计算一篇文章中单词出现的频率，并把输出频率最高的十五个单词输出来
#include
#include"string.h"
#define WORDNUM 65535 // max word number
#define WORDLENGTH 15 // word max length #define PRINTWORDNUM 15 //out put max word number
char differentword[WORDNUM][WORDLENGTH]={0};//different word info
int differentCount[WORDNUM];//equal word number int iIndex = 0;//current file word total number int iWordCount =0;//current file different word number
void GetWordInfo(FILE *fpRead);
void SetWordList(char word[WORDLENGTH]); void OrderWordList(); void OutPutWordList();
void DealWithWord()
{ FILE *fpRead;
if((fpRead = fopen("D:\word.txt", "r")) == NULL)
{ printf("Cannot read file %s\n", "D:\word.txt"); return; }
GetWordInfo(fpRead);
OrderWordList(); OutPutWordList(); }
void GetWordInfo(FILE *fpRead)
{ int jIndex = 0; int i = 0; char ch; char word[WORDLENGTH]={0};
while((ch=fgetc(fpRead))!=EOF)
{ //putchar(ch); if ((ch >= 65 && ch <=90) ||(ch >= 97 && ch <=122)) { if (jIndex < WORDLENGTH) { word[jIndex] = ch; jIndex ++; } } else { if (jIndex != 0) { SetWordList(word);
jIndex = 0;
iIndex ++; } } } }
void SetWordList(char word[WORDLENGTH])
{ int i; int iEqual = 0;
if (iIndex == 0)
{ strcpy(differentword[0],word); differentCount[0] = 0; iWordCount ++; }
for (i = 0; i < iWordCount; i ++)
{ if (strcmp(differentword[i],word) == 0) { differentCount[i] ++; iEqual = 1; } }
if (iEqual == 0)
{ strcpy(differentword[iWordCount],word); differentCount[iWordCount] ++; iWordCount ++; }
for (i = 0; i
word[i] = '\0'; } }
void OrderWordList()
{ int iCurrent = 0; int i,j,tempValue; char wordTemp[WORDLENGTH]={0};
for (i = 0; i < iWordCount; i ++)
{ iCurrent = i; for (j = i + 1; j < iWordCount; j ++) { if (differentCount[iCurrent] < differentCount[j]) { iCurrent = j; } }
if (iCurrent != i)
{ tempValue = differentCount[iCurrent]; differentCount[iCurrent] = differentCount[i]; differentCount[i] = tempValue;
strcpy(wordTemp,differentword[iCurrent]);
strcpy(differentword[iCurrent],differentword[i]); strcpy(differentword[i],wordTemp); } } }
void OutPutWordList()
{ int i; for (i = 0; i < iWordCount && i < PRINTWORDNUM; i ++) { printf("%s %d\n", differentword[i], differentCount[i]); } }
int main(int argc, char* argv[])
{ // 从英語文章中统计出单词出現の频率,打印出前15个出现频率高的单词
DealWithWord();
getchar();
}
|
