Eclipse Community Forums: TMF (Xtext) » get source position of tokens

Home » Modeling » TMF (Xtext) » get source position of tokens

get source position of tokens [message #793167]

Tue, 07 February 2012 16:05

Eclipse User

How one can get the source position (line, column) of a token?

For my runtime environment, i wish to inform the user about the source location in case of failure.

Thanks!

Bodo

[Updated on: Wed, 08 February 2012 00:26] by Moderator

Re: get source postion of tokens [message #793183 is a reply to message #793167]

Tue, 07 February 2012 16:27

Eclipse User

Hi,

can you give use more context please? from which place do you want to do this? can you give us an example?

~Christian

Re: get source postion of tokens [message #793332 is a reply to message #793167]

Tue, 07 February 2012 21:28

Eclipse User

When you parse the source, you get a tree of INode. An INode knows about
the position in the text.

When there are parse errors, the reported issues carry information about
position.

When there are validation or linking errors, the reported issues carry
information about positions.

If these are not enough to help you find what you want, then more
information about what you are trying to do is required.

- henrik

On 2012-07-02 22:05, Bodo Koch wrote:
> How one can get the source position (line, column) of a token?
>
> For my runtime environment, i wich to inform the user about the source
> location in case of failure.
>
> Thanks!
>
> Bodo

Re: get source postion of tokens [message #793428 is a reply to message #793183]

Wed, 08 February 2012 00:42

Eclipse User

Hi,

Let say a tiny scripting language can deal with some kind of objects. To refer to object members one can specify the grammar like:

ObjectRef: objId=ID ('.' objRef+=ID)*;

If the user writes sentences like:

a b

he/she will get a syntax error in the eclipse/xtext IDE.

If the user writes a syntactically correct sentence like

a.z

but z is not a member of object a one will try to catch that error by the validation phase and issue an error.

BUT if also the static semantic is correct like

a.b

but a or b is not instantiated one will get an null pointer exception during execution time of this script. In a normal eclipse/java IDE (if the program runs in this context) one will get an error message in the control window and can jump direct to the erroneous statement. I do not expect this comfort in exclipse/xtext because this compiled script will not run in that context, but I want to inform the user at what source code position this error occurs.

During the validation one can access the abstract syntax tree members with the methods or.getObjId() and or.getObjRef(), if or is of class ObjectRef. How one can get the source position info out of this elements?

Thanks, and hopefully this explanation helps

Bodo

[Updated on: Wed, 08 February 2012 00:46] by Moderator

Re: get source postion of tokens [message #793712 is a reply to message #793428]

Wed, 08 February 2012 08:14

Eclipse User

So, your evaluator simply needs to get the INode for the part of the
model that causes the problem.

If you are generating code, you must naturally ensure that there are
references in the generated code that allows you to map that back to the
source.

- henrik

On 2012-08-02 6:42, Bodo Koch wrote:
> Hi,
>
> Let say a tiny scripting language can deal with some kind of objects. To
> refer to object members one can specify the grammar like:
> objId=ID ('.' objRef+=ID)*
> If the user writes sentences like:
> a b
> he/she will get a syntax error in the eclipse/xtext IDE.
>
> If the user writes a syntactically correct sentence like
> a.z
> but z is not a member of object a one will try to catch that error by
> the validation phase and issue an error.
>
> BUT if also the static semantic is correct like
> a.b
> but a or b is not instantiated one will get an null pointer exception
> during execution time of this script. In a normal eclipse/java IDE (if
> the program runs in this context) one will get an error message in the
> control window and can jump direct to the erroneous statement. I do not
> expect this comfort in exclipse/xtext because this compiled script will
> not run in that context, but I want to inform the user at what source
> code position this error occurs.
>
> Thanks, and hopefully this explanation helps
>
> Bodo
>

Re: get source postion of tokens [message #794528 is a reply to message #793712]

Thu, 09 February 2012 07:02

Eclipse User

Henrik Lindberg wrote on Wed, 08 February 2012 08:14

So, your evaluator simply needs to get the INode for the part of the
model that causes the problem.

If you are generating code, you must naturally ensure that there are
references in the generated code that allows you to map that back to the
source.

- henrik

Hm, this hint doesn't realy helped me. I tried the following code:

/* wie geil ist das denn? */	
public String getTokenPosition (EObject e)
	{
        CompositeNode cn  = NodeUtil.getNodeAdapter(e).getParserNode();
        return (cn.getLine() + "/" + cn.getOffset(
        		) + "/" + cn.getLength());	
	}

Can I do this with die INode stuff some how?

Thanks

Bodo

Re: get source postion of tokens [message #794629 is a reply to message #794528]

Thu, 09 February 2012 09:20

Eclipse User

Hi, we are talking about Xtext 2.x you are talking about Xtext <=
1.0.x regards Christian

Re: get source postion of tokens [message #794778 is a reply to message #794629]

Thu, 09 February 2012 12:45

Eclipse User

Christian Dietrich wrote on Thu, 09 February 2012 09:20