ObjFW  Diff

@implementation OFXMLParser
+ (void)initialize
{
	size_t i;

	const SEL selectors_[] = {
		@selector(OF_parseOutsideTagWithBuffer:i:last:),
		@selector(OF_parseTagOpenedWithBuffer:i:last:),
		@selector(OF_parseInProcessingInstructionsWithBuffer:i:last:),
		@selector(OF_parseInTagNameWithBuffer:i:last:),
		@selector(OF_parseInCloseTagNameWithBuffer:i:last:),
		@selector(OF_parseInTagWithBuffer:i:last:),
		@selector(OF_parseInAttributeNameWithBuffer:i:last:),
		@selector(OF_parseExpectDelimiterWithBuffer:i:last:),
		@selector(OF_parseInAttributeValueWithBuffer:i:last:),
		@selector(OF_parseExpectCloseWithBuffer:i:last:),
		@selector(OF_parseExpectSpaceOrCloseWithBuffer:i:last:),
		@selector(OF_parseInExclamationMarkWithBuffer:i:last:),
		@selector(OF_parseInCDATAOpeningWithBuffer:i:last:),
		@selector(OF_parseInCDATA1WithBuffer:i:last:),
		@selector(OF_parseInCDATA2WithBuffer:i:last:),
		@selector(OF_parseInCommentOpeningWithBuffer:i:last:),
		@selector(OF_parseInComment1WithBuffer:i:last:),
		@selector(OF_parseInComment2WithBuffer:i:last:),
		@selector(OF_parseInDoctypeWithBuffer:i:last:),
	};
	memcpy(selectors, selectors_, sizeof(selectors_));

	for (i = 0; i < OF_XMLPARSER_NUM_STATES; i++) {
		if (![self instancesRespondToSelector: selectors[i]])
			@throw [OFInitializationFailedException
			    exceptionWithClass: self];

/*
 * The following methods handle the different states of the parser. They are
 * looked up in +[initialize] and put in a lookup table to speed things up.
 * One dispatch for every character would be way too slow!
 */

/* Not in a tag */
- (void)OF_parseOutsideTagWithBuffer: (const char*)buffer
				   i: (size_t*)i
				last: (size_t*)last
{
	size_t length;

	if ((finishedParsing || [previous count] < 1) && buffer[*i] != ' ' &&
	    buffer[*i] != '\t' && buffer[*i] != '\n' && buffer[*i] != '\r' &&
	    buffer[*i] != '<')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]

	[cache removeAllItems];

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

/* Tag was just opened */
- (void)OF_parseTagOpenedWithBuffer: (const char*)buffer
				  i: (size_t*)i
			       last: (size_t*)last
{
	if (finishedParsing && buffer[*i] != '!' && buffer[*i] != '?')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	switch (buffer[*i]) {
	case '?':

		acceptProlog = NO;
		(*i)--;
		break;
	}
}

/* <?xml […]?> */
- (BOOL)OF_parseXMLProcessingInstructions: (OFString*)pi
{
	const char *cString;
	size_t i, last, length;
	int piState = 0;
	OFString *attribute = nil;
	OFMutableString *value = nil;
	char piDelimiter = 0;

	if (piState != 0)
		return NO;

	return YES;
}

/* Inside processing instructions */
- (void)OF_parseInProcessingInstructionsWithBuffer: (const char*)buffer
						 i: (size_t*)i
					      last: (size_t*)last
{
	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);

		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];

		[delegate parser: self
		    foundProcessingInstructions: pi];

		objc_autoreleasePoolPop(pool);

		[cache 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;

	if (buffer[*i] != ' ' && buffer[*i] != '\t' && buffer[*i] != '\n' &&

	objc_autoreleasePoolPop(pool);

	[cache 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;
	OFString *ns;

	    : OF_XMLPARSER_EXPECT_SPACE_OR_CLOSE);

	if ([previous count] == 0)
		finishedParsing = YES;
}

/* Inside a tag, name found */
- (void)OF_parseInTagWithBuffer: (const char*)buffer
			      i: (size_t*)i
			   last: (size_t*)last
{
	void *pool;
	OFString *ns;
	OFXMLAttribute **attributesObjects;
	size_t j, attributesCount;

	if (buffer[*i] != '>' && buffer[*i] != '/') {

	*last = *i + 1;
	state = (buffer[*i] == '/'
	    ? OF_XMLPARSER_EXPECT_CLOSE
	    : OF_XMLPARSER_OUTSIDE_TAG);
}

