Fix multiple bugs in .sh.match (#455)

This commit backports all of the relevant .sh.match bugfixes from
ksh93v-. Most of the .sh.match rewrite is from versions 2012-08-24
and 2012-10-04, with patches from later releases of 93v- and
ksh2020 also applied. Note that there are still some remaining bugs
in .sh.match, although now the total count of .sh.match bugs should
be less that before.

These are the relevant changes in the ksh93v- changelog that were
backported:
12-08-07  .sh.match no longer gets set for patterns in PS4 during
          set -x.
12-08-10  Rewrote .sh.match expansions fixing several bugs and
          improving performance.
12-08-22  .sh.match now handles subpatterns that had no matches with
          ${var//pattern} correctly.
12-08-21  A bug in setting .sh.match after ${var//pattern/string}
          when string is empty has been fixed.
12-08-21  A bug in setting .sh.match after [[ string == pattern ]]
          has been fixed.
12-08-31  A bug that could cause a core dump after
          typeset -m var=.sh.match has been fixed.
12-09-10  Fixed a bug in typeset -m the .sh.match is being renamed.
12-09-07  Fixed a bug in .sh.match code that coud cause the shell
          to quitely
13-02-21  The 12-01-16 bug fix prevented .sh.match from being used
          in the replacement string. The previous code was restored
          and a different fix which prevented .sh.match from being
          computed for nested replacement has been used instead.
13-05-28  Fixed two bug for typeset -c and typeset -m for variable
          .sh.match.

Changes:
- The SHOPT_2DMATCH option has been removed. This was already the
  default behavior previously, and now it's documented in the man
  page.
- init.c: Backported the sh_setmatch() rewrite from 93v- 2012-08-24
  and 2012-10-04.
- Backported the libast 93v- strngrpmatch() function, as the
  .sh.match rewrite requires this API.
- Backported the sh_match regression tests from ksh93v-, with many
  other sh_match tests backported from ksh2020. Much of the sh_match
  script is based on code from Roland Mainz:
  https://marc.info/?l=ast-developers&m=134606574109162&w=2
  https://marc.info/?l=ast-developers&m=134490505607093
- tests/{substring,treemove}.sh: Backported other relevant .sh.match
  fixes, with tests added to the substring and treemove test scripts.
- tests/types.sh: One of the (now reverted) memory leak bugfixes
  introduced a CI test failure in this script, so for that test the
  error message has been improved.
- string/strmatch.c: The original ksh93v- code for the strngrpmatch()
  changes introduced a crash that could occur because strlen would
  be used on a null pointer. This has been fixed by avoiding strlen
  if the string is null.

One nice side effect of these changes is a considerable performance
improvement in the shbench[1] gsub benchmark (results from 20
iterations with CCFLAGS=-Os):
--------------------------------------------------
name      /tmp/ksh-current     /tmp/ksh-matchfixes
--------------------------------------------------
gsub.ksh  0.883 [0.822-0.959]  0.457 [0.442-0.505]
--------------------------------------------------

Despite all of the many fixes and improvements in the backported
93v- .sh.match code, there are a few remaining bugs:

- .sh.match is printed with a default [0] subscript (see also
  https://github.com/ksh93/ksh/issues/308#issuecomment-1025016088):
     $ arch/*/bin/ksh -c 'echo ${!.sh.match}'
       .sh.match[0]
  This bug appears to have been introduced by the changes from
  ksh93v- 2012-08-24.
- The wrong variable name is given for 'parameter not set' errors
  (from https://marc.info/?l=ast-developers&m=134489094602596):
     $ arch/*/bin/ksh -u
     $ x=1234
     $ true "${x//~(X)([012])|([345])/}"
     $ compound co
     $ typeset -m co.array=.sh.match
     $ printf "%q\n" "${co.array[2][0]}"
     arch/linux.i386-64/bin/ksh: co.array[2][(null)]: parameter not set
- .sh.match leaks out of subshells. Further information and a
  reproducer can be found here:
  https://marc.info/?l=ast-developers&m=136292897330187

[1]: https://github.com/ksh-community/shbench

.github/workflows/ci.yml

@@ -25,7 +25,7 @@ jobs:
         LANG=ja_JP.SJIS  script -q -e -c "bin/shtests --locale --nocompile" &&
         : disable most SHOPTs, rebuild ksh &&
         sed --regexp-extended --in-place=.orig \
-          '/^SHOPT (2DMATCH|AUDIT|BGX|BRACEPAT|DEVFD|DYNAMIC|EDPREDICT|ESH|FIXEDARRAY|HISTEXPAND|MULTIBYTE|NAMESPACE|OPTIMIZE|SPAWN|STATS|SUID_EXEC|VSH)=/ s/=1?/=0/' \
+          '/^SHOPT (AUDIT|BGX|BRACEPAT|DEVFD|DYNAMIC|EDPREDICT|ESH|FIXEDARRAY|HISTEXPAND|MULTIBYTE|NAMESPACE|OPTIMIZE|SPAWN|STATS|SUID_EXEC|VSH)=/ s/=1?/=0/' \
           src/cmd/ksh93/SHOPT.sh &&
         bin/package make &&
         : default regression tests with SHOPTs disabled &&

NEWS

@@ -27,6 +27,9 @@ Any uppercase BUG_* names are modernish shell bug IDs.
   associative array when read back in by the shell. Elements that are sparse
   indexed arrays are now prefixed with "typeset -a".
 
