xref: /illumos-gate/usr/src/cmd/awk_xpg4/awk2.c (revision 2a8bcb4e)
17c478bd9Sstevel@tonic-gate /*
27c478bd9Sstevel@tonic-gate  * CDDL HEADER START
37c478bd9Sstevel@tonic-gate  *
47c478bd9Sstevel@tonic-gate  * The contents of this file are subject to the terms of the
5*79777a7dSnakanon  * Common Development and Distribution License (the "License").
6*79777a7dSnakanon  * You may not use this file except in compliance with the License.
77c478bd9Sstevel@tonic-gate  *
87c478bd9Sstevel@tonic-gate  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
97c478bd9Sstevel@tonic-gate  * or http://www.opensolaris.org/os/licensing.
107c478bd9Sstevel@tonic-gate  * See the License for the specific language governing permissions
117c478bd9Sstevel@tonic-gate  * and limitations under the License.
127c478bd9Sstevel@tonic-gate  *
137c478bd9Sstevel@tonic-gate  * When distributing Covered Code, include this CDDL HEADER in each
147c478bd9Sstevel@tonic-gate  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
157c478bd9Sstevel@tonic-gate  * If applicable, add the following below this CDDL HEADER, with the
167c478bd9Sstevel@tonic-gate  * fields enclosed by brackets "[]" replaced with your own identifying
177c478bd9Sstevel@tonic-gate  * information: Portions Copyright [yyyy] [name of copyright owner]
187c478bd9Sstevel@tonic-gate  *
197c478bd9Sstevel@tonic-gate  * CDDL HEADER END
207c478bd9Sstevel@tonic-gate  */
217c478bd9Sstevel@tonic-gate /*
22*79777a7dSnakanon  * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
23cb4658fbSceastha  * Use is subject to license terms.
24cb4658fbSceastha  */
25cb4658fbSceastha 
26cb4658fbSceastha /*
277c478bd9Sstevel@tonic-gate  * Copyright 1986, 1994 by Mortice Kern Systems Inc.  All rights reserved.
287c478bd9Sstevel@tonic-gate  */
297c478bd9Sstevel@tonic-gate 
30cb4658fbSceastha /*
31cb4658fbSceastha  * awk -- process input files, field extraction, output
32cb4658fbSceastha  *
33cb4658fbSceastha  * Based on MKS awk(1) ported to be /usr/xpg4/bin/awk with POSIX/XCU4 changes
34cb4658fbSceastha  */
35cb4658fbSceastha 
367c478bd9Sstevel@tonic-gate #include "awk.h"
377c478bd9Sstevel@tonic-gate #include "y.tab.h"
387c478bd9Sstevel@tonic-gate 
397c478bd9Sstevel@tonic-gate static FILE	*awkinfp;		/* Input file pointer */
407c478bd9Sstevel@tonic-gate static int	reclen;			/* Length of last record */
417c478bd9Sstevel@tonic-gate static int	exstat;			/* Exit status */
427c478bd9Sstevel@tonic-gate 
437c478bd9Sstevel@tonic-gate static FILE	*openfile(NODE *np, int flag, int fatal);
447c478bd9Sstevel@tonic-gate static FILE	*newfile(void);
457c478bd9Sstevel@tonic-gate static NODE	*nextarg(NODE **npp);
467c478bd9Sstevel@tonic-gate static void	adjust_buf(wchar_t **, int *, wchar_t **, char *, size_t);
477c478bd9Sstevel@tonic-gate static void	awk_putwc(wchar_t, FILE *);
487c478bd9Sstevel@tonic-gate 
497c478bd9Sstevel@tonic-gate /*
507c478bd9Sstevel@tonic-gate  * mainline for awk execution
517c478bd9Sstevel@tonic-gate  */
527c478bd9Sstevel@tonic-gate void
awk()537c478bd9Sstevel@tonic-gate awk()
547c478bd9Sstevel@tonic-gate {
557c478bd9Sstevel@tonic-gate 	running = 1;
567c478bd9Sstevel@tonic-gate 	dobegin();
577c478bd9Sstevel@tonic-gate 	while (nextrecord(linebuf, awkinfp) > 0)
587c478bd9Sstevel@tonic-gate 		execute(yytree);
597c478bd9Sstevel@tonic-gate 	doend(exstat);
607c478bd9Sstevel@tonic-gate }
617c478bd9Sstevel@tonic-gate 
627c478bd9Sstevel@tonic-gate /*
637c478bd9Sstevel@tonic-gate  * "cp" is the buffer to fill.  There is a special case if this buffer is
647c478bd9Sstevel@tonic-gate  * "linebuf" ($0)
657c478bd9Sstevel@tonic-gate  * Return 1 if OK, zero on EOF, -1 on error.
667c478bd9Sstevel@tonic-gate  */
677c478bd9Sstevel@tonic-gate int
nextrecord(wchar_t * cp,FILE * fp)687c478bd9Sstevel@tonic-gate nextrecord(wchar_t *cp, FILE *fp)
697c478bd9Sstevel@tonic-gate {
707c478bd9Sstevel@tonic-gate 	wchar_t *ep = cp;
717c478bd9Sstevel@tonic-gate 
727c478bd9Sstevel@tonic-gate nextfile:
737c478bd9Sstevel@tonic-gate 	if (fp == FNULL && (fp = newfile()) == FNULL)
747c478bd9Sstevel@tonic-gate 		return (0);
757c478bd9Sstevel@tonic-gate 	if ((*awkrecord)(ep, NLINE, fp) == NULL) {
767c478bd9Sstevel@tonic-gate 		if (fp == awkinfp) {
777c478bd9Sstevel@tonic-gate 			if (fp != stdin)
787c478bd9Sstevel@tonic-gate 				(void) fclose(awkinfp);
797c478bd9Sstevel@tonic-gate 			awkinfp = fp = FNULL;
807c478bd9Sstevel@tonic-gate 			goto nextfile;
817c478bd9Sstevel@tonic-gate 		}
827c478bd9Sstevel@tonic-gate 		if (ferror(fp))
837c478bd9Sstevel@tonic-gate 			return (-1);
847c478bd9Sstevel@tonic-gate 		return (0);
857c478bd9Sstevel@tonic-gate 	}
867c478bd9Sstevel@tonic-gate 	if (fp == awkinfp) {
877c478bd9Sstevel@tonic-gate 		if (varNR->n_flags & FINT)
88cb4658fbSceastha 			++varNR->n_int;
89cb4658fbSceastha 		else
907c478bd9Sstevel@tonic-gate 			(void) exprreduce(incNR);
917c478bd9Sstevel@tonic-gate 		if (varFNR->n_flags & FINT)
92cb4658fbSceastha 			++varFNR->n_int;
93cb4658fbSceastha 		else
947c478bd9Sstevel@tonic-gate 			(void) exprreduce(incFNR);
957c478bd9Sstevel@tonic-gate 	}
967c478bd9Sstevel@tonic-gate 	if (cp == linebuf) {
977c478bd9Sstevel@tonic-gate 		lbuflen = reclen;
987c478bd9Sstevel@tonic-gate 		splitdone = 0;
997c478bd9Sstevel@tonic-gate 		if (needsplit)
1007c478bd9Sstevel@tonic-gate 			fieldsplit();
1017c478bd9Sstevel@tonic-gate 	}
1027c478bd9Sstevel@tonic-gate 	/* if record length is too long then bail out */
1037c478bd9Sstevel@tonic-gate 	if (reclen > NLINE - 2) {
1047c478bd9Sstevel@tonic-gate 		awkerr(gettext("Record too long (LIMIT: %d bytes)"),
1057c478bd9Sstevel@tonic-gate 		    NLINE - 1);
1067c478bd9Sstevel@tonic-gate 		/* Not Reached */
1077c478bd9Sstevel@tonic-gate 	}
1087c478bd9Sstevel@tonic-gate 	return (1);
1097c478bd9Sstevel@tonic-gate }
1107c478bd9Sstevel@tonic-gate 
1117c478bd9Sstevel@tonic-gate /*
112cb4658fbSceastha  * isclvar()
113cb4658fbSceastha  *
114cb4658fbSceastha  * Returns 1 if the input string, arg, is a variable assignment,
115cb4658fbSceastha  * otherwise returns 0.
116cb4658fbSceastha  *
117cb4658fbSceastha  * An argument to awk can be either a pathname of a file, or a variable
118cb4658fbSceastha  * assignment.  An operand that begins with an undersore or alphabetic
119cb4658fbSceastha  * character from the portable character set, followed by a sequence of
120cb4658fbSceastha  * underscores, digits, and alphabetics from the portable character set,
121cb4658fbSceastha  * followed by the '=' character, shall specify a variable assignment
122cb4658fbSceastha  * rather than a pathname.
