xref: /dflybsd-src/contrib/mdocml/man.c (revision 54ba96075f5891e4574304da6ba88f1a1afe520b)
1*54ba9607SSascha Wildner /*	$Id: man.c,v 1.187 2019/01/05 00:36:50 schwarze Exp $ */
280387638SSascha Wildner /*
360e1e752SSascha Wildner  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4*54ba9607SSascha Wildner  * Copyright (c) 2013-2015, 2017-2019 Ingo Schwarze <schwarze@openbsd.org>
5070c62a6SFranco Fichtner  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
680387638SSascha Wildner  *
780387638SSascha Wildner  * Permission to use, copy, modify, and distribute this software for any
880387638SSascha Wildner  * purpose with or without fee is hereby granted, provided that the above
980387638SSascha Wildner  * copyright notice and this permission notice appear in all copies.
1080387638SSascha Wildner  *
11*54ba9607SSascha Wildner  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1280387638SSascha Wildner  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13*54ba9607SSascha Wildner  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1480387638SSascha Wildner  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
1580387638SSascha Wildner  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
1680387638SSascha Wildner  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
1780387638SSascha Wildner  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1880387638SSascha Wildner  */
1980387638SSascha Wildner #include "config.h"
2080387638SSascha Wildner 
2180387638SSascha Wildner #include <sys/types.h>
2280387638SSascha Wildner 
2380387638SSascha Wildner #include <assert.h>
24070c62a6SFranco Fichtner #include <ctype.h>
2580387638SSascha Wildner #include <stdarg.h>
2680387638SSascha Wildner #include <stdlib.h>
2780387638SSascha Wildner #include <stdio.h>
2880387638SSascha Wildner #include <string.h>
2980387638SSascha Wildner 
30070c62a6SFranco Fichtner #include "mandoc_aux.h"
31*54ba9607SSascha Wildner #include "mandoc.h"
32*54ba9607SSascha Wildner #include "roff.h"
33*54ba9607SSascha Wildner #include "man.h"
3480387638SSascha Wildner #include "libmandoc.h"
35*54ba9607SSascha Wildner #include "roff_int.h"
36*54ba9607SSascha Wildner #include "libman.h"
3780387638SSascha Wildner 
38*54ba9607SSascha Wildner static	char		*man_hasc(char *);
39*54ba9607SSascha Wildner static	int		 man_ptext(struct roff_man *, int, char *, int);
40*54ba9607SSascha Wildner static	int		 man_pmacro(struct roff_man *, int, char *, int);
4180387638SSascha Wildner 
4280387638SSascha Wildner 
4380387638SSascha Wildner int
man_parseln(struct roff_man * man,int ln,char * buf,int offs)44*54ba9607SSascha Wildner man_parseln(struct roff_man *man, int ln, char *buf, int offs)
4580387638SSascha Wildner {
4680387638SSascha Wildner 
47*54ba9607SSascha Wildner 	if (man->last->type != ROFFT_EQN || ln > man->last->line)
48f88b6c16SFranco Fichtner 		man->flags |= MAN_NEWLINE;
4960e1e752SSascha Wildner 
50*54ba9607SSascha Wildner 	return roff_getcontrol(man->roff, buf, &offs) ?
51f88b6c16SFranco Fichtner 	    man_pmacro(man, ln, buf, offs) :
52*54ba9607SSascha Wildner 	    man_ptext(man, ln, buf, offs);
5380387638SSascha Wildner }
5480387638SSascha Wildner 
5580387638SSascha Wildner /*
56*54ba9607SSascha Wildner  * If the string ends with \c, return a pointer to the backslash.
57*54ba9607SSascha Wildner  * Otherwise, return NULL.
