xref: /freebsd/usr.bin/sed/compile.c (revision 0ea56610b435c9f689878d2e3bb282c485b29021)
19b50d902SRodney W. Grimes /*-
29b50d902SRodney W. Grimes  * Copyright (c) 1992 Diomidis Spinellis.
39b50d902SRodney W. Grimes  * Copyright (c) 1992, 1993
49b50d902SRodney W. Grimes  *	The Regents of the University of California.  All rights reserved.
59b50d902SRodney W. Grimes  *
69b50d902SRodney W. Grimes  * This code is derived from software contributed to Berkeley by
79b50d902SRodney W. Grimes  * Diomidis Spinellis of Imperial College, University of London.
89b50d902SRodney W. Grimes  *
99b50d902SRodney W. Grimes  * Redistribution and use in source and binary forms, with or without
109b50d902SRodney W. Grimes  * modification, are permitted provided that the following conditions
119b50d902SRodney W. Grimes  * are met:
129b50d902SRodney W. Grimes  * 1. Redistributions of source code must retain the above copyright
139b50d902SRodney W. Grimes  *    notice, this list of conditions and the following disclaimer.
149b50d902SRodney W. Grimes  * 2. Redistributions in binary form must reproduce the above copyright
159b50d902SRodney W. Grimes  *    notice, this list of conditions and the following disclaimer in the
169b50d902SRodney W. Grimes  *    documentation and/or other materials provided with the distribution.
179b50d902SRodney W. Grimes  * 3. All advertising materials mentioning features or use of this software
189b50d902SRodney W. Grimes  *    must display the following acknowledgement:
199b50d902SRodney W. Grimes  *	This product includes software developed by the University of
209b50d902SRodney W. Grimes  *	California, Berkeley and its contributors.
219b50d902SRodney W. Grimes  * 4. Neither the name of the University nor the names of its contributors
229b50d902SRodney W. Grimes  *    may be used to endorse or promote products derived from this software
239b50d902SRodney W. Grimes  *    without specific prior written permission.
249b50d902SRodney W. Grimes  *
259b50d902SRodney W. Grimes  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
269b50d902SRodney W. Grimes  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
279b50d902SRodney W. Grimes  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
289b50d902SRodney W. Grimes  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
299b50d902SRodney W. Grimes  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
309b50d902SRodney W. Grimes  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
319b50d902SRodney W. Grimes  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
329b50d902SRodney W. Grimes  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
339b50d902SRodney W. Grimes  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
349b50d902SRodney W. Grimes  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
359b50d902SRodney W. Grimes  * SUCH DAMAGE.
369b50d902SRodney W. Grimes  */
379b50d902SRodney W. Grimes 
389b50d902SRodney W. Grimes #ifndef lint
3973a08bb2SPhilippe Charnier #if 0
409b50d902SRodney W. Grimes static char sccsid[] = "@(#)compile.c	8.1 (Berkeley) 6/6/93";
4173a08bb2SPhilippe Charnier #endif
4273a08bb2SPhilippe Charnier static const char rcsid[] =
43c3aac50fSPeter Wemm   "$FreeBSD$";
449b50d902SRodney W. Grimes #endif /* not lint */
459b50d902SRodney W. Grimes 
469b50d902SRodney W. Grimes #include <sys/types.h>
479b50d902SRodney W. Grimes #include <sys/stat.h>
489b50d902SRodney W. Grimes 
499b50d902SRodney W. Grimes #include <ctype.h>
5073a08bb2SPhilippe Charnier #include <err.h>
519b50d902SRodney W. Grimes #include <fcntl.h>
529b50d902SRodney W. Grimes #include <limits.h>
539b50d902SRodney W. Grimes #include <regex.h>
549b50d902SRodney W. Grimes #include <stdio.h>
559b50d902SRodney W. Grimes #include <stdlib.h>
569b50d902SRodney W. Grimes #include <string.h>
579b50d902SRodney W. Grimes 
589b50d902SRodney W. Grimes #include "defs.h"
599b50d902SRodney W. Grimes #include "extern.h"
609b50d902SRodney W. Grimes 
619b50d902SRodney W. Grimes #define LHSZ	128
629b50d902SRodney W. Grimes #define	LHMASK	(LHSZ - 1)
639b50d902SRodney W. Grimes static struct labhash {
649b50d902SRodney W. Grimes 	struct	labhash *lh_next;
659b50d902SRodney W. Grimes 	u_int	lh_hash;
669b50d902SRodney W. Grimes 	struct	s_command *lh_cmd;
679b50d902SRodney W. Grimes 	int	lh_ref;
689b50d902SRodney W. Grimes } *labels[LHSZ];
699b50d902SRodney W. Grimes 
709b50d902SRodney W. Grimes static char	 *compile_addr __P((char *, struct s_addr *));
71ce19262dSJordan K. Hubbard static char	 *compile_ccl __P((char **, char *));
729b50d902SRodney W. Grimes static char	 *compile_delimited __P((char *, char *));
739b50d902SRodney W. Grimes static char	 *compile_flags __P((char *, struct s_subst *));
749b50d902SRodney W. Grimes static char	 *compile_re __P((char *, regex_t **));
759b50d902SRodney W. Grimes static char	 *compile_subst __P((char *, struct s_subst *));
769b50d902SRodney W. Grimes static char	 *compile_text __P((void));
779b50d902SRodney W. Grimes static char	 *compile_tr __P((char *, char **));
789b50d902SRodney W. Grimes static struct s_command
79ce19262dSJordan K. Hubbard 		**compile_stream __P((struct s_command **));
809b50d902SRodney W. Grimes static char	 *duptoeol __P((char *, char *));
819b50d902SRodney W. Grimes static void	  enterlabel __P((struct s_command *));
829b50d902SRodney W. Grimes static struct s_command
839b50d902SRodney W. Grimes 		 *findlabel __P((char *));
849b50d902SRodney W. Grimes static void	  fixuplabel __P((struct s_command *, struct s_command *));
859b50d902SRodney W. Grimes static void	  uselabel __P((void));
869b50d902SRodney W. Grimes 
879b50d902SRodney W. Grimes /*
889b50d902SRodney W. Grimes  * Command specification.  This is used to drive the command parser.
