]> git.cameronkatri.com Git - mandoc.git/blob - mandoc.c
Implement the roff(7) .shift and .return requests,
[mandoc.git] / mandoc.c
1 /* $Id: mandoc.c,v 1.107 2018/08/20 18:06:56 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2015, 2017, 2018 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 <errno.h>
25 #include <limits.h>
26 #include <stdlib.h>
27 #include <stdio.h>
28 #include <string.h>
29 #include <time.h>
30
31 #include "mandoc_aux.h"
32 #include "mandoc.h"
33 #include "roff.h"
34 #include "libmandoc.h"
35
36 static int a2time(time_t *, const char *, const char *);
37 static char *time2a(time_t);
38
39
40 enum mandoc_esc
41 mandoc_escape(const char **end, const char **start, int *sz)
42 {
43 const char *local_start;
44 int local_sz, c, i;
45 char term;
46 enum mandoc_esc gly;
47
48 /*
49 * When the caller doesn't provide return storage,
50 * use local storage.
51 */
52
53 if (NULL == start)
54 start = &local_start;
55 if (NULL == sz)
56 sz = &local_sz;
57
58 /*
59 * Beyond the backslash, at least one input character
60 * is part of the escape sequence. With one exception
61 * (see below), that character won't be returned.
62 */
63
64 gly = ESCAPE_ERROR;
65 *start = ++*end;
66 *sz = 0;
67 term = '\0';
68
69 switch ((*start)[-1]) {
70 /*
71 * First the glyphs. There are several different forms of
72 * these, but each eventually returns a substring of the glyph
73 * name.
74 */
75 case '(':
76 gly = ESCAPE_SPECIAL;
77 *sz = 2;
78 break;
79 case '[':
80 gly = ESCAPE_SPECIAL;
81 term = ']';
82 break;
83 case 'C':
84 if ('\'' != **start)
85 return ESCAPE_ERROR;
86 *start = ++*end;
87 gly = ESCAPE_SPECIAL;
88 term = '\'';
89 break;
90
91 /*
92 * Escapes taking no arguments at all.
93 */
94 case 'd':
95 case 'u':
96 case ',':
97 case '/':
98 return ESCAPE_IGNORE;
99 case 'p':
100 return ESCAPE_BREAK;
101
102 /*
103 * The \z escape is supposed to output the following
104 * character without advancing the cursor position.
105 * Since we are mostly dealing with terminal mode,
106 * let us just skip the next character.
107 */
108 case 'z':
109 return ESCAPE_SKIPCHAR;
110
111 /*
112 * Handle all triggers matching \X(xy, \Xx, and \X[xxxx], where
113 * 'X' is the trigger. These have opaque sub-strings.
114 */
115 case 'F':
116 case 'g':
117 case 'k':
118 case 'M':
119 case 'm':
120 case 'n':
121 case 'V':
122 case 'Y':
123 gly = ESCAPE_IGNORE;
124 /* FALLTHROUGH */
125 case 'f':
126 if (ESCAPE_ERROR == gly)
127 gly = ESCAPE_FONT;
128 switch (**start) {
129 case '(':
130 *start = ++*end;
131 *sz = 2;
132 break;
133 case '[':
134 *start = ++*end;
135 term = ']';
136 break;
137 default:
138 *sz = 1;
139 break;
140 }
141 break;
142 case '*':
143 if (strncmp(*start, "(.T", 3) != 0)
144 abort();
145 gly = ESCAPE_DEVICE;
146 *start = ++*end;
147 *sz = 2;
148 break;
149
150 /*
151 * These escapes are of the form \X'Y', where 'X' is the trigger
152 * and 'Y' is any string. These have opaque sub-strings.
153 * The \B and \w escapes are handled in roff.c, roff_res().
154 */
155 case 'A':
156 case 'b':
157 case 'D':
158 case 'R':
159 case 'X':
160 case 'Z':
161 gly = ESCAPE_IGNORE;
162 /* FALLTHROUGH */
163 case 'o':
164 if (**start == '\0')
165 return ESCAPE_ERROR;
166 if (gly == ESCAPE_ERROR)
167 gly = ESCAPE_OVERSTRIKE;
168 term = **start;
169 *start = ++*end;
170 break;
171
172 /*
173 * These escapes are of the form \X'N', where 'X' is the trigger
174 * and 'N' resolves to a numerical expression.
