xref: /dragonfly/contrib/mdocml/mandoc.c (revision 0f63dfd0)
1 /*	$Id: mandoc.c,v 1.70 2013/11/10 21:34:04 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2011, 2012, 2013 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 <errno.h>
27 #include <limits.h>
28 #include <stdlib.h>
29 #include <stdio.h>
30 #include <string.h>
31 #include <time.h>
32 
33 #include "mandoc.h"
34 #include "libmandoc.h"
35 
36 #define DATESIZE 32
37 
38 static	int	 a2time(time_t *, const char *, const char *);
39 static	char	*time2a(time_t);
40 
41 
42 enum mandoc_esc
43 mandoc_escape(const char const **end, const char const **start, int *sz)
44 {
45 	const char	*local_start;
46 	int		 local_sz;
47 	char		 term;
48 	enum mandoc_esc	 gly;
49 
50 	/*
51 	 * When the caller doesn't provide return storage,
52 	 * use local storage.
53 	 */
54 
55 	if (NULL == start)
56 		start = &local_start;
57 	if (NULL == sz)
58 		sz = &local_sz;
59 
60 	/*
61 	 * Beyond the backslash, at least one input character
62 	 * is part of the escape sequence.  With one exception
63 	 * (see below), that character won't be returned.
64 	 */
65 
66 	gly = ESCAPE_ERROR;
67 	*start = ++*end;
68 	*sz = 0;
69 	term = '\0';
70 
71 	switch ((*start)[-1]) {
72 	/*
73 	 * First the glyphs.  There are several different forms of
74 	 * these, but each eventually returns a substring of the glyph
75 	 * name.
76 	 */
77 	case ('('):
78 		gly = ESCAPE_SPECIAL;
79 		*sz = 2;
80 		break;
81 	case ('['):
82 		gly = ESCAPE_SPECIAL;
83 		/*
84 		 * Unicode escapes are defined in groff as \[uXXXX] to
85 		 * \[u10FFFF], where the contained value must be a valid
86 		 * Unicode codepoint.  Here, however, only check whether
87 		 * it's not a zero-width escape.
88 		 */
89 		if ('u' == (*start)[0] && ']' != (*start)[1])
90 			gly = ESCAPE_UNICODE;
91 		term = ']';
92 		break;
93 	case ('C'):
94 		if ('\'' != **start)
95 			return(ESCAPE_ERROR);
96 		*start = ++*end;
97 		if ('u' == (*start)[0] && '\'' != (*start)[1])
98 			gly = ESCAPE_UNICODE;
99 		else
100 			gly = ESCAPE_SPECIAL;
101 		term = '\'';
102 		break;
103 
104 	/*
105 	 * The \z escape is supposed to output the following
106 	 * character without advancing the cursor position.
107 	 * Since we are mostly dealing with terminal mode,
108 	 * let us just skip the next character.
109 	 */
110 	case ('z'):
111 		return(ESCAPE_SKIPCHAR);
112 
113 	/*
114 	 * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
115 	 * 'X' is the trigger.  These have opaque sub-strings.
116 	 */
117 	case ('F'):
118 		/* FALLTHROUGH */
119 	case ('g'):
120 		/* FALLTHROUGH */
121 	case ('k'):
122 		/* FALLTHROUGH */
123 	case ('M'):
124 		/* FALLTHROUGH */
125 	case ('m'):
126 		/* FALLTHROUGH */
127 	case ('n'):
128 		/* FALLTHROUGH */
129 	case ('V'):
130 		/* FALLTHROUGH */
131 	case ('Y'):
132 		gly = ESCAPE_IGNORE;
133 		/* FALLTHROUGH */
134 	case ('f'):
135 		if (ESCAPE_ERROR == gly)
136 			gly = ESCAPE_FONT;
137 		switch (**start) {
138 		case ('('):
139 			*start = ++*end;
140 			*sz = 2;
141 			break;
142 		case ('['):
143 			*start = ++*end;
144 			term = ']';
145 			break;
146 		default:
147 			*sz = 1;
148 			break;
149 		}
150 		break;
151 
152 	/*
153 	 * These escapes are of the form \X'Y', where 'X' is the trigger
154 	 * and 'Y' is any string.  These have opaque sub-strings.
155 	 */
156 	case ('A'):
157 		/* FALLTHROUGH */
158 	case ('b'):
159 		/* FALLTHROUGH */
160 	case ('D'):
161 		/* FALLTHROUGH */
162 	case ('o'):
163 		/* FALLTHROUGH */
164 	case ('R'):
165 		/* FALLTHROUGH */
166 	case ('X'):
167 		/* FALLTHROUGH */
168 	case ('Z'):
169 		if ('\'' != **start)
170 			return(ESCAPE_ERROR);
171 		gly = ESCAPE_IGNORE;
172 		*start = ++*end;
173 		term = '\'';
174 		break;
175 
176 	/*
177 	 * These escapes are of the form \X'N', where 'X' is the trigger
178 	 * and 'N' resolves to a numerical expression.
