xref: /openbsd-src/usr.bin/mandoc/man.c (revision f1c726aa8f5cc9c8ab96782dcfdfd0d2875eb42e)
1*f1c726aaSschwarze /* $OpenBSD: man.c,v 1.137 2022/08/16 22:59:48 schwarze Exp $ */
2f73abda9Skristaps /*
3cfaf94d5Sschwarze  * Copyright (c) 2013-2015,2017-2019,2022 Ingo Schwarze <schwarze@openbsd.org>
42791bd1cSschwarze  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
539c2a57eSschwarze  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
6f73abda9Skristaps  *
7f73abda9Skristaps  * Permission to use, copy, modify, and distribute this software for any
8a6464425Sschwarze  * purpose with or without fee is hereby granted, provided that the above
9a6464425Sschwarze  * copyright notice and this permission notice appear in all copies.
10f73abda9Skristaps  *
11d1982c71Sschwarze  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
12a6464425Sschwarze  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13d1982c71Sschwarze  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
14a6464425Sschwarze  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15a6464425Sschwarze  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16a6464425Sschwarze  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17a6464425Sschwarze  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18f73abda9Skristaps  */
19e81991a1Sschwarze #include <sys/types.h>
20e81991a1Sschwarze 
21f73abda9Skristaps #include <assert.h>
22fd5b70b0Sschwarze #include <ctype.h>
23f73abda9Skristaps #include <stdarg.h>
24f73abda9Skristaps #include <stdlib.h>
25f73abda9Skristaps #include <stdio.h>
26f73abda9Skristaps #include <string.h>
27f73abda9Skristaps 
284f4f7972Sschwarze #include "mandoc_aux.h"
29d1982c71Sschwarze #include "mandoc.h"
30d1982c71Sschwarze #include "roff.h"
31d1982c71Sschwarze #include "man.h"
32a66b65d0Sschwarze #include "libmandoc.h"
33fa2127f9Sschwarze #include "roff_int.h"
34d1982c71Sschwarze #include "libman.h"
35f73abda9Skristaps 
3606ce2d6aSschwarze static	char		*man_hasc(char *);
37ede1b9d0Sschwarze static	int		 man_ptext(struct roff_man *, int, char *, int);
38ede1b9d0Sschwarze static	int		 man_pmacro(struct roff_man *, int, char *, int);
39f73abda9Skristaps 
40f73abda9Skristaps 
41f73abda9Skristaps int
man_parseln(struct roff_man * man,int ln,char * buf,int offs)42ede1b9d0Sschwarze man_parseln(struct roff_man *man, int ln, char *buf, int offs)
43f73abda9Skristaps {
44f73abda9Skristaps 
45d1982c71Sschwarze 	if (man->last->type != ROFFT_EQN || ln > man->last->line)
467ead8a4eSschwarze 		man->flags |= MAN_NEWLINE;
47aad3ea0cSschwarze 
48526e306bSschwarze 	return roff_getcontrol(man->roff, buf, &offs) ?
497ead8a4eSschwarze 	    man_pmacro(man, ln, buf, offs) :
50526e306bSschwarze 	    man_ptext(man, ln, buf, offs);
51f73abda9Skristaps }
52f73abda9Skristaps 
5306ce2d6aSschwarze /*
5406ce2d6aSschwarze  * If the string ends with \c, return a pointer to the backslash.
5506ce2d6aSschwarze  * Otherwise, return NULL.
5606ce2d6aSschwarze  */
5706ce2d6aSschwarze static char *
man_hasc(char * start)5806ce2d6aSschwarze man_hasc(char *start)
592791bd1cSschwarze {
6006ce2d6aSschwarze 	char	*cp, *ep;
6106ce2d6aSschwarze 
6206ce2d6aSschwarze 	ep = strchr(start, '\0') - 2;
6306ce2d6aSschwarze 	if (ep < start || ep[0] != '\\' || ep[1] != 'c')
6406ce2d6aSschwarze 		return NULL;
6506ce2d6aSschwarze 	for (cp = ep; cp > start; cp--)
6606ce2d6aSschwarze 		if (cp[-1] != '\\')
6706ce2d6aSschwarze 			break;
6806ce2d6aSschwarze 	return (ep - cp) % 2 ? NULL : ep;
6906ce2d6aSschwarze }
7006ce2d6aSschwarze 
71cfaf94d5Sschwarze /*
72cfaf94d5Sschwarze  * Rewind all open next-line scopes.
