Merge from dzindra.

PDFKitten/CMap.h

@@ -10,10 +10,7 @@ extern NSValue *rangeValue(unsigned int from, unsigned int to);
 @end
 
 @interface CMap : NSObject {
-	NSMutableArray *offsets;
-    NSMutableDictionary *chars;
 	NSMutableDictionary *context;
-	NSString *currentEndToken;
 
 	/* CMap ranges */
 	NSMutableArray *codeSpaceRanges;

PDFKitten/CMap.m

@@ -21,6 +21,14 @@ + (Operator *)operatorWithStart:(NSString *)start end:(NSString *)end handler:(S
 	return op;
 }
 
+- (void)dealloc
+{
+    [start release];
+    [end release];
+
+    [super dealloc];
+}
+
 @synthesize start, end, handler;
 @end
 
@@ -29,10 +37,10 @@ - (void)handleCodeSpaceRange:(NSString *)string;
 - (void)handleCharacter:(NSString *)string;
 - (void)handleCharacterRange:(NSString *)string;
 - (void)parse:(NSString *)cMapString;
-@property (readonly) NSCharacterSet *tokenDelimiterSet;
-@property (nonatomic, retain) NSMutableDictionary *context;
-@property (nonatomic, readonly) NSCharacterSet *tagSet;
-@property (nonatomic, readonly) NSSet *operators;
+@property(nonatomic, retain) NSMutableDictionary *context;
+@property(readonly) NSCharacterSet *tokenDelimiterSet;
+@property(readonly) NSCharacterSet *tagSet;
+@property(readonly) NSSet *operators;
 @end
 
 @implementation CMap
@@ -69,12 +77,14 @@ - (BOOL)isInCodeSpaceRange:(unichar)cid
 	return NO;
 }
 
+#pragma mark - Public methods
+
 /**!
  * Returns the unicode value mapped by the given character ID
  */
 - (unichar)unicodeCharacter:(unichar)cid
 {
-	if (![self isInCodeSpaceRange:cid]) return 0;
+	if (![self isInCodeSpaceRange:cid]) return (unichar) NSNotFound;
 
 	NSArray	*mappedRanges = [self.characterRangeMappings allKeys];
 	for (NSValue *rangeValue in mappedRanges)
@@ -86,89 +96,66 @@ - (unichar)unicodeCharacter:(unichar)cid
 			return cid + [offsetValue intValue];
 		}
 	}
-
-	NSArray *mappedValues = [self.characterMappings allKeys];
-	for (NSNumber *from in mappedValues)
-	{
-		if ([from intValue] == cid)
-		{
-			return [[self.characterMappings objectForKey:from] intValue];
-		}
-	}
-
-	return (unichar) NSNotFound;
+
+    NSNumber *result = [self.characterMappings objectForKey:[NSNumber numberWithInt:cid]];
+    if (result) {
+        return [result intValue];
+    }
+
+    return (unichar) NSNotFound;
 }
 
