xref: /illumos-gate/usr/src/cmd/sed/compile.c (revision 84441f85)
1*84441f85SGarrett D'Amore /*
2*84441f85SGarrett D'Amore  * Copyright 2010 Nexenta Systems, Inc.  All rights reserved.
3*84441f85SGarrett D'Amore  * Copyright (c) 1992 Diomidis Spinellis.
4*84441f85SGarrett D'Amore  * Copyright (c) 1992, 1993
5*84441f85SGarrett D'Amore  *	The Regents of the University of California.  All rights reserved.
6*84441f85SGarrett D'Amore  *
7*84441f85SGarrett D'Amore  * This code is derived from software contributed to Berkeley by
8*84441f85SGarrett D'Amore  * Diomidis Spinellis of Imperial College, University of London.
9*84441f85SGarrett D'Amore  *
10*84441f85SGarrett D'Amore  * Redistribution and use in source and binary forms, with or without
11*84441f85SGarrett D'Amore  * modification, are permitted provided that the following conditions
12*84441f85SGarrett D'Amore  * are met:
13*84441f85SGarrett D'Amore  * 1. Redistributions of source code must retain the above copyright
14*84441f85SGarrett D'Amore  *    notice, this list of conditions and the following disclaimer.
15*84441f85SGarrett D'Amore  * 2. Redistributions in binary form must reproduce the above copyright
16*84441f85SGarrett D'Amore  *    notice, this list of conditions and the following disclaimer in the
17*84441f85SGarrett D'Amore  *    documentation and/or other materials provided with the distribution.
18*84441f85SGarrett D'Amore  * 4. Neither the name of the University nor the names of its contributors
19*84441f85SGarrett D'Amore  *    may be used to endorse or promote products derived from this software
20*84441f85SGarrett D'Amore  *    without specific prior written permission.
21*84441f85SGarrett D'Amore  *
22*84441f85SGarrett D'Amore  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23*84441f85SGarrett D'Amore  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24*84441f85SGarrett D'Amore  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25*84441f85SGarrett D'Amore  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26*84441f85SGarrett D'Amore  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27*84441f85SGarrett D'Amore  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28*84441f85SGarrett D'Amore  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29*84441f85SGarrett D'Amore  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30*84441f85SGarrett D'Amore  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31*84441f85SGarrett D'Amore  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32*84441f85SGarrett D'Amore  * SUCH DAMAGE.
33*84441f85SGarrett D'Amore  */
34*84441f85SGarrett D'Amore 
35*84441f85SGarrett D'Amore #include <sys/types.h>
36*84441f85SGarrett D'Amore #include <sys/stat.h>
37*84441f85SGarrett D'Amore 
38*84441f85SGarrett D'Amore #include <ctype.h>
39*84441f85SGarrett D'Amore #include <err.h>
40*84441f85SGarrett D'Amore #include <errno.h>
41*84441f85SGarrett D'Amore #include <fcntl.h>
42*84441f85SGarrett D'Amore #include <limits.h>
43*84441f85SGarrett D'Amore #include <regex.h>
44*84441f85SGarrett D'Amore #include <stdio.h>
45*84441f85SGarrett D'Amore #include <stdlib.h>
46*84441f85SGarrett D'Amore #include <string.h>
47*84441f85SGarrett D'Amore #include <wchar.h>
48*84441f85SGarrett D'Amore #include <libintl.h>
49*84441f85SGarrett D'Amore #include <note.h>
50*84441f85SGarrett D'Amore 
51*84441f85SGarrett D'Amore #include "defs.h"
52*84441f85SGarrett D'Amore #include "extern.h"
53*84441f85SGarrett D'Amore 
54*84441f85SGarrett D'Amore #define	LHSZ	128
55*84441f85SGarrett D'Amore #define	LHMASK	(LHSZ - 1)
56*84441f85SGarrett D'Amore static struct labhash {
57*84441f85SGarrett D'Amore 	struct	labhash *lh_next;
58*84441f85SGarrett D'Amore 	uint_t	lh_hash;
59*84441f85SGarrett D'Amore 	struct	s_command *lh_cmd;
60*84441f85SGarrett D'Amore 	int	lh_ref;
61*84441f85SGarrett D'Amore } *labels[LHSZ];
62*84441f85SGarrett D'Amore 
63*84441f85SGarrett D'Amore static char	 *compile_addr(char *, struct s_addr *);
64*84441f85SGarrett D'Amore static char	 *compile_ccl(char **, char *);
65*84441f85SGarrett D'Amore static char	 *compile_delimited(char *, char *, int);
66*84441f85SGarrett D'Amore static char	 *compile_flags(char *, struct s_subst *);
67*84441f85SGarrett D'Amore static regex_t	 *compile_re(char *, int);
68*84441f85SGarrett D'Amore static char	 *compile_subst(char *, struct s_subst *);
69*84441f85SGarrett D'Amore static char	 *compile_text(void);
70*84441f85SGarrett D'Amore static char	 *compile_tr(char *, struct s_tr **);
71*84441f85SGarrett D'Amore static struct s_command
72*84441f85SGarrett D'Amore 		**compile_stream(struct s_command **);
73*84441f85SGarrett D'Amore static char	 *duptoeol(char *, const char *);
74*84441f85SGarrett D'Amore static void	  enterlabel(struct s_command *);
75*84441f85SGarrett D'Amore static struct s_command
76*84441f85SGarrett D'Amore 		 *findlabel(char *);
77*84441f85SGarrett D'Amore static void	  fixuplabel(struct s_command *, struct s_command *);
78*84441f85SGarrett D'Amore static void	  uselabel(void);
79*84441f85SGarrett D'Amore 
80*84441f85SGarrett D'Amore /*
81*84441f85SGarrett D'Amore  * Command specification.  This is used to drive the command parser.
