xref: /dragonfly/contrib/mdocml/mandoc.c (revision 0dace59e)
1 /*	$Id: mandoc.c,v 1.68 2013/08/08 20:07:47 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 **end, const char **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 		gly = ESCAPE_SPECIAL;
97 		*start = ++*end;
98 		term = '\'';
99 		break;
100 
101 	/*
102 	 * The \z escape is supposed to output the following
103 	 * character without advancing the cursor position.
104 	 * Since we are mostly dealing with terminal mode,
105 	 * let us just skip the next character.
106 	 */
107 	case ('z'):
108 		return(ESCAPE_SKIPCHAR);
109 
110 	/*
111 	 * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
112 	 * 'X' is the trigger.  These have opaque sub-strings.
113 	 */
114 	case ('F'):
115 		/* FALLTHROUGH */
116 	case ('g'):
117 		/* FALLTHROUGH */
118 	case ('k'):
119 		/* FALLTHROUGH */
120 	case ('M'):
121 		/* FALLTHROUGH */
122 	case ('m'):
123 		/* FALLTHROUGH */
124 	case ('n'):
125 		/* FALLTHROUGH */
126 	case ('V'):
127 		/* FALLTHROUGH */
128 	case ('Y'):
129 		gly = ESCAPE_IGNORE;
130 		/* FALLTHROUGH */
131 	case ('f'):
132 		if (ESCAPE_ERROR == gly)
133 			gly = ESCAPE_FONT;
134 		switch (**start) {
135 		case ('('):
136 			*start = ++*end;
137 			*sz = 2;
138 			break;
139 		case ('['):
140 			*start = ++*end;
141 			term = ']';
142 			break;
143 		default:
144 			*sz = 1;
145 			break;
146 		}
147 		break;
148 
149 	/*
150 	 * These escapes are of the form \X'Y', where 'X' is the trigger
151 	 * and 'Y' is any string.  These have opaque sub-strings.
152 	 */
153 	case ('A'):
154 		/* FALLTHROUGH */
155 	case ('b'):
156 		/* FALLTHROUGH */
157 	case ('D'):
158 		/* FALLTHROUGH */
159 	case ('o'):
160 		/* FALLTHROUGH */
161 	case ('R'):
162 		/* FALLTHROUGH */
163 	case ('X'):
164 		/* FALLTHROUGH */
165 	case ('Z'):
166 		if ('\'' != **start)
167 			return(ESCAPE_ERROR);
168 		gly = ESCAPE_IGNORE;
169 		*start = ++*end;
170 		term = '\'';
171 		break;
172 
173 	/*
174 	 * These escapes are of the form \X'N', where 'X' is the trigger
175 	 * and 'N' resolves to a numerical expression.
176 	 */
177 	case ('B'):
178 		/* FALLTHROUGH */
179 	case ('h'):
180 		/* FALLTHROUGH */
181 	case ('H'):
182 		/* FALLTHROUGH */
183 	case ('L'):
184 		/* FALLTHROUGH */
185 	case ('l'):
186 		gly = ESCAPE_NUMBERED;
187 		/* FALLTHROUGH */
188 	case ('S'):
189 		/* FALLTHROUGH */
190 	case ('v'):
191 		/* FALLTHROUGH */
192 	case ('w'):
193 		/* FALLTHROUGH */
194 	case ('x'):
195 		if ('\'' != **start)
196 			return(ESCAPE_ERROR);
197 		if (ESCAPE_ERROR == gly)
198 			gly = ESCAPE_IGNORE;
199 		*start = ++*end;
200 		term = '\'';
201 		break;
202 
203 	/*
204 	 * Special handling for the numbered character escape.
205 	 * XXX Do any other escapes need similar handling?
206 	 */
207 	case ('N'):
208 		if ('\0' == **start)
209 			return(ESCAPE_ERROR);
210 		(*end)++;
211 		if (isdigit((unsigned char)**start)) {
212 			*sz = 1;
213 			return(ESCAPE_IGNORE);
214 		}
215 		(*start)++;
216 		while (isdigit((unsigned char)**end))
217 			(*end)++;
218 		*sz = *end - *start;
219 		if ('\0' != **end)
220 			(*end)++;
221 		return(ESCAPE_NUMBERED);
222 
223 	/*
224 	 * Sizes get a special category of their own.