899b50d902SRodney W. Grimes  */
909b50d902SRodney W. Grimes struct s_format {
919b50d902SRodney W. Grimes 	char code;				/* Command code */
929b50d902SRodney W. Grimes 	int naddr;				/* Number of address args */
939b50d902SRodney W. Grimes 	enum e_args args;			/* Argument type */
949b50d902SRodney W. Grimes };
959b50d902SRodney W. Grimes 
969b50d902SRodney W. Grimes static struct s_format cmd_fmts[] = {
979b50d902SRodney W. Grimes 	{'{', 2, GROUP},
98ce19262dSJordan K. Hubbard 	{'}', 0, ENDGROUP},
999b50d902SRodney W. Grimes 	{'a', 1, TEXT},
1009b50d902SRodney W. Grimes 	{'b', 2, BRANCH},
1019b50d902SRodney W. Grimes 	{'c', 2, TEXT},
1029b50d902SRodney W. Grimes 	{'d', 2, EMPTY},
1039b50d902SRodney W. Grimes 	{'D', 2, EMPTY},
1049b50d902SRodney W. Grimes 	{'g', 2, EMPTY},
1059b50d902SRodney W. Grimes 	{'G', 2, EMPTY},
1069b50d902SRodney W. Grimes 	{'h', 2, EMPTY},
1079b50d902SRodney W. Grimes 	{'H', 2, EMPTY},
1089b50d902SRodney W. Grimes 	{'i', 1, TEXT},
1099b50d902SRodney W. Grimes 	{'l', 2, EMPTY},
1109b50d902SRodney W. Grimes 	{'n', 2, EMPTY},
1119b50d902SRodney W. Grimes 	{'N', 2, EMPTY},
1129b50d902SRodney W. Grimes 	{'p', 2, EMPTY},
1139b50d902SRodney W. Grimes 	{'P', 2, EMPTY},
1149b50d902SRodney W. Grimes 	{'q', 1, EMPTY},
1159b50d902SRodney W. Grimes 	{'r', 1, RFILE},
1169b50d902SRodney W. Grimes 	{'s', 2, SUBST},
1179b50d902SRodney W. Grimes 	{'t', 2, BRANCH},
1189b50d902SRodney W. Grimes 	{'w', 2, WFILE},
1199b50d902SRodney W. Grimes 	{'x', 2, EMPTY},
1209b50d902SRodney W. Grimes 	{'y', 2, TR},
1219b50d902SRodney W. Grimes 	{'!', 2, NONSEL},
1229b50d902SRodney W. Grimes 	{':', 0, LABEL},
1239b50d902SRodney W. Grimes 	{'#', 0, COMMENT},
1249b50d902SRodney W. Grimes 	{'=', 1, EMPTY},
1259b50d902SRodney W. Grimes 	{'\0', 0, COMMENT},
1269b50d902SRodney W. Grimes };
1279b50d902SRodney W. Grimes 
1289b50d902SRodney W. Grimes /* The compiled program. */
1299b50d902SRodney W. Grimes struct s_command *prog;
1309b50d902SRodney W. Grimes 
1319b50d902SRodney W. Grimes /*
1329b50d902SRodney W. Grimes  * Compile the program into prog.
1339b50d902SRodney W. Grimes  * Initialise appends.
1349b50d902SRodney W. Grimes  */
1359b50d902SRodney W. Grimes void
1369b50d902SRodney W. Grimes compile()
1379b50d902SRodney W. Grimes {
138ce19262dSJordan K. Hubbard 	*compile_stream(&prog) = NULL;
1399b50d902SRodney W. Grimes 	fixuplabel(prog, NULL);
1409b50d902SRodney W. Grimes 	uselabel();
1410ea56610SMike Heffner 	if (appendnum == 0)
1420ea56610SMike Heffner 		appends = NULL;
1430ea56610SMike Heffner 	else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
1440ea56610SMike Heffner 	    NULL)
1458e33c0a0SDavid E. O'Brien 		err(1, "malloc");
1468e33c0a0SDavid E. O'Brien 	if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
1478e33c0a0SDavid E. O'Brien 		err(1, "malloc");
1489b50d902SRodney W. Grimes }
1499b50d902SRodney W. Grimes 
1509b50d902SRodney W. Grimes #define EATSPACE() do {							\
1519b50d902SRodney W. Grimes 	if (p)								\
152726aebe5SAndrey A. Chernov 		while (*p && isspace((unsigned char)*p))                \
1539b50d902SRodney W. Grimes 			p++;						\
1549b50d902SRodney W. Grimes 	} while (0)
1559b50d902SRodney W. Grimes 
1569b50d902SRodney W. Grimes static struct s_command **
157ce19262dSJordan K. Hubbard compile_stream(link)
1589b50d902SRodney W. Grimes 	struct s_command **link;
1599b50d902SRodney W. Grimes {
160ce19262dSJordan K. Hubbard 	register char *p;
1619b50d902SRodney W. Grimes 	static char lbuf[_POSIX2_LINE_MAX + 1];	/* To save stack */
162ce19262dSJordan K. Hubbard 	struct s_command *cmd, *cmd2, *stack;
1639b50d902SRodney W. Grimes 	struct s_format *fp;
1649b50d902SRodney W. Grimes 	int naddr;				/* Number of addresses */
1659b50d902SRodney W. Grimes 
166ce19262dSJordan K. Hubbard 	stack = 0;
1679b50d902SRodney W. Grimes 	for (;;) {
168c56690efSArchie Cobbs 		if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
169ce19262dSJordan K. Hubbard 			if (stack != 0)
17073a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: unexpected EOF (pending }'s)",
17173a08bb2SPhilippe Charnier 							linenum, fname);
1729b50d902SRodney W. Grimes 			return (link);
1739b50d902SRodney W. Grimes 		}
1749b50d902SRodney W. Grimes 
1759b50d902SRodney W. Grimes semicolon:	EATSPACE();
1769b50d902SRodney W. Grimes 		if (p && (*p == '#' || *p == '\0'))
1779b50d902SRodney W. Grimes 			continue;
1788e33c0a0SDavid E. O'Brien 		if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
1798e33c0a0SDavid E. O'Brien 			err(1, "malloc");
1809b50d902SRodney W. Grimes 		link = &cmd->next;
1819b50d902SRodney W. Grimes 		cmd->nonsel = cmd->inrange = 0;
1829b50d902SRodney W. Grimes 		/* First parse the addresses */
1839b50d902SRodney W. Grimes 		naddr = 0;
1849b50d902SRodney W. Grimes 
1859b50d902SRodney W. Grimes /* Valid characters to start an address */
1869b50d902SRodney W. Grimes #define	addrchar(c)	(strchr("0123456789/\\$", (c)))
1879b50d902SRodney W. Grimes 		if (addrchar(*p)) {
1889b50d902SRodney W. Grimes 			naddr++;
1898e33c0a0SDavid E. O'Brien 			if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
1908e33c0a0SDavid E. O'Brien 				err(1, "malloc");
1919b50d902SRodney W. Grimes 			p = compile_addr(p, cmd->a1);
1929b50d902SRodney W. Grimes 			EATSPACE();				/* EXTENSION */
1939b50d902SRodney W. Grimes 			if (*p == ',') {
1949b50d902SRodney W. Grimes 				p++;
1959b50d902SRodney W. Grimes 				EATSPACE();			/* EXTENSION */
196ce19262dSJordan K. Hubbard 				naddr++;
1978e33c0a0SDavid E. O'Brien 				if ((cmd->a2 = malloc(sizeof(struct s_addr)))
1988e33c0a0SDavid E. O'Brien 				    == NULL)
1998e33c0a0SDavid E. O'Brien 					err(1, "malloc");
2009b50d902SRodney W. Grimes 				p = compile_addr(p, cmd->a2);
201ce19262dSJordan K. Hubbard 				EATSPACE();
202ce19262dSJordan K. Hubbard 			} else
203ce19262dSJordan K. Hubbard 				cmd->a2 = 0;
204ce19262dSJordan K. Hubbard 		} else
205ce19262dSJordan K. Hubbard 			cmd->a1 = cmd->a2 = 0;
2069b50d902SRodney W. Grimes 
2079b50d902SRodney W. Grimes nonsel:		/* Now parse the command */
2089b50d902SRodney W. Grimes 		if (!*p)
20973a08bb2SPhilippe Charnier 			errx(1, "%lu: %s: command expected", linenum, fname);
2109b50d902SRodney W. Grimes 		cmd->code = *p;
2119b50d902SRodney W. Grimes 		for (fp = cmd_fmts; fp->code; fp++)
2129b50d902SRodney W. Grimes 			if (fp->code == *p)
2139b50d902SRodney W. Grimes 				break;
2149b50d902SRodney W. Grimes 		if (!fp->code)
21573a08bb2SPhilippe Charnier 			errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
2169b50d902SRodney W. Grimes 		if (naddr > fp->naddr)
21773a08bb2SPhilippe Charnier 			errx(1,
21873a08bb2SPhilippe Charnier 				"%lu: %s: command %c expects up to %d address(es), found %d",
21973a08bb2SPhilippe Charnier 				linenum, fname, *p, fp->naddr, naddr);
2209b50d902SRodney W. Grimes 		switch (fp->args) {
2219b50d902SRodney W. Grimes 		case NONSEL:			/* ! */
2229b50d902SRodney W. Grimes 			p++;
223ce19262dSJordan K. Hubbard 			EATSPACE();
224ce19262dSJordan K. Hubbard 			cmd->nonsel = ! cmd->nonsel;
2259b50d902SRodney W. Grimes 			goto nonsel;
2269b50d902SRodney W. Grimes 		case GROUP:			/* { */
2279b50d902SRodney W. Grimes 			p++;
2289b50d902SRodney W. Grimes 			EATSPACE();
229ce19262dSJordan K. Hubbard 			cmd->next = stack;
230ce19262dSJordan K. Hubbard 			stack = cmd;
231ce19262dSJordan K. Hubbard 			link = &cmd->u.c;
232ce19262dSJordan K. Hubbard 			if (*p)
233ce19262dSJordan K. Hubbard 				goto semicolon;
2349b50d902SRodney W. Grimes 			break;
235ce19262dSJordan K. Hubbard 		case ENDGROUP:
236ce19262dSJordan K. Hubbard 			/*
237ce19262dSJordan K. Hubbard 			 * Short-circuit command processing, since end of
238ce19262dSJordan K. Hubbard 			 * group is really just a noop.