82*84441f85SGarrett D'Amore  */
83*84441f85SGarrett D'Amore struct s_format {
84*84441f85SGarrett D'Amore 	char code;				/* Command code */
85*84441f85SGarrett D'Amore 	int naddr;				/* Number of address args */
86*84441f85SGarrett D'Amore 	enum e_args args;			/* Argument type */
87*84441f85SGarrett D'Amore };
88*84441f85SGarrett D'Amore 
89*84441f85SGarrett D'Amore static struct s_format cmd_fmts[] = {
90*84441f85SGarrett D'Amore 	{'{', 2, GROUP},
91*84441f85SGarrett D'Amore 	{'}', 0, ENDGROUP},
92*84441f85SGarrett D'Amore 	{'a', 1, TEXT},
93*84441f85SGarrett D'Amore 	{'b', 2, BRANCH},
94*84441f85SGarrett D'Amore 	{'c', 2, TEXT},
95*84441f85SGarrett D'Amore 	{'d', 2, EMPTY},
96*84441f85SGarrett D'Amore 	{'D', 2, EMPTY},
97*84441f85SGarrett D'Amore 	{'g', 2, EMPTY},
98*84441f85SGarrett D'Amore 	{'G', 2, EMPTY},
99*84441f85SGarrett D'Amore 	{'h', 2, EMPTY},
100*84441f85SGarrett D'Amore 	{'H', 2, EMPTY},
101*84441f85SGarrett D'Amore 	{'i', 1, TEXT},
102*84441f85SGarrett D'Amore 	{'l', 2, EMPTY},
103*84441f85SGarrett D'Amore 	{'n', 2, EMPTY},
104*84441f85SGarrett D'Amore 	{'N', 2, EMPTY},
105*84441f85SGarrett D'Amore 	{'p', 2, EMPTY},
106*84441f85SGarrett D'Amore 	{'P', 2, EMPTY},
107*84441f85SGarrett D'Amore 	{'q', 1, EMPTY},
108*84441f85SGarrett D'Amore 	{'r', 1, RFILE},
109*84441f85SGarrett D'Amore 	{'s', 2, SUBST},
110*84441f85SGarrett D'Amore 	{'t', 2, BRANCH},
111*84441f85SGarrett D'Amore 	{'w', 2, WFILE},
112*84441f85SGarrett D'Amore 	{'x', 2, EMPTY},
113*84441f85SGarrett D'Amore 	{'y', 2, TR},
114*84441f85SGarrett D'Amore 	{'!', 2, NONSEL},
115*84441f85SGarrett D'Amore 	{':', 0, LABEL},
116*84441f85SGarrett D'Amore 	{'#', 0, COMMENT},
117*84441f85SGarrett D'Amore 	{'=', 1, EMPTY},
118*84441f85SGarrett D'Amore 	{'\0', 0, COMMENT},
119*84441f85SGarrett D'Amore };
120*84441f85SGarrett D'Amore 
121*84441f85SGarrett D'Amore /* The compiled program. */
122*84441f85SGarrett D'Amore struct s_command *prog;
123*84441f85SGarrett D'Amore 
124*84441f85SGarrett D'Amore /*
125*84441f85SGarrett D'Amore  * Compile the program into prog.
126*84441f85SGarrett D'Amore  * Initialise appends.
127*84441f85SGarrett D'Amore  */
128*84441f85SGarrett D'Amore void
compile(void)129*84441f85SGarrett D'Amore compile(void)
130*84441f85SGarrett D'Amore {
131*84441f85SGarrett D'Amore 	*compile_stream(&prog) = NULL;
132*84441f85SGarrett D'Amore 	fixuplabel(prog, NULL);
133*84441f85SGarrett D'Amore 	uselabel();
134*84441f85SGarrett D'Amore 	if (appendnum == 0)
135*84441f85SGarrett D'Amore 		appends = NULL;
136*84441f85SGarrett D'Amore 	else if ((appends = malloc(sizeof (struct s_appends) * appendnum)) ==
137*84441f85SGarrett D'Amore 	    NULL)
138*84441f85SGarrett D'Amore 		err(1, "malloc");
139*84441f85SGarrett D'Amore 	if ((match = malloc((maxnsub + 1) * sizeof (regmatch_t))) == NULL)
140*84441f85SGarrett D'Amore 		err(1, "malloc");
141*84441f85SGarrett D'Amore }
142*84441f85SGarrett D'Amore 
143*84441f85SGarrett D'Amore #define	EATSPACE() do {							\
144*84441f85SGarrett D'Amore 	if (p)								\
145*84441f85SGarrett D'Amore 		while (*p && isspace((unsigned char)*p))                \
146*84441f85SGarrett D'Amore 			p++;						\
147*84441f85SGarrett D'Amore 	_NOTE(CONSTCOND)						\
148*84441f85SGarrett D'Amore } while (0)
149*84441f85SGarrett D'Amore 
150*84441f85SGarrett D'Amore static struct s_command **
compile_stream(struct s_command ** link)151*84441f85SGarrett D'Amore compile_stream(struct s_command **link)
152*84441f85SGarrett D'Amore {
153*84441f85SGarrett D'Amore 	char *p;
154*84441f85SGarrett D'Amore 	static char lbuf[_POSIX2_LINE_MAX + 1];	/* To save stack */
155*84441f85SGarrett D'Amore 	struct s_command *cmd, *cmd2, *stack;
156*84441f85SGarrett D'Amore 	struct s_format *fp;
157*84441f85SGarrett D'Amore 	char re[_POSIX2_LINE_MAX + 1];
158*84441f85SGarrett D'Amore 	int naddr;				/* Number of addresses */
159*84441f85SGarrett D'Amore 
160*84441f85SGarrett D'Amore 	stack = 0;
161*84441f85SGarrett D'Amore 	for (;;) {
162*84441f85SGarrett D'Amore 		if ((p = cu_fgets(lbuf, sizeof (lbuf), NULL)) == NULL) {
163*84441f85SGarrett D'Amore 			if (stack != 0)
164*84441f85SGarrett D'Amore 				fatal(_("unexpected EOF (pending }'s)"));
165*84441f85SGarrett D'Amore 			return (link);
166*84441f85SGarrett D'Amore 		}
167*84441f85SGarrett D'Amore 
168*84441f85SGarrett D'Amore semicolon:	EATSPACE();
169*84441f85SGarrett D'Amore 		if (p) {
170*84441f85SGarrett D'Amore 			if (*p == '#' || *p == '\0')
171*84441f85SGarrett D'Amore 				continue;
172*84441f85SGarrett D'Amore 			else if (*p == ';') {
173*84441f85SGarrett D'Amore 				p++;
174*84441f85SGarrett D'Amore 				goto semicolon;
175*84441f85SGarrett D'Amore 			}
176*84441f85SGarrett D'Amore 		}
177*84441f85SGarrett D'Amore 		if ((*link = cmd = malloc(sizeof (struct s_command))) == NULL)