5880387638SSascha Wildner  */
59*54ba9607SSascha Wildner static char *
man_hasc(char * start)60*54ba9607SSascha Wildner man_hasc(char *start)
6180387638SSascha Wildner {
62*54ba9607SSascha Wildner 	char	*cp, *ep;
6380387638SSascha Wildner 
64*54ba9607SSascha Wildner 	ep = strchr(start, '\0') - 2;
65*54ba9607SSascha Wildner 	if (ep < start || ep[0] != '\\' || ep[1] != 'c')
66*54ba9607SSascha Wildner 		return NULL;
67*54ba9607SSascha Wildner 	for (cp = ep; cp > start; cp--)
68*54ba9607SSascha Wildner 		if (cp[-1] != '\\')
69*54ba9607SSascha Wildner 			break;
70*54ba9607SSascha Wildner 	return (ep - cp) % 2 ? NULL : ep;
7180387638SSascha Wildner }
7280387638SSascha Wildner 
7380387638SSascha Wildner void
man_descope(struct roff_man * man,int line,int offs,char * start)74*54ba9607SSascha Wildner man_descope(struct roff_man *man, int line, int offs, char *start)
7580387638SSascha Wildner {
76*54ba9607SSascha Wildner 	/* Trailing \c keeps next-line scope open. */
7780387638SSascha Wildner 
78*54ba9607SSascha Wildner 	if (start != NULL && man_hasc(start) != NULL)
79*54ba9607SSascha Wildner 		return;
8080387638SSascha Wildner 
8180387638SSascha Wildner 	/*
8280387638SSascha Wildner 	 * Co-ordinate what happens with having a next-line scope open:
83*54ba9607SSascha Wildner 	 * first close out the element scopes (if applicable),
84*54ba9607SSascha Wildner 	 * then close out the block scope (also if applicable).
8580387638SSascha Wildner 	 */
8680387638SSascha Wildner 
87*54ba9607SSascha Wildner 	if (man->flags & MAN_ELINE) {
88*54ba9607SSascha Wildner 		while (man->last->parent->type != ROFFT_ROOT &&
89*54ba9607SSascha Wildner 		    man_macro(man->last->parent->tok)->flags & MAN_ESCOPED)
90*54ba9607SSascha Wildner 			man_unscope(man, man->last->parent);
91f88b6c16SFranco Fichtner 		man->flags &= ~MAN_ELINE;
9280387638SSascha Wildner 	}
93*54ba9607SSascha Wildner 	if ( ! (man->flags & MAN_BLINE))
94*54ba9607SSascha Wildner 		return;
95*54ba9607SSascha Wildner 	man_unscope(man, man->last->parent);
96*54ba9607SSascha Wildner 	roff_body_alloc(man, line, offs, man->last->tok);
97*54ba9607SSascha Wildner 	man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
9880387638SSascha Wildner }
9980387638SSascha Wildner 
10080387638SSascha Wildner static int
man_ptext(struct roff_man * man,int line,char * buf,int offs)101*54ba9607SSascha Wildner man_ptext(struct roff_man *man, int line, char *buf, int offs)
10280387638SSascha Wildner {
10380387638SSascha Wildner 	int		 i;
104*54ba9607SSascha Wildner 	char		*ep;
10580387638SSascha Wildner 
106*54ba9607SSascha Wildner 	/* In no-fill mode, whitespace is preserved on text lines. */
10780387638SSascha Wildner 
108*54ba9607SSascha Wildner 	if (man->flags & ROFF_NOFILL) {
109*54ba9607SSascha Wildner 		roff_word_alloc(man, line, offs, buf + offs);
110*54ba9607SSascha Wildner 		man_descope(man, line, offs, buf + offs);
111*54ba9607SSascha Wildner 		return 1;
11280387638SSascha Wildner 	}
11380387638SSascha Wildner 
114*54ba9607SSascha Wildner 	for (i = offs; buf[i] == ' '; i++)
11580387638SSascha Wildner 		/* Skip leading whitespace. */ ;
11680387638SSascha Wildner 
1177888c61dSFranco Fichtner 	/*
118*54ba9607SSascha Wildner 	 * Blank lines are ignored in next line scope
119*54ba9607SSascha Wildner 	 * and right after headings and cancel preceding \c,
1207888c61dSFranco Fichtner 	 * but add a single vertical space elsewhere.
