boost split with a single character or just one string

I wish to split a string on a single character or a string. I would like to use boost::split since boost string is our standard for basic s开发者_如何学JAVAtring handling (I don't wish to mix several techniques).

In the single character case I could do split(vec,str,is_any_of(':')) but I'd like to know if there is a way to specify just a single character. It may improve performance, but more importantly I think the code would be clearer with just a single character, since is_any_of conveys a different meaning that what I want.

For matching against a string I don't know what syntax to use. I don't wish to to construct a regex; some simple syntax like split(vec,str,match_str("::") would be good.

I was looking for the same answer but I couldn't find one. Finally I managed to produce one on my own.

You can use std::equal_to to form the predicate you need. Here's an example:

boost::split(container, str, std::bind1st(std::equal_to<char>(), ','));

This is exactly how I do it when I need to split a string using a single character.

In the following code, let me assume using namespace boost for brevity.
As for splitting on a character, if only algorithm/string is allowed, is_from_range might serve the purpose:

split(vec,str, is_from_range(':',':'));

Alternatively, if lambda is allowed:

split(vec,str, lambda::_1 == ':');

or if preparing a dedicated predicate is allowed:

struct match_char {
  char c;
  match_char(char c) : c(c) {}
  bool operator()(char x) const { return x == c; }
};

split(vec,str, match_char(':'));

As for matching against a string, as David Rodri'guez mentioned, there seems not to be the way with split. If iter_split is allowed, probably the following code will meet the purpose:

iter_split(vec,str, first_finder("::"));

On the simple token, I would just leave is_any_of as it is quite easy to understand what is_any_of( single_option ) means. If you really feel like changing it, the third element is a functor, so you could pass an equals functor to the split function.

That approach will not really work with multiple tokens, as the iteration is meant to be characater by character. I don't know the library enough to offer prebuilt alternatives, but you can implement the functionality on top of split_iterators