pageinit.cpp

自己用Markov模型做的一个整句物笔输入法的原型

 // PageInit.cpp : implementation file
//

#include "stdafx.h"
#include "HikerWBSet.h"
#include "PageInit.h"
#include "MessWnd.h"
#include <AFXTEMPL.H>
#include "softkeyboard.h"

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

#define LINE_LEN 256
#define CODE_LEN 20
#define WORD_LEN 21
//#define WORD_NUM 65536
#define POST_SUM 0.9f
#define AUTOADD2CCODE	TRUE
#define DEFAULT_PRIOR 1.0f
#define TRANS_SUMLMT 0.6f				//how much probabilitie assigned to exist words


struct CLItem
{
	CString code;
	int		numw;				//duplicate word number
	int		numw2;				//duplicate word number including binary code word
	CLItem ()
	{
		code = "";
		numw = 0;
		numw2 = 0;
	}
	CLItem (CString cd, int nw, int nw2)
	{
		code = cd;
		numw = nw;
		numw2 = nw2;
	}
};

struct WLItem
{
	CString word;
	float	pp;
	WLItem ()
	{
		word = "";
		pp = 0.0f;
	}
	WLItem (CString wd, float p)
	{
		word = wd;
 		pp = p;
	}
};

struct CWLItem
{
	CString code;
	CString word;
	float	post;		//人为设定一初始后验概率
	float	post2;		//人为设定一初始后验概率
	BOOL	c2style;			//二元码扩展条目
	CWLItem ()
	{
		code = "";
		word = "";
		post = 0.0f;
		post2 = 0.0f;
		c2style = FALSE;
	}
	CWLItem (CString cd, CString wd, float p, float p2, BOOL c2s)
	{
		code = cd;
		word = wd;
		post = p;
		post2 = p2;
		c2style = c2s;
	}
};

struct WCLItem
{
	CString word;
	CString code;
	float	cp;		//人为设定一初始后验概率
	float	cp2;		//人为设定一初始后验概率
	BOOL	c2style;
	WCLItem ()
	{
		word = "";
		code = "";
		cp = 0.0f;
		cp2 = 0.0f;
		c2style = FALSE;
	}
	WCLItem (CString wd, CString cd, float p, float p2, BOOL c2s)
	{
		word = wd;
		code = cd;
		cp = p;
		cp2 = p2;
		c2style = c2s;
	}
};

struct WWLItem
{
	CString word;
	CString wordc;
	float	tp;		//人为设定一初始后验概率
	WWLItem ()
	{
		word = "";
		wordc = "";
		tp = 0.0f;
	}
	WWLItem (CString wd, CString wdc, float p)
	{
		word = wd;
		wordc = wdc;
		tp = p;
	}
};


//first level words
static unsigned char p1CWord[] = "工了以在有地一上不是中国同民为这我的要和产发人经主";

static unsigned char gKeyArr[SOFTKEYBOARD_NUM*2][SOFTKEYBOARD_LEN] = {
		/*PC keyboard*/
		{
			SOFTKEY_PCC
		},
		{
			SOFTKEY_PCL
		},
		/*Greece letter*/																																								
		{
			SOFTKEY_GLC
		},
		{
			SOFTKEY_GLL
		},
		/*Russia letter*/																																							
		{
			SOFTKEY_RLC
		},
		{
			SOFTKEY_RLL
		},
		/*chinese spell letter*/																																									
		{
			SOFTKEY_NON
		},	
		{
			SOFTKEY_CSL
		},
		/*phonetic notation*/																																										
		{
			SOFTKEY_NON
		},	
		{
			SOFTKEY_CPN
		},
		/*Japanese hiragana */																																								
		{
			SOFTKEY_JHC
		},		
		{
			SOFTKEY_JHL
		},
		/*Japanese katakana*/																																			
		{
			SOFTKEY_JKC
		},
		{
			SOFTKEY_JKL
		},
		/*punctuation*/																																						
		{
			SOFTKEY_NON
		},
		{
			SOFTKEY_PCT
		},
		/*serial number*/																					
		{
			SOFTKEY_SNC
		},
		{
			SOFTKEY_SNL
		},
		/*mathematics notation*/
		{
			SOFTKEY_NON
		},
		{
			SOFTKEY_MAT
		},
		/*special notation*/																																								
		{
			SOFTKEY_NON
		},
		{
			SOFTKEY_SPC
		},
		/*全角字符*/
		{
			SOFTKEY_SBC
		},
		{
			SOFTKEY_SBL
		},
};


/////////////////////////////////////////////////////////////////////////////
// CPageInit property page

IMPLEMENT_DYNCREATE(CPageInit, CPropertyPage)

CPageInit::CPageInit() : CPropertyPage(CPageInit::IDD)
{
	//{{AFX_DATA_INIT(CPageInit)
	m_csFileName = _T("");
	m_csFileName2 = _T("");
	//}}AFX_DATA_INIT
}

CPageInit::~CPageInit()
{
}

void CPageInit::DoDataExchange(CDataExchange* pDX)
{
	CPropertyPage::DoDataExchange(pDX);
	//{{AFX_DATA_MAP(CPageInit)
	DDX_Text(pDX, IDC_EDIT_FILENAME, m_csFileName);
	DDV_MaxChars(pDX, m_csFileName, 256);
	DDX_Text(pDX, IDC_EDIT_FILENAME2, m_csFileName2);
	//}}AFX_DATA_MAP
}


