ObjFW  Diff

#include "config.h"

#include <string.h>

#import "OFXMLParser.h"
#import "OFString.h"
#import "OFArray.h"
#import "OFAutoreleasePool.h"
#import "OFExceptions.h"
#import "macros.h"

int _OFXMLParser_reference;

static OF_INLINE OFString*
transform_string(OFMutableString *cache,
    OFObject <OFXMLUnescapingDelegate> *handler)
{
	/* TODO: Support for xml:space */

	[cache removeLeadingAndTrailingWhitespaces];
	return [cache stringByXMLUnescapingWithHandler: handler];
}

static OF_INLINE OFString*
parse_numeric_entity(const char *entity, size_t length)
{
	of_unichar_t c;
	size_t i;
	char buf[5];

		return nil;
	buf[i] = 0;

	return [OFString stringWithCString: buf
				    length: i];
}

@implementation OFXMLParser
+ xmlParser
{
	return [[[self alloc] init] autorelease];
}

- init
{
	self = [super init];

	@try {
		cache = [[OFMutableString alloc] init];
		previous = [[OFMutableArray alloc] init];
	} @catch (OFException *e) {
		/* We can't use [super dealloc] on OS X here. Compiler bug? */
		[self dealloc];
		@throw e;
	}

	return self;
}

- (void)dealloc
{
	[delegate release];

	[cache release];
	[name release];
	[prefix release];
	[ns release];
	[attrs release];
	[attrName release];
	[attrPrefix release];
	[previous release];

	[super dealloc];
}

- (id)delegate
{
	return [[delegate retain] autorelease];
}

- (void)setDelegate: (OFObject <OFXMLParserDelegate>*)delegate_
{
	[delegate_ retain];
	[delegate release];
	delegate = delegate_;
}

- (void)parseBuffer: (const char*)buf
	   withSize: (size_t)size
{
	OFAutoreleasePool *pool;
	size_t i, last, len;

	last = 0;

	for (i = 0; i < size; i++) {
		switch (state) {
		/* Not in a tag */
		case OF_XMLPARSER_OUTSIDE_TAG:
			if (buf[i] == '<') {
				len = i - last;

				if (len > 0)
					[cache appendCString: buf + last
						  withLength: len];

				if ([cache cStringLength] > 0) {
					OFString *str;

					pool = [[OFAutoreleasePool alloc] init];
					str = transform_string(cache, self);
					[delegate xmlParser: self
						foundString: str];
					[pool release];
				}

				[cache setToCString: ""];

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

		/* Tag was just opened */
		case OF_XMLPARSER_TAG_OPENED:
			if (buf[i] == '/') {
				last = i + 1;
				state = OF_XMLPARSER_IN_CLOSE_TAG_NAME;
			} else if(buf[i] == '!') {
				last = i + 1;
				state = OF_XMLPARSER_IN_COMMENT_1;
			} else {
				state = OF_XMLPARSER_IN_TAG_NAME;
				i--;
			}
			break;

		/* Inside a tag, no name yet */
		case OF_XMLPARSER_IN_TAG_NAME:
			if (buf[i] == ' ' || buf[i] == '>' || buf[i] == '/') {
				const char *cache_c, *tmp;
				size_t cache_len;

				len = i - last;
				if (len > 0)
					[cache appendCString: buf + last
						  withLength: len];
				cache_c = [cache cString];
				cache_len = [cache cStringLength];

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

				if (buf[i] == '>' || buf[i] == '/') {
					pool = [[OFAutoreleasePool alloc] init];

					[delegate xmlParser: self
					didStartTagWithName: name
						     prefix: prefix
						  namespace: ns
						 attributes: nil];

					if (buf[i] == '/')
						[delegate xmlParser: self
						  didEndTagWithName: name
							     prefix: prefix
							  namespace: ns];
					else
						[previous addObject:
						    [[cache copy] autorelease]];

					[pool release];

