PATCH: Clean up subscripting details

["Bart Schaefer" <schaefer@xxxxxxxxxxxxxxxxxxxxxxx>]

I discovered that ${A[foo]=bar} gave "not an identifier" because of the
tokenized [ ] that getindex() was leaving in its rewritten input string.
As the code now knows where the close bracket must be, it's possible to
untokenize again, and also optimize by skipping some loops that walk the
string looking for the end.  That makes up a little for the extra cost
incurred by parse_subscript().

I also discovered that testing isident() in bin_typeset() was not entirely
redundant; doing it it typeset_single() avoids an extra parse_subscript().

Then there's this:

	zsh-4.0.1-pre-2% noglob typeset a[3]=three
	zsh-4.0.1-pre-2% echo $#a
	3
	zsh-4.0.1-pre-2% echo X${^a}X
	XX XX XthreeX

This has appeared to work for some time now -- I don't know exactly how
far back, though it's the same in 3.0.[6-8] -- but has the side-effect of
creating a spurious parameter whose name is 'a[3]':

	zsh-4.0.1-pre-2% set | grep -w a
	a=('' '' baz)
	'a[3]'=''

There's no way to reference this parameter or give it a value.  Similarly:

	zsh-4.0.1-pre-2% typeset 12foo=bar
	zsh-4.0.1-pre-2% set | grep foo
	12foo=''
	zsh-4.0.1-pre-2% echo $#
	12

Another spurious parameter that you can't use, plus assignment to the
positional parameter $12.

The patch below improves isident() a bit further to require that when an
identifier begins with a digit, it must consist of nothing but digits.
It also makes the typeset slice-assignment explicitly legal, but avoids
creating the bogus extra parameter.  This means you can do

	func () {
	  noglob local a[1]=one a[2]=two a[3]=three
	  # same as: local -a a; a=(one two three)
	}

to create and assign to a local array.  Note, however, that you can't
change the type of a parameter this way -- if `a' is already a scalar in
the local scope, the above will insert substrings into the value rather
than create array elements.

Finally, the lex.c hunks fix this discrepancy:

	zsh-4.0.1-pre-2% ((\[))
	[: ']' expected

	% ((\[))
	zsh: bad output format specification

That is, one of my changes broke zsh's heuristic for determining whether
((...)) is a math expression or a subshell in a subshell; the patch fixes
that, restoring the 4.0.1-pre-2 interpretation.

Now I'm going to try to write some documentation.

diff -ru -x CVS common/Src/builtin.c zsh-4.0/Src/builtin.c
--- common/Src/builtin.c	Tue Apr 17 19:54:53 2001
+++ zsh-4.0/Src/builtin.c	Sat Apr 21 00:01:06 2001
@@ -1691,7 +1691,7 @@
 		pm->env = NULL;
 	    }
 	    if (value && !(pm = setsparam(pname, ztrdup(value))))
-		return 0;
+		return NULL;
 	} else if (value) {
 	    zwarnnam(cname, "can't assign new value for array %s", pname, 0);
 	    return NULL;
@@ -1782,7 +1782,27 @@
 	    pm->ct = auxlen;
 	else
 	    pm->ct = 0;
-    } else {
+    } else if (strchr(pname, '[')) {
+	if (on & PM_READONLY) {
+	    zerrnam(cname,
+		    "%s: can't create readonly array elements", pname, 0);
+	    return NULL;
+	} else if (PM_TYPE(on) == PM_SCALAR) {
+	    /*
+	     * This will either complain about bad identifiers, or will set
+	     * a hash element or array slice.  This once worked by accident,
+	     * creating a stray parameter along the way via createparam(),
+	     * now called below in the isident() branch.
+	     */
+	    if (!(pm = setsparam(pname, ztrdup(value ? value : ""))))
+		return NULL;
+	    value = NULL;
+	} else {
+	    zerrnam(cname,
+		    "%s: array elements must be scalar", pname, 0);
+	    return NULL;
+	}
+    } else if (isident(pname)) {
 	/*
 	 * Create a new node for a parameter with the flags in `on' minus the
 	 * readonly flag
@@ -1790,6 +1810,9 @@
 	pm = createparam(pname, on & ~PM_READONLY);
 	DPUTS(!pm, "BUG: parameter not created");
 	pm->ct = auxlen;
+    } else {
+	zerr("not an identifier: %s", pname, 0);
+	return NULL;
     }
 
     if (altpm && PM_TYPE(pm->flags) == PM_SCALAR) {
@@ -1808,8 +1831,14 @@
     else if (on & PM_LOCAL)
 	pm->level = locallevel;
     if (value && !(pm->flags & (PM_ARRAY|PM_HASHED))) {
+	Param ipm = pm;
 	if (!(pm = setsparam(pname, ztrdup(value))))
-	    return 0;
+	    return NULL;
+	if (pm != ipm) {
+	    DPUTS(ipm->flags != pm->flags,
+		  "BUG: parameter recreated with wrong flags");
+	    unsetparam_pm(ipm, 0, 1);
+	}
     } else if (newspecial && !(pm->old->flags & PM_NORESTORE)) {
 	/*
 	 * We need to use the special setting function to re-initialise
diff -ru -x CVS common/Src/lex.c zsh-4.0/Src/lex.c
--- common/Src/lex.c	Thu Apr 19 22:11:32 2001
+++ zsh-4.0/Src/lex.c	Fri Apr 20 22:06:39 2001
@@ -1303,9 +1303,9 @@
 	    if (c != '\n') {
 		if (c == '$' || c == '\\' || (c == '}' && !intick && bct) ||
 		    c == endchar || c == '`' ||
-		    (math && (c == '[' || c == ']' ||
-			      c == '(' || c == ')' ||
-			      c == '{' || c == '}')))
+		    (endchar == ']' && (c == '[' || c == ']' ||
+					c == '(' || c == ')' ||
+					c == '{' || c == '}')))
 		    add(Bnull);
 		else {
 		    /* lexstop is implicitly handled here */
@@ -1390,7 +1390,7 @@
 		err = (!brct-- && math);
 	    break;
 	case '"':
-	    if (intick || (!endchar && !bct))
+	    if (intick || endchar == ']' || (!endchar && !bct))
 		break;
 	    if (bct) {
 		add(Dnull);
diff -ru -x CVS common/Src/params.c zsh-4.0/Src/params.c
--- common/Src/params.c	Thu Apr 19 22:11:32 2001
+++ zsh-4.0/Src/params.c	Sat Apr 21 00:37:09 2001
@@ -765,10 +765,17 @@
     if (!*s)			/* empty string is definitely not valid */
 	return 0;
 
-    /* find the first character in `s' not in the iident type table */
-    for (ss = s; *ss; ss++)
-	if (!iident(*ss))
-	    break;
+    if (idigit(*s)) {
+	/* If the first character is `s' is a digit, then all must be */
+	for (ss = ++s; *ss; ss++)
+	    if (!idigit(*ss))
+		break;
+    } else {
+	/* Find the first character in `s' not in the iident type table */
+	for (ss = s; *ss; ss++)
+	    if (!iident(*ss))
+		break;
+    }
 
     /* If the next character is not [, then it is *
      * definitely not a valid identifier.         */
@@ -1171,7 +1178,7 @@
     int start, end, inv = 0;
     char *s = *pptr, *tbrack;
 
-    *s++ = Inbrack;
+    *s++ = '[';
     s = parse_subscript(s);	/* Error handled after untokenizing */
     /* Now we untokenize everthing except INULL() markers so we can check *
      * for the '*' and '@' special subscripts.  The INULL()s are removed  *
@@ -1191,7 +1198,7 @@
 	return 1;
     }
     s = *pptr + 1;
-    if ((s[0] == '*' || s[0] == '@') && s[1] == Outbrack) {
+    if ((s[0] == '*' || s[0] == '@') && s + 1 == tbrack) {
 	if ((v->isarr || IS_UNSET_VALUE(v)) && s[0] == '@')
 	    v->isarr |= SCANPM_ISVAR_AT;
 	v->start = 0;
@@ -1223,12 +1230,11 @@
 	    }
 	    if (*s == ',') {
 		zerr("invalid subscript", NULL, 0);
-		while (*s && *s != Outbrack)
-		    s++;
-		*pptr = s;
+		*tbrack = ']';
+		*pptr = tbrack+1;
 		return 1;
 	    }
-	    if (*s == Outbrack)
+	    if (s == tbrack)
 		s++;
 	} else {
 	    int com;
@@ -1243,7 +1249,7 @@
 		start--;
 	    else if (start == 0 && end == 0)
 		end++;
-	    if (*s == Outbrack) {
+	    if (s == tbrack) {
 		s++;
 		if (v->isarr && start == end-1 && !com &&
 		    (!(v->isarr & SCANPM_MATCHMANY) ||
@@ -1256,6 +1262,7 @@
 		s = *pptr;
 	}
     }
+    *tbrack = ']';
     *pptr = s;
     return 0;
 }
diff -ru -x CVS common/Test/D06subscript.ztst zsh-4.0/Test/D06subscript.ztst
--- common/Test/D06subscript.ztst	Thu Apr 19 22:11:32 2001
+++ zsh-4.0/Test/D06subscript.ztst	Sat Apr 21 09:44:36 2001
@@ -128,3 +128,11 @@
 >obrack obrack
 >] ]
 >backcbrack
+
+  print -R ${A[${A[(r)\\\\\\\\\]]}]::=zounds}
+  print -R ${A[${A[(r)\\\\\\\\\]]}]}
+  print -R $A[\\\\\]]
+0:Associative array substitution-assignment with reverse pattern subscript key
+>zounds
+>zounds
+>zounds

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net