C++ How to convert string to char*

Question

I need to convert a string to a char * for use in strtok_s and have been unable to figure it out. c_str() converts to a const char *, which is incompatible.

Also, if someone could explain to me why the second strtok_s function (inside the loop) is necessary, it'd be a great help. Why do i need to explicitly advance the token rather than, for example, the while loop it is in, which fetches each line of a file consecutively, implicitly.

while( getline(myFile, line) ) { // Only one line anyway. . . is there a better way?
    char * con = line.c_str();
    token = strtok_s( con, "#", &next_token);
    while ((token != NULL))
    {
        printf( " %s\n", token );
        token = strtok_s( NULL, "#", &next_token);
    }
}

related question.

Answer 1

Use strdup() to copy the const char * returned by c_str() into a char * (remember to free() it afterwards)

Note that strdup() and free() are C, not C++, functions and you'd be better off using methods of std::string instead.

The second strtok_s() is needed because otherwise your loop won't terminate (token's value won't change).

Answer 2

As Daniel said, you could go with

strdup(line.c_str());

Which is better then the strcpy I originally proposed since it allocates the necessary space

Answer 3

You can't convert to a char * because that would that would allow you to write to std::string's internal buffer. To avoid making std::string's implementation visible, this isn't allowed.

Instead of strtok, try a more "C++-like" way of tokenizing strings. See this question:

How do I tokenize a string in C++?

Answer 4

strtok() is a badly designed function to begin with. Check your documentation to see if you have a better one. BTW, never use strtok() in any sort of threaded environment unless your docs specifically say it's safe, since it stores state in between calls and modifies the string it's called on. I assume strtok_s() is a safer version, but it's not going to be a really safe one.

To convert a std::string into the char *, you can do:

char * temp_line = new char[line.size() + 1];  // +1 char for '\0' terminator
strcpy(temp_line, line.c_str());

and use temp_line. Your installation may have a strdup() function, which will duplicate the above.

The reason you need two calls to strtok_s() is that they do different things. The first one tells strtok_s() what string it needs to work on, and the second one continues with the same string. That's the reason for the NULL argument; it tells strtok_s() to keep going with the original string.

Therefore, you need one call to get the first token, and then one for each subsequent token. They could be combined with something like

char * temp_string_pointer = temp_line;
while ((token = strtok_s( con, "#", &next_token)) != NULL)
{
   temp_string_pointer = NULL;

and so on, since that would call strtok_s() once with the string pointer and after that with NULL. Don't use temp_line for this, since you want to delete[] temp_line; after processing.

You may think this is a lot of fiddling around, but that's what strtok() and relatives usually entail.

Answer 5

strtok works like this:

First call return string from beginning unril the delimiter or all the string if no delimiter were found:

token = strtok_s(con, "#", &next_token);

Second call using with NULL allow you to continue parsing the same string to find the next delimiter:

token = strtok_s(NULL, "#", &next_token);

If you reach the end of the string next call will return NULL;

Answer 6

Whenever you have a std::string and what you need is a (modifiable) character array, then std::vector<char> is what you need:

void f(char* buffer, std::size_t buffer_size);

void g(std::string& str)
{
  std::vector<char> buffer(str.begin(),str.end());
  // buffer.push_back('\0');    // use this if you need a zero-terminated string
  f(&buffer[0], buffer.size()); // if you added zero-termination, consider it for the size
  str.assign(buffer.begin(), buffer.end());
}

Answer 7

the 2nd strtok call is inside the loop. It advances you token pointer so that you print out tokens one-by-one, until you've printed out all of them, the pointer becomes null and you exit the loop.

To answer the 1st part of your question, as other have suggested, c_str() only gives you the internal buffer pointer - you can't modify that, that's why it's const. If you want to modify it, you need to allocate your own buffer and copy the string's contents into it.

Answer 8

If you really need to access the string's internal buffer here is how: &*string.begin(). Direct access to string's buffer is useful in some cases, here you can see such a case.

Answer 9

You can easily write a conversion routine that will tokenize a string and return a vector of sub-strings:

std::vector<std::string> parse(const std::string& str, const char delimiter)
{
    std::vector<std::string> r;

    if(str.empty())
        return r;

    size_t prev = 0, curr = 0;

    do
    {
        if(std::string::npos == (curr = str.find(delimiter, prev)))
            curr = str.length();

        r.push_back(str.substr(prev, curr - prev));
        prev = curr + 1;
    }
    while(prev < (int)str.length());
    return r;
}

Answer 10

I think you can first convert string to const char*, then copy the const char* to a char* buffer for further use.