1217888c61dSFranco Fichtner 	 */
1227888c61dSFranco Fichtner 
123*54ba9607SSascha Wildner 	if (buf[i] == '\0') {
124*54ba9607SSascha Wildner 		if (man->flags & (MAN_ELINE | MAN_BLINE)) {
125*54ba9607SSascha Wildner 			mandoc_msg(MANDOCERR_BLK_BLANK, line, 0, NULL);
126*54ba9607SSascha Wildner 			return 1;
1277888c61dSFranco Fichtner 		}
128*54ba9607SSascha Wildner 		if (man->last->tok == MAN_SH || man->last->tok == MAN_SS)
129*54ba9607SSascha Wildner 			return 1;
130*54ba9607SSascha Wildner 		if (man->last->type == ROFFT_TEXT &&
131*54ba9607SSascha Wildner 		    ((ep = man_hasc(man->last->string)) != NULL)) {
132*54ba9607SSascha Wildner 			*ep = '\0';
133*54ba9607SSascha Wildner 			return 1;
134*54ba9607SSascha Wildner 		}
135*54ba9607SSascha Wildner 		roff_elem_alloc(man, line, offs, ROFF_sp);
136*54ba9607SSascha Wildner 		man->next = ROFF_NEXT_SIBLING;
137*54ba9607SSascha Wildner 		return 1;
13880387638SSascha Wildner 	}
13980387638SSascha Wildner 
14080387638SSascha Wildner 	/*
14180387638SSascha Wildner 	 * Warn if the last un-escaped character is whitespace. Then
14280387638SSascha Wildner 	 * strip away the remaining spaces (tabs stay!).
14380387638SSascha Wildner 	 */
14480387638SSascha Wildner 
14580387638SSascha Wildner 	i = (int)strlen(buf);
14680387638SSascha Wildner 	assert(i);
14780387638SSascha Wildner 
14880387638SSascha Wildner 	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
14980387638SSascha Wildner 		if (i > 1 && '\\' != buf[i - 2])
150*54ba9607SSascha Wildner 			mandoc_msg(MANDOCERR_SPACE_EOL, line, i - 1, NULL);
15180387638SSascha Wildner 
15280387638SSascha Wildner 		for (--i; i && ' ' == buf[i]; i--)
15380387638SSascha Wildner 			/* Spin back to non-space. */ ;
15480387638SSascha Wildner 
15580387638SSascha Wildner 		/* Jump ahead of escaped whitespace. */
15680387638SSascha Wildner 		i += '\\' == buf[i] ? 2 : 1;
15780387638SSascha Wildner 
15880387638SSascha Wildner 		buf[i] = '\0';
15980387638SSascha Wildner 	}
160*54ba9607SSascha Wildner 	roff_word_alloc(man, line, offs, buf + offs);
16180387638SSascha Wildner 
16280387638SSascha Wildner 	/*
16380387638SSascha Wildner 	 * End-of-sentence check.  If the last character is an unescaped
16480387638SSascha Wildner 	 * EOS character, then flag the node as being the end of a
16580387638SSascha Wildner 	 * sentence.  The front-end will know how to interpret this.