178*84441f85SGarrett D'Amore 			err(1, "malloc");
179*84441f85SGarrett D'Amore 		link = &cmd->next;
180*84441f85SGarrett D'Amore 		cmd->startline = cmd->nonsel = 0;
181*84441f85SGarrett D'Amore 		/* First parse the addresses */
182*84441f85SGarrett D'Amore 		naddr = 0;
183*84441f85SGarrett D'Amore 
184*84441f85SGarrett D'Amore /* Valid characters to start an address */
185*84441f85SGarrett D'Amore #define	addrchar(c)	(strchr("0123456789/\\$", (c)))
186*84441f85SGarrett D'Amore 		if (addrchar(*p)) {
187*84441f85SGarrett D'Amore 			naddr++;
188*84441f85SGarrett D'Amore 			if ((cmd->a1 = malloc(sizeof (struct s_addr))) == NULL)
189*84441f85SGarrett D'Amore 				err(1, "malloc");
190*84441f85SGarrett D'Amore 			p = compile_addr(p, cmd->a1);
191*84441f85SGarrett D'Amore 			EATSPACE();				/* EXTENSION */
192*84441f85SGarrett D'Amore 			if (*p == ',') {
193*84441f85SGarrett D'Amore 				p++;
194*84441f85SGarrett D'Amore 				EATSPACE();			/* EXTENSION */
195*84441f85SGarrett D'Amore 				naddr++;
196*84441f85SGarrett D'Amore 				if ((cmd->a2 = malloc(sizeof (struct s_addr)))
197*84441f85SGarrett D'Amore 				    == NULL)
198*84441f85SGarrett D'Amore 					err(1, "malloc");
199*84441f85SGarrett D'Amore 				p = compile_addr(p, cmd->a2);
200*84441f85SGarrett D'Amore 				EATSPACE();
201*84441f85SGarrett D'Amore 			} else
202*84441f85SGarrett D'Amore 				cmd->a2 = 0;
203*84441f85SGarrett D'Amore 		} else
204*84441f85SGarrett D'Amore 			cmd->a1 = cmd->a2 = 0;
205*84441f85SGarrett D'Amore 
206*84441f85SGarrett D'Amore nonsel:		/* Now parse the command */
207*84441f85SGarrett D'Amore 		if (!*p)
208*84441f85SGarrett D'Amore 			fatal(_("command expected"));
209*84441f85SGarrett D'Amore 		cmd->code = *p;
210*84441f85SGarrett D'Amore 		for (fp = cmd_fmts; fp->code; fp++)
211*84441f85SGarrett D'Amore 			if (fp->code == *p)
212*84441f85SGarrett D'Amore 				break;
213*84441f85SGarrett D'Amore 		if (!fp->code)
214*84441f85SGarrett D'Amore 			fatal(_("invalid command code %c"), *p);
215*84441f85SGarrett D'Amore 		if (naddr > fp->naddr)
216*84441f85SGarrett D'Amore 			fatal(_("command %c expects up to %d address(es), "
217*84441f85SGarrett D'Amore 			    "found %d"), *p, fp->naddr, naddr);
218*84441f85SGarrett D'Amore 		switch (fp->args) {
219*84441f85SGarrett D'Amore 		case NONSEL:			/* ! */
220*84441f85SGarrett D'Amore 			p++;
221*84441f85SGarrett D'Amore 			EATSPACE();
222*84441f85SGarrett D'Amore 			cmd->nonsel = 1;
223*84441f85SGarrett D'Amore 			goto nonsel;
224*84441f85SGarrett D'Amore 		case GROUP:			/* { */
225*84441f85SGarrett D'Amore 			p++;
226*84441f85SGarrett D'Amore 			EATSPACE();
227*84441f85SGarrett D'Amore 			cmd->next = stack;
228*84441f85SGarrett D'Amore 			stack = cmd;
229*84441f85SGarrett D'Amore 			link = &cmd->u.c;
230*84441f85SGarrett D'Amore 			if (*p)
231*84441f85SGarrett D'Amore 				goto semicolon;
232*84441f85SGarrett D'Amore 			break;
233*84441f85SGarrett D'Amore 		case ENDGROUP:
234*84441f85SGarrett D'Amore 			/*
235*84441f85SGarrett D'Amore 			 * Short-circuit command processing, since end of
236*84441f85SGarrett D'Amore 			 * group is really just a noop.
237*84441f85SGarrett D'Amore 			 */
238*84441f85SGarrett D'Amore 			cmd->nonsel = 1;
239*84441f85SGarrett D'Amore 			if (stack == 0)
240*84441f85SGarrett D'Amore 				fatal(_("unexpected }"));
241*84441f85SGarrett D'Amore 			cmd2 = stack;
242*84441f85SGarrett D'Amore 			stack = cmd2->next;
243*84441f85SGarrett D'Amore 			cmd2->next = cmd;
244*84441f85SGarrett D'Amore 			/*FALLTHROUGH*/
245*84441f85SGarrett D'Amore 		case EMPTY:		/* d D g G h H l n N p P q x = \0 */
246*84441f85SGarrett D'Amore 			p++;
247*84441f85SGarrett D'Amore 			EATSPACE();
248*84441f85SGarrett D'Amore 			if (*p == ';') {
249*84441f85SGarrett D'Amore 				p++;
250*84441f85SGarrett D'Amore 				link = &cmd->next;
251*84441f85SGarrett D'Amore 				goto semicolon;
252*84441f85SGarrett D'Amore 			}
253*84441f85SGarrett D'Amore 			if (*p)
254*84441f85SGarrett D'Amore 				fatal(_("extra characters at the end of %c "
255*84441f85SGarrett D'Amore 				    "command"), cmd->code);
256*84441f85SGarrett D'Amore 			break;
257*84441f85SGarrett D'Amore 		case TEXT:			/* a c i */
258*84441f85SGarrett D'Amore 			p++;
259*84441f85SGarrett D'Amore 			EATSPACE();
260*84441f85SGarrett D'Amore 			if (*p != '\\')
261*84441f85SGarrett D'Amore 				fatal(_("command %c expects \\ "
262*84441f85SGarrett D'Amore 				    "followed by text"), cmd->code);
263*84441f85SGarrett D'Amore 			p++;
264*84441f85SGarrett D'Amore 			EATSPACE();
265*84441f85SGarrett D'Amore 			if (*p)
266*84441f85SGarrett D'Amore 				fatal(_("extra characters after \\ "
267*84441f85SGarrett D'Amore 				    "at the end of %c command"),
268*84441f85SGarrett D'Amore 				    cmd->code);
269*84441f85SGarrett D'Amore 			cmd->t = compile_text();