BEGIN_MESSAGE_MAP(CPageInit, CPropertyPage)
	//{{AFX_MSG_MAP(CPageInit)
	ON_BN_CLICKED(IDC_BUTTON_CONVERT, OnButtonConvert)
	ON_BN_CLICKED(IDC_BUTTON_FILEOPEN, OnButtonFileopen)
	ON_BN_CLICKED(IDC_BUTTON_FILEOPEN2, OnButtonFileopen2)
	ON_EN_CHANGE(IDC_EDIT_FILENAME, OnChangeEditFilename)
	ON_EN_CHANGE(IDC_EDIT_FILENAME2, OnChangeEditFilename2)
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CPageInit message handlers

static int ReadLine(CFile* file, unsigned char* buffl)
{
	unsigned char c;
	unsigned char* bp = buffl;
	int len = 0;

	for ( ; file->Read(&c, 1); )
	{
		
		if (c == 13)
		{
			file->Read(&c, 1);	//skip char(10)
			break;
		}
		else
		{
			*(bp++) = c;
			len++;
		}
	}

	return len;
}

static int ReadCode(unsigned char*& blp,unsigned char* buffc)
{
	unsigned char c;
	unsigned char* bcp = buffc;
	
	int i = 0;
	for ( ; ; )
	{
		c = *(blp++);
		if (c == 32)
			break;
		else if (c >= 'a' && c <= 'z')
		{
			*(bcp++) = c;
		}

		if (bcp - buffc > CODE_LEN - 1)
			break;
	}

	return bcp - buffc;
}

static int ReadWord(unsigned char*& blp,unsigned char* buffw)
{
	unsigned char c;
	unsigned char* bwp = buffw;
	int len = 0;
	int fstbyte = 1;		//mark for double byte chinese

	int i = 0;
	for ( ; ; )
	{
		c = *(blp++);
		if (c == 0)
			break;
		else if (c == 32 && (bwp - buffw) != 0)
			break;
		else		//???
		{
			if (c >= 128 && fstbyte == 1)	//first byte should be no less than 128
			{
				*(bwp++) = c;
				fstbyte = 0;
			}
			else if (fstbyte == 0)
			{
				*(bwp++) = c;
				fstbyte = 1;
			}

			if (bwp-buffw > WORD_LEN-1)
				break;
		}
	}

	return bwp - buffw;
}

static int CalcLineNum(CFile* file)
{
	int numl = 0;
	unsigned char c;
	DWORD oldpos = file->GetPosition();

	for ( ;; )
	{
		if (file->Read(&c, 1))
		{
			if (c == 10)
				numl++;
		}
		else
			break;
	}

	file->Seek(oldpos, CFile::begin);
	return numl;
}

static int GetCodeId(CString code)
{
	unsigned char c;
	int id = 0;
	int len = min(code.GetLength(), 4);
	for (int i = 0; i < len; i++)
	{
		c = code.GetAt(i) - 'a'+1;
		id += (c << (5*i));
	}

	return id;
}

static int GetWordId(CString word)
{
	unsigned char c;
	int id = 0;
	int len = 2;
	for (int i = 0; i < len; i++)
	{
		c = word.GetAt(i);
		id += (c << (i<<3));		//8*i
	}

	return id;
}

static BOOL Is1CWord(CString last_strw)
{
	unsigned char c1, c2;

	if (2 != last_strw.GetLength())
		return FALSE;

	c1 = last_strw[0];
	c2 = last_strw[1];
	for (int i = 0; i < 50; i += 2)
	{
		if (c1 == p1CWord[i] && c2 == p1CWord[i+1])
			return TRUE;
	}
	
	return FALSE;
}

//generate next speical code according to the given code.
//special code is leaded by character 'z' and generated automatically, which range is from zaaa to zyyy
static void NextSpcCode(CString& csSpcCode)
{
	for (int i = 3; i > 0; i--)
	{
		if (csSpcCode[i] < 'y')
		{
			csSpcCode.SetAt(i, csSpcCode[i]+1);
			break;
		}
		else
		{
			for (int j = i; j < 4; j++)
				csSpcCode.SetAt(j, 'a');
		}
	}
}

//find the code's iword-th position in code-word table
static POSITION CWFind(CString code, CString word, 
					   CList<CWLItem, CWLItem&> *listcw,
					   CMap<int, int&, POSITION, POSITION&> *cwidx)
{
	CWLItem item;
	POSITION pos = 0;
	int id = GetCodeId(code);

	cwidx->Lookup(id, pos);
	for( ; pos; listcw->GetNext(pos))
	{
		item = listcw->GetAt(pos);
		if (item.code != code)
			break;

		if (item.word == word)
			return pos;
	}
	return FALSE;
}

//find the code-word item in code-word table.
//if the item not found, return FALSE, outherwise, TRUE
//posloc gives the position when found, 
//if the code exists, posloc gives the code's the last appearing position,
//otherwise, it gives last appearing position+1, or 0;
static BOOL CWLocate(CString code, CString word, 
					   CList<CWLItem, CWLItem&> *listcw,
					   CMap<int, int&, POSITION, POSITION&> *cwidx, POSITION& posloc)
{	
	posloc = 0;
	int id = GetCodeId(code);
	cwidx->Lookup(id, posloc);

	for (; posloc; listcw->GetNext(posloc))
	{
		CWLItem item = listcw->GetAt(posloc);
		if ( item.code != code)
			break;
		else if (item.word == word)