-- (NSSet *)operators
-{
-	@synchronized (self)
+- (unichar)cidCharacter:(unichar)unicode {
+    __block unichar result = NSNotFound;
+
+    [self.characterRangeMappings enumerateKeysAndObjectsUsingBlock:^(NSValue *rangeValue, NSNumber *offset, BOOL *stop) {
+        NSRange range = [rangeValue rangeValue];
+        range.location += [offset intValue];
+        if (unicode >= range.location && unicode <= NSMaxRange(range)) {
+            result = unicode - [offset intValue];
+            *stop = YES;
+        }
+    }];
+    if (result != NSNotFound)
+        return result;
+
+    NSArray *keys = [self.characterMappings allKeysForObject:[NSNumber numberWithInt:unicode]];
+    if (keys.count) {
+        if (keys.count > 1) {
+            NSLog(@"more keys for character %C keys = %@", unicode, keys);
+        }
+        return [[keys lastObject]intValue];
+    } else {
+        return NSNotFound;
+    }
+/*
+    // Look up the offsets dictionary for this unicode
+    for (NSDictionary *dict in offsets)
 	{
-		if (!sharedOperators)
-		{
-			sharedOperators = [[NSMutableSet alloc] initWithObjects:
-							   [Operator operatorWithStart:@"begincodespacerange" 
-														end:@"endcodespacerange"
-													handler:@selector(handleCodeSpaceRange:)],
-							   [Operator operatorWithStart:@"beginbfchar" 
-													   end:@"endbfchar" 
-												   handler:@selector(handleCharacter:)],
-							   [Operator operatorWithStart:@"beginbfrange" 
-													   end:@"endbfrange" 
-												   handler:@selector(handleCharacterRange:)],
-			nil];
-		}
-		return sharedOperators;
-	}
+        int firstChar = [[dict objectForKey:@"First"] intValue];
+        int lastChar = [[dict objectForKey:@"Last"] intValue];
+        int offset = [[dict objectForKey:@"Offset"] intValue];
+
+        for (int i = 0 ; i <= lastChar-firstChar ; i++) {
+            unichar dictUnicode = offset+i;
+            if (dictUnicode == unicode) {
+                return i;
+            }
+        }
+	} */
 }
 
 #pragma mark -
 #pragma mark Scanner
 
 - (Operator *)operatorWithStartingToken:(NSString *)token {
-	NSString *content = nil;
-    NSCharacterSet *newLineSet = [NSCharacterSet newlineCharacterSet];
-    NSCharacterSet *tagSet = [NSCharacterSet characterSetWithCharactersInString:@"<>"];
-    NSString *separatorString = @"> <";
-
-    chars = [[NSMutableDictionary alloc] init];    
-    NSScanner *rangeScanner = [NSScanner scannerWithString:content];
-    while (![rangeScanner isAtEnd])
-    {
-        NSString *line = nil;
-        [rangeScanner scanUpToCharactersFromSet:newLineSet intoString:&line];
-        line = [line stringByTrimmingCharactersInSet:tagSet];
-        NSArray *parts = [line componentsSeparatedByString:separatorString];
-
-        NSUInteger from, to;
-        NSScanner *scanner = [NSScanner scannerWithString:[parts objectAtIndex:0]];
-        [scanner scanHexInt:&from];
-        NSNumber *fromNumber = [NSNumber numberWithInt:from];
-
-        NSString *toString = [parts objectAtIndex:1];
-        int charLen = 4;
-        if ([toString length] > charLen) {
-            NSMutableArray *toArray = [NSMutableArray arrayWithCapacity: [toString length]/4 + ([toString length]%4 > 0 ? 1 : 0)];
-            NSRange range;
-            NSString *nextTo;
-            for (int offset = 0; offset < [toString length]/4; offset++) {
-                range = NSMakeRange(offset * charLen, charLen);
-                nextTo = [toString substringWithRange: range];
-                scanner = [NSScanner scannerWithString: nextTo];
-                [scanner scanHexInt: &to];
-                [toArray addObject: [NSNumber numberWithInt:to]];
+	if (token) {
+        for (Operator *op in self.operators) {
+            if ([op.start isEqualToString:token]) {
+                return op;
             }
-            [chars setObject: toArray forKey:fromNumber];
-        }
-        else 
-        {
-            scanner = [NSScanner scannerWithString:[parts objectAtIndex:1]];
-            [scanner scanHexInt:&to];
-            NSNumber *toNumber = [NSNumber numberWithInt:to];
-            [chars setObject:toNumber forKey:fromNumber];
-
         }
     }
-
     return nil;
 }
 
@@ -320,6 +307,23 @@ - (void)handleCharacterRange:(NSString *)token
 #pragma mark -
 #pragma mark Accessor methods
 
