Added whitespace normalization

Cocoanetics · Jan 17, 2011 · 7c41248 · 7c41248
1 parent 04a449d
commit 7c41248
Show file tree

Hide file tree

Showing 6 changed files with 115 additions and 24 deletions.
diff --git a/Classes/NSAttributedString+HTML.m b/Classes/NSAttributedString+HTML.m
@@ -209,12 +209,6 @@ - (id)initWithHTML:(NSData *)data options:(NSDictionary *)options documentAttrib
 
 				CTParagraphStyleRef paragraphStyle = createParagraphStyle(0, 0, 0, 0);
 
-			//	[localAttributes setObject:(id)paragraphStyle forKey:(id)kCTParagraphStyleAttributeName];
-
-			//	NSString *fontColor = [currentTagAttributes objectForKey:@"color"];
-
-	//					[attributes setObject:(id)[color CGColor] forKey:(id)kCTForegroundColorAttributeName];
-
 				NSMutableDictionary *localAttributes = [NSMutableDictionary dictionaryWithObjectsAndKeys:attachment, @"DTTextAttachment",
 												 (id)embeddedObjectRunDelegate, kCTRunDelegateAttributeName, 
 												 (id)paragraphStyle, kCTParagraphStyleAttributeName, nil];
@@ -510,6 +504,9 @@ - (id)initWithHTML:(NSData *)data options:(NSDictionary *)options documentAttrib
 
 			if ([scanner scanUpToString:@"<" intoString:&tagContents])
 			{
+				tagContents = [tagContents  stringByNormalizingWhitespace];
+
+
 				NSMutableDictionary *fontAttributes = [NSMutableDictionary dictionary];
 				NSMutableDictionary *fontStyleAttributes = [NSMutableDictionary dictionary];
 
@@ -647,18 +644,6 @@ - (id)initWithHTML:(NSData *)data options:(NSDictionary *)options documentAttrib
 				}
 
 
-				// TODO: Needs better handling of whitespace compression and adding space between tags if there are newlines
-				if (![tagContents hasPrefix:@" "])
-				{
-
-
-					if ([[tmpString string] length] && ![[tmpString string] hasSuffix:@" "] && ![[tmpString string] hasSuffix:@"\n"])
-					{
-						tagContents = [@" " stringByAppendingString:tagContents];
-					}
-				}
-
-
 				// Add newline after block contents if a new block follows
 				NSString *nextTag = [scanner peekNextTagSkippingClosingTags:YES];
 
@@ -691,6 +676,23 @@ - (id)initWithHTML:(NSData *)data options:(NSDictionary *)options documentAttrib
 					}
 					needsNewLineBefore = NO;
 				}
+				else // might be a continuation of a paragraph, then we might need space before it
+				{
+				// TODO: Needs better handling of whitespace compression and adding space between tags if there are newlines
+				if (![tagContents hasPrefix:@" "])
+				{
+					NSString *stringSoFar = [tmpString string];
+
+					if ([stringSoFar length] && ![stringSoFar hasSuffix:@" "] && ![stringSoFar hasSuffix:@"\n"]  && ![stringSoFar hasSuffix:UNICODE_LINE_FEED])
+					{
+						// add space prefix unless punctuation character
+						if (![tagContents hasPrefixCharacterFromSet:[NSCharacterSet punctuationCharacterSet]])
+						{
+							tagContents = [@" " stringByAppendingString:tagContents];
+						}
+					}
+				}
+				}
 
 				NSAttributedString *tagString = [[NSAttributedString alloc] initWithString:tagContents attributes:attributes];
 				[tmpString appendAttributedString:tagString];

diff --git a/Classes/NSAttributedStringHTMLTest.m b/Classes/NSAttributedStringHTMLTest.m
@@ -71,6 +71,24 @@ - (void)testImageParagraphs
 	STAssertEqualObjects(resultOnIOS, resultOnMac, @"Output on List Test differs");
 }
 
+- (void)testSpaceNormalization
+{
+	NSString *html = @"<p>Now there is some <b>bold</b>\ntext and  spaces\n    should be normalized.</p>";	
+
+	NSData *data = [html dataUsingEncoding:NSUTF8StringEncoding];
+
+	NSAttributedString *string = [[NSAttributedString alloc] initWithHTML:data documentAttributes:NULL];
+	NSData *dump = [[string string] dataUsingEncoding:NSUTF8StringEncoding];
+	NSString *resultOnIOS = [dump description];
+
+	NSString *resultOnMac = @"<4e6f7720 74686572 65206973 20736f6d 6520626f 6c642074 65787420 616e6420 73706163 65732073 686f756c 64206265 206e6f72 6d616c69 7a65642e 0a>";
+
+	STAssertEqualObjects(resultOnIOS, resultOnMac, @"Output on List Test differs");
+}
+
+
+
+
 
 
 @end
diff --git a/Classes/NSString+HTML.h b/Classes/NSString+HTML.h
@@ -10,5 +10,7 @@
 