123cb4658fbSceastha  */
124cb4658fbSceastha int
isclvar(wchar_t * arg)125cb4658fbSceastha isclvar(wchar_t *arg)
126cb4658fbSceastha {
127cb4658fbSceastha 	wchar_t	*tmpptr = arg;
128cb4658fbSceastha 
129cb4658fbSceastha 	if (tmpptr != NULL) {
130cb4658fbSceastha 
131cb4658fbSceastha 		/* Begins with an underscore or alphabetic character */
132cb4658fbSceastha 		if (iswalpha(*tmpptr) || *tmpptr == '_') {
133cb4658fbSceastha 
134cb4658fbSceastha 			/*
135cb4658fbSceastha 			 * followed by a sequence of underscores, digits,
136cb4658fbSceastha 			 * and alphabetics
137cb4658fbSceastha 			 */
138cb4658fbSceastha 			for (tmpptr++; *tmpptr; tmpptr++) {
139baaf2753Sceastha 				if (!(iswalnum(*tmpptr) || (*tmpptr == '_'))) {
140cb4658fbSceastha 					break;
141cb4658fbSceastha 				}
142cb4658fbSceastha 			}
143cb4658fbSceastha 			return (*tmpptr == '=');
144cb4658fbSceastha 		}
145cb4658fbSceastha 	}
146cb4658fbSceastha 
147cb4658fbSceastha 	return (0);
148cb4658fbSceastha }
149cb4658fbSceastha 
150cb4658fbSceastha /*
1517c478bd9Sstevel@tonic-gate  * Return the next file from the command line.
1527c478bd9Sstevel@tonic-gate  * Return FNULL when no more files.
1537c478bd9Sstevel@tonic-gate  * Sets awkinfp variable to the new current input file.
1547c478bd9Sstevel@tonic-gate  */
1557c478bd9Sstevel@tonic-gate static FILE *
newfile()1567c478bd9Sstevel@tonic-gate newfile()
1577c478bd9Sstevel@tonic-gate {
1587c478bd9Sstevel@tonic-gate 	static int argindex = 1;
1597c478bd9Sstevel@tonic-gate 	static int filedone;
160cb4658fbSceastha 	wchar_t *ap;
161cb4658fbSceastha 	int argc;
1627c478bd9Sstevel@tonic-gate 	wchar_t *arg;
1637c478bd9Sstevel@tonic-gate 	extern void strescape(wchar_t *);
1647c478bd9Sstevel@tonic-gate 
1657c478bd9Sstevel@tonic-gate 	argc = (int)exprint(varARGC);
1667c478bd9Sstevel@tonic-gate 	for (;;) {
1677c478bd9Sstevel@tonic-gate 		if (argindex >= argc) {
1687c478bd9Sstevel@tonic-gate 			if (filedone)
1697c478bd9Sstevel@tonic-gate 				return (FNULL);
1707c478bd9Sstevel@tonic-gate 			++filedone;
1717c478bd9Sstevel@tonic-gate 			awkinfp = stdin;
1727c478bd9Sstevel@tonic-gate 			arg = M_MB_L("-");
1737c478bd9Sstevel@tonic-gate 			break;
1747c478bd9Sstevel@tonic-gate 		}
1757c478bd9Sstevel@tonic-gate 		constant->n_int = argindex++;
1767c478bd9Sstevel@tonic-gate 		arg = (wchar_t *)exprstring(ARGVsubi);
177cb4658fbSceastha 		/*
178cb4658fbSceastha 		 * If the argument contains a '=', determine if the
179cb4658fbSceastha 		 * argument needs to be treated as a variable assignment
180cb4658fbSceastha 		 * or as the pathname of a file.
181cb4658fbSceastha 		 */
182cb4658fbSceastha 		if (((ap = wcschr(arg, '=')) != NULL) && isclvar(arg)) {
1837c478bd9Sstevel@tonic-gate 			*ap = '\0';
1847c478bd9Sstevel@tonic-gate 			strescape(ap+1);
1857c478bd9Sstevel@tonic-gate 			strassign(vlook(arg), linebuf, FALLOC|FSENSE,
1867c478bd9Sstevel@tonic-gate 			    wcslen(linebuf));
1877c478bd9Sstevel@tonic-gate 			*ap = '=';
1887c478bd9Sstevel@tonic-gate 			continue;
1897c478bd9Sstevel@tonic-gate 		}
1907c478bd9Sstevel@tonic-gate 		if (arg[0] == '\0')
1917c478bd9Sstevel@tonic-gate 			continue;
1927c478bd9Sstevel@tonic-gate 		++filedone;
1937c478bd9Sstevel@tonic-gate 		if (arg[0] == '-' && arg[1] == '\0') {
1947c478bd9Sstevel@tonic-gate 			awkinfp = stdin;
1957c478bd9Sstevel@tonic-gate 			break;
1967c478bd9Sstevel@tonic-gate 		}
1977c478bd9Sstevel@tonic-gate 		if ((awkinfp = fopen(mbunconvert(arg), r)) == FNULL) {
1987c478bd9Sstevel@tonic-gate 			(void) fprintf(stderr, gettext("input file \"%s\""),
1997c478bd9Sstevel@tonic-gate 			    mbunconvert(arg));
2007c478bd9Sstevel@tonic-gate 			exstat = 1;
2017c478bd9Sstevel@tonic-gate 			continue;
2027c478bd9Sstevel@tonic-gate 		}
2037c478bd9Sstevel@tonic-gate 		break;
2047c478bd9Sstevel@tonic-gate 	}
2057c478bd9Sstevel@tonic-gate 	strassign(varFILENAME, arg, FALLOC, wcslen(arg));
2067c478bd9Sstevel@tonic-gate 	if (varFNR->n_flags & FINT)
207cb4658fbSceastha 		varFNR->n_int = 0;
208cb4658fbSceastha 	else
2097c478bd9Sstevel@tonic-gate 		(void) exprreduce(clrFNR);
2107c478bd9Sstevel@tonic-gate 	return (awkinfp);
2117c478bd9Sstevel@tonic-gate }
2127c478bd9Sstevel@tonic-gate 
2137c478bd9Sstevel@tonic-gate /*
2147c478bd9Sstevel@tonic-gate  * Default record reading code
2157c478bd9Sstevel@tonic-gate  * Uses fgets for potential speedups found in some (e.g. MKS)
2167c478bd9Sstevel@tonic-gate  * stdio packages.
2177c478bd9Sstevel@tonic-gate  */
2187c478bd9Sstevel@tonic-gate wchar_t *
defrecord(wchar_t * bp,int lim,FILE * fp)2197c478bd9Sstevel@tonic-gate defrecord(wchar_t *bp, int lim, FILE *fp)
2207c478bd9Sstevel@tonic-gate {
221cb4658fbSceastha 	wchar_t *endp;
2227c478bd9Sstevel@tonic-gate 
2237c478bd9Sstevel@tonic-gate 	if (fgetws(bp, lim, fp) == NULL) {
2247c478bd9Sstevel@tonic-gate 		*bp = '\0';
2257c478bd9Sstevel@tonic-gate 		return (NULL);
2267c478bd9Sstevel@tonic-gate 	}
227cb4658fbSceastha /*
228cb4658fbSceastha  * XXXX
229cb4658fbSceastha  *	switch (fgetws(bp, lim, fp)) {
230cb4658fbSceastha  *	case M_FGETS_EOF:
231cb4658fbSceastha  *		*bp = '\0';
232cb4658fbSceastha  *		return (NULL);
233cb4658fbSceastha  *	case M_FGETS_BINARY:
234cb4658fbSceastha  *		awkerr(gettext("file is binary"));
235cb4658fbSceastha  *	case M_FGETS_LONG:
236cb4658fbSceastha  *		awkerr(gettext("line too long: limit %d"),
237cb4658fbSceastha  *			lim);
238cb4658fbSceastha  *	case M_FGETS_ERROR:
239cb4658fbSceastha  *		awkperr(gettext("error reading file"));
240cb4658fbSceastha  *	}
2417c478bd9Sstevel@tonic-gate  */
2427c478bd9Sstevel@tonic-gate 
2437c478bd9Sstevel@tonic-gate 	if (*(endp = (bp + (reclen = wcslen(bp))-1)) == '\n') {
2447c478bd9Sstevel@tonic-gate 		*endp = '\0';
2457c478bd9Sstevel@tonic-gate 		reclen--;
2467c478bd9Sstevel@tonic-gate 	}
2477c478bd9Sstevel@tonic-gate 	return (bp);
2487c478bd9Sstevel@tonic-gate }
2497c478bd9Sstevel@tonic-gate 
2507c478bd9Sstevel@tonic-gate /*
2517c478bd9Sstevel@tonic-gate  * Read a record separated by one character in the RS.
2527c478bd9Sstevel@tonic-gate  * Compatible calling sequence with fgets, but don't include
2537c478bd9Sstevel@tonic-gate  * record separator character in string.
