reimplement has_globbing() with proper bracket expression parsing,

and take ahead parsing collating symbols, equivalence classes and
character classes already (heck my first draft of this already did
better than GNU bash, ksh93 I still don’t grok its code at all)

check.t

@@ -1,4 +1,4 @@
-# $MirOS: src/bin/mksh/check.t,v 1.785 2017/04/29 15:18:25 tg Exp $
+# $MirOS: src/bin/mksh/check.t,v 1.786 2017/04/29 21:49:04 tg Exp $
 # -*- mode: sh -*-
 #-
 # Copyright © 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
@@ -1962,15 +1962,11 @@ expected-stdout:
 name: eglob-bad-1
 description:
 	Check that globbing isn't done when glob has syntax error
-file-setup: file 644 "abcx"
+file-setup: file 644 "@(a[b|)c]foo"
-file-setup: file 644 "abcz"
-file-setup: file 644 "bbc"
 stdin:
-	echo !([*)*
+	echo @(a[b|)c]*
-	echo +(a|b[)*
 expected-stdout:
-	!([*)*
+	@(a[b|)c]*
-	+(a|b[)*
 ---
 name: eglob-bad-2
 description:
@@ -2057,9 +2053,11 @@ stdin:
 	case foo in *(a|b[)) echo yes;; *) echo no;; esac
 	case foo in *(a|b[)|f*) echo yes;; *) echo no;; esac
 	case '*(a|b[)' in *(a|b[)) echo yes;; *) echo no;; esac
+	case 'aab[b[ab[a' in *(a|b[)) echo yes;; *) echo no;; esac
 expected-stdout:
 	no
 	yes
+	no
 	yes
 ---
 name: eglob-trim-1
@@ -2354,7 +2352,7 @@ expected-stdout:
 ---
 name: glob-bad-1
 description:
-	Check that globbing isn't done when glob has syntax error
+	Check that [ matches itself if it's not a valid bracket expr
 file-setup: dir 755 "[x"
 file-setup: file 644 "[x/foo"
 stdin:
@@ -2362,8 +2360,8 @@ stdin:
 	echo *[x
 	echo [x/*
 expected-stdout:
-	[*
+	[x
-	*[x
+	[x
 	[x/foo
 ---
 name: glob-bad-2
@@ -2403,24 +2401,31 @@ file-setup: file 644 "abc"
 file-setup: file 644 "bbc"
 file-setup: file 644 "cbc"
 file-setup: file 644 "-bc"
+file-setup: file 644 "!bc"
+file-setup: file 644 "^bc"
+file-setup: file 644 "+bc"
+file-setup: file 644 ",bc"
+file-setup: file 644 "0bc"
+file-setup: file 644 "1bc"
 stdin:
 	echo [ab-]*
 	echo [-ab]*
 	echo [!-ab]*
 	echo [!ab]*
 	echo []ab]*
-	:>'./!bc'
-	:>'./^bc'
 	echo [^ab]*
-	echo [!ab]*
+	echo [+--]*
+	echo [--1]*
+
 expected-stdout:
 	-bc abc bbc
 	-bc abc bbc
-	cbc
+	!bc +bc ,bc 0bc 1bc ^bc cbc
-	-bc cbc
+	!bc +bc ,bc -bc 0bc 1bc ^bc cbc
 	abc bbc
 	^bc abc bbc
-	!bc -bc ^bc cbc
+	+bc ,bc -bc
+	-bc 0bc 1bc
 ---
 name: glob-range-2
 description:

eval.c

@@ -23,7 +23,7 @@
 
 #include "sh.h"
 
-__RCSID("$MirOS: src/bin/mksh/eval.c,v 1.207 2017/04/28 00:38:29 tg Exp $");
+__RCSID("$MirOS: src/bin/mksh/eval.c,v 1.208 2017/04/29 21:49:06 tg Exp $");
 
 /*
  * string expansion
@@ -1666,7 +1666,7 @@ globit(XString *xs,	/* dest string */
 	 * directory isn't readable - if no globbing is needed, only execute
 	 * permission should be required (as per POSIX)).
 	 */
-	if (!has_globbing(sp, se)) {
+	if (!has_globbing(sp)) {
 		XcheckN(*xs, xp, se - sp + 1);
 		debunk(xp, sp, Xnleft(*xs, xp));
 		xp += strlen(xp);

misc.c

@@ -32,7 +32,7 @@
 #include <grp.h>
 #endif
 
-__RCSID("$MirOS: src/bin/mksh/misc.c,v 1.270 2017/04/28 21:43:30 tg Exp $");
+__RCSID("$MirOS: src/bin/mksh/misc.c,v 1.271 2017/04/29 21:49:07 tg Exp $");
 
