Add support for multibyte characters.

1 files changed, 112 insertions, 49 deletions

diff --git a/lib/libc/gen/fnmatch.c b/lib/libc/gen/fnmatch.c
index a07a2d0..3c3377f 100644
--- a/lib/libc/gen/fnmatch.c
+++ b/lib/libc/gen/fnmatch.c
@@ -45,10 +45,22 @@ __FBSDID("$FreeBSD$");
  * Compares a filename or pathname to a pattern.
  */
 
-#include <ctype.h>
+/*
+ * Some notes on multibyte character support:
+ * 1. Patterns with illegal byte sequences match nothing.
+ * 2. Illegal byte sequences in the "string" argument are handled by treating
+ *    them as single-byte characters with a value of the first byte of the
+ *    sequence cast to wchar_t.
+ * 3. Multibyte conversion state objects (mbstate_t) are passed around and
+ *    used for most, but not all, conversions. Further work will be required
+ *    to support state-dependent encodings.
+ */
+
 #include <fnmatch.h>
+#include <limits.h>
 #include <string.h>
-#include <stdio.h>
+#include <wchar.h>
+#include <wctype.h>
 
 #include "collate.h"
 
@@ -58,33 +70,57 @@ __FBSDID("$FreeBSD$");
 #define RANGE_NOMATCH   0
 #define RANGE_ERROR     (-1)
 
-static int rangematch(const char *, char, int, char **);
+static int rangematch(const char *, wchar_t, int, char **, mbstate_t *);
+static int fnmatch1(const char *, const char *, int, mbstate_t, mbstate_t);
 
 int
 fnmatch(pattern, string, flags)
 	const char *pattern, *string;
 	int flags;
 {
+	static const mbstate_t initial;
+
+	return (fnmatch1(pattern, string, flags, initial, initial));
+}
+
+static int
+fnmatch1(pattern, string, flags, patmbs, strmbs)
+	const char *pattern, *string;
+	int flags;
+	mbstate_t patmbs, strmbs;
+{
 	const char *stringstart;
 	char *newp;
-	char c, test;
+	char c;
+	wchar_t pc, sc;
+	size_t pclen, sclen;
 
-	for (stringstart = string;;)
-		switch (c = *pattern++) {
+	for (stringstart = string;;) {
+		pclen = mbrtowc(&pc, pattern, MB_LEN_MAX, &patmbs);
+		if (pclen == (size_t)-1 || pclen == (size_t)-2)
+			return (FNM_NOMATCH);
+		pattern += pclen;
+		sclen = mbrtowc(&sc, string, MB_LEN_MAX, &strmbs);
+		if (sclen == (size_t)-1 || sclen == (size_t)-2) {
+			sc = (unsigned char)*string;
+			sclen = 1;
+			memset(&strmbs, 0, sizeof(strmbs));
+		}
+		switch (pc) {
 		case EOS:
-			if ((flags & FNM_LEADING_DIR) && *string == '/')
+			if ((flags & FNM_LEADING_DIR) && sc == '/')
 				return (0);
-			return (*string == EOS ? 0 : FNM_NOMATCH);
+			return (sc == EOS ? 0 : FNM_NOMATCH);
 		case '?':
-			if (*string == EOS)
+			if (sc == EOS)
 				return (FNM_NOMATCH);
-			if (*string == '/' && (flags & FNM_PATHNAME))
+			if (sc == '/' && (flags & FNM_PATHNAME))
 				return (FNM_NOMATCH);
-			if (*string == '.' && (flags & FNM_PERIOD) &&
+			if (sc == '.' && (flags & FNM_PERIOD) &&
 			    (string == stringstart ||
 			    ((flags & FNM_PATHNAME) && *(string - 1) == '/')))
 				return (FNM_NOMATCH);
-			++string;
+			string += sclen;
 			break;
 		case '*':
 			c = *pattern;
@@ -92,7 +128,7 @@ fnmatch(pattern, string, flags)
 			while (c == '*')
 				c = *++pattern;
 
-			if (*string == '.' && (flags & FNM_PERIOD) &&
+			if (sc == '.' && (flags & FNM_PERIOD) &&
 			    (string == stringstart ||
 			    ((flags & FNM_PATHNAME) && *(string - 1) == '/')))
 				return (FNM_NOMATCH);
@@ -112,25 +148,35 @@ fnmatch(pattern, string, flags)
 			}
 
 			/* General case, use recursion. */
-			while ((test = *string) != EOS) {
-				if (!fnmatch(pattern, string, flags & ~FNM_PERIOD))
+			while (sc != EOS) {
+				if (!fnmatch1(pattern, string,
+				    flags & ~FNM_PERIOD, patmbs, strmbs))
 					return (0);
-				if (test == '/' && flags & FNM_PATHNAME)
+				sclen = mbrtowc(&sc, string, MB_LEN_MAX,
+				    &strmbs);
+				if (sclen == (size_t)-1 ||
+				    sclen == (size_t)-2) {
+					sc = (unsigned char)*string;
+					sclen = 1;
+					memset(&strmbs, 0, sizeof(strmbs));
+				}
+				if (sc == '/' && flags & FNM_PATHNAME)
 					break;
-				++string;
+				string += sclen;
 			}
 			return (FNM_NOMATCH);
 		case '[':
-			if (*string == EOS)
+			if (sc == EOS)
 				return (FNM_NOMATCH);
-			if (*string == '/' && (flags & FNM_PATHNAME))
+			if (sc == '/' && (flags & FNM_PATHNAME))
 				return (FNM_NOMATCH);
-			if (*string == '.' && (flags & FNM_PERIOD) &&
+			if (sc == '.' && (flags & FNM_PERIOD) &&
 			    (string == stringstart ||
 			    ((flags & FNM_PATHNAME) && *(string - 1) == '/')))
 				return (FNM_NOMATCH);
 
