parser.jjt

velocity 的脚本语言的全部代码集合

        if (!inComment)
        {
            lparen++;

            /*
             * If in REFERENCE and we have seen the dot, then move
             * to REFMOD2 -> Modifier()
             */

            if (curLexState == REFMODIFIER )
                SwitchTo( REFMOD2 );
        }
   }
}

<*>
MORE :
{
    /*
     *   Note : DOLLARBANG is a duplicate of DOLLAR.  They must be identical.
     */

    <DOLLAR: ("\\")* "$">
    {
        if (! inComment)
        {
            /*
             * if we find ourselves in REFERENCE, we need to pop down
             * to end the previous ref
             */

            if (curLexState == REFERENCE)
            {
                inReference = false;
                stateStackPop();
            }

            inReference = true;

            if ( debugPrint )
                System.out.print( "$  : going to " + REFERENCE );

            stateStackPush();
            SwitchTo(REFERENCE);
        }
    }

|   <DOLLARBANG: ("\\")* "$" ("\\")* "!">
    {
        if (! inComment)
        {
            /*
             * if we find ourselves in REFERENCE, we need to pop down
             * to end the previous ref
             */

            if (curLexState == REFERENCE)
            {
                inReference = false;
                stateStackPop();
            }

            inReference = true;

            if ( debugPrint )
                System.out.print( "$!  : going to " + REFERENCE );

            stateStackPush();
            SwitchTo(REFERENCE);
        }
    }

|   "##"
   {
        if (!inComment)
        {
            if (curLexState == REFERENCE)
            {
                inReference = false;
                stateStackPop();
            }

            inComment = true;
            stateStackPush();
            SwitchTo(IN_SINGLE_LINE_COMMENT);
        }
     }

|   <"#**" ~["#"]>
    {
        input_stream.backup(1);
        inComment = true;
        stateStackPush();
        SwitchTo( IN_FORMAL_COMMENT);
    }

|   "#*"
    {
        inComment=true;
        stateStackPush();
        SwitchTo( IN_MULTI_LINE_COMMENT );
    }

|   <HASH : "#" >
    {
        if (! inComment)
        {
            /*
             * We can have the situation where #if($foo)$foo#end.
             * We need to transition out of REFERENCE before going to DIRECTIVE.
             * I don't really like this, but I can't think of a legal way
             * you are going into DIRECTIVE while in REFERENCE.  -gmj
             */

            if (curLexState == REFERENCE || curLexState == REFMODIFIER )
            {
                inReference = false;
                stateStackPop();
            }

            inDirective = true;

            if ( debugPrint )
                System.out.print("# :  going to " + DIRECTIVE );

            stateStackPush();
            SwitchTo(PRE_DIRECTIVE);
        }
    }
}

TOKEN :
{
    <DOUBLE_ESCAPE : "\\\\">
|   <ESCAPE: "\\" >
|   <TEXT: (~["$", "#", "\\"])+ >
}

/* -----------------------------------------------------------------------
 *
 *   *_COMMENT Lexical tokens
 *
 *-----------------------------------------------------------------------*/
<IN_SINGLE_LINE_COMMENT>
TOKEN :
{
  <SINGLE_LINE_COMMENT: "\n" | "\r" | "\r\n" >
  {
     inComment = false;
     stateStackPop();
  }

}

<IN_FORMAL_COMMENT>
TOKEN :
{
  <FORMAL_COMMENT: "*#" >
  {
    inComment = false;
    stateStackPop();
  }
}

<IN_MULTI_LINE_COMMENT>
TOKEN :
{
  <MULTI_LINE_COMMENT: "*#" >
  {
    inComment = false;
    stateStackPop();
  }
}

<IN_SINGLE_LINE_COMMENT,IN_FORMAL_COMMENT,IN_MULTI_LINE_COMMENT>
MORE :
{
  < ~[] >
}

/* -----------------------------------------------------------------------
 *
 *  DIRECTIVE Lexical State (some of it, anyway)
 *
 * ---------------------------------------------------------------------- */

<DIRECTIVE,REFMOD2>
TOKEN:
{
    <WHITESPACE : ([" ","\t"])+ >
}

<DIRECTIVE,REFMOD2>
TOKEN :
{
//   <STRING_LITERAL: ( "\"" ( ~["\"","\n","\r"] )* "\"" ) | ( "'" ( ~["'","\n","\r"] )* "'" ) >
  < STRING_LITERAL:
      ("\""
        (   (~["\"","\\","\n","\r"])
          | ("\\"
              ( ["n","t","b","r","f","\\","'","\""]
              | ["0"-"7"] ( ["0"-"7"] )?
              | ["0"-"3"] ["0"-"7"] ["0"-"7"]
              )
            )
          | ( "\\" (" ")* "\n")
        )*
        "\""
      )
    |
    ("\'"
        (   (~["\'","\n","\r"])
          | ( "\\" (" ")* "\n")
        )*
        "\'"
      )
  >

