]> git.cameronkatri.com Git - mandoc.git/blob - mandoc.c
Have mandoc-db accumulate manual page descriptions (`Nd' in -mdoc parlance)
[mandoc.git] / mandoc.c
1 /* $Id: mandoc.c,v 1.46 2011/04/09 15:35:30 kristaps Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 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 <sys/types.h>
23
24 #include <assert.h>
25 #include <ctype.h>
26 #include <stdlib.h>
27 #include <stdio.h>
28 #include <string.h>
29 #include <time.h>
30
31 #include "mandoc.h"
32 #include "libmandoc.h"
33
34 #define DATESIZE 32
35
36 static int a2time(time_t *, const char *, const char *);
37 static char *time2a(time_t);
38 static int numescape(const char *);
39
40 /*
41 * Pass over recursive numerical expressions. This context of this
42 * function is important: it's only called within character-terminating
43 * escapes (e.g., \s[xxxyyy]), so all we need to do is handle initial
44 * recursion: we don't care about what's in these blocks.
45 * This returns the number of characters skipped or -1 if an error
46 * occurs (the caller should bail).
47 */
48 static int
49 numescape(const char *start)
50 {
51 int i;
52 size_t sz;
53 const char *cp;
54
55 i = 0;
56
57 /* The expression consists of a subexpression. */
58
59 if ('\\' == start[i]) {
60 cp = &start[++i];
61 /*
62 * Read past the end of the subexpression.
63 * Bail immediately on errors.
64 */
65 if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
66 return(-1);
67 return(i + cp - &start[i]);
68 }
69
70 if ('(' != start[i++])
71 return(0);
72
73 /*
74 * A parenthesised subexpression. Read until the closing
75 * parenthesis, making sure to handle any nested subexpressions
76 * that might ruin our parse.
77 */
78
79 while (')' != start[i]) {
80 sz = strcspn(&start[i], ")\\");
81 i += (int)sz;
82
83 if ('\0' == start[i])
84 return(-1);
85 else if ('\\' != start[i])
86 continue;
87
88 cp = &start[++i];
89 if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
90 return(-1);
91 i += cp - &start[i];
92 }
93
94 /* Read past the terminating ')'. */
95 return(++i);
96 }
97
98 /*
99 * Handle an escaped sequeence. This should be called with any
100 * string subsequent a `\'. Pass a pointer to this substring as "end";
101 * it will be set to the supremum of the parsed escape sequence. If
102 * this returns ESCAPE_ERROR, the string is bogus and should be thrown
103 * away. If not ESCAPE_ERROR or ESCAPE_IGNORE, "start" is set to the
104 * first relevant character of the substring (font, glyph, whatever) of
105 * length sz. Both "start" and "sz" may be NULL.
106 */
107 enum mandoc_esc
108 mandoc_escape(const char **end, const char **start, int *sz)
109 {
110 char c, term, numeric;
111 int i, lim, ssz, rlim;
112 const char *cp, *rstart;
113 enum mandoc_esc gly;
114
115 cp = *end;
116 rstart = cp;
117 if (start)
118 *start = rstart;
119 i = lim = 0;
120 gly = ESCAPE_ERROR;
121 term = numeric = '\0';
122
123 switch ((c = cp[i++])) {
124 /*
125 * First the glyphs. There are several different forms of
126 * these, but each eventually returns a substring of the glyph
127 * name.
128 */
129 case ('('):
130 gly = ESCAPE_SPECIAL;
131 lim = 2;
132 break;
133 case ('['):
134 gly = ESCAPE_SPECIAL;
135 term = ']';
136 break;
137 case ('C'):
138 if ('\'' != cp[i])
139 return(ESCAPE_ERROR);
140 gly = ESCAPE_SPECIAL;
141 term = '\'';
142 break;
143
144 /*
145 * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
146 * 'X' is the trigger. These have opaque sub-strings.
147 */
148 case ('F'):
149 /* FALLTHROUGH */
150 case ('g'):
151 /* FALLTHROUGH */
152 case ('k'):
153 /* FALLTHROUGH */
154 case ('M'):
155 /* FALLTHROUGH */
156 case ('m'):
157 /* FALLTHROUGH */
158 case ('n'):
159 /* FALLTHROUGH */
160 case ('V'):
161 /* FALLTHROUGH */
162 case ('Y'):
163 if (ESCAPE_ERROR == gly)
164 gly = ESCAPE_IGNORE;
165 /* FALLTHROUGH */
166 case ('*'):
167 if (ESCAPE_ERROR == gly)
168 gly = ESCAPE_PREDEF;
169 /* FALLTHROUGH */
170 case ('f'):
171 if (ESCAPE_ERROR == gly)
172 gly = ESCAPE_FONT;
173
174 rstart= &cp[i];
175 if (start)
176 *start = rstart;
177
178 switch (cp[i++]) {
179 case ('('):
180 lim = 2;
181 break;
182 case ('['):
183 term = ']';
184 break;
185 default:
186 lim = 1;
187 i--;
188 break;
189 }
190 break;
191
192 /*
193 * These escapes are of the form \X'Y', where 'X' is the trigger
194 * and 'Y' is any string. These have opaque sub-strings.