73cfaf94d5Sschwarze  */
7463f82d49Sschwarze void
man_descope(struct roff_man * man,int line,int offs,char * start)7506ce2d6aSschwarze man_descope(struct roff_man *man, int line, int offs, char *start)
7606ce2d6aSschwarze {
77*f1c726aaSschwarze 	/* First close out all next-line element scopes, if any. */
782791bd1cSschwarze 
79eae2af84Sschwarze 	if (man->flags & MAN_ELINE) {
8063f82d49Sschwarze 		while (man->last->parent->type != ROFFT_ROOT &&
8163f82d49Sschwarze 		    man_macro(man->last->parent->tok)->flags & MAN_ESCOPED)
82eae2af84Sschwarze 			man_unscope(man, man->last->parent);
8363f82d49Sschwarze 		man->flags &= ~MAN_ELINE;
842791bd1cSschwarze 	}
85*f1c726aaSschwarze 
86*f1c726aaSschwarze 	/* Trailing \c keeps next-line block scope open. */
87*f1c726aaSschwarze 
88*f1c726aaSschwarze 	if (start != NULL && man_hasc(start) != NULL)
89*f1c726aaSschwarze 		return;
90*f1c726aaSschwarze 
91*f1c726aaSschwarze 	/* Close out the next-line block scope, if there is one. */
92*f1c726aaSschwarze 
93eae2af84Sschwarze 	if ( ! (man->flags & MAN_BLINE))
94eae2af84Sschwarze 		return;
95eae2af84Sschwarze 	man_unscope(man, man->last->parent);
96fa2127f9Sschwarze 	roff_body_alloc(man, line, offs, man->last->tok);
9794a3c318Sschwarze 	man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
982791bd1cSschwarze }
992791bd1cSschwarze 
100f73abda9Skristaps static int
man_ptext(struct roff_man * man,int line,char * buf,int offs)101ede1b9d0Sschwarze man_ptext(struct roff_man *man, int line, char *buf, int offs)
102f73abda9Skristaps {
103a4c002ecSschwarze 	int		 i;
1043361af39Sschwarze 	char		*ep;
105f73abda9Skristaps 
1060438bfdfSschwarze 	/* In no-fill mode, whitespace is preserved on text lines. */
107e35cb253Sschwarze 
1080438bfdfSschwarze 	if (man->flags & ROFF_NOFILL) {
10969c34eaaSschwarze 		roff_word_alloc(man, line, offs, buf + offs);
11006ce2d6aSschwarze 		man_descope(man, line, offs, buf + offs);
111526e306bSschwarze 		return 1;
112fbaebbd2Sschwarze 	}
113fbaebbd2Sschwarze 
114eae2af84Sschwarze 	for (i = offs; buf[i] == ' '; i++)
115fbaebbd2Sschwarze 		/* Skip leading whitespace. */ ;
1168521b0bcSschwarze 
117eca45ce0Sschwarze 	/*
1183361af39Sschwarze 	 * Blank lines are ignored in next line scope
1193361af39Sschwarze 	 * and right after headings and cancel preceding \c,
1203361af39Sschwarze 	 * but add a single vertical space elsewhere.
121eca45ce0Sschwarze 	 */
122eca45ce0Sschwarze 
123eae2af84Sschwarze 	if (buf[i] == '\0') {
1243361af39Sschwarze 		if (man->flags & (MAN_ELINE | MAN_BLINE)) {
125a5a5f808Sschwarze 			mandoc_msg(MANDOCERR_BLK_BLANK, line, 0, NULL);
1263361af39Sschwarze 			return 1;
1273361af39Sschwarze 		}
1283361af39Sschwarze 		if (man->last->tok == MAN_SH || man->last->tok == MAN_SS)
1293361af39Sschwarze 			return 1;
13006ce2d6aSschwarze 		if (man->last->type == ROFFT_TEXT &&
13106ce2d6aSschwarze 		    ((ep = man_hasc(man->last->string)) != NULL)) {
1323361af39Sschwarze 			*ep = '\0';
1333361af39Sschwarze 			return 1;
1343361af39Sschwarze 		}
1356561cb23Sschwarze 		roff_elem_alloc(man, line, offs, ROFF_sp);
136ede1b9d0Sschwarze 		man->next = ROFF_NEXT_SIBLING;
137526e306bSschwarze 		return 1;
138fbaebbd2Sschwarze 	}
139fbaebbd2Sschwarze 
140a4c002ecSschwarze 	/*
141a4c002ecSschwarze 	 * Warn if the last un-escaped character is whitespace. Then
142a4c002ecSschwarze 	 * strip away the remaining spaces (tabs stay!).
