xref: /openbsd-src/usr.bin/mandoc/mdoc_argv.c (revision ce7e376cbf6005724ddbab52a1080263dbd6d918)
1*ce7e376cSschwarze /*	$OpenBSD: mdoc_argv.c,v 1.76 2019/07/11 16:56:52 schwarze Exp $ */
2f73abda9Skristaps /*
319a69263Sschwarze  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4*ce7e376cSschwarze  * Copyright (c) 2012, 2014-2019 Ingo Schwarze <schwarze@openbsd.org>
5f73abda9Skristaps  *
6f73abda9Skristaps  * Permission to use, copy, modify, and distribute this software for any
7a6464425Sschwarze  * purpose with or without fee is hereby granted, provided that the above
8a6464425Sschwarze  * copyright notice and this permission notice appear in all copies.
9f73abda9Skristaps  *
10d1982c71Sschwarze  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11a6464425Sschwarze  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12d1982c71Sschwarze  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13a6464425Sschwarze  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14a6464425Sschwarze  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15a6464425Sschwarze  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16a6464425Sschwarze  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17f73abda9Skristaps  */
18f73abda9Skristaps #include <sys/types.h>
19f73abda9Skristaps 
20f73abda9Skristaps #include <assert.h>
21f73abda9Skristaps #include <stdlib.h>
22f73abda9Skristaps #include <stdio.h>
23f73abda9Skristaps #include <string.h>
24f73abda9Skristaps 
254f4f7972Sschwarze #include "mandoc_aux.h"
26d1982c71Sschwarze #include "mandoc.h"
27d1982c71Sschwarze #include "roff.h"
28d1982c71Sschwarze #include "mdoc.h"
29a66b65d0Sschwarze #include "libmandoc.h"
309398f94cSschwarze #include "roff_int.h"
31d1982c71Sschwarze #include "libmdoc.h"
32f73abda9Skristaps 
3319a69263Sschwarze #define	MULTI_STEP	 5 /* pre-allocate argument values */
34a5e11edeSschwarze #define	DELIMSZ		 6 /* max possible size of a delimiter */
35a5e11edeSschwarze 
36a5e11edeSschwarze enum	argsflag {
37a5e11edeSschwarze 	ARGSFL_NONE = 0,
38a5e11edeSschwarze 	ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
39a5e11edeSschwarze 	ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
40a5e11edeSschwarze };
41a5e11edeSschwarze 
42a5e11edeSschwarze enum	argvflag {
43a5e11edeSschwarze 	ARGV_NONE, /* no args to flag (e.g., -split) */
44a5e11edeSschwarze 	ARGV_SINGLE, /* one arg to flag (e.g., -file xxx)  */
4522972b14Sschwarze 	ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
46a5e11edeSschwarze };
47f73abda9Skristaps 
48296801eeSschwarze struct	mdocarg {
49296801eeSschwarze 	enum argsflag	 flags;
50296801eeSschwarze 	const enum mdocargt *argvs;
51296801eeSschwarze };
52296801eeSschwarze 
53296801eeSschwarze static	void		 argn_free(struct mdoc_arg *, int);
54ede1b9d0Sschwarze static	enum margserr	 args(struct roff_man *, int, int *,
55a5e11edeSschwarze 				char *, enum argsflag, char **);
56a5e11edeSschwarze static	int		 args_checkpunct(const char *, int);
57ede1b9d0Sschwarze static	void		 argv_multi(struct roff_man *, int,
58f73abda9Skristaps 				struct mdoc_argv *, int *, char *);
59ede1b9d0Sschwarze static	void		 argv_single(struct roff_man *, int,
60f73abda9Skristaps 				struct mdoc_argv *, int *, char *);
61f73abda9Skristaps 
6219a69263Sschwarze static	const enum argvflag argvflags[MDOC_ARG_MAX] = {
63f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Split */
64f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Nosplit */
65f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Ragged */
66f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Unfilled */
67f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Literal */
684175bdabSschwarze 	ARGV_SINGLE,	/* MDOC_File */
6922972b14Sschwarze 	ARGV_SINGLE,	/* MDOC_Offset */
70f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Bullet */
71f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Dash */
72f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Hyphen */
73f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Item */
74f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Enum */
75f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Tag */
76f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Diag */
77f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Hang */
78f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Ohang */
79f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Inset */
80f73abda9Skristaps 	ARGV_MULTI,	/* MDOC_Column */
81f73abda9Skristaps 	ARGV_SINGLE,	/* MDOC_Width */
82f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Compact */
8354263317Sschwarze 	ARGV_NONE,	/* MDOC_Std */
84f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Filled */
85f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Words */
86f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Emphasis */
87f73abda9Skristaps 	ARGV_NONE,	/* MDOC_Symbolic */
88f73abda9Skristaps 	ARGV_NONE	/* MDOC_Symbolic */
89f73abda9Skristaps };
90f73abda9Skristaps 
91a35fc07aSschwarze static	const enum mdocargt args_Ex[] = {
92a35fc07aSschwarze 	MDOC_Std,
93a35fc07aSschwarze 	MDOC_ARG_MAX
94a35fc07aSschwarze };
95a35fc07aSschwarze 
96a35fc07aSschwarze static	const enum mdocargt args_An[] = {
97a35fc07aSschwarze 	MDOC_Split,
98a35fc07aSschwarze 	MDOC_Nosplit,
99a35fc07aSschwarze 	MDOC_ARG_MAX
100a35fc07aSschwarze };
101a35fc07aSschwarze 
102a35fc07aSschwarze static	const enum mdocargt args_Bd[] = {
103a35fc07aSschwarze 	MDOC_Ragged,
104a35fc07aSschwarze 	MDOC_Unfilled,
