/*
	"Flatten" a string of HTML, removing all HTML tags
	Original Source: <http://cocoa.karelia.com/Foundation_Categories/NSString/_Flatten__a_string_.m>
	(See copyright notice at <http://cocoa.karelia.com>)
*/

/*"	Remove HTML tags to turn a fragment of HTML into a piece of plain text.  All characters between < and > are removed.
	Escape sequences such as &amp; are converted into their character equivalents.
"*/

// Note: This is slow approach that I'm now considering obsolete.
// See this URL for a better approach:
// http://sugarmaplesoftware.com/25/strip-html-tags/

- (NSString *) flattenHTML
{
	NSString *result = self;
	
	if (![self isEqualToString:@""])	// if empty string, don't do this!  You get junk.
	{
		// HACK -- IF SHORT LENGTH, USE MACROMAN -- FOR SOME REASON UNICODE FAILS FOR "" AND "-" AND "CNN" ...

		int encoding = ([self length] > 3) ? NSUnicodeStringEncoding : NSMacOSRomanStringEncoding;
		NSAttributedString *attrString;
		NSData *theData = [self dataUsingEncoding:encoding];
		if (nil != theData)	// this returned nil once; not sure why; so handle this case.
		{
			NSDictionary *encodingDict = [NSDictionary dictionaryWithObject:[NSNumber numberWithInt:encoding] forKey:@"CharacterEncoding"];
			attrString
				= [[NSAttributedString alloc]
					initWithHTML:theData documentAttributes:&encodingDict];
			result = [[[attrString string] retain] autorelease];	// keep only this
			[attrString release];	// don't do autorelease since this is so deep down.
		}
	}
	return result;
}