cifafenxi.cpp

一个很不错的可视化词法分析器 用C++写的

// Compiler.cpp: implementation of the CCompiler class.
//
//////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "compiler.h"
#include "cifafenxi.h"

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

//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////

CCompiler::CCompiler()
{
	m_ntCifaLen=0;
	m_aSourse[0]=0;
	m_Interface.SourseBuffer=m_aSourse;
	m_Interface.LexicalDuad=m_tCifa;
	m_Interface.ErrorMessageTable=m_tErrorMessage;

}

CCompiler::~CCompiler()
{

}

void CCompiler::Initial()
{
	::strcpy(m_QuadOpCodeTable[0], "_End_");
	::strcpy(m_QuadOpCodeTable[1], "in");
	::strcpy(m_QuadOpCodeTable[2], "call");
	::strcpy(m_QuadOpCodeTable[3], "out");
	::strcpy(m_QuadOpCodeTable[4], "ret");
	::strcpy(m_QuadOpCodeTable[5], ":=I");
	::strcpy(m_QuadOpCodeTable[6], ":=R");
	::strcpy(m_QuadOpCodeTable[7], "par");
	::strcpy(m_QuadOpCodeTable[8], "valueI");
	::strcpy(m_QuadOpCodeTable[9], "valueR");
	::strcpy(m_QuadOpCodeTable[10],"addr");
	::strcpy(m_QuadOpCodeTable[11],"j");
	::strcpy(m_QuadOpCodeTable[12],"j<");
	::strcpy(m_QuadOpCodeTable[13],"j<=");
	::strcpy(m_QuadOpCodeTable[14],"j>");
	::strcpy(m_QuadOpCodeTable[15],"j>=");
	::strcpy(m_QuadOpCodeTable[16],"j=");
	::strcpy(m_QuadOpCodeTable[17],"j<>");
	::strcpy(m_QuadOpCodeTable[18],"`");
	::strcpy(m_QuadOpCodeTable[19],"+I");
	::strcpy(m_QuadOpCodeTable[20],"+R");
	::strcpy(m_QuadOpCodeTable[21],"*I");
	::strcpy(m_QuadOpCodeTable[22],"*R");
	::strcpy(m_QuadOpCodeTable[23],"`");
	::strcpy(m_QuadOpCodeTable[24],"`");
	::strcpy(m_QuadOpCodeTable[25],"`");
	::strcpy(m_QuadOpCodeTable[26],"`");
	::strcpy(m_QuadOpCodeTable[27],"`");
	::strcpy(m_QuadOpCodeTable[28],"`");
	::strcpy(m_QuadOpCodeTable[29],"_Error_");
	::strcpy(m_tKeyWord[0],"");
	::strcpy(m_tKeyWord[1],"");//标识符
	::strcpy(m_tKeyWord[2],"");//正整数
	::strcpy(m_tKeyWord[3],"program");
	::strcpy(m_tKeyWord[4],"var");
	::strcpy(m_tKeyWord[5],"procedure");
	::strcpy(m_tKeyWord[6],"begin");
	::strcpy(m_tKeyWord[7],"end");
	::strcpy(m_tKeyWord[8],"if");
	::strcpy(m_tKeyWord[9],"then");
	::strcpy(m_tKeyWord[10],"else");
	::strcpy(m_tKeyWord[11],"while");
	::strcpy(m_tKeyWord[12],"do");
	::strcpy(m_tKeyWord[13],"call");
	::strcpy(m_tKeyWord[14],"integer");
	::strcpy(m_tKeyWord[15],"real");
	::strcpy(m_tKeyWord[16],"+");
	::strcpy(m_tKeyWord[17],"-");
	::strcpy(m_tKeyWord[18],"*");
	::strcpy(m_tKeyWord[19],"/");
	::strcpy(m_tKeyWord[20],"~");
	::strcpy(m_tKeyWord[21],"&");
	::strcpy(m_tKeyWord[22],"|");
	::strcpy(m_tKeyWord[23],"<");
	::strcpy(m_tKeyWord[24],"<=");
	::strcpy(m_tKeyWord[25],">");
	::strcpy(m_tKeyWord[26],">=");
	::strcpy(m_tKeyWord[27],"=");
	::strcpy(m_tKeyWord[28],"<>");
	::strcpy(m_tKeyWord[29],":=");
	::strcpy(m_tKeyWord[30],";");
	::strcpy(m_tKeyWord[31],".");
	::strcpy(m_tKeyWord[32],",");
