Changeset 18

Timestamp:

09/16/07 4:04:53 PM (10 months ago)

Author:

aaron

Message:

• added test cases and target for running tests
  
• modified ICUMatcher performReplacementWithString to use realloc when the
  buffer size needs to be increased instead of freeing and allocing a new buffer
  
• chagned memory alloc/free from C functions to NSMallocZone and friends
  
• made ints from [NSString length] unsigned
  
• added comments for NSStringICUAdditions
  

Files:

• trunk/CocoaICU Tests-Info.plist (added)
• trunk/CocoaICU.xcodeproj/project.pbxproj (modified) (12 diffs)
• trunk/source/CocoaICU.m (modified) (2 diffs)
• trunk/source/ICUMatcher.m (modified) (6 diffs)
• trunk/source/ICUPattern.m (modified) (5 diffs)
• trunk/source/NSStringICUAdditions.h (modified) (1 diff)
• trunk/source/NSStringICUAdditions.m (modified) (2 diffs)
• trunk/source/TestCases.h (added)
• trunk/source/TestCases.m (added)

trunk/CocoaICU.xcodeproj/project.pbxproj

@@ -8 +8 @@
-
-/* Begin PBXBuildFile section */
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
-/* End PBXBuildFile section */
-
@@ -22 +28 @@
-                939322570B1E6AD000DD0AF0 /* ICUMatcher.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = ICUMatcher.h; path = source/ICUMatcher.h; sourceTree = "<group>"; };
-                939322580B1E6AD000DD0AF0 /* ICUMatcher.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; name = ICUMatcher.m; path = source/ICUMatcher.m; sourceTree = "<group>"; };
+                93AB9E800C98A24C00B11910 /* TestCases.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = TestCases.h; path = source/TestCases.h; sourceTree = "<group>"; };
+                93AB9E810C98A24C00B11910 /* TestCases.m */ = {isa = PBXFileReference; fileEncoding = 4; includeInIndex = 1; lastKnownFileType = sourcecode.c.objc; name = TestCases.m; path = source/TestCases.m; sourceTree = "<group>"; };
+                93AB9E8E0C98A30C00B11910 /* CocoaICU Tests.octest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = "CocoaICU Tests.octest"; sourceTree = BUILT_PRODUCTS_DIR; };
+                93AB9E8F0C98A30C00B11910 /* CocoaICU Tests-Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.xml; path = "CocoaICU Tests-Info.plist"; sourceTree = "<group>"; };
+                93BC7A0C0C9D563E00181C32 /* ICUTool */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = ICUTool; sourceTree = BUILT_PRODUCTS_DIR; };
-                93E2C0CE0B18D142001193AC /* CocoaICU.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; name = CocoaICU.m; path = source/CocoaICU.m; sourceTree = "<group>"; };
-                93E2C0CF0B18D142001193AC /* ICUPattern.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = ICUPattern.h; path = source/ICUPattern.h; sourceTree = "<group>"; };
@@ -27 +38 @@
-                93E2C0D30B18D142001193AC /* NSStringICUAdditions.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = NSStringICUAdditions.h; path = source/NSStringICUAdditions.h; sourceTree = "<group>"; };
-                93E2C0D40B18D142001193AC /* NSStringICUAdditions.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; name = NSStringICUAdditions.m; path = source/NSStringICUAdditions.m; sourceTree = "<group>"; };
-                93E2C4F00B1A53F0001193AC /* CocoaICU */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = CocoaICU; sourceTree = BUILT_PRODUCTS_DIR; };
-/* End PBXFileReference section */
-
-/* Begin PBXFrameworksBuildPhase section */
-E2C4EE0B1A53F0001193AC
+AB9E8B0C98A30C00B11910
-                        isa = PBXFrameworksBuildPhase;
-                        buildActionMask = 2147483647;
-                        files = (
-
-
+
+
+
+
+
+
+
+
+
+
+
-                        );
-                        runOnlyForDeploymentPostprocessing = 0;
@@ -47 +66 @@
-                        children = (
-                                93E2C0DE0B18D17D001193AC /* Other */,
+                                93AB9E7D0C98A23900B11910 /* Tests */,
-                                08FB7795FE84155DC02AAC07 /* Source */,
-                                08FB779DFE84155DC02AAC07 /* External Frameworks and Libraries */,
-                                1AB674ADFE9D54B511CA2CBB /* Products */,
+                                93AB9E8F0C98A30C00B11910 /* CocoaICU Tests-Info.plist */,
-                        );
-                        name = CocoaICU;
@@ -79 +100 @@
