20-Coding-Patterns-DesignGurus.org/1_longest_substring_with_k_distinct_characters.cpp

/*/ problem source:
 * https://www.geeksforgeeks.org/find-the-longest-substring-with-k-unique-characters-in-a-given-string/

 Possible references include:
https://stackoverflow.com/a/7304184 - custom delimeter for istream
 */
#include <fstream>
#include <iostream>
#include <istream>
#include <locale>
#include <ostream>
#include <sstream>
#include <string>
#include <unordered_map>
#include <vector>

struct semicolon_is_space : std::ctype<char> {
  // this struct/class is adapted from: https://stackoverflow.com/a/7304184
  semicolon_is_space() : std::ctype<char>(get_table()){};
  static mask const *get_table() {
    static mask rc[table_size];
    rc[(int)';'] = std::ctype_base::space;
    rc[(int)'\n'] = std::ctype_base::space;
    return &rc[0];
  };
};
class input {
public:
  std::string s;
  unsigned long k;
  friend std::istream &operator>>(std::istream &in, input &i);
  friend std::ostream &operator<<(std::ostream &out, const input &i);
};

std::istream &operator>>(std::istream &in, input &i) {
  std::getline(in, i.s, '\n'); // An inelegant hack to skip newline character
  std::getline(in, i.s, ';');
  in >> i.k;
  return in;
}

std::ostream &operator<<(std::ostream &out, const input &i) {
  out << "String:\n" << i.s << "\nK: " << i.k << std::endl;
  return out;
}

class result {
public:
  unsigned long longest_start = 0;
  unsigned long longest_end = 0;
  bool match_found = false;
  bool operator==(const result r);
  friend std::istream &operator>>(std::istream &in, result &r);
  friend std::ostream &operator<<(std::ostream &out, const result &r);
};

bool result::operator==(const result r) {
  return longest_start == r.longest_start && longest_end == r.longest_end &&
         match_found == r.match_found;
}

std::istream &operator>>(std::istream &in, result &r) {
  in >> r.longest_start >> r.longest_end >> r.match_found;
  return in;
}

std::ostream &operator<<(std::ostream &out, const result &r) {
  out << "longest_start: " << r.longest_start
      << " longest_end: " << r.longest_end << " match_found: ";
  if (r.match_found)
    out << "true";
  else
    out << "false";
  out << std::endl;
  return out;
}

class test_case {
public:
  input i;
  result r;
  friend std::istream &operator>>(std::istream &in, test_case &t);
  friend std::ostream &operator<<(std::ostream &out, const test_case &t);
};

std::istream &operator>>(std::istream &in, test_case &t) {
  in >> t.i >> t.r;
  return in;
}

std::ostream &operator<<(std::ostream &out, const test_case &t) {
  out << t.i << t.r;
  return out;
}

result find(const std::string &s, const unsigned long k) {
  result r;
  std::unordered_map<char, unsigned long> char_count;
  unsigned long start = 0, end = 0;
  for (; end < s.length(); end++) {
    char_count[s[end]] += 1;
    while (char_count.size() > k) {
      char_count[s[start]]--;
      if (char_count[s[start]] == 0) {
        char_count.erase(s[start]);
      }
      start++;
    }
    if (char_count.size() == k &&
        end - start > r.longest_end - r.longest_start) {
      r.longest_start = start, r.longest_end = end;
      r.match_found = true;
    }
  }

  if (r.match_found) {
    std::cout << "Longest substring is \""
              << s.substr(r.longest_start, r.longest_end - r.longest_start + 1)
              << "\" with length " << r.longest_end - r.longest_start + 1
              << ".\n";
  } else {
    std::cout << "Could not find any match, not enough unique characters.\n";
  }
  return r;
}

int main(int argc, char *argv[]) {
  std::vector<test_case> test_cases;
  semicolon_is_space delimeter;
  for (int i = 1; i < argc; i++) {
    test_case t;
    std::ifstream f(argv[i]);
    f.imbue(std::locale(f.getloc(), new semicolon_is_space));
    while (f.good()) {
      f >> t;
      if (t.i.s.length()) { // skip empty line/string inputs
        std::cout << t;
        test_cases.push_back(t);
      }
    }
  }

  std::cout << "\n===================================\n" << std::endl;

  for (const test_case &t : test_cases) {
    if (find(t.i.s, t.i.k) == t.r) {
      std::cout << "Test case with string \"" << t.i.s << "\" and k=" << t.i.k
                << " passed.\n";
    } else {
      std::cout << std::flush;
      std::cerr << "TEST CASE WITH STRING \"" << t.i.s << "\" AND k=" << t.i.k
                << " FAILED.\n";
    }
  }
  return 0;
}