boost::split to split only if delimiter is present in c++

Question

I am using boost::split but if the string has no delimiter then it returns the string in the vector. I want it to not return anything if delimiter is not there in the string.

#include <iostream>
#include <string>
#include <bits/stdc++.h>
#include <boost/algorithm/string.hpp>
using namespace std;
  
int main()
{
    string input("abcd");
    vector<string> result;
    boost::split(result, input, boost::is_any_of("\t"));
  
    for (int i = 0; i < result.size(); i++)
        cout << result[i] << endl;
    return 0;
}

Output is abcd. I want the vector to be empty if delimiter is not present in the string. please suggest.

Answer 1

It looks a bit like you might need a validating parser. Regex could be a good starting stone, but I'd suggest a parser generator, as in all likelihood you require more

My crystal ball whispers that you might be parsing command line output or CSV/TSV files

This is what you could do with Boost Spirit X3:

template <typename Cont>
bool parse_columns(std::string_view input, Cont& container,
                   unsigned required = 2) {
    namespace x3 = boost::spirit::x3;

    auto valid = [required](auto& ctx) {
        x3::_pass(ctx) = x3::_val(ctx).size() >= required;
    };

    auto delim = x3::char_('\t');
    auto field = *(~delim);
    auto rule
        = x3::rule<struct _, Cont, true>{"rule"} 
        = (field % delim)[valid];

    return parse(begin(input), end(input), rule, container);
}

Here's a live demo with test-cases:

Live On Compiler Explorer

#include <boost/spirit/home/x3.hpp>
#include <fmt/ranges.h>

template <typename Cont>
bool parse_columns(std::string_view input, Cont& container,
                   unsigned required = 2) {
    namespace x3 = boost::spirit::x3;

    auto valid = [required](auto& ctx) {
        x3::_pass(ctx) = x3::_val(ctx).size() >= required;
    };

    auto delim = x3::char_('\t');
    auto field = *(~delim);
    auto rule
        = x3::rule<struct _, Cont, true>{"rule"} 
        = (field % delim)[valid];

    return parse(begin(input), end(input), rule, container);
}

int main() {
    for (auto input : {
             "",
             "\t",
             "abcd\t",
             "ab cd\tef",
             "\tef",
             "ab\tc\t\tdef",
             "abcd",
         }) {
        std::vector<std::string> columns;

        if (parse_columns(input, columns)) {
            fmt::print("'{}' -> {}\n", input, columns);
        } else {
            fmt::print("'{}' -> not matched\n", input);
        }
    }
}

Prints

'' -> not matched
'   ' -> {"", ""}
'abcd   ' -> {"abcd", ""}
'ab cd  ef' -> {"ab cd", "ef"}
'   ef' -> {"", "ef"}
'ab c       def' -> {"ab", "c", "", "def"}
'abcd' -> not matched

Tweaks

• To treat repeated \t as a single delimiter, just change field % delim to field % +delim
• You can easily replace with another container, like std::set

Live On Compiler Explorer