 #define KSH_CHVT_FLAG
 #ifdef MKSH_SMALL
@@ -634,7 +634,7 @@ gmatchx(const char *s, const char *p, bool isfile)
 	 * isfile is false iff no syntax check has been done on
 	 * the pattern. If check fails, just do a strcmp().
 	 */
-	if (!isfile && !has_globbing(p, pe)) {
+	if (!isfile && !has_globbing(p)) {
 		size_t len = pe - p + 1;
 		char tbuf[64];
 		char *t = len <= sizeof(tbuf) ? tbuf : alloc(len, ATEMP);
@@ -661,7 +661,7 @@ gmatchx(const char *s, const char *p, bool isfile)
  * Syntax errors are:
  *	- [ with no closing ]
  *	- imbalanced $(...) expression
- *	- [...] and *(...) not nested (eg, [a$(b|]c), *(a[b|c]d))
+ *	- [...] and *(...) not nested (eg, @(a[b|)c], *(a[b|c]d))
  */
 /*XXX
  * - if no magic,
@@ -672,60 +672,80 @@ gmatchx(const char *s, const char *p, bool isfile)
  *	return ?
  * - return ?
  */
-int
+bool
-has_globbing(const char *xp, const char *xpe)
+has_globbing(const char *pat)
 {
-	const unsigned char *p = (const unsigned char *) xp;
+	unsigned char c, subc;
-	const unsigned char *pe = (const unsigned char *) xpe;
-	int c;
-	int nest = 0, bnest = 0;
 	bool saw_glob = false;
-	/* inside [...] */
+	unsigned int nest = 0;
-	bool in_bracket = false;
+	const unsigned char *p = (const unsigned char *)pat;
+	const unsigned char *s;
 
-	for (; p < pe; p++) {
+	while ((c = *p++)) {
-		if (!ISMAGIC(*p))
+		/* regular character? ok. */
+		if (!ISMAGIC(c))
 			continue;
-		if ((c = *++p) == '*' || c == '?')
+		/* MAGIC + NUL? abort. */
+		if (!(c = *p++))
+			return (false);
+		/* some specials */
+		if (c == '*' || c == '?') {
+			/* easy glob, accept */
 			saw_glob = true;
-		else if (c == '[') {
+		} else if (c == '[') {
-			if (!in_bracket) {
+			/* bracket expression; eat negation and initial ] */
+			if (ISMAGIC(p[0]) && p[1] == '!')
+				p += 2;
+			if (ISMAGIC(p[0]) && p[1] == ']')
+				p += 2;
+			/* check next string part */
+			s = p;
+			while ((c = *s++)) {
+				/* regular chars are ok */
+				if (!ISMAGIC(c))
+					continue;
+				/* MAGIC + NUL cannot happen */
+				if (!(c = *s++))
+					return (false);
+				/* terminating bracket? */
+				if (c == ']') {
+					/* accept and continue */
+					p = s;
 					saw_glob = true;
-				in_bracket = true;
+					break;
-				if (ISMAGIC(p[1]) && p[2] == '!')
-					p += 2;
-				if (ISMAGIC(p[1]) && p[2] == ']')
-					p += 2;
 				}
-			/*XXX Do we need to check ranges here? POSIX Q */
+				/* collating, equivalence or character class */
-		} else if (c == ']') {
+				if (c == '[' && (
-			if (in_bracket) {
+				    *s == '.' || *s == '=' || *s == ':')) {
-				if (bnest)
+					/* must stop with exactly the same c */
-					/* [a*(b]) */
+					subc = *s++;
-					return (0);
+					/* arbitrarily many chars in betwixt */
-				in_bracket = false;
+					while ((c = *s++))
+						/* but only this sequence... */
+						if (c == subc && ISMAGIC(*s) &&
+						    s[1] == ']') {
+							/* accept, terminate */
+							s += 2;
+							break;
+						}
+					/* EOS without: reject bracket expr */
+					if (!c)
+						break;
+					/* continue; */
+				}
+				/* anything else just goes on */
 			}
 		} else if ((c & 0x80) && ctype(c & 0x7F, C_PATMO | C_SPC)) {
+			/* opening pattern */
 			saw_glob = true;
-			if (in_bracket)
+			++nest;
-				bnest++;
+		} else if (c == /*(*/')') {
-			else
+			/* closing pattern */
-				nest++;
+			if (nest)
-		} else if (c == '|') {
+				--nest;
-			if (in_bracket && !bnest)
-				/* *(a[foo|bar]) */
-				return (0);
-		} else if (c == /*(*/ ')') {
-			if (in_bracket) {
-				if (!bnest--)
-					/* *(a[b)c] */
-					return (0);
-			} else if (nest)
-				nest--;
 		}
-		/* else must be MAGIC followed by MAGIC or one of: ]{},!- */
 	}
-	return (saw_glob && !in_bracket && !nest);
+	return (saw_glob && !nest);
 }
 
 /* Function must return either 0 or 1 (assumed by code for 0x80|'!') */
@@ -774,7 +794,7 @@ do_gmatch(const unsigned char *s, const unsigned char *se,
 			return (0);
 
 		/**
-		 * [*+?@!](pattern|pattern|..)
+		 * [+*?@!](pattern|pattern|..)
 		 * This is also needed for ${..%..}, etc.
 		 */
 

sh.h

@@ -175,7 +175,7 @@
 #endif
 
 #ifdef EXTERN
-__RCSID("$MirOS: src/bin/mksh/sh.h,v 1.832 2017/04/29 14:20:24 tg Exp $");
+__RCSID("$MirOS: src/bin/mksh/sh.h,v 1.833 2017/04/29 21:49:07 tg Exp $");
 #endif
 #define MKSH_VERSION "R55 2017/04/27"
 
@@ -2528,7 +2528,7 @@ void change_xtrace(unsigned char, bool);
 int parse_args(const char **, int, bool *);
 int getn(const char *, int *);
 int gmatchx(const char *, const char *, bool);
-int has_globbing(const char *, const char *) MKSH_A_PURE;
+bool has_globbing(const char *) MKSH_A_PURE;
 int ascstrcmp(const void *, const void *) MKSH_A_PURE;
 int ascpstrcmp(const void *, const void *) MKSH_A_PURE;
 void ksh_getopt_reset(Getopt *, int);