ObjFW  Diff

#include "config.h"

#define OF_XML_PARSER_M

#include <string.h>

#import "OFXMLParser.h"
#import "OFString.h"
#import "OFArray.h"
#import "OFArray.h"
#import "OFCharacterSet.h"
#import "OFDictionary.h"
#import "OFData.h"
#import "OFXMLAttribute.h"
#import "OFStream.h"
#import "OFDictionary.h"
#ifdef OF_HAVE_FILES
# import "OFFile.h"
#endif
#import "OFStream.h"
#import "OFString.h"
#import "OFSystemInfo.h"
#import "OFXMLAttribute.h"

#import "OFInitializationFailedException.h"
#import "OFInvalidArgumentException.h"
#import "OFInvalidEncodingException.h"
#import "OFInvalidFormatException.h"
#import "OFMalformedXMLException.h"
#import "OFOutOfRangeException.h"
#import "OFUnboundPrefixException.h"

@interface OFXMLParser () <OFStringXMLUnescapingDelegate>
@end

static void inByteOrderMarkState(OFXMLParser *);
static void outsideTagState(OFXMLParser *);
static void tagOpenedState(OFXMLParser *);
static void inProcessingInstructionsState(OFXMLParser *);
static void inProcessingInstructionState(OFXMLParser *);
static void inTagNameState(OFXMLParser *);
static void inCloseTagNameState(OFXMLParser *);
static void inTagState(OFXMLParser *);
static void inAttributeNameState(OFXMLParser *);
static void expectAttributeEqualSignState(OFXMLParser *);
static void expectAttributeDelimiterState(OFXMLParser *);
static void inAttributeValueState(OFXMLParser *);

static void inCommentState2(OFXMLParser *);
static void inDOCTYPEState(OFXMLParser *);
typedef void (*state_function_t)(OFXMLParser *);
static state_function_t lookupTable[] = {
	[OF_XMLPARSER_IN_BYTE_ORDER_MARK] = inByteOrderMarkState,
	[OF_XMLPARSER_OUTSIDE_TAG] = outsideTagState,
	[OF_XMLPARSER_TAG_OPENED] = tagOpenedState,
	[OF_XMLPARSER_IN_PROCESSING_INSTRUCTIONS] =
	[OF_XMLPARSER_IN_PROCESSING_INSTRUCTION] = inProcessingInstructionState,
	    inProcessingInstructionsState,
	[OF_XMLPARSER_IN_TAG_NAME] = inTagNameState,
	[OF_XMLPARSER_IN_CLOSE_TAG_NAME] = inCloseTagNameState,
	[OF_XMLPARSER_IN_TAG] = inTagState,
	[OF_XMLPARSER_IN_ATTRIBUTE_NAME] = inAttributeNameState,
	[OF_XMLPARSER_EXPECT_ATTRIBUTE_EQUAL_SIGN] =
	    expectAttributeEqualSignState,
	[OF_XMLPARSER_EXPECT_ATTRIBUTE_DELIMITER] =

	if (self->_finishedParsing && self->_data[self->_i] != '!' &&
	    self->_data[self->_i] != '?')
		@throw [OFMalformedXMLException exceptionWithParser: self];

	switch (self->_data[self->_i]) {
	case '?':
		self->_last = self->_i + 1;
		self->_state = OF_XMLPARSER_IN_PROCESSING_INSTRUCTIONS;
		self->_state = OF_XMLPARSER_IN_PROCESSING_INSTRUCTION;
		self->_level = 0;
		break;
	case '/':
		self->_last = self->_i + 1;
		self->_state = OF_XMLPARSER_IN_CLOSE_TAG_NAME;
		self->_acceptProlog = false;
		break;

		self->_i--;
		break;
	}
}

/* <?xml […]?> */
static bool
parseXMLProcessingInstructions(OFXMLParser *self, OFString *pi)
parseXMLProcessingInstruction(OFXMLParser *self, OFString *data)
{
	const char *cString;
	size_t length, last;
	int PIState = 0;
	OFString *attribute = nil;
	OFMutableString *value = nil;
	char piDelimiter = 0;
	bool hasVersion = false;

	if (!self->_acceptProlog)
		return false;

	self->_acceptProlog = false;

	pi = [pi substringFromIndex: 3];
	pi = pi.stringByDeletingEnclosingWhitespaces;

	cString = pi.UTF8String;
	length = pi.UTF8StringLength;
	cString = data.UTF8String;
	length = data.UTF8StringLength;

	last = 0;
	for (size_t i = 0; i < length; i++) {
		switch (PIState) {
		case 0:
			if (cString[i] == ' ' || cString[i] == '\t' ||
			    cString[i] == '\r' || cString[i] == '\n')

	if (PIState != 0 || !hasVersion)
		return false;

	return true;
}

/* Inside processing instructions */
/* Inside processing instruction */
static void
inProcessingInstructionsState(OFXMLParser *self)
inProcessingInstructionState(OFXMLParser *self)
{
	if (self->_data[self->_i] == '?')
		self->_level = 1;
	else if (self->_level == 1 && self->_data[self->_i] == '>') {
		void *pool = objc_autoreleasePoolPush();
		OFString *PI;
		OFString *PI, *target, *data = nil;
		OFCharacterSet *whitespaceCS;
		size_t pos;

		appendToBuffer(self->_buffer, self->_data + self->_last,
		    self->_encoding, self->_i - self->_last);
		PI = transformString(self, self->_buffer, 1, false);

		whitespaceCS = [OFCharacterSet
		    characterSetWithCharactersInString: @" \r\n\r"];
		pos = [PI indexOfCharacterFromSet: whitespaceCS];
		if (pos != OF_NOT_FOUND) {
			target = [PI substringToIndex: pos];
			data = [[PI substringFromIndex: pos + 1]
			    stringByDeletingEnclosingWhitespaces];
		if ([PI isEqual: @"xml"] || [PI hasPrefix: @"xml "] ||
		    [PI hasPrefix: @"xml\t"] || [PI hasPrefix: @"xml\r"] ||
		    [PI hasPrefix: @"xml\n"])
			if (!parseXMLProcessingInstructions(self, PI))

			if (data.length == 0)
				data = nil;
		} else
			target = PI;

		if ([target caseInsensitiveCompare: @"xml"] == OF_ORDERED_SAME)
			if (!parseXMLProcessingInstruction(self, data))
				@throw [OFMalformedXMLException
				    exceptionWithParser: self];

		if ([self->_delegate respondsToSelector:
		    @selector(parser:foundProcessingInstructions:)])
		if ([self->_delegate respondsToSelector: @selector(
		    parser:foundProcessingInstructionWithTarget:data:)])
			[self->_delegate parser: self
			    foundProcessingInstructions: PI];
			    foundProcessingInstructionWithTarget: target
							    data: data];

		objc_autoreleasePoolPop(pool);

		[self->_buffer removeAllItems];

		self->_last = self->_i + 1;
		self->_state = OF_XMLPARSER_OUTSIDE_TAG;

				length: tmp - bufferCString];
	} else {
		self->_name = [bufferString copy];
		self->_prefix = nil;
	}

	if (self->_data[self->_i] == '>' || self->_data[self->_i] == '/') {
		OFString *namespace;

		namespace = namespaceForPrefix(self->_prefix,
		OFString *namespace = namespaceForPrefix(self->_prefix,
		    self->_namespaces);

		if (self->_prefix != nil && namespace == nil)
			@throw [OFUnboundPrefixException
			    exceptionWithPrefix: self->_prefix
					 parser: self];