105a35fc07aSschwarze 	MDOC_Filled,
106a35fc07aSschwarze 	MDOC_Literal,
107a35fc07aSschwarze 	MDOC_File,
108a35fc07aSschwarze 	MDOC_Offset,
109a35fc07aSschwarze 	MDOC_Compact,
110a35fc07aSschwarze 	MDOC_Centred,
111a35fc07aSschwarze 	MDOC_ARG_MAX
112a35fc07aSschwarze };
113a35fc07aSschwarze 
114a35fc07aSschwarze static	const enum mdocargt args_Bf[] = {
115a35fc07aSschwarze 	MDOC_Emphasis,
116a35fc07aSschwarze 	MDOC_Literal,
117a35fc07aSschwarze 	MDOC_Symbolic,
118a35fc07aSschwarze 	MDOC_ARG_MAX
119a35fc07aSschwarze };
120a35fc07aSschwarze 
121a35fc07aSschwarze static	const enum mdocargt args_Bk[] = {
122a35fc07aSschwarze 	MDOC_Words,
123a35fc07aSschwarze 	MDOC_ARG_MAX
124a35fc07aSschwarze };
125a35fc07aSschwarze 
126a35fc07aSschwarze static	const enum mdocargt args_Bl[] = {
127a35fc07aSschwarze 	MDOC_Bullet,
128a35fc07aSschwarze 	MDOC_Dash,
129a35fc07aSschwarze 	MDOC_Hyphen,
130a35fc07aSschwarze 	MDOC_Item,
131a35fc07aSschwarze 	MDOC_Enum,
132a35fc07aSschwarze 	MDOC_Tag,
133a35fc07aSschwarze 	MDOC_Diag,
134a35fc07aSschwarze 	MDOC_Hang,
135a35fc07aSschwarze 	MDOC_Ohang,
136a35fc07aSschwarze 	MDOC_Inset,
137a35fc07aSschwarze 	MDOC_Column,
138a35fc07aSschwarze 	MDOC_Width,
139a35fc07aSschwarze 	MDOC_Offset,
140a35fc07aSschwarze 	MDOC_Compact,
141a35fc07aSschwarze 	MDOC_Nested,
142a35fc07aSschwarze 	MDOC_ARG_MAX
143a35fc07aSschwarze };
144a35fc07aSschwarze 
14516fe0cfcSschwarze static	const struct mdocarg mdocargs[MDOC_MAX - MDOC_Dd] = {
146296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Dd */
147296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Dt */
148296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Os */
149296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Sh */
150296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ss */
151296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Pp */
152296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* D1 */
153296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Dl */
154296801eeSschwarze 	{ ARGSFL_NONE, args_Bd }, /* Bd */
155296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ed */
156296801eeSschwarze 	{ ARGSFL_NONE, args_Bl }, /* Bl */
157296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* El */
158296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* It */
159296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ad */
160296801eeSschwarze 	{ ARGSFL_DELIM, args_An }, /* An */
16114a309e3Sschwarze 	{ ARGSFL_DELIM, NULL }, /* Ap */
162296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ar */
1637dad7ffdSschwarze 	{ ARGSFL_DELIM, NULL }, /* Cd */
164296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Cm */
165296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Dv */
166296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Er */
167296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ev */
168296801eeSschwarze 	{ ARGSFL_NONE, args_Ex }, /* Ex */
169296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Fa */
170296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Fd */
171296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Fl */
172296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Fn */
173296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ft */
174296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ic */
175cf9fe9abSschwarze 	{ ARGSFL_DELIM, NULL }, /* In */
176296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Li */
177296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Nd */
178296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Nm */
179296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Op */
180551cd4a8Sschwarze 	{ ARGSFL_DELIM, NULL }, /* Ot */
181296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Pa */
182296801eeSschwarze 	{ ARGSFL_NONE, args_Ex }, /* Rv */
183296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* St */
184296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Va */
185296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Vt */
186296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Xr */
187296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %A */
188296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %B */
189296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %D */
190296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %I */
191296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %J */
192296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %N */
193296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %O */
194296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %P */
195296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %R */
196296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %T */
197296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %V */
198296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ac */
199296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ao */
200296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Aq */
201296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* At */
202296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Bc */
203296801eeSschwarze 	{ ARGSFL_NONE, args_Bf }, /* Bf */
204296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Bo */
205296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Bq */
206296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Bsx */
207296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Bx */
208296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Db */
209296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Dc */
