scan.cpp

c语言编写

#include "globals.h"
#include "util.h"
#include "scan.h"
#include <stdio.h>
extern FILE* source;
extern int  EchoSource;
typedef enum
{START,INCOMMENT,INNUM,INID,DONE} StateType;

char tokenString[MAXTOKENLEN+1];

#define BUFLEN 256
int lineno = 0;
static char lineBuf[BUFLEN];
static int linepos = 0;
static int bufsize = 0;

static void ungetNextChar(void)
{linepos--;}

static struct
{
	char* str;
	TokenType tok;
}reservedWords[MAXRESERVED]={{"if",IF},{"int",INT},
{"else",ELSE},{"return",RETURN},{"void",VOID},
{"while",WHILE}};


static TokenType reservedLookup(char *s)
{
	int i;
	for(i=0;i<MAXRESERVED;i++)
		if(!strcmp(s,reservedWords[i].str))
			return reservedWords[i].tok;
	return ID;
}	


static char getNextChar(void)
{
    if(!lineno)
	{
		lineno++;
		if(fgets(lineBuf,BUFLEN-1,source))
		{
			if(EchoSource) 
			{
				fprintf(listing,"line %d: %s\n",lineno,lineBuf);
			}
			bufsize = strlen(lineBuf);
			linepos = 0;
			return lineBuf[linepos++];
		}
		else { linepos++; return EOF;}
	}
	else if(!(linepos < bufsize)&&lineBuf[linepos-1]==10)//每一行都以\n结束，最后以行例外
	{
		lineno++;
		if(fgets(lineBuf,BUFLEN-1,source))
		{
			if(EchoSource) 
			{
				fprintf(listing,"line %d: %s\n",lineno,lineBuf);
			}
			bufsize = strlen(lineBuf);
			linepos = 0;
			return lineBuf[linepos++];
		}
		else { linepos++; return EOF;}
	}	
	else if(!(linepos < bufsize))
		return EOF;
	else 
		return lineBuf[linepos++];
}

TokenType getToken(void)
{
	int tokenStringIndex=0;
	TokenType currentToken;
	StateType state=START;
	int save;
	char c;
	while(state!=DONE)
	{
		save=TRUE;
		c=getNextChar();
		switch(state)
		{
		case START:
			if(isdigit(c))
				state=INNUM;
			else if(isalpha(c))
				state=INID;
			else if((c==' ')||(c=='\t')||(c=='\n'))
				save=FALSE;
			else if(c=='/')
			{
				c=getNextChar();
				if(c=='*')
				{
					save=FALSE;
					state=INCOMMENT;
				}
				else
				{
					currentToken=OVER;
					state=DONE;
					ungetNextChar();
				}
			}
			else if(c=='<')
			{
				state=DONE;
				tokenString[tokenStringIndex++]=c;
				c=getNextChar();
				if(c=='=')
				{
				    currentToken=LTOREQ;
				}
				else
				{
					save=FALSE;
					currentToken=LT;
					ungetNextChar();
				}
			}
			else if(c=='>')
			{
				state=DONE;
				tokenString[tokenStringIndex++]=c;
				c=getNextChar();
				
				if(c=='=')
				{
				    currentToken=BIOREQ;
					
				//	c=next;
				}
				else
				{
					//flag=1;//预取了一个字符
				//	tokenString[tokenStringIndex++]=c;
				//	c=next;
					currentToken=BI;
					save=FALSE;
					ungetNextChar();
				}
			//	next=c;
			}
			else if(c=='=')
			{
				state=DONE;
				c=getNextChar();
				if(c=='=')
				{
				    currentToken=EQ;
				}
				else
				{
					currentToken=ASSIGN;
					save=FALSE;
					ungetNextChar();
				}
			}
			else if(c=='!')
			{
				state=DONE;
				c=getNextChar();
				if(c=='=')
				{
				    currentToken=NEQ;
				//	tokenString[tokenStringIndex++]=c;
				//	c=next;
				}
				else
				{
				//	flag=1;//预取了一个字符	
				//	tokenString[tokenStringIndex++]=c;
				//	c=next;
					currentToken=ERROR;
					save=FALSE;
					ungetNextChar();
				}
			}
			else
			{
				state=DONE;
				switch(c)
				{
				case EOF:
					save=FALSE;
					currentToken=ENDFILE;
					break;
				case '+':
					currentToken=PLUS;
					break;
				case '-':
					currentToken=MINUS;
					break;
				case '(':
					currentToken=LROUNDBRA;
					break;
				case ')':
					currentToken=RROUNDBRA;
					break;
				case '[':
					currentToken=LSQUARPAREN;
					break;
				case ']':
					currentToken=RSQUARPAREN;
					break;
				case '{':
					currentToken=LBRAC;
					break;
				case '}':
					currentToken=RBRAC;
					break;
				case ';':
					currentToken=SEMI;
					break;
				case ',':
					currentToken=COMA;
					break;
				case '*':
					currentToken=TIMES;
					break;
				default:
					currentToken=ERROR;
					break;
				}
			}
			break;
		case INCOMMENT:
			save=FALSE;
			if(c=='*')
			{
				c=getNextChar();
				if(c=='/')
				{
					state=START;
				}
				else
				{
					//flag=1;//预取了一个字符	
				//	c=next;
					ungetNextChar();
				}
			}
			break;
		case INNUM:
			if(!isdigit(c))
			{
			//	flag=1;
				save=FALSE;
				state=DONE;
				currentToken=NUM;
				ungetNextChar();
			}
			break;
		case INID:
			if(!isalpha(c))
			{
			//	flag=1;
				save=FALSE;
				state=DONE;
				currentToken=ID;
				ungetNextChar();

			}
			break;
		case DONE:
		default:
			fprintf(listing,"Scanner bug:state=%d\n",state);
			state=DONE;
			currentToken=ERROR;
			break;
		}
		if((save)&&(tokenStringIndex<=MAXTOKENLEN))
			tokenString[tokenStringIndex++]=c;
		if(state == DONE)
		{
			tokenString[tokenStringIndex] = '\0';
			if(currentToken == ID)
				currentToken = reservedLookup(tokenString);
		}
	}//循环结束
	if(TraceScan)
	{
		fprintf(listing,"%d:\n",lineno);
		printToken(currentToken,tokenString);
	}
	return currentToken;
}