ObjFW  Check-in [c53d661daf]

 * unescape XML. The callback is supposed to return a substitution for the
 * entity or nil if it is unknown to the callback as well, in which case an
 * exception will be thrown.
 *
 * \param entity The name of the entity that is unknown
 * \return A substitution for the entity or nil
 */
-	   (OFString*)string: (OFString*)str
  containsUnknownEntityNamed: (OFString*)entity;
@end

/**
 * \brief A category for unescaping XML in strings.
 */
@interface OFString (XMLUnescaping)
/**
 * Unescapes XML in the string.
 */
- (OFString*)stringByXMLUnescaping;

/**
 * Unescapes XML in the string and uses the specified handler for unknown
 * entities.
 *
 * \param h An OFXMLUnescapingDelegate as a handler for unknown entities
 */
- (OFString*)stringByXMLUnescapingWithDelegate:
    (OFObject <OFStringXMLUnescapingDelegate>*)delegate;
@end

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

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

- (OFString*)stringByXMLUnescapingWithDelegate:
    (OFObject <OFStringXMLUnescapingDelegate>*)delegate
{
	size_t i, last;
	BOOL in_entity;
	OFMutableString *ret;

	last = 0;
	in_entity = NO;

				if (tmp == nil)
					@throw [OFInvalidEncodingException
					    newWithClass: isa];

				[ret appendString: tmp];
				[pool release];
			} else if (delegate != nil) {
				OFAutoreleasePool *pool;
				OFString *n, *tmp;

				pool = [[OFAutoreleasePool alloc] init];

				n = [OFString stringWithCString: entity
							 length: len];
				tmp =	  [delegate string: self
				containsUnknownEntityNamed: n];

				if (tmp == nil)
					@throw [OFInvalidEncodingException
					    newWithClass: isa];

				[ret appendString: tmp];
				[pool release];

 *
 * \param parser The parser which found a new tag
 * \param name The name of the tag which just started
 * \param prefix The prefix of the tag which just started or nil
 * \param ns The namespace of the tag which just started or nil
 * \param attrs The attributes included in the tag which just started or nil
 */
-    (void)parser: (OFXMLParser*)parser
  didStartElement: (OFString*)name
       withPrefix: (OFString*)prefix
	namespace: (OFString*)ns
       attributes: (OFArray*)attrs;

/**
 * This callback is called when the XML parser found the end of a tag.
 *
 * \param parser The parser which found the end of a tag
 * \param name The name of the tag which just ended
 * \param prefix The prefix of the tag which just ended or nil
 * \param ns The namespace of the tag which just ended or nil
 */
-  (void)parser: (OFXMLParser*)parser
  didEndElement: (OFString*)name
     withPrefix: (OFString*)prefix
      namespace: (OFString*)ns;

/**
 * This callback is called when the XML parser found characters.
 *
 * In case there are comments or CDATA, it is possible that this callback is
 * called multiple times in a row.
 *
 * \param parser The parser which found a string
 * \param string The string the XML parser found
 */
-    (void)parser: (OFXMLParser*)parser
  foundCharacters: (OFString*)string;

/**
 * This callback is called when the XML parser found CDATA.
 *
 * \param parser The parser which found a string
 * \param string The string the XML parser found
 */
- (void)parser: (OFXMLParser*)parser
    foundCDATA: (OFString*)cdata;

/**
 * This callback is called when the XML parser found a comment.
 *
 * \param parser The parser which found a comment
 * \param comment The comment the XML parser found
 */
- (void)parser: (OFXMLParser*)parser
  foundComment: (OFString*)comment;

/**
 * This callback is called when the XML parser found an entity it doesn't know.
 * The callback is supposed to return a substitution for the entity or nil if
 * it is not known to the callback as well, in which case an exception will be
 * risen.
 *
 * \param parser The parser which found an unknown entity
 * \param entity The name of the entity the XML parser didn't know
 * \return A substitution for the entity or nil
 */
-	(OFString*)parser: (OFXMLParser*)parser
  foundUnknownEntityNamed: (OFString*)entity;
@end

/**
 * \brief An event-based XML parser.
 *
 * OFXMLParser is an event-based XML parser which calls the delegate's callbacks
 * as soon asit finds something, thus suitable for streams as well.