175 */
176 case 'h':
177 case 'H':
178 case 'L':
179 case 'l':
180 case 'S':
181 case 'v':
182 case 'x':
183 if (strchr(" %&()*+-./0123456789:<=>", **start)) {
184 if ('\0' != **start)
185 ++*end;
186 return ESCAPE_ERROR;
187 }
188 switch ((*start)[-1]) {
189 case 'h':
190 gly = ESCAPE_HORIZ;
191 break;
192 case 'l':
193 gly = ESCAPE_HLINE;
194 break;
195 default:
196 gly = ESCAPE_IGNORE;
197 break;
198 }
199 term = **start;
200 *start = ++*end;
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 *start = ++*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 case '3':
247 case '2':
248 case '1':
249 *sz = (*end)[-1] == 's' &&
250 isdigit((unsigned char)(*end)[1]) ? 2 : 1;
251 break;
252 default:
253 *sz = 1;
254 break;
255 }
256
257 break;
258
259 /*
260 * Anything else is assumed to be a glyph.
261 * In this case, pass back the character after the backslash.
262 */
263 default:
264 gly = ESCAPE_SPECIAL;
265 *start = --*end;
266 *sz = 1;
267 break;
268 }
269
270 assert(ESCAPE_ERROR != gly);
271
272 /*
273 * Read up to the terminating character,
274 * paying attention to nested escapes.
275 */
276
277 if ('\0' != term) {
278 while (**end != term) {
279 switch (**end) {
280 case '\0':
281 return ESCAPE_ERROR;
282 case '\\':
283 (*end)++;
284 if (ESCAPE_ERROR ==
285 mandoc_escape(end, NULL, NULL))
286 return ESCAPE_ERROR;
287 break;
288 default:
289 (*end)++;
290 break;
291 }
292 }
293 *sz = (*end)++ - *start;
294 } else {
295 assert(*sz > 0);
296 if ((size_t)*sz > strlen(*start))
297 return ESCAPE_ERROR;
298 *end += *sz;
299 }
300
301 /* Run post-processors. */
302
303 switch (gly) {
304 case ESCAPE_FONT:
305 if (*sz == 2) {
306 if (**start == 'C') {
307 /*
308 * Treat constant-width font modes
309 * just like regular font modes.
310 */
311 (*start)++;
312 (*sz)--;
313 } else {
314 if ((*start)[0] == 'B' && (*start)[1] == 'I')
315 gly = ESCAPE_FONTBI;
316 break;
317 }
318 } else if (*sz != 1) {
319 if (*sz == 0)
320 gly = ESCAPE_FONTPREV;
321 break;
322 }
323
324 switch (**start) {
325 case '3':
326 case 'B':
327 gly = ESCAPE_FONTBOLD;
328 break;
329 case '2':
330 case 'I':
331 gly = ESCAPE_FONTITALIC;
332 break;
333 case 'P':
334 gly = ESCAPE_FONTPREV;
335 break;
336 case '1':
337 case 'R':
338 gly = ESCAPE_FONTROMAN;
339 break;
340 }
341 break;
342 case ESCAPE_SPECIAL:
343 if (**start == 'c') {
344 if (*sz == 1) {
345 gly = ESCAPE_NOSPACE;
346 break;
347 }
348 if (*sz < 6 || *sz > 7 ||
349 strncmp(*start, "char", 4) != 0 ||
350 (int)strspn(*start + 4, "0123456789") + 4 < *sz)
351 break;
352 c = 0;
353 for (i = 4; i < *sz; i++)
354 c = 10 * c + ((*start)[i] - '0');
355 if (c < 0x21 || (c > 0x7e && c < 0xa0) || c > 0xff)
356 break;
357 *start += 4;
358 *sz -= 4;
359 gly = ESCAPE_NUMBERED;
360 break;
361 }
362
363 /*
364 * Unicode escapes are defined in groff as \[u0000]
365 * to \[u10FFFF], where the contained value must be
366 * a valid Unicode codepoint. Here, however, only
367 * check the length and range.
