POSIX character class support for fnmatch(3) and glob(3). OK deraadt@

3 files changed, 149 insertions, 8 deletions

diff --git a/lib/libc/gen/charclass.h b/lib/libc/gen/charclass.h
new file mode 100644
index 00000000000..5edb2c1144c
--- /dev/null
+++ b/lib/libc/gen/charclass.h
@@ -0,0 +1,29 @@
+/*
+ * Public domain, 2008, Todd C. Miller <Todd.Miller@courtesan.com>
+ *
+ * $OpenBSD: charclass.h,v 1.1 2008/10/01 23:04:13 millert Exp $
+ */
+
+/*
+ * POSIX character class support for fnmatch() and glob().
+ */
+static struct cclass {
+	const char *name;
+	int (*isctype)(int);
+} cclasses[] = {
+	{ "alnum",	isalnum },
+	{ "alpha",	isalpha },
+	{ "blank",	isblank },
+	{ "cntrl",	iscntrl },
+	{ "digit",	isdigit },
+	{ "graph",	isgraph },
+	{ "lower",	islower },
+	{ "print",	isprint },
+	{ "punct",	ispunct },
+	{ "space",	isspace },
+	{ "upper",	isupper },
+	{ "xdigit",	isxdigit },
+	{ NULL,		NULL }
+};
+
+#define NCCLASSES	(sizeof(cclasses) / sizeof(cclasses[0]) - 1)
diff --git a/lib/libc/gen/fnmatch.c b/lib/libc/gen/fnmatch.c
index 5a282f1e596..2388f826382 100644
--- a/lib/libc/gen/fnmatch.c
+++ b/lib/libc/gen/fnmatch.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: fnmatch.c,v 1.13 2006/03/31 05:34:14 deraadt Exp $	*/
+/*	$OpenBSD: fnmatch.c,v 1.14 2008/10/01 23:04:13 millert Exp $	*/
 
 /*
  * Copyright (c) 1989, 1993, 1994
@@ -42,6 +42,8 @@
 #include <string.h>
 #include <fnmatch.h>
 
+#include "charclass.h"
+
 #define	EOS	'\0'
 
 #define	RANGE_MATCH	1
@@ -49,6 +51,7 @@
 #define	RANGE_ERROR	(-1)
 
 static int rangematch(const char *, char, int, char **);
+static int classmatch(const char *, char, int, const char **);
 
 int
 fnmatch(const char *pattern, const char *string, int flags)
@@ -153,7 +156,7 @@ fnmatch(const char *pattern, const char *string, int flags)
 static int
 rangematch(const char *pattern, char test, int flags, char **newp)
 {
-	int negate, ok;
+	int negate, ok, rv;
 	char c, c2;
 
 	/*
@@ -177,6 +180,17 @@ rangematch(const char *pattern, char test, int flags, char **newp)
 	ok = 0;
 	c = *pattern++;
 	do {
+		if (c == '[' && *pattern == ':') {
+			do {
+				rv = classmatch(pattern + 1, test,
+				    (flags & FNM_CASEFOLD), &pattern);
+				if (rv == RANGE_MATCH)
+					ok = 1;
+				c = *pattern++;
+			} while (rv != RANGE_ERROR && c == '[' && *pattern == ':');
+			if (c == ']')
+				break;
+		}
 		if (c == '\\' && !(flags & FNM_NOESCAPE))
 			c = *pattern++;
 		if (c == EOS)
@@ -203,3 +217,35 @@ rangematch(const char *pattern, char test, int flags, char **newp)
 	*newp = (char *)pattern;
 	return (ok == negate ? RANGE_NOMATCH : RANGE_MATCH);
 }
+
+static int
+classmatch(const char *pattern, char test, int foldcase, const char **ep)
+{
+	struct cclass *cc;
+	const char *colon;
+	size_t len;
+	int rval = RANGE_NOMATCH;
+
+	if ((colon = strchr(pattern, ':')) == NULL || colon[1] != ']') {
+		*ep = pattern - 2;
+		return(RANGE_ERROR);
+	}
+	*ep = colon + 2;
+	len = (size_t)(colon - pattern);
+
+	if (foldcase && strncmp(pattern, "upper:]", 7) == 0)
+		pattern = "lower:]";
+	for (cc = cclasses; cc->name != NULL; cc++) {
+		if (!strncmp(pattern, cc->name, len) && cc->name[len] == '\0') {
+			if (cc->isctype((unsigned char)test))
+				rval = RANGE_MATCH;
+			break;
+		}
+	}
+	if (cc->name == NULL) {
+		/* invalid character class, return EOS */
+		*ep = colon + strlen(colon);
+		rval = RANGE_ERROR;
+	}
+	return(rval);
+}
diff --git a/lib/libc/gen/glob.c b/lib/libc/gen/glob.c
index 67b534f3b2c..59c26629659 100644
--- a/lib/libc/gen/glob.c
+++ b/lib/libc/gen/glob.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: glob.c,v 1.26 2005/11/28 17:50:12 deraadt Exp $ */
+/*	$OpenBSD: glob.c,v 1.27 2008/10/01 23:04:13 millert Exp $ */
 /*
  * Copyright (c) 1989, 1993
  *	The Regents of the University of California.  All rights reserved.
@@ -69,6 +69,8 @@
 #include <string.h>
 #include <unistd.h>
 