179 	 */
180 	case ('B'):
181 		/* FALLTHROUGH */
182 	case ('h'):
183 		/* FALLTHROUGH */
184 	case ('H'):
185 		/* FALLTHROUGH */
186 	case ('L'):
187 		/* FALLTHROUGH */
188 	case ('l'):
189 		gly = ESCAPE_NUMBERED;
190 		/* FALLTHROUGH */
191 	case ('S'):
192 		/* FALLTHROUGH */
193 	case ('v'):
194 		/* FALLTHROUGH */
195 	case ('w'):
196 		/* FALLTHROUGH */
197 	case ('x'):
198 		if ('\'' != **start)
199 			return(ESCAPE_ERROR);
200 		if (ESCAPE_ERROR == gly)
201 			gly = ESCAPE_IGNORE;
202 		*start = ++*end;
203 		term = '\'';
204 		break;
205 
206 	/*
207 	 * Special handling for the numbered character escape.
208 	 * XXX Do any other escapes need similar handling?
209 	 */
210 	case ('N'):
211 		if ('\0' == **start)
212 			return(ESCAPE_ERROR);
213 		(*end)++;
214 		if (isdigit((unsigned char)**start)) {
215 			*sz = 1;
216 			return(ESCAPE_IGNORE);
217 		}
218 		(*start)++;
219 		while (isdigit((unsigned char)**end))
220 			(*end)++;
221 		*sz = *end - *start;
222 		if ('\0' != **end)
223 			(*end)++;
224 		return(ESCAPE_NUMBERED);
225 
226 	/*
227 	 * Sizes get a special category of their own.
228 	 */
229 	case ('s'):
230 		gly = ESCAPE_IGNORE;
231 
232 		/* See +/- counts as a sign. */
233 		if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
234 			(*end)++;
235 
236 		switch (**end) {
237 		case ('('):
238 			*start = ++*end;
239 			*sz = 2;
240 			break;
241 		case ('['):
242 			*start = ++*end;
243 			term = ']';
244 			break;
245 		case ('\''):
246 			*start = ++*end;
247 			term = '\'';
248 			break;
249 		default:
250 			*sz = 1;
251 			break;
252 		}
253 
254 		break;
255 
256 	/*
257 	 * Anything else is assumed to be a glyph.
258 	 * In this case, pass back the character after the backslash.
259 	 */
260 	default:
261 		gly = ESCAPE_SPECIAL;
262 		*start = --*end;
263 		*sz = 1;
264 		break;
265 	}
266 
267 	assert(ESCAPE_ERROR != gly);
268 
269 	/*
270 	 * Read up to the terminating character,
271 	 * paying attention to nested escapes.
272 	 */
273 
274 	if ('\0' != term) {
275 		while (**end != term) {
276 			switch (**end) {
277 			case ('\0'):
278 				return(ESCAPE_ERROR);
279 			case ('\\'):
280 				(*end)++;
281 				if (ESCAPE_ERROR ==
282 				    mandoc_escape(end, NULL, NULL))
283 					return(ESCAPE_ERROR);
284 				break;
285 			default:
286 				(*end)++;
287 				break;
288 			}
289 		}
290 		*sz = (*end)++ - *start;
291 	} else {
292 		assert(*sz > 0);
293 		if ((size_t)*sz > strlen(*start))
294 			return(ESCAPE_ERROR);
295 		*end += *sz;
296 	}
297 
298 	/* Run post-processors. */
299 
300 	switch (gly) {
301 	case (ESCAPE_FONT):
302 		if (2 == *sz) {
303 			if ('C' == **start) {
304 				/*
305 				 * Treat constant-width font modes
306 				 * just like regular font modes.