#import "OFXMLAttribute.h"
#import "OFAutoreleasePool.h"
#import "OFExceptions.h"
#import "macros.h"

static OF_INLINE OFString*
transform_string(OFMutableString *cache,
    OFObject <OFStringXMLUnescapingDelegate> *delegate)
{
	[cache replaceOccurrencesOfString: @"\r\n"
			       withString: @"\n"];
	[cache replaceOccurrencesOfString: @"\r"
			       withString: @"\n"];
	return [cache stringByXMLUnescapingWithDelegate: delegate];
}

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

					    length: len];

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

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

				[cache setToCString: ""];

				last = i + 1;
				state = OF_XMLPARSER_TAG_OPENED;

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

					[delegate parser: self
					 didStartElement: name
					      withPrefix: prefix
					       namespace: ns
					      attributes: nil];

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

					[pool release];

					[name release];

				ns = namespace_for_prefix(prefix, namespaces);
				if (prefix != nil && ns == nil)
					@throw [OFUnboundNamespaceException
					    newWithClass: isa
						  prefix: prefix];
				[namespaces removeNObjects: 1];

				[delegate parser: self
				   didEndElement: name
				      withPrefix: prefix
				       namespace: ns];

				[pool release];

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

				ns = namespace_for_prefix(prefix, namespaces);

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

				[delegate parser: self
				 didStartElement: name
				      withPrefix: prefix
				       namespace: ns
				      attributes: attrs];

				if (buf[i] == '/') {
					[delegate parser: self
					   didEndElement: name
					      withPrefix: prefix
					       namespace: ns];
					[namespaces removeNObjects: 1];
				} else if (prefix != nil) {
					OFString *str = [OFString
					    stringWithFormat: @"%s:%s",
							      [prefix cString],
							      [name cString]];
					[previous addObject: str];

				    appendCStringWithoutUTF8Checking: buf + last
							      length: i - last];
				cdata = [[cache mutableCopy] autorelease];
				len = [cdata length];

				[cdata removeCharactersFromIndex: len - 2
							 toIndex: len];
				[delegate parser: self
				      foundCDATA: cdata];
				[pool release];

				[cache setToCString: ""];

				last = i + 1;
				state = OF_XMLPARSER_OUTSIDE_TAG;
			} else if (buf[i] != ']')

				    appendCStringWithoutUTF8Checking: buf + last
							      length: i - last];
				comment = [[cache mutableCopy] autorelease];
				len = [comment length];

				[comment removeCharactersFromIndex: len - 2
							   toIndex: len];
				[delegate parser: self
				    foundComment: comment];
				[pool release];

				[cache setToCString: ""];

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

	len = size - last;
	/* In OF_XMLPARSER_IN_TAG, there can be only spaces */
	if (len > 0 && state != OF_XMLPARSER_IN_TAG)
		[cache appendCStringWithoutUTF8Checking: buf + last
						 length: len];
}

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

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

-  (void)parser: (OFXMLParser*)parser
  didEndElement: (OFString*)name
     withPrefix: (OFString*)prefix
      namespace: (OFString*)ns
{
}

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

- (void)parser: (OFXMLParser*)parser
    foundCDATA: (OFString*)cdata
{
}

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

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

	@" \t\t  \t\t  \t \t"
};

