ObjFW  Check-in [bbe98ea6c2]

#import "OFObject.h"
#import "OFString.h"
#import "OFXMLAttribute.h"

@class OFXMLParser;
@class OFArray;
@class OFMutableArray;
@class OFDataArray;
@class OFStream;

#if defined(OF_HAVE_PROPERTIES) && defined(OF_HAVE_BLOCKS)
typedef void (^of_xml_parser_processing_instructions_block_t)(
    OFXMLParser *parser, OFString *pi);
typedef void (^of_xml_parser_element_start_block_t)(OFXMLParser *parser,
    OFString *name, OFString *prefix, OFString *ns, OFArray *attributes);

		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;
	OFString *name;
	OFString *prefix;
	OFMutableArray *namespaces;
	OFMutableArray *attributes;
	OFString *attributeName;
	OFString *attributePrefix;
	char delimiter;

#include <string.h>
#include <unistd.h>

#import "OFXMLParser.h"
#import "OFString.h"
#import "OFArray.h"
#import "OFDictionary.h"
#import "OFDataArray.h"
#import "OFXMLAttribute.h"
#import "OFStream.h"
#import "OFFile.h"
#import "OFAutoreleasePool.h"

#import "OFInitializationFailedException.h"
#import "OFMalformedXMLException.h"
#import "OFUnboundNamespaceException.h"

#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,
    of_string_encoding_t encoding, size_t length)
{
	if (OF_LIKELY(encoding == OF_STRING_ENCODING_UTF_8))

		[cache addNItems: length
		      fromCArray: string];
	else {
		OFAutoreleasePool *pool = [[OFAutoreleasePool alloc] init];
		OFString *tmp = [OFString stringWithCString: string
						   encoding: encoding
						     length: length];
		[cache addNItems: [tmp cStringLength]
		      fromCArray: [tmp cString]];
		[pool release];
	}
}

static OFString*
transform_string(OFDataArray *cache, size_t cut, BOOL unescape,
    OFObject <OFStringXMLUnescapingDelegate> *delegate)
{
	OFMutableString *ret = [OFMutableString
	    stringWithCString: [cache cArray]
		       length: [cache count]];

	[ret replaceOccurrencesOfString: @"\r\n"
			     withString: @"\n"];
	[ret replaceOccurrencesOfString: @"\r"
			     withString: @"\n"];

	if (cut > 0) {
		size_t length = [ret length];

		[ret deleteCharactersFromIndex: length - cut
				       toIndex: length];
	}

	if (unescape)
		return [ret stringByXMLUnescapingWithDelegate: delegate];

	ret->isa = [OFString class];
	return ret;
}

static OFString*
namespace_for_prefix(OFString *prefix, OFArray *namespaces)
{
	OFDictionary **cArray = [namespaces cArray];
	ssize_t i;

{
	self = [super init];

	@try {
		OFAutoreleasePool *pool;
		OFMutableDictionary *dict;

		cache = [[OFBigDataArray alloc] init];
		previous = [[OFMutableArray alloc] init];
		namespaces = [[OFMutableArray alloc] init];

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

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

	if ((length = *i - *last) > 0)
		cache_append(cache, buffer + *last, encoding, length);

	if ([cache count] > 0) {
		OFString *characters;
		OFAutoreleasePool *pool;

		pool = [[OFAutoreleasePool alloc] init];
		characters = transform_string(cache, 0, YES, self);

#if defined(OF_HAVE_PROPERTIES) && defined(OF_HAVE_BLOCKS)
		if (charactersHandler != NULL)
			charactersHandler(self, characters);
		else
#endif
			[delegate parser: self
			 foundCharacters: characters];

		[pool release];
	}

	[cache removeNItems: [cache count]];

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

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

					parser: self];

		[delegate parser: self
		    foundProcessingInstructions: pi];

		[pool release];

		[cache removeNItems: [cache count]];

		*last = *i + 1;
		state = OF_XMLPARSER_OUTSIDE_TAG;
	} else
		level = 0;
}

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

	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);

	pool = [[OFAutoreleasePool alloc] init];

	cacheCString = [cache cArray];
	cacheLength = [cache count];
	cacheString = [OFString stringWithCString: cacheCString
					   length: cacheLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
		name = [[OFString alloc] initWithCString: tmp + 1
						  length: cacheLength -
							  (tmp - cacheCString) -
							  1];
		prefix = [[OFString alloc] initWithCString: cacheCString
						    length: tmp - cacheCString];
	} else {
		name = [cacheString copy];
		prefix = nil;
	}

	if (buffer[*i] == '>' || buffer[*i] == '/') {
		OFAutoreleasePool *pool2;
		OFString *ns;

		ns = namespace_for_prefix(prefix, namespaces);

		if (prefix != nil && ns == nil)
			@throw
			    [OFUnboundNamespaceException newWithClass: isa
							       prefix: prefix];

		pool2 = [[OFAutoreleasePool alloc] init];

