Simple DSL editor with code completion using PEG.js
Marcin Stefaniuk
Using single PEG grammar I've defined a transforming parser and code
editor with simple content completion. All built with excellent
PEG.js.
The language as its output gives metamodel for Concrete editor (http://
concrete-editor.org/). You can check the demo on
http://mstefaniuk.github.com/Concrete-Freetext/editor/index.html. Base
implementation details are described on
http://eutechne.stefaniuk.info/2011/10/dsl-editor-with-code-completion-in.html.
I have few suggestion that can be taken into account in next versions
of PEG.js.
1. To make grammar more readable maybe implementation can be
optionally separated as in Mouse - Java PEG parser (http://
www.romanredz.se/freesoft.htm). It gives possibility to use single
grammar with pluggable parsing logic.
2. It would be more convenient to receive parser using constructor
than via global variable. Constructor gives possibility to provide
configuration object that can be used internally by parser.
Configuration can be external model, resources or parsing settings.
3. To build a code completion I need to slighty modify generated
parser to provide in error structure also expectations list (https://
gist.github.com/1281239). Is it possible to add it in next version?
Jamon
Marcin Stefaniuk
Ace editor. It has simple definitions of syntax coloring based on
regex. Most of these you can get directly from your grammar or even
you can build small highlighting rules extractor using grammar AST
built by PEG.js itself.
Regards
Marcin
Ulric Wilfred
David Majda
2011/10/12 Marcin Stefaniuk <marcin.s...@gmail.com>:
> I would like to present small DSL I've created as proof-of-concept.
> Using single PEG grammar I've defined a transforming parser and code
> editor with simple content completion. All built with excellent
> PEG.js.
>
> The language as its output gives metamodel for Concrete editor (http://
> concrete-editor.org/). You can check the demo on
> http://mstefaniuk.github.com/Concrete-Freetext/editor/index.html. Base
> implementation details are described on
> http://eutechne.stefaniuk.info/2011/10/dsl-editor-with-code-completion-in.html.
I looked only briefly, but it looks like a very nice project.
> I have few suggestion that can be taken into account in next versions
> of PEG.js.
> 1. To make grammar more readable maybe implementation can be
> optionally separated as in Mouse - Java PEG parser (http://
> www.romanredz.se/freesoft.htm). It gives possibility to use single
> grammar with pluggable parsing logic.
Please file an issue on GitHub. I'd really appreciate more specific
thoughts about how would you like this to work.
> 2. It would be more convenient to receive parser using constructor
> than via global variable. Constructor gives possibility to provide
> configuration object that can be used internally by parser.
> Configuration can be external model, resources or parsing settings.
Why wouldn't a configuration object passed to the "parse" method be sufficient?
> 3. To build a code completion I need to slighty modify generated
> parser to provide in error structure also expectations list (https://
> gist.github.com/1281239). Is it possible to add it in next version?
Sure, I'll reply to the mail with the patch you sent me later.
--
David Majda
Entropy fighter
http://majda.cz/
Marcin Stefaniuk
> > I have few suggestion that can be taken into account in next versions
> > of PEG.js.
> > 1. To make grammar more readable maybe implementation can be
> > optionally separated as in Mouse - Java PEG parser (http://
> >www.romanredz.se/freesoft.htm). It gives possibility to use single
> > grammar with pluggable parsing logic.
>
> Please file an issue on GitHub. I'd really appreciate more specific
> thoughts about how would you like this to work.
> > 2. It would be more convenient to receive parser using constructor
> > than via global variable. Constructor gives possibility to provide
> > configuration object that can be used internally by parser.
> > Configuration can be external model, resources or parsing settings.
>
> Why wouldn't a configuration object passed to the "parse" method be sufficient?
Regards
--
Marcin Stefaniuk
mar...@stefaniuk.info
http://eutechne.stefaniuk.info
David Majda
>> > than via global variable. Constructor gives possibility to provide
>> > configuration object that can be used internally by parser.
>> > Configuration can be external model, resources or parsing settings.
>>
>> Why wouldn't a configuration object passed to the "parse" method be sufficient?
> Sorry for silly question but how could I do it now?
question really should have been something like:
Would the ability to pass a configuration object to the "parse"
method (which would be visible in the parser) be a sufficient solution
for you?
Sorry for being unclear.
Marcin Stefaniuk
Would the ability to pass a configuration object to the "parse"
method (which would be visible in the parser) be a sufficient solution
for you?