    {
        /*
         *  - if we are in DIRECTIVE and haven't seen ( yet, then also drop out.
         *      don't forget to account for the beloved yet wierd #set
         *  - finally, if we are in REFMOD2 (remember : $foo.bar( ) then " is ok!
         */

         if( curLexState == DIRECTIVE && !inSet && lparen == 0)
            stateStackPop();
    }
}

<REFERENCE,DIRECTIVE,REFMODIFIER,REFMOD2>
TOKEN:
{
   <TRUE: "true">
|   <FALSE: "false">
}

<DIRECTIVE>
TOKEN :
{
    <NEWLINE: "\n" | "\r" | "\r\n" >
    {
        if ( debugPrint )
            System.out.println(" NEWLINE :");

        stateStackPop();

        if (inSet)
            inSet = false;

        if (inDirective)
            inDirective = false;
    }
}


<DIRECTIVE>
TOKEN :
{
    <MINUS: "-">
|   <PLUS: "+">
|   <MULTIPLY: "*">
|   <DIVIDE: "/">
|   <MODULUS: "%">
|   <LOGICAL_AND: "&&">
|   <LOGICAL_OR: "||">
|   <LOGICAL_LT: "<">
|   <LOGICAL_LE: "<=">
|   <LOGICAL_GT: ">">
|   <LOGICAL_GE: ">=">
|   <LOGICAL_EQUALS: "==">
|   <LOGICAL_NOT_EQUALS: "!=">
|   <LOGICAL_NOT: "!">
|   <EQUALS: "=" >
}

<PRE_DIRECTIVE>
TOKEN :
{
    <END: "end" ( ( " " | "\t" )* ( "\n" | "\r" | "\r\n" ) )? >
    {
        inDirective = false;
        stateStackPop();
    }

|   <IF_DIRECTIVE: "if">
    {
        SwitchTo(DIRECTIVE);
    }

|   <ELSEIF_DIRECTIVE: "elseif">
    {
        SwitchTo(DIRECTIVE);
    }

|   <ELSE_DIRECTIVE: "else" ( ( " " | "\t" )* ( "\n" | "\r" | "\r\n" ) )?  >
     {
        inDirective = false;
        stateStackPop();
    }

|   <STOP_DIRECTIVE: "stop">
    {
        matchedToken.kind = EOF;
        fileDepth = 0;
    }
}

<PRE_DIRECTIVE,DIRECTIVE,REFMOD2>
TOKEN:
{
   <#DIGIT: [ "0"-"9" ] >
|   <NUMBER_LITERAL: ("-")? (<DIGIT>)+ >
    {
        /*
         * check to see if we are in set
         *    ex.  #set $foo = $foo + 3
         *  because we want to handle the \n after
         */

        if ( lparen == 0 && !inSet && curLexState != REFMOD2)
        {
            stateStackPop();
        }

    }
}

<PRE_DIRECTIVE,DIRECTIVE>
TOKEN:
{
    <#LETTER: [ "a"-"z", "A" - "Z" ] >
|   <#DIRECTIVE_CHAR: [ "a"-"z", "A"-"Z", "0"-"9", "_" ] >
|   <WORD:  ( <LETTER> | ["_"]) (<DIRECTIVE_CHAR>)* >

}

/* -----------------------------------------------------------------------
 *
 *  REFERENCE Lexical States
 *
 *  This is more than a single state, because of the  structure of
 *  the VTL references.  We use three states because the set of tokens
 *  for each state can be different.
 *
 *  $foo.bar( "arg" )
 *  ^   ^   ^
 *  |   |   |
 *  ----------- >  REFERENCE : state initiated by the '$' character.  Continues
 *      |   |       until end of the reference, or the . character.
 *      |------ >  REFMODIFIER : state switched to when the <DOT> is encountered.
 *          |       note that this is a switch, not a push. See notes at bottom
 *          |       re stateStack.
 *          |-- >  REFMOD2 : state switch to when the LPAREN is encountered.
 *                  again, this is a switch, not a push.
 *
 * ---------------------------------------------------------------------------- */