239ce19262dSJordan K. Hubbard 			 */
240ce19262dSJordan K. Hubbard 			cmd->nonsel = 1;
241ce19262dSJordan K. Hubbard 			if (stack == 0)
24273a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: unexpected }", linenum, fname);
243ce19262dSJordan K. Hubbard 			cmd2 = stack;
244ce19262dSJordan K. Hubbard 			stack = cmd2->next;
245ce19262dSJordan K. Hubbard 			cmd2->next = cmd;
246ce19262dSJordan K. Hubbard 			/*FALLTHROUGH*/
2479b50d902SRodney W. Grimes 		case EMPTY:		/* d D g G h H l n N p P q x = \0 */
2489b50d902SRodney W. Grimes 			p++;
2499b50d902SRodney W. Grimes 			EATSPACE();
2509b50d902SRodney W. Grimes 			if (*p == ';') {
2519b50d902SRodney W. Grimes 				p++;
2529b50d902SRodney W. Grimes 				link = &cmd->next;
2539b50d902SRodney W. Grimes 				goto semicolon;
2549b50d902SRodney W. Grimes 			}
2559b50d902SRodney W. Grimes 			if (*p)
25673a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: extra characters at the end of %c command",
25773a08bb2SPhilippe Charnier 						linenum, fname, cmd->code);
2589b50d902SRodney W. Grimes 			break;
2599b50d902SRodney W. Grimes 		case TEXT:			/* a c i */
2609b50d902SRodney W. Grimes 			p++;
2619b50d902SRodney W. Grimes 			EATSPACE();
2629b50d902SRodney W. Grimes 			if (*p != '\\')
26373a08bb2SPhilippe Charnier 				errx(1,
26473a08bb2SPhilippe Charnier "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
2659b50d902SRodney W. Grimes 			p++;
2669b50d902SRodney W. Grimes 			EATSPACE();
2679b50d902SRodney W. Grimes 			if (*p)
26873a08bb2SPhilippe Charnier 				errx(1,
26973a08bb2SPhilippe Charnier 				"%lu: %s: extra characters after \\ at the end of %c command",
27073a08bb2SPhilippe Charnier 				linenum, fname, cmd->code);
2719b50d902SRodney W. Grimes 			cmd->t = compile_text();
2729b50d902SRodney W. Grimes 			break;
2739b50d902SRodney W. Grimes 		case COMMENT:			/* \0 # */
2749b50d902SRodney W. Grimes 			break;
2759b50d902SRodney W. Grimes 		case WFILE:			/* w */
2769b50d902SRodney W. Grimes 			p++;
2779b50d902SRodney W. Grimes 			EATSPACE();
2789b50d902SRodney W. Grimes 			if (*p == '\0')
27973a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: filename expected", linenum, fname);
2809b50d902SRodney W. Grimes 			cmd->t = duptoeol(p, "w command");
2819b50d902SRodney W. Grimes 			if (aflag)
2829b50d902SRodney W. Grimes 				cmd->u.fd = -1;
2839b50d902SRodney W. Grimes 			else if ((cmd->u.fd = open(p,
2849b50d902SRodney W. Grimes 			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
2859b50d902SRodney W. Grimes 			    DEFFILEMODE)) == -1)
28673a08bb2SPhilippe Charnier 				err(1, "%s", p);
2879b50d902SRodney W. Grimes 			break;
2889b50d902SRodney W. Grimes 		case RFILE:			/* r */
2899b50d902SRodney W. Grimes 			p++;
2909b50d902SRodney W. Grimes 			EATSPACE();
2919b50d902SRodney W. Grimes 			if (*p == '\0')
29273a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: filename expected", linenum, fname);
2939b50d902SRodney W. Grimes 			else
2949b50d902SRodney W. Grimes 				cmd->t = duptoeol(p, "read command");
2959b50d902SRodney W. Grimes 			break;
2969b50d902SRodney W. Grimes 		case BRANCH:			/* b t */
2979b50d902SRodney W. Grimes 			p++;
2989b50d902SRodney W. Grimes 			EATSPACE();
2999b50d902SRodney W. Grimes 			if (*p == '\0')
3009b50d902SRodney W. Grimes 				cmd->t = NULL;
3019b50d902SRodney W. Grimes 			else
3029b50d902SRodney W. Grimes 				cmd->t = duptoeol(p, "branch");
3039b50d902SRodney W. Grimes 			break;
3049b50d902SRodney W. Grimes 		case LABEL:			/* : */
3059b50d902SRodney W. Grimes 			p++;
3069b50d902SRodney W. Grimes 			EATSPACE();
3079b50d902SRodney W. Grimes 			cmd->t = duptoeol(p, "label");
3089b50d902SRodney W. Grimes 			if (strlen(p) == 0)
30973a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: empty label", linenum, fname);
3109b50d902SRodney W. Grimes 			enterlabel(cmd);
3119b50d902SRodney W. Grimes 			break;
3129b50d902SRodney W. Grimes 		case SUBST:			/* s */
3139b50d902SRodney W. Grimes 			p++;
3149b50d902SRodney W. Grimes 			if (*p == '\0' || *p == '\\')
31573a08bb2SPhilippe Charnier 				errx(1,
31673a08bb2SPhilippe Charnier "%lu: %s: substitute pattern can not be delimited by newline or backslash",
31773a08bb2SPhilippe Charnier 					linenum, fname);
3188e33c0a0SDavid E. O'Brien 			if ((cmd->u.s = malloc(sizeof(struct s_subst))) == NULL)
3198e33c0a0SDavid E. O'Brien 				err(1, "malloc");
3209b50d902SRodney W. Grimes 			p = compile_re(p, &cmd->u.s->re);
3219b50d902SRodney W. Grimes 			if (p == NULL)
32273a08bb2SPhilippe Charnier 				errx(1,
32373a08bb2SPhilippe Charnier 				"%lu: %s: unterminated substitute pattern", linenum, fname);
3249b50d902SRodney W. Grimes 			--p;
3259b50d902SRodney W. Grimes 			p = compile_subst(p, cmd->u.s);
3269b50d902SRodney W. Grimes 			p = compile_flags(p, cmd->u.s);
3279b50d902SRodney W. Grimes 			EATSPACE();
3289b50d902SRodney W. Grimes 			if (*p == ';') {
3299b50d902SRodney W. Grimes 				p++;
3309b50d902SRodney W. Grimes 				link = &cmd->next;
3319b50d902SRodney W. Grimes 				goto semicolon;
3329b50d902SRodney W. Grimes 			}