225 	 */
226 	case ('s'):
227 		gly = ESCAPE_IGNORE;
228 
229 		/* See +/- counts as a sign. */
230 		if ('+' == **end || '-' == **end || ASCII_HYPH == **end)
231 			(*end)++;
232 
233 		switch (**end) {
234 		case ('('):
235 			*start = ++*end;
236 			*sz = 2;
237 			break;
238 		case ('['):
239 			*start = ++*end;
240 			term = ']';
241 			break;
242 		case ('\''):
243 			*start = ++*end;
244 			term = '\'';
245 			break;
246 		default:
247 			*sz = 1;
248 			break;
249 		}
250 
251 		break;
252 
253 	/*
254 	 * Anything else is assumed to be a glyph.
255 	 * In this case, pass back the character after the backslash.
256 	 */
257 	default:
258 		gly = ESCAPE_SPECIAL;
259 		*start = --*end;
260 		*sz = 1;
261 		break;
262 	}
263 
264 	assert(ESCAPE_ERROR != gly);
265 
266 	/*
267 	 * Read up to the terminating character,
268 	 * paying attention to nested escapes.
269 	 */
270 
271 	if ('\0' != term) {
272 		while (**end != term) {
273 			switch (**end) {
274 			case ('\0'):
275 				return(ESCAPE_ERROR);
276 			case ('\\'):
277 				(*end)++;
278 				if (ESCAPE_ERROR ==
279 				    mandoc_escape(end, NULL, NULL))
280 					return(ESCAPE_ERROR);
281 				break;
282 			default:
283 				(*end)++;
284 				break;
285 			}
286 		}
287 		*sz = (*end)++ - *start;
288 	} else {
289 		assert(*sz > 0);
290 		if ((size_t)*sz > strlen(*start))
291 			return(ESCAPE_ERROR);
292 		*end += *sz;
293 	}
294 
295 	/* Run post-processors. */
296 
297 	switch (gly) {
298 	case (ESCAPE_FONT):
299 		if (2 == *sz) {
300 			if ('C' == **start) {
301 				/*
302 				 * Treat constant-width font modes
303 				 * just like regular font modes.
304 				 */
305 				(*start)++;
306 				(*sz)--;
307 			} else {
308 				if ('B' == (*start)[0] && 'I' == (*start)[1])
309 					gly = ESCAPE_FONTBI;
310 				break;
311 			}
312 		} else if (1 != *sz)
313 			break;
314 
315 		switch (**start) {
316 		case ('3'):
317 			/* FALLTHROUGH */
318 		case ('B'):
319 			gly = ESCAPE_FONTBOLD;
320 			break;
321 		case ('2'):
322 			/* FALLTHROUGH */
323 		case ('I'):
324 			gly = ESCAPE_FONTITALIC;
325 			break;
326 		case ('P'):
327 			gly = ESCAPE_FONTPREV;
328 			break;
329 		case ('1'):
330 			/* FALLTHROUGH */
331 		case ('R'):
332 			gly = ESCAPE_FONTROMAN;
333 			break;
334 		}
335 		break;
336 	case (ESCAPE_SPECIAL):
337 		if (1 == *sz && 'c' == **start)
338 			gly = ESCAPE_NOSPACE;
339 		break;
340 	default:
341 		break;
342 	}
343 
344 	return(gly);
345 }
346 
347 void *
348 mandoc_calloc(size_t num, size_t size)
349 {
350 	void		*ptr;
351 
352 	ptr = calloc(num, size);
353 	if (NULL == ptr) {
354 		perror(NULL);
355 		exit((int)MANDOCLEVEL_SYSERR);
356 	}
357 
358 	return(ptr);
359 }
360 
361 
362 void *
363 mandoc_malloc(size_t size)
364 {
365 	void		*ptr;
366 
367 	ptr = malloc(size);
368 	if (NULL == ptr) {
369 		perror(NULL);
370 		exit((int)MANDOCLEVEL_SYSERR);
371 	}
372 
373 	return(ptr);
374 }
375 
376 
377 void *
378 mandoc_realloc(void *ptr, size_t size)
379 {
380 
381 	ptr = realloc(ptr, size);
382 	if (NULL == ptr) {
383 		perror(NULL);
384 		exit((int)MANDOCLEVEL_SYSERR);
385 	}
386 
387 	return(ptr);
388 }
389 
390 char *
391 mandoc_strndup(const char *ptr, size_t sz)
392 {
393 	char		*p;
394 
395 	p = mandoc_malloc(sz + 1);
396 	memcpy(p, ptr, sz);
397 	p[(int)sz] = '\0';
398 	return(p);
399 }
400 
401 char *
402 mandoc_strdup(const char *ptr)
403 {
404 	char		*p;
405 
406 	p = strdup(ptr);
407 	if (NULL == p) {
408 		perror(NULL);
409 		exit((int)MANDOCLEVEL_SYSERR);
410 	}
411 
412 	return(p);
413 }
414 
415 /*
416  * Parse a quoted or unquoted roff-style request or macro argument.
