中文詞法分析的簡單程式 (轉)

中文詞法分析的簡單程式 (轉)[@more@]

6.0 下編譯透過  disc 為辭典  expression 為輸入的短語

 

// Code.cpp : Defines the entry point for the console application.
//

#include "stdafx.h"
#include "iostream.h"
#include "string.h"
#include ".h"

int Count(char* words);  算有多少個字元

struct dict
{
 char syn[8],word[6];
 int de,wcode;
} dic[19];

int main(int argc, char* argv[])
{
 strcpy(dic[0].syn,"祈使動詞");
  dic[0].scode=0;
 strcpy(dic[0].word,"請");
  dic[0].wcode=0;
  strcpy(dic[1].syn,"祈使動詞");
  dic[1].scode=0;
 strcpy(dic[1].word,"把");
  dic[1].wcode=1;
  strcpy(dic[2].syn,"代詞");
 dic[2].scode=1;
 strcpy(dic[2].word,"你");
 dic[2].wcode=0;
 strcpy(dic[3].syn,"動詞");
 dic[3].scode=2;
 strcpy(dic[3].word,"走");
 dic[3].wcode=0;
 strcpy(dic[4].syn,"動詞");
 dic[4].scode=2;
 strcpy(dic[4].word,"跑");
 dic[4].wcode=1;
 strcpy(dic[5].syn,"動詞");
 dic[5].scode=2;
 strcpy(dic[5].word,"跳");
 dic[5].wcode=2;
 strcpy(dic[6].syn,"動詞");
 dic[6].scode=2;
 strcpy(dic[6].word,"轉");
 dic[6].wcode=3;
 strcpy(dic[7].syn,"動詞");
 dic[7].scode=2;
 strcpy(dic[7].word,"抬");
 dic[7].wcode=4;
 strcpy(dic[8].syn,"方向詞");
 dic[8].scode=3;
 strcpy(dic[8].word,"向左");
 dic[8].wcode=0;
 strcpy(dic[9].syn,"方向詞");
 dic[9].scode=3;
 strcpy(dic[9].word,"向右");
 dic[9].wcode=1;
 strcpy(dic[10].syn,"方位詞");
 dic[10].scode=4;
 strcpy(dic[10].word,"向前");
 dic[10].wcode=0;
 strcpy(dic[11].syn,"方位詞");
 dic[11].scode=4;
 strcpy(dic[11].word,"側");
 dic[11].wcode=1;
 strcpy(dic[12].syn,"名詞");
 dic[12].scode=5;
 strcpy(dic[12].word,"左手");
 dic[12].wcode=0;
 strcpy(dic[13].syn,"名詞");
 dic[13].scode=5;
 strcpy(dic[13].word,"右手");
 dic[13].wcode=1;
 strcpy(dic[14].syn,"名詞");
 dic[14].scode=5;
 strcpy(dic[14].word,"左腿");
 dic[14].wcode=2;
 strcpy(dic[15].syn,"名詞");
 dic[15].scode=5;
 strcpy(dic[15].word,"右腿");
 dic[15].wcode=3;
 strcpy(dic[16].syn,"副詞");
 dic[16].scode=6;
 strcpy(dic[16].word,"快");
 dic[16].wcode=0;
 strcpy(dic[17].syn,"副詞");
 dic[17].scode=6;
 strcpy(dic[17].word,"慢");
 dic[17].wcode=1;
 strcpy(dic[18].syn,"其他詞");
 dic[18].scode=7;
 strcpy(dic[18].word,"起來");
 dic[18].wcode=0;
//////////////////////--Init Data End
///////////////////////////////////////////

 char expression[30],phrase[40]; 入句子的分詞
 char *p1,*p2; 
 PWORD pw_exp1,pw_exp2,pw_old;   pw_old Can RollBack
 int n_WordNub = 0,n_InputNub = 0; 
 
 strcpy(expression,"請你向前走快抬左手側左腿");
 
 for(int i=0; i<40; i++)
 phrase[i] = ' ' ;
 phrase[39] = '';

 p2 = expression;
 p1 = phrase;
 pw_old = pw_exp1 = (PWORD)p2; 

算輸入的字數以確定查詢的次數
 n_InputNub = CountWord(p2);
 
  printf("分析結果: n");

  for(int nub=0; nub {
 for(int i=0; i<19; i++)  // 查詢詞表
 {
 pw_exp2 = (PWORD)dic[i].word;
 for(int j=0; j {
 if (*pw_exp1 == *pw_exp2)
 {
 pw_exp1++;
 pw_exp2++;
 }
 else
 {
 pw_exp1 = pw_old; 
 }
 }

 if ((char*)pw_exp1 != p2)  表找到單詞
 {
 for(int x=0; x {
 *p1 = *(p2+x);
 p1++;
 }
 *p1 = '/';  入分隔符
 p1++;
 p2 =(char*) pw_exp1;  詢下一個
 pw_old = pw_exp1;  RollBack
 }

 if (*p2 == '') //完了!
 {
 break;
 }

 }
 }

詢完辭典!
 if( ((char*)pw_exp1 == p2) && (*p2 != '') ) 表裡沒有找到
 {
 cout< }
 else{
 cout< cout< 
 return 0;
}

int CountWord(char* words)
{
 int n = 0;
 while(*words != '')
 {
 words++;
 n++;
 }
 return int(n/2);
}