210296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Do */
211296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Dq */
212296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ec */
213296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ef */
214296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Em */
215296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Eo */
216296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Fx */
217296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ms */
218296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* No */
219296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ns */
220296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Nx */
221296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ox */
222296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Pc */
223296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Pf */
224296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Po */
225296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Pq */
226296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Qc */
227296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ql */
228296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Qo */
229296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Qq */
230296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Re */
231296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Rs */
232296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Sc */
233296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* So */
234296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Sq */
235296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Sm */
236296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Sx */
237296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Sy */
238296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Tn */
239296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Ux */
240296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Xc */
241296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Xo */
242296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Fo */
2437dad7ffdSschwarze 	{ ARGSFL_DELIM, NULL }, /* Fc */
244296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Oo */
245296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Oc */
246296801eeSschwarze 	{ ARGSFL_NONE, args_Bk }, /* Bk */
247296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ek */
248296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Bt */
249296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Hf */
250551cd4a8Sschwarze 	{ ARGSFL_DELIM, NULL }, /* Fr */
251296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ud */
2527dad7ffdSschwarze 	{ ARGSFL_DELIM, NULL }, /* Lb */
253296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Lp */
254296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Lk */
255296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Mt */
256296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Brq */
257296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Bro */
258296801eeSschwarze 	{ ARGSFL_DELIM, NULL }, /* Brc */
259296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %C */
260296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Es */
261551cd4a8Sschwarze 	{ ARGSFL_DELIM, NULL }, /* En */
26270610dabSschwarze 	{ ARGSFL_DELIM, NULL }, /* Dx */
263296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %Q */
264296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* %U */
265296801eeSschwarze 	{ ARGSFL_NONE, NULL }, /* Ta */
266296801eeSschwarze };
267296801eeSschwarze 
268296801eeSschwarze 
269f73abda9Skristaps /*
270a53559d2Sschwarze  * Parse flags and their arguments from the input line.
271a53559d2Sschwarze  * These come in the form -flag [argument ...].
272a53559d2Sschwarze  * Some flags take no argument, some one, some multiple.
273f73abda9Skristaps  */
274a53559d2Sschwarze void
mdoc_argv(struct roff_man * mdoc,int line,enum roff_tok tok,struct mdoc_arg ** reta,int * pos,char * buf)27514a309e3Sschwarze mdoc_argv(struct roff_man *mdoc, int line, enum roff_tok tok,
276a53559d2Sschwarze 	struct mdoc_arg **reta, int *pos, char *buf)
277f73abda9Skristaps {
278a53559d2Sschwarze 	struct mdoc_argv	  tmpv;
279a53559d2Sschwarze 	struct mdoc_argv	**retv;
280a53559d2Sschwarze 	const enum mdocargt	 *argtable;
281a53559d2Sschwarze 	char			 *argname;
282a53559d2Sschwarze 	int			  ipos, retc;
283a53559d2Sschwarze 	char			  savechar;
284f73abda9Skristaps 
285a53559d2Sschwarze 	*reta = NULL;
286a53559d2Sschwarze 
287a53559d2Sschwarze 	/* Which flags does this macro support? */
288a53559d2Sschwarze 
28914a309e3Sschwarze 	assert(tok >= MDOC_Dd && tok < MDOC_MAX);
29016fe0cfcSschwarze 	argtable = mdocargs[tok - MDOC_Dd].argvs;
291a53559d2Sschwarze 	if (argtable == NULL)
292a53559d2Sschwarze 		return;
293a53559d2Sschwarze 
294a53559d2Sschwarze 	/* Loop over the flags on the input line. */
295a53559d2Sschwarze 
296a53559d2Sschwarze 	ipos = *pos;
297a53559d2Sschwarze 	while (buf[ipos] == '-') {
298f73abda9Skristaps 
299296801eeSschwarze 		/* Seek to the first unescaped space. */
300f73abda9Skristaps 
301a53559d2Sschwarze 		for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
302a53559d2Sschwarze 			if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
303f73abda9Skristaps 				break;
304f73abda9Skristaps 
305296801eeSschwarze 		/*
306a53559d2Sschwarze 		 * We want to nil-terminate the word to look it up.