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

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

				[cache setToCString: ""];
				last = i + 1;
			}
			break;

		/* Inside a close tag, no name yet */
		case OF_XMLPARSER_IN_CLOSE_TAG_NAME:
			if (buf[i] == ' ' || buf[i] == '>') {
				const char *cache_c, *tmp;
				size_t cache_len;

				len = i - last;
				if (len > 0)
					[cache appendCString: buf + last
						  withLength: len];
				cache_c = [cache cString];
				cache_len = [cache cStringLength];

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

				if (![[previous lastObject] isEqual: cache])
					@throw [OFMalformedXMLException
					    newWithClass: isa];
				[previous removeNObjects: 1];

				[cache setToCString: ""];

				pool = [[OFAutoreleasePool alloc] init];

				[delegate xmlParser: self
				  didEndTagWithName: name
					     prefix: prefix
					  namespace: ns];

				[pool release];

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

				last = i + 1;
				state = (buf[i] == ' '
				    ? OF_XMLPARSER_EXPECT_SPACE_OR_CLOSE
				    : OF_XMLPARSER_OUTSIDE_TAG);
			}
			break;

		/* Inside a tag, name found */
		case OF_XMLPARSER_IN_TAG:
			if (buf[i] == '>' || buf[i] == '/') {
				pool = [[OFAutoreleasePool alloc] init];

				[delegate xmlParser: self
				didStartTagWithName: name
					     prefix: prefix
					  namespace: ns
					 attributes: attrs];

				if (buf[i] == '/')
					[delegate xmlParser: self
					  didEndTagWithName: name
						     prefix: prefix
						  namespace: ns];
				else if (prefix != nil) {
					OFString *str = [OFString
					    stringWithFormat: @"%s:%s",
							      [prefix cString],
							      [name cString]];
					[previous addObject: str];
				} else
					[previous addObject: name];

				[pool release];

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

				last = i + 1;
				state = (buf[i] == '/'
				    ? OF_XMLPARSER_EXPECT_CLOSE
				    : OF_XMLPARSER_OUTSIDE_TAG);
			} else if (buf[i] != ' ') {
				last = i;
				state = OF_XMLPARSER_IN_ATTR_NAME;
				i--;
			}
			break;

		/* Looking for attribute name */
		case OF_XMLPARSER_IN_ATTR_NAME:
			if (buf[i] == '=') {
				const char *cache_c, *tmp;
				size_t cache_len;

				len = i - last;
				if (len > 0)
					[cache appendCString: buf + last
						  withLength: len];

				cache_c = [cache cString];
				cache_len = [cache cStringLength];

				if ((tmp = memchr(cache_c, ':',
				    cache_len)) != NULL ) {
					attrName = [[OFString alloc]
					    initWithCString: tmp + 1
						     length: cache_len - (tmp -
							     cache_c) - 1];
					attrPrefix = [[OFString alloc]
					    initWithCString: cache_c
						     length: tmp - cache_c];
				} else {
					attrName = [cache copy];
					attrPrefix = nil;
				}

				[cache setToCString: ""];

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

		/* Expecting delimiter */
		case OF_XMLPARSER_EXPECT_DELIM:
			if (buf[i] != '\'' && buf[i] != '"')
				@throw [OFMalformedXMLException
				    newWithClass: isa];

			delim = buf[i];
			last = i + 1;
			state = OF_XMLPARSER_IN_ATTR_VALUE;
			break;

		/* Looking for attribute value */
		case OF_XMLPARSER_IN_ATTR_VALUE:
			if (buf[i] == delim) {
				OFString *attr_val;

				len = i - last;
				if (len > 0)
					[cache appendCString: buf + last
						  withLength: len];

				if (attrs == nil)
					attrs = [[OFMutableArray alloc] init];

				pool = [[OFAutoreleasePool alloc] init];
				attr_val = [cache
				    stringByXMLUnescapingWithHandler: self];
				[attrs addObject: [OFXMLAttribute
				    attributeWithName: attrName
					       prefix: attrPrefix
					    namespace: nil
					  stringValue: attr_val]];
				[pool release];

