| How to do semantic-based highlighting? [message #225713] |
Sat, 10 April 2004 04:55  |
Eclipse User |
|
|
|
I'm looking for an example (or tutorial) how to do semantic-based
highlighting for a text editor. All examples I found only use a rule
based approach.
Let's assume I'd like to highlight this piece of code
class ... def ... end ... end
and I'd like to color the matching class/end pair in one color and the
matching def/end pair in another color.
I created my own ITokenScanner which is able to distinguish the "end"
tokens based on its own internal state. This works quite well, although
I need to parse the whole partition not only the the modified part.
Now I tried to partion the document into code and comment regions.
Unfortunately, now matching tokens may be part of different partitions
and my highlighting approach fails.
What's the right way to do this?
bye
--
Stefan Matthias Aust // "Zweifel sind der Ansporn des Denkens..." -U
|
|
|
| Re: How to do semantic-based highlighting? [message #227420 is a reply to message #225713] |
Tue, 13 April 2004 04:56  |
| Eclipse User |
|
|
|
Not a solution, but a general remark:
Depending on the complexity of your (highlighting) grammar, you will
need to reparse more or less of your document. For example, keyword
highlighting in Java only needs local information, but semantic
highlighting (different colors for static, final etc. fields) needs
information about the entire file, basically an AST, which is hard to
update incrementally.
So, depending on your grammar, updating the highlighting in sync with
typing may not be possible or fast enough - in the Java editor,
highlighting that needs an AST (mark occurences etc.) is done in a
background thread.
-tom
Stefan Matthias Aust wrote:
> I'm looking for an example (or tutorial) how to do semantic-based
> highlighting for a text editor. All examples I found only use a rule
> based approach.
>
> Let's assume I'd like to highlight this piece of code
>
> class ... def ... end ... end
>
> and I'd like to color the matching class/end pair in one color and the
> matching def/end pair in another color.
>
> I created my own ITokenScanner which is able to distinguish the "end"
> tokens based on its own internal state. This works quite well, although
> I need to parse the whole partition not only the the modified part.
>
> Now I tried to partion the document into code and comment regions.
> Unfortunately, now matching tokens may be part of different partitions
> and my highlighting approach fails.
>
> What's the right way to do this?
>
>
> bye
|
|
|
| Re: How to do semantic-based highlighting? [message #230625 is a reply to message #227420] |
Sun, 18 April 2004 03:40 |
| Eclipse User |
|
|
|
Tom Eicher wrote:
> Not a solution, but a general remark:
> [Parse whole file, use background thread]
Thanks for your answer. I was already afraid that it would be along
that lines. I'll try to understand how the Java editor does the
background reconciling and analysis then...
bye
--
Stefan Matthias Aust // "Zweifel sind der Ansporn des Denkens..." -U
|
|
|
Powered by
FUDForum. Page generated in 0.02217 seconds
Home