[cdt-dev] Language plugin extending Scanner

["Peter Watkins" <peter.watkins@xxxxxxxxxxxxx>]

I've written a language plugin to support
Cosmic's dialect of C. I had to copy a fair portion of the GCC language
parser to make my own parser that extends ISourceCodeParser. It works fine
except in one case: assembly code. Cosmic doesn't have GCC's asm() statement.
Instead, it uses two types of preprocessor keywords:

#pragma asm
#pragma endasm

or

#asm
#endasm


...to denote a block of assembly. I
can add the latter pair of keywords to my ScannerExtensionConfiguration,
but CPreprocessor doesn't know what do with them of course.

I have a few questions:

1) I assume all the # lines above are
consumed by the scanner, and my parser will never see them. Is this correct?

2) In the case of #if #endif, the scanner
inserts tINACTIVE_CODE_START into the token stream for the parser to see,
right? To handle my assembly block preprocessor instructions, should I
do something similar (i.e. insert tASSEMBLY_START and tASSEMBLY_END)?

3) If I do need to go about extending
the scanner in my language plugin, how do I do that? IScanner "is
not intended to be implemented by clients" and CPreprocessor "...should
not be used directly, rather than that you should be using the IScanner
interface." Furthermore, AbstractCLikeLanguage.createScanner() is
marked final, so even if I could make my own scanner, how would my language
provide it?

Thanks,

Peter