3339b50d902SRodney W. Grimes 			break;
3349b50d902SRodney W. Grimes 		case TR:			/* y */
3359b50d902SRodney W. Grimes 			p++;
3369b50d902SRodney W. Grimes 			p = compile_tr(p, (char **)&cmd->u.y);
3379b50d902SRodney W. Grimes 			EATSPACE();
3389b50d902SRodney W. Grimes 			if (*p == ';') {
3399b50d902SRodney W. Grimes 				p++;
3409b50d902SRodney W. Grimes 				link = &cmd->next;
3419b50d902SRodney W. Grimes 				goto semicolon;
3429b50d902SRodney W. Grimes 			}
3439b50d902SRodney W. Grimes 			if (*p)
34473a08bb2SPhilippe Charnier 				errx(1,
34573a08bb2SPhilippe Charnier "%lu: %s: extra text at the end of a transform command", linenum, fname);
3469b50d902SRodney W. Grimes 			break;
3479b50d902SRodney W. Grimes 		}
3489b50d902SRodney W. Grimes 	}
3499b50d902SRodney W. Grimes }
3509b50d902SRodney W. Grimes 
3519b50d902SRodney W. Grimes /*
3529b50d902SRodney W. Grimes  * Get a delimited string.  P points to the delimeter of the string; d points
3539b50d902SRodney W. Grimes  * to a buffer area.  Newline and delimiter escapes are processed; other
3549b50d902SRodney W. Grimes  * escapes are ignored.
3559b50d902SRodney W. Grimes  *
3569b50d902SRodney W. Grimes  * Returns a pointer to the first character after the final delimiter or NULL
3579b50d902SRodney W. Grimes  * in the case of a non-terminated string.  The character array d is filled
3589b50d902SRodney W. Grimes  * with the processed string.
3599b50d902SRodney W. Grimes  */
3609b50d902SRodney W. Grimes static char *
3619b50d902SRodney W. Grimes compile_delimited(p, d)
3629b50d902SRodney W. Grimes 	char *p, *d;
3639b50d902SRodney W. Grimes {
3649b50d902SRodney W. Grimes 	char c;
3659b50d902SRodney W. Grimes 
3669b50d902SRodney W. Grimes 	c = *p++;
3679b50d902SRodney W. Grimes 	if (c == '\0')
3689b50d902SRodney W. Grimes 		return (NULL);
3699b50d902SRodney W. Grimes 	else if (c == '\\')
37073a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: \\ can not be used as a string delimiter",
37173a08bb2SPhilippe Charnier 				linenum, fname);
3729b50d902SRodney W. Grimes 	else if (c == '\n')
37373a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: newline can not be used as a string delimiter",
37473a08bb2SPhilippe Charnier 				linenum, fname);
3759b50d902SRodney W. Grimes 	while (*p) {
376ce19262dSJordan K. Hubbard 		if (*p == '[') {
377ce19262dSJordan K. Hubbard 			if ((d = compile_ccl(&p, d)) == NULL)
37873a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
379ce19262dSJordan K. Hubbard 			continue;
380ce19262dSJordan K. Hubbard 		} else if (*p == '\\' && p[1] == '[') {
381ce19262dSJordan K. Hubbard 			*d++ = *p++;
382ce19262dSJordan K. Hubbard 		} else if (*p == '\\' && p[1] == c)
3839b50d902SRodney W. Grimes 			p++;
3849b50d902SRodney W. Grimes 		else if (*p == '\\' && p[1] == 'n') {
3859b50d902SRodney W. Grimes 			*d++ = '\n';
3869b50d902SRodney W. Grimes 			p += 2;
3879b50d902SRodney W. Grimes 			continue;
3889b50d902SRodney W. Grimes 		} else if (*p == '\\' && p[1] == '\\')
3899b50d902SRodney W. Grimes 			*d++ = *p++;
3909b50d902SRodney W. Grimes 		else if (*p == c) {
3919b50d902SRodney W. Grimes 			*d = '\0';
3929b50d902SRodney W. Grimes 			return (p + 1);
3939b50d902SRodney W. Grimes 		}
3949b50d902SRodney W. Grimes 		*d++ = *p++;
3959b50d902SRodney W. Grimes 	}
3969b50d902SRodney W. Grimes 	return (NULL);
3979b50d902SRodney W. Grimes }
3989b50d902SRodney W. Grimes 
399ce19262dSJordan K. Hubbard 
400ce19262dSJordan K. Hubbard /* compile_ccl: expand a POSIX character class */
401ce19262dSJordan K. Hubbard static char *
402ce19262dSJordan K. Hubbard compile_ccl(sp, t)
403ce19262dSJordan K. Hubbard 	char **sp;
404ce19262dSJordan K. Hubbard 	char *t;
405ce19262dSJordan K. Hubbard {
406ce19262dSJordan K. Hubbard 	int c, d;
407ce19262dSJordan K. Hubbard 	char *s = *sp;
408ce19262dSJordan K. Hubbard 
409ce19262dSJordan K. Hubbard 	*t++ = *s++;
410ce19262dSJordan K. Hubbard 	if (*s == '^')
411ce19262dSJordan K. Hubbard 		*t++ = *s++;
412ce19262dSJordan K. Hubbard 	if (*s == ']')
413ce19262dSJordan K. Hubbard 		*t++ = *s++;
414ce19262dSJordan K. Hubbard 	for (; *s && (*t = *s) != ']'; s++, t++)
415ce19262dSJordan K. Hubbard 		if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
416ce19262dSJordan K. Hubbard 			*++t = *++s, t++, s++;
417ce19262dSJordan K. Hubbard 			for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
418ce19262dSJordan K. Hubbard 				if ((c = *s) == '\0')
419ce19262dSJordan K. Hubbard 					return NULL;
420ce19262dSJordan K. Hubbard 		} else if (*s == '\\' && s[1] == 'n')
421ce19262dSJordan K. Hubbard 			    *t = '\n', s++;
422ce19262dSJordan K. Hubbard 	return (*s == ']') ? *sp = ++s, ++t : NULL;
423ce19262dSJordan K. Hubbard }
424ce19262dSJordan K. Hubbard 
4259b50d902SRodney W. Grimes /*
4269b50d902SRodney W. Grimes  * Get a regular expression.  P points to the delimiter of the regular
4279b50d902SRodney W. Grimes  * expression; repp points to the address of a regexp pointer.  Newline
4289b50d902SRodney W. Grimes  * and delimiter escapes are processed; other escapes are ignored.