+- The rewritten .sh.match code from ksh93v- has been backported to ksh93u+m,
+  fixing many bugs and improving performance by a considerable amount.
+
 2022-02-05:
 
 - Fixed: for indexed arrays, given an unset array member a[i] with i > 0,

src/cmd/ksh93/README

@@ -20,8 +20,6 @@ options where no feature probe is available, probe is the same as off.
 
 The options have the following defaults and meanings:
 
-    2DMATCH      on  Two-dimensional ${.sh.match} for ${var//pat/str}.
-
     ACCT         off Shell accounting.
                      Noted by "L" in the version string when enabled.
                      See README-AUDIT.md.

src/cmd/ksh93/SHOPT.sh

@@ -5,7 +5,6 @@
 # For a more complete description of the options, see src/cmd/ksh93/README.
 #
 
-SHOPT 2DMATCH=1				# two dimensional ${.sh.match} for ${var//pat/str}
 SHOPT ACCT=0				# accounting
 SHOPT ACCTFILE=0			# per-user accounting info
 SHOPT AUDIT=1				# enable auditing per SHOPT_AUDITFILE

src/cmd/ksh93/bltins/test.c

@@ -83,7 +83,7 @@ static int e3(struct test*);
 
 static int test_strmatch(const char *str, const char *pat)
 {
-	regoff_t match[2*(MATCH_MAX+1)],n;
+	int match[2*(MATCH_MAX+1)],n;
 	register int c, m=0;
 	register const char *cp=pat; 
 	while(c = *cp++)
@@ -99,9 +99,9 @@ static int test_strmatch(const char *str, const char *pat)
 		match[0] = 0;
 	if(m >  elementsof(match)/2)
 		m = elementsof(match)/2;
-	n = strgrpmatch(str, pat, match, m, STR_GROUP|STR_MAXIMAL|STR_LEFT|STR_RIGHT);
+	n = strgrpmatch(str, pat, (ssize_t*)match, m, STR_GROUP|STR_MAXIMAL|STR_LEFT|STR_RIGHT|STR_INT);
 	if(m==0 && n==1)
-		match[1] = strlen(str);
+		match[1] = (int)strlen(str);
 	if(n)
 		sh_setmatch(str, -1, n, match, 0);
 	return(n);

src/cmd/ksh93/include/defs.h

@@ -29,8 +29,8 @@
 #define defs_h_defined
 
 #include	<ast.h>
-#if !defined(AST_VERSION) || AST_VERSION < 20220201
+#if !defined(AST_VERSION) || AST_VERSION < 20220208
-#error libast version 20220201 or later is required
+#error libast version 20220208 or later is required
 #endif
 #if !_lib_fork
 #error In 2021, ksh joined the 21st century and started requiring fork(2).

src/cmd/ksh93/include/shell.h

@@ -327,6 +327,7 @@ struct Shell_s
 	char		instance;	/* in set_instance */
 	char		decomma;	/* decimal_point=',' */
 	char		redir0;		/* redirect of 0 */
+	char		intrace;	/* set when trace expands PS4 */
 	char		*readscript;	/* set before reading a script */
 	int		subdup;		/* bitmask for dups of 1 */
 	int		*inpipe;	/* input pipe pointer */

src/cmd/ksh93/sh.1

@@ -1695,6 +1695,13 @@ element stores the complete match and the
 element stores the
 .IR i\^ -th
 submatch.
+For
+.B //
+the array is two dimensional with the first subscript indicating the
+most recent match and subpattern match and the second script indicating
+which match with
+.B 0
+representing the first match.
 The
 .B .sh.match
 variable

src/cmd/ksh93/sh/array.c

@@ -1176,7 +1176,7 @@ Namval_t *nv_putsub(Namval_t *np,register char *sp,register long mode)
 	if(!ap || !ap->header.fun)
 #endif /* SHOPT_FIXEDARRAY */
 	{
-		if(sp)
+		if(sp && sp!=Empty)
 		{
 			if(ap && ap->xp && !strmatch(sp,"+([0-9])"))
 			{
@@ -1185,7 +1185,11 @@ Namval_t *nv_putsub(Namval_t *np,register char *sp,register long mode)
 				size = nv_getnum(mp);
 			}
 			else
+			{
+				Dt_t *root = sh.last_root;
 				size = (int)sh_arith((char*)sp);
+				sh.last_root = root;
+			}
 		}
 		if(size <0 && ap)
 			size += array_maxindex(np);

src/cmd/ksh93/sh/fault.c

@@ -585,6 +585,7 @@ void sh_exit(register int xno)
 	if(!pp)
 		sh_done(sig);
 	sh.arithrecursion = 0;
+	sh.intrace = 0;
 	sh.prefix = 0;
 #if SHOPT_TYPEDEF
 	sh.mktype = 0;

src/cmd/ksh93/sh/init.c

@@ -176,10 +176,13 @@ struct match
 	const char	*v;
 	char		*val;
 	char		*rval[2];
-	regoff_t	*match;
+	int		*match;
-	char		node[NV_MINSZ+sizeof(char*)+sizeof(Dtlink_t)];
+	char		*nodes;
-	regoff_t	first;
+	char		*names;
+	int		first;
 	int		vsize;
+	int		vlen;
+	int		msize;
 	int		nmatch;
 	int		index;
 	int		lastsub[2];
@@ -775,76 +778,118 @@ static void put_lastarg(Namval_t* np,const char *val,int flags,Namfun_t *fp)
 	np->nvenv = 0;
 }
 
