770
1771 static enum rofferr
1772 roff_rm(ROFF_ARGS)
1773 {
1774 const char *name;
1775 char *cp;
1776 size_t namesz;
1777
1778 cp = buf->buf + pos;
1779 while (*cp != '\0') {
1780 name = cp;
1781 namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1782 roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1783 if (name[namesz] == '\\')
1784 break;
1785 }
1786 return(ROFF_IGN);
1787 }
1788
1789 static enum rofferr
1790 roff_it(ROFF_ARGS)
1791 {
1792 char *cp;
1793 size_t len;
1794 int iv;
1795
1796 /* Parse the number of lines. */
1797 cp = buf->buf + pos;
1798 len = strcspn(cp, " \t");
1799 cp[len] = '\0';
1800 if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1801 mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1802 ln, ppos, buf->buf + 1);
1803 return(ROFF_IGN);
1804 }
1805 cp += len + 1;
1806
1807 /* Arm the input line trap. */
1808 roffit_lines = iv;
1809 roffit_macro = mandoc_strdup(cp);
1810 return(ROFF_IGN);
1811 }
1812
1813 static enum rofferr
1814 roff_Dd(ROFF_ARGS)
1815 {
1816 const char *const *cp;
1817
1818 if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1819 for (cp = __mdoc_reserved; *cp; cp++)
1820 roff_setstr(r, *cp, NULL, 0);
1821
1822 if (r->format == 0)
1823 r->format = MPARSE_MDOC;
1824
1825 return(ROFF_CONT);
1826 }
1827
1828 static enum rofferr
1829 roff_TH(ROFF_ARGS)
1830 {
1831 const char *const *cp;
1832
1833 if ((r->options & MPARSE_QUICK) == 0)
1834 for (cp = __man_reserved; *cp; cp++)
1835 roff_setstr(r, *cp, NULL, 0);
1836
1837 if (r->format == 0)
1838 r->format = MPARSE_MAN;
1839
1840 return(ROFF_CONT);
1841 }
1842
1843 static enum rofferr
1844 roff_TE(ROFF_ARGS)
1845 {
1846
1847 if (NULL == r->tbl)
1848 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1849 ln, ppos, "TE");
1850 else
1851 tbl_end(&r->tbl);
1852
1853 return(ROFF_IGN);
1854 }
1855
1856 static enum rofferr
1857 roff_T_(ROFF_ARGS)
1858 {
1859
1860 if (NULL == r->tbl)
1861 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1862 ln, ppos, "T&");
1863 else
1864 tbl_restart(ppos, ln, r->tbl);
1865
1866 return(ROFF_IGN);
1867 }
1868
1869 /*
1870 * Handle in-line equation delimiters.
1871 */
1872 static enum rofferr
1873 roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1874 {
1875 char *cp1, *cp2;
1876 const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1877
1878 /*
1879 * Outside equations, look for an opening delimiter.
1880 * If we are inside an equation, we already know it is
1881 * in-line, or this function wouldn't have been called;
1882 * so look for a closing delimiter.
1883 */
1884
1885 cp1 = buf->buf + pos;
1886 cp2 = strchr(cp1, r->eqn == NULL ?
1887 r->last_eqn->odelim : r->last_eqn->cdelim);
1888 if (cp2 == NULL)
1889 return(ROFF_CONT);
1890
1891 *cp2++ = '\0';
1892 bef_pr = bef_nl = aft_nl = aft_pr = "";
1893
1894 /* Handle preceding text, protecting whitespace. */
1895
1896 if (*buf->buf != '\0') {
1897 if (r->eqn == NULL)
1898 bef_pr = "\\&";
1899 bef_nl = "\n";
1900 }
1901
1902 /*
1903 * Prepare replacing the delimiter with an equation macro
1904 * and drop leading white space from the equation.