2547c478bd9Sstevel@tonic-gate  */
2557c478bd9Sstevel@tonic-gate wchar_t *
charrecord(wchar_t * abp,int alim,FILE * fp)2567c478bd9Sstevel@tonic-gate charrecord(wchar_t *abp, int alim, FILE *fp)
2577c478bd9Sstevel@tonic-gate {
258cb4658fbSceastha 	wchar_t *bp;
259cb4658fbSceastha 	wint_t c;
260cb4658fbSceastha 	int limit = alim;
261cb4658fbSceastha 	wint_t endc;
2627c478bd9Sstevel@tonic-gate 
2637c478bd9Sstevel@tonic-gate 	bp = abp;
2647c478bd9Sstevel@tonic-gate 	endc = *(wchar_t *)varRS->n_string;
2657c478bd9Sstevel@tonic-gate 	while (--limit > 0 && (c = getwc(fp)) != endc && c != WEOF)
2667c478bd9Sstevel@tonic-gate 		*bp++ = c;
2677c478bd9Sstevel@tonic-gate 	*bp = '\0';
2687c478bd9Sstevel@tonic-gate 	reclen = bp-abp;
2697c478bd9Sstevel@tonic-gate 	return (c == WEOF && bp == abp ? NULL : abp);
2707c478bd9Sstevel@tonic-gate }
2717c478bd9Sstevel@tonic-gate 
2727c478bd9Sstevel@tonic-gate /*
2737c478bd9Sstevel@tonic-gate  * Special routine for multiple line records.
2747c478bd9Sstevel@tonic-gate  */
2757c478bd9Sstevel@tonic-gate wchar_t *
multirecord(wchar_t * abp,int limit,FILE * fp)2767c478bd9Sstevel@tonic-gate multirecord(wchar_t *abp, int limit, FILE *fp)
2777c478bd9Sstevel@tonic-gate {
278cb4658fbSceastha 	wchar_t *bp;
279cb4658fbSceastha 	int c;
2807c478bd9Sstevel@tonic-gate 
2817c478bd9Sstevel@tonic-gate 	while ((c = getwc(fp)) == '\n')
2827c478bd9Sstevel@tonic-gate 		;
2837c478bd9Sstevel@tonic-gate 	bp = abp;
2847c478bd9Sstevel@tonic-gate 	if (c != WEOF) do {
2857c478bd9Sstevel@tonic-gate 		if (--limit == 0)
2867c478bd9Sstevel@tonic-gate 			break;
2877c478bd9Sstevel@tonic-gate 		if (c == '\n' && bp[-1] == '\n')
2887c478bd9Sstevel@tonic-gate 			break;
2897c478bd9Sstevel@tonic-gate 
2907c478bd9Sstevel@tonic-gate 		*bp++ = c;
2917c478bd9Sstevel@tonic-gate 	} while ((c = getwc(fp)) != WEOF);
2927c478bd9Sstevel@tonic-gate 	*bp = '\0';
2937c478bd9Sstevel@tonic-gate 	if (bp > abp)
2947c478bd9Sstevel@tonic-gate 		*--bp = '\0';
2957c478bd9Sstevel@tonic-gate 	reclen = bp-abp;
2967c478bd9Sstevel@tonic-gate 	return (c == WEOF && bp == abp ? NULL : abp);
2977c478bd9Sstevel@tonic-gate }
2987c478bd9Sstevel@tonic-gate 
2997c478bd9Sstevel@tonic-gate /*
3007c478bd9Sstevel@tonic-gate  * Look for fields separated by spaces, tabs or newlines.
3017c478bd9Sstevel@tonic-gate  * Extract the next field, given pointer to start address.
3027c478bd9Sstevel@tonic-gate  * Return pointer to beginning of field or NULL.
3037c478bd9Sstevel@tonic-gate  * Reset end of field reference, which is the beginning of the
3047c478bd9Sstevel@tonic-gate  * next field.
3057c478bd9Sstevel@tonic-gate  */
3067c478bd9Sstevel@tonic-gate wchar_t *
whitefield(wchar_t ** endp)3077c478bd9Sstevel@tonic-gate whitefield(wchar_t **endp)
3087c478bd9Sstevel@tonic-gate {
309cb4658fbSceastha 	wchar_t *sp;
310cb4658fbSceastha 	wchar_t *ep;
3117c478bd9Sstevel@tonic-gate 
3127c478bd9Sstevel@tonic-gate 	sp = *endp;
3137c478bd9Sstevel@tonic-gate 	while (*sp == ' ' || *sp == '\t' || *sp == '\n')
3147c478bd9Sstevel@tonic-gate 		++sp;
3157c478bd9Sstevel@tonic-gate 	if (*sp == '\0')
3167c478bd9Sstevel@tonic-gate 		return (NULL);
317cb4658fbSceastha 	for (ep = sp; *ep != ' ' && *ep != '\0' && *ep != '\t' &&
318cb4658fbSceastha 	    *ep != '\n'; ++ep)
3197c478bd9Sstevel@tonic-gate 		;
3207c478bd9Sstevel@tonic-gate 	*endp = ep;
3217c478bd9Sstevel@tonic-gate 	return (sp);
3227c478bd9Sstevel@tonic-gate }
3237c478bd9Sstevel@tonic-gate 
3247c478bd9Sstevel@tonic-gate /*
3257c478bd9Sstevel@tonic-gate  * Look for fields separated by non-whitespace characters.
3267c478bd9Sstevel@tonic-gate  * Same calling sequence as whitefield().
3277c478bd9Sstevel@tonic-gate  */
3287c478bd9Sstevel@tonic-gate wchar_t *
blackfield(wchar_t ** endp)3297c478bd9Sstevel@tonic-gate blackfield(wchar_t **endp)
3307c478bd9Sstevel@tonic-gate {
331cb4658fbSceastha 	wchar_t *cp;
332cb4658fbSceastha 	int endc;
3337c478bd9Sstevel@tonic-gate 
3347c478bd9Sstevel@tonic-gate 	endc = *(wchar_t *)varFS->n_string;
3357c478bd9Sstevel@tonic-gate 	cp = *endp;
3367c478bd9Sstevel@tonic-gate 	if (*cp == '\0')
3377c478bd9Sstevel@tonic-gate 		return (NULL);
3387c478bd9Sstevel@tonic-gate 	if (*cp == endc && fcount != 0)
3397c478bd9Sstevel@tonic-gate 		cp++;
3407c478bd9Sstevel@tonic-gate 	if ((*endp = wcschr(cp, endc)) == NULL)
3417c478bd9Sstevel@tonic-gate 		*endp = wcschr(cp, '\0');
3427c478bd9Sstevel@tonic-gate 	return (cp);
3437c478bd9Sstevel@tonic-gate }
3447c478bd9Sstevel@tonic-gate 
3457c478bd9Sstevel@tonic-gate /*
3467c478bd9Sstevel@tonic-gate  * This field separation routine uses the same logic as
3477c478bd9Sstevel@tonic-gate  * blackfield but uses a regular expression to separate
3487c478bd9Sstevel@tonic-gate  * the fields.
3497c478bd9Sstevel@tonic-gate  */
3507c478bd9Sstevel@tonic-gate wchar_t *
refield(wchar_t ** endpp)3517c478bd9Sstevel@tonic-gate refield(wchar_t **endpp)
3527c478bd9Sstevel@tonic-gate {
353cb4658fbSceastha 	wchar_t *cp, *start;
354cb4658fbSceastha 	int flags;
3557c478bd9Sstevel@tonic-gate 	static	REGWMATCH_T match[10];
3567c478bd9Sstevel@tonic-gate 	int result;
3577c478bd9Sstevel@tonic-gate 
3587c478bd9Sstevel@tonic-gate 	cp = *endpp;
3597c478bd9Sstevel@tonic-gate 	if (*cp == '\0') {
3607c478bd9Sstevel@tonic-gate 		match[0].rm_ep = NULL;
3617c478bd9Sstevel@tonic-gate 		return (NULL);
3627c478bd9Sstevel@tonic-gate 	}
3637c478bd9Sstevel@tonic-gate 	if (match[0].rm_ep != NULL) {
3647c478bd9Sstevel@tonic-gate 		flags = REG_NOTBOL;
3657c478bd9Sstevel@tonic-gate 		cp = (wchar_t *)match[0].rm_ep;
3667c478bd9Sstevel@tonic-gate 	} else
3677c478bd9Sstevel@tonic-gate 		flags = 0;
3687c478bd9Sstevel@tonic-gate 	start = cp;
3697c478bd9Sstevel@tonic-gate again:
3707c478bd9Sstevel@tonic-gate 	switch ((result = REGWEXEC(resep, cp, 10, match, flags))) {
3717c478bd9Sstevel@tonic-gate 	case REG_OK:
3727c478bd9Sstevel@tonic-gate 		/*
3737c478bd9Sstevel@tonic-gate 		 * Check to see if a null string was matched. If this is the
3747c478bd9Sstevel@tonic-gate 		 * case, then move the current pointer beyond this position.