270*84441f85SGarrett D'Amore 			break;
271*84441f85SGarrett D'Amore 		case COMMENT:			/* \0 # */
272*84441f85SGarrett D'Amore 			break;
273*84441f85SGarrett D'Amore 		case WFILE:			/* w */
274*84441f85SGarrett D'Amore 			p++;
275*84441f85SGarrett D'Amore 			EATSPACE();
276*84441f85SGarrett D'Amore 			if (*p == '\0')
277*84441f85SGarrett D'Amore 				fatal(_("filename expected"));
278*84441f85SGarrett D'Amore 			cmd->t = duptoeol(p, "w command");
279*84441f85SGarrett D'Amore 			if (aflag)
280*84441f85SGarrett D'Amore 				cmd->u.fd = -1;
281*84441f85SGarrett D'Amore 			else if ((cmd->u.fd = open(p,
282*84441f85SGarrett D'Amore 			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 0666)) == -1)
283*84441f85SGarrett D'Amore 				err(1, "%s", p);
284*84441f85SGarrett D'Amore 			break;
285*84441f85SGarrett D'Amore 		case RFILE:			/* r */
286*84441f85SGarrett D'Amore 			p++;
287*84441f85SGarrett D'Amore 			EATSPACE();
288*84441f85SGarrett D'Amore 			if (*p == '\0')
289*84441f85SGarrett D'Amore 				fatal(_("filename expected"));
290*84441f85SGarrett D'Amore 			else
291*84441f85SGarrett D'Amore 				cmd->t = duptoeol(p, "read command");
292*84441f85SGarrett D'Amore 			break;
293*84441f85SGarrett D'Amore 		case BRANCH:			/* b t */
294*84441f85SGarrett D'Amore 			p++;
295*84441f85SGarrett D'Amore 			EATSPACE();
296*84441f85SGarrett D'Amore 			if (*p == '\0')
297*84441f85SGarrett D'Amore 				cmd->t = NULL;
298*84441f85SGarrett D'Amore 			else
299*84441f85SGarrett D'Amore 				cmd->t = duptoeol(p, "branch");
300*84441f85SGarrett D'Amore 			break;
301*84441f85SGarrett D'Amore 		case LABEL:			/* : */
302*84441f85SGarrett D'Amore 			p++;
303*84441f85SGarrett D'Amore 			EATSPACE();
304*84441f85SGarrett D'Amore 			cmd->t = duptoeol(p, "label");
305*84441f85SGarrett D'Amore 			if (strlen(p) == 0)
306*84441f85SGarrett D'Amore 				fatal(_("empty label"));
307*84441f85SGarrett D'Amore 			enterlabel(cmd);
308*84441f85SGarrett D'Amore 			break;
309*84441f85SGarrett D'Amore 		case SUBST:			/* s */
310*84441f85SGarrett D'Amore 			p++;
311*84441f85SGarrett D'Amore 			if (*p == '\0' || *p == '\\')
312*84441f85SGarrett D'Amore 				fatal(_("substitute pattern can not "
313*84441f85SGarrett D'Amore 				    "be delimited by newline or backslash"));
314*84441f85SGarrett D'Amore 			if ((cmd->u.s = calloc(1, sizeof (struct s_subst))) ==
315*84441f85SGarrett D'Amore 			    NULL)
316*84441f85SGarrett D'Amore 				err(1, "malloc");
317*84441f85SGarrett D'Amore 			p = compile_delimited(p, re, 0);
318*84441f85SGarrett D'Amore 			if (p == NULL)
319*84441f85SGarrett D'Amore 				fatal(_("unterminated substitute pattern"));
320*84441f85SGarrett D'Amore 
321*84441f85SGarrett D'Amore 			/* Compile RE with no case sensitivity temporarily */
322*84441f85SGarrett D'Amore 			if (*re == '\0')
323*84441f85SGarrett D'Amore 				cmd->u.s->re = NULL;
324*84441f85SGarrett D'Amore 			else
325*84441f85SGarrett D'Amore 				cmd->u.s->re = compile_re(re, 0);
326*84441f85SGarrett D'Amore 			--p;
327*84441f85SGarrett D'Amore 			p = compile_subst(p, cmd->u.s);
328*84441f85SGarrett D'Amore 			p = compile_flags(p, cmd->u.s);
329*84441f85SGarrett D'Amore 
330*84441f85SGarrett D'Amore 			/* Recompile RE with case sens. from "I" flag if any */
331*84441f85SGarrett D'Amore 			if (*re == '\0')
332*84441f85SGarrett D'Amore 				cmd->u.s->re = NULL;
333*84441f85SGarrett D'Amore 			else
334*84441f85SGarrett D'Amore 				cmd->u.s->re = compile_re(re, cmd->u.s->icase);
335*84441f85SGarrett D'Amore 			EATSPACE();
336*84441f85SGarrett D'Amore 			if (*p == ';') {
337*84441f85SGarrett D'Amore 				p++;
338*84441f85SGarrett D'Amore 				link = &cmd->next;
339*84441f85SGarrett D'Amore 				goto semicolon;
340*84441f85SGarrett D'Amore 			}
341*84441f85SGarrett D'Amore 			break;
342*84441f85SGarrett D'Amore 		case TR:			/* y */
343*84441f85SGarrett D'Amore 			p++;
344*84441f85SGarrett D'Amore 			p = compile_tr(p, &cmd->u.y);
345*84441f85SGarrett D'Amore 			EATSPACE();
346*84441f85SGarrett D'Amore 			if (*p == ';') {
347*84441f85SGarrett D'Amore 				p++;
348*84441f85SGarrett D'Amore 				link = &cmd->next;
349*84441f85SGarrett D'Amore 				goto semicolon;
350*84441f85SGarrett D'Amore 			}
351*84441f85SGarrett D'Amore 			if (*p)
352*84441f85SGarrett D'Amore 				fatal(_("extra text at the end of a "
353*84441f85SGarrett D'Amore 				    "transform command"));
354*84441f85SGarrett D'Amore 			break;
355*84441f85SGarrett D'Amore 		}
356*84441f85SGarrett D'Amore 	}
357*84441f85SGarrett D'Amore }
358*84441f85SGarrett D'Amore 
359*84441f85SGarrett D'Amore /*
360*84441f85SGarrett D'Amore  * Get a delimited string.  P points to the delimeter of the string; d points
361*84441f85SGarrett D'Amore  * to a buffer area.  Newline and delimiter escapes are processed; other
362*84441f85SGarrett D'Amore  * escapes are ignored.