-			switch (rangematch(pattern, *string, flags, &newp)) {
+			switch (rangematch(pattern, sc, flags, &newp,
+			    &patmbs)) {
 			case RANGE_ERROR:
 				goto norm;
 			case RANGE_MATCH:
@@ -139,41 +185,47 @@ fnmatch(pattern, string, flags)
 			case RANGE_NOMATCH:
 				return (FNM_NOMATCH);
 			}
-			++string;
+			string += sclen;
 			break;
 		case '\\':
 			if (!(flags & FNM_NOESCAPE)) {
-				if ((c = *pattern++) == EOS) {
-					c = '\\';
-					--pattern;
-				}
+				pclen = mbrtowc(&pc, pattern, MB_LEN_MAX,
+				    &patmbs);
+				if (pclen == (size_t)-1 || pclen == (size_t)-2)
+					return (FNM_NOMATCH);
+				if (pclen == 0)
+					pc = '\\';
+				pattern += pclen;
 			}
 			/* FALLTHROUGH */
 		default:
 		norm:
-			if (c == *string)
+			if (pc == sc)
 				;
 			else if ((flags & FNM_CASEFOLD) &&
-				 (tolower((unsigned char)c) ==
-				  tolower((unsigned char)*string)))
+				 (towlower(pc) == towlower(sc)))
 				;
 			else
 				return (FNM_NOMATCH);
-			string++;
+			string += sclen;
 			break;
 		}
+	}
 	/* NOTREACHED */
 }
 
 static int
-rangematch(pattern, test, flags, newp)
+rangematch(pattern, test, flags, newp, patmbs)
 	const char *pattern;
-	char test;
+	wchar_t test;
 	int flags;
 	char **newp;
+	mbstate_t *patmbs;
 {
 	int negate, ok;
-	char c, c2;
+	wchar_t c, c2;
+	size_t pclen;
+	const char *origpat;
 
 	/*
 	 * A bracket expression starting with an unquoted circumflex
@@ -186,7 +238,7 @@ rangematch(pattern, test, flags, newp)
 		++pattern;
 
 	if (flags & FNM_CASEFOLD)
-		test = tolower((unsigned char)test);
+		test = towlower(test);
 
 	/*
 	 * A right bracket shall lose its special meaning and represent
@@ -194,29 +246,40 @@ rangematch(pattern, test, flags, newp)
 	 * -- POSIX.2 2.8.3.2
 	 */
 	ok = 0;
-	c = *pattern++;
-	do {
-		if (c == '\\' && !(flags & FNM_NOESCAPE))
-			c = *pattern++;
-		if (c == EOS)
+	origpat = pattern;
+	for (;;) {
+		if (*pattern == ']' && pattern > origpat) {
+			pattern++;
+			break;
+		} else if (*pattern == '\0') {
 			return (RANGE_ERROR);
-
-		if (c == '/' && (flags & FNM_PATHNAME))
+		} else if (*pattern == '/' && (flags & FNM_PATHNAME)) {
+			pattern++;
+			return (RANGE_NOMATCH);
+		} else if (*pattern == '\\' && !(flags & FNM_NOESCAPE))
+			pattern++;
+		pclen = mbrtowc(&c, pattern, MB_LEN_MAX, patmbs);
+		if (pclen == (size_t)-1 || pclen == (size_t)-2)
 			return (RANGE_NOMATCH);
+		pattern += pclen;
 
 		if (flags & FNM_CASEFOLD)
-			c = tolower((unsigned char)c);
+			c = towlower(c);
 
-		if (*pattern == '-'
-		    && (c2 = *(pattern+1)) != EOS && c2 != ']') {
-			pattern += 2;
-			if (c2 == '\\' && !(flags & FNM_NOESCAPE))
-				c2 = *pattern++;
+		if (*pattern == '-' && *(pattern + 1) != EOS &&
+		    *(pattern + 1) != ']') {
+			if (*++pattern == '\\' && !(flags & FNM_NOESCAPE))
+				if (*pattern != EOS)
+					pattern++;
+			pclen = mbrtowc(&c2, pattern, MB_LEN_MAX, patmbs);
+			if (pclen == (size_t)-1 || pclen == (size_t)-2)
+				return (RANGE_NOMATCH);
+			pattern += pclen;
 			if (c2 == EOS)
 				return (RANGE_ERROR);
 
 			if (flags & FNM_CASEFOLD)
-				c2 = tolower((unsigned char)c2);
+				c2 = towlower(c2);
 
 			if (__collate_load_error ?
 			    c <= test && test <= c2 :
@@ -226,7 +289,7 @@ rangematch(pattern, test, flags, newp)
 				ok = 1;
 		} else if (c == test)
 			ok = 1;
-	} while ((c = *pattern++) != ']');
+	}
 
 	*newp = (char *)pattern;
 	return (ok == negate ? RANGE_NOMATCH : RANGE_MATCH);