ghidra scripting: ghidra unable to resolve string type

Question

I have following code:

#include <stdio.h>

char somes[10] = "good job!";
char somes1[8] = "another";
char somes2 [5] = "job";

int func()
{
  printf("You did a %s\n", somes);
  // func(10);
}

int main()
{
  printf("You did a %s\n", somes1);
  printf("You did a %s\n", somes2);
  printf("garb");
  printf("sharp");
  func();
}

I want to get sizes of strings defined in the code. Ghidra behaves strangely in such cases, even with symbol information. For e.g. in the above program, ghidra accurately detects lengths (or sizes of) all strings except. "garb" - it says that its size is 1 and defined as a single character "g" (s.getObject()). I feel this very strange, as even with debugging information, ghidra fails to detect this. Is there anything I can do to get such sizes (or correct string representations)?

I am doing something like:

symbols = set(currentProgram.getSymbolTable().getAllSymbols(True))
for s in symbols:
    if s.isGlobal():
        // gives me size of object
        print(s.getObject().getLength())

Note that I have also posted this question here - https://github.com/NationalSecurityAgency/ghidra/issues/2274

Answer 1

If you are trying to change and individual string, you can right click on the first character of the string and under Data, convert it to TerminatedCString.

If you have a problem with how Ghidra is recognizing strings, you can tweak the settings in the ASCII Strings analyzer to have it recognize shorter strings, change alignments ettings or use a different word model.