195 */
196 case ('A'):
197 /* FALLTHROUGH */
198 case ('b'):
199 /* FALLTHROUGH */
200 case ('D'):
201 /* FALLTHROUGH */
202 case ('o'):
203 /* FALLTHROUGH */
204 case ('R'):
205 /* FALLTHROUGH */
206 case ('X'):
207 /* FALLTHROUGH */
208 case ('Z'):
209 if ('\'' != cp[i++])
210 return(ESCAPE_ERROR);
211 gly = ESCAPE_IGNORE;
212 term = '\'';
213 break;
214
215 /*
216 * These escapes are of the form \X'N', where 'X' is the trigger
217 * and 'N' resolves to a numerical expression.
218 */
219 case ('B'):
220 /* FALLTHROUGH */
221 case ('h'):
222 /* FALLTHROUGH */
223 case ('H'):
224 /* FALLTHROUGH */
225 case ('L'):
226 /* FALLTHROUGH */
227 case ('l'):
228 /* FALLTHROUGH */
229 case ('N'):
230 if (ESCAPE_ERROR == gly)
231 gly = ESCAPE_NUMBERED;
232 /* FALLTHROUGH */
233 case ('S'):
234 /* FALLTHROUGH */
235 case ('v'):
236 /* FALLTHROUGH */
237 case ('w'):
238 /* FALLTHROUGH */
239 case ('x'):
240 if (ESCAPE_ERROR == gly)
241 gly = ESCAPE_IGNORE;
242 if ('\'' != cp[i++])
243 return(ESCAPE_ERROR);
244 term = numeric = '\'';
245 break;
246
247 /*
248 * Sizes get a special category of their own.
249 */
250 case ('s'):
251 gly = ESCAPE_IGNORE;
252
253 rstart = &cp[i];
254 if (start)
255 *start = rstart;
256
257 /* See +/- counts as a sign. */
258 c = cp[i];
259 if ('+' == c || '-' == c || ASCII_HYPH == c)
260 ++i;
261
262 switch (cp[i++]) {
263 case ('('):
264 lim = 2;
265 break;
266 case ('['):
267 term = numeric = ']';
268 break;
269 case ('\''):
270 term = numeric = '\'';
271 break;
272 default:
273 lim = 1;
274 i--;
275 break;
276 }
277
278 /* See +/- counts as a sign. */
279 c = cp[i];
280 if ('+' == c || '-' == c || ASCII_HYPH == c)
281 ++i;
282
283 break;
284
285 /*
286 * Anything else is assumed to be a glyph.
287 */
288 default:
289 gly = ESCAPE_SPECIAL;
290 lim = 1;
291 i--;
292 break;
293 }
294
295 assert(ESCAPE_ERROR != gly);
296
297 rstart = &cp[i];
298 if (start)
299 *start = rstart;
300
301 /*
302 * If a terminating block has been specified, we need to
303 * handle the case of recursion, which could have their
304 * own terminating blocks that mess up our parse. This, by the
305 * way, means that the "start" and "size" values will be
306 * effectively meaningless.
307 */
308
309 ssz = 0;
310 if (numeric && -1 == (ssz = numescape(&cp[i])))
311 return(ESCAPE_ERROR);
312
313 i += ssz;
314 rlim = -1;
315
316 /*
317 * We have a character terminator. Try to read up to that
318 * character. If we can't (i.e., we hit the nil), then return
319 * an error; if we can, calculate our length, read past the
320 * terminating character, and exit.
321 */
322
323 if ('\0' != term) {
324 *end = strchr(&cp[i], term);
325 if ('\0' == *end)
326 return(ESCAPE_ERROR);
327
328 rlim = *end - &cp[i];
329 if (sz)
330 *sz = rlim;
331 (*end)++;
332 goto out;
333 }
334
335 assert(lim > 0);
336
337 /*
338 * We have a numeric limit. If the string is shorter than that,
339 * stop and return an error. Else adjust our endpoint, length,
340 * and return the current glyph.