+static void match2d(struct match *mp)
+{
+	Namval_t	*np;
+	int		i;
+	Namarr_t	*ap;
+	nv_disc(SH_MATCHNOD, &mp->hdr, NV_POP);
+	np = nv_namptr(mp->nodes, 0);
+	for(i=0; i < mp->nmatch; i++)
+	{
+		np->nvname = mp->names + 3 * i;
+		if(i > 9)
+		{
+			*np->nvname = '0' + i / 10;
+			np->nvname[1] = '0' + (i % 10);
+		}
+		else
+			*np->nvname = '0' + i;
+		nv_putsub(np, (char*)0, 1);
+		nv_putsub(np, (char*)0, 0);
+		nv_putsub(SH_MATCHNOD, (char*)0, i);
+		nv_arraychild(SH_MATCHNOD, np, 0);
+		np = nv_namptr(np + 1, 0);
+	}
+	if(ap = nv_arrayptr(SH_MATCHNOD))
+		ap->nelem = mp->nmatch;
+}
+
 /*
  * store the most recent value for use in .sh.match
  * treat .sh.match as a two dimensional array
  */
-void sh_setmatch(const char *v, int vsize, int nmatch, regoff_t match[],int index)
+void sh_setmatch(const char *v, int vsize, int nmatch, int match[], int index)
 {
-	struct match	*mp = &((Init_t*)sh.init_context)->SH_MATCH_init;
+	Init_t		*ip = sh.init_context;
-	Namval_t	*np = (Namval_t*)(&(mp->node[0]));
+	struct match	*mp = &ip->SH_MATCH_init;
-	register int	i,n,x;
+	register int	i,n,x, savesub=sh.subshell;
-	unsigned int	savesub = sh.subshell;
 	Namarr_t	*ap = nv_arrayptr(SH_MATCHNOD);
-	Namarr_t	*ap_save = ap;
+	Namval_t	*np;
-	/* do not crash if .sh.match is unset */
+	if(sh.intrace)
-	if(!ap)
 		return;
 	sh.subshell = 0;
-#if !SHOPT_2DMATCH
+	if(index<0)
-	index = 0;
-#else
-	if(index==0)
-#endif /* !SHOPT_2DMATCH */
 	{
-		if(ap->hdr.next != &mp->hdr)
+		np = nv_namptr(mp->nodes,0);
+		if(mp->index==0)
+			match2d(mp);
+		for(i=0; i < mp->nmatch; i++)
 		{
-			free((void*)ap);
+			nv_disc(np,&mp->hdr,NV_LAST);
-			ap = nv_arrayptr(np);
+			nv_putsub(np,(char*)0,mp->index);
-			SH_MATCHNOD->nvfun = &ap->hdr;
+			for(x=mp->index; x >=0; x--)
-		}
-		if(ap)
-		{
-			ap->nelem &= ~ARRAY_SCAN;
-			i = array_elem(ap);
-			ap->nelem++;
-			while(--i>= 0)
 			{
-				nv_putsub(SH_MATCHNOD, (char*)0,i);
+				n = i + x*mp->nmatch;
+				if(mp->match[2*n+1]>mp->match[2*n])
+					nv_putsub(np,Empty,ARRAY_ADD|x);
+			}
+			if((ap=nv_arrayptr(np)) && array_elem(ap)==0)
+			{
+				nv_putsub(SH_MATCHNOD,(char*)0,i);
 				_nv_unset(SH_MATCHNOD,NV_RDONLY);
 			}
-			ap->nelem--;
+			np = nv_namptr(np+1,0);
+		}
+		sh.subshell = savesub;
+		return;
+	}
+	mp->index = index;
+	if(index==0)
+	{
+		if(mp->nodes)
+		{
+			np = nv_namptr(mp->nodes,0);
+			for(i=0; i < mp->nmatch; i++)
+			{
+				if(np->nvfun && np->nvfun != &mp->hdr)
+				{
+					free((void*)np->nvfun);
+					np->nvfun = 0;
+				}
+				np = nv_namptr(np+1,0);
+			}
+			free((void*)mp->nodes);
+			mp->nodes = 0;
+		}
+		mp->vlen = 0;
+		if(ap && ap->hdr.next != &mp->hdr)
+			free((void*)ap);
+		SH_MATCHNOD->nvalue.cp = 0;
+		SH_MATCHNOD->nvfun = 0;
+		if(!(mp->nmatch=nmatch) && !v)
+		{
+			sh.subshell = savesub;
+			return;
+		}
+		mp->nodes = sh_calloc(mp->nmatch*(NV_MINSZ+sizeof(void*)+3),1);
+		mp->names = mp->nodes + mp->nmatch*(NV_MINSZ+sizeof(void*));
+		np = nv_namptr(mp->nodes,0);
+		nv_disc(SH_MATCHNOD,&mp->hdr,NV_LAST);
+		for(i=nmatch; --i>=0;)
+		{
+			if(match[2*i]>=0)
+				nv_putsub(SH_MATCHNOD,Empty,ARRAY_ADD|i);
 		}
-		if(!nv_hasdisc(SH_MATCHNOD,mp->hdr.disc))
-			nv_disc(SH_MATCHNOD,&mp->hdr,NV_LAST);
-		if(nmatch)
-			nv_putsub(SH_MATCHNOD, NIL(char*), (nmatch-1)|ARRAY_FILL|ARRAY_SETSUB);
-		ap_save->nelem = mp->nmatch = nmatch;
 		mp->v = v;
 		mp->first = match[0];
 	}
-#if SHOPT_2DMATCH
 	else
 	{
 		if(index==1)
-		{
+			match2d(mp);
-			np->nvalue.cp = Empty;
-			np->nvfun = SH_MATCHNOD->nvfun;
-			nv_onattr(np,NV_NOFREE|NV_ARRAY);
-			SH_MATCHNOD->nvfun = 0;
-			for(i=0; i < mp->nmatch; i++)
-			{
-				nv_putsub(SH_MATCHNOD, (char*)0, i);
-				nv_arraychild(SH_MATCHNOD, np,0);
-			}
-			ap_save->nelem = mp->nmatch;
-		}
-		ap = nv_arrayptr(np);
-		nv_putsub(np, NIL(char*), index|ARRAY_FILL|ARRAY_SETSUB);
 	}
-#endif /* SHOPT_2DMATCH */
 	sh.subshell = savesub;
-	index *= 2*mp->nmatch;
 	if(mp->nmatch)
 	{
 		for(n=mp->first+(mp->v-v),vsize=0,i=0; i < 2*nmatch; i++)
@@ -852,33 +897,39 @@ void sh_setmatch(const char *v, int vsize, int nmatch, regoff_t match[],int inde
 			if(match[i]>=0 && (match[i] - n) > vsize)
 				vsize = match[i] -n;
 		}
+		index *= 2*mp->nmatch;
 		i = (index+2*mp->nmatch)*sizeof(match[0]);
-		if((i+vsize) >= mp->vsize)
+		if(i >= mp->msize)
+		{
+			if(mp->msize)
+				mp->match = (int*)sh_realloc(mp->match,2*i);
+			else
+				mp->match = (int*)sh_malloc(2*i);
+			mp->msize = 2*i;
+		}
+		if(vsize >= mp->vsize)
 		{
 			if(mp->vsize)
-				mp->match = (int*)sh_realloc(mp->match,i+vsize+1);
+				mp->val = (char*)sh_realloc(mp->val,x=2*vsize);
 			else
-				mp->match = (int*)sh_malloc(i+vsize+1);
+				mp->val =  (char*)sh_malloc(x=vsize+1);
-			mp->vsize = i+vsize+1;
+			mp->vsize = x;
 		}
-		mp->val =  ((char*)mp->match)+i; 
 		memcpy(mp->match+index,match,nmatch*2*sizeof(match[0]));
-		for(x=0,i=0; i < 2*nmatch; i++)
+		for(i=0; i < 2*nmatch; i++)
 		{
 			if(match[i]>=0)
 				mp->match[index+i] -= n;
-			else
-				x=1;
-
 		}
-		ap_save->nelem -= x;
 		while(i < 2*mp->nmatch)
 			mp->match[index+i++] = -1;
-		memcpy(mp->val,v+n,vsize);
+		if(index==0)
-		mp->val[vsize] = 0;
+			v+= mp->first;
+		memcpy(mp->val+mp->vlen,v,vsize-mp->vlen);
+		mp->val[mp->vlen=vsize] = 0;
 		mp->lastsub[0] = mp->lastsub[1] = -1;
 	}
-} 
+}
 
