ObjFW  Check-in [265a0928ca]

		OF_XMLPARSER_IN_CDATA_2,
		OF_XMLPARSER_IN_COMMENT_OPENING,
		OF_XMLPARSER_IN_COMMENT_1,
		OF_XMLPARSER_IN_COMMENT_2,
		OF_XMLPARSER_IN_DOCTYPE,
		OF_XMLPARSER_NUM_STATES
	} _state;
	OFDataArray *_cache;
	OFDataArray *_buffer;
	OFString *_name, *_prefix;
	OFMutableArray *_namespaces, *_attributes;
	OFString *_attributeName, *_attributePrefix;
	char _delimiter;
	OFMutableArray *_previous;
	size_t _level;
	BOOL _acceptProlog;

#import "macros.h"

typedef void (*state_function)(id, SEL, const char*, size_t*, size_t*);
static SEL selectors[OF_XMLPARSER_NUM_STATES];
static state_function lookupTable[OF_XMLPARSER_NUM_STATES];

static OF_INLINE void
cache_append(OFDataArray *cache, const char *string,
buffer_append(OFDataArray *buffer, const char *string,
    of_string_encoding_t encoding, size_t length)
{
	if (OF_LIKELY(encoding == OF_STRING_ENCODING_UTF_8))
		[cache addItems: string
			  count: length];
		[buffer addItems: string
			   count: length];
	else {
		void *pool = objc_autoreleasePoolPush();
		OFString *tmp = [OFString stringWithCString: string
						   encoding: encoding
						     length: length];
		[cache addItems: [tmp UTF8String]
			  count: [tmp UTF8StringLength]];
		[buffer addItems: [tmp UTF8String]
			   count: [tmp UTF8StringLength]];
		objc_autoreleasePoolPop(pool);
	}
}

static OFString*
transform_string(OFDataArray *cache, size_t cut, BOOL unescape,
transform_string(OFDataArray *buffer, size_t cut, BOOL unescape,
    id <OFStringXMLUnescapingDelegate> delegate)
{
	char *items;
	size_t i, length;
	BOOL hasEntities = NO;
	OFString *ret;

	items = [cache items];
	length = [cache count] - cut;
	items = [buffer items];
	length = [buffer count] - cut;

	for (i = 0; i < length; i++) {
		if (items[i] == '\r') {
			if (i + 1 < length && items[i + 1] == '\n') {
				[cache removeItemAtIndex: i];
				items = [cache items];
				[buffer removeItemAtIndex: i];
				items = [buffer items];

				i--;
				length--;
			} else
				items[i] = '\n';
		} else if (items[i] == '&')
			hasEntities = YES;

{
	self = [super init];

	@try {
		void *pool;
		OFMutableDictionary *dict;

		_cache = [[OFBigDataArray alloc] init];
		_buffer = [[OFBigDataArray alloc] init];
		_previous = [[OFMutableArray alloc] init];
		_namespaces = [[OFMutableArray alloc] init];
		_attributes = [[OFMutableArray alloc] init];

		pool = objc_autoreleasePoolPush();
		dict = [OFMutableDictionary dictionaryWithKeysAndObjects:
		    @"xml", @"http://www.w3.org/XML/1998/namespace",

	}

	return self;
}

- (void)dealloc
{
	[_cache release];
	[_buffer release];
	[_name release];
	[_prefix release];
	[_namespaces release];
	[_attributes release];
	[_attributeName release];
	[_attributePrefix release];
	[_previous release];

			_lineNumber++;

		_lastCarriageReturn = (buffer[i] == '\r');
	}

	/* In OF_XMLPARSER_IN_TAG, there can be only spaces */
	if (length - last > 0 && _state != OF_XMLPARSER_IN_TAG)
		cache_append(_cache, buffer + last, _encoding, length - last);
		buffer_append(_buffer, buffer + last, _encoding, length - last);
}

- (void)parseString: (OFString*)string
{
	[self parseBuffer: [string UTF8String]
		   length: [string UTF8StringLength]];
}

