如何在 C/C++ 的源代码中获取浮点文字的位置？

Question

I would like to write a function whose input is a piece of C/C++ code, and whose output is the exact of locations of floating-point literals. Preferred implementation languages are Java or Python, although this question is language-agnostic.

Example input program:

#include<stdio.h>
#include<string.h>
int main() {
   float x = 10.3;
   int y = 28;
   printf("The float value : %f\n", x);
   printf("The sum of float and int variable : %f\n", (x+y));
   return 0;
}

The output should be something like 4, 13-16 indicating the line and offset location of the float literal 10.3. The reason I need to get this location is to automatically transfer the literal to another one for the purpose of developing a bug analysis and tracking tool.

Any idea how we can precisely get the locations?

I tried to apply a simple regular expression to capture decimals like "10.3" but it would also capture floats in comments, like // APACHE-common-2.0, and strings, like printf( "%2.5g", x);, also regular expression-based solutions would miss floats in "double x=0;".

Answer 1

You may be looking for something like Kythe , it is a tool Google build to index and search their source code.

As I understand it (I have only seen a presentation, not worked with the tool myself), it is build on top of LLVM(?) and uses the compiler to build and extract a graph of the code. This then enables somebody to search for f.ex. places where a specific function is called - not just functions that are named the same, but call-sites that actually link to "this specific definition".