commit my WIP for the Beltane Snapshot of the Mainframe Korn Shell, not going to make finishing it tonight ☹

check.t

@@ -1,4 +1,4 @@
-# $MirOS: src/bin/mksh/check.t,v 1.786 2017/04/29 21:49:04 tg Exp $
+# $MirOS: src/bin/mksh/check.t,v 1.787 2017/05/01 19:44:26 tg Exp $
 # -*- mode: sh -*-
 #-
 # Copyright © 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
@@ -30,58 +30,62 @@
 # (2013/12/02 20:39:44) http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/regress/bin/ksh/?sortby=date
 
 expected-stdout:
-	@(#)MIRBSD KSH R55 2017/04/27
+	@(#)MIRBSD KSH R55 2017/05/01
 description:
-	Check version of shell.
+	Check base version of full shell
 stdin:
-	echo $KSH_VERSION
+	echo ${KSH_VERSION%%' +'*}
 name: KSH_VERSION
-category: !shell:legacy-yes,!shell:textmode-yes,!shell:ebcdic-yes
+category: !shell:legacy-yes
 ---
 expected-stdout:
-	@(#)LEGACY KSH R55 2017/04/27
+	@(#)LEGACY KSH R55 2017/05/01
 description:
-	Check version of legacy shell.
+	Check base version of legacy shell
 stdin:
-	echo $KSH_VERSION
+	echo ${KSH_VERSION%%' +'*}
 name: KSH_VERSION-legacy
-category: !shell:legacy-no,!shell:textmode-yes,!shell:ebcdic-yes
+category: !shell:legacy-no
 ---
-expected-stdout:
+name: KSH_VERSION-ascii
-	@(#)MIRBSD KSH R55 2017/04/27 +EBCDIC
 description:
-	Check version of shell.
+	Check that the shell version tag does not include EBCDIC
+category: !shell:ebcdic-yes
 stdin:
-	echo $KSH_VERSION
+	for x in $KSH_VERSION; do
+		[[ $x = '+EBCDIC' ]] && exit 1
+	done
+	exit 0
+---
 name: KSH_VERSION-ebcdic
-category: !shell:legacy-yes,!shell:textmode-yes,!shell:ebcdic-no
----
-expected-stdout:
-	@(#)LEGACY KSH R55 2017/04/27 +EBCDIC
 description:
-	Check version of legacy shell.
+	Check that the shell version tag includes EBCDIC
+category: !shell:ebcdic-no
 stdin:
-	echo $KSH_VERSION
+	for x in $KSH_VERSION; do
-name: KSH_VERSION-legacy-ebcdic
+		[[ $x = '+EBCDIC' ]] && exit 0
-category: !shell:legacy-no,!shell:textmode-yes,!shell:ebcdic-no
+	done
+	exit 1
 ---
-expected-stdout:
+name: KSH_VERSION-binmode
-	@(#)MIRBSD KSH R55 2017/04/27 +TEXTMODE
 description:
-	Check version of shell.
+	Check that the shell version tag does not include TEXTMODE
+category: !shell:textmode-yes
 stdin:
-	echo $KSH_VERSION
+	for x in $KSH_VERSION; do
+		[[ $x = '+TEXTMODE' ]] && exit 1
+	done
+	exit 0
+---
 name: KSH_VERSION-textmode
-category: !shell:legacy-yes,!shell:textmode-no
----
-expected-stdout:
-	@(#)LEGACY KSH R55 2017/04/27 +TEXTMODE
 description:
-	Check version of legacy shell.
+	Check that the shell version tag includes TEXTMODE
+category: !shell:textmode-no
 stdin:
-	echo $KSH_VERSION
+	for x in $KSH_VERSION; do
-name: KSH_VERSION-legacy-textmode
+		[[ $x = '+TEXTMODE' ]] && exit 0
-category: !shell:legacy-no,!shell:textmode-no
+	done
+	exit 1
 ---
 name: selftest-1
 description:
@@ -2353,16 +2357,25 @@ expected-stdout:
 name: glob-bad-1
 description:
 	Check that [ matches itself if it's not a valid bracket expr
+	but does not prevent globbing, while backslash-escaping does
 file-setup: dir 755 "[x"
 file-setup: file 644 "[x/foo"
 stdin:
 	echo [*
 	echo *[x
 	echo [x/*
+	:>'ab[x'
+	:>'a[a-z][x'
+	echo a[a-z][*
+	echo a[a-z]*
+	echo a[a\-z]*
 expected-stdout:
 	[x
 	[x
 	[x/foo
+	ab[x
+	ab[x
+	a[a-z]*
 ---
 name: glob-bad-2
 description:
@@ -2470,11 +2483,33 @@ file-setup: file 644 "cbc"
 file-setup: file 644 "dbc"
 file-setup: file 644 "ebc"
 file-setup: file 644 "-bc"
+file-setup: file 644 "@bc"
 stdin:
 	echo [a-c-e]*
+	#XXX TODO: echo [a--@]*
+	# -> @bc
 expected-stdout:
 	-bc abc bbc cbc ebc
 ---
+name: glob-word-1
+description:
+	Check BSD word boundary matches
+stdin:
+	t() { [[ $1 = *[[:\<:]]bar[[:\>:]]* ]]; echo =$?; }
+	t 'foo bar baz'
+	t 'foobar baz'
+	t 'foo barbaz'
+	t 'bar'
+	t '_bar'
+	t 'bar_'
+expected-stdout:
+	=0
+	=1
+	=1
+	=0
+	=1
+	=1
+---
 name: glob-trim-1
 description:
 	Check against a regression from fixing IFS-subst-2