363*84441f85SGarrett D'Amore  *
364*84441f85SGarrett D'Amore  * Returns a pointer to the first character after the final delimiter or NULL
365*84441f85SGarrett D'Amore  * in the case of a non-terminated string.  The character array d is filled
366*84441f85SGarrett D'Amore  * with the processed string.
367*84441f85SGarrett D'Amore  */
368*84441f85SGarrett D'Amore static char *
compile_delimited(char * p,char * d,int is_tr)369*84441f85SGarrett D'Amore compile_delimited(char *p, char *d, int is_tr)
370*84441f85SGarrett D'Amore {
371*84441f85SGarrett D'Amore 	char c;
372*84441f85SGarrett D'Amore 
373*84441f85SGarrett D'Amore 	c = *p++;
374*84441f85SGarrett D'Amore 	if (c == '\0')
375*84441f85SGarrett D'Amore 		return (NULL);
376*84441f85SGarrett D'Amore 	else if (c == '\\')
377*84441f85SGarrett D'Amore 		fatal(_("\\ can not be used as a string delimiter"));
378*84441f85SGarrett D'Amore 	else if (c == '\n')
379*84441f85SGarrett D'Amore 		fatal(_("newline can not be used as a string delimiter"));
380*84441f85SGarrett D'Amore 	while (*p) {
381*84441f85SGarrett D'Amore 		if (*p == '[' && *p != c) {
382*84441f85SGarrett D'Amore 			if ((d = compile_ccl(&p, d)) == NULL)
383*84441f85SGarrett D'Amore 				fatal(_("unbalanced brackets ([])"));
384*84441f85SGarrett D'Amore 			continue;
385*84441f85SGarrett D'Amore 		} else if (*p == '\\' && p[1] == '[') {
386*84441f85SGarrett D'Amore 			*d++ = *p++;
387*84441f85SGarrett D'Amore 		} else if (*p == '\\' && p[1] == c)
388*84441f85SGarrett D'Amore 			p++;
389*84441f85SGarrett D'Amore 		else if (*p == '\\' && p[1] == 'n') {
390*84441f85SGarrett D'Amore 			*d++ = '\n';
391*84441f85SGarrett D'Amore 			p += 2;
392*84441f85SGarrett D'Amore 			continue;
393*84441f85SGarrett D'Amore 		} else if (*p == '\\' && p[1] == '\\') {
394*84441f85SGarrett D'Amore 			if (is_tr)
395*84441f85SGarrett D'Amore 				p++;
396*84441f85SGarrett D'Amore 			else
397*84441f85SGarrett D'Amore 				*d++ = *p++;
398*84441f85SGarrett D'Amore 		} else if (*p == c) {
399*84441f85SGarrett D'Amore 			*d = '\0';
400*84441f85SGarrett D'Amore 			return (p + 1);
401*84441f85SGarrett D'Amore 		}
402*84441f85SGarrett D'Amore 		*d++ = *p++;
403*84441f85SGarrett D'Amore 	}
404*84441f85SGarrett D'Amore 	return (NULL);
405*84441f85SGarrett D'Amore }
406*84441f85SGarrett D'Amore 
407*84441f85SGarrett D'Amore 
408*84441f85SGarrett D'Amore /* compile_ccl: expand a POSIX character class */
409*84441f85SGarrett D'Amore static char *
compile_ccl(char ** sp,char * t)410*84441f85SGarrett D'Amore compile_ccl(char **sp, char *t)
411*84441f85SGarrett D'Amore {
412*84441f85SGarrett D'Amore 	int c, d;
413*84441f85SGarrett D'Amore 	char *s = *sp;
414*84441f85SGarrett D'Amore 
415*84441f85SGarrett D'Amore 	*t++ = *s++;
416*84441f85SGarrett D'Amore 	if (*s == '^')
417*84441f85SGarrett D'Amore 		*t++ = *s++;
418*84441f85SGarrett D'Amore 	if (*s == ']')
419*84441f85SGarrett D'Amore 		*t++ = *s++;
420*84441f85SGarrett D'Amore 	for (; *s && (*t = *s) != ']'; s++, t++)
421*84441f85SGarrett D'Amore 		if (*s == '[' &&
422*84441f85SGarrett D'Amore 		    ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
423*84441f85SGarrett D'Amore 			*++t = *++s, t++, s++;
424*84441f85SGarrett D'Amore 			for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
425*84441f85SGarrett D'Amore 				if ((c = *s) == '\0')
426*84441f85SGarrett D'Amore 					return (NULL);
427*84441f85SGarrett D'Amore 		}
428*84441f85SGarrett D'Amore 	return ((*s == ']') ? *sp = ++s, ++t : NULL);
429*84441f85SGarrett D'Amore }
430*84441f85SGarrett D'Amore 
431*84441f85SGarrett D'Amore /*
432*84441f85SGarrett D'Amore  * Compiles the regular expression in RE and returns a pointer to the compiled
433*84441f85SGarrett D'Amore  * regular expression.
434*84441f85SGarrett D'Amore  * Cflags are passed to regcomp.
435*84441f85SGarrett D'Amore  */
436*84441f85SGarrett D'Amore static regex_t *
compile_re(char * re,int case_insensitive)437*84441f85SGarrett D'Amore compile_re(char *re, int case_insensitive)
438*84441f85SGarrett D'Amore {
439*84441f85SGarrett D'Amore 	regex_t *rep;
440*84441f85SGarrett D'Amore 	int eval, flags;
441*84441f85SGarrett D'Amore 
442*84441f85SGarrett D'Amore 
443*84441f85SGarrett D'Amore 	flags = rflags;
444*84441f85SGarrett D'Amore 	if (case_insensitive)
445*84441f85SGarrett D'Amore 		flags |= REG_ICASE;
446*84441f85SGarrett D'Amore 	if ((rep = malloc(sizeof (regex_t))) == NULL)
447*84441f85SGarrett D'Amore 		err(1, "malloc");
448*84441f85SGarrett D'Amore 	if ((eval = regcomp(rep, re, flags)) != 0)
449*84441f85SGarrett D'Amore 		fatal(_("RE error: %s"), strregerror(eval, rep));
450*84441f85SGarrett D'Amore 	if (maxnsub < rep->re_nsub)
451*84441f85SGarrett D'Amore 		maxnsub = rep->re_nsub;
452*84441f85SGarrett D'Amore 	return (rep);
453*84441f85SGarrett D'Amore }
454*84441f85SGarrett D'Amore 
455*84441f85SGarrett D'Amore /*
456*84441f85SGarrett D'Amore  * Compile the substitution string of a regular expression and set res to
457*84441f85SGarrett D'Amore  * point to a saved copy of it.  Nsub is the number of parenthesized regular
458*84441f85SGarrett D'Amore  * expressions.