16680387638SSascha Wildner 	 */
16780387638SSascha Wildner 
16880387638SSascha Wildner 	assert(i);
169070c62a6SFranco Fichtner 	if (mandoc_eos(buf, (size_t)i))
170*54ba9607SSascha Wildner 		man->last->flags |= NODE_EOS;
17180387638SSascha Wildner 
172*54ba9607SSascha Wildner 	man_descope(man, line, offs, buf + offs);
173*54ba9607SSascha Wildner 	return 1;
17480387638SSascha Wildner }
17580387638SSascha Wildner 
17680387638SSascha Wildner static int
man_pmacro(struct roff_man * man,int ln,char * buf,int offs)177*54ba9607SSascha Wildner man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
17880387638SSascha Wildner {
179*54ba9607SSascha Wildner 	struct roff_node *n;
180*54ba9607SSascha Wildner 	const char	*cp;
181*54ba9607SSascha Wildner 	size_t		 sz;
182*54ba9607SSascha Wildner 	enum roff_tok	 tok;
183*54ba9607SSascha Wildner 	int		 ppos;
184070c62a6SFranco Fichtner 	int		 bline;
18580387638SSascha Wildner 
186*54ba9607SSascha Wildner 	/* Determine the line macro. */
18780387638SSascha Wildner 
18860e1e752SSascha Wildner 	ppos = offs;
189*54ba9607SSascha Wildner 	tok = TOKEN_NONE;
190*54ba9607SSascha Wildner 	for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
191*54ba9607SSascha Wildner 		offs++;
192*54ba9607SSascha Wildner 	if (sz > 0 && sz < 4)
193*54ba9607SSascha Wildner 		tok = roffhash_find(man->manmac, buf + ppos, sz);
194*54ba9607SSascha Wildner 	if (tok == TOKEN_NONE) {
195*54ba9607SSascha Wildner 		mandoc_msg(MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
196*54ba9607SSascha Wildner 		return 1;
19780387638SSascha Wildner 	}
19880387638SSascha Wildner 
199*54ba9607SSascha Wildner 	/* Skip a leading escape sequence or tab. */
20080387638SSascha Wildner 
201*54ba9607SSascha Wildner 	switch (buf[offs]) {
202*54ba9607SSascha Wildner 	case '\\':
203*54ba9607SSascha Wildner 		cp = buf + offs + 1;
204*54ba9607SSascha Wildner 		mandoc_escape(&cp, NULL, NULL);
205*54ba9607SSascha Wildner 		offs = cp - buf;
206*54ba9607SSascha Wildner 		break;
207*54ba9607SSascha Wildner 	case '\t':
208*54ba9607SSascha Wildner 		offs++;
209*54ba9607SSascha Wildner 		break;
210*54ba9607SSascha Wildner 	default:
211*54ba9607SSascha Wildner 		break;
212*54ba9607SSascha Wildner 	}
213*54ba9607SSascha Wildner 
214*54ba9607SSascha Wildner 	/* Jump to the next non-whitespace word. */
215*54ba9607SSascha Wildner 
216*54ba9607SSascha Wildner 	while (buf[offs] == ' ')
21760e1e752SSascha Wildner 		offs++;
21880387638SSascha Wildner 
21980387638SSascha Wildner 	/*
22080387638SSascha Wildner 	 * Trailing whitespace.  Note that tabs are allowed to be passed
22180387638SSascha Wildner 	 * into the parser as "text", so we only warn about spaces here.
22280387638SSascha Wildner 	 */
22380387638SSascha Wildner 
224*54ba9607SSascha Wildner 	if (buf[offs] == '\0' && buf[offs - 1] == ' ')
225*54ba9607SSascha Wildner 		mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);
22680387638SSascha Wildner 
22780387638SSascha Wildner 	/*
228*54ba9607SSascha Wildner 	 * Some macros break next-line scopes; otherwise, remember
229*54ba9607SSascha Wildner 	 * whether we are in next-line scope for a block head.
23080387638SSascha Wildner 	 */
23180387638SSascha Wildner 
232*54ba9607SSascha Wildner 	man_breakscope(man, tok);
233070c62a6SFranco Fichtner 	bline = man->flags & MAN_BLINE;
23480387638SSascha Wildner 
235*54ba9607SSascha Wildner 	/*
236*54ba9607SSascha Wildner 	 * If the line in next-line scope ends with \c, keep the
237*54ba9607SSascha Wildner 	 * next-line scope open for the subsequent input line.