#if defined(OF_HAVE_PROPERTIES) && defined(OF_HAVE_BLOCKS)
		if (elementStartHandler != NULL)
			elementStartHandler(self, name, prefix, ns, nil);
		else
#endif
			[delegate parser: self

				   didEndElement: name
				      withPrefix: prefix
				       namespace: ns];

			if ([previous count] == 0)
				finishedParsing = YES;
		} else
			[previous addObject: cacheString];

		[pool2 release];

		[name release];
		[prefix release];
		name = prefix = nil;

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


	[pool release];

	if (buffer[*i] != '/') {
		pool = [[OFAutoreleasePool alloc] init];
		[namespaces addObject: [OFMutableDictionary dictionary]];
		[pool release];
	}

	[cache removeNItems: [cache count]];
	*last = *i + 1;
}

/* Inside a close tag, no name yet */
- (void)_parseInCloseTagNameWithBuffer: (const char*)buffer
				     i: (size_t*)i
				  last: (size_t*)last
{
	OFAutoreleasePool *pool;
	const char *cacheCString, *tmp;
	size_t length, cacheLength;
	OFString *cacheString;
	OFString *ns;

	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);

	pool = [[OFAutoreleasePool alloc] init];

	cacheCString = [cache cArray];
	cacheLength = [cache count];
	cacheString = [OFString stringWithCString: cacheCString
					   length: cacheLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
		name = [[OFString alloc] initWithCString: tmp + 1
						  length: cacheLength -
							  (tmp - cacheCString) -
							  1];
		prefix = [[OFString alloc] initWithCString: cacheCString
						    length: tmp - cacheCString];
	} else {
		name = [cacheString copy];
		prefix = nil;
	}

	if (![[previous lastObject] isEqual: cacheString])
		@throw [OFMalformedXMLException newWithClass: isa
						      parser: self];

	[previous removeLastObject];

	[cache removeNItems: [cache count]];

	ns = namespace_for_prefix(prefix, namespaces);
	if (prefix != nil && ns == nil)
		@throw [OFUnboundNamespaceException newWithClass: isa
							  prefix: prefix];



#if defined(OF_HAVE_PROPERTIES) && defined(OF_HAVE_BLOCKS)
	if (elementEndHandler != NULL)
		elementEndHandler(self, name, prefix, ns);
	else
#endif
		[delegate parser: self
		   didEndElement: name

}

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

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

	if ((length = *i - *last) > 0)
		cache_append(cache, buffer + *last, encoding, length);

	pool = [[OFAutoreleasePool alloc] init];

	cacheString = [OFMutableString stringWithCString: [cache cArray]
						  length: [cache count]];
	[cacheString deleteEnclosingWhitespaces];
	/* Prevent a useless copy later */
	cacheString->isa = [OFString class];

	cacheCString = [cacheString cString];
	cacheLength = [cacheString cStringLength];

	if ((tmp = memchr(cacheCString, ':', cacheLength)) != NULL) {
		attributeName = [[OFString alloc]
		    initWithCString: tmp + 1
			     length: cacheLength - (tmp - cacheCString) - 1];
		attributePrefix = [[OFString alloc]
		    initWithCString: cacheCString
			     length: tmp - cacheCString];
	} else {
		attributeName = [cacheString copy];
		attributePrefix = nil;
	}

	[pool release];

	[cache removeNItems: [cache count]];

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

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

	[attributes addObject:
	    [OFXMLAttribute attributeWithName: attributeName
				    namespace: attributePrefix
				  stringValue: attributeValue]];

	[pool release];

	[cache removeNItems: [cache count]];
	[attributeName release];
	[attributePrefix release];
	attributeName = attributePrefix = nil;

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

	else
#endif
		[delegate parser: self
		      foundCDATA: CDATA];

	[pool release];

	[cache removeNItems: [cache count]];

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

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

	else
#endif
		[delegate parser: self
		    foundComment: comment];

	[pool release];

	[cache removeNItems: [cache count]];

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

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