@interface EntityHandler: OFObject <OFStringXMLUnescapingDelegate>
@end

@implementation EntityHandler
-	   (OFString*)string: (OFString*)string
  containsUnknownEntityNamed: (OFString*)entity
{
	if ([entity isEqual: @"foo"])
		return @"bar";

	return nil;
}
@end

	EXPECT_EXCEPTION(@"Detect invalid entities in -[stringByXMLUnescaping] "
	    @"#5", OFInvalidEncodingException, [@"&#g;" stringByXMLUnescaping])
	EXPECT_EXCEPTION(@"Detect invalid entities in -[stringByXMLUnescaping] "
	    @"#6", OFInvalidEncodingException, [@"&#xg;" stringByXMLUnescaping])

	TEST(@"-[stringByXMLUnescapingWithHandler:]",
	    (h = [[[EntityHandler alloc] init] autorelease]) &&
	    [[@"x&foo;y" stringByXMLUnescapingWithDelegate: h]
	    isEqual: @"xbary"])

	[pool drain];
}
@end

static OFString *module = @"OFXMLParser";
static int i = 0;

enum event_type {
	TAG_START,
	TAG_END,
	STRING,
	CDATA,
	COMMENT
};

@implementation TestsAppDelegate (OFXMLParser)
- (void)parserCallbackWithEventType: (enum event_type)et
			       name: (OFString*)name
			     prefix: (OFString*)prefix
			  namespace: (OFString*)ns
			 attributes: (OFArray*)attrs
			     string: (OFString*)string
			    comment: (OFString*)comment
{
	OFString *msg;

	i++;
	msg = [OFString stringWithFormat: @"Parsing part #%d", i];

	switch (i) {
	case 1:
		TEST(msg, et == TAG_START && [name isEqual: @"root"] &&
		    prefix == nil && ns == nil && [attrs count] == 0)
		break;
	case 2:
		TEST(msg, et == STRING && [string isEqual: @"\n "])
		break;
	case 3:
		TEST(msg, et == CDATA && [string isEqual: @"f<oo"])
		break;
	case 4:
		TEST(msg, et == TAG_START && [name isEqual: @"bar"] &&
		    prefix == nil && ns == nil && attrs == nil)
		break;
	case 5:
		TEST(msg, et == TAG_END && [name isEqual: @"bar"] &&

	case 30:
		TEST(msg, et == TAG_END && [name isEqual: @"root"] &&
		    prefix == nil && ns == nil);
		break;
	}
}

-    (void)parser: (OFXMLParser*)parser
  didStartElement: (OFString*)name
       withPrefix: (OFString*)prefix
	namespace: (OFString*)ns
       attributes: (OFArray*)attrs
{
	[self parserCallbackWithEventType: TAG_START
				     name: name
				   prefix: prefix
				namespace: ns
			       attributes: attrs
				   string: nil
				  comment: nil];
}

-  (void)parser: (OFXMLParser*)parser
  didEndElement: (OFString*)name
     withPrefix: (OFString*)prefix
      namespace: (OFString*)ns
{
	[self parserCallbackWithEventType: TAG_END
				     name: name
				   prefix: prefix
				namespace: ns
			       attributes: nil
				   string: nil
				  comment: nil];
}

-    (void)parser: (OFXMLParser*)parser
  foundCharacters: (OFString*)string
{
	[self parserCallbackWithEventType: STRING
				     name: nil
				   prefix: nil
				namespace: nil
			       attributes: nil
				   string: string
				  comment: nil];
}

- (void)parser: (OFXMLParser*)parser
    foundCDATA: (OFString*)cdata
{
	[self parserCallbackWithEventType: CDATA
				     name: nil
				   prefix: nil
				namespace: nil
			       attributes: nil
				   string: cdata
				  comment: nil];
}

- (void)parser: (OFXMLParser*)parser
  foundComment: (OFString*)comment
{
	[self parserCallbackWithEventType: COMMENT
				     name: nil
				   prefix: nil
				namespace: nil
			       attributes: nil
				   string: nil
				  comment: comment];
}

-	(OFString*)parser: (OFXMLParser*)parser
  foundUnknownEntityNamed: (OFString*)entity
{
	if ([entity isEqual: @"foo"])
		return @"foobar";

	return nil;
}