 static char* get_match(register Namval_t* np, Namfun_t *fp)
 {
@@ -886,6 +937,8 @@ static char* get_match(register Namval_t* np, Namfun_t *fp)
 	int		sub,sub2=0,n,i =!mp->index;
 	char		*val;
 	sub = nv_aindex(SH_MATCHNOD);
+	if(sub<0)
+		sub = 0;
 	if(np!=SH_MATCHNOD)
 		sub2 = nv_aindex(np);
 	if(sub>=mp->nmatch)
@@ -915,7 +968,15 @@ static char* get_match(register Namval_t* np, Namfun_t *fp)
 	return(mp->rval[i]);
 }
 
-static const Namdisc_t SH_MATCH_disc  = { sizeof(struct match), 0, get_match };
+static char *name_match(Namval_t *np, Namfun_t *fp)
+{
+	int sub = nv_aindex(SH_MATCHNOD);
+	sfprintf(sh.strbuf,".sh.match[%d]",sub);
+	return(sfstruse(sh.strbuf));
+}
+
+static const Namdisc_t SH_MATCH_disc = { sizeof(struct match), 0, get_match,
+	0,0,0,0,name_match };
 
 static char* get_version(register Namval_t* np, Namfun_t *fp)
 {

src/cmd/ksh93/sh/macro.c

@@ -46,11 +46,6 @@
 #include	"national.h"
 #include	"streval.h"
 
-#undef STR_GROUP
-#ifndef STR_GROUP
-#   define STR_GROUP	0
-#endif
-
 #if _WINIX
     static int Skip;
 #endif /* _WINIX */
@@ -74,6 +69,7 @@ typedef struct  _mac_
 	char		arith;		/* set for ((...)) */
 	char		arrayok;	/* $x[] ok for arrays */
 	char		subcopy;	/* set when copying subscript */
+	char		macsub;		/* set to 1 when running mac_substitute */
 	int		dotdot;		/* set for .. in subscript */
 	void		*nvwalk;	/* for name space walking */
 } Mac_t;
@@ -96,7 +92,7 @@ typedef struct  _mac_
 #define M_TYPE		8	/* ${@var}	*/
 