143a4c002ecSschwarze 	 */
144fbaebbd2Sschwarze 
145a4c002ecSschwarze 	i = (int)strlen(buf);
146a4c002ecSschwarze 	assert(i);
147fbaebbd2Sschwarze 
148a4c002ecSschwarze 	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
149a4c002ecSschwarze 		if (i > 1 && '\\' != buf[i - 2])
150a5a5f808Sschwarze 			mandoc_msg(MANDOCERR_SPACE_EOL, line, i - 1, NULL);
1518521b0bcSschwarze 
152a4c002ecSschwarze 		for (--i; i && ' ' == buf[i]; i--)
153a4c002ecSschwarze 			/* Spin back to non-space. */ ;
154fbaebbd2Sschwarze 
155a4c002ecSschwarze 		/* Jump ahead of escaped whitespace. */
156a4c002ecSschwarze 		i += '\\' == buf[i] ? 2 : 1;
1578521b0bcSschwarze 
158a4c002ecSschwarze 		buf[i] = '\0';
159fbaebbd2Sschwarze 	}
16069c34eaaSschwarze 	roff_word_alloc(man, line, offs, buf + offs);
161fbaebbd2Sschwarze 
162bc49dbe1Sschwarze 	/*
163bc49dbe1Sschwarze 	 * End-of-sentence check.  If the last character is an unescaped
164bc49dbe1Sschwarze 	 * EOS character, then flag the node as being the end of a
165bc49dbe1Sschwarze 	 * sentence.  The front-end will know how to interpret this.
166bc49dbe1Sschwarze 	 */
167bc49dbe1Sschwarze 
168bc49dbe1Sschwarze 	assert(i);
169f29433e9Sschwarze 	if (mandoc_eos(buf, (size_t)i))
170c4b0939cSschwarze 		man->last->flags |= NODE_EOS;
171bc49dbe1Sschwarze 
17206ce2d6aSschwarze 	man_descope(man, line, offs, buf + offs);
173526e306bSschwarze 	return 1;
174f73abda9Skristaps }
175f73abda9Skristaps 
176609398efSschwarze static int
man_pmacro(struct roff_man * man,int ln,char * buf,int offs)177ede1b9d0Sschwarze man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
178f73abda9Skristaps {
1793a0d07afSschwarze 	struct roff_node *n;
1805bb18d70Sschwarze 	const char	*cp;
1816050a3daSschwarze 	size_t		 sz;
1826050a3daSschwarze 	enum roff_tok	 tok;
1836050a3daSschwarze 	int		 ppos;
184bd39224aSschwarze 	int		 bline;
1856050a3daSschwarze 
1866050a3daSschwarze 	/* Determine the line macro. */
187f73abda9Skristaps 
188a35fc07aSschwarze 	ppos = offs;
1896050a3daSschwarze 	tok = TOKEN_NONE;
1906050a3daSschwarze 	for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
1916050a3daSschwarze 		offs++;
1926050a3daSschwarze 	if (sz > 0 && sz < 4)
1936050a3daSschwarze 		tok = roffhash_find(man->manmac, buf + ppos, sz);
1942d6f95d3Sschwarze 	if (tok == TOKEN_NONE) {
195a5a5f808Sschwarze 		mandoc_msg(MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
196526e306bSschwarze 		return 1;
197f73abda9Skristaps 	}
198f73abda9Skristaps 
1995bb18d70Sschwarze 	/* Skip a leading escape sequence or tab. */
2005bb18d70Sschwarze 
2015bb18d70Sschwarze 	switch (buf[offs]) {
2025bb18d70Sschwarze 	case '\\':
2035bb18d70Sschwarze 		cp = buf + offs + 1;
2045bb18d70Sschwarze 		mandoc_escape(&cp, NULL, NULL);
2055bb18d70Sschwarze 		offs = cp - buf;
2065bb18d70Sschwarze 		break;
2075bb18d70Sschwarze 	case '\t':
2085bb18d70Sschwarze 		offs++;
2095bb18d70Sschwarze 		break;
2105bb18d70Sschwarze 	default:
2115bb18d70Sschwarze 		break;
2125bb18d70Sschwarze 	}
2135bb18d70Sschwarze 
2145bb18d70Sschwarze 	/* Jump to the next non-whitespace word. */
215f73abda9Skristaps 
2169b423113Sschwarze 	while (buf[offs] == ' ')
217a35fc07aSschwarze 		offs++;
218f73abda9Skristaps 
219a4c002ecSschwarze 	/*
220a4c002ecSschwarze 	 * Trailing whitespace.  Note that tabs are allowed to be passed
221a4c002ecSschwarze 	 * into the parser as "text", so we only warn about spaces here.