		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	if (buffer[*i] != '<')
		return;

	if ((length = *i - *last) > 0)
		cache_append(_cache, buffer + *last, _encoding, length);
		buffer_append(_buffer, buffer + *last, _encoding, length);

	if ([_cache count] > 0) {
	if ([_buffer count] > 0) {
		void *pool = objc_autoreleasePoolPush();
		OFString *characters = transform_string(_cache, 0, YES, self);
		OFString *characters = transform_string(_buffer, 0, YES, self);

		if ([_delegate respondsToSelector:
		    @selector(parser:foundCharacters:)])
			[_delegate parser: self
			  foundCharacters: characters];

		objc_autoreleasePoolPop(pool);
	}

	[_cache removeAllItems];
	[_buffer removeAllItems];

	*last = *i + 1;
	_state = OF_XMLPARSER_TAG_OPENED;
}

/* Tag was just opened */
- (void)OF_parseTagOpenedWithBuffer: (const char*)buffer

{
	if (buffer[*i] == '?')
		_level = 1;
	else if (_level == 1 && buffer[*i] == '>') {
		void *pool = objc_autoreleasePoolPush();
		OFString *PI;

		cache_append(_cache, buffer + *last, _encoding, *i - *last);
		PI = transform_string(_cache, 1, NO, nil);
		buffer_append(_buffer, buffer + *last, _encoding, *i - *last);
		PI = transform_string(_buffer, 1, NO, nil);

		if ([PI isEqual: @"xml"] || [PI hasPrefix: @"xml "] ||
		    [PI hasPrefix: @"xml\t"] || [PI hasPrefix: @"xml\r"] ||
		    [PI hasPrefix: @"xml\n"])
			if (![self OF_parseXMLProcessingInstructions: PI])
				@throw [OFMalformedXMLException
				    exceptionWithClass: [self class]
						parser: self];

		if ([_delegate respondsToSelector:
		    @selector(parser:foundProcessingInstructions:)])
			[_delegate		 parser: self
			    foundProcessingInstructions: PI];

		objc_autoreleasePoolPop(pool);

		[_cache removeAllItems];
		[_buffer removeAllItems];

		*last = *i + 1;
		_state = OF_XMLPARSER_OUTSIDE_TAG;
	} else
		_level = 0;
}

/* Inside a tag, no name yet */
- (void)OF_parseInTagNameWithBuffer: (const char*)buffer
				  i: (size_t*)i
			       last: (size_t*)last
{
	void *pool;
	const char *cacheCString, *tmp;
	size_t length, cacheLength;
	OFString *cacheString;
	const char *bufferCString, *tmp;
	size_t length, bufferLength;
	OFString *bufferString;

	if (buffer[*i] != ' ' && buffer[*i] != '\t' && buffer[*i] != '\n' &&
	    buffer[*i] != '\r' && buffer[*i] != '>' && buffer[*i] != '/')
		return;

	if ((length = *i - *last) > 0)
		cache_append(_cache, buffer + *last, _encoding, length);
		buffer_append(_buffer, buffer + *last, _encoding, length);

	pool = objc_autoreleasePoolPush();

	cacheCString = [_cache items];
	cacheLength = [_cache count];
	cacheString = [OFString stringWithUTF8String: cacheCString
					      length: cacheLength];
	bufferCString = [_buffer items];
	bufferLength = [_buffer count];
	bufferString = [OFString stringWithUTF8String: bufferCString
					       length: bufferLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
	if ((tmp = memchr(bufferCString, ':', bufferLength)) != NULL) {
		_name = [[OFString alloc]
		    initWithUTF8String: tmp + 1
				length: bufferLength -
				length: cacheLength - (tmp - cacheCString) - 1];
					(tmp - bufferCString) - 1];
		_prefix = [[OFString alloc]
		    initWithUTF8String: cacheCString
				length: tmp - cacheCString];
		    initWithUTF8String: bufferCString
				length: tmp - bufferCString];
	} else {
		_name = [cacheString copy];
		_name = [bufferString copy];
		_prefix = nil;
	}

	if (buffer[*i] == '>' || buffer[*i] == '/') {
		OFString *namespace;

		namespace = namespace_for_prefix(_prefix, _namespaces);

				    didEndElement: _name
					   prefix: _prefix
					namespace: namespace];

			if ([_previous count] == 0)
				_finishedParsing = YES;
		} else
			[_previous addObject: cacheString];
			[_previous addObject: bufferString];

