📄 lexer.cpp
字号:
#include"init.h"
void parse()
{
char ch, *arr="";
char filenamein[20],filenameout[20];
int i=0,j=0,count=0;
FILE *fpin , *output;
int type[]={0,1,2,3}; /* 关键字，标示符，数字，符号*/
printf(" Input the name of your input file: ");
scanf("%s",filenamein);
if((fpin=fopen(filenamein, "r"))==NULL)
{
printf(" cannot open the file\n");
exit(0);
}
printf(" Input the name of your output file: ");
scanf("%s",filenameout);
if((output=fopen(filenameout,"w"))==NULL)
{
printf(" cannot open the file\n");
exit(0);
}
ch=fgetc(fpin);
while(ch!=EOF)
{
i=0;
if( ch==' '|| ch =='\t') /* 去除空格*/
ch=fgetc(fpin);
else if( ch=='\n') /* 去除回车符*/
{
lineno++;
ch=fgetc(fpin);
}
else
if( isalpha(ch)||ch=='_') /* 开头是字母或下划线*/
{
while(isalpha(ch)||isdigit(ch)||ch=='_')
{
arr[i++] = ch;
ch=fgetc(fpin);
}
arr[i]='\0';
if( Find(arr))
{
fprintf(output,"%-10s\t%5d\t%-5s \n",arr,type[1],"id"); /* 标示符*/
}
else
{
fprintf(output,"%-10s\t%5d\t%-5s\n",arr,type[0],"kword"); /* 是关键字*/
}
continue;
}
else
if(isdigit(ch)) /* 分析数字的情况*/
{
while(isdigit(ch))
{
arr[i]=ch;
i++;
ch=fgetc(fpin);
}
if (ch=='.') /*出现非点时*/
{
arr[i]=ch; /*将点加入*/
i++;
ch=fgetc(fpin);/*继续读取*/
if (isdigit(ch))
{
while(isdigit(ch)) /*读取的是数字时，加入*/
{
arr[i]=ch;
i++;
ch=fgetc(fpin);
}
/*如果是数字加点，以后再加非数字的情况出现时，就是错误*/
if(isalpha(ch))
{
while(isdigit(ch)||isalpha(ch)||ch=='_')
{
arr[i++]=ch;
ch=fgetc(fpin);
}
arr[i]='\0';
fprintf(output,"%-10s\t unexpect in %-5d line ",arr,lineno);
continue; /* 退出当前循环 if */
}
/*当出现结束符时，就收入为实数内*/
else
{
arr[i]='\0';
fprintf(output,"%-10s\t%5d\t%-5s\n",arr,type[2],"num");/* 实数 */
continue;
}
} /* 157 line if */
} /* 152 line if */
/*如果是字符,则判断为标识错误*/
else if(isalpha(ch))
{
while(isdigit(ch)||isalpha(ch)||ch=='.')
{
arr[i++]=ch;
ch=fgetc(fpin);
}
arr[i]='\0';
fprintf(output,"%-10s\t unexpect in %-5d line ",arr,lineno);
continue;
}
/*如果读取的为结束时，就判断为常数*/
else
{
arr[i]='\0';
fprintf(output,"%-10s\t%5d\t%-5s\n",arr,type[2],"num");
continue;
}
} /* 142 else if */
else if( ch=='>')
{
arr[i++]=ch;
ch=fgetc(fpin);
if( ch=='=')
{
fprintf(output,"%-10s\t%5d\t%-5d\n",">=",type[3],25); /* 为运算符 ">="*/
ch=fgetc(fpin);
continue;
}
else
{
fprintf(output,"%-10s\t%5d\t%-5d\n",">",type[3],24); /* 为运算符 ">"*/
ch=fgetc(fpin);
continue;
}
}
else if (ch=='<')
{
arr[i++]=ch;
ch=fgetc(fpin);
if( ch=='=')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","<=",type[3],23); /* 为运算符 "<="*/
ch=fgetc(fpin);
continue;
}
if( ch=='<')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","<<",type[3],20); /* 为运算符 "<<"*/
ch=fgetc(fpin);
continue;
}
fprintf(output,"%-10s\t%5d\t%-5d\n","<",type[3],22); /* 为运算符 "<"*/
continue;
}
else if(ch=='"') /* 处理字符串的问题 */
{
arr[i++]=ch;
ch=fgetc(fpin);
while(ch!='"'&&ch!=EOF)
{
arr[i++]=ch;
ch=fgetc(fpin);
}
if(ch=='"')
{
arr[i++]=ch;
ch=fgetc(fpin);
arr[i]='\0';
fprintf(output,"%-10s\t\t\t字符串\n",arr);
continue;
}
}
else
if(ch=='-')
{
arr[i++]=ch;
ch=fgetc(fpin);
if( ch=='-')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","--",type[3],12); /* 为运算符 "--"*/
ch=fgetc(fpin);
continue;
}
if(ch=='>')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","->",type[3],8); /* 为运算符 "->"*/
ch=fgetc(fpin);
continue;
}
arr[i]='\0';
fprintf(output, "%-10s\t%5d\t%-5d\n",arr,type[3],19); /* 为运算符 "-"*/
ch=fgetc(fpin);
continue;
}
else
if(ch=='+')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='+')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","++",type[3],11); /* 为运算符 "++" */
ch=fgetc(fpin);
continue;
}
arr[i]='\0';
fprintf(output, "%-10s\t%5d\t%-5d\n",arr,type[3],18); /* 为运算符 "+" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='*')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","*",type[3],15); /* 为运算符 "*" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='/')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='*') /*若为注释的开始，标示出包含在里面的所有字符*/