 static noreturn void	mac_error(Namval_t*);
-static int	substring(const char*, const char*, int[], int);
+static int	substring(const char*, size_t, const char*, int[], int);
 static void	copyto(Mac_t*, int, int);
 static void	comsubst(Mac_t*, Shnode_t*, int);
 static int	varsub(Mac_t*);
@@ -862,7 +858,22 @@ done:
 static void mac_substitute(Mac_t *mp, register char *cp,char *str,register int subexp[],int subsize)
 {
 	register int	c,n;
-	register char *first=cp;
+	register char *first=fcseek(0);
+	char		*ptr;
+	Mac_t		savemac;
+	n = stktell(sh.stk);
+	savemac = *mp;
+	mp->pattern = 3;
+	mp->split = 0;
+	mp->macsub++;
+	fcsopen(cp);
+	copyto(mp,0,0);
+	sfputc(sh.stk,0);
+	ptr = cp = sh_strdup(stkptr(sh.stk,n));
+	stkseek(sh.stk,n);
+	*mp = savemac;
+	fcsopen(first);
+	first = cp;
 	while(1)
 	{
 		while((c= *cp++) && c!=ESCAPE);
@@ -890,6 +901,7 @@ static void mac_substitute(Mac_t *mp, register char *cp,char *str,register int s
 	}
 	if(n=cp-first-1)
 		mac_copy(mp,first,n);
+	free(ptr);
 }
 
 #if  SHOPT_FILESCAN
@@ -1362,7 +1374,7 @@ retry1:
 					ap = nv_arrayptr(np=nq);
 				if(ap)
 				{
-					nv_putsub(np,v,ARRAY_SCAN);
+					np = nv_putsub(np,v,ARRAY_SCAN);
 					v = stkptr(stkp,mp->dotdot);
 					dolmax =1;
 					if(array_assoc(ap))
@@ -1796,22 +1808,7 @@ retry1:
 		}
 		pattern = sh_strdup(argp);
 		if((type=='/' || c=='/') && (repstr = mac_getstring(pattern)))
-		{
-			Mac_t	savemac;
-			char	*first = fcseek(0);
-			int	n = stktell(stkp);
-			savemac = *mp;
-			fcsopen(repstr);
-			mp->pattern = 3;
-			mp->split = 0;
-			copyto(mp,0,0);
-			sfputc(stkp,0);
-			repstr = sh_strdup(stkptr(stkp,n));
 			replen = strlen(repstr);
-			stkseek(stkp,n);
-			*mp = savemac;
-			fcsopen(first);
-		}
 		if(v || c=='/' && offset>=0)
 			stkseek(stkp,offset);
 	}
@@ -1822,29 +1819,30 @@ retry2:
 	if(v && (!nulflg || *v ) && c!='+')
 	{
 		int ofs_size = 0;
-		regoff_t match[2*(MATCH_MAX+1)];
+		int match[2*(MATCH_MAX+1)],index;
-		int nmatch, nmatch_prev, vsize_last;
+		int nmatch, nmatch_prev, vsize_last, tsize;
-		char *vlast = NIL(char*);
+		char *vlast = NIL(char*), *oldv;
 		while(1)
 		{
 			if(!v)
 				v= "";
 			if(c=='/' || c=='#' || c== '%')
 			{
-				int index = 0;
 				flag = (type || c=='/')?(STR_GROUP|STR_MAXIMAL):STR_GROUP;
 				if(c!='/')
 					flag |= STR_LEFT;
-				nmatch = 0;
+				index = nmatch = 0;
+				tsize = (int)strlen(v);
 				while(1)
 				{
-					vsize = strlen(v);
+					vsize = tsize;
+					oldv = v;
 					nmatch_prev = nmatch;
 					if(c=='%')
-						nmatch=substring(v,pattern,match,flag&STR_MAXIMAL);
+						nmatch=substring(v,tsize,pattern,match,flag&STR_MAXIMAL);
 					else
-						nmatch=strgrpmatch(v,pattern,match,elementsof(match)/2,flag);
+						nmatch=strngrpmatch(v,vsize,pattern,(ssize_t*)match,elementsof(match)/2,flag|STR_INT);
-					if(nmatch && replen>0)
+					if(nmatch && repstr && !mp->macsub)
 						sh_setmatch(v,vsize,nmatch,match,index++);
 					if(nmatch)
 					{
@@ -1871,13 +1869,16 @@ retry2:
 							mac_copy(mp,v,1);
 							v++;
 						}
+						tsize -= v-oldv;
 						continue;
 					}
 					vsize = -1;
 					break;
 				}
-				if(replen==0)
+				if(!mp->macsub && (!repstr || (nmatch==0 && index==0)))
 					sh_setmatch(vlast,vsize_last,nmatch,match,index++);
+				if(!mp->macsub && index>0 && c=='/' && type)
+					sh_setmatch(0,0,nmatch,0,-1);
 			}
 			if(vsize)
 				mac_copy(mp,v,vsize>0?vsize:strlen(v));
@@ -2044,8 +2045,6 @@ retry2:
 		nv_close(np);
 	if(pattern)
 		free(pattern);
-	if(repstr)
-		free(repstr);
 	if(idx)
 		free(idx);
 	return(1);
@@ -2534,27 +2533,27 @@ static void endfield(register Mac_t *mp,int split)
  * Finds the right substring of STRING using the expression PAT
  * the longest substring is found when FLAG is set.
  */
-static int substring(register const char *string,const char *pat,int match[], int flag)
+static int substring(register const char *string,size_t len,const char *pat,int match[], int flag)
 {
 	register const char *sp=string;
-	register int size,len,nmatch,n;
+	register int size,nmatch,n;
 	int smatch[2*(MATCH_MAX+1)];
 	if(flag)
 	{
-		if(n=strgrpmatch(sp,pat,smatch,elementsof(smatch)/2,STR_RIGHT|STR_MAXIMAL))
+		if(n=strngrpmatch(sp,len,pat,(ssize_t*)smatch,elementsof(smatch)/2,STR_RIGHT|STR_MAXIMAL|STR_INT))
 		{
 			memcpy(match,smatch,n*2*sizeof(smatch[0]));
 			return(n);
 		}
 		return(0);
 	}
-	size = len = strlen(sp);
+	size = (int)len;
 	sp += size;
 	while(sp>=string)
 	{
 		if(mbwide())
 			sp = lastchar(string,sp);
-		if(n=strgrpmatch(sp,pat,smatch,elementsof(smatch)/2,STR_RIGHT|STR_LEFT|STR_MAXIMAL))
+		if(n=strgrpmatch(sp,pat,(ssize_t*)smatch,elementsof(smatch)/2,STR_RIGHT|STR_LEFT|STR_MAXIMAL|STR_INT))
 		{
 			nmatch = n;
 			memcpy(match,smatch,n*2*sizeof(smatch[0]));