307 				 */
308 				(*start)++;
309 				(*sz)--;
310 			} else {
311 				if ('B' == (*start)[0] && 'I' == (*start)[1])
312 					gly = ESCAPE_FONTBI;
313 				break;
314 			}
315 		} else if (1 != *sz)
316 			break;
317 
318 		switch (**start) {
319 		case ('3'):
320 			/* FALLTHROUGH */
321 		case ('B'):
322 			gly = ESCAPE_FONTBOLD;
323 			break;
324 		case ('2'):
325 			/* FALLTHROUGH */
326 		case ('I'):
327 			gly = ESCAPE_FONTITALIC;
328 			break;
329 		case ('P'):
330 			gly = ESCAPE_FONTPREV;
331 			break;
332 		case ('1'):
333 			/* FALLTHROUGH */
334 		case ('R'):
335 			gly = ESCAPE_FONTROMAN;
336 			break;
337 		}
338 		break;
339 	case (ESCAPE_SPECIAL):
340 		if (1 == *sz && 'c' == **start)
341 			gly = ESCAPE_NOSPACE;
342 		break;
343 	default:
344 		break;
345 	}
346 
347 	return(gly);
348 }
349 
350 void *
351 mandoc_calloc(size_t num, size_t size)
352 {
353 	void		*ptr;
354 
355 	ptr = calloc(num, size);
356 	if (NULL == ptr) {
357 		perror(NULL);
358 		exit((int)MANDOCLEVEL_SYSERR);
359 	}
360 
361 	return(ptr);
362 }
363 
364 
365 void *
366 mandoc_malloc(size_t size)
367 {
368 	void		*ptr;
369 
370 	ptr = malloc(size);
371 	if (NULL == ptr) {
372 		perror(NULL);
373 		exit((int)MANDOCLEVEL_SYSERR);
374 	}
375 
376 	return(ptr);
377 }
378 
379 
380 void *
381 mandoc_realloc(void *ptr, size_t size)
382 {
383 
384 	ptr = realloc(ptr, size);
385 	if (NULL == ptr) {
386 		perror(NULL);
387 		exit((int)MANDOCLEVEL_SYSERR);
388 	}
389 
390 	return(ptr);
391 }
392 
393 char *
394 mandoc_strndup(const char *ptr, size_t sz)
395 {
396 	char		*p;
397 
398 	p = mandoc_malloc(sz + 1);
399 	memcpy(p, ptr, sz);
400 	p[(int)sz] = '\0';
401 	return(p);
402 }
403 
404 char *
405 mandoc_strdup(const char *ptr)
406 {
407 	char		*p;
408 
409 	p = strdup(ptr);
410 	if (NULL == p) {
411 		perror(NULL);
412 		exit((int)MANDOCLEVEL_SYSERR);
413 	}
414 
415 	return(p);
416 }
417 
418 /*
419  * Parse a quoted or unquoted roff-style request or macro argument.
420  * Return a pointer to the parsed argument, which is either the original
421  * pointer or advanced by one byte in case the argument is quoted.
422  * Null-terminate the argument in place.
423  * Collapse pairs of quotes inside quoted arguments.
424  * Advance the argument pointer to the next argument,
425  * or to the null byte terminating the argument line.
426  */
427 char *
428 mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
429 {
430 	char	 *start, *cp;
431 	int	  quoted, pairs, white;
432 
433 	/* Quoting can only start with a new word. */
434 	start = *cpp;
435 	quoted = 0;
436 	if ('"' == *start) {
437 		quoted = 1;
438 		start++;
439 	}
440 
441 	pairs = 0;
442 	white = 0;
443 	for (cp = start; '\0' != *cp; cp++) {
444 
445 		/*
446 		 * Move the following text left
447 		 * after quoted quotes and after "\\" and "\t".
448 		 */
449 		if (pairs)
450 			cp[-pairs] = cp[0];
451 
452 		if ('\\' == cp[0]) {
453 			/*
454 			 * In copy mode, translate double to single
455 			 * backslashes and backslash-t to literal tabs.
