cpp-string-utils/include/string_utils/tokenizer.h

//
//  tokenizer.hpp
//  string-utils
//
//  Created by Sam Jaffe on 10/8/20.
//  Copyright © 2020 Sam Jaffe. All rights reserved.
//

#pragma once

#include <string>
#include <vector>

namespace string_utils {

class tokenizer {
public:
  static constexpr size_t const infinite_outputs{~size_t(0)};
  struct quote {
    char on;
    std::string escaped;
  };
private:
  std::string divider_;
  quote quote_;
  size_t max_outputs_{infinite_outputs};
  bool truncate_{false};
  bool ignore_empty_tokens_{true};
  bool escapable_{false};

public:
  tokenizer(std::string divider, struct quote quote = {'\0', ""});
    
  tokenizer &max_outputs(size_t new_max_outputs);
  tokenizer &truncate(bool new_truncate_overage);
  tokenizer &ignore_empty_tokens(bool new_ignore_empty_tokens);
  tokenizer &escapable(bool new_escapable);

  std::vector<std::string> operator()(std::string const &input) const;
  
private:
  size_t max_outputs() const;
};

inline auto split(std::string const & data, std::string const & on,
                  size_t max = tokenizer::infinite_outputs) {
  return tokenizer{on}.max_outputs(max)(data);
}

}