	::strcpy(m_tKeyWord[33],"(");
	::strcpy(m_tKeyWord[34],")");
	::strcpy(m_tKeyWord[35],":");
	::strcpy(m_tKeyWord[36],"/*");
	::strcpy(m_tKeyWord[37],"*/");
	::strcpy(m_tKeyWord[38],"");
	::strcpy(m_tKeyWord[39],"");
	::strcpy(m_tKeyWord[40],"");
	::strcpy(m_tKeyWord[41],"");
	::strcpy(m_tKeyWord[42],"");
	::strcpy(m_tKeyWord[43],"");
	::strcpy(m_tKeyWord[44],"");
	::strcpy(m_tKeyWord[45],"");
	::strcpy(m_tKeyWord[46],"");
	::strcpy(m_tKeyWord[47],"");
	::strcpy(m_tKeyWord[48],"");
	::strcpy(m_tKeyWord[49],"");
/*
	::strcpy(m_tErrorMessage[0],"正确！");
	::strcpy(m_tErrorMessage[1],"数字位数过长！");
	::strcpy(m_tErrorMessage[2],"非法字符！");
	::strcpy(m_tErrorMessage[3],"缺少保留字“ program ”！");
	::strcpy(m_tErrorMessage[4],"program 后缺少标识符！");
	::strcpy(m_tErrorMessage[5],"缺少“ ; ”符号！");
	::strcpy(m_tErrorMessage[6],"缺少程序结束符“ . ”符号！");
	::strcpy(m_tErrorMessage[7],"缺少《程序体》，应为 begin,integer,real,procedure");
	::strcpy(m_tErrorMessage[8],"缺少保留字“ begin ”！");
	::strcpy(m_tErrorMessage[9],"缺少《语句》，应为 begin,ID,if,while,call");
	::strcpy(m_tErrorMessage[10],"缺少保留字“ end ”！");
	::strcpy(m_tErrorMessage[11],"缺少《说明》，应为 integer,real,procedure");
	::strcpy(m_tErrorMessage[12],"缺少保留字 integer 或 real！");
	::strcpy(m_tErrorMessage[13],"procedure 后缺少标识符！");
	::strcpy(m_tErrorMessage[14],"简单变量说明后缺少标识符！");
	::strcpy(m_tErrorMessage[15],"缺少保留字“ procedure ”！");
	::strcpy(m_tErrorMessage[16],"缺少 “ ( ”符号！");
	::strcpy(m_tErrorMessage[17],"缺少 “ ) ”符号！");
	::strcpy(m_tErrorMessage[18],"缺少形参标识符 ！");
	::strcpy(m_tErrorMessage[19],"缺少 “ : ”符号！");
	::strcpy(m_tErrorMessage[20],"缺少标识符！");
	::strcpy(m_tErrorMessage[21],"缺少赋值符号“ := ”！");
	::strcpy(m_tErrorMessage[22],"缺少保留字“ if ”！");
	::strcpy(m_tErrorMessage[23],"缺少保留字“ then ”！");
	::strcpy(m_tErrorMessage[24],"缺少保留字“ while ”！");
	::strcpy(m_tErrorMessage[25],"缺少保留字“ do ”！");
	::strcpy(m_tErrorMessage[26],"缺少保留字“ call ”！");
	::strcpy(m_tErrorMessage[27],"缺少被调过程名标识符！");
	::strcpy(m_tErrorMessage[28],"缺少《因子》，应为 (,ID,NUMBER");
	::strcpy(m_tErrorMessage[29],"缺少《布尔表达式》，应为 ~,(,ID,NUMBER");
	::strcpy(m_tErrorMessage[30],"缺少《关系》！");
	::strcpy(m_tErrorMessage[31],"变量名不能和过程名相同！");
	::strcpy(m_tErrorMessage[32],"标识符重复定义！");
	::strcpy(m_tErrorMessage[33],"未定义标识符！");
	::strcpy(m_tErrorMessage[34],"不能直接引用过程名！");
	::strcpy(m_tErrorMessage[35],"不能从real转换为integer类型！");
	::strcpy(m_tErrorMessage[36],"不能用常数作实参！");
	::strcpy(m_tErrorMessage[37],"变参应为变量！");
	::strcpy(m_tErrorMessage[38],"实参个数不足！");
	::strcpy(m_tErrorMessage[39],"只有integer和integer才能比较！");
	::strcpy(m_tErrorMessage[40],"不能这样调用过程！");
	::strcpy(m_tErrorMessage[41],"");
	::strcpy(m_tErrorMessage[42],"");
	::strcpy(m_tErrorMessage[43],"");
	::strcpy(m_tErrorMessage[96],"源程序结束符 end. 后还有多余单词！");
	::strcpy(m_tErrorMessage[97],"语法错误太多，终止语法分析！");
	::strcpy(m_tErrorMessage[98],"源程序不正常结束！");
	::strcpy(m_tErrorMessage[99],"内存不足！词法分析终止！");
*/
}

char CCompiler::GetCurCh()//get ch and point does not ++
{
	return m_aSourse[m_pCurCh];
}

void CCompiler::NextCh()//point ++
{
	m_pCurCh++;
}

char CCompiler::GetCh()//get ch and point ++
{
	return m_aSourse[m_pCurCh++];
}

char CCompiler::GetNextCh()//get next ch and point does not ++
{
	if (m_aSourse[m_pCurCh]==0) return 0;
	return m_aSourse[m_pCurCh+1];
}

void CCompiler::CiFaFengXi()
{
	BOOL flag=FALSE;
	char token[MAX_LEX_LEN];
	int k,v;
	Initial();
	for (int i=0;i<m_ntCifaLen;i++)
		delete m_tCifa[i];
	m_pCurCh=0;
	m_ntCifaLen=0;
	while (GetCurCh())//GetSYM
	{
		if (flag)
		{
			while (!((GetCurCh()=='*')&&(GetNextCh()=='/')))
				if (GetCurCh()) NextCh();else break;
			if (GetCurCh())
			{
				NextCh();NextCh();
				flag=FALSE;
			}
		}
		while (1)
		{
			while ((GetCurCh()==32) || (GetCurCh()==9))
				NextCh();
			if (!((GetCurCh()==13) && (GetNextCh()==10)))
				break;
			NextCh();NextCh();
		}
		if (isalpha(GetCurCh()))//alpha
		{
			k=0;
			while (1)
			{
				if (k<MAX_LEX_LEN-1) token[k++]=GetCh();
				else NextCh();
				if (!isalnum(GetCurCh())) break;
			}
			token[k]=0;
			v=FindInKeyWordTab(token);
			if (v)
			{
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh-k;
				m_ntCifaLen++;
			}
			else
			{
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=FindInCifaTab(token);
				m_tCifa[m_ntCifaLen]->type=1;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh-k;
				m_ntCifaLen++;
			}
		}
		else if (isdigit(GetCurCh()))//digit
		{
			k=0;
			BOOL error=FALSE;
			while (1)
			{
				if (k<=MAX_NUMBER_LEN) token[k++]=GetCh();
				else
				{
					error=TRUE;
					NextCh();
				}
				if (!isdigit(GetCurCh())) break;
			}
			token[k]=0;
			v=::strtol(token,NULL,10);
			if (error)
			{
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=1;
				m_tCifa[m_ntCifaLen]->type=0;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh-k;
				m_ntCifaLen++;
			}
			else
			{
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=v;
				m_tCifa[m_ntCifaLen]->type=2;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh-k;
				m_ntCifaLen++;
			}
		}
		else
		switch (GetCurCh())//other
		{
		case '+':
		case '-':
		case '*':
		case '~':
		case '&':
		case '|':
		case '=':
		case ';':
		case '.':
		case ',':
		case '(':
		case ')':
			token[0]=GetCurCh();
			token[1]=0;
			v=FindInKeyWordTab(token);
			m_tCifa[m_ntCifaLen]=new CifaResult;
			m_tCifa[m_ntCifaLen]->value=0;
			m_tCifa[m_ntCifaLen]->type=v;
			::strcpy(m_tCifa[m_ntCifaLen]->text,token);
			m_tCifa[m_ntCifaLen]->address=m_pCurCh;
			m_ntCifaLen++;
			NextCh();
			break;
		case '\0':
			break;
		case '/':
			switch (GetNextCh())
			{
			case '*':
				NextCh();
				NextCh();