456 			 */
457 			switch (cp[1]) {
458 			case ('t'):
459 				cp[0] = '\t';
460 				/* FALLTHROUGH */
461 			case ('\\'):
462 				pairs++;
463 				cp++;
464 				break;
465 			case (' '):
466 				/* Skip escaped blanks. */
467 				if (0 == quoted)
468 					cp++;
469 				break;
470 			default:
471 				break;
472 			}
473 		} else if (0 == quoted) {
474 			if (' ' == cp[0]) {
475 				/* Unescaped blanks end unquoted args. */
476 				white = 1;
477 				break;
478 			}
479 		} else if ('"' == cp[0]) {
480 			if ('"' == cp[1]) {
481 				/* Quoted quotes collapse. */
482 				pairs++;
483 				cp++;
484 			} else {
485 				/* Unquoted quotes end quoted args. */
486 				quoted = 2;
487 				break;
488 			}
489 		}
490 	}
491 
492 	/* Quoted argument without a closing quote. */
493 	if (1 == quoted)
494 		mandoc_msg(MANDOCERR_BADQUOTE, parse, ln, *pos, NULL);
495 
496 	/* Null-terminate this argument and move to the next one. */
497 	if (pairs)
498 		cp[-pairs] = '\0';
499 	if ('\0' != *cp) {
500 		*cp++ = '\0';
501 		while (' ' == *cp)
502 			cp++;
503 	}
504 	*pos += (int)(cp - start) + (quoted ? 1 : 0);
505 	*cpp = cp;
506 
507 	if ('\0' == *cp && (white || ' ' == cp[-1]))
508 		mandoc_msg(MANDOCERR_EOLNSPACE, parse, ln, *pos, NULL);
509 
510 	return(start);
511 }
512 
513 static int
514 a2time(time_t *t, const char *fmt, const char *p)
515 {
516 	struct tm	 tm;
517 	char		*pp;
518 
519 	memset(&tm, 0, sizeof(struct tm));
520 
521 	pp = NULL;
522 #ifdef	HAVE_STRPTIME
523 	pp = strptime(p, fmt, &tm);
524 #endif
525 	if (NULL != pp && '\0' == *pp) {
526 		*t = mktime(&tm);
527 		return(1);
528 	}
529 
530 	return(0);
531 }
532 
533 static char *
534 time2a(time_t t)
535 {
536 	struct tm	*tm;
537 	char		*buf, *p;
538 	size_t		 ssz;
539 	int		 isz;
540 
541 	tm = localtime(&t);
542 
543 	/*
544 	 * Reserve space:
545 	 * up to 9 characters for the month (September) + blank
546 	 * up to 2 characters for the day + comma + blank
547 	 * 4 characters for the year and a terminating '\0'
548 	 */
549 	p = buf = mandoc_malloc(10 + 4 + 4 + 1);
550 
551 	if (0 == (ssz = strftime(p, 10 + 1, "%B ", tm)))
552 		goto fail;
553 	p += (int)ssz;
554 
555 	if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)))
556 		goto fail;
557 	p += isz;
558 
559 	if (0 == strftime(p, 4 + 1, "%Y", tm))
560 		goto fail;
561 	return(buf);
562 
563 fail:
564 	free(buf);
565 	return(NULL);
566 }
567 
568 char *
569 mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
570 {
571 	char		*out;
572 	time_t		 t;
573 
574 	if (NULL == in || '\0' == *in ||
575 	    0 == strcmp(in, "$" "Mdocdate$")) {
576 		mandoc_msg(MANDOCERR_NODATE, parse, ln, pos, NULL);
577 		time(&t);
578 	}
579 	else if (a2time(&t, "%Y-%m-%d", in))
580 		t = 0;
581 	else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
582 	    !a2time(&t, "%b %d, %Y", in)) {
583 		mandoc_msg(MANDOCERR_BADDATE, parse, ln, pos, NULL);
584 		t = 0;
585 	}
586 	out = t ? time2a(t) : NULL;
587 	return(out ? out : mandoc_strdup(in));
588 }
589 
590 int
591 mandoc_eos(const char *p, size_t sz, int enclosed)
592 {
593 	const char *q;
594 	int found;
595 
596 	if (0 == sz)
597 		return(0);
598 
599 	/*
600 	 * End-of-sentence recognition must include situations where
601 	 * some symbols, such as `)', allow prior EOS punctuation to
602 	 * propagate outward.
603 	 */
604 
605 	found = 0;
606 	for (q = p + (int)sz - 1; q >= p; q--) {
607 		switch (*q) {
608 		case ('\"'):
609 			/* FALLTHROUGH */
610 		case ('\''):
611 			/* FALLTHROUGH */
612 		case (']'):
613 			/* FALLTHROUGH */
614 		case (')'):
615 			if (0 == found)
616 				enclosed = 1;
617 			break;
618 		case ('.'):
619 			/* FALLTHROUGH */
620 		case ('!'):
621 			/* FALLTHROUGH */
622 		case ('?'):
623 			found = 1;
624 			break;
625 		default:
626 			return(found && (!enclosed || isalnum((unsigned char)*q)));
627 		}
628 	}
629 
630 	return(found && !enclosed);
631 }
632 
633 /*
634  * Convert a string to a long that may not be <0.
635  * If the string is invalid, or is less than 0, return -1.
636  */
637 int
638 mandoc_strntoi(const char *p, size_t sz, int base)
639 {
640 	char		 buf[32];
641 	char		*ep;
642 	long		 v;
643 
644 	if (sz > 31)
645 		return(-1);
646 
647 	memcpy(buf, p, sz);
648 	buf[(int)sz] = '\0';
649 
650 	errno = 0;
651 	v = strtol(buf, &ep, base);
652 
653 	if (buf[0] == '\0' || *ep != '\0')
654 		return(-1);
655 
656 	if (v > INT_MAX)
657 		v = INT_MAX;
658 	if (v < INT_MIN)
659 		v = INT_MIN;
660 
661 	return((int)v);
662 }
663