-                        isa = PBXGroup;
-                        children = (
-
+
+
-                        );
-                        name = Products;
+                        sourceTree = "<group>";
+                };
+                93AB9E7D0C98A23900B11910 /* Tests */ = {
+                        isa = PBXGroup;
+                        children = (
+                                93AB9E800C98A24C00B11910 /* TestCases.h */,
+                                93AB9E810C98A24C00B11910 /* TestCases.m */,
+                        );
+                        name = Tests;
-                        sourceTree = "<group>";
-                };
@@ -96 +127 @@
-
-/* Begin PBXNativeTarget section */
-E2C4EF0B1A53F0001193AC /* CocoaICU
+AB9E8D0C98A30C00B11910 /* CocoaICU Tests
-                        isa = PBXNativeTarget;
-E2C5010B1A5403001193AC /* Build configuration list for PBXNativeTarget "CocoaICU
+AB9E900C98A30D00B11910 /* Build configuration list for PBXNativeTarget "CocoaICU Tests
-                        buildPhases = (
-
-
+
+
+
+
-                        );
-                        buildRules = (
@@ -107 +140 @@
-                        dependencies = (
-                        );
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-                        productType = "com.apple.product-type.tool";
-                };
@@ -122 +172 @@
-                        projectDirPath = "";
-                        targets = (
-
+
+
-                        );
-                };
-/* End PBXProject section */
-
+/* Begin PBXResourcesBuildPhase section */
+                93AB9E890C98A30C00B11910 /* Resources */ = {
+                        isa = PBXResourcesBuildPhase;
+                        buildActionMask = 2147483647;
+                        files = (
+                        );
+                        runOnlyForDeploymentPostprocessing = 0;
+                };
+/* End PBXResourcesBuildPhase section */
+
+/* Begin PBXShellScriptBuildPhase section */
+                93AB9E8C0C98A30C00B11910 /* ShellScript */ = {
+                        isa = PBXShellScriptBuildPhase;
+                        buildActionMask = 2147483647;
+                        files = (
+                        );
+                        inputPaths = (
+                        );
+                        outputPaths = (
+                        );
+                        runOnlyForDeploymentPostprocessing = 0;
+                        shellPath = /bin/sh;
+                        shellScript = "# Run the unit tests in this test bundle.\n\"${SYSTEM_DEVELOPER_DIR}/Tools/RunUnitTests\"\n";
+                };
+/* End PBXShellScriptBuildPhase section */
+
-/* Begin PBXSourcesBuildPhase section */
-E2C4ED0B1A53F0001193AC
+AB9E8A0C98A30C00B11910
-                        isa = PBXSourcesBuildPhase;
-                        buildActionMask = 2147483647;
-                        files = (
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-                        );
-                        runOnlyForDeploymentPostprocessing = 0;
@@ -166 +254 @@
-                        name = Release;
-                };
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-                        isa = XCBuildConfiguration;
-                        buildSettings = {
@@ -177 +317 @@
-                                INSTALL_PATH = "$(HOME)/bin";
-                                PREBINDING = NO;
-CocoaICU
+ICUTool
-                                ZERO_LINK = YES;
-                        };
-                        name = Debug;
-                };
-E2C5030B1A5403001193AC
+BC7A9B0C9D569200181C32
-                        isa = XCBuildConfiguration;
-                        buildSettings = {
@@ -191 +331 @@
-                                INSTALL_PATH = "$(HOME)/bin";
-                                PREBINDING = NO;
-CocoaICU
+ICUTool
-                                ZERO_LINK = NO;
-                        };
@@ -208 +348 @@
-                        defaultConfigurationName = Release;
-                };
-E2C5010B1A5403001193AC /* Build configuration list for PBXNativeTarget "CocoaICU
+AB9E900C98A30D00B11910 /* Build configuration list for PBXNativeTarget "CocoaICU Tests
-                        isa = XCConfigurationList;
-                        buildConfigurations = (
-
-
+
+
+
+
+
+
+
+
+
+
+
-                        );
-                        defaultConfigurationIsVisible = 0;