3757c478bd9Sstevel@tonic-gate 		 */
3767c478bd9Sstevel@tonic-gate 		if (match[0].rm_sp == match[0].rm_ep) {
3777c478bd9Sstevel@tonic-gate 			cp = (wchar_t *)match[0].rm_sp;
3787c478bd9Sstevel@tonic-gate 			if (*cp++ != '\0') {
3797c478bd9Sstevel@tonic-gate 				goto again;
3807c478bd9Sstevel@tonic-gate 			}
3817c478bd9Sstevel@tonic-gate 		}
3827c478bd9Sstevel@tonic-gate 		*endpp = (wchar_t *)match[0].rm_sp;
3837c478bd9Sstevel@tonic-gate 		break;
3847c478bd9Sstevel@tonic-gate 	case REG_NOMATCH:
3857c478bd9Sstevel@tonic-gate 		match[0].rm_ep = NULL;
3867c478bd9Sstevel@tonic-gate 		*endpp = wcschr(cp, '\0');
3877c478bd9Sstevel@tonic-gate 		break;
3887c478bd9Sstevel@tonic-gate 	default:
389*79777a7dSnakanon 		(void) REGWERROR(result, resep, (char *)linebuf,
390cb4658fbSceastha 		    sizeof (linebuf));
3917c478bd9Sstevel@tonic-gate 		awkerr(gettext("error splitting record: %s"),
3927c478bd9Sstevel@tonic-gate 		    (char *)linebuf);
3937c478bd9Sstevel@tonic-gate 	}
3947c478bd9Sstevel@tonic-gate 	return (start);
3957c478bd9Sstevel@tonic-gate }
3967c478bd9Sstevel@tonic-gate 
3977c478bd9Sstevel@tonic-gate /*
3987c478bd9Sstevel@tonic-gate  * do begin processing
3997c478bd9Sstevel@tonic-gate  */
4007c478bd9Sstevel@tonic-gate void
dobegin()4017c478bd9Sstevel@tonic-gate dobegin()
4027c478bd9Sstevel@tonic-gate {
403cb4658fbSceastha 	/*
4047c478bd9Sstevel@tonic-gate 	 * Free all keyword nodes to save space.
4057c478bd9Sstevel@tonic-gate 	 */
4067c478bd9Sstevel@tonic-gate 	{
4077c478bd9Sstevel@tonic-gate 		NODE *np;
4087c478bd9Sstevel@tonic-gate 		int nbuck;
409cb4658fbSceastha 		NODE *knp;
4107c478bd9Sstevel@tonic-gate 
4117c478bd9Sstevel@tonic-gate 		np = NNULL;
4127c478bd9Sstevel@tonic-gate 		nbuck = 0;
4137c478bd9Sstevel@tonic-gate 		while ((knp = symwalk(&nbuck, &np)) != NNULL)
4147c478bd9Sstevel@tonic-gate 			if (knp->n_type == KEYWORD)
4157c478bd9Sstevel@tonic-gate 				delsymtab(knp, 1);
4167c478bd9Sstevel@tonic-gate 	}
417cb4658fbSceastha 	/*
4187c478bd9Sstevel@tonic-gate 	 * Copy ENVIRON array only if needed.
4197c478bd9Sstevel@tonic-gate 	 * Note the convoluted work to assign to an array
4207c478bd9Sstevel@tonic-gate 	 * and that the temporary nodes will be freed by
4217c478bd9Sstevel@tonic-gate 	 * freetemps() because we are "running".
4227c478bd9Sstevel@tonic-gate 	 */
4237c478bd9Sstevel@tonic-gate 	if (needenviron) {
424cb4658fbSceastha 		char **app;
425cb4658fbSceastha 		wchar_t *name, *value;
426cb4658fbSceastha 		NODE *namep = stringnode(_null, FSTATIC, 0);
427cb4658fbSceastha 		NODE *valuep = stringnode(_null, FSTATIC, 0);
428cb4658fbSceastha 		NODE *ENVsubname = node(INDEX, varENVIRON, namep);
4297c478bd9Sstevel@tonic-gate 		extern char **environ;
4307c478bd9Sstevel@tonic-gate 
4317c478bd9Sstevel@tonic-gate 		/* (void) m_setenv(); XXX what's this do? */
432cb4658fbSceastha 		for (app = environ; *app != NULL; /* empty */) {
4337c478bd9Sstevel@tonic-gate 			name = mbstowcsdup(*app++);
4347c478bd9Sstevel@tonic-gate 
4357c478bd9Sstevel@tonic-gate 			if ((value = wcschr(name, '=')) != NULL) {
4367c478bd9Sstevel@tonic-gate 				*value++ = '\0';
4377c478bd9Sstevel@tonic-gate 				valuep->n_strlen = wcslen(value);
4387c478bd9Sstevel@tonic-gate 				valuep->n_string = value;
4397c478bd9Sstevel@tonic-gate 			} else {
4407c478bd9Sstevel@tonic-gate 				valuep->n_strlen = 0;
4417c478bd9Sstevel@tonic-gate 				valuep->n_string = _null;
4427c478bd9Sstevel@tonic-gate 			}
4437c478bd9Sstevel@tonic-gate 			namep->n_strlen = wcslen(namep->n_string = name);
4447c478bd9Sstevel@tonic-gate 			(void) assign(ENVsubname, valuep);
4457c478bd9Sstevel@tonic-gate 			if (value != NULL)
4467c478bd9Sstevel@tonic-gate 				value[-1] = '=';
4477c478bd9Sstevel@tonic-gate 		}
4487c478bd9Sstevel@tonic-gate 	}
4497c478bd9Sstevel@tonic-gate 	phase = BEGIN;
4507c478bd9Sstevel@tonic-gate 	execute(yytree);
4517c478bd9Sstevel@tonic-gate 	phase = 0;
4527c478bd9Sstevel@tonic-gate 	if (npattern == 0)
4537c478bd9Sstevel@tonic-gate 		doend(0);
4547c478bd9Sstevel@tonic-gate 	/*
4557c478bd9Sstevel@tonic-gate 	 * Delete all pattern/action rules that are BEGIN at this
4567c478bd9Sstevel@tonic-gate 	 * point to save space.
4577c478bd9Sstevel@tonic-gate 	 * NOTE: this is not yet implemented.
4587c478bd9Sstevel@tonic-gate 	 */
4597c478bd9Sstevel@tonic-gate }
4607c478bd9Sstevel@tonic-gate 
4617c478bd9Sstevel@tonic-gate /*
4627c478bd9Sstevel@tonic-gate  * Do end processing.
4637c478bd9Sstevel@tonic-gate  * Exit with a status
4647c478bd9Sstevel@tonic-gate  */
4657c478bd9Sstevel@tonic-gate void
doend(int s)4667c478bd9Sstevel@tonic-gate doend(int s)
4677c478bd9Sstevel@tonic-gate {
468cb4658fbSceastha 	OFILE *op;
4697c478bd9Sstevel@tonic-gate 
4707c478bd9Sstevel@tonic-gate 	if (phase != END) {
4717c478bd9Sstevel@tonic-gate 		phase = END;
4727c478bd9Sstevel@tonic-gate 		awkinfp = stdin;
4737c478bd9Sstevel@tonic-gate 		execute(yytree);
4747c478bd9Sstevel@tonic-gate 	}
4757c478bd9Sstevel@tonic-gate 	for (op = &ofiles[0]; op < &ofiles[NIOSTREAM]; op++)
4767c478bd9Sstevel@tonic-gate 		if (op->f_fp != FNULL)
4777c478bd9Sstevel@tonic-gate 			awkclose(op);
4787c478bd9Sstevel@tonic-gate 	if (awkinfp == stdin)
4797c478bd9Sstevel@tonic-gate 		(void) fflush(awkinfp);
4807c478bd9Sstevel@tonic-gate 	exit(s);
4817c478bd9Sstevel@tonic-gate }
4827c478bd9Sstevel@tonic-gate 
4837c478bd9Sstevel@tonic-gate /*
4847c478bd9Sstevel@tonic-gate  * Print statement.
4857c478bd9Sstevel@tonic-gate  */
4867c478bd9Sstevel@tonic-gate void
s_print(NODE * np)4877c478bd9Sstevel@tonic-gate s_print(NODE *np)
4887c478bd9Sstevel@tonic-gate {
489cb4658fbSceastha 	FILE *fp;
4907c478bd9Sstevel@tonic-gate 	NODE *listp;
491cb4658fbSceastha 	char *ofs;
492cb4658fbSceastha 	int notfirst = 0;
4937c478bd9Sstevel@tonic-gate 
4947c478bd9Sstevel@tonic-gate 	fp = openfile(np->n_right, 1, 1);
4957c478bd9Sstevel@tonic-gate 	if (np->n_left == NNULL)
4967c478bd9Sstevel@tonic-gate 		(void) fputs(mbunconvert(linebuf), fp);
4977c478bd9Sstevel@tonic-gate 	else {
4987c478bd9Sstevel@tonic-gate 		ofs = wcstombsdup((isstring(varOFS->n_flags)) ?