417  * Return a pointer to the parsed argument, which is either the original
418  * pointer or advanced by one byte in case the argument is quoted.
419  * Null-terminate the argument in place.
420  * Collapse pairs of quotes inside quoted arguments.
421  * Advance the argument pointer to the next argument,
422  * or to the null byte terminating the argument line.
423  */
424 char *
425 mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
426 {
427 	char	 *start, *cp;
428 	int	  quoted, pairs, white;
429 
430 	/* Quoting can only start with a new word. */
431 	start = *cpp;
432 	quoted = 0;
433 	if ('"' == *start) {
434 		quoted = 1;
435 		start++;
436 	}
437 
438 	pairs = 0;
439 	white = 0;
440 	for (cp = start; '\0' != *cp; cp++) {
441 
442 		/*
443 		 * Move the following text left
444 		 * after quoted quotes and after "\\" and "\t".
445 		 */
446 		if (pairs)
447 			cp[-pairs] = cp[0];
448 
449 		if ('\\' == cp[0]) {
450 			/*
451 			 * In copy mode, translate double to single
452 			 * backslashes and backslash-t to literal tabs.
453 			 */
454 			switch (cp[1]) {
455 			case ('t'):
456 				cp[0] = '\t';
457 				/* FALLTHROUGH */
458 			case ('\\'):
459 				pairs++;
460 				cp++;
461 				break;
462 			case (' '):
463 				/* Skip escaped blanks. */
464 				if (0 == quoted)
465 					cp++;
466 				break;
467 			default:
468 				break;
469 			}
470 		} else if (0 == quoted) {
471 			if (' ' == cp[0]) {
472 				/* Unescaped blanks end unquoted args. */
473 				white = 1;
474 				break;
475 			}
476 		} else if ('"' == cp[0]) {
477 			if ('"' == cp[1]) {
478 				/* Quoted quotes collapse. */
479 				pairs++;
480 				cp++;
481 			} else {
482 				/* Unquoted quotes end quoted args. */
483 				quoted = 2;
484 				break;
485 			}
486 		}
487 	}
488 
489 	/* Quoted argument without a closing quote. */
490 	if (1 == quoted)
491 		mandoc_msg(MANDOCERR_BADQUOTE, parse, ln, *pos, NULL);
492 
493 	/* Null-terminate this argument and move to the next one. */
494 	if (pairs)
495 		cp[-pairs] = '\0';
496 	if ('\0' != *cp) {
497 		*cp++ = '\0';
498 		while (' ' == *cp)
499 			cp++;
500 	}
501 	*pos += (int)(cp - start) + (quoted ? 1 : 0);
502 	*cpp = cp;
503 
504 	if ('\0' == *cp && (white || ' ' == cp[-1]))
505 		mandoc_msg(MANDOCERR_EOLNSPACE, parse, ln, *pos, NULL);
506 
507 	return(start);
508 }
509 
510 static int
511 a2time(time_t *t, const char *fmt, const char *p)
512 {
513 	struct tm	 tm;
514 	char		*pp;
515 
516 	memset(&tm, 0, sizeof(struct tm));
517 
518 	pp = NULL;
519 #ifdef	HAVE_STRPTIME
520 	pp = strptime(p, fmt, &tm);
521 #endif
522 	if (NULL != pp && '\0' == *pp) {
523 		*t = mktime(&tm);
524 		return(1);
525 	}
526 
527 	return(0);
528 }
529 
530 static char *
531 time2a(time_t t)
532 {
533 	struct tm	*tm;
534 	char		*buf, *p;
