strtok how to also include delimiters as tokens

Question

Right now I have code set up to divide up my string into tokens with delimiters of ,;= and space. I would also like to include the special characters as tokens.

char * cstr = new char [str.length()+1];
strcpy (cstr, str.c_str());

char * p = strtok (cstr," ");

while (p!=0)
{
    whichType(p);
    p = strtok(NULL," ,;=");
}

So right now if I print out the tokens of a string such as, asd sdf qwe wer,sdf;wer it would be

asd
sdf
qwe
wer
sdf
wer

I want it to look like

asd
sdf
qwe
wer
,
sdf
;
wer

Any help would be great. Thanks

Answer 1

You need more flexibility. (Besides, strtok is a bad, error prone interface).

Here's a flexible algorithm that generates tokens, copying them to an output iterator. This means you can use it to fill a container of your choice, or print it directly to an output stream (which is what I'll use as a demo).

The behaviour is specified in option flags:

enum tokenize_options
{
    tokenize_skip_empty_tokens              = 1 << 0,
    tokenize_include_delimiters             = 1 << 1,
    tokenize_exclude_whitespace_delimiters  = 1 << 2,
    //
    tokenize_options_none    = 0,
    tokenize_default_options =   tokenize_skip_empty_tokens 
                               | tokenize_exclude_whitespace_delimiters
                               | tokenize_include_delimiters,
};

Not how I actually distilled an extra requirement that you hadn't named, but your sample implies: you want the delimiters output as tokens unless they're whitespace (' '). This is what the third option comes in for: tokenize_exclude_whitespace_delimiters.

Now here's the real meat:

template <typename Input, typename Delimiters, typename Out>
Out tokenize(
        Input const& input,
        Delimiters const& delim,
        Out out,
        tokenize_options options = tokenize_default_options
        )
{
    // decode option flags
    const bool includeDelim   = options & tokenize_include_delimiters;
    const bool excludeWsDelim = options & tokenize_exclude_whitespace_delimiters;
    const bool skipEmpty      = options & tokenize_skip_empty_tokens;

    using namespace std;
    string accum;

    for(auto it = begin(input), last = end(input); it != last; ++it)
    {
        if (find(begin(delim), end(delim), *it) == end(delim))
        {
            accum += *it;
        }
        else
        {
            // output the token
            if (!(skipEmpty && accum.empty()))
                *out++ = accum;   // optionally skip if `accum.empty()`?

            // output the delimiter
            bool isWhitespace = std::isspace(*it) || (*it == '\0'); 
            if (includeDelim && !(excludeWsDelim && isWhitespace))
            {
                *out++ = { *it }; // dump the delimiter as a separate token
            }

            accum.clear();
        }
    }

    if (!accum.empty())
        *out++ = accum;

    return out;
}

A full demo is Live on Ideone (default options) and Live on Coliru (no options)

int main()
{
    // let's print tokens to stdout
    std::ostringstream oss;
    std::ostream_iterator<std::string> out(oss, "\n"); 

    tokenize("asd sdf qwe wer,sdf;wer", " ;,", out/*, tokenize_options_none*/);

    std::cout << oss.str();
    // that's all, folks
}

Prints:

asd
sdf
qwe
wer
,
sdf
;
wer