{
arr[i++]=ch;
count=0;
ch=fgetc(fpin);
while(count!=2)
{ /*当扫描到‘*’且紧接着下一个字符为‘/’才是注释的结束*/
count=0;
while(ch!='*')
{
arr[i++]=ch;
ch=fgetc(fpin);
}
count++;
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='/')
count++;
else
{
ch=fgetc(fpin);
}
}
arr[i++]=ch;
arr[i]='\0';
fprintf(output,"%s\t 注释\n",arr);
continue;
}
else
if(ch=='=')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","/=",type[3],34); /* 为运算符 "/=" */
continue;
}
else
{
fprintf(output, "%-10s\t%5d\t%-5d\n","/",type[3],16); /* 为运算符 "/" */
ch=fgetc(fpin);
continue;
}
}
else
if(ch=='%')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='=')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","%=",type[3],42); /* 为运算符 "%=" */
continue;
}
else
{
fprintf(output, "%-10s\t%5d\t%-5d\n","%",type[3],17); /* 为运算符 "%" */
ch=fgetc(fpin);
continue;
}
}
else
if(ch=='&')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='&')
{
arr[i++]=ch;
arr[i]='\0';
fprintf(output,"%-10s\t%5d\t%-5d\n",arr,type[3],28); /* 为运算符 "&&" */
ch=fgetc(fpin);
continue;
}
arr[i]='\0';
fprintf(output,"%-10s\t%5d\t%-5d\n",arr,type[3],13); /* 为运算符 "&" */
ch=fgetc(fpin);
continue;
}
else
if(ch==';')
{
fprintf(output, "%-10s\t%5d\t%-5d\n",";",type[3],36); /* 为限界符 ";" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='(')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","(",type[3],4); /* 为限界符 "(" */
ch=fgetc(fpin);
continue;
}
else
if(ch==')')
{
fprintf(output, "%-10s\t%5d\t%-5d\n",")",type[3],5); /* 为限界符 ")" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='[')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","[",type[3],6); /* 为限界符 "[" */
ch=fgetc(fpin);
continue;
}
else
if(ch==']')
{
fprintf(output,"%-10s\t%5d\t%-5d\n","]",type[3],7); /* 为限界符 "]" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='{')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","{",type[3],37); /* 为限界符 "{" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='}')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","}",type[3],38); /* 为限界符 "}" */
ch=fgetc(fpin);
continue;
}
else
if(ch=='.')
{
fprintf(output, "%-10s\t%5d\t%-5d\n",".",type[3],9); /* 为限界符 "." */
ch=fgetc(fpin);
continue;
}
else
if(ch==',')
{
fprintf(output, "%-10s\t%5d\t%-5d\n",",",type[3],35); /* 为限界符 "," */
ch=fgetc(fpin);
continue;
}
else
if(ch=='=')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='=')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","==",type[3],26); /* 为运算符 "==" */
ch=fgetc(fpin);
continue;
}
fprintf(output, "%-10s\t%5d\t%-5d\n","=",type[3],30); /* 为运算符 "=" */
continue;
}
else
if(ch=='!')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='=')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","!=",type[3],27); /* 为运算符 "!=" */
ch=fgetc(fpin);
continue;
}
fprintf(output, "%-10s\t%5d\t%-5d\n","!",type[3],10); /* 为运算符 "!" */
continue;
}
else
if(ch=='|')
{
arr[i++]=ch;
ch=fgetc(fpin);
if(ch=='|')
{
fprintf(output, "%-10s\t%5d\t%-5d\n","|",type[3],43); /* 为运算符 "|" */
ch=fgetc(fpin);
continue;
}
fprintf(output, "%-10s\t%5d\t%-5d\n","||",type[3],29); /* 为运算符 "||" */
continue;
}
else
{
fprintf(output,"error in %d lines unknown character %c \n",lineno,ch);
ch=fgetc(fpin);
continue;
}
}
fclose(fpin);
fclose(output);
}
int main()
{
printf("****** you can look up help.txt for reading the result ******\n");
printf("****** the result will be saved in the file you named it ******\n\n\n");
init();
Help();
parse();
}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -