wordcount.cpp

#include "wordcount.hpp"

#include <algorithm>
#include <fstream>
#include <iostream>

// Assumptions
// 1. Function should read the input from the file, i.e. caching the input is
// not allowed.
// 2. The input is always encoded in UTF-8.
// 3. Break only on space, tab and newline (do not break on non-breaking space).
// 4. Sort words by frequency AND secondary sort in alphabetical order.

// Implementation rules
// 1. You can add new files but dependencies are generally not allowed unless it
// is a header-only library.
// 2. Your submission must be single-threaded, however feel free to implement
// multi-threaded version (optional).

#ifdef SOLUTION
//
// Your solution here.
//
#else
// Baseline solution.
// Do not change it - you can use for quickly checking speedups
// of your solution agains the baseline, see check_speedup.py
std::vector<WordCount> wordcount(std::string filePath) {
  std::unordered_map<std::string, int> m;
  m.max_load_factor(0.5);

  std::vector<WordCount> mvec;

  std::ifstream inFile{filePath};
  if (!inFile) {
    std::cerr << "Invalid input file: " << filePath << "\n";
    return mvec;
  }

  std::string s;
  while (inFile >> s)
    m[s]++;

  mvec.reserve(m.size());
  for (auto &p : m)
    mvec.emplace_back(WordCount{p.second, move(p.first)});

  std::sort(mvec.begin(), mvec.end(), std::greater<WordCount>());
  return mvec;
}
#endif