4299b50d902SRodney W. Grimes  * Returns a pointer to the first character after the final delimiter
4309b50d902SRodney W. Grimes  * or NULL in the case of a non terminated regular expression.  The regexp
4319b50d902SRodney W. Grimes  * pointer is set to the compiled regular expression.
4329b50d902SRodney W. Grimes  * Cflags are passed to regcomp.
4339b50d902SRodney W. Grimes  */
4349b50d902SRodney W. Grimes static char *
4359b50d902SRodney W. Grimes compile_re(p, repp)
4369b50d902SRodney W. Grimes 	char *p;
4379b50d902SRodney W. Grimes 	regex_t **repp;
4389b50d902SRodney W. Grimes {
4399b50d902SRodney W. Grimes 	int eval;
4409b50d902SRodney W. Grimes 	char re[_POSIX2_LINE_MAX + 1];
4419b50d902SRodney W. Grimes 
4429b50d902SRodney W. Grimes 	p = compile_delimited(p, re);
4439b50d902SRodney W. Grimes 	if (p && strlen(re) == 0) {
4449b50d902SRodney W. Grimes 		*repp = NULL;
4459b50d902SRodney W. Grimes 		return (p);
4469b50d902SRodney W. Grimes 	}
4478e33c0a0SDavid E. O'Brien 	if ((*repp = malloc(sizeof(regex_t))) == NULL)
4488e33c0a0SDavid E. O'Brien 		err(1, "malloc");
449175de1e6SBrian Feldman 	if (p && (eval = regcomp(*repp, re, rflags)) != 0)
45073a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: RE error: %s",
45173a08bb2SPhilippe Charnier 				linenum, fname, strregerror(eval, *repp));
4529b50d902SRodney W. Grimes 	if (maxnsub < (*repp)->re_nsub)
4539b50d902SRodney W. Grimes 		maxnsub = (*repp)->re_nsub;
4549b50d902SRodney W. Grimes 	return (p);
4559b50d902SRodney W. Grimes }
4569b50d902SRodney W. Grimes 
4579b50d902SRodney W. Grimes /*
4589b50d902SRodney W. Grimes  * Compile the substitution string of a regular expression and set res to
4599b50d902SRodney W. Grimes  * point to a saved copy of it.  Nsub is the number of parenthesized regular
4609b50d902SRodney W. Grimes  * expressions.
4619b50d902SRodney W. Grimes  */
4629b50d902SRodney W. Grimes static char *
4639b50d902SRodney W. Grimes compile_subst(p, s)
4649b50d902SRodney W. Grimes 	char *p;
4659b50d902SRodney W. Grimes 	struct s_subst *s;
4669b50d902SRodney W. Grimes {
4679b50d902SRodney W. Grimes 	static char lbuf[_POSIX2_LINE_MAX + 1];
4689b50d902SRodney W. Grimes 	int asize, ref, size;
4699b50d902SRodney W. Grimes 	char c, *text, *op, *sp;
470c56690efSArchie Cobbs 	int more = 0;
4719b50d902SRodney W. Grimes 
4729b50d902SRodney W. Grimes 	c = *p++;			/* Terminator character */
4739b50d902SRodney W. Grimes 	if (c == '\0')
4749b50d902SRodney W. Grimes 		return (NULL);
4759b50d902SRodney W. Grimes 
4769b50d902SRodney W. Grimes 	s->maxbref = 0;
4779b50d902SRodney W. Grimes 	s->linenum = linenum;
4789b50d902SRodney W. Grimes 	asize = 2 * _POSIX2_LINE_MAX + 1;
4798e33c0a0SDavid E. O'Brien 	if ((text = malloc(asize)) == NULL)
4808e33c0a0SDavid E. O'Brien 		err(1, "malloc");
4819b50d902SRodney W. Grimes 	size = 0;
4829b50d902SRodney W. Grimes 	do {
4839b50d902SRodney W. Grimes 		op = sp = text + size;
4849b50d902SRodney W. Grimes 		for (; *p; p++) {
4859b50d902SRodney W. Grimes 			if (*p == '\\') {
4869b50d902SRodney W. Grimes 				p++;
4879b50d902SRodney W. Grimes 				if (strchr("123456789", *p) != NULL) {
4889b50d902SRodney W. Grimes 					*sp++ = '\\';
4899b50d902SRodney W. Grimes 					ref = *p - '0';
4909b50d902SRodney W. Grimes 					if (s->re != NULL &&
4919b50d902SRodney W. Grimes 					    ref > s->re->re_nsub)
49273a08bb2SPhilippe Charnier 						errx(1, "%lu: %s: \\%c not defined in the RE",
49373a08bb2SPhilippe Charnier 								linenum, fname, *p);
4949b50d902SRodney W. Grimes 					if (s->maxbref < ref)
4959b50d902SRodney W. Grimes 						s->maxbref = ref;
4969b50d902SRodney W. Grimes 				} else if (*p == '&' || *p == '\\')
4979b50d902SRodney W. Grimes 					*sp++ = '\\';
4989b50d902SRodney W. Grimes 			} else if (*p == c) {
499c56690efSArchie Cobbs 				if (*++p == '\0' && more) {
500c56690efSArchie Cobbs 					if (cu_fgets(lbuf, sizeof(lbuf), &more))
5011a6583daSArchie Cobbs 						p = lbuf;
5021a6583daSArchie Cobbs 				}
5039b50d902SRodney W. Grimes 				*sp++ = '\0';
5049b50d902SRodney W. Grimes 				size += sp - op;
5058e33c0a0SDavid E. O'Brien 				if ((s->new = realloc(text, size)) == NULL)
5068e33c0a0SDavid E. O'Brien 					err(1, "realloc");
5079b50d902SRodney W. Grimes 				return (p);
5089b50d902SRodney W. Grimes 			} else if (*p == '\n') {
50973a08bb2SPhilippe Charnier 				errx(1,
51073a08bb2SPhilippe Charnier "%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
5119b50d902SRodney W. Grimes 				/* NOTREACHED */
5129b50d902SRodney W. Grimes 			}
5139b50d902SRodney W. Grimes 			*sp++ = *p;
5149b50d902SRodney W. Grimes 		}
5159b50d902SRodney W. Grimes 		size += sp - op;
5169b50d902SRodney W. Grimes 		if (asize - size < _POSIX2_LINE_MAX + 1) {
5179b50d902SRodney W. Grimes 			asize *= 2;
5188e33c0a0SDavid E. O'Brien 			if ((text = realloc(text, asize)) == NULL)
5198e33c0a0SDavid E. O'Brien 				err(1, "realloc");
5209b50d902SRodney W. Grimes 		}
521c56690efSArchie Cobbs 	} while (cu_fgets(p = lbuf, sizeof(lbuf), &more));
52273a08bb2SPhilippe Charnier 	errx(1, "%lu: %s: unterminated substitute in regular expression",
52373a08bb2SPhilippe Charnier 			linenum, fname);
5249b50d902SRodney W. Grimes 	/* NOTREACHED */
5259b50d902SRodney W. Grimes }
5269b50d902SRodney W. Grimes 
5279b50d902SRodney W. Grimes /*
5289b50d902SRodney W. Grimes  * Compile the flags of the s command
5299b50d902SRodney W. Grimes  */
5309b50d902SRodney W. Grimes static char *
5319b50d902SRodney W. Grimes compile_flags(p, s)
5329b50d902SRodney W. Grimes 	char *p;
5339b50d902SRodney W. Grimes 	struct s_subst *s;
5349b50d902SRodney W. Grimes {
5359b50d902SRodney W. Grimes 	int gn;			/* True if we have seen g or n */
