typedef std::string Word;
typedef int Freq;
typedef int ParagraphNum; 
typedef int LocationInParagraph; 
struct WordElement
{
    Word word_content; // 单词
    Freq word_frequency; // 词频
    ParagraphNum word_para_num; // 单词第一次出现的段落数（因为getline以回车符结束）
    LocationInParagraph word_location; // 单词第一次在段落中出现的位置

};

class WordAnalysisInArticle {
public:
    // 读取英文文本，将词频存入无排序列表 word_unsorted
    void read_file(FileName & kFilename_1);

    // 按内容、按词频 排序
    void sort_by_content();
    void sort_by_frequency();

    // 将排序后的词频表写进文本
    void write_word_sorted_by_content(FileName &kFilename_2);
    void write_word_sorted_by_freq(FileName &kFilename_3);

    // 将具有相同后缀的 word 写入 word_suffix_map
    void add_word_suffix_map(const Word &word);
    // 从 unsorted_word 建立 word_suffix_map
    void build_map_from_word_unsorted();
    // 将 word_suffix_map 写入 文件中
    void write_map_file(FileName &kFilename_4);

    // 搜索函数
    // 搜索 word 在文本中第一次出现的段落及位置
    bool search_word(const Word &word);

    // 析构函数
    ~WordAnalysisInArticle();

private:
    std::vector<WordElement> word_unsorted; // 单词列表
    std::vector<WordElement> word_sorted_by_freq; // 词频排序表
    std::vector<WordElement> word_sorted_by_content; // 词频排序表

    // 常见后缀 Suffix
    SuffixVector suffix_vector = { "ment", "ance", "bility", "hood", "sion", "tion", "ism", "ness", "tude", "ble", "ive", "ish", "like", "ful", "ous", "less", "ent", "ize", "ise",  "most"};

    // multimap ： Key：后缀（Suffix） ；Value：单词 （word）
    WordSuffixMap word_suffix_map;
};

        {
                "the", "a", "an", "other", "another",
                "is", "are", "am", "was", "were", "been",
                "be", "i", "you", "he", "she", "they", "it", "all", "one",
                "me", "him", "her", "them", "we", "this", "those", "these", "our", "yourself", "yourselves",
                "my", "his", "hers", "their", "for", "as", "of", "go",
                "what", "where", "who", "when", "how", "which", "that", "will", "its", "why",
                "whatever", "wherever", "whoever", "whenever", "however", "but", "ever", "and",
                "at", "in", "on", "by", "up", "down", "about", "along", "upon", "through", "to",
                "isnt", "do", "doesnt", "dont","did", "wasnt", "werent", "im", "youre", "with", "toward",
                "has", "have", "off", "from","had", "gone", "any", "some", "just", "so", "then",
                "here", "there", "out", "off", "ok", "until", "end", "begin", "almost", "always", "often",
                "sometimes", "good", "say", "said", "went", "hello", "some", "any", "than",
                "now", "today", "says", "make", "take","okay",
                "not", "no", "if", "into", "would", "never", "too", "us", "may", "maybe", "might", "can", "should", "shall",
                "could", "under", "beyond", "though", "more", "much", "many", "such", "yes",
                "man", "men", "woman", "women", "also", "himself", "herself", "ourselves", "themselves"
        };

// 判断是否具有相同后缀 Suffix
static bool WithSameSuffix(const Word &word, const Suffix &suffix);
// 去掉标点符号及转小写字母
static std::string  cleanup_str(const Word&);

// 按词频大小比较
bool comp_frequency(const WordElement &word_element_1, const WordElement &word_element_2);
// 按字母表顺序比较
bool comp_word(const WordElement &word_element_1, const WordElement &word_element_2)

#include <iostream>
#include <vector>
#include <fstream>
#include <string>
#include <string.h>
#include <iomanip>
#include <sstream>
#include <unordered_set>

#include "WordElement.h"
#include "FunctionWord.h"
#include "WordAnalysisInArticle.h"


int main()
{
    WordAnalysisInArticle word_freq_calculate;

    // 处理英文文本
    word_freq_calculate.read_file("A_Game_Of_Thrones_1.txt");

    // word_freq_calculate.search_word("dark");


    // 输出到 WordSortedbyContent.txt
    std::string str_word = "WordSortedbyContent.txt";
    std::ofstream ofs_word(str_word.c_str());
    // 清除待写入文本内容
    if (ofs_word.good())
        {
            remove(str_word.c_str());
        }
    word_freq_calculate.write_word_sorted_by_content(str_word);


    // 输出到 WordSortedbyFreqVector.txt
    std::string str_word_frequency = "WordSortedbyFrequency.txt";
    std::ofstream ofs_word_frequency(str_word_frequency.c_str());
    // 清除待写入文本内容
    if (ofs_word_frequency.good())
    {
        remove(str_word_frequency.c_str());
    }
    word_freq_calculate.write_word_sorted_by_freq(str_word_frequency);


    // 输出到 WordWithSameSuffix.txt
    std::string str_word_with_same_suffix = "WordWithSameSuffix.txt";
    std::ofstream ofs_word_with_same_suffix(str_word_with_same_suffix.c_str());
    // 清除待写入文本内容
    if (ofs_word_with_same_suffix.good())
    {
        remove(str_word_with_same_suffix.c_str());
    }

    word_freq_calculate.write_map_file(str_word_with_same_suffix);


    // 搜索特定单词
    word_freq_calculate.search_word("Especially");
    word_freq_calculate.search_word("man-at-arms");
    word_freq_calculate.search_word("ClionProject");

}