src/cmd/ksh93/sh/name.c

@@ -3355,7 +3355,28 @@ int nv_rename(register Namval_t *np, int flags)
 		}
 		else
 			mp = np;
-		nv_clone(nr,mp,(flags&NV_MOVE)|NV_COMVAR);
+		if(nr==SH_MATCHNOD)
+		{
+			Sfio_t *iop;
+			Dt_t *save_root = sh.var_tree;
+			int trace = sh_isoption(SH_XTRACE);
+			sfprintf(sh.strbuf,"typeset -a %s=",nv_name(mp));
+			nv_outnode(nr,sh.strbuf,-1,0);
+			sfwrite(sh.strbuf,")\n",2);
+			cp = sfstruse(sh.strbuf);
+			iop = sfopen((Sfio_t*)0,cp,"s");
+			if(trace)
+				sh_offoption(SH_XTRACE);
+			sh.var_tree = last_root;
+			sh_eval(iop,SH_READEVAL);
+			sh.var_tree = save_root;
+			if(trace)
+				sh_onoption(SH_XTRACE);
+			if(flags&NV_MOVE)
+				sh_setmatch(0,0,0,0,0);
+		}
+		else
+			nv_clone(nr,mp,(flags&NV_MOVE)|NV_COMVAR);
 		mp->nvenv = nvenv;
 		if(flags&NV_MOVE)
 		{
@@ -3632,6 +3653,8 @@ char *nv_name(register Namval_t *np)
 #endif /* SHOPT_NAMESPACE */
 		return(np->nvname);
 	}
+	if(!np->nvname)
+		goto skip;
 #if SHOPT_FIXEDARRAY
 	ap = nv_arrayptr(np);
 #endif /* SHOPT_FIXEDARRAY */
@@ -3651,6 +3674,7 @@ char *nv_name(register Namval_t *np)
 		sh.last_table = nv_parent(np);
 	else if(!nv_isref(np))
 	{
+	skip:
 		for(fp= np->nvfun ; fp; fp=fp->next)
 		if(fp->disc && fp->disc->namef)
 		{

src/cmd/ksh93/sh/nvtree.c

@@ -670,11 +670,12 @@ void nv_outnode(Namval_t *np, Sfio_t* out, int indent, int special)
 static void outval(char *name, const char *vname, struct Walk *wp)
 {
 	register Namval_t *np, *nq=0, *last_table=sh.last_table;
-        register Namfun_t *fp;
+	register Namfun_t *fp;
 	int isarray=0, special=0,mode=0;
+	Dt_t *root = wp->root?wp->root:sh.var_base;
 	if(*name!='.' || vname[strlen(vname)-1]==']')
 		mode = NV_ARRAY;
-	if(!(np=nv_open(vname,wp->root,mode|NV_VARNAME|NV_NOADD|NV_NOASSIGN|NV_NOFAIL|wp->noscope)))
+	if(!(np=nv_open(vname,root,mode|NV_VARNAME|NV_NOADD|NV_NOASSIGN|NV_NOFAIL|wp->noscope)))
 	{
 		sh.last_table = last_table;
 		return;

src/cmd/ksh93/sh/xec.c

@@ -2816,9 +2816,11 @@ int sh_trace(register char *argv[], register int nl)
 			cp = "+ ";
 		else
 		{
+			sh.intrace = 1;
 			sh_offoption(SH_XTRACE);
 			cp = sh_mactry(cp);
 			sh_onoption(SH_XTRACE);
+			sh.intrace = 0;
 		}
 		if(*cp)
 			sfputr(sfstderr,cp,-1);

src/cmd/ksh93/tests/substring.sh

@@ -252,7 +252,7 @@ then	err_exit '${var//+(\S)/Q} not workding'
 fi
 
 var=$($SHELL -c 'v=/vin:/usr/vin r=vin; : ${v//vin/${r//v/b}};typeset -p .sh.match') 2> /dev/null
-((SHOPT_2DMATCH)) && exp='typeset -a .sh.match=((vin vin) )' || exp='typeset -a .sh.match=(vin)'
+exp='typeset -a .sh.match=((vin vin) )'
 [[ $var == "$exp" ]] || err_exit '.sh.match not correct when replacement pattern contains a substring match' \
 	"(expected $(printf %q "$exp"), got $(printf %q "$var"))"
 