+- (NSSet *)operators {
+    if (!sharedOperators) {
+        sharedOperators = [[NSMutableSet alloc] initWithObjects:
+                [Operator operatorWithStart:@"begincodespacerange"
+                        end:@"endcodespacerange"
+                        handler:@selector(handleCodeSpaceRange:)],
+                [Operator operatorWithStart:@"beginbfchar"
+                        end:@"endbfchar"
+                        handler:@selector(handleCharacter:)],
+                [Operator operatorWithStart:@"beginbfrange"
+                        end:@"endbfrange"
+                        handler:@selector(handleCharacterRange:)],
+                nil];
+    }
+    return sharedOperators;
+}
+
 - (NSCharacterSet *)tagSet {
 	if (!sharedTagSet) {
 		sharedTagSet = [[NSCharacterSet characterSetWithCharactersInString:@"<>"] retain];
@@ -355,48 +359,15 @@ - (NSMutableDictionary *)characterRangeMappings {
 	return characterRangeMappings;
 }
 
-- (unichar)cidCharacter:(unichar)unicode 
-{
-    //TODO: search in range dictionary
-
-    // Look up the offsets dictionary for this unicode
-    for (NSDictionary *dict in offsets)
-	{
-        int firstChar = [[dict objectForKey:@"First"] intValue];
-        int lastChar = [[dict objectForKey:@"Last"] intValue];
-        int offset = [[dict objectForKey:@"Offset"] intValue];
-
-        for (int i = 0 ; i <= lastChar-firstChar ; i++) {
-            unichar dictUnicode = offset+i;
-            if (dictUnicode == unicode) {
-                return i;
-            }
-        }
-	}
-
-    if (chars) {
-        NSEnumerator *keys = [chars keyEnumerator];
-        NSObject *value;
-        NSObject *key;
-        while (key = [keys nextObject]) {
-            value = [chars objectForKey: key];
-            if ([value isKindOfClass: [NSNumber class]]) {
-                if ([(NSNumber *)value intValue] == unicode) {
-                    return [(NSNumber *)key intValue];
-                }
-            }
-        }        
-    }
-    return unicode;
-}
-
 - (void)dealloc
 {
-	[offsets release];
+    [context release];
+    [characterMappings release];
+    [characterRangeMappings release];
 	[codeSpaceRanges release];
 	[super dealloc];
 }
 
-@synthesize operators, context;
+@synthesize context;
 @synthesize codeSpaceRanges, characterMappings, characterRangeMappings;
 @end

PDFKitten/FontDescriptor.m

@@ -91,11 +91,11 @@ - (id)initWithPDFDictionary:(CGPDFDictionaryRef)dict
 		{
 			CGPDFDataFormat format;
 			NSData *data = (NSData *) CGPDFStreamCopyData(fontFileStream, &format);
-
+			/*
 	 		NSString *path = [NSSearchPathForDirectoriesInDomains(NSDocumentDirectory, NSUserDomainMask, YES) lastObject];
 			path = [path stringByAppendingPathComponent:@"fontfile"];
 			[data writeToFile:path atomically:YES];
-
+			  */
 			fontFile = [[FontFile alloc] initWithData:data];
 			[data release];
 		}

PDFKitten/Scanner.h

@@ -8,6 +8,7 @@
 @interface Scanner : NSObject <StringDetectorDelegate> {
 	CGPDFPageRef pdfPage;
 	NSMutableArray *selections;
+    Selection *possibleSelection;
 
 	StringDetector *stringDetector;
 	FontCollection *fontCollection;

PDFKitten/Scanner.m

@@ -18,6 +18,7 @@ - (id)initWithPage:(CGPDFPageRef)page {
 }
 
 - (NSArray *)select:(NSString *)keyword {
+    self.content = [NSMutableString string];
 	self.stringDetector = [StringDetector detectorWithKeyword:keyword delegate:self];
 	[self.selections removeAllObjects];
     self.renderingStateStack = [RenderingStateStack stack];
@@ -110,20 +111,25 @@ - (void)detector:(StringDetector *)detector didScanCharacter:(unichar)character
 	[self.renderingState translateTextPosition:CGSizeMake(width, 0)];
 }
 