222a4c002ecSschwarze 	 */
2238521b0bcSschwarze 
224eae2af84Sschwarze 	if (buf[offs] == '\0' && buf[offs - 1] == ' ')
225a5a5f808Sschwarze 		mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);
2268521b0bcSschwarze 
22787e50d6cSschwarze 	/*
22845fad4aeSschwarze 	 * Some macros break next-line scopes; otherwise, remember
22945fad4aeSschwarze 	 * whether we are in next-line scope for a block head.
23087e50d6cSschwarze 	 */
231fbaebbd2Sschwarze 
23245fad4aeSschwarze 	man_breakscope(man, tok);
233bd39224aSschwarze 	bline = man->flags & MAN_BLINE;
234a52dde15Sschwarze 
235129c7e1fSschwarze 	/*
236129c7e1fSschwarze 	 * If the line in next-line scope ends with \c, keep the
237129c7e1fSschwarze 	 * next-line scope open for the subsequent input line.
238129c7e1fSschwarze 	 * That is not at all portable, only groff >= 1.22.4
239129c7e1fSschwarze 	 * does it, but *if* this weird idiom occurs in a manual
240129c7e1fSschwarze 	 * page, that's very likely what the author intended.
241129c7e1fSschwarze 	 */
242129c7e1fSschwarze 
24363f82d49Sschwarze 	if (bline && man_hasc(buf + offs))
244129c7e1fSschwarze 		bline = 0;
245129c7e1fSschwarze 
246a52dde15Sschwarze 	/* Call to handler... */
247f73abda9Skristaps 
24816fe0cfcSschwarze 	(*man_macro(tok)->fp)(man, tok, ln, ppos, &offs, buf);
249f73abda9Skristaps 
250f1da507fSschwarze 	/* In quick mode (for mandocdb), abort after the NAME section. */
251f1da507fSschwarze 
252eae2af84Sschwarze 	if (man->quick && tok == MAN_SH) {
2534c84596fSschwarze 		n = man->last;
254d1982c71Sschwarze 		if (n->type == ROFFT_BODY &&
2554c84596fSschwarze 		    strcmp(n->prev->child->string, "NAME"))
256526e306bSschwarze 			return 2;
2574c84596fSschwarze 	}
258f1da507fSschwarze 
25987e50d6cSschwarze 	/*
260bd39224aSschwarze 	 * If we are in a next-line scope for a block head,
261bd39224aSschwarze 	 * close it out now and switch to the body,
262bd39224aSschwarze 	 * unless the next-line scope is allowed to continue.
26387e50d6cSschwarze 	 */
26487e50d6cSschwarze 
26563f82d49Sschwarze 	if (bline == 0 ||
26663f82d49Sschwarze 	    (man->flags & MAN_BLINE) == 0 ||
26763f82d49Sschwarze 	    man->flags & MAN_ELINE ||
26816fe0cfcSschwarze 	    man_macro(tok)->flags & MAN_NSCOPED)
269526e306bSschwarze 		return 1;
270e35cb253Sschwarze 
271eae2af84Sschwarze 	man_unscope(man, man->last->parent);
272fa2127f9Sschwarze 	roff_body_alloc(man, ln, ppos, man->last->tok);
27394a3c318Sschwarze 	man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
274526e306bSschwarze 	return 1;
275f73abda9Skristaps }
276f73abda9Skristaps 
277cfaf94d5Sschwarze /*
278cfaf94d5Sschwarze  * Rewind open next-line scopes
279cfaf94d5Sschwarze  * unless the tok request or macro is allowed inside them.
280cfaf94d5Sschwarze  */
28145fad4aeSschwarze void
man_breakscope(struct roff_man * man,int tok)282ede1b9d0Sschwarze man_breakscope(struct roff_man *man, int tok)
28345fad4aeSschwarze {
2843a0d07afSschwarze 	struct roff_node *n;
28545fad4aeSschwarze 
28645fad4aeSschwarze 	/*
28745fad4aeSschwarze 	 * An element next line scope is open,
28845fad4aeSschwarze 	 * and the new macro is not allowed inside elements.
