Eclipse Community Forums: TMF (Xtext) » How to force the editor to parse the entire editor content

Help

Home

Home » Modeling » TMF (Xtext) » How to force the editor to parse the entire editor content

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

How to force the editor to parse the entire editor content [message #1776163]

Mon, 13 November 2017 10:21

Markus Holzem

Messages: 7
Registered: November 2017

Junior Member

Hello,

I created a COBOL like syntax with Xtext. I got the syntax and an adapted lexer to handle COBOL's comment and numbering areas (column 1-6, 73-80) and the COBOL comment marker "*" on column 7. Everything is fine until I start to edit the file.

Taking the following COBOL code as an example:

      * Comment
        linkage section.
        01 parameter PIC X(20).
        procedure division using parameter.

When I start to edit e.g. the 3rd line the lexer receives only

01 parameter PIC X(20).
        procedure division using parameter.

to be parsed. After that the first few columns are interpreted as comments by my lexer and the highlighting is wrong. The reduced call seems to be used only to do the styling of the text in the editor. Afterwards a check is processed on the entire document for syntax checking.

Is there a possibility to force the editor parse the entire document instead of the part beginning with the modified line?

Kind regards,
Markus

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776196 is a reply to message #1776163]

Mon, 13 November 2017 15:44

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

Hi,
id have a look at org.eclipse.xtext.parser.impl.PartialParsingHelper.reparse(IParser, IParseResult, ReplaceRegion) respectively org.eclipse.xtext.parser.AbstractParser.reparse(IParseResult, ReplaceRegion)

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

[Updated on: Mon, 13 November 2017 15:44]

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776224 is a reply to message #1776196]

Tue, 14 November 2017 06:58

Markus Holzem

Messages: 7
Registered: November 2017

Junior Member

Thanks for the hint, but unfortunately the process does not pass PartialParsingHelper.reparse.

While editing my stack trace looks like

Thread [main] (Suspended)	
	DocumentTokenSource.computeDamageRegion(DocumentEvent) line: 286	
	DocumentTokenSource.updateStructure(DocumentEvent) line: 249	
	XtextDocument.fireDocumentChanged(DocumentEvent) line: 700	
	XtextDocument(AbstractDocument).replace(int, int, String, long) line: 1102	
	XtextDocument(AbstractDocument).replace(int, int, String) line: 1120	
	ProjectionTextStore.replace(int, int, String) line: 103	
	ProjectionDocument(AbstractDocument).replace(int, int, String, long) line: 1095	
	ProjectionDocument(AbstractDocument).replace(int, int, String) line: 1120	
	ProjectionDocument.replace(int, int, String) line: 625	
	DefaultDocumentAdapter.replaceTextRange(int, int, String) line: 233	
	StyledText.modifyContent(Event, boolean) line: 7355	
	...

The DocumentEvent holds the XtextDocument and an offset starting at the position where I type in the editor.

I think I have to override org.eclipse.xtext.ui.editor.model.DocumentTokenSource.getRepairEntryData(DocumentEvent) to achieve what I want.

Markus

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776226 is a reply to message #1776224]

Tue, 14 November 2017 07:16

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

but where is the parser involved?

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

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776243 is a reply to message #1776226]

Tue, 14 November 2017 11:51

Markus Holzem

Messages: 7
Registered: November 2017

Junior Member

The only place to get access to line and character position in line is the lexer.
The adapted lexer can be configured in the ???RuntimeModule class (overriding bindLexer, etc.)
The adapted lexer is accessed in DocumentTokenSource.getRepairEntryData using DocumentTokenSource.createTokenSource. This is where the string containing the document(part) is provided to the lexer.

On further investigation I found BacktrackingLexerDocumentTokenSource doing exactly what I need. But how do I provide an object of BacktrackingLexerDocumentTokenSource instead of DocumentTokenSource while XtextDocument is created?

I simply cannot find the piece of code where and how to define the binding. Could you help me there?

Markus

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776250 is a reply to message #1776243]

Tue, 14 November 2017 13:20

Christian Dietrich

Messages: 14665
Registered: July 2009

Senior Member

sry have no idea on that

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

Report message to a moderator

Re: How to force the editor to parse the entire editor content [message #1776251 is a reply to message #1776243]

Tue, 14 November 2017 13:22

Markus Holzem

Messages: 7
Registered: November 2017

Junior Member

Got it:

class ...UiModule extends AbstractApiUiModule {
  ...
  def Class<? extends DocumentTokenSource> bindDocumentTokenSource() {
    return typeof(BacktrackingLexerDocumentTokenSource);
  }
  ...

solves the problem.

Markus

Report message to a moderator

Previous Topic:	Content Assist for external information
Next Topic:	[DELETE PLEASE] Configurable issue code is not registered - check binding for...

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Fri Apr 26 01:04:00 GMT 2024

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter