ObjFW  Diff

#include "config.h"

#define OF_XML_PARSER_M

#include <string.h>

#import "OFXMLParser.h"
#import "OFArray.h"
#import "OFCharacterSet.h"

#import "OFData.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 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 expectTagCloseState(OFXMLParser *);
static void expectSpaceOrTagCloseState(OFXMLParser *);
static void inExclamationMarkState(OFXMLParser *);
static void inCDATAOpeningState(OFXMLParser *);
static void inCDATAState(OFXMLParser *);
static void inCommentOpeningState(OFXMLParser *);
static void inCommentState1(OFXMLParser *);
static void inCommentState2(OFXMLParser *);
static void inDOCTYPEState(OFXMLParser *);
typedef void (*StateFunction)(OFXMLParser *);
static StateFunction lookupTable[] = {
	[OFXMLParserStateInByteOrderMark] = inByteOrderMarkState,
	[OFXMLParserStateOutsideTag] = outsideTagState,
	[OFXMLParserStateTagOpened] = tagOpenedState,
	[OFXMLParserStateInProcessingInstruction] =
	    inProcessingInstructionState,
	[OFXMLParserStateInTagName] = inTagNameState,
	[OFXMLParserStateInCloseTagName] = inCloseTagNameState,
	[OFXMLParserStateInTag] = inTagState,
	[OFXMLParserStateInAttributeName] = inAttributeNameState,
	[OFXMLParserStateExpectAttributeEqualSign] =
	    expectAttributeEqualSignState,
	[OFXMLParserStateExpectAttributeDelimiter] =
	    expectAttributeDelimiterState,
	[OFXMLParserStateInAttributeValue] = inAttributeValueState,
	[OFXMLParserStateExpectTagClose] = expectTagCloseState,
	[OFXMLParserStateExpectSpaceOrTagClose] = expectSpaceOrTagCloseState,
	[OFXMLParserStateInExclamationMark] = inExclamationMarkState,
	[OFXMLParserStateInCDATAOpening] = inCDATAOpeningState,
	[OFXMLParserStateInCDATA] = inCDATAState,
	[OFXMLParserStateInCommentOpening] = inCommentOpeningState,
	[OFXMLParserStateInComment1] = inCommentState1,
	[OFXMLParserStateInComment2] = inCommentState2,
	[OFXMLParserStateInDOCTYPE] = inDOCTYPEState
};

static OF_INLINE void
appendToBuffer(OFMutableData *buffer, const char *string,
    OFStringEncoding encoding, size_t length)
{
	if OF_LIKELY(encoding == OFStringEncodingUTF8)
		[buffer addItems: string count: length];
	else {
		void *pool = objc_autoreleasePoolPush();
		OFString *tmp = [OFString stringWithCString: string
						   encoding: encoding
						     length: length];
		[buffer addItems: tmp.UTF8String count: tmp.UTF8StringLength];

		dict = [OFMutableDictionary dictionaryWithKeysAndObjects:
		    @"xml", @"http://www.w3.org/XML/1998/namespace",
		    @"xmlns", @"http://www.w3.org/2000/xmlns/", nil];
		[_namespaces addObject: dict];

		_acceptProlog = true;
		_lineNumber = 1;
		_encoding = OFStringEncodingUTF8;
		_depthLimit = 32;

		objc_autoreleasePoolPop(pool);
	} @catch (id e) {
		[self release];
		@throw e;
	}

		if (_data[_i] == '\r' || (_data[_i] == '\n' &&
		    !_lastCarriageReturn))
			_lineNumber++;

		_lastCarriageReturn = (_data[_i] == '\r');
	}

	/* In OFXMLParserStateInTag, there can be only spaces */
	if (length - _last > 0 && _state != OFXMLParserStateInTag)
		appendToBuffer(_buffer, _data + _last, _encoding,
		    length - _last);
}

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

- (void)parseStream: (OFStream *)stream
{
	size_t pageSize = [OFSystemInfo pageSize];
	char *buffer = OFAllocMemory(1, pageSize);

	@try {
		while (!stream.atEndOfStream) {
			size_t length = [stream readIntoBuffer: buffer
							length: pageSize];
			[self parseBuffer: buffer length: length];
		}
	} @finally {
		OFFreeMemory(buffer);
	}
}

static void
inByteOrderMarkState(OFXMLParser *self)
{
	if (self->_data[self->_i] != "\xEF\xBB\xBF"[self->_level]) {
		if (self->_level == 0) {
			self->_state = OFXMLParserStateOutsideTag;
			self->_i--;
			return;
		}

		@throw [OFMalformedXMLException exceptionWithParser: self];
	}

	if (self->_level++ == 2)
		self->_state = OFXMLParserStateOutsideTag;

	self->_last = self->_i + 1;
}

/* Not in a tag */
static void
outsideTagState(OFXMLParser *self)

		objc_autoreleasePoolPop(pool);
	}

	[self->_buffer removeAllItems];

	self->_last = self->_i + 1;
	self->_state = OFXMLParserStateTagOpened;
}

/* Tag was just opened */
static void
tagOpenedState(OFXMLParser *self)
{
	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 = OFXMLParserStateInProcessingInstruction;
		self->_level = 0;
		break;
	case '/':
		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateInCloseTagName;
		self->_acceptProlog = false;
		break;
	case '!':
		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateInExclamationMark;
		self->_acceptProlog = false;
		break;
	default:
		if (self->_depthLimit > 0 &&
		    self->_previous.count >= self->_depthLimit)
			@throw [OFOutOfRangeException exception];

		self->_state = OFXMLParserStateInTagName;
		self->_acceptProlog = false;
		self->_i--;
		break;
	}
}

/* <?xml […]?> */
static bool
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;




	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')

				hasVersion = true;
			}

			if ([attribute isEqual: @"encoding"]) {
				@try {
					self->_encoding =
					    OFStringEncodingParseName(value);
				} @catch (OFInvalidArgumentException *e) {
					@throw [OFInvalidEncodingException
					    exception];
				}
			}

			last = i + 1;
			PIState = 0;

			break;
		}
	}

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

	return true;
}

/* Inside processing instruction */
static void
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, *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 != OFNotFound) {
			target = [PI substringToIndex: pos];
			data = [[PI substringFromIndex: pos + 1]
			    stringByDeletingEnclosingWhitespaces];

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

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

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

		objc_autoreleasePoolPop(pool);

		[self->_buffer removeAllItems];

		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateOutsideTag;
	} else
		self->_level = 0;
}

/* Inside a tag, no name yet */
static void
inTagNameState(OFXMLParser *self)

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

	if (self->_data[self->_i] == '>' || self->_data[self->_i] == '/') {


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

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

			[self->_previous addObject: bufferString];

		[self->_name release];
		[self->_prefix release];
		self->_name = self->_prefix = nil;

		self->_state = (self->_data[self->_i] == '/'
		    ? OFXMLParserStateExpectTagClose
		    : OFXMLParserStateOutsideTag);
	} else
		self->_state = OFXMLParserStateInTag;

	if (self->_data[self->_i] != '/')
		[self->_namespaces addObject: [OFMutableDictionary dictionary]];

	objc_autoreleasePoolPop(pool);

	[self->_buffer removeAllItems];

	[self->_namespaces removeLastObject];
	[self->_name release];
	[self->_prefix release];
	self->_name = self->_prefix = nil;

	self->_last = self->_i + 1;
	self->_state = (self->_data[self->_i] == '>'
	    ? OFXMLParserStateOutsideTag
	    : OFXMLParserStateExpectSpaceOrTagClose);

	if (self->_previous.count == 0)
		self->_finishedParsing = true;
}

/* Inside a tag, name found */
static void
inTagState(OFXMLParser *self)
{
	void *pool;
	OFString *namespace;
	OFXMLAttribute *const *attributesObjects;
	size_t attributesCount;

	if (self->_data[self->_i] != '>' && self->_data[self->_i] != '/') {
		if (self->_data[self->_i] != ' ' &&
		    self->_data[self->_i] != '\t' &&
		    self->_data[self->_i] != '\n' &&
		    self->_data[self->_i] != '\r') {
			self->_last = self->_i;
			self->_state = OFXMLParserStateInAttributeName;
			self->_i--;
		}

		return;
	}

	attributesObjects = self->_attributes.objects;

	[self->_name release];
	[self->_prefix release];
	[self->_attributes removeAllObjects];
	self->_name = self->_prefix = nil;

	self->_last = self->_i + 1;
	self->_state = (self->_data[self->_i] == '/'
	    ? OFXMLParserStateExpectTagClose : OFXMLParserStateOutsideTag);

}

/* Looking for attribute name */
static void
inAttributeNameState(OFXMLParser *self)
{
	void *pool;

	objc_autoreleasePoolPop(pool);

	[self->_buffer removeAllItems];

	self->_last = self->_i + 1;
	self->_state = (self->_data[self->_i] == '='
	    ? OFXMLParserStateExpectAttributeDelimiter
	    : OFXMLParserStateExpectAttributeEqualSign);
}

/* Expecting equal sign of an attribute */
static void
expectAttributeEqualSignState(OFXMLParser *self)
{
	if (self->_data[self->_i] == '=') {
		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateExpectAttributeDelimiter;
		return;
	}

	if (self->_data[self->_i] != ' '  && self->_data[self->_i] != '\t' &&
	    self->_data[self->_i] != '\n' && self->_data[self->_i] != '\r')
		@throw [OFMalformedXMLException exceptionWithParser: self];
}

	    self->_data[self->_i] == '\n' || self->_data[self->_i] == '\r')
		return;

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

	self->_delimiter = self->_data[self->_i];
	self->_state = OFXMLParserStateInAttributeValue;
}

/* Looking for attribute value */
static void
inAttributeValueState(OFXMLParser *self)
{
	void *pool;

	[self->_buffer removeAllItems];
	[self->_attributeName release];
	[self->_attributePrefix release];
	self->_attributeName = self->_attributePrefix = nil;

	self->_last = self->_i + 1;
	self->_state = OFXMLParserStateInTag;
}

/* Expecting closing '>' */
static void
expectTagCloseState(OFXMLParser *self)
{
	if (self->_data[self->_i] == '>') {
		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateOutsideTag;
	} else
		@throw [OFMalformedXMLException exceptionWithParser: self];
}

/* Expecting closing '>' or space */
static void
expectSpaceOrTagCloseState(OFXMLParser *self)
{
	if (self->_data[self->_i] == '>') {
		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateOutsideTag;
	} else if (self->_data[self->_i] != ' ' &&
	    self->_data[self->_i] != '\t' && self->_data[self->_i] != '\n' &&
	    self->_data[self->_i] != '\r')
		@throw [OFMalformedXMLException exceptionWithParser: self];
}

/* In <! */
static void
inExclamationMarkState(OFXMLParser *self)
{
	if (self->_finishedParsing && self->_data[self->_i] != '-')
		@throw [OFMalformedXMLException exceptionWithParser: self];

	if (self->_data[self->_i] == '-')
		self->_state = OFXMLParserStateInCommentOpening;
	else if (self->_data[self->_i] == '[') {
		self->_state = OFXMLParserStateInCDATAOpening;
		self->_level = 0;
	} else if (self->_data[self->_i] == 'D') {
		self->_state = OFXMLParserStateInDOCTYPE;
		self->_level = 0;
	} else
		@throw [OFMalformedXMLException exceptionWithParser: self];

	self->_last = self->_i + 1;
}

/* CDATA */
static void
inCDATAOpeningState(OFXMLParser *self)
{
	if (self->_data[self->_i] != "CDATA["[self->_level])
		@throw [OFMalformedXMLException exceptionWithParser: self];

	if (++self->_level == 6) {
		self->_state = OFXMLParserStateInCDATA;
		self->_level = 0;
	}

	self->_last = self->_i + 1;
}

static void

			[self->_delegate parser: self foundCDATA: CDATA];

		objc_autoreleasePoolPop(pool);

		[self->_buffer removeAllItems];

		self->_last = self->_i + 1;
		self->_state = OFXMLParserStateOutsideTag;
	} else
		self->_level = 0;
}

/* Comment */
static void
inCommentOpeningState(OFXMLParser *self)
{
	if (self->_data[self->_i] != '-')
		@throw [OFMalformedXMLException exceptionWithParser: self];

	self->_last = self->_i + 1;
	self->_state = OFXMLParserStateInComment1;
	self->_level = 0;
}

static void
inCommentState1(OFXMLParser *self)
{
	if (self->_data[self->_i] == '-')
		self->_level++;
	else
		self->_level = 0;

	if (self->_level == 2)
		self->_state = OFXMLParserStateInComment2;
}

static void
inCommentState2(OFXMLParser *self)
{
	void *pool;
	OFString *comment;

		[self->_delegate parser: self foundComment: comment];

	objc_autoreleasePoolPop(pool);

	[self->_buffer removeAllItems];

	self->_last = self->_i + 1;
	self->_state = OFXMLParserStateOutsideTag;
}

/* In <!DOCTYPE ...> */
static void
inDOCTYPEState(OFXMLParser *self)
{
	if ((self->_level < 6 &&
	    self->_data[self->_i] != "OCTYPE"[self->_level]) ||
	    (self->_level == 6 && self->_data[self->_i] != ' ' &&
	    self->_data[self->_i] != '\t' && self->_data[self->_i] != '\n' &&
	    self->_data[self->_i] != '\r'))
		@throw [OFMalformedXMLException exceptionWithParser: self];

	self->_level++;

	if (self->_level > 6 && self->_data[self->_i] == '>')
		self->_state = OFXMLParserStateOutsideTag;

	self->_last = self->_i + 1;
}

- (size_t)lineNumber
{
	return _lineNumber;