C++ 实现类似 Python 字符串的 split 函数

my_split.cc

#include <iostream>
#include <string>
#include <vector>

std::vector<std::string> split_naive(const std::string& s, char delim) {
  std::vector<std::string> ret;
  std::string token;
  for (char ch : s) {
    if (ch == delim) {
      ret.push_back(token);
      token.clear();
    } else {
      token += ch;
    }
  }
  ret.push_back(token);

  return ret;
}

void print_vector(const std::vector<std::string>& v) {
  std::cout << "[";
  if (v.size() > 0) {
    std::cout << "'" << v[0] << "'";
  }
  for (size_t i = 1; i < v.size(); i++) {
    std::cout << ", '" << v[i] << "'";
  }
  std::cout << "]\n";
}

int main() {
  std::string test = ",123,,456,";
  std::vector<std::string> tokens = split_naive(test, ',');
  print_vector(tokens);

  return 0;
}

NVIDIA DALI 中实现的字符串 split（dali/core/common.cc）：

std::vector<std::string> string_split(const std::string &s, const char delim) {
    std::vector<std::string> ret;
    size_t pos = 0;
    while (pos != std::string::npos) {
        size_t newpos = s.find(delim, pos);
        ret.push_back(s.substr(pos, newpos - pos));
        if (newpos != std::string::npos) {
            pos = newpos + 1;
        } else {
            pos = newpos;
        }
    }
    return ret;
}

我的和 DALI 的性能对比：https://quick-bench.com/q/ljKZItp5YvqVl0-EvrwpZDx1sSc