Re: lua parser generator or c-grammar parser

[Dave Dodge <dododge@...>]

On Mon, Jul 25, 2005 at 10:59:24AM +0700, Antero Vipunen wrote:
> Well, even simple C-grammar parser would be enough for my needs.(I 
> mean it should parse C files, and be written in lua)

Note that the C grammar is a bit tricky because of the preprocessor.
Consider what happens if the code has macros, stringization, token
pasting, or #if statements with mathematical expressions.  If you
implement your own preprocessor, you also have the issue that
conditional code might expect compiler-specific predefined values.

If you want to follow the standard completely, you also run into
unusual things like:

  - Trigraphs

  - Alternate punctuator spellings such as <: and %:%:

  - The fact that string-literal concatenation is assumed to take place
    before tokenization

  - Preprocessor tokenization is context-sensitive. "yes\no" tokenizes
    as a header-name when it appears on an #include line but as a
    string-literal elsewhere.  The rules for handling the backslash
    are different in those two contexts.

I only mention this stuff because I was working on a C parser (sorry,
not in Lua) in recent weeks and ran into all of this myself.

                                                  -Dave Dodge