@@ -683,5 +683,24 @@ Errors=$?
 # Test for a crash after unsetting ${.sh.match} then matching a pattern
 $SHELL -c 'unset .sh.match; [[ bar == ba* ]]' || err_exit 'crash after unsetting .sh.match then trying to match a pattern'
 
+# Tests for ${.sh.match} backported from ksh93v-
+unset v d
+v=abbbc
+d="${v/~(E)b{2,4}/dummy}"
+[[ ${.sh.match} == bbb ]] || err_exit '.sh.match wrong after ${s/~(E)b{2,4}/dummy}'
+[[ $d == adummyc ]] || err_exit '${s/~(E)b{2,4}/dummy} not working'
+
+x=1234
+: "${x//~(X)([012])|([345])/}"
+[[ ${.sh.match[1][600..602]} ]] && err_exit '${.sh.match[0][600..602]} is not the empty string'
+
+: "${x//~(X)([012])|([345])/}"
+x=$(print -v .sh.match)
+compound co
+typeset -m co.array=.sh.match
+[[ $x == "$(print -v co.array)" ]] || err_exit 'typeset -m for .sh.match to compound variable not working (1)'
+: "${x//~(X)([345])|([012])/}"
+[[ $x == "$(print -v co.array)" ]] || err_exit 'typeset -m for .sh.match to compound variable not working (2)'
+
 # ======
 exit $((Errors<125?Errors:125))

src/cmd/ksh93/tests/treemove.sh

@@ -2,7 +2,7 @@
 #                                                                      #
 #               This software is part of the ast package               #
 #          Copyright (c) 1982-2011 AT&T Intellectual Property          #
-#          Copyright (c) 2020-2021 Contributors to ksh 93u+m           #
+#          Copyright (c) 2020-2022 Contributors to ksh 93u+m           #
 #                      and is licensed under the                       #
 #                 Eclipse Public License, Version 1.0                  #
 #                    by AT&T Intellectual Property                     #
@@ -152,4 +152,21 @@ then	if	[[ $(kill -l $exitval) == SEGV ]]
 	else	err_exit 'typeset -m "c.board[1][i]=el" gives wrong value'
 	fi
 fi
+
+function f2
+{
+	nameref mar=$1 exp=$2
+	typeset dummy x="-1a2b3c4d9u"
+	dummy="${x//~(E)([[:digit:]])|([[:alpha:]])/D}"
+	exp=${ print -v .sh.match;}
+	typeset -m "mar=.sh.match"
+}
+function f1
+{
+	typeset matchar exp
+	f2 matchar exp
+	[[ ${ print -v matchar;} == "$exp" ]] || err_exit 'moving .sh.match to a function local variable using a name reference fails'
+}
+f1
+
 exit $((Errors<125?Errors:125))

src/cmd/ksh93/tests/types.sh

@@ -552,7 +552,8 @@ compound -a b.ca
 b_t b.ca[4].b
 exp='typeset -C b=(typeset -C -a ca=( [4]=(b_t b=(a_t b=(a=hello))));)'
 got=$(typeset -p b)
-[[ $got == "$exp" ]] || err_exit 'typeset -p of nested type not correct'
+[[ $got == "$exp" ]] || err_exit 'typeset -p of nested type not correct' \
+	"(expected $(printf %q "$exp"), got $(printf %q "$got"))"
 
 typeset -T u_t=(
 	integer dummy 

src/lib/libast/features/api

@@ -1,8 +1,8 @@
 iff AST_API
 
-ver ast 20220201
+ver ast 20220208
 
-api ast 20120528 regexec regnexec regrexec regsubexec strgrpmatch
+api ast 20120528 regexec regnexec regrexec regsubexec
 
 api ast 20120411 cmdopen
 

src/lib/libast/features/map.c

@@ -2,7 +2,7 @@
 *                                                                      *
 *               This software is part of the ast package               *
 *          Copyright (c) 1985-2011 AT&T Intellectual Property          *
-*          Copyright (c) 2020-2021 Contributors to ksh 93u+m           *
+*          Copyright (c) 2020-2022 Contributors to ksh 93u+m           *
 *                      and is licensed under the                       *
 *                 Eclipse Public License, Version 1.0                  *
 *                    by AT&T Intellectual Property                     *
@@ -322,6 +322,8 @@ main()
 	printf("#define strgid		_ast_strgid\n");
 	printf("#undef	strgrpmatch\n");
 	printf("#define strgrpmatch	_ast_strgrpmatch\n");
+	printf("#undef	strngrpmatch\n");
+	printf("#define strngrpmatch	_ast_strngrpmatch\n");
 	printf("#undef	strhash\n");
 	printf("#define strhash		_ast_strhash\n");
 	printf("#undef	strkey\n");

src/lib/libast/include/ast.h

@@ -169,11 +169,12 @@ typedef struct
  * strgrpmatch() flags
  */
 
-#define STR_MAXIMAL	01		/* maximal match		*/
+#define STR_MAXIMAL	0x01		/* maximal match		*/
-#define STR_LEFT	02		/* implicit left anchor		*/
+#define STR_LEFT	0x02		/* implicit left anchor		*/
-#define STR_RIGHT	04		/* implicit right anchor	*/
+#define STR_RIGHT	0x04		/* implicit right anchor	*/
-#define STR_ICASE	010		/* ignore case			*/
+#define STR_ICASE	0x08		/* ignore case			*/
-#define STR_GROUP	020		/* (|&) inside [@|&](...) only	*/
+#define STR_GROUP	0x10		/* (|&) inside [@|&](...) only	*/
+#define STR_INT		0x20		/* int* match array		*/
 