307a53559d2Sschwarze 		 * But we may not have a flag, in which case we need
308a53559d2Sschwarze 		 * to restore the line as-is.  So keep around the
309a53559d2Sschwarze 		 * stray byte, which we'll reset upon exiting.
310296801eeSschwarze 		 */
311f73abda9Skristaps 
312a53559d2Sschwarze 		if ((savechar = buf[ipos]) != '\0')
313a53559d2Sschwarze 			buf[ipos++] = '\0';
314296801eeSschwarze 
315296801eeSschwarze 		/*
316a53559d2Sschwarze 		 * Now look up the word as a flag.  Use temporary
317a53559d2Sschwarze 		 * storage that we'll copy into the node's flags.
318296801eeSschwarze 		 */
319f73abda9Skristaps 
320a53559d2Sschwarze 		while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
321a53559d2Sschwarze 			if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
322296801eeSschwarze 				break;
323f73abda9Skristaps 
324a53559d2Sschwarze 		/* If it isn't a flag, restore the saved byte. */
325a53559d2Sschwarze 
326a53559d2Sschwarze 		if (tmpv.arg == MDOC_ARG_MAX) {
327a53559d2Sschwarze 			if (savechar != '\0')
328a53559d2Sschwarze 				buf[ipos - 1] = savechar;
329a53559d2Sschwarze 			break;
330f73abda9Skristaps 		}
331f73abda9Skristaps 
332a53559d2Sschwarze 		/* Read to the next word (the first argument). */
333296801eeSschwarze 
334a53559d2Sschwarze 		while (buf[ipos] == ' ')
335a53559d2Sschwarze 			ipos++;
336f73abda9Skristaps 
337a53559d2Sschwarze 		/* Parse the arguments of the flag. */
338a53559d2Sschwarze 
339a53559d2Sschwarze 		tmpv.line  = line;
3406b7a500cSschwarze 		tmpv.pos   = *pos;
341a53559d2Sschwarze 		tmpv.sz    = 0;
342a53559d2Sschwarze 		tmpv.value = NULL;
343a53559d2Sschwarze 
344a53559d2Sschwarze 		switch (argvflags[tmpv.arg]) {
34549aff9f8Sschwarze 		case ARGV_SINGLE:
346a53559d2Sschwarze 			argv_single(mdoc, line, &tmpv, &ipos, buf);
347296801eeSschwarze 			break;
34849aff9f8Sschwarze 		case ARGV_MULTI:
349a53559d2Sschwarze 			argv_multi(mdoc, line, &tmpv, &ipos, buf);
350296801eeSschwarze 			break;
35149aff9f8Sschwarze 		case ARGV_NONE:
352296801eeSschwarze 			break;
353296801eeSschwarze 		}
354f73abda9Skristaps 
355a53559d2Sschwarze 		/* Append to the return values. */
356f73abda9Skristaps 
357a53559d2Sschwarze 		if (*reta == NULL)
358a53559d2Sschwarze 			*reta = mandoc_calloc(1, sizeof(**reta));
359f73abda9Skristaps 
360a53559d2Sschwarze 		retc = ++(*reta)->argc;
361a53559d2Sschwarze 		retv = &(*reta)->argv;
362a53559d2Sschwarze 		*retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
363a53559d2Sschwarze 		memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
364f73abda9Skristaps 
365a53559d2Sschwarze 		/* Prepare for parsing the next flag. */
366a53559d2Sschwarze 
367a53559d2Sschwarze 		*pos = ipos;
36816fe0cfcSschwarze 		argtable = mdocargs[tok - MDOC_Dd].argvs;
369a53559d2Sschwarze 	}
370f73abda9Skristaps }
371f73abda9Skristaps 
372f73abda9Skristaps void
mdoc_argv_free(struct mdoc_arg * p)373f73abda9Skristaps mdoc_argv_free(struct mdoc_arg *p)
374f73abda9Skristaps {
37569d1e7afSschwarze 	int		 i;
376f73abda9Skristaps 
377f73abda9Skristaps 	if (NULL == p)
378f73abda9Skristaps 		return;
379f73abda9Skristaps 
380f73abda9Skristaps 	if (p->refcnt) {
381f73abda9Skristaps 		--(p->refcnt);
382f73abda9Skristaps 		if (p->refcnt)
383f73abda9Skristaps 			return;
384f73abda9Skristaps 	}
385f73abda9Skristaps 	assert(p->argc);
386f73abda9Skristaps 
387a4c002ecSschwarze 	for (i = (int)p->argc - 1; i >= 0; i--)
388a35fc07aSschwarze 		argn_free(p, i);
389f73abda9Skristaps 
390f73abda9Skristaps 	free(p->argv);