5369b50d902SRodney W. Grimes 	char wfile[_POSIX2_LINE_MAX + 1], *q;
5379b50d902SRodney W. Grimes 
5389b50d902SRodney W. Grimes 	s->n = 1;				/* Default */
5399b50d902SRodney W. Grimes 	s->p = 0;
5409b50d902SRodney W. Grimes 	s->wfile = NULL;
5419b50d902SRodney W. Grimes 	s->wfd = -1;
5429b50d902SRodney W. Grimes 	for (gn = 0;;) {
5439b50d902SRodney W. Grimes 		EATSPACE();			/* EXTENSION */
5449b50d902SRodney W. Grimes 		switch (*p) {
5459b50d902SRodney W. Grimes 		case 'g':
5469b50d902SRodney W. Grimes 			if (gn)
54773a08bb2SPhilippe Charnier 				errx(1,
54873a08bb2SPhilippe Charnier "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
5499b50d902SRodney W. Grimes 			gn = 1;
5509b50d902SRodney W. Grimes 			s->n = 0;
5519b50d902SRodney W. Grimes 			break;
5529b50d902SRodney W. Grimes 		case '\0':
5539b50d902SRodney W. Grimes 		case '\n':
5549b50d902SRodney W. Grimes 		case ';':
5559b50d902SRodney W. Grimes 			return (p);
5569b50d902SRodney W. Grimes 		case 'p':
5579b50d902SRodney W. Grimes 			s->p = 1;
5589b50d902SRodney W. Grimes 			break;
5599b50d902SRodney W. Grimes 		case '1': case '2': case '3':
5609b50d902SRodney W. Grimes 		case '4': case '5': case '6':
5619b50d902SRodney W. Grimes 		case '7': case '8': case '9':
5629b50d902SRodney W. Grimes 			if (gn)
56373a08bb2SPhilippe Charnier 				errx(1,
56473a08bb2SPhilippe Charnier "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
5659b50d902SRodney W. Grimes 			gn = 1;
5669b50d902SRodney W. Grimes 			/* XXX Check for overflow */
5679b50d902SRodney W. Grimes 			s->n = (int)strtol(p, &p, 10);
5689b50d902SRodney W. Grimes 			break;
5699b50d902SRodney W. Grimes 		case 'w':
5709b50d902SRodney W. Grimes 			p++;
5719b50d902SRodney W. Grimes #ifdef HISTORIC_PRACTICE
5729b50d902SRodney W. Grimes 			if (*p != ' ') {
57373a08bb2SPhilippe Charnier 				warnx("%lu: %s: space missing before w wfile", linenum, fname);
5749b50d902SRodney W. Grimes 				return (p);
5759b50d902SRodney W. Grimes 			}
5769b50d902SRodney W. Grimes #endif
5779b50d902SRodney W. Grimes 			EATSPACE();
5789b50d902SRodney W. Grimes 			q = wfile;
5799b50d902SRodney W. Grimes 			while (*p) {
5809b50d902SRodney W. Grimes 				if (*p == '\n')
5819b50d902SRodney W. Grimes 					break;
5829b50d902SRodney W. Grimes 				*q++ = *p++;
5839b50d902SRodney W. Grimes 			}
5849b50d902SRodney W. Grimes 			*q = '\0';
5859b50d902SRodney W. Grimes 			if (q == wfile)
58673a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: no wfile specified", linenum, fname);
5879b50d902SRodney W. Grimes 			s->wfile = strdup(wfile);
5889b50d902SRodney W. Grimes 			if (!aflag && (s->wfd = open(wfile,
5899b50d902SRodney W. Grimes 			    O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
5909b50d902SRodney W. Grimes 			    DEFFILEMODE)) == -1)
59173a08bb2SPhilippe Charnier 				err(1, "%s", wfile);
5929b50d902SRodney W. Grimes 			return (p);
5939b50d902SRodney W. Grimes 		default:
59473a08bb2SPhilippe Charnier 			errx(1, "%lu: %s: bad flag in substitute command: '%c'",
59573a08bb2SPhilippe Charnier 					linenum, fname, *p);
5969b50d902SRodney W. Grimes 			break;
5979b50d902SRodney W. Grimes 		}
5989b50d902SRodney W. Grimes 		p++;
5999b50d902SRodney W. Grimes 	}
6009b50d902SRodney W. Grimes }
6019b50d902SRodney W. Grimes 
6029b50d902SRodney W. Grimes /*
6039b50d902SRodney W. Grimes  * Compile a translation set of strings into a lookup table.
6049b50d902SRodney W. Grimes  */
6059b50d902SRodney W. Grimes static char *
6069b50d902SRodney W. Grimes compile_tr(p, transtab)
6079b50d902SRodney W. Grimes 	char *p;
6089b50d902SRodney W. Grimes 	char **transtab;
6099b50d902SRodney W. Grimes {
6109b50d902SRodney W. Grimes 	int i;
6119b50d902SRodney W. Grimes 	char *lt, *op, *np;
6129b50d902SRodney W. Grimes 	char old[_POSIX2_LINE_MAX + 1];
6139b50d902SRodney W. Grimes 	char new[_POSIX2_LINE_MAX + 1];
6149b50d902SRodney W. Grimes 
6159b50d902SRodney W. Grimes 	if (*p == '\0' || *p == '\\')
61673a08bb2SPhilippe Charnier 		errx(1,
61773a08bb2SPhilippe Charnier 	"%lu: %s: transform pattern can not be delimited by newline or backslash",
61873a08bb2SPhilippe Charnier 			linenum, fname);
6199b50d902SRodney W. Grimes 	p = compile_delimited(p, old);
62073a08bb2SPhilippe Charnier 	if (p == NULL)
62173a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: unterminated transform source string",
62273a08bb2SPhilippe Charnier 				linenum, fname);
6239b50d902SRodney W. Grimes 	p = compile_delimited(--p, new);
62473a08bb2SPhilippe Charnier 	if (p == NULL)
62573a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: unterminated transform target string",
62673a08bb2SPhilippe Charnier 				linenum, fname);
6279b50d902SRodney W. Grimes 	EATSPACE();
62873a08bb2SPhilippe Charnier 	if (strlen(new) != strlen(old))
62973a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: transform strings are not the same length",
63073a08bb2SPhilippe Charnier 				linenum, fname);
6319b50d902SRodney W. Grimes 	/* We assume characters are 8 bits */
6328e33c0a0SDavid E. O'Brien 	if ((lt = malloc(UCHAR_MAX)) == NULL)
6338e33c0a0SDavid E. O'Brien 		err(1, "malloc");
6349b50d902SRodney W. Grimes 	for (i = 0; i <= UCHAR_MAX; i++)
6359b50d902SRodney W. Grimes 		lt[i] = (char)i;
6369b50d902SRodney W. Grimes 	for (op = old, np = new; *op; op++, np++)
6379b50d902SRodney W. Grimes 		lt[(u_char)*op] = *np;
6389b50d902SRodney W. Grimes 	*transtab = lt;
6399b50d902SRodney W. Grimes 	return (p);
6409b50d902SRodney W. Grimes }
6419b50d902SRodney W. Grimes 
6429b50d902SRodney W. Grimes /*
6439b50d902SRodney W. Grimes  * Compile the text following an a or i command.