		[_name release];
		[_prefix release];
		_name = _prefix = nil;

		_state = (buffer[*i] == '/'
		    ? OF_XMLPARSER_EXPECT_CLOSE
		    : OF_XMLPARSER_OUTSIDE_TAG);
	} else
		_state = OF_XMLPARSER_IN_TAG;

	if (buffer[*i] != '/')
		[_namespaces addObject: [OFMutableDictionary dictionary]];

	objc_autoreleasePoolPop(pool);

	[_cache removeAllItems];
	[_buffer removeAllItems];
	*last = *i + 1;
}

/* Inside a close tag, no name yet */
- (void)OF_parseInCloseTagNameWithBuffer: (const char*)buffer
				       i: (size_t*)i
				    last: (size_t*)last
{
	void *pool;
	const char *cacheCString, *tmp;
	size_t length, cacheLength;
	OFString *cacheString, *namespace;
	const char *bufferCString, *tmp;
	size_t length, bufferLength;
	OFString *bufferString, *namespace;

	if (buffer[*i] != ' ' && buffer[*i] != '\t' && buffer[*i] != '\n' &&
	    buffer[*i] != '\r' && buffer[*i] != '>')
		return;

	if ((length = *i - *last) > 0)
		cache_append(_cache, buffer + *last, _encoding, length);
		buffer_append(_buffer, buffer + *last, _encoding, length);

	pool = objc_autoreleasePoolPush();

	cacheCString = [_cache items];
	cacheLength = [_cache count];
	cacheString = [OFString stringWithUTF8String: cacheCString
					      length: cacheLength];
	bufferCString = [_buffer items];
	bufferLength = [_buffer count];
	bufferString = [OFString stringWithUTF8String: bufferCString
					       length: bufferLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
	if ((tmp = memchr(bufferCString, ':', bufferLength)) != NULL) {
		_name = [[OFString alloc]
		    initWithUTF8String: tmp + 1
				length: bufferLength -
				length: cacheLength - (tmp - cacheCString) - 1];
					(tmp - bufferCString) - 1];
		_prefix = [[OFString alloc]
		    initWithUTF8String: cacheCString
				length: tmp - cacheCString];
		    initWithUTF8String: bufferCString
				length: tmp - bufferCString];
	} else {
		_name = [cacheString copy];
		_name = [bufferString copy];
		_prefix = nil;
	}

	if (![[_previous lastObject] isEqual: cacheString])
	if (![[_previous lastObject] isEqual: bufferString])
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	[_previous removeLastObject];

	[_cache removeAllItems];
	[_buffer removeAllItems];

	namespace = namespace_for_prefix(_prefix, _namespaces);
	if (_prefix != nil && namespace == nil)
		@throw [OFUnboundNamespaceException
		    exceptionWithClass: [self class]
				prefix: _prefix];

/* Looking for attribute name */
- (void)OF_parseInAttributeNameWithBuffer: (const char*)buffer
					i: (size_t*)i
				     last: (size_t*)last
{
	void *pool;
	OFMutableString *cacheString;
	const char *cacheCString, *tmp;
	size_t length, cacheLength;
	OFMutableString *bufferString;
	const char *bufferCString, *tmp;
	size_t length, bufferLength;

	if (buffer[*i] != '=')
		return;

	if ((length = *i - *last) > 0)
		cache_append(_cache, buffer + *last, _encoding, length);
		buffer_append(_buffer, buffer + *last, _encoding, length);

	pool = objc_autoreleasePoolPush();

	cacheString = [OFMutableString stringWithUTF8String: [_cache items]
						     length: [_cache count]];
	[cacheString deleteEnclosingWhitespaces];
	bufferString = [OFMutableString stringWithUTF8String: [_buffer items]
						      length: [_buffer count]];
	[bufferString deleteEnclosingWhitespaces];
	/* Prevent a useless copy later */
	[cacheString makeImmutable];
	[bufferString makeImmutable];

