[antlr-interest] Re: strings and vocab?
lgcraymer
lgc at mail1.jpl.nasa.gov
Tue Apr 13 11:11:40 PDT 2004
--- In antlr-interest at yahoogroups.com, ronald.petty at m... wrote:
> So in your example, if I have input "fake input ; some more fakeinput"
>
> //parser
> semi : ";"
>
> //lexer
> SEMI : ';'
>
>
> When it reads the ";" from the input stream SEMI is matched from the Lexer
> and not a LITERAL_;, this is because the Lexer is used first and sets the
> Token type right?
>
> So in the end does this mean that STRING literals are just tokens also?
Not quite, but close. There has to be a rule to match characters to build the token, usually called something like TEXT or ID, and the
literal is just a retyped one of those.
> The main difference is that there is no Lexer matching rule generated? And
Yes for 2.x.x. 3.x will probably generate matching rules.
> if there is no rule you need to test for the literal if you want to get a
> precreated Token type, correct? So where does the Token { } field in the
> grammar come into play. Is this just setting up more string literals? If
> so are the Tokens sections in Lexer or Parser mean the same thing? They
> both set up Tokens?
That is pretty much true.
--Loring
> Ugh. I need time to read the source code of Antlr.
>
> Thanks for helping.
> Ron
>
>
>
>
>
> "lgcraymer" <lgc at m...>
> 04/12/2004 06:01 PM
> Please respond to
> antlr-interest at yahoogroups.com
>
>
> To
> antlr-interest at yahoogroups.com
> cc
>
> Subject
> [antlr-interest] Re: strings and vocab?
>
>
>
>
>
>
> This one has to be thought of in implementation terms. For any lexer rule
> in which testLiterals is true: tokens are constructed and
> then checked against a hash table of literals. If the table contains a
> corresponding literal definition, then the token type is changed to
> match the literal; if not, it is given the default token type for that
> rule. Note that this is independent of the parser. I believe that the
> current implementation requires that all literals be defined in the same
> file as the lexer grammar.
>
> Rules for which testLiterals=false are not checked against the hash table.
> So if you have a rule
> SEMI : ':' ;
> and the literal ";" in the parser grammar, you will get strange
> results--the literal ";" has a different token type than the SEMI rule;
> since
> table lookup does not occur, you will never see the LITERAL_; value in the
> parser.
>
> --Loring
>
>
> --- In antlr-interest at yahoogroups.com, ronald.petty at m... wrote:
> > Alright, I give up :(. What is the secret to Antlr, jk. I am still
> > having some trouble getting started with Antlr, and I believe most of my
>
> > confusion comes from how strings/tokens/vocab is done.
> >
> > I was reading the java.g grammar and was wonding, in the parser there is
>
> > the rule
> >
> > builtInType
> > : "void"
> > | "boolean"
> > | "byte"
> > ..
> > ;
> >
> > Then in the Lexer there is
> >
> > IDENT
> > options { testLiterals=true; }
> > :
> ('a'..'z'|'A'..'Z'|'_'|'$')('a'..'z'|'A'..'Z'|'_'|'0'..'9'|'$')*
> > ;
> >
> > NUM_INT
> > {boolean isDecimal=false; Token t=null;}
> > : '.' {_ttype=DOT;}
> > ( ('0'..'9')+ (EXPONENT)? (f1:FLOAT_SUFFIX
> {t=f1;})?
> > {
> > ......
> >
> > protected
> > FLOAT_SUFFIX
> > : 'f'|'F'|'d'|'D'
> > ;
> >
> >
> > When the parser says, give me next token (nextToken), the Lexer will eat
>
> > the next token based on the Lexer rules. Now if the string "void" comes
>
> > in, the Lexer says, let me check if there is a literal yet for this
> token.
> > However I do not see what is going on here. The word "void" in the
> > parser may not have been seen yet (calling builtinType). I have read
> teh
> > vocab document, but still don't think I understand. I have tried using
> > tokens {} and don't understand why that works. Could someone explain
> > these simple concepts? I know I am missing something very simple here.
> I
> > can follow along the grammars just fine, but I don't understand real
> > workings on these issues, espically how or where you check Identifiers
> vs.
> > Keywords (I have read a dozen things, and none of them seem to explain
> it
> > in a way I can follow).
> >
> > Also does protected mean that the Lexer will never call FLOAT_SUFFIX
> > directly,if it is trying to get the nextToken, it will only try to get
> it
> > from the FLOAT_SUFFIX call in NUM_INT. Correct? Is this to keep
> similiar
> > issues like (IDENT vs Keywords) from happening?
> >
> > Thanks Ron
> >
> > ps. When I get this all figured out, I will write another tutorial
> > hopefully documenting the same issues I have, maybe help someone one day
>
> > :)
> >
> >
> **************************************************************************************
> > This communication is intended solely for the addressee and is
> > confidential. If you are not the intended recipient, any disclosure,
> > copying, distribution or any action taken or omitted to be taken in
> > reliance on it, is prohibited and may be unlawful. Unless indicated
> > to the contrary: it does not constitute professional advice or
> > opinions upon which reliance may be made by the addressee or any
> > other party, and it should be considered to be a work in progress.
> >
> **************************************************************************************
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>
> **************************************************************************************
> This communication is intended solely for the addressee and is
> confidential. If you are not the intended recipient, any disclosure,
> copying, distribution or any action taken or omitted to be taken in
> reliance on it, is prohibited and may be unlawful. Unless indicated
> to the contrary: it does not constitute professional advice or
> opinions upon which reliance may be made by the addressee or any
> other party, and it should be considered to be a work in progress.
> **************************************************************************************
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/antlr-interest/
<*> To unsubscribe from this group, send an email to:
antlr-interest-unsubscribe at yahoogroups.com
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
More information about the antlr-interest
mailing list