341 */
342
343 if ((size_t)lim > strlen(&cp[i]))
344 return(ESCAPE_ERROR);
345
346 rlim = lim;
347 if (sz)
348 *sz = rlim;
349
350 *end = &cp[i] + lim;
351
352 out:
353 assert(rlim >= 0 && rstart);
354
355 /* Run post-processors. */
356
357 switch (gly) {
358 case (ESCAPE_FONT):
359 if (1 != rlim)
360 break;
361 switch (*rstart) {
362 case ('3'):
363 /* FALLTHROUGH */
364 case ('B'):
365 gly = ESCAPE_FONTBOLD;
366 break;
367 case ('2'):
368 /* FALLTHROUGH */
369 case ('I'):
370 gly = ESCAPE_FONTITALIC;
371 break;
372 case ('P'):
373 gly = ESCAPE_FONTPREV;
374 break;
375 case ('1'):
376 /* FALLTHROUGH */
377 case ('R'):
378 gly = ESCAPE_FONTROMAN;
379 break;
380 }
381 break;
382 case (ESCAPE_SPECIAL):
383 if (1 != rlim)
384 break;
385 if ('c' == *rstart)
386 gly = ESCAPE_NOSPACE;
387 break;
388 default:
389 break;
390 }
391
392 return(gly);
393 }
394
395 void *
396 mandoc_calloc(size_t num, size_t size)
397 {
398 void *ptr;
399
400 ptr = calloc(num, size);
401 if (NULL == ptr) {
402 perror(NULL);
403 exit((int)MANDOCLEVEL_SYSERR);
404 }
405
406 return(ptr);
407 }
408
409
410 void *
411 mandoc_malloc(size_t size)
412 {
413 void *ptr;
414
415 ptr = malloc(size);
416 if (NULL == ptr) {
417 perror(NULL);
418 exit((int)MANDOCLEVEL_SYSERR);
419 }
420
421 return(ptr);
422 }
423
424
425 void *
426 mandoc_realloc(void *ptr, size_t size)
427 {
428
429 ptr = realloc(ptr, size);
430 if (NULL == ptr) {
431 perror(NULL);
432 exit((int)MANDOCLEVEL_SYSERR);
433 }
434
435 return(ptr);
436 }
437
438
439 char *
440 mandoc_strdup(const char *ptr)
441 {
442 char *p;
443
444 p = strdup(ptr);
445 if (NULL == p) {
446 perror(NULL);
447 exit((int)MANDOCLEVEL_SYSERR);
448 }
449
450 return(p);
451 }
452
453 /*
454 * Parse a quoted or unquoted roff-style request or macro argument.
455 * Return a pointer to the parsed argument, which is either the original
456 * pointer or advanced by one byte in case the argument is quoted.
457 * Null-terminate the argument in place.
458 * Collapse pairs of quotes inside quoted arguments.
459 * Advance the argument pointer to the next argument,
460 * or to the null byte terminating the argument line.
461 */
462 char *
463 mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
464 {
465 char *start, *cp;
466 int quoted, pairs, white;
467
468 /* Quoting can only start with a new word. */
469 start = *cpp;
470 if ('"' == *start) {
471 quoted = 1;
472 start++;
473 } else
474 quoted = 0;
475
476 pairs = 0;
477 white = 0;
478 for (cp = start; '\0' != *cp; cp++) {
479 /* Move left after quoted quotes and escaped backslashes. */
480 if (pairs)
481 cp[-pairs] = cp[0];
482 if ('\\' == cp[0]) {
483 if ('\\' == cp[1]) {
484 /* Poor man's copy mode. */
485 pairs++;
486 cp++;
487 } else if (0 == quoted && ' ' == cp[1])
488 /* Skip escaped blanks. */
489 cp++;
490 } else if (0 == quoted) {
491 if (' ' == cp[0]) {
492 /* Unescaped blanks end unquoted args. */
493 white = 1;
494 break;
495 }
496 } else if ('"' == cp[0]) {
497 if ('"' == cp[1]) {
498 /* Quoted quotes collapse. */
499 pairs++;
500 cp++;
501 } else {
502 /* Unquoted quotes end quoted args. */
503 quoted = 2;
504 break;
505 }
506 }
507 }
508
509 /* Quoted argument without a closing quote. */
510 if (1 == quoted)
511 mandoc_msg(MANDOCERR_BADQUOTE, parse, ln, *pos, NULL);
512
513 /* Null-terminate this argument and move to the next one. */
514 if (pairs)
515 cp[-pairs] = '\0';
516 if ('\0' != *cp) {
517 *cp++ = '\0';
518 while (' ' == *cp)
519 cp++;
520 }
521 *pos += (int)(cp - start) + (quoted ? 1 : 0);
522 *cpp = cp;
523
524 if ('\0' == *cp && (white || ' ' == cp[-1]))
525 mandoc_msg(MANDOCERR_EOLNSPACE, parse, ln, *pos, NULL);