391f73abda9Skristaps 	free(p);
392f73abda9Skristaps }
393f73abda9Skristaps 
394a35fc07aSschwarze static void
argn_free(struct mdoc_arg * p,int iarg)395a35fc07aSschwarze argn_free(struct mdoc_arg *p, int iarg)
39669d1e7afSschwarze {
397ddce0b0cSschwarze 	struct mdoc_argv *arg;
39869d1e7afSschwarze 	int		  j;
39969d1e7afSschwarze 
400ddce0b0cSschwarze 	arg = &p->argv[iarg];
401ddce0b0cSschwarze 
40269d1e7afSschwarze 	if (arg->sz && arg->value) {
403a4c002ecSschwarze 		for (j = (int)arg->sz - 1; j >= 0; j--)
40469d1e7afSschwarze 			free(arg->value[j]);
40569d1e7afSschwarze 		free(arg->value);
40669d1e7afSschwarze 	}
40769d1e7afSschwarze 
40869d1e7afSschwarze 	for (--p->argc; iarg < (int)p->argc; iarg++)
40969d1e7afSschwarze 		p->argv[iarg] = p->argv[iarg+1];
41069d1e7afSschwarze }
41169d1e7afSschwarze 
412a4c002ecSschwarze enum margserr
mdoc_args(struct roff_man * mdoc,int line,int * pos,char * buf,enum roff_tok tok,char ** v)413ede1b9d0Sschwarze mdoc_args(struct roff_man *mdoc, int line, int *pos,
41414a309e3Sschwarze 	char *buf, enum roff_tok tok, char **v)
415f73abda9Skristaps {
4163a0d07afSschwarze 	struct roff_node *n;
4179d7f0ab7Sschwarze 	enum argsflag	  fl;
418f73abda9Skristaps 
41916fe0cfcSschwarze 	fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok - MDOC_Dd].flags;
42068ab41b0Sschwarze 
421f73abda9Skristaps 	/*
4226093755cSschwarze 	 * We know that we're in an `It', so it's reasonable to expect
4236093755cSschwarze 	 * us to be sitting in a `Bl'.  Someday this may not be the case
4246093755cSschwarze 	 * (if we allow random `It's sitting out there), so provide a
4256093755cSschwarze 	 * safe fall-back into the default behaviour.
426f73abda9Skristaps 	 */
427f73abda9Skristaps 
428c720427dSschwarze 	if (tok == MDOC_It) {
429c720427dSschwarze 		for (n = mdoc->last; n != NULL; n = n->parent) {
430c720427dSschwarze 			if (n->tok != MDOC_Bl)
431c720427dSschwarze 				continue;
432c720427dSschwarze 			if (n->norm->Bl.type == LIST_column)
433a5e11edeSschwarze 				fl = ARGSFL_TABSEP;
434f73abda9Skristaps 			break;
435f73abda9Skristaps 		}
436c720427dSschwarze 	}
437f73abda9Skristaps 
438526e306bSschwarze 	return args(mdoc, line, pos, buf, fl, v);
439f73abda9Skristaps }
440f73abda9Skristaps 
441a4c002ecSschwarze static enum margserr
args(struct roff_man * mdoc,int line,int * pos,char * buf,enum argsflag fl,char ** v)442ede1b9d0Sschwarze args(struct roff_man *mdoc, int line, int *pos,
443a5e11edeSschwarze 		char *buf, enum argsflag fl, char **v)
444f73abda9Skristaps {
44590fd17c2Sschwarze 	char		*p;
446e94357f9Sschwarze 	char		*v_local;
4474e221707Sschwarze 	int		 pairs;
448f73abda9Skristaps 
44990fd17c2Sschwarze 	if (buf[*pos] == '\0') {
45090fd17c2Sschwarze 		if (mdoc->flags & MDOC_PHRASELIT &&
451407d12bbSschwarze 		    ! (mdoc->flags & MDOC_PHRASE)) {
452a5a5f808Sschwarze 			mandoc_msg(MANDOCERR_ARG_QUOTE, line, *pos, NULL);
4537ead8a4eSschwarze 			mdoc->flags &= ~MDOC_PHRASELIT;
45490fd17c2Sschwarze 		}
455*ce7e376cSschwarze 		mdoc->flags &= ~MDOC_PHRASEQL;
456526e306bSschwarze 		return ARGS_EOLN;
45766a16cf6Sschwarze 	}
458f73abda9Skristaps 
459e94357f9Sschwarze 	if (v == NULL)
460e94357f9Sschwarze 		v = &v_local;
46190fd17c2Sschwarze 	*v = buf + *pos;
46219a69263Sschwarze 
46390fd17c2Sschwarze 	if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos))
464526e306bSschwarze 		return ARGS_PUNCT;
465f73abda9Skristaps 
466f73abda9Skristaps 	/*
46790fd17c2Sschwarze 	 * Tabs in `It' lines in `Bl -column' can't be escaped.