238*54ba9607SSascha Wildner 	 * That is not at all portable, only groff >= 1.22.4
239*54ba9607SSascha Wildner 	 * does it, but *if* this weird idiom occurs in a manual
240*54ba9607SSascha Wildner 	 * page, that's very likely what the author intended.
241*54ba9607SSascha Wildner 	 */
242*54ba9607SSascha Wildner 
243*54ba9607SSascha Wildner 	if (bline && man_hasc(buf + offs))
244*54ba9607SSascha Wildner 		bline = 0;
245*54ba9607SSascha Wildner 
24680387638SSascha Wildner 	/* Call to handler... */
24780387638SSascha Wildner 
248*54ba9607SSascha Wildner 	(*man_macro(tok)->fp)(man, tok, ln, ppos, &offs, buf);
24980387638SSascha Wildner 
250070c62a6SFranco Fichtner 	/* In quick mode (for mandocdb), abort after the NAME section. */
25180387638SSascha Wildner 
252*54ba9607SSascha Wildner 	if (man->quick && tok == MAN_SH) {
253070c62a6SFranco Fichtner 		n = man->last;
254*54ba9607SSascha Wildner 		if (n->type == ROFFT_BODY &&
255070c62a6SFranco Fichtner 		    strcmp(n->prev->child->string, "NAME"))
256*54ba9607SSascha Wildner 			return 2;
25780387638SSascha Wildner 	}
25880387638SSascha Wildner 
25980387638SSascha Wildner 	/*
260070c62a6SFranco Fichtner 	 * If we are in a next-line scope for a block head,
261070c62a6SFranco Fichtner 	 * close it out now and switch to the body,
262070c62a6SFranco Fichtner 	 * unless the next-line scope is allowed to continue.
26380387638SSascha Wildner 	 */
26480387638SSascha Wildner 
265*54ba9607SSascha Wildner 	if (bline == 0 ||
266*54ba9607SSascha Wildner 	    (man->flags & MAN_BLINE) == 0 ||
267*54ba9607SSascha Wildner 	    man->flags & MAN_ELINE ||
268*54ba9607SSascha Wildner 	    man_macro(tok)->flags & MAN_NSCOPED)
269*54ba9607SSascha Wildner 		return 1;
27080387638SSascha Wildner 
271*54ba9607SSascha Wildner 	man_unscope(man, man->last->parent);
272*54ba9607SSascha Wildner 	roff_body_alloc(man, ln, ppos, man->last->tok);
273*54ba9607SSascha Wildner 	man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
274*54ba9607SSascha Wildner 	return 1;
27536342e81SSascha Wildner }
276070c62a6SFranco Fichtner 
277070c62a6SFranco Fichtner void
man_breakscope(struct roff_man * man,int tok)278*54ba9607SSascha Wildner man_breakscope(struct roff_man *man, int tok)
279070c62a6SFranco Fichtner {
280*54ba9607SSascha Wildner 	struct roff_node *n;
281070c62a6SFranco Fichtner 
282*54ba9607SSascha Wildner 	/*
283*54ba9607SSascha Wildner 	 * An element next line scope is open,
284*54ba9607SSascha Wildner 	 * and the new macro is not allowed inside elements.
285*54ba9607SSascha Wildner 	 * Delete the element that is being broken.