4997c478bd9Sstevel@tonic-gate 		    (wchar_t *)varOFS->n_string :
5007c478bd9Sstevel@tonic-gate 		    (wchar_t *)exprstring(varOFS));
5017c478bd9Sstevel@tonic-gate 		listp = np->n_left;
5027c478bd9Sstevel@tonic-gate 		while ((np = getlist(&listp)) != NNULL) {
5037c478bd9Sstevel@tonic-gate 			if (notfirst++)
5047c478bd9Sstevel@tonic-gate 				(void) fputs(ofs, fp);
5057c478bd9Sstevel@tonic-gate 			np = exprreduce(np);
5067c478bd9Sstevel@tonic-gate 			if (np->n_flags & FINT)
5077c478bd9Sstevel@tonic-gate 				(void) fprintf(fp, "%lld", (INT)np->n_int);
5087c478bd9Sstevel@tonic-gate 			else if (isstring(np->n_flags))
509cb4658fbSceastha 				(void) fprintf(fp, "%S", np->n_string);
5107c478bd9Sstevel@tonic-gate 			else
5117c478bd9Sstevel@tonic-gate 				(void) fprintf(fp,
5127c478bd9Sstevel@tonic-gate 				    mbunconvert((wchar_t *)exprstring(varOFMT)),
5137c478bd9Sstevel@tonic-gate 				    (double)np->n_real);
5147c478bd9Sstevel@tonic-gate 		}
5157c478bd9Sstevel@tonic-gate 		free(ofs);
5167c478bd9Sstevel@tonic-gate 	}
5177c478bd9Sstevel@tonic-gate 	(void) fputs(mbunconvert(isstring(varORS->n_flags) ?
5187c478bd9Sstevel@tonic-gate 	    (wchar_t *)varORS->n_string : (wchar_t *)exprstring(varORS)),
5197c478bd9Sstevel@tonic-gate 	    fp);
5207c478bd9Sstevel@tonic-gate 	if (ferror(fp))
5217c478bd9Sstevel@tonic-gate 		awkperr("error on print");
5227c478bd9Sstevel@tonic-gate }
5237c478bd9Sstevel@tonic-gate 
5247c478bd9Sstevel@tonic-gate /*
5257c478bd9Sstevel@tonic-gate  * printf statement.
5267c478bd9Sstevel@tonic-gate  */
5277c478bd9Sstevel@tonic-gate void
s_prf(NODE * np)5287c478bd9Sstevel@tonic-gate s_prf(NODE *np)
5297c478bd9Sstevel@tonic-gate {
530cb4658fbSceastha 	FILE *fp;
5317c478bd9Sstevel@tonic-gate 
5327c478bd9Sstevel@tonic-gate 	fp = openfile(np->n_right, 1, 1);
5337c478bd9Sstevel@tonic-gate 	(void) xprintf(np->n_left, fp, (wchar_t **)NULL);
5347c478bd9Sstevel@tonic-gate 	if (ferror(fp))
5357c478bd9Sstevel@tonic-gate 		awkperr("error on printf");
5367c478bd9Sstevel@tonic-gate }
5377c478bd9Sstevel@tonic-gate 
5387c478bd9Sstevel@tonic-gate /*
5397c478bd9Sstevel@tonic-gate  * Get next input line.
5407c478bd9Sstevel@tonic-gate  * Read into variable on left of node (or $0 if NULL).
5417c478bd9Sstevel@tonic-gate  * Read from pipe or file on right of node (or from regular
5427c478bd9Sstevel@tonic-gate  * input if NULL).
5437c478bd9Sstevel@tonic-gate  * This is an oddball inasmuch as it is a function
5447c478bd9Sstevel@tonic-gate  * but parses more like the keywords print, etc.
5457c478bd9Sstevel@tonic-gate  */
5467c478bd9Sstevel@tonic-gate NODE *
f_getline(NODE * np)5477c478bd9Sstevel@tonic-gate f_getline(NODE *np)
5487c478bd9Sstevel@tonic-gate {
549cb4658fbSceastha 	wchar_t *cp;
550cb4658fbSceastha 	INT ret;
551cb4658fbSceastha 	FILE *fp;
552cb4658fbSceastha 	size_t len;
5537c478bd9Sstevel@tonic-gate 
5547c478bd9Sstevel@tonic-gate 	if (np->n_right == NULL && phase == END) {
5557c478bd9Sstevel@tonic-gate 		/* Pretend we've reached end of (the non-existant) file. */
556cb4658fbSceastha 		return (intnode(0));
5577c478bd9Sstevel@tonic-gate 	}
5587c478bd9Sstevel@tonic-gate 
5597c478bd9Sstevel@tonic-gate 	if ((fp = openfile(np->n_right, 0, 0)) != FNULL) {
5607c478bd9Sstevel@tonic-gate 		if (np->n_left == NNULL) {
5617c478bd9Sstevel@tonic-gate 			ret = nextrecord(linebuf, fp);
5627c478bd9Sstevel@tonic-gate 		} else {
5637c478bd9Sstevel@tonic-gate 			cp = emalloc(NLINE * sizeof (wchar_t));
5647c478bd9Sstevel@tonic-gate 			ret = nextrecord(cp, fp);
5657c478bd9Sstevel@tonic-gate 			np = np->n_left;
5667c478bd9Sstevel@tonic-gate 			len = wcslen(cp);
5677c478bd9Sstevel@tonic-gate 			cp = erealloc(cp, (len+1)*sizeof (wchar_t));
5687c478bd9Sstevel@tonic-gate 			if (isleaf(np->n_flags)) {
5697c478bd9Sstevel@tonic-gate 				if (np->n_type == PARM)
5707c478bd9Sstevel@tonic-gate 					np = np->n_next;
5717c478bd9Sstevel@tonic-gate 				strassign(np, cp, FNOALLOC, len);
5727c478bd9Sstevel@tonic-gate 			} else
573cb4658fbSceastha 				(void) assign(np, stringnode(cp,
574cb4658fbSceastha 				    FNOALLOC, len));
5757c478bd9Sstevel@tonic-gate 		}
5767c478bd9Sstevel@tonic-gate 	} else
5777c478bd9Sstevel@tonic-gate 		ret = -1;
5787c478bd9Sstevel@tonic-gate 	return (intnode(ret));
5797c478bd9Sstevel@tonic-gate }
5807c478bd9Sstevel@tonic-gate 
5817c478bd9Sstevel@tonic-gate /*
5827c478bd9Sstevel@tonic-gate  * Open a file.  Flag is non-zero for output.