6449b50d902SRodney W. Grimes  */
6459b50d902SRodney W. Grimes static char *
6469b50d902SRodney W. Grimes compile_text()
6479b50d902SRodney W. Grimes {
64849e65599SBruce Evans 	int asize, esc_nl, size;
6499b50d902SRodney W. Grimes 	char *text, *p, *op, *s;
6509b50d902SRodney W. Grimes 	char lbuf[_POSIX2_LINE_MAX + 1];
6519b50d902SRodney W. Grimes 
6529b50d902SRodney W. Grimes 	asize = 2 * _POSIX2_LINE_MAX + 1;
6538e33c0a0SDavid E. O'Brien 	if ((text = malloc(asize)) == NULL)
6548e33c0a0SDavid E. O'Brien 		err(1, "malloc");
6559b50d902SRodney W. Grimes 	size = 0;
656c56690efSArchie Cobbs 	while (cu_fgets(lbuf, sizeof(lbuf), NULL)) {
6579b50d902SRodney W. Grimes 		op = s = text + size;
6589b50d902SRodney W. Grimes 		p = lbuf;
6599b50d902SRodney W. Grimes 		EATSPACE();
66049e65599SBruce Evans 		for (esc_nl = 0; *p != '\0'; p++) {
66149e65599SBruce Evans 			if (*p == '\\' && p[1] != '\0' && *++p == '\n')
66249e65599SBruce Evans 				esc_nl = 1;
6639b50d902SRodney W. Grimes 			*s++ = *p;
6649b50d902SRodney W. Grimes 		}
6659b50d902SRodney W. Grimes 		size += s - op;
66649e65599SBruce Evans 		if (!esc_nl) {
6679b50d902SRodney W. Grimes 			*s = '\0';
6689b50d902SRodney W. Grimes 			break;
6699b50d902SRodney W. Grimes 		}
6709b50d902SRodney W. Grimes 		if (asize - size < _POSIX2_LINE_MAX + 1) {
6719b50d902SRodney W. Grimes 			asize *= 2;
6728e33c0a0SDavid E. O'Brien 			if ((text = realloc(text, asize)) == NULL)
6738e33c0a0SDavid E. O'Brien 				err(1, "realloc");
6749b50d902SRodney W. Grimes 		}
6759b50d902SRodney W. Grimes 	}
67613ede3c0SBrian Somers 	text[size] = '\0';
6778e33c0a0SDavid E. O'Brien 	if ((p = realloc(text, size + 1)) == NULL)
6788e33c0a0SDavid E. O'Brien 		err(1, "realloc");
6798e33c0a0SDavid E. O'Brien 	return (p);
6809b50d902SRodney W. Grimes }
6819b50d902SRodney W. Grimes 
6829b50d902SRodney W. Grimes /*
6839b50d902SRodney W. Grimes  * Get an address and return a pointer to the first character after
6849b50d902SRodney W. Grimes  * it.  Fill the structure pointed to according to the address.
6859b50d902SRodney W. Grimes  */
6869b50d902SRodney W. Grimes static char *
6879b50d902SRodney W. Grimes compile_addr(p, a)
6889b50d902SRodney W. Grimes 	char *p;
6899b50d902SRodney W. Grimes 	struct s_addr *a;
6909b50d902SRodney W. Grimes {
6919b50d902SRodney W. Grimes 	char *end;
6929b50d902SRodney W. Grimes 
6939b50d902SRodney W. Grimes 	switch (*p) {
6949b50d902SRodney W. Grimes 	case '\\':				/* Context address */
6959b50d902SRodney W. Grimes 		++p;
6969b50d902SRodney W. Grimes 		/* FALLTHROUGH */
6979b50d902SRodney W. Grimes 	case '/':				/* Context address */
6989b50d902SRodney W. Grimes 		p = compile_re(p, &a->u.r);
6999b50d902SRodney W. Grimes 		if (p == NULL)
70073a08bb2SPhilippe Charnier 			errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
7019b50d902SRodney W. Grimes 		a->type = AT_RE;
7029b50d902SRodney W. Grimes 		return (p);
7039b50d902SRodney W. Grimes 
7049b50d902SRodney W. Grimes 	case '$':				/* Last line */
7059b50d902SRodney W. Grimes 		a->type = AT_LAST;
7069b50d902SRodney W. Grimes 		return (p + 1);
7079b50d902SRodney W. Grimes 						/* Line number */
7089b50d902SRodney W. Grimes 	case '0': case '1': case '2': case '3': case '4':
7099b50d902SRodney W. Grimes 	case '5': case '6': case '7': case '8': case '9':
7109b50d902SRodney W. Grimes 		a->type = AT_LINE;
7119b50d902SRodney W. Grimes 		a->u.l = strtol(p, &end, 10);
7129b50d902SRodney W. Grimes 		return (end);
7139b50d902SRodney W. Grimes 	default:
71473a08bb2SPhilippe Charnier 		errx(1, "%lu: %s: expected context address", linenum, fname);
7159b50d902SRodney W. Grimes 		return (NULL);
7169b50d902SRodney W. Grimes 	}
7179b50d902SRodney W. Grimes }
7189b50d902SRodney W. Grimes 
7199b50d902SRodney W. Grimes /*
7209b50d902SRodney W. Grimes  * duptoeol --
7219b50d902SRodney W. Grimes  *	Return a copy of all the characters up to \n or \0.
7229b50d902SRodney W. Grimes  */
7239b50d902SRodney W. Grimes static char *
7249b50d902SRodney W. Grimes duptoeol(s, ctype)
7259b50d902SRodney W. Grimes 	register char *s;
7269b50d902SRodney W. Grimes 	char *ctype;
7279b50d902SRodney W. Grimes {
7289b50d902SRodney W. Grimes 	size_t len;
7299b50d902SRodney W. Grimes 	int ws;
7308e33c0a0SDavid E. O'Brien 	char *p, *start;
7319b50d902SRodney W. Grimes 
7329b50d902SRodney W. Grimes 	ws = 0;
7339b50d902SRodney W. Grimes 	for (start = s; *s != '\0' && *s != '\n'; ++s)
734726aebe5SAndrey A. Chernov 		ws = isspace((unsigned char)*s);
7359b50d902SRodney W. Grimes 	*s = '\0';
7369b50d902SRodney W. Grimes 	if (ws)
73773a08bb2SPhilippe Charnier 		warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
7389b50d902SRodney W. Grimes 	len = s - start + 1;
7398e33c0a0SDavid E. O'Brien 	if ((p = malloc(len)) == NULL)
7408e33c0a0SDavid E. O'Brien 		err(1, "malloc");
7418e33c0a0SDavid E. O'Brien 	return (memmove(p, start, len));
7429b50d902SRodney W. Grimes }
7439b50d902SRodney W. Grimes 
7449b50d902SRodney W. Grimes /*
7459b50d902SRodney W. Grimes  * Convert goto label names to addresses, and count a and r commands, in
7469b50d902SRodney W. Grimes  * the given subset of the script.  Free the memory used by labels in b
7479b50d902SRodney W. Grimes  * and t commands (but not by :).