286*54ba9607SSascha Wildner 	 */
287*54ba9607SSascha Wildner 
288*54ba9607SSascha Wildner 	if (man->flags & MAN_ELINE && (tok < MAN_TH ||
289*54ba9607SSascha Wildner 	    (man_macro(tok)->flags & MAN_NSCOPED) == 0)) {
290*54ba9607SSascha Wildner 		n = man->last;
291*54ba9607SSascha Wildner 		if (n->type == ROFFT_TEXT)
292*54ba9607SSascha Wildner 			n = n->parent;
293*54ba9607SSascha Wildner 		if (n->tok < MAN_TH ||
294*54ba9607SSascha Wildner 		    (man_macro(n->tok)->flags & (MAN_NSCOPED | MAN_ESCOPED))
295*54ba9607SSascha Wildner 		     == MAN_NSCOPED)
296*54ba9607SSascha Wildner 			n = n->parent;
297*54ba9607SSascha Wildner 
298*54ba9607SSascha Wildner 		mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
299*54ba9607SSascha Wildner 		    "%s breaks %s", roff_name[tok], roff_name[n->tok]);
300*54ba9607SSascha Wildner 
301*54ba9607SSascha Wildner 		roff_node_delete(man, n);
302*54ba9607SSascha Wildner 		man->flags &= ~MAN_ELINE;
303070c62a6SFranco Fichtner 	}
304070c62a6SFranco Fichtner 
305*54ba9607SSascha Wildner 	/*
306*54ba9607SSascha Wildner 	 * Weird special case:
307*54ba9607SSascha Wildner 	 * Switching fill mode closes section headers.
308*54ba9607SSascha Wildner 	 */
309070c62a6SFranco Fichtner 
310*54ba9607SSascha Wildner 	if (man->flags & MAN_BLINE &&
311*54ba9607SSascha Wildner 	    (tok == ROFF_nf || tok == ROFF_fi) &&
312*54ba9607SSascha Wildner 	    (man->last->tok == MAN_SH || man->last->tok == MAN_SS)) {
313*54ba9607SSascha Wildner 		n = man->last;
314*54ba9607SSascha Wildner 		man_unscope(man, n);
315*54ba9607SSascha Wildner 		roff_body_alloc(man, n->line, n->pos, n->tok);
316*54ba9607SSascha Wildner 		man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
317070c62a6SFranco Fichtner 	}
318070c62a6SFranco Fichtner 
319*54ba9607SSascha Wildner 	/*
320*54ba9607SSascha Wildner 	 * A block header next line scope is open,
321*54ba9607SSascha Wildner 	 * and the new macro is not allowed inside block headers.
322*54ba9607SSascha Wildner 	 * Delete the block that is being broken.
323*54ba9607SSascha Wildner 	 */
324070c62a6SFranco Fichtner 
325*54ba9607SSascha Wildner 	if (man->flags & MAN_BLINE && tok != ROFF_nf && tok != ROFF_fi &&
326*54ba9607SSascha Wildner 	    (tok < MAN_TH || man_macro(tok)->flags & MAN_XSCOPE)) {
327*54ba9607SSascha Wildner 		n = man->last;
328*54ba9607SSascha Wildner 		if (n->type == ROFFT_TEXT)
329*54ba9607SSascha Wildner 			n = n->parent;
330*54ba9607SSascha Wildner 		if (n->tok < MAN_TH ||
331*54ba9607SSascha Wildner 		    (man_macro(n->tok)->flags & MAN_XSCOPE) == 0)
332*54ba9607SSascha Wildner 			n = n->parent;
333070c62a6SFranco Fichtner 
334*54ba9607SSascha Wildner 		assert(n->type == ROFFT_HEAD);
335*54ba9607SSascha Wildner 		n = n->parent;
336*54ba9607SSascha Wildner 		assert(n->type == ROFFT_BLOCK);
337*54ba9607SSascha Wildner 		assert(man_macro(n->tok)->flags & MAN_BSCOPED);
338070c62a6SFranco Fichtner 
339*54ba9607SSascha Wildner 		mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
340*54ba9607SSascha Wildner 		    "%s breaks %s", roff_name[tok], roff_name[n->tok]);
341070c62a6SFranco Fichtner 
342*54ba9607SSascha Wildner 		roff_node_delete(man, n);
343*54ba9607SSascha Wildner 		man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
344070c62a6SFranco Fichtner 	}
345070c62a6SFranco Fichtner }
346