trunk/source/CocoaICU.m

@@ -4 +4 @@
-#import "NSStringICUAdditions.h"
-
+void runTest();
-
-
-
-
-
-
-
-
-
+
-int main (int argc, const char * argv[]) {
-    NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
-
-
-
-
-
-
-
-
+
-
-        [pool release];
@@ -28 +16 @@
-}
-
-
-
-
-
-
+
-        
-        while([matcher findNext]) {
-                NSRange r = [matcher rangeOfMatch];
-                NSLog(@"match = %@", [searchString substringWithRange:r]);
-        }
-}
-
-void findAndReplaceFirst(){
-        NSString *searchString = [NSString stringWithUTF8String:"庭庢庭庢圭圀庭庢"];
-        NSString *patternString = [NSString stringWithUTF8String:"([庭庢]+)"];
-        ICUPattern *p = [ICUPattern patternWithString:patternString];
-        ICUMatcher *matcher = [ICUMatcher matcherWithPattern:p overString:searchString];
-        NSString *result = [matcher replaceFirstWithString:[NSString stringWithUTF8String:"匇匈匇匊"]];
-        NSLog(@"result = %@", result);
-}
-
-void findAndReplaceAll() {
-        NSString *searchString = [NSString stringWithUTF8String:"❾➈➈➈➈➈➈❾➈➈➈➈➈➈❾➈➈➈➈➈➈"];
-        NSString *patternString = [NSString stringWithUTF8String:"(❾➈➈➈➈➈➈)"];
-        ICUPattern *p = [ICUPattern patternWithString:patternString];
-        ICUMatcher *matcher = [ICUMatcher matcherWithPattern:p overString:searchString];
-        NSString *result = [matcher replaceAllWithString:[NSString stringWithUTF8String:"ŞŞËËËËà à à à À"]];
-        NSLog(@"result = %@", result);
-}
-
-void split() {
-        NSString *searchString = @"bbdfdbbababababaklsdjfababababab";
-        NSString *patternString = @"[ab]+";
-        ICUPattern *p = [ICUPattern patternWithString:patternString];
-        NSArray *r = [p componentsSplitFromString:searchString];
-        NSLog(@"result = %@", r);
-}
-
-void unicodeFind() {
-        NSString *searchString = [NSString stringWithUTF8String:"❾➈➈➈➈➈➈â
-“â
-“ŊŘÀÀÀÀÀŔŔĚĚĚÀÀÀŞŞŞËËËËà à à à ÀÀÀÀÀ"];
-        NSString *patternString = [NSString stringWithUTF8String:"À+"];
-        ICUPattern *p = [ICUPattern patternWithString:patternString];
-        ICUMatcher *matcher = [ICUMatcher matcherWithPattern:p overString:searchString];
-        while([matcher findNext]) {
-                NSLog(@"match = %@", [searchString substringWithRange:[matcher rangeOfMatch]]);
-        }
-}
-
-void groupFindAndReplace() {
-        NSString *searchString = @"ababcdcdababcdcdabefghiabbababbacdcdccdcd";
-        NSString *patternString = @"([ab]+)([cd]+)";
-        ICUPattern *p = [ICUPattern patternWithString:patternString];
-        ICUMatcher *matcher = [ICUMatcher matcherWithPattern:p overString:searchString];
-        while([matcher findNext]) {
-                int i;
-                for(i=0;i<=[matcher numberOfGroups];i++) {
-                        NSLog(@"group %i : %@", i, [matcher groupAtIndex:i]);
-                }
-        }
-}
-
-void advancedFindAndReplace() {
-        NSString *searchString = @"aabbcc";
-        NSString *patternString = @"(a+)..(c+)";
-        NSString *replacedText = [searchString replaceOccurrencesOfPattern:patternString withString:@" $2  $1"];
-        NSLog(@"%@", replacedText);
-}
-

trunk/source/ICUMatcher.m

@@ -110 +110 @@
-        while(YES) { 
-                UErrorCode status = 0;
-calloc(
+(UChar *)NSZoneCalloc([self zone], 
-                int32_t buffSize = uregex_group(re, groupIndex, dest, groupSize, &status);
-
-                if(U_BUFFER_OVERFLOW_ERROR == status) {
-                        groupSize *= 2;
-free(
+NSZoneFree([self zone], 
-                        continue;
-                }
@@ -123 +123 @@
-                groupSize = InitialGroupSize; // reset to default
-                NSString *result = [[[NSString alloc] initWithBytes:dest length:buffSize*sizeof(UChar) encoding:[NSString nativeUTF16Encoding]] autorelease];
-free(
+NSZoneFree([self zone], 
-                return result;
-        }
@@ -150 +150 @@
-
--(NSString *)performReplacementWithString:(NSString *)aReplacementString replaceAll:(BOOL)replacingAll {
+
-        UErrorCode status = 0;
-        UChar *replacementText = [aReplacementString UTF16String];
-        URegularExpression *re = [[self pattern] re];
-
+unsigned 
-        
-        BOOL replacementCompleted = NO;
@@ -161 +162 @@
-        while(!replacementCompleted) {
-                
-
+
+
+
+
+
+
+
+
-                if(replacingAll)
-                        resultLength = uregex_replaceAll(re, replacementText, -1, destString, destStringBufferSize, &status);
@@ -167 +175 @@
-                        resultLength = uregex_replaceFirst(re, replacementText, -1, destString, destStringBufferSize, &status);
-
+                // realloc some more space if possible
-                if(status == U_BUFFER_OVERFLOW_ERROR) {
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
-                        [NSException raise:@"Find Exception"
-                                                format:@"Could not perform find and replace: %s", u_errorName(status)];
@@ -184 +200 @@
-                                                                                                 length:resultLength * sizeof(UChar)
-                                                                                           encoding:[NSString nativeUTF16Encoding]] autorelease];
-free(
+NSZoneFree([self zone], 
-        return result;  
-}

