文本文件单词统计

#include <iostream>
#include <cstdio>
#include <cstring>
#include <algorithm>

using namespace std;
#define LIST_INIT_SIZE 100
#define LISTINCREMENT 10

typedef struct{
    char word[21];
    int count;
}ElemType;

typedef struct{
    ElemType* elem;
    int length;
    int listsize;
}Sqlist;

void InitList(Sqlist &S)//初始化顺序表
{
    S.elem=(ElemType*)malloc((LIST_INIT_SIZE)*sizeof(ElemType));
    if(!S.elem){
        printf("malloc false\n");
        return;
    }else{
        S.listsize=LIST_INIT_SIZE;
        for(int i=0;i<S.listsize;i++){
            memset(S.elem[i].word,0,sizeof(S.elem[i].word));
            S.elem[i].count=0;
        }
        S.length=0;
    }
}

int LocateElem(Sqlist &S,char *s)//寻找单词
{
    bool flag=true;
    for(int i=0;i<S.length;i++){
        flag=true;
        for(int j=0;S.elem[i].word[j]||s[j];j++){

            char Selem,selem;//不区分大小写，全部按小写字母处理
            Selem=S.elem[i].word[j];
            selem=s[j];
            if(Selem>='A'&&Selem<='Z')
                Selem+=32;
            if(selem>='A'&&selem<='Z')
                selem+=32;

            if(Selem!=selem){//和这个单词不相等，看下一个
                flag=false;
                break;
            }
        }
        if(flag){//出现过这个单词
            S.elem[i].count++;
            return 0;
        }
    }

    return S.length+1;
}

void InsertList(Sqlist &S,int i,char *s)//i为单词长度
{
    int slength=i;
    if(!LocateElem(S,s))//要插入的单词已经出现过
        return;

    if(S.listsize<=S.length){//空间不足
        ElemType *newbase;
        newbase=(ElemType*)realloc(S.elem,(LISTINCREMENT)*sizeof(ElemType));
        if(!newbase){
            printf("malloc LISTINCREMENT false\n");
            return;
        }
        S.elem=newbase;
        S.listsize+=(LISTINCREMENT);
    }

        int p;
        bool flag=false;
        for(int i=0;i<S.length;){
            for(int j=0;S.elem[i].word[j]||s[j];j++){

                char Selem,selem;
                Selem=S.elem[i].word[j];
                selem=s[j];
                if(Selem>='A'&&Selem<='Z')
                    Selem+=32;
                if(selem>='A'&&selem<='Z')
                    selem+=32;

                if(selem==Selem)
                    continue;
                else if(selem<=Selem){//比当前的单词小，找到了合适的插入位置
                    p=i;
                    flag=true;
                    break;
                }else if(selem>=Selem){//比当前的单词大
                    i++;
                    break;
                }
            }
            if(flag)
                break;
        }

        if(flag){//如果找到了合适的插入位置则插入
            for(int i=S.length-1;i>=p;i--){
                S.elem[i+1]=S.elem[i];
            }
            memset(S.elem[p].word,0,sizeof(S.elem[p].word));
            memcpy(S.elem[p].word,s,slength);
            S.elem[p].count=1;
        }else{//比当前所有的单词都要大，插入到顺序表的最后一个位置
            memset(S.elem[S.length].word,0,sizeof(S.elem[S.length].word));
            memcpy(S.elem[S.length].word,s,slength);
            S.elem[S.length].count=1;
        }
        S.length++;
}

void PrintList(Sqlist &S,FILE *fp)//打印结果
{
    fprintf(fp,"\n");
    int res=0;
    for(int i=0;i<S.length;i++){
        //cout<<S.elem[i].word<<"         "<<S.elem[i].count<<endl;
        fprintf(fp,"%s      %d\n",S.elem[i].word,S.elem[i].count);
        res+=S.elem[i].count;
    }
    fprintf(fp,"ALL:%d\n",res);
    return;
}

int main(void)
{
    FILE *fp;
    fp=fopen("d:\\History.txt","r");//读文件

    Sqlist Q;//单词统计数组Q

    InitList(Q);

    char c;
    char s[21];
    memset(s,0,sizeof(s));
    int i=0;
    int titlecnt=0;
    while(fscanf(fp,"%c",&c)){

        if(c=='\n'&&titlecnt==1)
            break;
        else if(c=='\n'){
            titlecnt++;
            s[i]='\0';
            InsertList(Q,i,s);
            memset(s,0,sizeof(s));
            i=0;
            continue;
        }

        if(!((c>='A'&&c<='Z')||(c>='a'&&c<='z'))&&i){
            s[i]='\0';
            InsertList(Q,i,s);
            memset(s,0,sizeof(s));
            i=0;

        }else if(!((c>='A'&&c<='Z')||(c>='a'&&c<='z'))){
            continue;
        }else
            s[i++]=c;
    }

    fp=fopen("d:\\History.txt","a");//写入结果

    PrintList(Q,fp);
    fclose(fp);

    return 0;
}
//2020-5-28 LZH