+#include "charclass.h"
+
 #define	DOLLAR		'$'
 #define	DOT		'.'
 #define	EOS		'\0'
@@ -116,6 +118,7 @@ typedef char Char;
 #define	M_ONE		META('?')
 #define	M_RNG		META('-')
 #define	M_SET		META('[')
+#define	M_CLASS		META(':')
 #define	ismeta(c)	(((c)&M_QUOTE) != 0)
 
 
@@ -123,7 +126,8 @@ static int	 compare(const void *, const void *);
 static int	 g_Ctoc(const Char *, char *, u_int);
 static int	 g_lstat(Char *, struct stat *, glob_t *);
 static DIR	*g_opendir(Char *, glob_t *);
-static Char	*g_strchr(Char *, int);
+static Char	*g_strchr(const Char *, int);
+static int	 g_strncmp(const Char *, const char *, size_t);
 static int	 g_stat(Char *, struct stat *, glob_t *);
 static int	 glob0(const Char *, glob_t *);
 static int	 glob1(Char *, Char *, glob_t *, size_t *);
@@ -200,7 +204,7 @@ globexp1(const Char *pattern, glob_t *pglob)
 	if (pattern[0] == LBRACE && pattern[1] == RBRACE && pattern[2] == EOS)
 		return glob0(pattern, pglob);
 
-	while ((ptr = (const Char *) g_strchr((Char *) ptr, LBRACE)) != NULL)
+	while ((ptr = (const Char *) g_strchr(ptr, LBRACE)) != NULL)
 		if (!globexp2(ptr, pattern, pglob, &rv))
 			return rv;
 
@@ -375,6 +379,47 @@ globtilde(const Char *pattern, Char *patbuf, size_t patbuf_len, glob_t *pglob)
 	return patbuf;
 }
 
+static int
+g_strncmp(const Char *s1, const char *s2, size_t n)
+{
+	int rv = 0;
+
+	while (n--) {
+		rv = *(Char *)s1 - *(const unsigned char *)s2++;
+		if (rv)
+			break;
+		if (*s1++ == '\0')
+			break;
+	}
+	return rv;
+}
+
+static int
+g_charclass(const Char **patternp, Char **bufnextp)
+{
+	const Char *pattern = *patternp + 1;
+	Char *bufnext = *bufnextp;
+	const Char *colon;
+	struct cclass *cc;
+	size_t len;
+
+	if ((colon = g_strchr(pattern, ':')) == NULL || colon[1] != ']')
+		return 1;	/* not a character class */
+
+	len = (size_t)(colon - pattern);
+	for (cc = cclasses; cc->name != NULL; cc++) {
+		if (!g_strncmp(pattern, cc->name, len) && cc->name[len] == '\0')
+			break;
+	}
+	if (cc->name == NULL)
+		return -1;	/* invalid character class */
+	*bufnext++ = M_CLASS;
+	*bufnext++ = (Char)(cc - &cclasses[0]);
+	*bufnextp = bufnext;
+	*patternp += len + 3;
+
+	return 0;
+}
 
 /*
  * The main glob() routine: compiles the pattern (optionally processing
@@ -403,7 +448,7 @@ glob0(const Char *pattern, glob_t *pglob)
 			if (c == NOT)
 				++qpatnext;
 			if (*qpatnext == EOS ||
-			    g_strchr((Char *) qpatnext+1, RBRACKET) == NULL) {
+			    g_strchr(qpatnext+1, RBRACKET) == NULL) {
 				*bufnext++ = LBRACKET;
 				if (c == NOT)
 					--qpatnext;
@@ -414,6 +459,20 @@ glob0(const Char *pattern, glob_t *pglob)
 				*bufnext++ = M_NOT;
 			c = *qpatnext++;
 			do {
+				if (c == LBRACKET && *qpatnext == ':') {
+					do {
+						err = g_charclass(&qpatnext,
+						    &bufnext);
+						if (err)
+							break;
+						c = *qpatnext++;
+					} while (c == LBRACKET && *qpatnext == ':');
+					if (err == -1 &&
+					    !(pglob->gl_flags & GLOB_NOCHECK))
+						return GLOB_NOMATCH;
+					if (c == RBRACKET)
+						break;
+				}
 				*bufnext++ = CHAR(c);
 				if (*qpatnext == RANGE &&
 				    (c = qpatnext[1]) != RBRACKET) {
@@ -728,6 +787,13 @@ match(Char *name, Char *pat, Char *patend)
 			if ((negate_range = ((*pat & M_MASK) == M_NOT)) != EOS)
 				++pat;
 			while (((c = *pat++) & M_MASK) != M_END)
+				if ((c & M_MASK) == M_CLASS) {
+					int idx = *pat & M_MASK;
+					if (idx < NCCLASSES &&
+					    cclasses[idx].isctype(k))
+						ok = 1;
+					++pat;
+				}
 				if ((*pat & M_MASK) == M_RNG) {
 					if (c <= k && k <= pat[1])
 						ok = 1;
@@ -806,11 +872,11 @@ g_stat(Char *fn, struct stat *sb, glob_t *pglob)
 }
 
 static Char *
-g_strchr(Char *str, int ch)
+g_strchr(const Char *str, int ch)
 {
 	do {
 		if (*str == ch)
-			return (str);
+			return ((Char *)str);
 	} while (*str++);
 	return (NULL);
 }