5837c478bd9Sstevel@tonic-gate  */
5847c478bd9Sstevel@tonic-gate static FILE *
openfile(NODE * np,int flag,int fatal)5857c478bd9Sstevel@tonic-gate openfile(NODE *np, int flag, int fatal)
5867c478bd9Sstevel@tonic-gate {
587cb4658fbSceastha 	OFILE *op;
588cb4658fbSceastha 	char *cp;
589cb4658fbSceastha 	FILE *fp;
590cb4658fbSceastha 	int type;
591cb4658fbSceastha 	OFILE *fop;
5927c478bd9Sstevel@tonic-gate 
5937c478bd9Sstevel@tonic-gate 	if (np == NNULL) {
5947c478bd9Sstevel@tonic-gate 		if (flag)
5957c478bd9Sstevel@tonic-gate 			return (stdout);
5967c478bd9Sstevel@tonic-gate 		if (awkinfp == FNULL)
5977c478bd9Sstevel@tonic-gate 			awkinfp = newfile();
5987c478bd9Sstevel@tonic-gate 		return (awkinfp);
5997c478bd9Sstevel@tonic-gate 	}
6007c478bd9Sstevel@tonic-gate 	if ((type = np->n_type) == APPEND)
6017c478bd9Sstevel@tonic-gate 		type = WRITE;
6027c478bd9Sstevel@tonic-gate 	cp = mbunconvert(exprstring(np->n_left));
6037c478bd9Sstevel@tonic-gate 	fop = (OFILE *)NULL;
6047c478bd9Sstevel@tonic-gate 	for (op = &ofiles[0]; op < &ofiles[NIOSTREAM]; op++) {
6057c478bd9Sstevel@tonic-gate 		if (op->f_fp == FNULL) {
6067c478bd9Sstevel@tonic-gate 			if (fop == (OFILE *)NULL)
6077c478bd9Sstevel@tonic-gate 				fop = op;
6087c478bd9Sstevel@tonic-gate 			continue;
6097c478bd9Sstevel@tonic-gate 		}
610cb4658fbSceastha 		if (op->f_mode == type && strcmp(op->f_name, cp) == 0)
6117c478bd9Sstevel@tonic-gate 			return (op->f_fp);
6127c478bd9Sstevel@tonic-gate 	}
6137c478bd9Sstevel@tonic-gate 	if (fop == (OFILE *)NULL)
6147c478bd9Sstevel@tonic-gate 		awkerr(gettext("too many open streams to %s onto \"%s\""),
6157c478bd9Sstevel@tonic-gate 		    flag ? "print/printf" : "getline", cp);
6167c478bd9Sstevel@tonic-gate 	(void) fflush(stdout);
6177c478bd9Sstevel@tonic-gate 	op = fop;
6187c478bd9Sstevel@tonic-gate 	if (cp[0] == '-' && cp[1] == '\0') {
6197c478bd9Sstevel@tonic-gate 		fp = flag ? stdout : stdin;
6207c478bd9Sstevel@tonic-gate 	} else {
6217c478bd9Sstevel@tonic-gate 		switch (np->n_type) {
6227c478bd9Sstevel@tonic-gate 		case WRITE:
6237c478bd9Sstevel@tonic-gate 			if ((fp = fopen(cp, w)) != FNULL) {
6247c478bd9Sstevel@tonic-gate 				if (isatty(fileno(fp)))
6257c478bd9Sstevel@tonic-gate 					(void) setvbuf(fp, 0, _IONBF, 0);
6267c478bd9Sstevel@tonic-gate 			}
6277c478bd9Sstevel@tonic-gate 			break;
6287c478bd9Sstevel@tonic-gate 
6297c478bd9Sstevel@tonic-gate 		case APPEND:
6307c478bd9Sstevel@tonic-gate 			fp = fopen(cp, "a");
6317c478bd9Sstevel@tonic-gate 			break;
6327c478bd9Sstevel@tonic-gate 
6337c478bd9Sstevel@tonic-gate 		case PIPE:
6347c478bd9Sstevel@tonic-gate 			fp = popen(cp, w);
6357c478bd9Sstevel@tonic-gate 			(void) setvbuf(fp, (char *)0, _IOLBF, 0);
6367c478bd9Sstevel@tonic-gate 			break;
6377c478bd9Sstevel@tonic-gate 
6387c478bd9Sstevel@tonic-gate 		case PIPESYM:
6397c478bd9Sstevel@tonic-gate 			fp = popen(cp, r);
6407c478bd9Sstevel@tonic-gate 			break;
6417c478bd9Sstevel@tonic-gate 
6427c478bd9Sstevel@tonic-gate 		case LT:
6437c478bd9Sstevel@tonic-gate 			fp = fopen(cp, r);
6447c478bd9Sstevel@tonic-gate 			break;
6457c478bd9Sstevel@tonic-gate 
6467c478bd9Sstevel@tonic-gate 		default:
6477c478bd9Sstevel@tonic-gate 			awkerr(interr, "openfile");
6487c478bd9Sstevel@tonic-gate 		}
6497c478bd9Sstevel@tonic-gate 	}
6507c478bd9Sstevel@tonic-gate 	if (fp != FNULL) {
6517c478bd9Sstevel@tonic-gate 		op->f_name = strdup(cp);
6527c478bd9Sstevel@tonic-gate 		op->f_fp = fp;
6537c478bd9Sstevel@tonic-gate 		op->f_mode = type;
6547c478bd9Sstevel@tonic-gate 	} else if (fatal) {
6557c478bd9Sstevel@tonic-gate 		awkperr(flag ? gettext("output file \"%s\"") :
6567c478bd9Sstevel@tonic-gate 		    gettext("input file \"%s\""), cp);
6577c478bd9Sstevel@tonic-gate 	}
6587c478bd9Sstevel@tonic-gate 	return (fp);
6597c478bd9Sstevel@tonic-gate }
6607c478bd9Sstevel@tonic-gate 
6617c478bd9Sstevel@tonic-gate /*
6627c478bd9Sstevel@tonic-gate  * Close a stream.
6637c478bd9Sstevel@tonic-gate  */
6647c478bd9Sstevel@tonic-gate void
awkclose(OFILE * op)6657c478bd9Sstevel@tonic-gate awkclose(OFILE *op)
6667c478bd9Sstevel@tonic-gate {
6677c478bd9Sstevel@tonic-gate 	if (op->f_mode == PIPE || op->f_mode == PIPESYM)
6687c478bd9Sstevel@tonic-gate 		(void) pclose(op->f_fp);
6697c478bd9Sstevel@tonic-gate 	else if (fclose(op->f_fp) == EOF)
6707c478bd9Sstevel@tonic-gate 		awkperr("error on stream \"%s\"", op->f_name);
6717c478bd9Sstevel@tonic-gate 	op->f_fp = FNULL;
6727c478bd9Sstevel@tonic-gate 	free(op->f_name);
6737c478bd9Sstevel@tonic-gate 	op->f_name = NULL;
6747c478bd9Sstevel@tonic-gate }
6757c478bd9Sstevel@tonic-gate 
6767c478bd9Sstevel@tonic-gate /*
6777c478bd9Sstevel@tonic-gate  * Internal routine common to printf, sprintf.
6787c478bd9Sstevel@tonic-gate  * The node is that describing the arguments.
6797c478bd9Sstevel@tonic-gate  * Returns the number of characters written to file
6807c478bd9Sstevel@tonic-gate  * pointer `fp' or the length of the string return
6817c478bd9Sstevel@tonic-gate  * in cp. If cp is NULL then the file pointer is used. If
6827c478bd9Sstevel@tonic-gate  * cp points to a string pointer, a pointer to an allocated
6837c478bd9Sstevel@tonic-gate  * buffer will be returned in it.
6847c478bd9Sstevel@tonic-gate  */
6857c478bd9Sstevel@tonic-gate size_t
xprintf(NODE * np,FILE * fp,wchar_t ** cp)6867c478bd9Sstevel@tonic-gate xprintf(NODE *np, FILE *fp, wchar_t **cp)
6877c478bd9Sstevel@tonic-gate {
688cb4658fbSceastha 	wchar_t *fmt;
689cb4658fbSceastha 	int c;
6907c478bd9Sstevel@tonic-gate 	wchar_t *bptr = (wchar_t *)NULL;
6917c478bd9Sstevel@tonic-gate 	char fmtbuf[40];
692cb4658fbSceastha 	size_t length = 0;
693cb4658fbSceastha 	char *ofmtp;
694cb4658fbSceastha 	NODE *fnp;
695cb4658fbSceastha 	wchar_t *fmtsave;
6967c478bd9Sstevel@tonic-gate 	int slen;
6977c478bd9Sstevel@tonic-gate 	int cplen;
6987c478bd9Sstevel@tonic-gate 
6997c478bd9Sstevel@tonic-gate 	fnp = getlist(&np);
7007c478bd9Sstevel@tonic-gate 	if (isleaf(fnp->n_flags) && fnp->n_type == PARM)
7017c478bd9Sstevel@tonic-gate 		fnp = fnp->n_next;
7027c478bd9Sstevel@tonic-gate 	if (isstring(fnp->n_flags)) {
7037c478bd9Sstevel@tonic-gate 		fmt = fnp->n_string;
7047c478bd9Sstevel@tonic-gate 		fmtsave = NULL;
7057c478bd9Sstevel@tonic-gate 	} else
7067c478bd9Sstevel@tonic-gate 		fmtsave = fmt = (wchar_t *)strsave(exprstring(fnp));
7077c478bd9Sstevel@tonic-gate 
7087c478bd9Sstevel@tonic-gate 	/*
7097c478bd9Sstevel@tonic-gate 	 * if a char * pointer has been passed in then allocate an initial
7107c478bd9Sstevel@tonic-gate 	 * buffer for the string. Make it LINE_MAX plus the length of
7117c478bd9Sstevel@tonic-gate 	 * the format string but do reallocs only based LINE_MAX.