46890fd17c2Sschwarze 	 * Phrases are reparsed for `Ta' and other macros later.
469f73abda9Skristaps 	 */
470f73abda9Skristaps 
47190fd17c2Sschwarze 	if (fl == ARGSFL_TABSEP) {
47290fd17c2Sschwarze 		if ((p = strchr(*v, '\t')) != NULL) {
473407d12bbSschwarze 
474407d12bbSschwarze 			/*
475407d12bbSschwarze 			 * Words right before and right after
476407d12bbSschwarze 			 * tab characters are not parsed,
477407d12bbSschwarze 			 * unless there is a blank in between.
478407d12bbSschwarze 			 */
479407d12bbSschwarze 
480eca553adSschwarze 			if (p > buf && p[-1] != ' ')
481407d12bbSschwarze 				mdoc->flags |= MDOC_PHRASEQL;
482407d12bbSschwarze 			if (p[1] != ' ')
483407d12bbSschwarze 				mdoc->flags |= MDOC_PHRASEQN;
484407d12bbSschwarze 
485407d12bbSschwarze 			/*
486407d12bbSschwarze 			 * One or more blanks after a tab cause
487407d12bbSschwarze 			 * one leading blank in the next column.
488407d12bbSschwarze 			 * So skip all but one of them.
489407d12bbSschwarze 			 */
490407d12bbSschwarze 
49190fd17c2Sschwarze 			*pos += (int)(p - *v) + 1;
492407d12bbSschwarze 			while (buf[*pos] == ' ' && buf[*pos + 1] == ' ')
49390fd17c2Sschwarze 				(*pos)++;
494407d12bbSschwarze 
495407d12bbSschwarze 			/*
496407d12bbSschwarze 			 * A tab at the end of an input line
497407d12bbSschwarze 			 * switches to the next column.
498407d12bbSschwarze 			 */
499407d12bbSschwarze 
500407d12bbSschwarze 			if (buf[*pos] == '\0' || buf[*pos + 1] == '\0')
501407d12bbSschwarze 				mdoc->flags |= MDOC_PHRASEQN;
502fccfce9dSschwarze 		} else {
50390fd17c2Sschwarze 			p = strchr(*v, '\0');
50490fd17c2Sschwarze 			if (p[-1] == ' ')
50590fd17c2Sschwarze 				mandoc_msg(MANDOCERR_SPACE_EOL,
506a5a5f808Sschwarze 				    line, *pos, NULL);
50768ab41b0Sschwarze 			*pos += (int)(p - *v);
50868ab41b0Sschwarze 		}
509f73abda9Skristaps 
51090fd17c2Sschwarze 		/* Skip any trailing blank characters. */
51190fd17c2Sschwarze 		while (p > *v && p[-1] == ' ' &&
51290fd17c2Sschwarze 		    (p - 1 == *v || p[-2] != '\\'))
51390fd17c2Sschwarze 			p--;
51490fd17c2Sschwarze 		*p = '\0';
515f73abda9Skristaps 
516407d12bbSschwarze 		return ARGS_PHRASE;
517f73abda9Skristaps 	}
518f73abda9Skristaps 
51968ab41b0Sschwarze 	/*
52068ab41b0Sschwarze 	 * Process a quoted literal.  A quote begins with a double-quote
52168ab41b0Sschwarze 	 * and ends with a double-quote NOT preceded by a double-quote.
522d581c541Sschwarze 	 * NUL-terminate the literal in place.
5234e221707Sschwarze 	 * Collapse pairs of quotes inside quoted literals.
52468ab41b0Sschwarze 	 * Whitespace is NOT involved in literal termination.
