/*
    An HTML grammar for use with OkudaKit

    this grammar is intentionally very forgiving (non-strict)
    the point here is to highlight, not validate, HTML.
*/

@before {
        self.statementTerminator = nil;
        self.singleLineCommentMarker = nil;
        self.multiLineCommentStartMarker = @"<!--";
        self.multiLineCommentEndMarker = @"-->";
		self.blockStartMarker = nil;
		self.blockEndMarker = nil;
        self.braces = @"< > <!-- --> <? ?> <![CDATA[ ]]>";

        PKTokenizer *t = self.tokenizer;

        // symbols
        [t.symbolState add:@"<!--"];
        [t.symbolState add:@"-->"];
        [t.symbolState add:@"<?"];
        [t.symbolState add:@"?>"];
		[t.symbolState add:@"<![CDATA["];
		[t.symbolState add:@"]]>"];
		[t.symbolState add:@"<!DOCTYPE"];
		[t.symbolState add:@"]>"];

        // comments	
        [t setTokenizerState:t.commentState from:'<' to:'<'];
        [t.commentState addMultiLineStartMarker:@"<!--" endMarker:@"-->"];
        [t.commentState setFallbackState:t.delimitState from:'<' to:'<'];
        t.commentState.reportsCommentTokens = YES;

        // pi
        [t.delimitState addStartMarker:@"<?" endMarker:@"?>" allowedCharacterSet:nil];
        
        // doctype & cdata
        [t.delimitState addStartMarker:@"<![CDATA[" endMarker:@"]]>" allowedCharacterSet:nil];
        
        [t.delimitState setFallbackState:t.symbolState from:'<' to:'<'];
}

document        = any+;
any             = tag | dtd | cdata | pi | comment | text;

pi              = %{'<?', '?>'};
cdata           = %{'<![CDATA[', ']]>'};

text            = { ![LS(1) hasPrefix:@"<"] }? Any; // /[^<]+/;

tag		        = lt fwdSlash* tagName? (fwdSlash | attrNCName | string | num | tagSymbol)+ gt;

dtd             = doctype dtdContent* doctypeClose;

doctype         = '<!DOCTYPE';

internalDTDStartTag = '[';

doctypeClose    = internalDTDStartTag | gt;
dtdContent      = { !EQ(LS(1), @"[") && !EQ(LS(1), @">") }? Any;

tagName = elNCName (colon elNCName)?;
elNCName = Word;

string = QuotedString;
attrNCName = Word;

colon           = ':';
eq              = '=';
lt              = '<';
gt              = '>';
fwdSlash        = '/';

tagSymbol		= eq | colon | symbol;
symbol 			= Symbol;
num				= Number;

comment         = Comment;