535 	size_t		 ssz;
536 	int		 isz;
537 
538 	tm = localtime(&t);
539 
540 	/*
541 	 * Reserve space:
542 	 * up to 9 characters for the month (September) + blank
543 	 * up to 2 characters for the day + comma + blank
544 	 * 4 characters for the year and a terminating '\0'
545 	 */
546 	p = buf = mandoc_malloc(10 + 4 + 4 + 1);
547 
548 	if (0 == (ssz = strftime(p, 10 + 1, "%B ", tm)))
549 		goto fail;
550 	p += (int)ssz;
551 
552 	if (-1 == (isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)))
553 		goto fail;
554 	p += isz;
555 
556 	if (0 == strftime(p, 4 + 1, "%Y", tm))
557 		goto fail;
558 	return(buf);
559 
560 fail:
561 	free(buf);
562 	return(NULL);
563 }
564 
565 char *
566 mandoc_normdate(struct mparse *parse, char *in, int ln, int pos)
567 {
568 	char		*out;
569 	time_t		 t;
570 
571 	if (NULL == in || '\0' == *in ||
572 	    0 == strcmp(in, "$" "Mdocdate$")) {
573 		mandoc_msg(MANDOCERR_NODATE, parse, ln, pos, NULL);
574 		time(&t);
575 	}
576 	else if (a2time(&t, "%Y-%m-%d", in))
577 		t = 0;
578 	else if (!a2time(&t, "$" "Mdocdate: %b %d %Y $", in) &&
579 	    !a2time(&t, "%b %d, %Y", in)) {
580 		mandoc_msg(MANDOCERR_BADDATE, parse, ln, pos, NULL);
581 		t = 0;
582 	}
583 	out = t ? time2a(t) : NULL;
584 	return(out ? out : mandoc_strdup(in));
585 }
586 
587 int
588 mandoc_eos(const char *p, size_t sz, int enclosed)
589 {
590 	const char *q;
591 	int found;
592 
593 	if (0 == sz)
594 		return(0);
595 
596 	/*
597 	 * End-of-sentence recognition must include situations where
598 	 * some symbols, such as `)', allow prior EOS punctuation to
599 	 * propagate outward.
600 	 */
601 
602 	found = 0;
603 	for (q = p + (int)sz - 1; q >= p; q--) {
604 		switch (*q) {
605 		case ('\"'):
606 			/* FALLTHROUGH */
607 		case ('\''):
608 			/* FALLTHROUGH */
609 		case (']'):
610 			/* FALLTHROUGH */
611 		case (')'):
612 			if (0 == found)
613 				enclosed = 1;
614 			break;
615 		case ('.'):
616 			/* FALLTHROUGH */
617 		case ('!'):
618 			/* FALLTHROUGH */
619 		case ('?'):
620 			found = 1;
621 			break;
622 		default:
623 			return(found && (!enclosed || isalnum((unsigned char)*q)));
624 		}
625 	}
626 
627 	return(found && !enclosed);
628 }
629 
630 /*
631  * Convert a string to a long that may not be <0.
632  * If the string is invalid, or is less than 0, return -1.
633  */
634 int
635 mandoc_strntoi(const char *p, size_t sz, int base)
636 {
637 	char		 buf[32];
638 	char		*ep;
639 	long		 v;
640 
641 	if (sz > 31)
642 		return(-1);
643 
644 	memcpy(buf, p, sz);
645 	buf[(int)sz] = '\0';
646 
647 	errno = 0;
648 	v = strtol(buf, &ep, base);
649 
650 	if (buf[0] == '\0' || *ep != '\0')
651 		return(-1);
652 
653 	if (v > INT_MAX)
654 		v = INT_MAX;
655 	if (v < INT_MIN)
656 		v = INT_MIN;
657 
658 	return((int)v);
659 }
660