1*c1c95addSBrooks Davis /* $Id: mandoc.c,v 1.121 2022/05/19 15:37:47 schwarze Exp $ */
261d06d6bSBaptiste Daroussin /*
3*c1c95addSBrooks Davis * Copyright (c) 2010, 2011, 2015, 2017, 2018, 2019, 2020, 2021
4*c1c95addSBrooks Davis * Ingo Schwarze <schwarze@openbsd.org>
5*c1c95addSBrooks Davis * Copyright (c) 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
661d06d6bSBaptiste Daroussin *
761d06d6bSBaptiste Daroussin * Permission to use, copy, modify, and distribute this software for any
861d06d6bSBaptiste Daroussin * purpose with or without fee is hereby granted, provided that the above
961d06d6bSBaptiste Daroussin * copyright notice and this permission notice appear in all copies.
1061d06d6bSBaptiste Daroussin *
1161d06d6bSBaptiste Daroussin * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1261d06d6bSBaptiste Daroussin * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1361d06d6bSBaptiste Daroussin * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1461d06d6bSBaptiste Daroussin * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
1561d06d6bSBaptiste Daroussin * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
1661d06d6bSBaptiste Daroussin * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
1761d06d6bSBaptiste Daroussin * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18*c1c95addSBrooks Davis *
19*c1c95addSBrooks Davis * Utility functions to handle end of sentence punctuation
20*c1c95addSBrooks Davis * and dates and times, for use by mdoc(7) and man(7) parsers.
21*c1c95addSBrooks Davis * Utility functions to handle fonts and numbers,
22*c1c95addSBrooks Davis * for use by mandoc(1) parsers and formatters.
2361d06d6bSBaptiste Daroussin */
2461d06d6bSBaptiste Daroussin #include "config.h"
2561d06d6bSBaptiste Daroussin
2661d06d6bSBaptiste Daroussin #include <sys/types.h>
2761d06d6bSBaptiste Daroussin
2861d06d6bSBaptiste Daroussin #include <assert.h>
2961d06d6bSBaptiste Daroussin #include <ctype.h>
3061d06d6bSBaptiste Daroussin #include <errno.h>
3161d06d6bSBaptiste Daroussin #include <limits.h>
3261d06d6bSBaptiste Daroussin #include <stdlib.h>
3361d06d6bSBaptiste Daroussin #include <stdio.h>
3461d06d6bSBaptiste Daroussin #include <string.h>
3561d06d6bSBaptiste Daroussin #include <time.h>
3661d06d6bSBaptiste Daroussin
3761d06d6bSBaptiste Daroussin #include "mandoc_aux.h"
3861d06d6bSBaptiste Daroussin #include "mandoc.h"
3961d06d6bSBaptiste Daroussin #include "roff.h"
4061d06d6bSBaptiste Daroussin #include "libmandoc.h"
417295610fSBaptiste Daroussin #include "roff_int.h"
4261d06d6bSBaptiste Daroussin
4361d06d6bSBaptiste Daroussin static int a2time(time_t *, const char *, const char *);
4461d06d6bSBaptiste Daroussin static char *time2a(time_t);
4561d06d6bSBaptiste Daroussin
4661d06d6bSBaptiste Daroussin
4761d06d6bSBaptiste Daroussin enum mandoc_esc
mandoc_font(const char * cp,int sz)487295610fSBaptiste Daroussin mandoc_font(const char *cp, int sz)
497295610fSBaptiste Daroussin {
507295610fSBaptiste Daroussin switch (sz) {
517295610fSBaptiste Daroussin case 0:
527295610fSBaptiste Daroussin return ESCAPE_FONTPREV;
537295610fSBaptiste Daroussin case 1:
547295610fSBaptiste Daroussin switch (cp[0]) {
557295610fSBaptiste Daroussin case 'B':
567295610fSBaptiste Daroussin case '3':
577295610fSBaptiste Daroussin return ESCAPE_FONTBOLD;
587295610fSBaptiste Daroussin case 'I':
597295610fSBaptiste Daroussin case '2':
607295610fSBaptiste Daroussin return ESCAPE_FONTITALIC;
617295610fSBaptiste Daroussin case 'P':
627295610fSBaptiste Daroussin return ESCAPE_FONTPREV;
637295610fSBaptiste Daroussin case 'R':
647295610fSBaptiste Daroussin case '1':
657295610fSBaptiste Daroussin return ESCAPE_FONTROMAN;
667295610fSBaptiste Daroussin case '4':
677295610fSBaptiste Daroussin return ESCAPE_FONTBI;
687295610fSBaptiste Daroussin default:
697295610fSBaptiste Daroussin return ESCAPE_ERROR;
707295610fSBaptiste Daroussin }
717295610fSBaptiste Daroussin case 2:
727295610fSBaptiste Daroussin switch (cp[0]) {
737295610fSBaptiste Daroussin case 'B':
747295610fSBaptiste Daroussin switch (cp[1]) {
757295610fSBaptiste Daroussin case 'I':
767295610fSBaptiste Daroussin return ESCAPE_FONTBI;
777295610fSBaptiste Daroussin default:
787295610fSBaptiste Daroussin return ESCAPE_ERROR;
797295610fSBaptiste Daroussin }
807295610fSBaptiste Daroussin case 'C':
817295610fSBaptiste Daroussin switch (cp[1]) {
827295610fSBaptiste Daroussin case 'B':
836d38604fSBaptiste Daroussin return ESCAPE_FONTCB;
847295610fSBaptiste Daroussin case 'I':
856d38604fSBaptiste Daroussin return ESCAPE_FONTCI;
867295610fSBaptiste Daroussin case 'R':
877295610fSBaptiste Daroussin case 'W':
886d38604fSBaptiste Daroussin return ESCAPE_FONTCR;
897295610fSBaptiste Daroussin default:
907295610fSBaptiste Daroussin return ESCAPE_ERROR;
917295610fSBaptiste Daroussin }
927295610fSBaptiste Daroussin default:
937295610fSBaptiste Daroussin return ESCAPE_ERROR;
947295610fSBaptiste Daroussin }
957295610fSBaptiste Daroussin default:
967295610fSBaptiste Daroussin return ESCAPE_ERROR;
977295610fSBaptiste Daroussin }
987295610fSBaptiste Daroussin }
997295610fSBaptiste Daroussin
10061d06d6bSBaptiste Daroussin static int
a2time(time_t * t,const char * fmt,const char * p)10161d06d6bSBaptiste Daroussin a2time(time_t *t, const char *fmt, const char *p)
10261d06d6bSBaptiste Daroussin {
10361d06d6bSBaptiste Daroussin struct tm tm;
10461d06d6bSBaptiste Daroussin char *pp;
10561d06d6bSBaptiste Daroussin
10661d06d6bSBaptiste Daroussin memset(&tm, 0, sizeof(struct tm));
10761d06d6bSBaptiste Daroussin
10861d06d6bSBaptiste Daroussin pp = NULL;
10961d06d6bSBaptiste Daroussin #if HAVE_STRPTIME
11061d06d6bSBaptiste Daroussin pp = strptime(p, fmt, &tm);
11161d06d6bSBaptiste Daroussin #endif
11261d06d6bSBaptiste Daroussin if (NULL != pp && '\0' == *pp) {
11361d06d6bSBaptiste Daroussin *t = mktime(&tm);
11461d06d6bSBaptiste Daroussin return 1;
11561d06d6bSBaptiste Daroussin }
11661d06d6bSBaptiste Daroussin
11761d06d6bSBaptiste Daroussin return 0;
11861d06d6bSBaptiste Daroussin }
11961d06d6bSBaptiste Daroussin
12061d06d6bSBaptiste Daroussin static char *
time2a(time_t t)12161d06d6bSBaptiste Daroussin time2a(time_t t)
12261d06d6bSBaptiste Daroussin {
12361d06d6bSBaptiste Daroussin struct tm *tm;
12461d06d6bSBaptiste Daroussin char *buf, *p;
