Eclipse Community Forums: TMF (Xtext)

Help

Home

Home » Modeling » TMF (Xtext) » Antlr version

Show: Today's Messages :: Show Polls :: Message Navigator

Antlr version [message #1694827]

Fri, 08 May 2015 09:09

Anakreontas Mentis

Messages: 85
Registered: October 2014

Member

Hello.

Which version of antlr 3 does XText uses?

Report message to a moderator

Re: Antlr version [message #1694976 is a reply to message #1694827]

Mon, 11 May 2015 07:28

Sebastian Zarnekow

Messages: 3118
Registered: July 2009

Senior Member

It's a patched version of Antlr 3.2.1 to allow to run with Java8, too.

Why do you ask?

Best,
Sebastian
--
Looking for professional support for Xtext, Xtend or Eclipse Modeling?
Find help at http://xtext.itemis.com or xtext(@)itemis.com
Blog: zarnekow.blogspot.com
Twitter: @szarnekow
Google+: https://www.google.com/+SebastianZarnekow

Report message to a moderator

Re: Antlr version [message #1694977 is a reply to message #1694976]

Mon, 11 May 2015 07:36

Anakreontas Mentis

Messages: 85
Registered: October 2014

Member

I planned to implement a custom lexer and wanted to avoid using features of antlr not supported by the version XText uses.

Report message to a moderator

Re: Antlr version [message #1695028 is a reply to message #1694977]

Mon, 11 May 2015 13:12

Sebastian Zarnekow

Messages: 3118
Registered: July 2009

Senior Member

Could you outline the reason why you need a custom lexer or is that
beyond this question?

Best,
Sebastian
--
Looking for professional support for Xtext, Xtend or Eclipse Modeling?
Find help at http://xtext.itemis.com or xtext(@)itemis.com
Blog: zarnekow.blogspot.com
Twitter: @szarnekow
Google+: https://www.google.com/+SebastianZarnekow

Report message to a moderator

Re: Antlr version [message #1695041 is a reply to message #1695028]

Mon, 11 May 2015 14:39

Anakreontas Mentis

Messages: 85
Registered: October 2014

Member

The DSL I'm working on contains sections with different lexing rules. Keywords of one section might be identifiers in another one. My first approach was to include lexer modes in a custom lexer. This does not work since modes are not supported in Antlr3. The second approach was to disable some lexer rules based on a predicate. However, I discovered that syntax highlighting is disabled for keywords that are not described as literal strings in the parser even when I do not use a custom lexer.

For example if I change a rule in the parser:
Field : 'field' ',' ...
to
Field : T_FIELD T_COMMA
where T_FIELD is a lexer rule defined as
T_FIELD : 'field' ;

keyword field is no longer highlighted and the lexer rule is case sensitive even though I have set the parser fragment option:
fragment = org.eclipse.xtext.generator.parser.antlr.ex.rt.AntlrGeneratorFragment {
options = {
ignoreCase = true
//backtrackLexer = true
}
}

In a custom lexer the case sensitive matching would work but I do not want to lose syntax highlighting.

Finally I settled for the sub-optimal solution where I define a parser for both languages that recognizes as keywords the union of keywords applicable to each language.

Report message to a moderator

Re: Antlr version [message #1695282 is a reply to message #1695041]

Wed, 13 May 2015 09:52

Sebastian Zarnekow

Messages: 3118
Registered: July 2009

Senior Member

If you need to use keywords as identifiers in some parts of the grammar,
the recommend approach is to use datatype rules:

ValidID: ID | 'keyword';

ValidIDInAnotherGrammarPart: ID | 'keyword2';

Best,
Sebastian
--
Looking for professional support for Xtext, Xtend or Eclipse Modeling?
Find help at http://xtext.itemis.com or xtext(@)itemis.com
Blog: zarnekow.blogspot.com
Twitter: @szarnekow
Google+: https://www.google.com/+SebastianZarnekow

Report message to a moderator

Previous Topic:	Validation in Xtext
Next Topic:	code gen. /no duplicate

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Fri Apr 26 07:46:45 GMT 2024

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter