Recommend C front-end that preserves preprocessor directives.

Question

I'd like to start a project that involves transforming C code, but I'd like to include the preprocessor directives. I don't want to reinvent the wheel by writing my own C parser, so does anyone know of a front-end that can parse C preprocessor and C code, and produce an AST that can be used to re-generate (or pretty-print) the original source?

e.g.,:

#define FILENAME "filename"
#include <stdio.h>

FILE *f=0;
...
if (file_is_open) {
#ifdef CAN_OPEN_IT
    f = fopen(FILENAME, "r");
#else
    printf("Unable to open file.\n");
#endif
}

The above code should be parsed into some in-memory representation that can be used to re-generate the source. In other words, it should not be processed as normal C in two phases, first processing the PP directives and then parsing pure C code. Rather it should represent the whole compile-time logic including the preprocessor variables.

Answer 1

Take the GNU gcc compiler, the flags required to pre-process the source is gcc -E mysource.c, see here for further information. As for pretty printing it, there's indent and this explains the usage here, this is a bit old, but nonetheless worthy of mention. There is also cflow that can produce a map of the source.

Sorry if I misunderstood what you're looking for...

Hope this helps, Best regards, Tom.

Answer 2

Take a look at Clang. (See http://clang.llvm.org/features.html#applications .)

Answer 3

You can take look at the http://www.antlr.org/wiki/display/ANTLR3/ANTLR3+Code+Generation+-+C

Answer 4

Our DMS Software Reengineering Toolkit has a C front end (and a C++ front end) that:

• parses (compilable) C source code in a variety of dialects into ASTs,
• preserves the preprocessor directives in most cases as AST nodes
• can regenerate compilable C code (with comments and preprocessor directives) from the ASTs
• can collects thousands of files in a single image to allow cross-file analysis and transformation
• provides full symbol table construction and access
• provides procedural access to ASTs with a large AST manipulation library, including navigate, inspect, insert, delete, replace, match, ...
• provides source-to-source transformations using patterns written in the C notation that match against the ASTs

For C (not yet for C++), DMS also provides:

• control and data flow analysis
• local and global points-to analysis
• global call graph construction

DMS has been used to process extremely large C applications for the purposes of extracting facts and generating new, derived code from the original source base.

It can handle the OP's example directly.