// Document.cpp: implementation of the Document class.
//
//////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "frequency.h"
#include "Document.h"

#ifdef _DEBUG
#undef THIS_FILE
static char THIS_FILE[]=__FILE__;
#define new DEBUG_NEW
#endif

#include "Document.h"

//++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
//++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
//詞表操作

void Document::initTmpWordArray()
//詞匯表初始化
{	
	int i;
	for (i=0;i<=255;i++) tmpWordArray[i]=NULL;	
}

void Document::InCreaseWord(WordType *pWord)
//指定詞條頻度加1
{	
	if (pWord==NULL) return;
	pWord->count ++;
}

void Document::ArrangeWord(WordType * pWord,int intMethod,int Direction)
//intMethod=1 詞條頻率 =2 詞條文內(nèi)權(quán)重
//插入詞條按照詞頻或者詞權(quán)重對(duì)wordtype進(jìn)行排序
{	
	int i,j;
	if (intMethod==1)	
	{
		// 插入排序
		for (i=0;i<intWordCount[Direction-1];i++) //intWordCount描述lastWordArray[]中兩個(gè)方向的詞數(shù).by xfguo
			if (lastWordArray[Direction-1][i].count<=pWord->count)//Direction; 1 正向　；2 反向
			{ 
				for(j=intWordCount[Direction-1]-1;j>=i;j--) 
					lastWordArray[Direction-1][j+1]=lastWordArray[Direction-1][j]; 
				break;
			}
		lastWordArray[Direction-1][i]=*pWord;
		intWordCount[Direction-1]++;
	}
	else if (intMethod==2)	
	{	
		for (i=0;i<intWordCount[Direction-1];i++)
			if (lastWordArray[Direction-1][i].weight_count<=pWord->weight_count )
			{ 
				for(j=intWordCount[Direction-1]-1;j>=i;j--) 
					lastWordArray[Direction-1][j+1]=lastWordArray[Direction-1][j]; 
				break;
			}
		lastWordArray[Direction-1][i]=*pWord;
		intWordCount[Direction-1]++;
	}
	else if (intMethod==0)
	{	
		lastWordArray[Direction-1][intWordCount[Direction-1]]=*pWord;
		intWordCount[Direction-1]++;
	}
}

void Document::AddNewWord(char * strWord,int intWordCount,int intWordLength,float weight)
//加入新的詞匯
{	
	//if (IfExtendWordTable==2) return;
	if (strlen(strWord)>48 ) return ;
	WordType * pWordTmp,*pWordInsert,*pWordStart,*pTmp;	
	pWordTmp=(WordType *)malloc(sizeof(WordType));
    	pWordTmp->next=tmpWordArray[(unsigned char)(*strWord)];
	pWordTmp->count=intWordCount;
	pWordTmp->length=intWordLength;
	pWordTmp->weight=weight; //???
	strcpy(pWordTmp->word,strWord);

	//插入排序
	pWordStart=tmpWordArray[(unsigned char)(*strWord)];
	pWordInsert=NULL;

	while((pWordStart!=NULL)&&(pWordStart->length>intWordLength))
	{	
		pWordInsert=pWordStart;
		pWordStart=pWordStart->next;
	}
	
	if (pWordInsert==NULL)
	{	
		pTmp=tmpWordArray[(unsigned char)(*strWord)];
		tmpWordArray[(unsigned char)(*strWord)]=pWordTmp;
		pWordTmp->next=pTmp;
	}
	else
	{	
		pTmp=pWordInsert;
		pWordInsert->next=pWordTmp;
		pWordTmp->next =pWordStart;
	}
}

void Document::freeWordTable(WordType * pWord)
//詞匯表行的釋放
{	
	WordType *tmpWord;
	while (pWord!=NULL)
	{
		tmpWord=pWord->next;
		free(pWord);
		pWord=tmpWord;
	}

/*	WordType * q;
	while (pWord!=NULL)
	{	
		q=pWord->next;
	        free(pWord); 
		pWord=q;
	}
*/

}

void Document::ArrangeWordInTable(int YuZhi,int intMethod,int Direction)
//intMethod=1 詞條頻率 =2 詞條文內(nèi)權(quán)重 =0 不排序
//輸出詞條排序
//tmpWordArray[]按一定的順序插入到lastWordArray[]中
{	
	int i;
	intWordCount[Direction-1]=0;
	WordType *pWord;
	
	for (i=0;i<=255;i++) 
	{	pWord=tmpWordArray[i];
		while (pWord!=NULL)
		{	if (pWord->count>=YuZhi) ArrangeWord(pWord,intMethod,Direction);
			pWord=pWord->next;
		}
	}
}

void Document::ResetWordTable()//將詞頻設(shè)置為0
{
	int i;
	WordType *pWord;
	for (i=0;i<=255;i++) 
	{	pWord=tmpWordArray[i];
		while (pWord!=NULL)
		{	pWord->count=0;
			pWord->weight_count=0; 
			pWord=pWord->next;
		}
	}
}

void Document::ClearWordTable()//清空詞表內(nèi)的詞條
{	
	int i;
	WordType *pWord;
	for (i=0;i<=255;i++) 
	{	pWord=tmpWordArray[i];
		while (pWord!=NULL)
		{	pWord->count=0;
			pWord->weight_count=0; 
			pWord=pWord->next;
		}
	}
}