368 */
369 if (**start != 'u' || *sz < 5 || *sz > 7)
370 break;
371 if (*sz == 7 && ((*start)[1] != '1' || (*start)[2] != '0'))
372 break;
373 if (*sz == 6 && (*start)[1] == '0')
374 break;
375 if (*sz == 5 && (*start)[1] == 'D' &&
376 strchr("89ABCDEF", (*start)[2]) != NULL)
377 break;
378 if ((int)strspn(*start + 1, "0123456789ABCDEFabcdef")
379 + 1 == *sz)
380 gly = ESCAPE_UNICODE;
381 break;
382 default:
383 break;
384 }
385
386 return gly;
387 }
388
389 /*
390 * Parse a quoted or unquoted roff-style request or macro argument.
391 * Return a pointer to the parsed argument, which is either the original
392 * pointer or advanced by one byte in case the argument is quoted.
393 * NUL-terminate the argument in place.
394 * Collapse pairs of quotes inside quoted arguments.
395 * Advance the argument pointer to the next argument,
396 * or to the NUL byte terminating the argument line.
397 */
398 char *
399 mandoc_getarg(struct mparse *parse, char **cpp, int ln, int *pos)
400 {
401 char *start, *cp;
402 int quoted, pairs, white;
403
404 /* Quoting can only start with a new word. */
405 start = *cpp;
406 quoted = 0;
407 if ('"' == *start) {
408 quoted = 1;
409 start++;
410 }
411
412 pairs = 0;
413 white = 0;
414 for (cp = start; '\0' != *cp; cp++) {
415
416 /*
417 * Move the following text left
418 * after quoted quotes and after "\\" and "\t".
419 */
420 if (pairs)
421 cp[-pairs] = cp[0];
422
423 if ('\\' == cp[0]) {
424 /*
425 * In copy mode, translate double to single
426 * backslashes and backslash-t to literal tabs.
427 */
428 switch (cp[1]) {
429 case 't':
430 cp[0] = '\t';
431 /* FALLTHROUGH */
432 case '\\':
433 pairs++;
434 cp++;
435 break;
436 case ' ':
437 /* Skip escaped blanks. */
438 if (0 == quoted)
439 cp++;
440 break;
441 default:
442 break;
443 }
444 } else if (0 == quoted) {
445 if (' ' == cp[0]) {
446 /* Unescaped blanks end unquoted args. */
447 white = 1;
448 break;
449 }
450 } else if ('"' == cp[0]) {
451 if ('"' == cp[1]) {
452 /* Quoted quotes collapse. */
453 pairs++;
454 cp++;
455 } else {
456 /* Unquoted quotes end quoted args. */
457 quoted = 2;
458 break;
459 }
460 }
461 }
462
463 /* Quoted argument without a closing quote. */
464 if (1 == quoted)
465 mandoc_msg(MANDOCERR_ARG_QUOTE, parse, ln, *pos, NULL);
466
467 /* NUL-terminate this argument and move to the next one. */
468 if (pairs)
469 cp[-pairs] = '\0';
470 if ('\0' != *cp) {
471 *cp++ = '\0';
472 while (' ' == *cp)
473 cp++;
474 }
475 *pos += (int)(cp - start) + (quoted ? 1 : 0);
476 *cpp = cp;
477
478 if ('\0' == *cp && (white || ' ' == cp[-1]))
479 mandoc_msg(MANDOCERR_SPACE_EOL, parse, ln, *pos, NULL);
480
481 return start;
482 }
483
484 static int
485 a2time(time_t *t, const char *fmt, const char *p)
486 {
487 struct tm tm;
488 char *pp;
489
490 memset(&tm, 0, sizeof(struct tm));
491
492 pp = NULL;
493 #if HAVE_STRPTIME
494 pp = strptime(p, fmt, &tm);
495 #endif
496 if (NULL != pp && '\0' == *pp) {
497 *t = mktime(&tm);
498 return 1;
499 }
500
501 return 0;
502 }
503
504 static char *
505 time2a(time_t t)
506 {
507 struct tm *tm;
508 char *buf, *p;
509 size_t ssz;
510 int isz;
511
512 tm = localtime(&t);
513 if (tm == NULL)
514 return NULL;
515
516 /*
517 * Reserve space:
518 * up to 9 characters for the month (September) + blank
519 * up to 2 characters for the day + comma + blank
520 * 4 characters for the year and a terminating '\0'
521 */
522
523 p = buf = mandoc_malloc(10 + 4 + 4 + 1);
524
525 if ((ssz = strftime(p, 10 + 1, "%B ", tm)) == 0)
526 goto fail;
527 p += (int)ssz;
528
529 /*
530 * The output format is just "%d" here, not "%2d" or "%02d".