7127c478bd9Sstevel@tonic-gate 	 */
7137c478bd9Sstevel@tonic-gate 	if (cp != (wchar_t **)NULL) {
7147c478bd9Sstevel@tonic-gate 		cplen = LINE_MAX;
7157c478bd9Sstevel@tonic-gate 		bptr = *cp = emalloc(sizeof (wchar_t) * (cplen + wcslen(fmt)));
7167c478bd9Sstevel@tonic-gate 	}
7177c478bd9Sstevel@tonic-gate 
7187c478bd9Sstevel@tonic-gate 	while ((c = *fmt++) != '\0') {
7197c478bd9Sstevel@tonic-gate 		if (c != '%') {
7207c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
7217c478bd9Sstevel@tonic-gate 				awk_putwc(c, fp);
7227c478bd9Sstevel@tonic-gate 			else
7237c478bd9Sstevel@tonic-gate 				*bptr++ = c;
7247c478bd9Sstevel@tonic-gate 			++length;
7257c478bd9Sstevel@tonic-gate 			continue;
7267c478bd9Sstevel@tonic-gate 		}
7277c478bd9Sstevel@tonic-gate 		ofmtp = fmtbuf;
7287c478bd9Sstevel@tonic-gate 		*ofmtp++ = (char)c;
7297c478bd9Sstevel@tonic-gate 	nextc:
7307c478bd9Sstevel@tonic-gate 		switch (c = *fmt++) {
7317c478bd9Sstevel@tonic-gate 		case '%':
7327c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
7337c478bd9Sstevel@tonic-gate 				awk_putwc(c, fp);
7347c478bd9Sstevel@tonic-gate 			else
7357c478bd9Sstevel@tonic-gate 				*bptr++ = c;
7367c478bd9Sstevel@tonic-gate 			++length;
7377c478bd9Sstevel@tonic-gate 			continue;
7387c478bd9Sstevel@tonic-gate 
7397c478bd9Sstevel@tonic-gate 		case 'c':
7407c478bd9Sstevel@tonic-gate 			*ofmtp++ = 'w';
7417c478bd9Sstevel@tonic-gate 			*ofmtp++ = 'c';
7427c478bd9Sstevel@tonic-gate 			*ofmtp = '\0';
7437c478bd9Sstevel@tonic-gate 			fnp = exprreduce(nextarg(&np));
7447c478bd9Sstevel@tonic-gate 			if (isnumber(fnp->n_flags))
7457c478bd9Sstevel@tonic-gate 				c = exprint(fnp);
7467c478bd9Sstevel@tonic-gate 			else
7477c478bd9Sstevel@tonic-gate 				c = *(wchar_t *)exprstring(fnp);
7487c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
7497c478bd9Sstevel@tonic-gate 				length += fprintf(fp, fmtbuf, c);
7507c478bd9Sstevel@tonic-gate 			else {
7517c478bd9Sstevel@tonic-gate 				/*
7527c478bd9Sstevel@tonic-gate 				 * Make sure that the buffer is long
7537c478bd9Sstevel@tonic-gate 				 * enough to hold the formatted string.
7547c478bd9Sstevel@tonic-gate 				 */
7557c478bd9Sstevel@tonic-gate 				adjust_buf(cp, &cplen, &bptr, fmtbuf, 0);
7567c478bd9Sstevel@tonic-gate 				/*
7577c478bd9Sstevel@tonic-gate 				 * Since the call to adjust_buf() has already
7587c478bd9Sstevel@tonic-gate 				 * guaranteed that the buffer will be long
7597c478bd9Sstevel@tonic-gate 				 * enough, just pass in INT_MAX as
7607c478bd9Sstevel@tonic-gate 				 * the length.
7617c478bd9Sstevel@tonic-gate 				 */
7627c478bd9Sstevel@tonic-gate 				(void) wsprintf(bptr, (const char *) fmtbuf, c);
7637c478bd9Sstevel@tonic-gate 				bptr += (slen = wcslen(bptr));
7647c478bd9Sstevel@tonic-gate 				length += slen;
7657c478bd9Sstevel@tonic-gate 			}
7667c478bd9Sstevel@tonic-gate 			continue;
7677c478bd9Sstevel@tonic-gate /* XXXX Is this bogus? Figure out what s & S mean - look at original code */
7687c478bd9Sstevel@tonic-gate 		case 's':
7697c478bd9Sstevel@tonic-gate 		case 'S':
7707c478bd9Sstevel@tonic-gate 			*ofmtp++ = 'w';
7717c478bd9Sstevel@tonic-gate 			*ofmtp++ = 's';
7727c478bd9Sstevel@tonic-gate 			*ofmtp = '\0';
7737c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
7747c478bd9Sstevel@tonic-gate 				length += fprintf(fp, fmtbuf,
7757c478bd9Sstevel@tonic-gate 				    (wchar_t *)exprstring(nextarg(&np)));
7767c478bd9Sstevel@tonic-gate 			else {
7777c478bd9Sstevel@tonic-gate 				wchar_t *ts = exprstring(nextarg(&np));
7787c478bd9Sstevel@tonic-gate 
7797c478bd9Sstevel@tonic-gate 				adjust_buf(cp, &cplen, &bptr, fmtbuf,
7807c478bd9Sstevel@tonic-gate 				    wcslen(ts));
7817c478bd9Sstevel@tonic-gate 				(void) wsprintf(bptr, (const char *) fmtbuf,
7827c478bd9Sstevel@tonic-gate 				    ts);
7837c478bd9Sstevel@tonic-gate 				bptr += (slen = wcslen(bptr));
7847c478bd9Sstevel@tonic-gate 				length += slen;
7857c478bd9Sstevel@tonic-gate 			}
7867c478bd9Sstevel@tonic-gate 			continue;
7877c478bd9Sstevel@tonic-gate 
7887c478bd9Sstevel@tonic-gate 		case 'o':
7897c478bd9Sstevel@tonic-gate 		case 'O':
7907c478bd9Sstevel@tonic-gate 		case 'X':
7917c478bd9Sstevel@tonic-gate 		case 'x':
7927c478bd9Sstevel@tonic-gate 		case 'd':
7937c478bd9Sstevel@tonic-gate 		case 'i':
7947c478bd9Sstevel@tonic-gate 		case 'D':
7957c478bd9Sstevel@tonic-gate 		case 'U':
7967c478bd9Sstevel@tonic-gate 		case 'u':
7977c478bd9Sstevel@tonic-gate 			*ofmtp++ = 'l';
7987c478bd9Sstevel@tonic-gate 			*ofmtp++ = 'l'; /* now dealing with long longs */
7997c478bd9Sstevel@tonic-gate 			*ofmtp++ = c;
8007c478bd9Sstevel@tonic-gate 			*ofmtp = '\0';
8017c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
8027c478bd9Sstevel@tonic-gate 				length += fprintf(fp, fmtbuf,
8037c478bd9Sstevel@tonic-gate 				    exprint(nextarg(&np)));
8047c478bd9Sstevel@tonic-gate 			else {
8057c478bd9Sstevel@tonic-gate 				adjust_buf(cp, &cplen, &bptr, fmtbuf, 0);
8067c478bd9Sstevel@tonic-gate 				(void) wsprintf(bptr, (const char *) fmtbuf,
8077c478bd9Sstevel@tonic-gate 				    exprint(nextarg(&np)));
8087c478bd9Sstevel@tonic-gate 				bptr += (slen = wcslen(bptr));
8097c478bd9Sstevel@tonic-gate 				length += slen;
8107c478bd9Sstevel@tonic-gate 			}
8117c478bd9Sstevel@tonic-gate 			continue;
8127c478bd9Sstevel@tonic-gate 
8137c478bd9Sstevel@tonic-gate 		case 'e':
8147c478bd9Sstevel@tonic-gate 		case 'E':
8157c478bd9Sstevel@tonic-gate 		case 'f':
8167c478bd9Sstevel@tonic-gate 		case 'F':
8177c478bd9Sstevel@tonic-gate 		case 'g':
8187c478bd9Sstevel@tonic-gate 		case 'G':
8197c478bd9Sstevel@tonic-gate 			*ofmtp++ = c;
8207c478bd9Sstevel@tonic-gate 			*ofmtp = '\0';
8217c478bd9Sstevel@tonic-gate 			if (bptr == (wchar_t *)NULL)
8227c478bd9Sstevel@tonic-gate 				length += fprintf(fp, fmtbuf,
8237c478bd9Sstevel@tonic-gate 				    exprreal(nextarg(&np)));
8247c478bd9Sstevel@tonic-gate 			else {
8257c478bd9Sstevel@tonic-gate 				adjust_buf(cp, &cplen, &bptr, fmtbuf, 0);
8267c478bd9Sstevel@tonic-gate 				(void) wsprintf(bptr, (const char *) fmtbuf,
8277c478bd9Sstevel@tonic-gate 				    exprreal(nextarg(&np)));
8287c478bd9Sstevel@tonic-gate 				bptr += (slen = wcslen(bptr));
8297c478bd9Sstevel@tonic-gate 				length += slen;
8307c478bd9Sstevel@tonic-gate 			}
8317c478bd9Sstevel@tonic-gate 			continue;
8327c478bd9Sstevel@tonic-gate 
8337c478bd9Sstevel@tonic-gate 		case 'l':
8347c478bd9Sstevel@tonic-gate 		case 'L':
8357c478bd9Sstevel@tonic-gate 			break;
8367c478bd9Sstevel@tonic-gate 
8377c478bd9Sstevel@tonic-gate 		case '*':
8387c478bd9Sstevel@tonic-gate #ifdef M_BSD_SPRINTF
8397c478bd9Sstevel@tonic-gate 			sprintf(ofmtp, "%lld", (INT)exprint(nextarg(&np)));
8407c478bd9Sstevel@tonic-gate 			ofmtp += strlen(ofmtp);
8417c478bd9Sstevel@tonic-gate #else
842cb4658fbSceastha 			ofmtp += sprintf(ofmtp, "%lld",
843cb4658fbSceastha 			    (INT)exprint(nextarg(&np)));
8447c478bd9Sstevel@tonic-gate #endif
8457c478bd9Sstevel@tonic-gate 			break;
8467c478bd9Sstevel@tonic-gate 
8477c478bd9Sstevel@tonic-gate 		default:
8487c478bd9Sstevel@tonic-gate 			if (c == '\0') {
8497c478bd9Sstevel@tonic-gate 				*ofmtp = (wchar_t)NULL;
850cb4658fbSceastha 				(void) fprintf(fp, "%s", fmtbuf);
8517c478bd9Sstevel@tonic-gate 				continue;
852cb4658fbSceastha 			} else {
8537c478bd9Sstevel@tonic-gate 				*ofmtp++ = (wchar_t)c;
8547c478bd9Sstevel@tonic-gate 				break;
8557c478bd9Sstevel@tonic-gate 			}
8567c478bd9Sstevel@tonic-gate 		}
8577c478bd9Sstevel@tonic-gate 		goto nextc;
8587c478bd9Sstevel@tonic-gate 	}
8597c478bd9Sstevel@tonic-gate 	if (fmtsave != NULL)
8607c478bd9Sstevel@tonic-gate 		free(fmtsave);
8617c478bd9Sstevel@tonic-gate 	/*
8627c478bd9Sstevel@tonic-gate 	 * If printing to a character buffer then make sure it is
8637c478bd9Sstevel@tonic-gate 	 * null-terminated and only uses as much space as required.