1905 */
1906
1907 if (r->eqn == NULL) {
1908 while (*cp2 == ' ')
1909 cp2++;
1910 mac = ".EQ";
1911 } else
1912 mac = ".EN";
1913
1914 /* Handle following text, protecting whitespace. */
1915
1916 if (*cp2 != '\0') {
1917 aft_nl = "\n";
1918 if (r->eqn != NULL)
1919 aft_pr = "\\&";
1920 }
1921
1922 /* Do the actual replacement. */
1923
1924 buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1925 bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1926 free(buf->buf);
1927 buf->buf = cp1;
1928
1929 /* Toggle the in-line state of the eqn subsystem. */
1930
1931 r->eqn_inline = r->eqn == NULL;
1932 return(ROFF_REPARSE);
1933 }
1934
1935 static enum rofferr
1936 roff_EQ(ROFF_ARGS)
1937 {
1938 struct eqn_node *e;
1939
1940 assert(r->eqn == NULL);
1941 e = eqn_alloc(ppos, ln, r->parse);
1942
1943 if (r->last_eqn) {
1944 r->last_eqn->next = e;
1945 e->delim = r->last_eqn->delim;
1946 e->odelim = r->last_eqn->odelim;
1947 e->cdelim = r->last_eqn->cdelim;
1948 } else
1949 r->first_eqn = r->last_eqn = e;
1950
1951 r->eqn = r->last_eqn = e;
1952
1953 if (buf->buf[pos] != '\0')
1954 mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1955 ".EQ %s", buf->buf + pos);
1956
1957 return(ROFF_IGN);
1958 }
1959
1960 static enum rofferr
1961 roff_EN(ROFF_ARGS)
1962 {
1963
1964 mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1965 return(ROFF_IGN);
1966 }
1967
1968 static enum rofferr
1969 roff_TS(ROFF_ARGS)
1970 {
1971 struct tbl_node *tbl;
1972
1973 if (r->tbl) {
1974 mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1975 ln, ppos, "TS breaks TS");
1976 tbl_end(&r->tbl);
1977 }
1978
1979 tbl = tbl_alloc(ppos, ln, r->parse);
1980
1981 if (r->last_tbl)
1982 r->last_tbl->next = tbl;
1983 else
1984 r->first_tbl = r->last_tbl = tbl;
1985
1986 r->tbl = r->last_tbl = tbl;
1987 return(ROFF_IGN);
1988 }
1989
1990 static enum rofferr
1991 roff_cc(ROFF_ARGS)
1992 {
1993 const char *p;
1994
1995 p = buf->buf + pos;
1996
1997 if (*p == '\0' || (r->control = *p++) == '.')
1998 r->control = 0;
1999
2000 if (*p != '\0')
2001 mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2002
2003 return(ROFF_IGN);
2004 }
2005
2006 static enum rofferr
2007 roff_tr(ROFF_ARGS)
2008 {
2009 const char *p, *first, *second;
2010 size_t fsz, ssz;
2011 enum mandoc_esc esc;
2012
2013 p = buf->buf + pos;
2014
2015 if (*p == '\0') {
2016 mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2017 return(ROFF_IGN);
2018 }
2019
2020 while (*p != '\0') {
2021 fsz = ssz = 1;
2022
2023 first = p++;
2024 if (*first == '\\') {
2025 esc = mandoc_escape(&p, NULL, NULL);
2026 if (esc == ESCAPE_ERROR) {
2027 mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2028 ln, (int)(p - buf->buf), first);
2029 return(ROFF_IGN);
2030 }
2031 fsz = (size_t)(p - first);
2032 }
2033
2034 second = p++;
2035 if (*second == '\\') {
2036 esc = mandoc_escape(&p, NULL, NULL);
2037 if (esc == ESCAPE_ERROR) {
2038 mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2039 ln, (int)(p - buf->buf), second);
2040 return(ROFF_IGN);
2041 }
2042 ssz = (size_t)(p - second);
2043 } else if (*second == '\0') {
2044 mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
2045 ln, (int)(p - buf->buf), NULL);
2046 second = " ";
2047 p--;
2048 }
2049
2050 if (fsz > 1) {
2051 roff_setstrn(&r->xmbtab, first, fsz,
2052 second, ssz, 0);
2053 continue;
2054 }
2055
2056 if (r->xtab == NULL)
2057 r->xtab = mandoc_calloc(128,
2058 sizeof(struct roffstr));
2059
2060 free(r->xtab[(int)*first].p);
2061 r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
2062 r->xtab[(int)*first].sz = ssz;
2063 }
2064
2065 return(ROFF_IGN);
2066 }
2067
2068 static enum rofferr
2069 roff_so(ROFF_ARGS)
2070 {
2071 char *name;
2072
2073 name = buf->buf + pos;
2074 mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
2075
2076 /*
2077 * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2078 * opening anything that's not in our cwd or anything beneath
2079 * it. Thus, explicitly disallow traversing up the file-system
2080 * or using absolute paths.
