Eclipse Community Forums: TMF (Xtext) » Make a block ignored by Parser

Help

Home

Home » Modeling » TMF (Xtext) » Make a block ignored by Parser

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

Make a block ignored by Parser [message #1733818]

Wed, 01 June 2016 07:40

Jonathan L.

Messages: 24
Registered: May 2016

Junior Member

Hi,

I'm implementing an Editor for an existing Language.
In this Language a C integration is possible with the structure:

native do
/* C code*/
end

Since a real C integration would be difficult (or are there any recommendations?) I want to make the Parser ignoring the block between 'do' and 'end'.
How do I realize this?

I can't do it with the '->' operator in a terminal rule, because in other rules there are also 'do'-'end'-blocks used.

Can someone please help me with this?

Report message to a moderator

Re: Make a block ignored by Parser [message #1733854 is a reply to message #1733818]

Wed, 01 June 2016 13:36

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

sounds like a good usecase for a custom/stateful lexer.

or you enumerate all stuff like ( dont know if this works i a smooth way)

CODE: "keyword1"|...."keywordn" | ID | INT | ANY_OTHER | ....;

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator

Re: Make a block ignored by Parser [message #1733862 is a reply to message #1733854]

Wed, 01 June 2016 14:16

Uli Merkel

Messages: 250
Registered: June 2013

Senior Member

but do you not want to ignore "native do"->"end" ?

Report message to a moderator

Re: Make a block ignored by Parser [message #1733866 is a reply to message #1733862]

Wed, 01 June 2016 14:26

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

spaces in keywords a VERY bad thing in xtext

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

[Updated on: Wed, 01 June 2016 14:26]

Report message to a moderator

Re: Make a block ignored by Parser [message #1733897 is a reply to message #1733866]

Thu, 02 June 2016 04:55

Uli Merkel

Messages: 250
Registered: June 2013

Senior Member

Christian Dietrich wrote on Wed, 01 June 2016 10:26

spaces in keywords a VERY bad thing in xtext

... but will it do what is required?
I don't think enumerating all possible C-code elements are a feasible option.

[Updated on: Thu, 02 June 2016 04:56]

Report message to a moderator

Re: Make a block ignored by Parser [message #1733899 is a reply to message #1733897]

Thu, 02 June 2016 05:36

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

well you can wlqays give it a try.

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator

Re: Make a block ignored by Parser [message #1733906 is a reply to message #1733854]

Thu, 02 June 2016 07:00

Jonathan L.

Messages: 24
Registered: May 2016

Junior Member

Hi,

thanks for the quick answers.

Quote:

but do you not want to ignore "native do"->"end" ?

The meaning of the rule would look like

Native:
  ('native' | 'native/pre') 'do' -> 'end'
;

So the keywords are given and shall be highlighted, but everything between 'do' and 'end' shall be ignored and doesn't need to be highlighted. Basically like an ML_COMMENT, but without the '/*' and '*/'

Quote:

or you enumerate all stuff like ( dont know if this works i a smooth way)

CODE: "keyword1"|...."keywordn" | ID | INT | ANY_OTHER | ....;

do you mean something like this? :

Native:
	('native' | 'native/pre') 'do' e=enumRule 'end'
;
enum enumRule:
	ID | INT | STRING | ANY_OTHER
;

When I tried this it also doesn't accept anything within the block Confused

Report message to a moderator

Re: Make a block ignored by Parser [message #1733909 is a reply to message #1733906]

Thu, 02 June 2016 07:25

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

no i mean

Native:
('native' | 'native/pre') 'do' e=Stuff 'end'
;
Stuff:
(ID | INT | STRING | ANY_OTHER)*
;

Twitter : @chrdietrich
Blog : https://www.dietrich-it.de

Report message to a moderator

Re: Make a block ignored by Parser [message #1733919 is a reply to message #1733818]

Thu, 02 June 2016 08:47

Jonathan L.

Messages: 24
Registered: May 2016

Junior Member

Ah yes that works better thanks!
I guess I have to see if it will suffice.

But I got a weird problem.. In a smaller Test-"language" it works, but in my final implementation it produces the error: " extraneous input 'content of block' expecting 'end' "
in both the rules are the same...

Another idea I have is to do a workaround with the Validator or something. Is there a possibility to suppress error messages at certain positions?

Else I guess I will give the custom Lexer a try.

Thanks in advance!

Report message to a moderator