28945fad4aeSschwarze 	 * Delete the element that is being broken.
29045fad4aeSschwarze 	 */
29145fad4aeSschwarze 
292885ec428Sschwarze 	if (man->flags & MAN_ELINE && (tok < MAN_TH ||
29316fe0cfcSschwarze 	    (man_macro(tok)->flags & MAN_NSCOPED) == 0)) {
29445fad4aeSschwarze 		n = man->last;
295c4db4099Sschwarze 		if (n->type == ROFFT_TEXT)
296c4db4099Sschwarze 			n = n->parent;
297c4db4099Sschwarze 		if (n->tok < MAN_TH ||
29863f82d49Sschwarze 		    (man_macro(n->tok)->flags & (MAN_NSCOPED | MAN_ESCOPED))
29963f82d49Sschwarze 		     == MAN_NSCOPED)
30045fad4aeSschwarze 			n = n->parent;
301cfaf94d5Sschwarze 		for (;;) {
302a5a5f808Sschwarze 			mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
303a5a5f808Sschwarze 			    "%s breaks %s", roff_name[tok], roff_name[n->tok]);
304cfaf94d5Sschwarze 			if (n->parent->type != ROFFT_ELEM ||
305cfaf94d5Sschwarze 			    (man_macro(n->parent->tok)->flags &
306cfaf94d5Sschwarze 			     MAN_ESCOPED) == 0)
307cfaf94d5Sschwarze 				break;
308cfaf94d5Sschwarze 			n = n->parent;
309cfaf94d5Sschwarze 		}
310fa2127f9Sschwarze 		roff_node_delete(man, n);
31145fad4aeSschwarze 		man->flags &= ~MAN_ELINE;
31245fad4aeSschwarze 	}
31345fad4aeSschwarze 
31445fad4aeSschwarze 	/*
3159d496861Sschwarze 	 * Weird special case:
3169d496861Sschwarze 	 * Switching fill mode closes section headers.
3179d496861Sschwarze 	 */
3189d496861Sschwarze 
3199d496861Sschwarze 	if (man->flags & MAN_BLINE &&
3200438bfdfSschwarze 	    (tok == ROFF_nf || tok == ROFF_fi) &&
3219d496861Sschwarze 	    (man->last->tok == MAN_SH || man->last->tok == MAN_SS)) {
3229d496861Sschwarze 		n = man->last;
3239d496861Sschwarze 		man_unscope(man, n);
3249d496861Sschwarze 		roff_body_alloc(man, n->line, n->pos, n->tok);
32594a3c318Sschwarze 		man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
3269d496861Sschwarze 	}
3279d496861Sschwarze 
3289d496861Sschwarze 	/*
32945fad4aeSschwarze 	 * A block header next line scope is open,
33045fad4aeSschwarze 	 * and the new macro is not allowed inside block headers.
33145fad4aeSschwarze 	 * Delete the block that is being broken.
33245fad4aeSschwarze 	 */
33345fad4aeSschwarze 
3340438bfdfSschwarze 	if (man->flags & MAN_BLINE && tok != ROFF_nf && tok != ROFF_fi &&
3350438bfdfSschwarze 	    (tok < MAN_TH || man_macro(tok)->flags & MAN_XSCOPE)) {
33645fad4aeSschwarze 		n = man->last;
337d1982c71Sschwarze 		if (n->type == ROFFT_TEXT)
33845fad4aeSschwarze 			n = n->parent;
339c4db4099Sschwarze 		if (n->tok < MAN_TH ||
34063f82d49Sschwarze 		    (man_macro(n->tok)->flags & MAN_XSCOPE) == 0)
34145fad4aeSschwarze 			n = n->parent;
34245fad4aeSschwarze 
343d1982c71Sschwarze 		assert(n->type == ROFFT_HEAD);
34445fad4aeSschwarze 		n = n->parent;
345d1982c71Sschwarze 		assert(n->type == ROFFT_BLOCK);
34663f82d49Sschwarze 		assert(man_macro(n->tok)->flags & MAN_BSCOPED);
34745fad4aeSschwarze 
348a5a5f808Sschwarze 		mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
349a5a5f808Sschwarze 		    "%s breaks %s", roff_name[tok], roff_name[n->tok]);
35045fad4aeSschwarze 
351fa2127f9Sschwarze 		roff_node_delete(man, n);
35294a3c318Sschwarze 		man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
35345fad4aeSschwarze 	}
35445fad4aeSschwarze }
355