WordType * Document::FindWord(char *strWord)
//查詢表中有無(wú)所查的單詞，如果找到，返回指針位置，否則返回NULL
{	
	if (*strWord=='\0') return NULL;//出錯(cuò)誤
	WordType * pStartPlace;
	pStartPlace=tmpWordArray[(unsigned char)(*strWord)];
	while (pStartPlace!=NULL)
	{	
		if (strcmp(strWord,pStartPlace->word)==0) return pStartPlace;  //找到了
		pStartPlace=pStartPlace->next ;
	}
	return NULL;//沒(méi)有找到
}

void Document::freeTmpWordArray()
//詞匯表的釋放
{	
	int i;
	for (i=0;i<=255;i++) {freeWordTable(tmpWordArray[i]);tmpWordArray[i]=NULL;}
}

void Document::initArrangeTable()
//釋放臨時(shí)詞表
{	
	intWordCount[0]=0;
	intWordCount[1]=0;
}

//++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
//++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
//詞條文件操作

//篩選后導(dǎo)出詞條
void Document::WriteWordToFile(char * strFileName,int ifWriteFre,int Direction)
//ifWriteFre =1 輸出詞頻 =0 no
{	
	FILE *fWrite;
	if((fWrite=fopen(strFileName,"w"))==NULL) return ;
	int i;
	int count;
	char strWordTmp[50];
	count=intWordCount[Direction-1];

	for (i=0;i<count;i++)
	{	
		if (Direction==2)
			ReverseString(lastWordArray[Direction-1][i].word,strWordTmp);
		else	strcpy(strWordTmp,lastWordArray[Direction-1][i].word);
		if (ifWriteFre==1){
			//if (strWordTmp[0]<128)
			//	continue;
		 	fprintf(fWrite,"%s %d\n",strWordTmp,lastWordArray[Direction-1][i].count);
		}
		else 	fprintf(fWrite,"%s\n",strWordTmp);
	}
	fclose(fWrite);
}

void Document::copylastword(WordType *pWordtype,int *wordnum)
{             
	int i;
	int count=intWordCount[0];	
		
	*wordnum=intWordCount[0];
	
	for (i=0;i<intWordCount[0];i++)
	{	
		strcpy(pWordtype[i].word,lastWordArray[0][i].word);
                pWordtype[i].count =lastWordArray[0][i].count;
	}
} 

int Document::ReadFileFromText(char* strFileName)
//讀取文本文件內(nèi)容到字符串。自動(dòng)分配內(nèi)存。
{	
	/* if (strSource!=NULL) free(strSource); */
	FILE* fp;
	long lFileLength;
	if((fp=fopen(strFileName,"rb"))!=NULL)
	{           
		fseek(fp,0,SEEK_END);
		lFileLength=ftell(fp);
		rewind(fp);
		
		strSource=(char *)malloc(lFileLength+1);
		fread(strSource,lFileLength,1,fp);
		*(strSource+lFileLength)='\0';
		fclose(fp);
	
                //去掉換行 
		//直接在strsource上修改，那個(gè)更更好？？-

		char * strTmpSource;
		strTmpSource=(char *)malloc(lFileLength+1);
		int i,j;
		j=0;
		for (i=0;i<=lFileLength;i++)
		{
			if (*(strSource+i)!='\r' && *(strSource+i) !='\n')
			{  
				*(strTmpSource+j) = *(strSource+i);
			   	j ++;
			}
		}			
		lFileLength=j;
		free(strSource);
		strSource=(char *)malloc(lFileLength+1);
		*(strSource+lFileLength+1)='\0';
		strcpy(strSource,strTmpSource);
		free(strTmpSource);
		
	  	return 1;
	}
	else
	{
		//AfxMessageBox(strFileName);
		return 0;
	}
}
	
int Document::GetCharFromString(char * strSource,int intStartPlace,char * strReturn,int * intCharLength)
//取得源串中的從絕對(duì)地址開(kāi)始的字符
//返回值：0 末尾；1 字符 ；2 符號(hào) ；-1 出錯(cuò) ；4 回車(chē)
//                 漢語(yǔ)    
{	
	unsigned char char1,char2;
	char strVertueWord[1000];

	int intStrLength;
	WordType * pWord;
	int i;
	
	char1=(unsigned char)(*(strSource+intStartPlace));
	if (char1==0) return 0;
	char2=(unsigned char)(*(strSource+intStartPlace+1));

	if (char1>128) 
	{	
		if (char2==0) return -1;
		*strReturn=(char)(char1);
		*(strReturn+1)=(char)(char2);
		*(strReturn+2)='\0';
		*intCharLength=2;
		
		if (Direction==2) ReverseString("－—；／　。，：、！？的了在你我她它他到是不和《〉》〈～【】〖〗×∥■▲○●→─·…★“”‘’（）—",strVertueWord);
		else strcpy(strVertueWord,"；－—／　。，：、！？的了在你我她它他到是不和《〉 》〈～【】〖〗×∥■▲○●→─·…★“”‘’（）—");
		if (strstr(strVertueWord,strReturn)==NULL) return 1;
		else if ((strstr(strVertueWord,strReturn)-strVertueWord)%2==1) return 1;
		else return 2;
	}
	
	*strReturn=(char)(char1);