12561d06d6bSBaptiste Daroussin size_t ssz;
12661d06d6bSBaptiste Daroussin int isz;
12761d06d6bSBaptiste Daroussin
12845a5aec3SBaptiste Daroussin buf = NULL;
12961d06d6bSBaptiste Daroussin tm = localtime(&t);
13061d06d6bSBaptiste Daroussin if (tm == NULL)
13145a5aec3SBaptiste Daroussin goto fail;
13261d06d6bSBaptiste Daroussin
13361d06d6bSBaptiste Daroussin /*
13461d06d6bSBaptiste Daroussin * Reserve space:
13561d06d6bSBaptiste Daroussin * up to 9 characters for the month (September) + blank
13661d06d6bSBaptiste Daroussin * up to 2 characters for the day + comma + blank
13761d06d6bSBaptiste Daroussin * 4 characters for the year and a terminating '\0'
13861d06d6bSBaptiste Daroussin */
13961d06d6bSBaptiste Daroussin
14061d06d6bSBaptiste Daroussin p = buf = mandoc_malloc(10 + 4 + 4 + 1);
14161d06d6bSBaptiste Daroussin
14261d06d6bSBaptiste Daroussin if ((ssz = strftime(p, 10 + 1, "%B ", tm)) == 0)
14361d06d6bSBaptiste Daroussin goto fail;
14461d06d6bSBaptiste Daroussin p += (int)ssz;
14561d06d6bSBaptiste Daroussin
14661d06d6bSBaptiste Daroussin /*
14761d06d6bSBaptiste Daroussin * The output format is just "%d" here, not "%2d" or "%02d".
14861d06d6bSBaptiste Daroussin * That's also the reason why we can't just format the
14961d06d6bSBaptiste Daroussin * date as a whole with "%B %e, %Y" or "%B %d, %Y".
15061d06d6bSBaptiste Daroussin * Besides, the present approach is less prone to buffer
15161d06d6bSBaptiste Daroussin * overflows, in case anybody should ever introduce the bug
15261d06d6bSBaptiste Daroussin * of looking at LC_TIME.
15361d06d6bSBaptiste Daroussin */
15461d06d6bSBaptiste Daroussin
15545a5aec3SBaptiste Daroussin isz = snprintf(p, 4 + 1, "%d, ", tm->tm_mday);
15645a5aec3SBaptiste Daroussin if (isz < 0 || isz > 4)
15761d06d6bSBaptiste Daroussin goto fail;
15861d06d6bSBaptiste Daroussin p += isz;
15961d06d6bSBaptiste Daroussin
16061d06d6bSBaptiste Daroussin if (strftime(p, 4 + 1, "%Y", tm) == 0)
16161d06d6bSBaptiste Daroussin goto fail;
16261d06d6bSBaptiste Daroussin return buf;
16361d06d6bSBaptiste Daroussin
16461d06d6bSBaptiste Daroussin fail:
16561d06d6bSBaptiste Daroussin free(buf);
16645a5aec3SBaptiste Daroussin return mandoc_strdup("");
16761d06d6bSBaptiste Daroussin }
16861d06d6bSBaptiste Daroussin
16961d06d6bSBaptiste Daroussin char *
mandoc_normdate(struct roff_node * nch,struct roff_node * nbl)1706d38604fSBaptiste Daroussin mandoc_normdate(struct roff_node *nch, struct roff_node *nbl)
17161d06d6bSBaptiste Daroussin {
17261d06d6bSBaptiste Daroussin char *cp;
17361d06d6bSBaptiste Daroussin time_t t;
17461d06d6bSBaptiste Daroussin
1756d38604fSBaptiste Daroussin /* No date specified. */
17645a5aec3SBaptiste Daroussin
1776d38604fSBaptiste Daroussin if (nch == NULL) {
1786d38604fSBaptiste Daroussin if (nbl == NULL)
1796d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_MISSING, 0, 0, NULL);
1806d38604fSBaptiste Daroussin else
1816d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_MISSING, nbl->line,
1826d38604fSBaptiste Daroussin nbl->pos, "%s", roff_name[nbl->tok]);
1836d38604fSBaptiste Daroussin return mandoc_strdup("");