2081 */
2082
2083 if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
2084 mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2085 ".so %s", name);
2086 return(ROFF_ERR);
2087 }
2088
2089 *offs = pos;
2090 return(ROFF_SO);
2091 }
2092
2093 static enum rofferr
2094 roff_userdef(ROFF_ARGS)
2095 {
2096 const char *arg[9];
2097 char *cp, *n1, *n2;
2098 int i;
2099
2100 /*
2101 * Collect pointers to macro argument strings
2102 * and NUL-terminate them.
2103 */
2104 cp = buf->buf + pos;
2105 for (i = 0; i < 9; i++)
2106 arg[i] = *cp == '\0' ? "" :
2107 mandoc_getarg(r->parse, &cp, ln, &pos);
2108
2109 /*
2110 * Expand macro arguments.
2111 */
2112 buf->sz = 0;
2113 n1 = cp = mandoc_strdup(r->current_string);
2114 while ((cp = strstr(cp, "\\$")) != NULL) {
2115 i = cp[2] - '1';
2116 if (0 > i || 8 < i) {
2117 /* Not an argument invocation. */
2118 cp += 2;
2119 continue;
2120 }
2121 *cp = '\0';
2122 buf->sz = mandoc_asprintf(&n2, "%s%s%s",
2123 n1, arg[i], cp + 3) + 1;
2124 cp = n2 + (cp - n1);
2125 free(n1);
2126 n1 = n2;
2127 }
2128
2129 /*
2130 * Replace the macro invocation
2131 * by the expanded macro.
2132 */
2133 free(buf->buf);
2134 buf->buf = n1;
2135 if (buf->sz == 0)
2136 buf->sz = strlen(buf->buf) + 1;
2137
2138 return(buf->sz > 1 && buf->buf[buf->sz - 2] == '\n' ?
2139 ROFF_REPARSE : ROFF_APPEND);
2140 }
2141
2142 static size_t
2143 roff_getname(struct roff *r, char **cpp, int ln, int pos)
2144 {
2145 char *name, *cp;
2146 size_t namesz;
2147
2148 name = *cpp;
2149 if ('\0' == *name)
2150 return(0);
2151
2152 /* Read until end of name and terminate it with NUL. */
2153 for (cp = name; 1; cp++) {
2154 if ('\0' == *cp || ' ' == *cp) {
2155 namesz = cp - name;
2156 break;
2157 }
2158 if ('\\' != *cp)
2159 continue;
2160 namesz = cp - name;
2161 if ('{' == cp[1] || '}' == cp[1])
2162 break;
2163 cp++;
2164 if ('\\' == *cp)
2165 continue;
2166 mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2167 "%.*s", (int)(cp - name + 1), name);
2168 mandoc_escape((const char **)&cp, NULL, NULL);
2169 break;
2170 }
2171
2172 /* Read past spaces. */
2173 while (' ' == *cp)
2174 cp++;
2175
2176 *cpp = cp;
2177 return(namesz);
2178 }
2179
2180 /*
2181 * Store *string into the user-defined string called *name.
2182 * To clear an existing entry, call with (*r, *name, NULL, 0).