8647c478bd9Sstevel@tonic-gate 	 */
8657c478bd9Sstevel@tonic-gate 	if (bptr != (wchar_t *)NULL) {
8667c478bd9Sstevel@tonic-gate 		*bptr = '\0';
8677c478bd9Sstevel@tonic-gate 		*cp = erealloc(*cp, (length+1) * sizeof (wchar_t));
8687c478bd9Sstevel@tonic-gate 	}
8697c478bd9Sstevel@tonic-gate 	return (length);
8707c478bd9Sstevel@tonic-gate }
8717c478bd9Sstevel@tonic-gate 
8727c478bd9Sstevel@tonic-gate /*
8737c478bd9Sstevel@tonic-gate  * Return the next argument from the list.
8747c478bd9Sstevel@tonic-gate  */
8757c478bd9Sstevel@tonic-gate static NODE *
nextarg(NODE ** npp)8767c478bd9Sstevel@tonic-gate nextarg(NODE **npp)
8777c478bd9Sstevel@tonic-gate {
878cb4658fbSceastha 	NODE *np;
8797c478bd9Sstevel@tonic-gate 
8807c478bd9Sstevel@tonic-gate 	if ((np = getlist(npp)) == NNULL)
8817c478bd9Sstevel@tonic-gate 		awkerr(gettext("insufficient arguments to printf or sprintf"));
8827c478bd9Sstevel@tonic-gate 	if (isleaf(np->n_flags) && np->n_type == PARM)
8837c478bd9Sstevel@tonic-gate 		return (np->n_next);
8847c478bd9Sstevel@tonic-gate 	return (np);
8857c478bd9Sstevel@tonic-gate }
8867c478bd9Sstevel@tonic-gate 
8877c478bd9Sstevel@tonic-gate 
8887c478bd9Sstevel@tonic-gate /*
8897c478bd9Sstevel@tonic-gate  * Check and adjust the length of the buffer that has been passed in
8907c478bd9Sstevel@tonic-gate  * to make sure that it has space to accomodate the sequence string
8917c478bd9Sstevel@tonic-gate  * described in fmtstr. This routine is used by xprintf() to allow
8927c478bd9Sstevel@tonic-gate  * for arbitrarily long sprintf() strings.
8937c478bd9Sstevel@tonic-gate  *
8947c478bd9Sstevel@tonic-gate  * bp		= start of current buffer
8957c478bd9Sstevel@tonic-gate  * len		= length of current buffer
8967c478bd9Sstevel@tonic-gate  * offset	= offset in current buffer
8977c478bd9Sstevel@tonic-gate  * fmtstr	= format string to check
8987c478bd9Sstevel@tonic-gate  * slen		= size of string for %s formats
8997c478bd9Sstevel@tonic-gate  */
9007c478bd9Sstevel@tonic-gate static void
adjust_buf(wchar_t ** bp,int * len,wchar_t ** offset,char * fmtstr,size_t slen)9017c478bd9Sstevel@tonic-gate adjust_buf(wchar_t **bp, int *len, wchar_t **offset, char *fmtstr, size_t slen)
9027c478bd9Sstevel@tonic-gate {
9037c478bd9Sstevel@tonic-gate 	int ioff;
9047c478bd9Sstevel@tonic-gate 	int width = 0;
9057c478bd9Sstevel@tonic-gate 	int prec = 0;
9067c478bd9Sstevel@tonic-gate 
9077c478bd9Sstevel@tonic-gate 	do {
9087c478bd9Sstevel@tonic-gate 		fmtstr++;
909cb4658fbSceastha 	} while (strchr("-+ 0", *fmtstr) != (char *)0 || *fmtstr == ('#'));
9107c478bd9Sstevel@tonic-gate 	if (*fmtstr != '*') {
9117c478bd9Sstevel@tonic-gate 		if (isdigit(*fmtstr)) {
9127c478bd9Sstevel@tonic-gate 			width = *fmtstr-'0';
9137c478bd9Sstevel@tonic-gate 			while (isdigit(*++fmtstr))
9147c478bd9Sstevel@tonic-gate 				width = width * 10 + *fmtstr - '0';
9157c478bd9Sstevel@tonic-gate 		}
9167c478bd9Sstevel@tonic-gate 	} else
9177c478bd9Sstevel@tonic-gate 		fmtstr++;
9187c478bd9Sstevel@tonic-gate 	if (*fmtstr == '.') {
9197c478bd9Sstevel@tonic-gate 		if (*++fmtstr != '*') {
9207c478bd9Sstevel@tonic-gate 			prec = *fmtstr-'0';
9217c478bd9Sstevel@tonic-gate 			while (isdigit(*++fmtstr))
9227c478bd9Sstevel@tonic-gate 				prec = prec * 10 + *fmtstr - '0';
9237c478bd9Sstevel@tonic-gate 		} else
9247c478bd9Sstevel@tonic-gate 			fmtstr++;
9257c478bd9Sstevel@tonic-gate 	}
9267c478bd9Sstevel@tonic-gate 	if (strchr("Llh", *fmtstr) != (char *)0)
9277c478bd9Sstevel@tonic-gate 		fmtstr++;
928cb4658fbSceastha 	if (*fmtstr == 'S') {
9297c478bd9Sstevel@tonic-gate 		if (width && slen < width)
9307c478bd9Sstevel@tonic-gate 			slen = width;
9317c478bd9Sstevel@tonic-gate 		if (prec && slen > prec)
9327c478bd9Sstevel@tonic-gate 			slen = prec;
9337c478bd9Sstevel@tonic-gate 		width = slen+1;
9347c478bd9Sstevel@tonic-gate 	} else
9357c478bd9Sstevel@tonic-gate 		if (width == 0)
9367c478bd9Sstevel@tonic-gate 			width = NUMSIZE;
9377c478bd9Sstevel@tonic-gate 
9387c478bd9Sstevel@tonic-gate 	if (*offset+ width > *bp+ *len) {
9397c478bd9Sstevel@tonic-gate 		ioff = *offset-*bp;
9407c478bd9Sstevel@tonic-gate 		*len += width+1;
9417c478bd9Sstevel@tonic-gate 		*bp = erealloc(*bp, *len * sizeof (wchar_t));
9427c478bd9Sstevel@tonic-gate 		*offset = *bp+ioff;
9437c478bd9Sstevel@tonic-gate 	}
9447c478bd9Sstevel@tonic-gate }
9457c478bd9Sstevel@tonic-gate 
9467c478bd9Sstevel@tonic-gate static void
awk_putwc(wchar_t c,FILE * fp)9477c478bd9Sstevel@tonic-gate awk_putwc(wchar_t c, FILE *fp)
9487c478bd9Sstevel@tonic-gate {
9497c478bd9Sstevel@tonic-gate 	char mb[MB_LEN_MAX];
9507c478bd9Sstevel@tonic-gate 	size_t mbl;
9517c478bd9Sstevel@tonic-gate 
9527c478bd9Sstevel@tonic-gate 	if ((mbl = wctomb(mb, c)) > 0) {
9537c478bd9Sstevel@tonic-gate 		mb[mbl] = '\0';
9547c478bd9Sstevel@tonic-gate 		(void) fputs(mb, fp);
9557c478bd9Sstevel@tonic-gate 	} else
9567c478bd9Sstevel@tonic-gate 		awkerr(gettext("invalid wide character %x"), c);
9577c478bd9Sstevel@tonic-gate }
958