459*84441f85SGarrett D'Amore  */
460*84441f85SGarrett D'Amore static char *
compile_subst(char * p,struct s_subst * s)461*84441f85SGarrett D'Amore compile_subst(char *p, struct s_subst *s)
462*84441f85SGarrett D'Amore {
463*84441f85SGarrett D'Amore 	static char lbuf[_POSIX2_LINE_MAX + 1];
464*84441f85SGarrett D'Amore 	int asize;
465*84441f85SGarrett D'Amore 	uintptr_t size;
466*84441f85SGarrett D'Amore 	uchar_t ref;
467*84441f85SGarrett D'Amore 	char c, *text, *op, *sp;
468*84441f85SGarrett D'Amore 	int more = 1, sawesc = 0;
469*84441f85SGarrett D'Amore 
470*84441f85SGarrett D'Amore 	c = *p++;			/* Terminator character */
471*84441f85SGarrett D'Amore 	if (c == '\0')
472*84441f85SGarrett D'Amore 		return (NULL);
473*84441f85SGarrett D'Amore 
474*84441f85SGarrett D'Amore 	s->maxbref = 0;
475*84441f85SGarrett D'Amore 	s->linenum = linenum;
476*84441f85SGarrett D'Amore 	asize = 2 * _POSIX2_LINE_MAX + 1;
477*84441f85SGarrett D'Amore 	if ((text = malloc(asize)) == NULL)
478*84441f85SGarrett D'Amore 		err(1, "malloc");
479*84441f85SGarrett D'Amore 	size = 0;
480*84441f85SGarrett D'Amore 	do {
481*84441f85SGarrett D'Amore 		op = sp = text + size;
482*84441f85SGarrett D'Amore 		for (; *p; p++) {
483*84441f85SGarrett D'Amore 			if (*p == '\\' || sawesc) {
484*84441f85SGarrett D'Amore 				/*
485*84441f85SGarrett D'Amore 				 * If this is a continuation from the last
486*84441f85SGarrett D'Amore 				 * buffer, we won't have a character to
487*84441f85SGarrett D'Amore 				 * skip over.
488*84441f85SGarrett D'Amore 				 */
489*84441f85SGarrett D'Amore 				if (sawesc)
490*84441f85SGarrett D'Amore 					sawesc = 0;
491*84441f85SGarrett D'Amore 				else
492*84441f85SGarrett D'Amore 					p++;
493*84441f85SGarrett D'Amore 
494*84441f85SGarrett D'Amore 				if (*p == '\0') {
495*84441f85SGarrett D'Amore 					/*
496*84441f85SGarrett D'Amore 					 * This escaped character is continued
497*84441f85SGarrett D'Amore 					 * in the next part of the line.  Note
498*84441f85SGarrett D'Amore 					 * this fact, then cause the loop to
499*84441f85SGarrett D'Amore 					 * exit w/ normal EOL case and reenter
500*84441f85SGarrett D'Amore 					 * above with the new buffer.
501*84441f85SGarrett D'Amore 					 */
502*84441f85SGarrett D'Amore 					sawesc = 1;
503*84441f85SGarrett D'Amore 					p--;
504*84441f85SGarrett D'Amore 					continue;
505*84441f85SGarrett D'Amore 				} else if (strchr("123456789", *p) != NULL) {
506*84441f85SGarrett D'Amore 					*sp++ = '\\';
507*84441f85SGarrett D'Amore 					ref = *p - '0';
508*84441f85SGarrett D'Amore 					if (s->re != NULL &&
509*84441f85SGarrett D'Amore 					    ref > s->re->re_nsub)
510*84441f85SGarrett D'Amore 						fatal(_("not defined in "
511*84441f85SGarrett D'Amore 						    "the RE: \\%c"), *p);
512*84441f85SGarrett D'Amore 					if (s->maxbref < ref)
513*84441f85SGarrett D'Amore 						s->maxbref = ref;
514*84441f85SGarrett D'Amore 				} else if (*p == '&' || *p == '\\')
515*84441f85SGarrett D'Amore 					*sp++ = '\\';
516*84441f85SGarrett D'Amore 			} else if (*p == c) {
517*84441f85SGarrett D'Amore 				if (*++p == '\0' && more) {
518*84441f85SGarrett D'Amore 					if (cu_fgets(lbuf, sizeof (lbuf),
519*84441f85SGarrett D'Amore 					    &more))
520*84441f85SGarrett D'Amore 						p = lbuf;
521*84441f85SGarrett D'Amore 				}
522*84441f85SGarrett D'Amore 				*sp++ = '\0';
523*84441f85SGarrett D'Amore 				size += (uintptr_t)sp - (uintptr_t)op;
524*84441f85SGarrett D'Amore 				if ((s->new = realloc(text, size)) == NULL)
525*84441f85SGarrett D'Amore 					err(1, "realloc");
526*84441f85SGarrett D'Amore 				return (p);
527*84441f85SGarrett D'Amore 			} else if (*p == '\n') {
528*84441f85SGarrett D'Amore 				fatal(_("unescaped newline inside "
529*84441f85SGarrett D'Amore 				    "substitute pattern"));
530*84441f85SGarrett D'Amore 				/* NOTREACHED */
531*84441f85SGarrett D'Amore 			}
532*84441f85SGarrett D'Amore 			*sp++ = *p;
533*84441f85SGarrett D'Amore 		}
534*84441f85SGarrett D'Amore 		size += (uintptr_t)sp - (uintptr_t)op;
535*84441f85SGarrett D'Amore 		if (asize - size < _POSIX2_LINE_MAX + 1) {
536*84441f85SGarrett D'Amore 			asize *= 2;
537*84441f85SGarrett D'Amore 			if ((text = realloc(text, asize)) == NULL)
538*84441f85SGarrett D'Amore 				err(1, "realloc");
539*84441f85SGarrett D'Amore 		}
