Thread: Faster C coding thanks to utilities?

Hello,

I recently had to implement a deep copy for a cascade of C structs. Structs in structs are easy; structs containing variable length arrays of structs are tougher - you actually need to find the length of the array and copy the structs individually - because those structs also contain arrays of structs.

I did it by running all the C source code through a grep script that pulled out the struct definitions. Within struct definitions I put comments by the arrays I wanted copied and made gawk assemble these into structname_copy() functions - and structname_memdup and structname_memfree functions but that's a side issue. It worked fine but it felt unsatisfactory. The gawk relied on my code being written consistently - it didn't use a C parser, so had I changed my coding style but once the code would have broken. Maintenance is going to be hell.

Are there utilities for doing general parsing of C files, extracting chunks of code and similar? It would be nice to be able to have a set of C utilities that make this sort of thing as easy as running grep on a one-record-per-line data file. That way I gould generate the copy commands in a makefile and be confident that the code it creates will be good, no matter who else has monkeyed with the code in the meantime.

Regards,
Lock

Originally Posted by rommegrot

I did it by running all the C source code through a grep script that pulled out the struct definitions. Within struct definitions I put comments by the arrays I wanted copied and made gawk assemble these into structname_copy() functions - and structname_memdup and structname_memfree functions but that's a side issue. It worked fine but it felt unsatisfactory. The gawk relied on my code being written consistently - it didn't use a C parser, so had I changed my coding style but once the code would have broken. Maintenance is going to be hell.

That's actually a clever approach. Several C code tools rely on specially formatted comments. For instance Lint, and certain source code translation tools.

Are there utilities for doing general parsing of C files, extracting chunks of code and similar?

There are plenty of refactoring tools available, although I'm not sure that any of them will do exactly what you need. They will certainly get you the names and members of the relevant structures, though.

If you have programmers using different styles, I think you already have somewhat of a maintenance problem. I don't think introducing a tool is going to make things worse. In fact, it might open everybody's eyes to the advantage of having a consistent style.

Originally Posted by Salem

http://www.gccxml.org/HTML/Index.html
This would allow you to generate an unambiguous representation of the source, and not rely on the maintainers playing by the rules for detailed formatting.

Cool link. I had not seen that before.

That looks like a most complete answer, thanks! xmlpath and xmlgrep to hand I'll start playing with it!

Regards, Lock