				flag=TRUE;
				break;
			default:
				token[0]=GetCurCh();
				token[1]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
			}
			break;
		case '<':
			switch (GetNextCh())
			{
			case '=':
				token[0]=GetCurCh();
				token[1]=GetNextCh();
				token[2]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
				NextCh();
				break;
			case '>':
				token[0]=GetCurCh();
				token[1]=GetNextCh();
				token[2]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
				NextCh();
				break;
			default:
				token[0]=GetCurCh();
				token[1]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
			}
			break;
		case '>':
			switch (GetNextCh())
			{
			case '=':
				token[0]=GetCurCh();
				token[1]=GetNextCh();
				token[2]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
				NextCh();
				break;
			default:
				token[0]=GetCurCh();
				token[1]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
			}
			break;
		case ':':
			switch (GetNextCh())
			{
			case '=':
				token[0]=GetCurCh();
				token[1]=GetNextCh();
				token[2]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
				NextCh();
				break;
			default:
				token[0]=GetCurCh();
				token[1]=0;
				v=FindInKeyWordTab(token);
				m_tCifa[m_ntCifaLen]=new CifaResult;
				m_tCifa[m_ntCifaLen]->value=0;
				m_tCifa[m_ntCifaLen]->type=v;
				::strcpy(m_tCifa[m_ntCifaLen]->text,token);
				m_tCifa[m_ntCifaLen]->address=m_pCurCh;
				m_ntCifaLen++;
				NextCh();
			}
			break;
		default:
			token[0]=GetCurCh();
			token[1]=0;
			m_tCifa[m_ntCifaLen]=new CifaResult;
			m_tCifa[m_ntCifaLen]->value=2;
			m_tCifa[m_ntCifaLen]->type=0;
			::strcpy(m_tCifa[m_ntCifaLen]->text,token);
			m_tCifa[m_ntCifaLen]->address=m_pCurCh;
			m_ntCifaLen++;
			NextCh();
		}//switch
		if (m_ntCifaLen==LEX_DUAD_TABLE_LEN-2)//out of table space
		{
			m_tCifa[m_ntCifaLen]=new CifaResult;
			m_tCifa[m_ntCifaLen]->value=99;
			m_tCifa[m_ntCifaLen]->type=0;
			::strcpy(m_tCifa[m_ntCifaLen]->text,"");
			m_tCifa[m_ntCifaLen]->address=m_pCurCh-1;
			m_ntCifaLen++;
			break;
		}
	}//GetSYM
	m_tCifa[m_ntCifaLen]=new CifaResult;
	m_tCifa[m_ntCifaLen]->value=0;
	m_tCifa[m_ntCifaLen]->type=-1;//end node
	::strcpy(m_tCifa[m_ntCifaLen]->text,"");
	m_tCifa[m_ntCifaLen]->address=0;
	m_ntCifaLen++;
	return;
}

int CCompiler::FindInKeyWordTab(char * a)//查找关键字
{
	for (int i=0;i<KEY_WORD_TABLE_LEN;i++)
		if (!::stricmp(m_tKeyWord[i],a)) return i;
	return 0;
}

int CCompiler::FindInCifaTab(char * a)
{
	int v=1;
	for (int i=0;i<m_ntCifaLen;i++)
	{
		if (m_tCifa[i]->type==1)
		{
			v++;
			if (!::stricmp(m_tCifa[i]->text,a))
				return m_tCifa[i]->value;
		}
	}
	return v;
}
InterfaceType * CCompiler::GetTable()
{
	return &m_Interface;
}