/* 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;

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

	[cache removeAllItems];

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

/* Expecting delimiter */
- (void)OF_parseExpectDelimiterWithBuffer: (const char*)buffer
					i: (size_t*)i
				     last: (size_t*)last
{
	*last = *i + 1;

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

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

	delimiter = buffer[*i];
	state = OF_XMLPARSER_IN_ATTR_VALUE;
}

/* Looking for attribute value */
- (void)OF_parseInAttributeValueWithBuffer: (const char*)buffer
					 i: (size_t*)i
				      last: (size_t*)last
{
	void *pool;
	OFString *attributeValue;
	size_t length;

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

	attributeName = attributePrefix = nil;

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

/* Expecting closing '>' */
- (void)OF_parseExpectCloseWithBuffer: (const char*)buffer
				    i: (size_t*)i
				 last: (size_t*)last
{
	if (buffer[*i] == '>') {
		*last = *i + 1;
		state = OF_XMLPARSER_OUTSIDE_TAG;
	} else
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];
}

/* Expecting closing '>' or space */
- (void)OF_parseExpectSpaceOrCloseWithBuffer: (const char*)buffer
					   i: (size_t*)i
					last: (size_t*)last
{
	if (buffer[*i] == '>') {
		*last = *i + 1;
		state = OF_XMLPARSER_OUTSIDE_TAG;
	} else if (buffer[*i] != ' ' && buffer[*i] != '\t' &&
	    buffer[*i] != '\n' && buffer[*i] != '\r')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];
}

/* In <! */
- (void)OF_parseInExclamationMarkWithBuffer: (const char*)buffer
					  i: (size_t*)i
				       last: (size_t*)last
{
	if (finishedParsing && buffer[*i] != '-')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	if (buffer[*i] == '-')
		state = OF_XMLPARSER_IN_COMMENT_OPENING;

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

	*last = *i + 1;
}

/* CDATA */
- (void)OF_parseInCDATAOpeningWithBuffer: (const char*)buffer
				       i: (size_t*)i
				    last: (size_t*)last
{
	if (buffer[*i] != "CDATA["[level])
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	if (++level == 6) {
		state = OF_XMLPARSER_IN_CDATA_1;
		level = 0;
	}

	*last = *i + 1;
}

- (void)OF_parseInCDATA1WithBuffer: (const char*)buffer
				 i: (size_t*)i
			      last: (size_t*)last
{
	if (buffer[*i] == ']')
		level++;
	else
		level = 0;

	if (level == 2)
		state = OF_XMLPARSER_IN_CDATA_2;
}

- (void)OF_parseInCDATA2WithBuffer: (const char*)buffer
				 i: (size_t*)i
			      last: (size_t*)last
{
	void *pool;
	OFString *CDATA;

	if (buffer[*i] != '>') {
		state = OF_XMLPARSER_IN_CDATA_1;
		level = (buffer[*i] == ']' ? 1 : 0);

	[cache removeAllItems];

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

/* Comment */
- (void)OF_parseInCommentOpeningWithBuffer: (const char*)buffer
					 i: (size_t*)i
				      last: (size_t*)last
{
	if (buffer[*i] != '-')
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];

	*last = *i + 1;
	state = OF_XMLPARSER_IN_COMMENT_1;
	level = 0;
}

- (void)OF_parseInComment1WithBuffer: (const char*)buffer
				   i: (size_t*)i
				last: (size_t*)last
{
	if (buffer[*i] == '-')
		level++;
	else
		level = 0;

	if (level == 2)
		state = OF_XMLPARSER_IN_COMMENT_2;
}

- (void)OF_parseInComment2WithBuffer: (const char*)buffer
				   i: (size_t*)i
				last: (size_t*)last
{
	void *pool;
	OFString *comment;

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

	[cache removeAllItems];

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

/* In <!DOCTYPE ...> */
- (void)OF_parseInDoctypeWithBuffer: (const char*)buffer
				  i: (size_t*)i
			       last: (size_t*)last
{
	if ((level < 6 && buffer[*i] != "OCTYPE"[level]) ||
	    (level == 6 && buffer[*i] != ' ' && buffer[*i] != '\t' &&
	    buffer[*i] != '\n' && buffer[*i] != '\r'))
		@throw [OFMalformedXMLException exceptionWithClass: [self class]
							    parser: self];