-- (void)detectorDidStartMatching:(StringDetector *)stringDetector {
-	[self.selections addObject:[Selection selectionWithState:self.renderingState]];
+- (void)detectorDidStartMatching:(StringDetector *)detector {
+    possibleSelection = [[Selection selectionWithState:self.renderingState] retain];
 }
 
 - (void)detectorFoundString:(StringDetector *)detector {
-	Selection *selection = [self.selections lastObject];
-	selection.finalState = self.renderingState;
+    if (possibleSelection) {
+	    possibleSelection.finalState = self.renderingState;
+        [self.selections addObject:possibleSelection];
+        [possibleSelection release];
+        possibleSelection = nil;
+    }
 }
 
 - (RenderingState *)renderingState {
 	return [self.renderingStateStack topRenderingState];
 }
 
 - (void)dealloc {
+    [possibleSelection release];
 	[fontCollection release];
 	[selections release];
 	[renderingStateStack release];

PDFKitten/SimpleFont.m

@@ -48,31 +48,29 @@ - (void)setEncodingWithFontDictionary:(CGPDFDictionaryRef)dict
 - (NSString *)stringWithPDFString:(CGPDFStringRef)pdfString
 {
 	const unsigned char *bytes = CGPDFStringGetBytePtr(pdfString);
-	NSInteger length = CGPDFStringGetLength(pdfString);
-	NSData *rawBytes = [NSData dataWithBytes:bytes length:length];
+	NSUInteger length = CGPDFStringGetLength(pdfString);
 	if (!self.encoding && self.toUnicode)
 	{
 		// Use ToUnicode map
-		NSString *str = (NSString *) CGPDFStringCopyTextString(pdfString);
 		NSMutableString *unicodeString = [NSMutableString string];
 
 		// Translate to Unicode
-		for (int i = 0; i < [str length]; i++)
+		for (int i = 0; i < length; i++)
 		{
-			unichar cid = [str characterAtIndex:i];
-		 	[unicodeString appendFormat:@"%C", [self.toUnicode unicodeCharacter:cid]];
+			unichar cid = bytes[i];
+            unichar uni = [self.toUnicode unicodeCharacter:cid];
+            //NSLog(@"(%hu) %C -> (%hu) %C", cid, cid, uni, uni);
+            [unicodeString appendFormat:@"%C", uni];
 		}
 
-		[str release];
 		return unicodeString;
 	}
 	else if (!self.encoding)
 	{
 		return [super stringWithPDFString:pdfString];
-//		NSString *string = [[NSString alloc] initWithData:rawBytes encoding:NSMacOSRomanStringEncoding];
-//		return [string autorelease];
 	}
-	NSString *string = [[NSString alloc] initWithData:rawBytes encoding:self.encoding];
+    NSData *rawBytes = [NSData dataWithBytes:bytes length:length];
+	NSString *string = [[NSString alloc] initWithData:rawBytes encoding:nativeEncoding(self.encoding)];
 
 	return [string autorelease];
 }
@@ -121,4 +119,15 @@ - (void)setEncodingWithEncodingObject:(CGPDFObjectRef)object
 //	return [NSString stringWithCString:characters encoding:encoding];
 //}
 
+- (CGFloat)widthOfSpace
+{
+    unichar c = 0x20;
+    if (self.toUnicode) {
+        c = [self.toUnicode cidCharacter:c];
+        if (c == NSNotFound)
+            return 0;
+    }
+    return [self widthOfCharacter:c withFontSize:1.0];
+}
+
 @end

PDFKitten/StringDetector.h

@@ -22,7 +22,6 @@
 - (void)setKeyword:(NSString *)kword;
 - (void)reset;
 
-- (NSString *)appendPDFString:(CGPDFStringRef)string withFont:(Font *)font;
 - (NSString *)appendString:(NSString *)inputString;
 
 @property (nonatomic, assign) id<StringDetectorDelegate> delegate;