1846d38604fSBaptiste Daroussin }
1856d38604fSBaptiste Daroussin if (*nch->string == '\0') {
1866d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_MISSING, nch->line,
1876d38604fSBaptiste Daroussin nch->pos, "%s", roff_name[nbl->tok]);
1886d38604fSBaptiste Daroussin return mandoc_strdup("");
1896d38604fSBaptiste Daroussin }
1906d38604fSBaptiste Daroussin if (strcmp(nch->string, "$" "Mdocdate$") == 0)
19161d06d6bSBaptiste Daroussin return time2a(time(NULL));
19261d06d6bSBaptiste Daroussin
19361d06d6bSBaptiste Daroussin /* Valid mdoc(7) date format. */
19461d06d6bSBaptiste Daroussin
1956d38604fSBaptiste Daroussin if (a2time(&t, "$" "Mdocdate: %b %d %Y $", nch->string) ||
1966d38604fSBaptiste Daroussin a2time(&t, "%b %d, %Y", nch->string)) {
19761d06d6bSBaptiste Daroussin cp = time2a(t);
19861d06d6bSBaptiste Daroussin if (t > time(NULL) + 86400)
1996d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_FUTURE, nch->line,
2006d38604fSBaptiste Daroussin nch->pos, "%s %s", roff_name[nbl->tok], cp);
2016d38604fSBaptiste Daroussin else if (*nch->string != '$' &&
2026d38604fSBaptiste Daroussin strcmp(nch->string, cp) != 0)
2036d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_NORM, nch->line,
2046d38604fSBaptiste Daroussin nch->pos, "%s %s", roff_name[nbl->tok], cp);
20561d06d6bSBaptiste Daroussin return cp;
20661d06d6bSBaptiste Daroussin }
20761d06d6bSBaptiste Daroussin
20861d06d6bSBaptiste Daroussin /* In man(7), do not warn about the legacy format. */
20961d06d6bSBaptiste Daroussin
2106d38604fSBaptiste Daroussin if (a2time(&t, "%Y-%m-%d", nch->string) == 0)
2116d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_BAD, nch->line, nch->pos,
2126d38604fSBaptiste Daroussin "%s %s", roff_name[nbl->tok], nch->string);
21361d06d6bSBaptiste Daroussin else if (t > time(NULL) + 86400)
2146d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_FUTURE, nch->line, nch->pos,
2156d38604fSBaptiste Daroussin "%s %s", roff_name[nbl->tok], nch->string);
2166d38604fSBaptiste Daroussin else if (nbl->tok == MDOC_Dd)
2176d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_DATE_LEGACY, nch->line, nch->pos,
2186d38604fSBaptiste Daroussin "Dd %s", nch->string);
21961d06d6bSBaptiste Daroussin
22061d06d6bSBaptiste Daroussin /* Use any non-mdoc(7) date verbatim. */
22161d06d6bSBaptiste Daroussin
2226d38604fSBaptiste Daroussin return mandoc_strdup(nch->string);
22361d06d6bSBaptiste Daroussin }
22461d06d6bSBaptiste Daroussin
22561d06d6bSBaptiste Daroussin int
mandoc_eos(const char * p,size_t sz)22661d06d6bSBaptiste Daroussin mandoc_eos(const char *p, size_t sz)
22761d06d6bSBaptiste Daroussin {
22861d06d6bSBaptiste Daroussin const char *q;
22961d06d6bSBaptiste Daroussin int enclosed, found;
23061d06d6bSBaptiste Daroussin
23161d06d6bSBaptiste Daroussin if (0 == sz)
23261d06d6bSBaptiste Daroussin return 0;
23361d06d6bSBaptiste Daroussin
23461d06d6bSBaptiste Daroussin /*
23561d06d6bSBaptiste Daroussin * End-of-sentence recognition must include situations where
23661d06d6bSBaptiste Daroussin * some symbols, such as `)', allow prior EOS punctuation to
23761d06d6bSBaptiste Daroussin * propagate outward.