531 * That's also the reason why we can't just format the
532 * date as a whole with "%B %e, %Y" or "%B %d, %Y".
533 * Besides, the present approach is less prone to buffer
534 * overflows, in case anybody should ever introduce the bug
535 * of looking at LC_TIME.
536 */
537
538 if ((isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday)) == -1)
539 goto fail;
540 p += isz;
541
542 if (strftime(p, 4 + 1, "%Y", tm) == 0)
543 goto fail;
544 return buf;
545
546 fail:
547 free(buf);
548 return NULL;
549 }
550
551 char *
552 mandoc_normdate(struct roff_man *man, char *in, int ln, int pos)
553 {
554 char *cp;
555 time_t t;
556
557 /* No date specified: use today's date. */
558
559 if (in == NULL || *in == '\0' || strcmp(in, "$" "Mdocdate$") == 0) {
560 mandoc_msg(MANDOCERR_DATE_MISSING, man->parse, ln, pos, NULL);
561 return time2a(time(NULL));
562 }
563
564 /* Valid mdoc(7) date format. */
565
566 if (a2time(&t, "$" "Mdocdate: %b %d %Y $", in) ||
567 a2time(&t, "%b %d, %Y", in)) {
568 cp = time2a(t);
569 if (t > time(NULL) + 86400)
570 mandoc_msg(MANDOCERR_DATE_FUTURE, man->parse,
571 ln, pos, cp);
572 else if (*in != '$' && strcmp(in, cp) != 0)
573 mandoc_msg(MANDOCERR_DATE_NORM, man->parse,
574 ln, pos, cp);
575 return cp;
576 }
577
578 /* In man(7), do not warn about the legacy format. */
579
580 if (a2time(&t, "%Y-%m-%d", in) == 0)
581 mandoc_msg(MANDOCERR_DATE_BAD, man->parse, ln, pos, in);
582 else if (t > time(NULL) + 86400)
583 mandoc_msg(MANDOCERR_DATE_FUTURE, man->parse, ln, pos, in);
584 else if (man->macroset == MACROSET_MDOC)
585 mandoc_vmsg(MANDOCERR_DATE_LEGACY, man->parse,
586 ln, pos, "Dd %s", in);
587
588 /* Use any non-mdoc(7) date verbatim. */
589
590 return mandoc_strdup(in);
591 }
592
593 int
594 mandoc_eos(const char *p, size_t sz)
595 {
596 const char *q;
597 int enclosed, found;
598
599 if (0 == sz)
600 return 0;
601
602 /*
603 * End-of-sentence recognition must include situations where
604 * some symbols, such as `)', allow prior EOS punctuation to
605 * propagate outward.
606 */
607
608 enclosed = found = 0;
609 for (q = p + (int)sz - 1; q >= p; q--) {
610 switch (*q) {
611 case '\"':
612 case '\'':
613 case ']':
614 case ')':
615 if (0 == found)
616 enclosed = 1;
617 break;
618 case '.':
619 case '!':
620 case '?':
621 found = 1;
622 break;
623 default:
624 return found &&
625 (!enclosed || isalnum((unsigned char)*q));
626 }
627 }
628
629 return found && !enclosed;
630 }
631
632 /*
633 * Convert a string to a long that may not be <0.
634 * If the string is invalid, or is less than 0, return -1.
635 */
636 int
637 mandoc_strntoi(const char *p, size_t sz, int base)
638 {
639 char buf[32];
640 char *ep;
641 long v;
642
643 if (sz > 31)
644 return -1;
645
646 memcpy(buf, p, sz);
647 buf[(int)sz] = '\0';
648
649 errno = 0;
650 v = strtol(buf, &ep, base);
651
652 if (buf[0] == '\0' || *ep != '\0')
653 return -1;
654
655 if (v > INT_MAX)
656 v = INT_MAX;
657 if (v < INT_MIN)
658 v = INT_MIN;
659
660 return (int)v;
661 }