modifying language during lexical parse?

You probably can't do that in v3.1 however in v4.0 that we're working on a new feature is programmatic lexical parsers. So that means you can write a lexical parser in code and probably do this sort of thing.

I've got it to mostly work now -- when the document is modified, I loop through all the tokens, and any VariableTokens (which were initially found via regex matching their declarations) get added as explicit patterns. Then I re-parse the document with the updated language.

It's admittedly a bit klunky, but it works -- except that for some reason, the first time the document is loaded from my data source, the Document.Reparse() appears to have no effect.

--
Benjamin Lewis

Make sure you set SyntaxLanguage.IsUpdating to true before making any changes to the language and then reset it to false afterwards. Doing that will also reparse your document so no need to call Document.Reparse.

I have a feeling that you were missing the IsUpdating calls.

Nope; the calls are there. It works fine everywhere but at the initial document load. I'll try removing the call to Reparse, though.

Thanks,
--
Benjamin Lewis

It turns out the problem was unrelated -- my documents are being loaded from data grids, and I wasn't aware that when I select one row, the next N rows get loaded as well for efficiency. Since I'm reusing my language across documents, my modifications were being made properly but then getting cleared by subsequent loads.

You're correct, my call to Reparse() was superfluous.

I don't suppose there's some way to reuse a static base language across my documents, but to have local modifications on a per-document basis?

Thanks,
--
Benjamin Lewis

Unfortunately no since the language definition dictates how the lexical parser operates. So if you are using it in all your documents then any changes made will apply to all those documents. In this case if you are making document-specific changes to the language then you'll have to make a new language instance for each document.

The language updating and reparsing happens quickly enough that it's not an issue, so now I'm just calling my language updating routine when the document becomes visible/focused. Tastes great, less (memory) filling!

I suppose speed could become an issue for longer documents, but the code in these documents is for defining database stored procedures, and tends to be quite short.

Thanks again; this is starting to look great!
--
Benjamin Lewis