misc.c

@@ -32,7 +32,7 @@
 #include <grp.h>
 #endif
 
-__RCSID("$MirOS: src/bin/mksh/misc.c,v 1.273 2017/05/01 15:59:45 tg Exp $");
+__RCSID("$MirOS: src/bin/mksh/misc.c,v 1.274 2017/05/01 19:44:29 tg Exp $");
 
 #define KSH_CHVT_FLAG
 #ifdef MKSH_SMALL
@@ -917,8 +917,84 @@ do_gmatch(const unsigned char *s, const unsigned char *se,
 	return (s == se);
 }
 
+/*XXX this is a prime example for bsearch or a const hashtable */
+static const struct cclass {
+	const char *name;
+	uint32_t value;
+} cclasses[] = {
+	/* POSIX */
+	{ "alnum",	C_ALNUM	},
+	{ "alpha",	C_ALPHA	},
+	{ "blank",	C_BLANK	},
+	{ "cntrl",	C_CNTRL	},
+	{ "digit",	C_DIGIT	},
+	{ "graph",	C_GRAPH	},
+	{ "lower",	C_LOWER	},
+	{ "print",	C_PRINT	},
+	{ "punct",	C_PUNCT	},
+	{ "space",	C_SPACE	},
+	{ "upper",	C_UPPER	},
+	{ "xdigit",	C_SEDEC	},
+	/* BSD */
+	/* "<" and ">" are handled inline */
+	/* GNU bash */
+	{ "ascii",	C_ASCII	},
+	{ "word",	C_ALNUX	},
+	/* mksh */
+	{ "sh_alias",	C_ALIAS	},
+	{ "sh_edq",	C_EDQ	},
+	{ "sh_ifs",	C_IFS	},
+	{ "sh_ifsws",	C_IFSWS	},
+	{ "sh_nl",	C_NL	},
+	{ "sh_quote",	C_QUOTE	},
+	/* sentinel */
+	{ NULL,		0	}
+};
+
 static const unsigned char *
 gmatch_cclass(const unsigned char *p, unsigned char sub)
+#if 0
+gmatch_cclass(const unsigned char *pat, unsigned char sc)
+{
+	unsigned char c, subc;
+	const unsigned char *p = pat, *s;
+	bool found = false;
+	bool negated = false;
+
+	/* check for negation */
+	if (ISMAGIC(p[0]) && ord(p[1]) == ord('!')) {
+		p += 2;
+		negated = true;
+	}
+	/* make initial ] non-MAGIC */
+	if (ISMAGIC(p[0]) && ord(p[1]) == ord(']'))
+		++p;
+	/* iterate over bracket expression, debunk()ing on the fly */
+	while ((c = *p++)) {
+		/* non-regular character? */
+		if (ISMAGIC(c)) {
+			/* MAGIC + NUL cannot happen */
+			if (!(c = *p++))
+				break;
+			/* terminating bracket? */
+			if (ord(c) == ord(']')) {
+				/* accept and return */
+				return (found != negated ? p : NULL);
+			}
+			/* sub-bracket expressions */
+			if (ord(c) == ord('[') && (
+			    /* collating element? */
+			    ord(*p) == ord('.') ||
+			    /* equivalence class? */
+			    ord(*p) == ord('=') ||
+			    /* character class? */
+			    ord(*p) == ord(':'))) {
+				/* must stop with exactly the same c */
+				subc = *p++;
+				/* save away start of substring */
+				s = p;
+}
+#endif
 {
 	unsigned char c, d;
 	bool notp, found = false;