trunk/source/ICUPattern.m

@@ -82 +82 @@
-
-        if(re != NULL)
-free(
+NSZoneFree([self zone], 
-
-        if(textToSearch != NULL)
@@ -152 +152 @@
--(void)setRe:(URegularExpression *)p {
-        if(re != NULL)
-free(
+NSZoneFree([self zone], 
-
-        re = p;
@@ -191 +191 @@
-
-        while(!isDone) {
-calloc(
+(UChar *)NSZoneCalloc([self zone], 
-                int requiredCapacity = 0;
-                UChar *destFields[destFieldsCapacity];
@@ -201 +201 @@
-                                                                                          destFieldsCapacity,
-                                                                                          &status);
-
+
-                if(status == U_BUFFER_OVERFLOW_ERROR) { // buffer was too small, grow it
-free(
+NSZoneFree([self zone], 
-                        NSAssert(destCapacity * 2 < INT_MAX, @"Overflow occurred splitting string.");
-                        destCapacity = (destCapacity < requiredCapacity) ? requiredCapacity : destCapacity * 2;
@@ -210 +210 @@
-                        destFieldsCapacity *= 2;
-                        NSAssert(destFieldsCapacity *2 < INT_MAX, @"Overflow occurred splitting string.");
-free(
+NSZoneFree([self zone], 
-                        status = 0;
-                } else if(U_FAILURE(status)) {
-free(
+NSZoneFree([self zone], 
-                        isDone = YES;
-                } else {

trunk/source/NSStringICUAdditions.h

@@ -12 +12 @@
-@interface NSString (NSStringICUAdditions)
-
+/*!
+    @method     nativeUTF16Encoding
+    @abstract   The native UTF16 encoding on the given machine.
+    @discussion The native UTF16 encoding on the given machine.
+*/
-+(NSStringEncoding)nativeUTF16Encoding;
-
+/*!
+    @method     stringWithICUString:
+    @abstract   Create an NSString from a UTF16 encoded string.
+    @discussion Create an NSString from a UTF16 encoded string.
+*/
-+(NSString *)stringWithICUString:(void *)utf16EncodedString;
+
+/*!
+    @method     UTF16String
+    @abstract   Returns a UTF16 encoded string in the native encoding.
+    @discussion This string has a retain policy equivalent to UTF8String. In
+ other words, if you want to keep this string around beyond the given autorelease
+ context, you need to copy the returned string.
+*/
--(void *)UTF16String;
+
+
+/*!
+ @method     copyUTF16String
+ @abstract   returns a copy
+ @discussion Returns a UTF16 encoded string in the native encoding. The returned
+ buffer must eventually be freed.
+*/
--(void *)copyUTF16String;
-
-
-
-
-
+
+
+
+
+
+
+
--(NSArray *)findPattern:(NSString *)aRegex;
-
+/*!
+    @method     componentsSeparatedByPattern:
+    @abstract   Returns the components separated by the given pattern.
+    @discussion Break a string into components where components separated
+        by occurrences of the given pattern.
+*/
+-(NSArray *)componentsSeparatedByPattern:(NSString *)aRegex;
+
+/*!
+    @method     replaceOccurrencesOfPattern:withString:
+    @abstract   Replace occurrences of the pattern with the replacement text.
+    @discussion Replace occurrences of the pattern with the replacement text.
+   The replacement text may contain backrerferences.
+*/
+-(NSString *)replaceOccurrencesOfPattern:(NSString *)aPattern withString:(NSString *)replacementText;
+
+/*!
+    @method     matchesPattern:
+    @abstract   Returns YES if the string matches the entire pattern.
+    @discussion Returns YES if the string matches the entire pattern.
+*/
+-(BOOL)matchesPattern:(NSString *)aRegex;
+
-@end

trunk/source/NSStringICUAdditions.m

@@ -27 +27 @@
-@end
-
-//
+
-
--(NSString *)replaceOccurrencesOfPattern:(NSString *)aPattern withString:(NSString *)replacementText {
@@ -78 +78 @@
--(void *)UTF16String {
-        UChar *ret = (UChar *)[self cStringUsingEncoding:[NSString nativeUTF16Encoding]];
-
+
-        // for some reason, the null-terminator doesn't always show up at the right place and this
-        // causes extra characters to be created in the unicode string.  We remove them here by force.
-
+unsigned 
-        ret[len] = '\0';
-        return ret;