52568ab41b0Sschwarze 	 */
526f73abda9Skristaps 
527e94357f9Sschwarze 	if (mdoc->flags & MDOC_PHRASELIT ||
528e94357f9Sschwarze 	    (mdoc->flags & MDOC_PHRASE && buf[*pos] == '\"')) {
529e94357f9Sschwarze 		if ((mdoc->flags & MDOC_PHRASELIT) == 0) {
53068ab41b0Sschwarze 			*v = &buf[++(*pos)];
5317ead8a4eSschwarze 			mdoc->flags |= MDOC_PHRASELIT;
532e94357f9Sschwarze 		}
5334e221707Sschwarze 		pairs = 0;
53468ab41b0Sschwarze 		for ( ; buf[*pos]; (*pos)++) {
5354e221707Sschwarze 			/* Move following text left after quoted quotes. */
5364e221707Sschwarze 			if (pairs)
5374e221707Sschwarze 				buf[*pos - pairs] = buf[*pos];
53868ab41b0Sschwarze 			if ('\"' != buf[*pos])
53968ab41b0Sschwarze 				continue;
5404e221707Sschwarze 			/* Unquoted quotes end quoted args. */
54168ab41b0Sschwarze 			if ('\"' != buf[*pos + 1])
542f73abda9Skristaps 				break;
5434e221707Sschwarze 			/* Quoted quotes collapse. */
5444e221707Sschwarze 			pairs++;
545f73abda9Skristaps 			(*pos)++;
546f73abda9Skristaps 		}
5474e221707Sschwarze 		if (pairs)
5484e221707Sschwarze 			buf[*pos - pairs] = '\0';
549f73abda9Skristaps 
550407d12bbSschwarze 		if (buf[*pos] == '\0') {
551407d12bbSschwarze 			if ( ! (mdoc->flags & MDOC_PHRASE))
552dd5b31c3Sschwarze 				mandoc_msg(MANDOCERR_ARG_QUOTE,
553a5a5f808Sschwarze 				    line, *pos, NULL);
554232c32a3Sschwarze 			return ARGS_WORD;
55568ab41b0Sschwarze 		}
556f73abda9Skristaps 
5577ead8a4eSschwarze 		mdoc->flags &= ~MDOC_PHRASELIT;
55866a16cf6Sschwarze 		buf[(*pos)++] = '\0';
559f73abda9Skristaps 
56066a16cf6Sschwarze 		if ('\0' == buf[*pos])
561232c32a3Sschwarze 			return ARGS_WORD;
562f73abda9Skristaps 
56368ab41b0Sschwarze 		while (' ' == buf[*pos])
564f73abda9Skristaps 			(*pos)++;
565f73abda9Skristaps 
566a5e11edeSschwarze 		if ('\0' == buf[*pos])
567a5a5f808Sschwarze 			mandoc_msg(MANDOCERR_SPACE_EOL, line, *pos, NULL);
568f73abda9Skristaps 
569232c32a3Sschwarze 		return ARGS_WORD;
570f73abda9Skristaps 	}
571f73abda9Skristaps 
572a5e11edeSschwarze 	p = &buf[*pos];
573e94357f9Sschwarze 	*v = roff_getarg(mdoc->roff, &p, line, pos);
574e94357f9Sschwarze 	if (v == &v_local)
575e94357f9Sschwarze 		free(*v);
576f73abda9Skristaps 
577407d12bbSschwarze 	/*
578407d12bbSschwarze 	 * After parsing the last word in this phrase,
579407d12bbSschwarze 	 * tell lookup() whether or not to interpret it.
580407d12bbSschwarze 	 */
581407d12bbSschwarze 
582407d12bbSschwarze 	if (*p == '\0' && mdoc->flags & MDOC_PHRASEQL) {
583407d12bbSschwarze 		mdoc->flags &= ~MDOC_PHRASEQL;
584407d12bbSschwarze 		mdoc->flags |= MDOC_PHRASEQF;
585407d12bbSschwarze 	}
586e94357f9Sschwarze 	return ARGS_ALLOC;
587f73abda9Skristaps }
588f73abda9Skristaps 
58919a69263Sschwarze /*
59019a69263Sschwarze  * Check if the string consists only of space-separated closing
59119a69263Sschwarze  * delimiters.  This is a bit of a dance: the first must be a close
59219a69263Sschwarze  * delimiter, but it may be followed by middle delimiters.  Arbitrary
59319a69263Sschwarze  * whitespace may separate these tokens.