 /*
  * fmtquote() flags
@@ -371,8 +372,8 @@ extern int		strexp(char*, int);
 extern long		streval(const char*, char**, long(*)(const char*, char**));
 extern long		strexpr(const char*, char**, long(*)(const char*, char**, void*), void*);
 extern int		strgid(const char*);
-extern int		strgrpmatch(const char*, const char*, int*, int, int);
+extern int		strgrpmatch(const char*, const char*, ssize_t*, int, int);
-extern int		strgrpmatch_20120528(const char*, const char*, ssize_t*, int, int);
+extern int		strngrpmatch(const char*, size_t, const char*, ssize_t*, int, int);
 extern unsigned int	strhash(const char*);
 extern void*		strlook(const void*, size_t, const char*);
 extern int		strmatch(const char*, const char*);

src/lib/libast/man/astsa.3

@@ -124,7 +124,7 @@ Returns the error message text corresponding to the
 \fIerrno\fP.
 .Ss "char* strerror(int \fIerrno\fP)"
 Equivalent to fmterror(\fIerrno\fP).
-.Ss "int strgrpmatch(const char* \fIstring\fP, const char* \fIpattern\fP, int* \fIsub\fP, int \fInsub\fP, int \fIflags\fP)"
+.Ss "int strgrpmatch(const char* \fIstring\fP, const char* \fIpattern\fP, ssize_t* \fIsub\fP, int \fInsub\fP, int \fIflags\fP)"
 Matches the null-terminated \fIstring\fP against the null-terminated
 .BR ksh (1)
 augmented \fIpattern\fP.
@@ -148,8 +148,6 @@ Ignore case.
 .Tp
 \f3STR_GROUP\fP:
 (|&) inside [@|*|+{n,m}](...) only.
-.Ss "int strmatch(const char* \fIstring\fP, const char* \fIpattern\fP, int* \fIsub\fP, int \fInsub\fP, int \fIflags\fP)"
-Equivalent to strgrpmatch(\fIstring\fP,\fIpattern\fP,0,0,STR_MAXIMAL|STR_LEFT|STR_RIGHT).
 .SH "SEE ALSO"
 .BR ast (3),
 .BR ccode (3),

src/lib/libast/string/strmatch.c

@@ -2,7 +2,7 @@
 *                                                                      *
 *               This software is part of the ast package               *
 *          Copyright (c) 1985-2012 AT&T Intellectual Property          *
-*          Copyright (c) 2020-2021 Contributors to ksh 93u+m           *
+*          Copyright (c) 2020-2022 Contributors to ksh 93u+m           *
 *                      and is licensed under the                       *
 *                 Eclipse Public License, Version 1.0                  *
 *                    by AT&T Intellectual Property                     *
@@ -64,8 +64,6 @@ static struct State_s
 	int		nmatch;
 } matchstate;
 
-#define STR_INT		040000
-
 /*
  * subgroup match
  * 0 returned if no match
@@ -77,7 +75,7 @@ static struct State_s
  */
 
 int
-strgrpmatch(const char* b, const char* p, ssize_t* sub, int n, register int flags)
+strngrpmatch(const char* b, size_t z, const char* p, ssize_t* sub, int n, register int flags)
 {
 	register regex_t*	re;
 	register ssize_t*	end;
@@ -140,7 +138,7 @@ strgrpmatch(const char* b, const char* p, ssize_t* sub, int n, register int flag
 			return 0;
 		matchstate.nmatch = n;
 	}
-	if (regexec(re, b, n, matchstate.match, reflags & ~(REG_MINIMAL|REG_SHELL_GROUP|REG_LEFT|REG_RIGHT|REG_ICASE)))
+	if (regnexec(re, b, z, n, matchstate.match, reflags & ~(REG_MINIMAL|REG_SHELL_GROUP|REG_LEFT|REG_RIGHT|REG_ICASE)))
 		return 0;
 	if (!sub || n <= 0)
 		return 1;
@@ -176,7 +174,7 @@ strgrpmatch(const char* b, const char* p, ssize_t* sub, int n, register int flag
 int
 strmatch(const char* s, const char* p)
 {
-	return strgrpmatch(s, p, NiL, 0, STR_MAXIMAL|STR_LEFT|STR_RIGHT);
+	return strngrpmatch(s, s ? strlen(s) : 0, p, NiL, 0, STR_MAXIMAL|STR_LEFT|STR_RIGHT);
 }
 
 /*
@@ -192,7 +190,7 @@ strsubmatch(const char* s, const char* p, int flags)
 {
 	ssize_t	match[2];
 
-	return strgrpmatch(s, p, match, 1, (flags ? STR_MAXIMAL : 0)|STR_LEFT) ? (char*)s + match[1] : (char*)0;
+	return strngrpmatch(s, s ? strlen(s) : 0, p, match, 1, (flags ? STR_MAXIMAL : 0)|STR_LEFT) ? (char*)s + match[1] : (char*)0;
 }
 
 #undef	strgrpmatch
@@ -201,7 +199,7 @@ strsubmatch(const char* s, const char* p, int flags)
 #endif
 
 int
-strgrpmatch(const char* b, const char* p, int* sub, int n, int flags)
+strgrpmatch(const char* b, const char* p, ssize_t* sub, int n, int flags)
 {
-	return strgrpmatch_20120528(b, p, (ssize_t*)sub, n, flags|STR_INT);
+	return strngrpmatch(b, b ? strlen(b) : 0, p, sub, n, flags|STR_INT);
 }