c语言中文件中单词排序

Question

Answer 1

代码如下：如有帮助，请采纳一下，谢谢。

#include <stdio.h>
#include <string.h>
#include <stdlib.h>

struct WordNode
{
    char word[30];
    int nmb;
    struct WordNode* next;
    WordNode(){next = 0;}
};

class WordCount
{
public:
    WordCount(){head = 0;}
    ~WordCount()
    {
        WordNode* node = 0;
        while(head)
        {
            node = head->next;
            delete head;
            head = node;
        }
    }
    WordNode* findNode(char *p);     //查找单词
    void insertNode(WordNode* node); //插入单词
    void addCount(WordNode* node);  //单词的数量加1
    void display();   //显示
private:
    struct WordNode* head;
};

//判断是否为空
int isEmpty(char* p)
{
    while(*p)
    {
        if(*p > 0x20 )
            return 0;
        p++;
    }
    return 1;
}
//查找单词
WordNode* WordCount::findNode(char* p)
{
    struct WordNode* node = head;
    while(node)
    {
        if (strcmp(p,node->word) == 0)
        {
            return node;
        }else
        {
            node = node->next;
        }
    }
    return 0;
}
//插入单词
void WordCount::insertNode(struct WordNode* node)
{
    //printf(" >> insert node:%s\n",node->word);
    struct WordNode* P;
    struct WordNode* pre;
    if (head == 0)
    {
        head = node;
        return;
    }
    P = head;
    pre = head;
    while(P)
    {
        if (strcmp(P->word,node->word) < 0 )  //降序排列
        {
            if(P == head)
            {
                node->next = head;
                head = node; //插入头部
            }else
            {
                pre->next = node;
                node->next = P;
            }
            break;
        }else
        {
            pre = P;
            P = P->next;
            //插入尾部
            if(P==0)
            {
                pre->next = node;
                break;
            }
        }
    }
}
//单词的数量加1
void WordCount::addCount(struct WordNode* node)
{
    node->nmb += 1;
    //printf("   >> %s :%d\n",node->word,node->nmb);
}

//显示
void WordCount::display() 
{
    struct WordNode* node = head;
    while(node)
    {
        printf("%s\t%d\n",node->word,node->nmb);
        //cout << node->word << "\t" << node->nmb << endl;
        node = node->next;
    }
}
bool isctrl(char ch)
{
    if (ch < 0x1F || ch == 0x7F)
    {
        return true;
    }
    return false;
}
//删除空格，并将大写转变成小写
void toLower(char tt[])
{
    int i=0,j=0;
    while(tt[i] != '\0')
    {
        if( (tt[i] != ' ') && (!isctrl(tt[i])) )
        {
            if(tt[i] >= 'A' && tt[i] <= 'Z')
                tt[j++] = tt[i] + 32;
            else
                tt[j++] = tt[i];
        }
        i++;
    }
}
//buf是存储文件的缓冲区，lSize是文件大小
char* textFileRead(char* filename,int *lSize)
{
    char* buf;
    FILE *pf = fopen(filename,"r");
    fseek(pf,0,SEEK_END);
    *lSize = ftell(pf);
    // 用完后需要将内存free掉
    rewind(pf); 
    buf = new char[*lSize+1];
    *lSize = fread(buf,sizeof(char),*lSize,pf);
    buf[*lSize] = '\0';
    return buf;
}

int main()
{
    int size,i=0,j= 0;
    char text[50] = {0};
    WordCount cc;
    struct WordNode* node = 0;
    char* buf = textFileRead("textfile.txt",&size);
    
    if(size <= 0)
    {
        printf("文件打开失败，或者为空\n");
        return 0;
    }
    
    while( i <= size)
    {
        if (i == size)
        {
            text[j] = '\0';
            toLower(text);
            if( isEmpty(text) )
            {
                j = 0;
                i++;
                continue;
            }
            node = 0;
            node = cc.findNode(text);
            if (node)
            {
                cc.addCount(node);
            }else
            {
                node = (struct WordNode*)malloc(sizeof(WordNode));
                strcpy(node->word ,text);
                node->nmb = 1;
                node->next = 0;
                cc.insertNode(node);
            }
            break;
        }else
        {
            //此处不考虑中间含有.的单词，入地名，缩写等等
            if (buf[i] == ' ' || buf[i] == '\r' || buf[i] == ',' || buf[i] == '.' || buf[i] == '\n'|| buf[i] == '!' || isctrl(buf[i])|| buf[i] == ';')
            {
                text[j] = '\0';
                toLower(text);
                if(isEmpty(text))
                {
                    j = 0;
                    i++;
                    continue;
                }
                node = 0;
                node = cc.findNode(text);
                if (node)
                {
                    //printf("find ,add count\n");
                    cc.addCount(node);
                }else
                {
                    //printf("not find,create\n");
                    node = (struct WordNode*)malloc(sizeof(WordNode));//new WordNode;
                    strcpy(node->word,text);
                    node->nmb = 1;
                    node->next = 0;
                    cc.insertNode(node);
                }
                j = 0;
            }else
            {
                text[j++] = buf[i];
            }
            i++;
        }
    }

    cc.display();

    


    return 0;
}