<REFERENCE,REFMODIFIER,REFMOD2>
TOKEN :
{
    <#ALPHA_CHAR: ["a"-"z", "A"-"Z"] >
|   <#ALPHANUM_CHAR: [ "a"-"z", "A"-"Z", "0"-"9" ] >
|   <#IDENTIFIER_CHAR: [ "a"-"z", "A"-"Z", "0"-"9", "-", "_" ] >
|   <IDENTIFIER:  ( <ALPHA_CHAR> | ["_"]) (<IDENTIFIER_CHAR>)* >
|   <DOT: "." <ALPHA_CHAR>>
    {
        /*
         * push the alpha char back into the stream so the following identifier
         * is complete
         */

        input_stream.backup(1);

        /*
         * and munge the <DOT> so we just get a . when we have normal text that
         * looks like a ref.ident
         */

        matchedToken.image = ".";

        if ( debugPrint )
            System.out.print("DOT : switching to " + REFMODIFIER);
        SwitchTo(REFMODIFIER);

    }
|   <LCURLY: "{">
|   <RCURLY: "}">
    {
        stateStackPop();
    }
}

<REFERENCE,REFMODIFIER,REFMOD2>
SPECIAL_TOKEN :
{
    <REFERENCE_TERMINATOR: ~[] >
    {
        /*
         * push every terminator character back into the stream
         */

        input_stream.backup(1);

        inReference = false;

        if ( debugPrint )
            System.out.print("REF_TERM :");

        stateStackPop();
    }
}

<PRE_DIRECTIVE>
SPECIAL_TOKEN :
{
    <DIRECTIVE_TERMINATOR: ~[] >
    {
        if ( debugPrint )
            System.out.print("DIRECTIVE_TERM :");

        input_stream.backup(1);
        inDirective = false;
        stateStackPop();
    }
}

/**
 * This method is what starts the whole parsing
 * process. After the parsing is complete and
 * the template has been turned into an AST,
 * this method returns the root of AST which
 * can subsequently be traversed by a visitor
 * which implements the ParserVisitor interface
 * which is generated automatically by JavaCC
 */
SimpleNode process() : {}
{
   ( Statement() )* <EOF>
   { return jjtThis; }
}

/**
 * These are the types of statements that
 * are acceptable in Velocity templates.
 */
void Statement() #void : {}
{
    IfStatement()
|   StopStatement()
|   LOOKAHEAD(2) Reference()
|   Comment()
|   SetDirective()
|   EscapedDirective()
|   Escape()
|   Directive()
|   Text()
}

/**
 *  used to separate the notion of a valid directive that has been
 *  escaped, versus something that looks like a directive and
 *  is just schmoo.  This is important to do as a separate production
 *  that creates a node, because we want this, in either case, to stop
 *  the further parsing of the Directive() tree.
 */
void EscapedDirective() : {}
{
    {
        Token t = null;
    }

    t = <ESCAPE_DIRECTIVE>
    {
        /*
         *  churn and burn..
         */
        t.image = escapedDirective( t.image );
    }
}

/**
 *  Used to catch and process escape sequences in grammatical constructs
 *  as escapes outside of VTL are just characters.  Right now we have both
 *  this and the EscapeDirective() construction because in the EscapeDirective()
 *  case, we want to suck in the #<directive> and here we don't.  We just want
 *  the escapes to render correctly
 */
void Escape() : {}
{
    {
        Token t = null;
        int count = 0;
        boolean control = false;
    }

   ( LOOKAHEAD(2)  t = <DOUBLE_ESCAPE>
    {
        count++;
    }
   )+
    {
        /*
         * first, check to see if we have a control directive
         */
        switch(t.next.kind ) {
            case IF_DIRECTIVE :
            case ELSE_DIRECTIVE :
            case ELSEIF_DIRECTIVE :
            case END :
            case STOP_DIRECTIVE :
                control = true;
                break;
        }

        /*
         * if that failed, lets lookahead to see if we matched a PD or a VM
         */

        if ( isDirective( t.next.image.substring(1)))
            control = true;
        else if ( rsvc.isVelocimacro( t.next.image.substring(1), currentTemplateName))
            control = true;

        jjtThis.val = "";

        for( int i = 0; i < count; i++)
            jjtThis.val += ( control ? "\\" : "\\\\");
    }

}

void Comment() : {}
{
    <SINGLE_LINE_COMMENT>
|   <MULTI_LINE_COMMENT>
|   <FORMAL_COMMENT>
}

void NumberLiteral() : {}
{
    <NUMBER_LITERAL>
}

void StringLiteral() : {}
{
    <STRING_LITERAL>
}

/**
 * This method corresponds to variable
 * references in Velocity templates.
 * The following are examples of variable
 * references that may be found in a
 * template:
 *
 * $foo
 * $bar
 *
 */
void Identifier() : {}
{
    <IDENTIFIER>
}

void Word() : {}
{
    <WORD>
}

/**
 *   Supports the arguments for the Pluggable Directives