2183 * append == 0: replace mode
2184 * append == 1: single-line append mode
2185 * append == 2: multiline append mode, append '\n' after each call
2186 */
2187 static void
2188 roff_setstr(struct roff *r, const char *name, const char *string,
2189 int append)
2190 {
2191
2192 roff_setstrn(&r->strtab, name, strlen(name), string,
2193 string ? strlen(string) : 0, append);
2194 }
2195
2196 static void
2197 roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
2198 const char *string, size_t stringsz, int append)
2199 {
2200 struct roffkv *n;
2201 char *c;
2202 int i;
2203 size_t oldch, newch;
2204
2205 /* Search for an existing string with the same name. */
2206 n = *r;
2207
2208 while (n && (namesz != n->key.sz ||
2209 strncmp(n->key.p, name, namesz)))
2210 n = n->next;
2211
2212 if (NULL == n) {
2213 /* Create a new string table entry. */
2214 n = mandoc_malloc(sizeof(struct roffkv));
2215 n->key.p = mandoc_strndup(name, namesz);
2216 n->key.sz = namesz;
2217 n->val.p = NULL;
2218 n->val.sz = 0;
2219 n->next = *r;
2220 *r = n;
2221 } else if (0 == append) {
2222 free(n->val.p);
2223 n->val.p = NULL;
2224 n->val.sz = 0;
2225 }
2226
2227 if (NULL == string)
2228 return;
2229
2230 /*
2231 * One additional byte for the '\n' in multiline mode,
2232 * and one for the terminating '\0'.
2233 */
2234 newch = stringsz + (1 < append ? 2u : 1u);
2235
2236 if (NULL == n->val.p) {
2237 n->val.p = mandoc_malloc(newch);
2238 *n->val.p = '\0';
2239 oldch = 0;
2240 } else {
2241 oldch = n->val.sz;
2242 n->val.p = mandoc_realloc(n->val.p, oldch + newch);
2243 }
2244
2245 /* Skip existing content in the destination buffer. */
2246 c = n->val.p + (int)oldch;
2247
2248 /* Append new content to the destination buffer. */
2249 i = 0;
2250 while (i < (int)stringsz) {
2251 /*
2252 * Rudimentary roff copy mode:
2253 * Handle escaped backslashes.
2254 */
2255 if ('\\' == string[i] && '\\' == string[i + 1])
2256 i++;
2257 *c++ = string[i++];
2258 }
2259
2260 /* Append terminating bytes. */
2261 if (1 < append)
2262 *c++ = '\n';
2263
2264 *c = '\0';
2265 n->val.sz = (int)(c - n->val.p);
2266 }
2267
2268 static const char *
2269 roff_getstrn(const struct roff *r, const char *name, size_t len)
2270 {
2271 const struct roffkv *n;
2272 int i;
2273
2274 for (n = r->strtab; n; n = n->next)
2275 if (0 == strncmp(name, n->key.p, len) &&
2276 '\0' == n->key.p[(int)len])
2277 return(n->val.p);
2278
2279 for (i = 0; i < PREDEFS_MAX; i++)
2280 if (0 == strncmp(name, predefs[i].name, len) &&
2281 '\0' == predefs[i].name[(int)len])
2282 return(predefs[i].str);
2283
2284 return(NULL);
2285 }
2286
2287 static void
2288 roff_freestr(struct roffkv *r)
2289 {
2290 struct roffkv *n, *nn;
2291
2292 for (n = r; n; n = nn) {
2293 free(n->key.p);
2294 free(n->val.p);
2295 nn = n->next;
2296 free(n);
2297 }
2298 }
2299
2300 const struct tbl_span *
2301 roff_span(const struct roff *r)
2302 {
2303
2304 return(r->tbl ? tbl_span(r->tbl) : NULL);
2305 }
2306
2307 const struct eqn *
2308 roff_eqn(const struct roff *r)
2309 {
2310
2311 return(r->last_eqn ? &r->last_eqn->eqn : NULL);
2312 }
2313
2314 /*
2315 * Duplicate an input string, making the appropriate character
2316 * conversations (as stipulated by `tr') along the way.
2317 * Returns a heap-allocated string with all the replacements made.