540*84441f85SGarrett D'Amore 	} while (cu_fgets(p = lbuf, sizeof (lbuf), &more));
541*84441f85SGarrett D'Amore 	fatal(_("unterminated substitute in regular expression"));
542*84441f85SGarrett D'Amore 	return (NULL);
543*84441f85SGarrett D'Amore }
544*84441f85SGarrett D'Amore 
545*84441f85SGarrett D'Amore /*
546*84441f85SGarrett D'Amore  * Compile the flags of the s command
547*84441f85SGarrett D'Amore  */
548*84441f85SGarrett D'Amore static char *
compile_flags(char * p,struct s_subst * s)549*84441f85SGarrett D'Amore compile_flags(char *p, struct s_subst *s)
550*84441f85SGarrett D'Amore {
551*84441f85SGarrett D'Amore 	int gn;			/* True if we have seen g or n */
552*84441f85SGarrett D'Amore 	unsigned long nval;
553*84441f85SGarrett D'Amore 	char wfile[_POSIX2_LINE_MAX + 1], *q;
554*84441f85SGarrett D'Amore 
555*84441f85SGarrett D'Amore 	s->n = 1;				/* Default */
556*84441f85SGarrett D'Amore 	s->p = 0;
557*84441f85SGarrett D'Amore 	s->wfile = NULL;
558*84441f85SGarrett D'Amore 	s->wfd = -1;
559*84441f85SGarrett D'Amore 	s->icase = 0;
560*84441f85SGarrett D'Amore 	gn = 0;
561*84441f85SGarrett D'Amore 	for (;;) {
562*84441f85SGarrett D'Amore 		EATSPACE();			/* EXTENSION */
563*84441f85SGarrett D'Amore 		switch (*p) {
564*84441f85SGarrett D'Amore 		case 'g':
565*84441f85SGarrett D'Amore 			if (gn)
566*84441f85SGarrett D'Amore 				fatal(_("more than one number or "
567*84441f85SGarrett D'Amore 				    "'g' in substitute flags"));
568*84441f85SGarrett D'Amore 			gn = 1;
569*84441f85SGarrett D'Amore 			s->n = 0;
570*84441f85SGarrett D'Amore 			break;
571*84441f85SGarrett D'Amore 		case '\0':
572*84441f85SGarrett D'Amore 		case '\n':
573*84441f85SGarrett D'Amore 		case ';':
574*84441f85SGarrett D'Amore 			return (p);
575*84441f85SGarrett D'Amore 		case 'p':
576*84441f85SGarrett D'Amore 			s->p = 1;
577*84441f85SGarrett D'Amore 			break;
578*84441f85SGarrett D'Amore 		case 'I':
579*84441f85SGarrett D'Amore 			s->icase = 1;
580*84441f85SGarrett D'Amore 			break;
581*84441f85SGarrett D'Amore 		case '1': case '2': case '3':
582*84441f85SGarrett D'Amore 		case '4': case '5': case '6':
583*84441f85SGarrett D'Amore 		case '7': case '8': case '9':
584*84441f85SGarrett D'Amore 			if (gn)
585*84441f85SGarrett D'Amore 				fatal(_("more than one number or "
586*84441f85SGarrett D'Amore 				    "'g' in substitute flags"));
587*84441f85SGarrett D'Amore 			gn = 1;
588*84441f85SGarrett D'Amore 			errno = 0;
589*84441f85SGarrett D'Amore 			nval = strtol(p, &p, 10);
590*84441f85SGarrett D'Amore 			if (errno == ERANGE || nval > INT_MAX)
591*84441f85SGarrett D'Amore 				fatal(_("overflow in the 'N' substitute flag"));
592*84441f85SGarrett D'Amore 			s->n = nval;
593*84441f85SGarrett D'Amore 			p--;
594*84441f85SGarrett D'Amore 			break;
595*84441f85SGarrett D'Amore 		case 'w':
596*84441f85SGarrett D'Amore 			p++;
597*84441f85SGarrett D'Amore #ifdef HISTORIC_PRACTICE
598*84441f85SGarrett D'Amore 			if (*p != ' ') {
599*84441f85SGarrett D'Amore 				fatal(_("space missing before w wfile"));
600*84441f85SGarrett D'Amore 				return (p);
601*84441f85SGarrett D'Amore 			}
602*84441f85SGarrett D'Amore #endif
603*84441f85SGarrett D'Amore 			EATSPACE();
604*84441f85SGarrett D'Amore 			q = wfile;
605*84441f85SGarrett D'Amore 			while (*p) {
606*84441f85SGarrett D'Amore 				if (*p == '\n')
607*84441f85SGarrett D'Amore 					break;
608*84441f85SGarrett D'Amore 				*q++ = *p++;
609*84441f85SGarrett D'Amore 			}
610*84441f85SGarrett D'Amore 			*q = '\0';
611*84441f85SGarrett D'Amore 			if (q == wfile)
612*84441f85SGarrett D'Amore 				fatal(_("no wfile specified"));
613*84441f85SGarrett D'Amore 			s->wfile = strdup(wfile);
614*84441f85SGarrett D'Amore 			if (!aflag && (s->wfd = open(wfile,
615*84441f85SGarrett D'Amore 			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 0666)) == -1)
616*84441f85SGarrett D'Amore 				err(1, "%s", wfile);
617*84441f85SGarrett D'Amore 			return (p);
618*84441f85SGarrett D'Amore 		default:
619*84441f85SGarrett D'Amore 			fatal(_("bad flag in substitute command: '%c'"), *p);
620*84441f85SGarrett D'Amore 			break;
621*84441f85SGarrett D'Amore 		}
622*84441f85SGarrett D'Amore 		p++;
623*84441f85SGarrett D'Amore 	}
624*84441f85SGarrett D'Amore }
625*84441f85SGarrett D'Amore 
626*84441f85SGarrett D'Amore /*
627*84441f85SGarrett D'Amore  * Compile a translation set of strings into a lookup table.
