word_freq.cpp

//                                Copyright James P. McNellis 2013.                               //
//                   Distributed under the Boost Software License, Version 1.0.                   //
//     (See accompanying file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)    //

#include <algorithm>
#include <cctype>
#include <cstdio>
#include <map>
#include <memory>
#include <string>
#include <vector>
#include <windows.h>

struct close_handle_deleter
{
    typedef HANDLE pointer;
    void operator()(HANDLE h) { if (h != INVALID_HANDLE_VALUE && h != nullptr) { CloseHandle(h); } }
};

struct unmap_file_view_deleter
{
    void operator()(void* p) { if (p != nullptr) { UnmapViewOfFile(p); } }
};

typedef std::unique_ptr<HANDLE, close_handle_deleter>  raii_handle;
typedef std::unique_ptr<void, unmap_file_view_deleter> raii_file_view;
typedef std::map<std::string, unsigned long long>      count_map_type;

int main(int argc, char** argv)
{
    if (argc <= 1)
        return std::puts("no files to process\n"), 0;

    count_map_type word_counts;
    unsigned const files_processed(std::count_if(argv + 1, argv + argc, [&](char const* file_name) -> bool
    {
        raii_handle const file(CreateFile(file_name, GENERIC_READ, FILE_SHARE_READ, nullptr,
                                          OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, nullptr));
        if (file.get() == INVALID_HANDLE_VALUE)
            return std::printf("failed to open file %s\n", file_name), false;

        raii_handle const mapping(CreateFileMapping(file.get(), nullptr, PAGE_READONLY, 0, 0, nullptr));
        if (mapping.get() == nullptr)
            return std::printf("failed to map file %s\n", file_name), false;

        raii_file_view const view(MapViewOfFile(mapping.get(), FILE_MAP_READ, 0, 0, 0));
        if (view.get() == nullptr)
            return std::printf("failed to map view of file %s\n", file_name), false;

        LARGE_INTEGER file_size = { };
        if (GetFileSizeEx(file.get(), &file_size) == FALSE)
            return std::printf("failed to compute size of file %s\n", file_name), false;

        char const* const file_first(reinterpret_cast<char const*>(view.get()));
        char const* const file_last(file_first + file_size.QuadPart);

        char const* word_first(nullptr);
        for (char const* it(file_first); it != file_last; ++it)
        {
            if (*it < 0 || (!std::isalnum(*it) && *it != '_'))
            {
                if (word_first != nullptr)
                    ++word_counts[std::string(word_first, it)];

                word_first = nullptr;
            }
            else if (word_first == nullptr)
            {
                word_first = it;
            }
        }

        if (word_first != nullptr)
            ++word_counts[std::string(word_first, file_last)];

        return true;
    }));

    std::printf("%u word(s) in %u file(s)\n", static_cast<unsigned>(word_counts.size()), files_processed);
    std::for_each(begin(word_counts), end(word_counts), [&](count_map_type::const_reference e)
    {
        std::printf("%s: %llu\n", e.first.c_str(), e.second);
    });
}