23861d06d6bSBaptiste Daroussin */
23961d06d6bSBaptiste Daroussin
24061d06d6bSBaptiste Daroussin enclosed = found = 0;
24161d06d6bSBaptiste Daroussin for (q = p + (int)sz - 1; q >= p; q--) {
24261d06d6bSBaptiste Daroussin switch (*q) {
24361d06d6bSBaptiste Daroussin case '\"':
24461d06d6bSBaptiste Daroussin case '\'':
24561d06d6bSBaptiste Daroussin case ']':
24661d06d6bSBaptiste Daroussin case ')':
24761d06d6bSBaptiste Daroussin if (0 == found)
24861d06d6bSBaptiste Daroussin enclosed = 1;
24961d06d6bSBaptiste Daroussin break;
25061d06d6bSBaptiste Daroussin case '.':
25161d06d6bSBaptiste Daroussin case '!':
25261d06d6bSBaptiste Daroussin case '?':
25361d06d6bSBaptiste Daroussin found = 1;
25461d06d6bSBaptiste Daroussin break;
25561d06d6bSBaptiste Daroussin default:
25661d06d6bSBaptiste Daroussin return found &&
25761d06d6bSBaptiste Daroussin (!enclosed || isalnum((unsigned char)*q));
25861d06d6bSBaptiste Daroussin }
25961d06d6bSBaptiste Daroussin }
26061d06d6bSBaptiste Daroussin
26161d06d6bSBaptiste Daroussin return found && !enclosed;
26261d06d6bSBaptiste Daroussin }
26361d06d6bSBaptiste Daroussin
26461d06d6bSBaptiste Daroussin /*
26561d06d6bSBaptiste Daroussin * Convert a string to a long that may not be <0.
26661d06d6bSBaptiste Daroussin * If the string is invalid, or is less than 0, return -1.
26761d06d6bSBaptiste Daroussin */
26861d06d6bSBaptiste Daroussin int
mandoc_strntoi(const char * p,size_t sz,int base)26961d06d6bSBaptiste Daroussin mandoc_strntoi(const char *p, size_t sz, int base)
27061d06d6bSBaptiste Daroussin {
27161d06d6bSBaptiste Daroussin char buf[32];
27261d06d6bSBaptiste Daroussin char *ep;
27361d06d6bSBaptiste Daroussin long v;
27461d06d6bSBaptiste Daroussin
27561d06d6bSBaptiste Daroussin if (sz > 31)
27661d06d6bSBaptiste Daroussin return -1;
27761d06d6bSBaptiste Daroussin
27861d06d6bSBaptiste Daroussin memcpy(buf, p, sz);
27961d06d6bSBaptiste Daroussin buf[(int)sz] = '\0';
28061d06d6bSBaptiste Daroussin
28161d06d6bSBaptiste Daroussin errno = 0;
28261d06d6bSBaptiste Daroussin v = strtol(buf, &ep, base);
28361d06d6bSBaptiste Daroussin
28461d06d6bSBaptiste Daroussin if (buf[0] == '\0' || *ep != '\0')
28561d06d6bSBaptiste Daroussin return -1;
28661d06d6bSBaptiste Daroussin
28761d06d6bSBaptiste Daroussin if (v > INT_MAX)
28861d06d6bSBaptiste Daroussin v = INT_MAX;
28961d06d6bSBaptiste Daroussin if (v < INT_MIN)
29061d06d6bSBaptiste Daroussin v = INT_MIN;
29161d06d6bSBaptiste Daroussin
29261d06d6bSBaptiste Daroussin return (int)v;
29361d06d6bSBaptiste Daroussin }
294