				[cache setToCString: ""];
				[attrName release];
				[attrPrefix release];
				attrName = attrPrefix = nil;

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

		/* Expecting closing '>' */
		case OF_XMLPARSER_EXPECT_CLOSE:
			if (buf[i] == '>') {
				last = i + 1;
				state = OF_XMLPARSER_OUTSIDE_TAG;
			} else
				@throw [OFMalformedXMLException
				    newWithClass: isa];
			break;

		/* Expecting closing '>' or space */
		case OF_XMLPARSER_EXPECT_SPACE_OR_CLOSE:
			if (buf[i] == '>') {
				last = i + 1;
				state = OF_XMLPARSER_OUTSIDE_TAG;
			} else if (buf[i] != ' ')
				@throw [OFMalformedXMLException
				    newWithClass: isa];
			break;

		/* Comment */
		case OF_XMLPARSER_IN_COMMENT_1:
		case OF_XMLPARSER_IN_COMMENT_2:
			if (buf[i] != '-')
				@throw [OFMalformedXMLException
				    newWithClass: isa];
			last = i + 1;
			state++;
			break;
		case OF_XMLPARSER_IN_COMMENT_3:
			if (buf[i] == '-')
				state = OF_XMLPARSER_IN_COMMENT_4;
			break;
		case OF_XMLPARSER_IN_COMMENT_4:
			if (buf[i] == '-') {
				size_t cache_len;

				[cache appendCString: buf + last
					  withLength: i - last];
				cache_len = [cache length];

				pool = [[OFAutoreleasePool alloc] init];
				[cache removeCharactersFromIndex: cache_len - 1
							 toIndex: cache_len];
				[cache removeLeadingAndTrailingWhitespaces];
				[delegate xmlParser: self
				       foundComment: cache];
				[pool release];

				[cache setToCString: ""];

				last = i + 1;
				state = OF_XMLPARSER_EXPECT_CLOSE;
			} else
				state = OF_XMLPARSER_IN_COMMENT_3;

			break;
		}
	}

	len = size - last;
	/* In OF_XMLPARSER_IN_TAG, there can be only spaces */
	if (len > 0 && state != OF_XMLPARSER_IN_TAG)
		[cache appendCString: buf + last
			  withLength: len];
}

- (OFString*)foundUnknownEntityNamed: (OFString*)entity
{
	return [delegate xmlParser: self
	   foundUnknownEntityNamed: entity];
}
@end

@implementation OFString (OFXMLUnescaping)
- (OFString*)stringByXMLUnescaping
{
	return [self stringByXMLUnescapingWithHandler: nil];
}

- (OFString*)stringByXMLUnescapingWithHandler:

		@throw [OFInvalidEncodingException newWithClass: isa];

	[ret appendCStringWithoutUTF8Checking: string + last
				       length: i - last];

	return ret;
}
@end

@implementation OFObject (OFXMLParserDelegate)
-     (void)xmlParser: (OFXMLParser*)parser
  didStartTagWithName: (OFString*)name
	       prefix: (OFString*)prefix
	    namespace: (OFString*)ns
	   attributes: (OFArray*)attrs
{
}

-   (void)xmlParser: (OFXMLParser*)parser
  didEndTagWithName: (OFString*)name
	     prefix: (OFString*)prefix
	  namespace: (OFString*)ns
{
}

- (void)xmlParser: (OFXMLParser*)parser
      foundString: (OFString*)string
{
}

- (void)xmlParser: (OFXMLParser*)parser
     foundComment: (OFString*)comment
{
}

-    (OFString*)xmlParser: (OFXMLParser*)parser
  foundUnknownEntityNamed: (OFString*)entity
{
	return nil;
}
@end