database.get.string() incorrectly includes null terminator in string about ida-minsc

[arizvisa]

Okay. You caught me. I was being lazy and didn't want to implement the decoding of all the available strtypes since db.get.string() works regardless of whether a string is defined at the address or not and lets you specify the length too.

The reason why idaapi.get_strlit_contents(ea, idaapi.get_item_size(ea), idaapi.get_str_type(ea)) never made it into db.get.string is because it actually utf8 encodes utf-16le (or any mbcs really). This changes the bytes in the actual string when really it should return it as a unicode type. So to preserve those bytes, I always just used .rstrip() and then .decode() on all the strings I read from a database.

An example of this utf8-encoding on a utf-16le string (IDA 7.1.180227): 9a 3f 43 56 4f 54 57 3e 43 62 00 00

Python>repr(idaapi.get_strlit_contents(h(), idaapi.get_item_size(h()), idaapi.get_str_type(h())))
'\xe3\xbe\x9a\xe5\x99\x83\xe5\x91\x8f\xe3\xb9\x97\xe6\x89\x83'

But then with db.get.string(), if you process it you get the following correct string.

Python>repr(db.get.string().decode('utf-16le'))
u'\u3f9a\u5643\u544f\u3e57\u6243\x00'

I just didn't want to implement all of the ASCSTR types, but I guess I will.

Since you're only using C-style strings, you can put something like the following in your ~/.idapythonrc.py until I get it done.

db.get.string = staticmethod(fcompose(db.get.string, operator.methodcaller('rstrip')))

[arizvisa]

@x0rloser, PR #21 should fix this. I submitted the PR for review, then I reviewed my own code and realized that I'm just so awesome and so I merged it upstream because I'm always right and of course it fixes the issue...so, yeah...

Anyways, you should just need to do a git pull to re-sync, and you should be good to go. It supports the other stuff without needing UTF-8 like I mentioned. If that didn't fix it for you for some reason or there's some particular case I missed, just re-open this issue and lmk.

Thanks for your contribution, sir

[x0rloser]

can confirm is fixed (for C strings at least). awesomeness ftw ;)