2318 */
2319 char *
2320 roff_strdup(const struct roff *r, const char *p)
2321 {
2322 const struct roffkv *cp;
2323 char *res;
2324 const char *pp;
2325 size_t ssz, sz;
2326 enum mandoc_esc esc;
2327
2328 if (NULL == r->xmbtab && NULL == r->xtab)
2329 return(mandoc_strdup(p));
2330 else if ('\0' == *p)
2331 return(mandoc_strdup(""));
2332
2333 /*
2334 * Step through each character looking for term matches
2335 * (remember that a `tr' can be invoked with an escape, which is
2336 * a glyph but the escape is multi-character).
2337 * We only do this if the character hash has been initialised
2338 * and the string is >0 length.
2339 */
2340
2341 res = NULL;
2342 ssz = 0;
2343
2344 while ('\0' != *p) {
2345 if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2346 sz = r->xtab[(int)*p].sz;
2347 res = mandoc_realloc(res, ssz + sz + 1);
2348 memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2349 ssz += sz;
2350 p++;
2351 continue;
2352 } else if ('\\' != *p) {
2353 res = mandoc_realloc(res, ssz + 2);
2354 res[ssz++] = *p++;
2355 continue;
2356 }
2357
2358 /* Search for term matches. */
2359 for (cp = r->xmbtab; cp; cp = cp->next)
2360 if (0 == strncmp(p, cp->key.p, cp->key.sz))
2361 break;
2362
2363 if (NULL != cp) {
2364 /*
2365 * A match has been found.
2366 * Append the match to the array and move
2367 * forward by its keysize.
2368 */
2369 res = mandoc_realloc(res,
2370 ssz + cp->val.sz + 1);
2371 memcpy(res + ssz, cp->val.p, cp->val.sz);
2372 ssz += cp->val.sz;
2373 p += (int)cp->key.sz;
2374 continue;
2375 }
2376
2377 /*
2378 * Handle escapes carefully: we need to copy
2379 * over just the escape itself, or else we might
2380 * do replacements within the escape itself.
2381 * Make sure to pass along the bogus string.
2382 */
2383 pp = p++;
2384 esc = mandoc_escape(&p, NULL, NULL);
2385 if (ESCAPE_ERROR == esc) {
2386 sz = strlen(pp);
2387 res = mandoc_realloc(res, ssz + sz + 1);
2388 memcpy(res + ssz, pp, sz);
2389 break;
2390 }
2391 /*
2392 * We bail out on bad escapes.
2393 * No need to warn: we already did so when
2394 * roff_res() was called.
2395 */
2396 sz = (int)(p - pp);
2397 res = mandoc_realloc(res, ssz + sz + 1);
2398 memcpy(res + ssz, pp, sz);
2399 ssz += sz;
2400 }
2401
2402 res[(int)ssz] = '\0';
2403 return(res);
2404 }
2405
2406 int
2407 roff_getformat(const struct roff *r)
2408 {
2409
2410 return(r->format);
2411 }
2412
2413 /*
2414 * Find out whether a line is a macro line or not.
2415 * If it is, adjust the current position and return one; if it isn't,
2416 * return zero and don't change the current position.
2417 * If the control character has been set with `.cc', then let that grain
2418 * precedence.
2419 * This is slighly contrary to groff, where using the non-breaking
2420 * control character when `cc' has been invoked will cause the
2421 * non-breaking macro contents to be printed verbatim.
2422 */
2423 int
2424 roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2425 {
2426 int pos;
2427
2428 pos = *ppos;
2429
2430 if (0 != r->control && cp[pos] == r->control)
2431 pos++;
2432 else if (0 != r->control)
2433 return(0);
2434 else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2435 pos += 2;
2436 else if ('.' == cp[pos] || '\'' == cp[pos])
2437 pos++;
2438 else
2439 return(0);
2440
2441 while (' ' == cp[pos] || '\t' == cp[pos])
2442 pos++;
2443
2444 *ppos = pos;
2445 return(1);
2446 }
A git conversion mirror of mandoc.bsd.lv