 - (NSUInteger)integerValueFromHex;
 - (BOOL)isInlineTag;
+- (NSString *)stringByNormalizingWhitespace;
+- (BOOL)hasPrefixCharacterFromSet:(NSCharacterSet *)characterSet;
 
 @end
diff --git a/Classes/NSString+HTML.m b/Classes/NSString+HTML.m
@@ -32,4 +32,59 @@ - (BOOL)isInlineTag
 	return [inlineTags containsObject:[self lowercaseString]];
 }
 
+
+- (NSString *)stringByNormalizingWhitespace
+{
+	NSCharacterSet *whiteSpaceCharacterSet = [NSCharacterSet whitespaceAndNewlineCharacterSet];
+
+	NSScanner *scanner = [NSScanner scannerWithString:self];
+	[scanner setCharactersToBeSkipped:nil];
+
+	NSMutableArray *tokens = [NSMutableArray array];
+
+	NSString *prefix = @"";
+	if ([scanner scanCharactersFromSet:whiteSpaceCharacterSet intoString:NULL])
+	{
+		prefix = @" ";
+	}
+
+	NSString *suffix = @"";
+
+	while (![scanner isAtEnd])
+	{
+		NSString *string = nil;
+
+		if ([scanner scanUpToCharactersFromSet:whiteSpaceCharacterSet intoString:&string])
+		{
+			[tokens addObject:string];
+		}
+
+		if ([scanner scanCharactersFromSet:whiteSpaceCharacterSet intoString:NULL])
+		{
+			suffix = @" ";
+		}
+		else 
+		{
+			suffix = @"";
+		}
+	}
+
+	NSString *retStr = [NSString stringWithFormat:@"%@%@%@", prefix, [tokens componentsJoinedByString:@" "], suffix];
+
+	return retStr;
+}
+
+
+- (BOOL)hasPrefixCharacterFromSet:(NSCharacterSet *)characterSet
+{
+	if (![self length])
+	{
+		return NO;
+	}
+
+	unichar firstChar = [self characterAtIndex:0];
+
+	return [characterSet characterIsMember:firstChar];
+}
+
 @end
diff --git a/CoreTextExtensions.xcodeproj/project.pbxproj b/CoreTextExtensions.xcodeproj/project.pbxproj
@@ -61,6 +61,16 @@
 		B2B440C712E030DF00497B74 /* DTAttributedTextContentView.m in Sources */ = {isa = PBXBuildFile; fileRef = A73BC48D12DA078100F064C6 /* DTAttributedTextContentView.m */; };
 /* End PBXBuildFile section */
 
+/* Begin PBXContainerItemProxy section */
+		A78A3F9E12E47D4E007CE622 /* PBXContainerItemProxy */ = {
+			isa = PBXContainerItemProxy;
+			containerPortal = 29B97313FDCFA39411CA2CEA /* Project object */;
+			proxyType = 1;
+			remoteGlobalIDString = 6A5A32BF12DD108D0019AAF1 /* UnitTest */;
+			remoteInfo = UnitTest;
+		};
+/* End PBXContainerItemProxy section */
+
 /* Begin PBXFileReference section */
 		1D30AB110D05D00D00671497 /* Foundation.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Foundation.framework; path = System/Library/Frameworks/Foundation.framework; sourceTree = SDKROOT; };
 		1D3623240D0F684500981E51 /* DemoAppDelegate.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = DemoAppDelegate.h; sourceTree = "<group>"; };
@@ -290,6 +300,7 @@
 			buildRules = (
 			);
 			dependencies = (
+				A78A3F9F12E47D4E007CE622 /* PBXTargetDependency */,
 			);
 			name = CoreTextExtensions;
 			productName = CoreTextExtensions;
@@ -435,6 +446,14 @@
 		};
 /* End PBXSourcesBuildPhase section */
 
+/* Begin PBXTargetDependency section */
+		A78A3F9F12E47D4E007CE622 /* PBXTargetDependency */ = {
+			isa = PBXTargetDependency;
+			target = 6A5A32BF12DD108D0019AAF1 /* UnitTest */;
+			targetProxy = A78A3F9E12E47D4E007CE622 /* PBXContainerItemProxy */;
+		};
+/* End PBXTargetDependency section */
+
 /* Begin XCBuildConfiguration section */
 		1D6058940D05DD3E006BFB54 /* Debug */ = {
 			isa = XCBuildConfiguration;

diff --git a/Resources/CurrentTest.html b/Resources/CurrentTest.html
@@ -1,6 +1 @@
-<p>Now there is some <b>bold</b>
-text and  spaces    should be normalized.</p>
-
-<h1>List Oddity</h1>
-<ul><li>A List Item</li></ul>
-<p>This line below a list is too close if not a trick is used. I would want the extra spacing to be achieved by adding paragraph spacing before.</p>
+<p>Before</p><img src="Oliver.jpg"><h1>Header</h2><p>after</p><p>Some inline <img src="Oliver.jpg"> text.</p>