628*84441f85SGarrett D'Amore  */
629*84441f85SGarrett D'Amore static char *
compile_tr(char * p,struct s_tr ** py)630*84441f85SGarrett D'Amore compile_tr(char *p, struct s_tr **py)
631*84441f85SGarrett D'Amore {
632*84441f85SGarrett D'Amore 	struct s_tr *y;
633*84441f85SGarrett D'Amore 	int i;
634*84441f85SGarrett D'Amore 	const char *op, *np;
635*84441f85SGarrett D'Amore 	char old[_POSIX2_LINE_MAX + 1];
636*84441f85SGarrett D'Amore 	char new[_POSIX2_LINE_MAX + 1];
637*84441f85SGarrett D'Amore 	size_t oclen, oldlen, nclen, newlen;
638*84441f85SGarrett D'Amore 	mbstate_t mbs1, mbs2;
639*84441f85SGarrett D'Amore 
640*84441f85SGarrett D'Amore 	if ((*py = y = malloc(sizeof (*y))) == NULL)
641*84441f85SGarrett D'Amore 		err(1, NULL);
642*84441f85SGarrett D'Amore 	y->multis = NULL;
643*84441f85SGarrett D'Amore 	y->nmultis = 0;
644*84441f85SGarrett D'Amore 
645*84441f85SGarrett D'Amore 	if (*p == '\0' || *p == '\\')
646*84441f85SGarrett D'Amore 		fatal(_("transform pattern can not be delimited by "
647*84441f85SGarrett D'Amore 		    "newline or backslash"));
648*84441f85SGarrett D'Amore 	p = compile_delimited(p, old, 1);
649*84441f85SGarrett D'Amore 	if (p == NULL)
650*84441f85SGarrett D'Amore 		fatal(_("unterminated transform source string"));
651*84441f85SGarrett D'Amore 	p = compile_delimited(p - 1, new, 1);
652*84441f85SGarrett D'Amore 	if (p == NULL)
653*84441f85SGarrett D'Amore 		fatal(_("unterminated transform target string"));
654*84441f85SGarrett D'Amore 	EATSPACE();
655*84441f85SGarrett D'Amore 	op = old;
656*84441f85SGarrett D'Amore 	oldlen = mbsrtowcs(NULL, &op, 0, NULL);
657*84441f85SGarrett D'Amore 	if (oldlen == (size_t)-1)
658*84441f85SGarrett D'Amore 		err(1, NULL);
659*84441f85SGarrett D'Amore 	np = new;
660*84441f85SGarrett D'Amore 	newlen = mbsrtowcs(NULL, &np, 0, NULL);
661*84441f85SGarrett D'Amore 	if (newlen == (size_t)-1)
662*84441f85SGarrett D'Amore 		err(1, NULL);
663*84441f85SGarrett D'Amore 	if (newlen != oldlen)
664*84441f85SGarrett D'Amore 		fatal(_("transform strings are not the same length"));
665*84441f85SGarrett D'Amore 	if (MB_CUR_MAX == 1) {
666*84441f85SGarrett D'Amore 		/*
667*84441f85SGarrett D'Amore 		 * The single-byte encoding case is easy: generate a
668*84441f85SGarrett D'Amore 		 * lookup table.
669*84441f85SGarrett D'Amore 		 */
670*84441f85SGarrett D'Amore 		for (i = 0; i <= UCHAR_MAX; i++)
671*84441f85SGarrett D'Amore 			y->bytetab[i] = (char)i;
672*84441f85SGarrett D'Amore 		for (; *op; op++, np++)
673*84441f85SGarrett D'Amore 			y->bytetab[(uchar_t)*op] = *np;
674*84441f85SGarrett D'Amore 	} else {
675*84441f85SGarrett D'Amore 		/*
676*84441f85SGarrett D'Amore 		 * Multi-byte encoding case: generate a lookup table as
677*84441f85SGarrett D'Amore 		 * above, but only for single-byte characters. The first
678*84441f85SGarrett D'Amore 		 * bytes of multi-byte characters have their lookup table
679*84441f85SGarrett D'Amore 		 * entries set to 0, which causes do_tr() to search through
680*84441f85SGarrett D'Amore 		 * an auxiliary vector of multi-byte mappings.
681*84441f85SGarrett D'Amore 		 */
682*84441f85SGarrett D'Amore 		(void) memset(&mbs1, 0, sizeof (mbs1));
683*84441f85SGarrett D'Amore 		(void) memset(&mbs2, 0, sizeof (mbs2));
684*84441f85SGarrett D'Amore 		for (i = 0; i <= UCHAR_MAX; i++)
685*84441f85SGarrett D'Amore 			y->bytetab[i] = (btowc(i) != WEOF) ? (uchar_t)i : 0;
686*84441f85SGarrett D'Amore 		while (*op != '\0') {
687*84441f85SGarrett D'Amore 			oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
688*84441f85SGarrett D'Amore 			if (oclen == (size_t)-1 || oclen == (size_t)-2)
689*84441f85SGarrett D'Amore 				errx(1, "%s", strerror(EILSEQ));
690*84441f85SGarrett D'Amore 			nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
691*84441f85SGarrett D'Amore 			if (nclen == (size_t)-1 || nclen == (size_t)-2)
692*84441f85SGarrett D'Amore 				errx(1, "%s", strerror(EILSEQ));
693*84441f85SGarrett D'Amore 			if (oclen == 1 && nclen == 1)
694*84441f85SGarrett D'Amore 				y->bytetab[(uchar_t)*op] = *np;
695*84441f85SGarrett D'Amore 			else {
696*84441f85SGarrett D'Amore 				y->bytetab[(uchar_t)*op] = 0;
697*84441f85SGarrett D'Amore 				y->multis = realloc(y->multis,
698*84441f85SGarrett D'Amore 				    (y->nmultis + 1) * sizeof (*y->multis));
699*84441f85SGarrett D'Amore 				if (y->multis == NULL)
700*84441f85SGarrett D'Amore 					err(1, NULL);
701*84441f85SGarrett D'Amore 				i = y->nmultis++;
702*84441f85SGarrett D'Amore 				y->multis[i].fromlen = oclen;
703*84441f85SGarrett D'Amore 				(void) memcpy(y->multis[i].from, op, oclen);
704*84441f85SGarrett D'Amore 				y->multis[i].tolen = nclen;
705*84441f85SGarrett D'Amore 				(void) memcpy(y->multis[i].to, np, nclen);
706*84441f85SGarrett D'Amore 			}
707*84441f85SGarrett D'Amore 			op += oclen;
708*84441f85SGarrett D'Amore 			np += nclen;
709*84441f85SGarrett D'Amore 		}
710*84441f85SGarrett D'Amore 	}
711*84441f85SGarrett D'Amore 	return (p);
712*84441f85SGarrett D'Amore }
713*84441f85SGarrett D'Amore 
714*84441f85SGarrett D'Amore /*
715*84441f85SGarrett D'Amore  * Compile the text following an a or i command.
716*84441f85SGarrett D'Amore  */
717*84441f85SGarrett D'Amore static char *
compile_text(void)718*84441f85SGarrett D'Amore compile_text(void)
719*84441f85