	cacheCString = [cacheString UTF8String];
	cacheLength = [cacheString UTF8StringLength];
	bufferCString = [bufferString UTF8String];
	bufferLength = [bufferString UTF8StringLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
	if ((tmp = memchr(bufferCString, ':', bufferLength)) != NULL) {
		_attributeName = [[OFString alloc]
		    initWithUTF8String: tmp + 1
				length: bufferLength -
				length: cacheLength - (tmp - cacheCString) - 1];
					(tmp - bufferCString) - 1];
		_attributePrefix = [[OFString alloc]
		    initWithUTF8String: cacheCString
				length: tmp - cacheCString];
		    initWithUTF8String: bufferCString
				length: tmp - bufferCString];
	} else {
		_attributeName = [cacheString copy];
		_attributeName = [bufferString copy];
		_attributePrefix = nil;
	}

	objc_autoreleasePoolPop(pool);

	[_cache removeAllItems];
	[_buffer removeAllItems];

	*last = *i + 1;
	_state = OF_XMLPARSER_EXPECT_DELIM;
}

/* Expecting delimiter */
- (void)OF_parseExpectDelimiterWithBuffer: (const char*)buffer

	OFString *attributeValue;
	size_t length;

	if (buffer[*i] != _delimiter)
		return;

	if ((length = *i - *last) > 0)
		cache_append(_cache, buffer + *last, _encoding, length);
		buffer_append(_buffer, buffer + *last, _encoding, length);

	pool = objc_autoreleasePoolPush();
	attributeValue = transform_string(_cache, 0, YES, self);
	attributeValue = transform_string(_buffer, 0, YES, self);

	if (_attributePrefix == nil && [_attributeName isEqual: @"xmlns"])
		[[_namespaces lastObject] setObject: attributeValue
					     forKey: @""];
	if ([_attributePrefix isEqual: @"xmlns"])
		[[_namespaces lastObject] setObject: attributeValue
					     forKey: _attributeName];

	[_attributes addObject:
	    [OFXMLAttribute attributeWithName: _attributeName
				    namespace: _attributePrefix
				  stringValue: attributeValue]];

	objc_autoreleasePoolPop(pool);

	[_cache removeAllItems];
	[_buffer removeAllItems];
	[_attributeName release];
	[_attributePrefix release];
	_attributeName = _attributePrefix = nil;

	*last = *i + 1;
	_state = OF_XMLPARSER_IN_TAG;
}

		_level = (buffer[*i] == ']' ? 1 : 0);

		return;
	}

	pool = objc_autoreleasePoolPush();

	cache_append(_cache, buffer + *last, _encoding, *i - *last);
	CDATA = transform_string(_cache, 2, NO, nil);
	buffer_append(_buffer, buffer + *last, _encoding, *i - *last);
	CDATA = transform_string(_buffer, 2, NO, nil);

	if ([_delegate respondsToSelector: @selector(parser:foundCDATA:)])
		[_delegate parser: self
		       foundCDATA: CDATA];

	objc_autoreleasePoolPop(pool);

	[_cache removeAllItems];
	[_buffer removeAllItems];

	*last = *i + 1;
	_state = OF_XMLPARSER_OUTSIDE_TAG;
}

/* Comment */
- (void)OF_parseInCommentOpeningWithBuffer: (const char*)buffer

	if (buffer[*i] != '>')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	pool = objc_autoreleasePoolPush();

	cache_append(_cache, buffer + *last, _encoding, *i - *last);
	comment = transform_string(_cache, 2, NO, nil);
	buffer_append(_buffer, buffer + *last, _encoding, *i - *last);
	comment = transform_string(_buffer, 2, NO, nil);

	if ([_delegate respondsToSelector: @selector(parser:foundComment:)])
		[_delegate parser: self
		     foundComment: comment];

	objc_autoreleasePoolPop(pool);

	[_cache removeAllItems];
	[_buffer removeAllItems];

	*last = *i + 1;
	_state = OF_XMLPARSER_OUTSIDE_TAG;
}

/* In <!DOCTYPE ...> */
- (void)OF_parseInDoctypeWithBuffer: (const char*)buffer