7489b50d902SRodney W. Grimes  *
7499b50d902SRodney W. Grimes  * TODO: Remove } nodes
7509b50d902SRodney W. Grimes  */
7519b50d902SRodney W. Grimes static void
7529b50d902SRodney W. Grimes fixuplabel(cp, end)
7539b50d902SRodney W. Grimes 	struct s_command *cp, *end;
7549b50d902SRodney W. Grimes {
7559b50d902SRodney W. Grimes 
7569b50d902SRodney W. Grimes 	for (; cp != end; cp = cp->next)
7579b50d902SRodney W. Grimes 		switch (cp->code) {
7589b50d902SRodney W. Grimes 		case 'a':
7599b50d902SRodney W. Grimes 		case 'r':
7609b50d902SRodney W. Grimes 			appendnum++;
7619b50d902SRodney W. Grimes 			break;
7629b50d902SRodney W. Grimes 		case 'b':
7639b50d902SRodney W. Grimes 		case 't':
7649b50d902SRodney W. Grimes 			/* Resolve branch target. */
7659b50d902SRodney W. Grimes 			if (cp->t == NULL) {
7669b50d902SRodney W. Grimes 				cp->u.c = NULL;
7679b50d902SRodney W. Grimes 				break;
7689b50d902SRodney W. Grimes 			}
7699b50d902SRodney W. Grimes 			if ((cp->u.c = findlabel(cp->t)) == NULL)
77073a08bb2SPhilippe Charnier 				errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
7719b50d902SRodney W. Grimes 			free(cp->t);
7729b50d902SRodney W. Grimes 			break;
7739b50d902SRodney W. Grimes 		case '{':
7749b50d902SRodney W. Grimes 			/* Do interior commands. */
7759b50d902SRodney W. Grimes 			fixuplabel(cp->u.c, cp->next);
7769b50d902SRodney W. Grimes 			break;
7779b50d902SRodney W. Grimes 		}
7789b50d902SRodney W. Grimes }
7799b50d902SRodney W. Grimes 
7809b50d902SRodney W. Grimes /*
7819b50d902SRodney W. Grimes  * Associate the given command label for later lookup.
7829b50d902SRodney W. Grimes  */
7839b50d902SRodney W. Grimes static void
7849b50d902SRodney W. Grimes enterlabel(cp)
7859b50d902SRodney W. Grimes 	struct s_command *cp;
7869b50d902SRodney W. Grimes {
7879b50d902SRodney W. Grimes 	register struct labhash **lhp, *lh;
7889b50d902SRodney W. Grimes 	register u_char *p;
7899b50d902SRodney W. Grimes 	register u_int h, c;
7909b50d902SRodney W. Grimes 
7919b50d902SRodney W. Grimes 	for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
7929b50d902SRodney W. Grimes 		h = (h << 5) + h + c;
7939b50d902SRodney W. Grimes 	lhp = &labels[h & LHMASK];
7949b50d902SRodney W. Grimes 	for (lh = *lhp; lh != NULL; lh = lh->lh_next)
7959b50d902SRodney W. Grimes 		if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
79673a08bb2SPhilippe Charnier 			errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
7978e33c0a0SDavid E. O'Brien 	if ((lh = malloc(sizeof *lh)) == NULL)
7988e33c0a0SDavid E. O'Brien 		err(1, "malloc");
7999b50d902SRodney W. Grimes 	lh->lh_next = *lhp;
8009b50d902SRodney W. Grimes 	lh->lh_hash = h;
8019b50d902SRodney W. Grimes 	lh->lh_cmd = cp;
8029b50d902SRodney W. Grimes 	lh->lh_ref = 0;
8039b50d902SRodney W. Grimes 	*lhp = lh;
8049b50d902SRodney W. Grimes }
8059b50d902SRodney W. Grimes 
8069b50d902SRodney W. Grimes /*
8079b50d902SRodney W. Grimes  * Find the label contained in the command l in the command linked
8089b50d902SRodney W. Grimes  * list cp.  L is excluded from the search.  Return NULL if not found.
8099b50d902SRodney W. Grimes  */
8109b50d902SRodney W. Grimes static struct s_command *
8119b50d902SRodney W. Grimes findlabel(name)
8129b50d902SRodney W. Grimes 	char *name;
8139b50d902SRodney W. Grimes {
8149b50d902SRodney W. Grimes 	register struct labhash *lh;
8159b50d902SRodney W. Grimes 	register u_char *p;
8169b50d902SRodney W. Grimes 	register u_int h, c;
8179b50d902SRodney W. Grimes 
8189b50d902SRodney W. Grimes 	for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
8199b50d902SRodney W. Grimes 		h = (h << 5) + h + c;
8209b50d902SRodney W. Grimes 	for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
8219b50d902SRodney W. Grimes 		if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
8229b50d902SRodney W. Grimes 			lh->lh_ref = 1;
8239b50d902SRodney W. Grimes 			return (lh->lh_cmd);
8249b50d902SRodney W. Grimes 		}
8259b50d902SRodney W. Grimes 	}
8269b50d902SRodney W. Grimes 	return (NULL);
8279b50d902SRodney W. Grimes }
8289b50d902SRodney W. Grimes 
8299b50d902SRodney W. Grimes /*
8309b50d902SRodney W. Grimes  * Warn about any unused labels.  As a side effect, release the label hash
8319b50d902SRodney W. Grimes  * table space.
8329b50d902SRodney W. Grimes  */
8339b50d902SRodney W. Grimes static void
8349b50d902SRodney W. Grimes uselabel()
8359b50d902SRodney W. Grimes {
8369b50d902SRodney W. Grimes 	register struct labhash *lh, *next;
8379b50d902SRodney W. Grimes 	register int i;
8389b50d902SRodney W. Grimes 
8399b50d902SRodney W. Grimes 	for (i = 0; i < LHSZ; i++) {
8409b50d902SRodney W. Grimes 		for (lh = labels[i]; lh != NULL; lh = next) {
8419b50d902SRodney W. Grimes 			next = lh->lh_next;
8429b50d902SRodney W. Grimes 			if (!lh->lh_ref)
84373a08bb2SPhilippe Charnier 				warnx("%lu: %s: unused label '%s'",
84473a08bb2SPhilippe Charnier 				    linenum, fname, lh->lh_cmd->t);
8459b50d902SRodney W. Grimes 			free(lh);
8469b50d902SRodney W. Grimes 		}
8479b50d902SRodney W. Grimes 	}
8489b50d902SRodney W. Grimes }
849