59419a69263Sschwarze  */
59519a69263Sschwarze static int
args_checkpunct(const char * buf,int i)596a5e11edeSschwarze args_checkpunct(const char *buf, int i)
59719a69263Sschwarze {
598a5e11edeSschwarze 	int		 j;
599a5e11edeSschwarze 	char		 dbuf[DELIMSZ];
60019a69263Sschwarze 	enum mdelim	 d;
601f73abda9Skristaps 
60219a69263Sschwarze 	/* First token must be a close-delimiter. */
60319a69263Sschwarze 
604a5e11edeSschwarze 	for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
605a5e11edeSschwarze 		dbuf[j] = buf[i];
60619a69263Sschwarze 
60719a69263Sschwarze 	if (DELIMSZ == j)
608526e306bSschwarze 		return 0;
60919a69263Sschwarze 
610a5e11edeSschwarze 	dbuf[j] = '\0';
611a5e11edeSschwarze 	if (DELIM_CLOSE != mdoc_isdelim(dbuf))
612526e306bSschwarze 		return 0;
61319a69263Sschwarze 
614a5e11edeSschwarze 	while (' ' == buf[i])
61519a69263Sschwarze 		i++;
61619a69263Sschwarze 
61719a69263Sschwarze 	/* Remaining must NOT be open/none. */
61819a69263Sschwarze 
619a5e11edeSschwarze 	while (buf[i]) {
62019a69263Sschwarze 		j = 0;
621a5e11edeSschwarze 		while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
622a5e11edeSschwarze 			dbuf[j++] = buf[i++];
62319a69263Sschwarze 
62419a69263Sschwarze 		if (DELIMSZ == j)
625526e306bSschwarze 			return 0;
62619a69263Sschwarze 
627a5e11edeSschwarze 		dbuf[j] = '\0';
628a5e11edeSschwarze 		d = mdoc_isdelim(dbuf);
62919a69263Sschwarze 		if (DELIM_NONE == d || DELIM_OPEN == d)
630526e306bSschwarze 			return 0;
63119a69263Sschwarze 
632a5e11edeSschwarze 		while (' ' == buf[i])
63319a69263Sschwarze 			i++;
63419a69263Sschwarze 	}
63519a69263Sschwarze 
636526e306bSschwarze 	return '\0' == buf[i];
63719a69263Sschwarze }
63819a69263Sschwarze 
639f4086bd4Sschwarze static void
argv_multi(struct roff_man * mdoc,int line,struct mdoc_argv * v,int * pos,char * buf)640ede1b9d0Sschwarze argv_multi(struct roff_man *mdoc, int line,
641f73abda9Skristaps 		struct mdoc_argv *v, int *pos, char *buf)
642f73abda9Skristaps {
643a4c002ecSschwarze 	enum margserr	 ac;
644f73abda9Skristaps 	char		*p;
645f73abda9Skristaps 
646f73abda9Skristaps 	for (v->sz = 0; ; v->sz++) {
647f4086bd4Sschwarze 		if (buf[*pos] == '-')
648f73abda9Skristaps 			break;
6497ead8a4eSschwarze 		ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
650f4086bd4Sschwarze 		if (ac == ARGS_EOLN)
651f73abda9Skristaps 			break;
652f73abda9Skristaps 
653f4086bd4Sschwarze 		if (v->sz % MULTI_STEP == 0)
6548286bf36Sschwarze 			v->value = mandoc_reallocarray(v->value,
6558286bf36Sschwarze 			    v->sz + MULTI_STEP, sizeof(char *));
656a66b65d0Sschwarze 
657e94357f9Sschwarze 		if (ac != ARGS_ALLOC)
658e94357f9Sschwarze 			p = mandoc_strdup(p);
659e94357f9Sschwarze 		v->value[(int)v->sz] = p;
660f73abda9Skristaps 	}
661f73abda9Skristaps }
662f73abda9Skristaps 
663f4086bd4Sschwarze static void
argv_single(struct roff_man * mdoc,int line,struct mdoc_argv * v,int * pos,char * buf)664ede1b9d0Sschwarze argv_single(struct roff_man *mdoc, int line,
665f73abda9Skristaps 		struct mdoc_argv *v, int *pos, char *buf)
666f73abda9Skristaps {
667a4c002ecSschwarze 	enum margserr	 ac;
668f73abda9Skristaps 	char		*p;
669f73abda9Skristaps 
6707ead8a4eSschwarze 	ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
671f4086bd4Sschwarze 	if (ac == ARGS_EOLN)
672f4086bd4Sschwarze 		return;
673f73abda9Skristaps 
674e94357f9Sschwarze 	if (ac != ARGS_ALLOC)
675e94357f9Sschwarze 		p = mandoc_strdup(p);
676e94357f9Sschwarze 
677f73abda9Skristaps 	v->sz = 1;
678a66b65d0Sschwarze 	v->value = mandoc_malloc(sizeof(char *));
679e94357f9Sschwarze 	v->value[0] = p;
680f73abda9Skristaps }
681