526
527 return(start);
528 }
529
530 static int
531 a2time(time_t *t, const char *fmt, const char *p)
532 {
533 struct tm tm;
534 char *pp;
535
536 memset(&tm, 0, sizeof(struct tm));
537
538 pp = strptime(p, fmt, &tm);
539 if (NULL != pp && '\0' == *pp) {
540 *t = mktime(&tm);
541 return(1);
542 }
543
544 return(0);
545 }
546
547 static char *
548 time2a(time_t t)
549 {
550 struct tm tm;
551 char *buf, *p;
552 size_t ssz;
553 int isz;
554
555 localtime_r(&t, &tm);
556
557 /*
558 * Reserve space:
559 * up to 9 characters for the month (September) + blank
560 * up to 2 characters for the day + comma + blank
561 * 4 characters for the year and a terminating '\0'
562 */
563 p = buf = mandoc_malloc(10 + 4 + 4 + 1);
564
565 if (0 == (ssz = strftime(p, 10 + 1, "%B ", &tm)))
566 goto fail;
567 p += (int)ssz;
568
569 if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm.tm_mday)))
570 goto fail;
571 p += isz;
572
573 if (0 == strftime(p, 4 + 1, "%Y", &tm))
574 goto fail;
575 return(buf);
576
577 fail:
578 free(buf);
579 return(NULL);
580 }
581
582 char *
583 mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
584 {
585 char *out;
586 time_t t;
587
588 if (NULL == in || '\0' == *in ||
589 0 == strcmp(in, "$" "Mdocdate$")) {
590 mandoc_msg(MANDOCERR_NODATE, parse, ln, pos, NULL);
591 time(&t);
592 }
593 else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
594 !a2time(&t, "%b %d, %Y", in) &&
595 !a2time(&t, "%Y-%m-%d", in)) {
596 mandoc_msg(MANDOCERR_BADDATE, parse, ln, pos, NULL);
597 t = 0;
598 }
599 out = t ? time2a(t) : NULL;
600 return(out ? out : mandoc_strdup(in));
601 }
602
603 int
604 mandoc_eos(const char *p, size_t sz, int enclosed)
605 {
606 const char *q;
607 int found;
608
609 if (0 == sz)
610 return(0);
611
612 /*
613 * End-of-sentence recognition must include situations where
614 * some symbols, such as `)', allow prior EOS punctuation to
615 * propogate outward.
616 */
617
618 found = 0;
619 for (q = p + (int)sz - 1; q >= p; q--) {
620 switch (*q) {
621 case ('\"'):
622 /* FALLTHROUGH */
623 case ('\''):
624 /* FALLTHROUGH */
625 case (']'):
626 /* FALLTHROUGH */
627 case (')'):
628 if (0 == found)
629 enclosed = 1;
630 break;
631 case ('.'):
632 /* FALLTHROUGH */
633 case ('!'):
634 /* FALLTHROUGH */
635 case ('?'):
636 found = 1;
637 break;
638 default:
639 return(found && (!enclosed || isalnum((unsigned char)*q)));
640 }
641 }
642
643 return(found && !enclosed);
644 }
645
646 int
647 mandoc_hyph(const char *start, const char *c)
648 {
649
650 /*
651 * Choose whether to break at a hyphenated character. We only
652 * do this if it's free-standing within a word.
653 */
654
655 /* Skip first/last character of buffer. */
656 if (c == start || '\0' == *(c + 1))
657 return(0);
658 /* Skip first/last character of word. */
659 if ('\t' == *(c + 1) || '\t' == *(c - 1))
660 return(0);
661 if (' ' == *(c + 1) || ' ' == *(c - 1))
662 return(0);
663 /* Skip double invocations. */
664 if ('-' == *(c + 1) || '-' == *(c - 1))
665 return(0);
666 /* Skip escapes. */
667 if ('\\' == *(c - 1))
668 return(0);
669
670 return(1);
671 }
672
673 /*
674 * Find out whether a line is a macro line or not. If it is, adjust the
675 * current position and return one; if it isn't, return zero and don't
676 * change the current position.
677 */
678 int
679 mandoc_getcontrol(const char *cp, int *ppos)
680 {
681 int pos;
682
683 pos = *ppos;
684
685 if ('\\' == cp[pos] && '.' == cp[pos + 1])
686 pos += 2;
687 else if ('.' == cp[pos] || '\'' == cp[pos])
688 pos++;
689 else
690 return(0);
691
692 while (' ' == cp[pos] || '\t' == cp[pos])
693 pos++;
694
695 *ppos = pos;
696 return(1);
697 }