C语言实现词法分析器

这是笔者按照课本中的思路实现的词法分析程序，具体细节和意义请参考
《编译技术》张莉等著.–北京：高等教育出版社,2016.9.ISBN: 978-7-04-046317-0
注：本程序使用文件读入的方法进行字符的读取，每次识别一个词，返回值一个词的类型，类型码在注释中给出。
代码如下：

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/*
#define BEGINSY 1   //begin
#define ENDSY 2     //end
#define IFSY 3      //if
#define THENSY 4    //then
#define ELSE 5      //else
#define IDSY 20     //标识符
#define INTSY 21    //整常数
#define PLUSSY 22  //+
#define MINUSSY 23  //-
#define STARSY 24   /
#define DIVISY 25   // /
#define LPARSY 26   //(
#define RPARSY 27   //)
#define COMMASY 28  //,
#define SEMISY 29   //;
#define COLONSY 30  //:
#define ASSIGNSY 31 //:=
#define EQUSY 32    //=
*/
char CHAR ;
char TOKEN[50] ;
enum SYMBOL{BEGINSY = 1 , ENDSY = 2, IFSY = 3, THENSY = 4, ELSE = 5, IDSY = 20,
            INTSY = 21, PLUSSY = 22, MINUSSY = 23, STARSY = 24, DIVISY = 25,
            LPARSY = 26, RPARSY = 27, COMMASY = 28, SEMISY = 29, COLONSY = 30,
            ASSIGNSY = 31, EQUSY = 32} SYMBOL;

int NUM ;

int topow(int x){
    int i ;
    int result = 1 ;
    for(i = 0 ; i < x ; i++){
        result *= 10 ;
    }
    return result ;
}
void clearToken(){
    int i = 0 ;
    while(i < 50){
        TOKEN[i] = '\0' ;
        i++ ;
    }
}
void retract(FILE* f){
    if(!feof(f)){
        fseek(f, -1, SEEK_CUR) ;
    }
}
void catToken(){
    int i = 0 ;
    while(TOKEN[i] != '\0'){
        i++ ;
    }
    TOKEN[i] = CHAR ;
}
int transNum(){
    int num = 0 ;
    int i = 0 ;
    int result = 0 ;
    while(TOKEN[i] != '\0'){
        i++ ;
    }
    num = i - 1 ;
    i-- ;
    while(i >= 0){
        result += (TOKEN[i] - '0') * topow(num - i) ;
        i-- ;
    }
    return result ;
}
int compare(char des[], char token[]){//字符串比较，区分大小写
    int i ;
    if(strlen(des) != strlen(token)){
        return 0 ;
    }
    for(i = 0 ; token[i] != '\0' ; i++){
        if(des[i] != token[i]){
            return 0 ;
        }
    }
    return 1 ;
}
int reserver(){
    char a[] = {'B','E','G','I','N','\0'} ;
    char b[] = {'E','N','D','\0'} ;
    char c[] = {'I','F','\0'} ;
    char d[] = {'T','H','E','N','\0'} ;
    char e[] = {'E','L','S','E','\0'} ;
    if(compare(a, TOKEN) == 1){
        return BEGINSY ;
    }
    else if(compare(b, TOKEN) == 1){
        return ENDSY ;
    }
    else if(compare(c, TOKEN) == 1){
        return IFSY ;
    }
    else if(compare(d, TOKEN) == 1){
        return THENSY ;
    }
    else if(compare(e, TOKEN) == 1){
        return ELSE ;
    }
    else{
        return 0 ;
    }
}

int isSpace(){
    return CHAR == ' ' ? 1 : 0 ;
}
int isNewline(){
    return CHAR == '\n' ? 1 : 0 ;
}
int isTab(){
    return CHAR == '\t' ? 1 : 0 ;
}
int isLetter(){
    return (CHAR >= 'a' && CHAR <= 'z') ||
           (CHAR >= 'A' && CHAR <= 'Z') ? 1 : 0 ;
}
int isDigit(){
    return CHAR <= '9' && CHAR >= '0' ? 1 : 0 ;
}
int isColon(){
    return CHAR == ':' ? 1 : 0 ;
}
int isPlus(){
    return CHAR == '+' ? 1 : 0 ;
}
int isMinus(){
    return CHAR == '-' ? 1 : 0 ;
}
int isStar(){
    return CHAR == '*' ? 1 : 0 ;
}
int isLpar(){
    return CHAR == '(' ? 1 : 0 ;
}
int isRpar(){
    return CHAR == ')' ? 1 : 0 ;
}
int isComma(){
    return CHAR == ',' ? 1 : 0 ;
}
int isSemi(){
    return CHAR == ';' ? 1 : 0 ;
}
int isDivi(){
    return CHAR == '/' ? 1 : 0 ;
}
int isEqu(){
    return CHAR == '=' ? 1 : 0 ;
}
void error(){
    printf("NOT MATCHED!\n") ;
}

int getsym(FILE* f){
    clearToken() ;
    while(isSpace() || isNewline() || isTab()){
        CHAR = fgetc(f) ;
        if(CHAR == EOF){
            break ;
        }
    }
    if(isLetter()){
        while(isLetter() || isDigit()){
            catToken() ;
            CHAR = fgetc(f) ;
        }
        retract(f) ;
        int resultValue = reserver() ;
        if(resultValue == 0){
            SYMBOL = IDSY ;
        }
        else{
            SYMBOL = resultValue ;
        }
    }
    else if(isDigit()){
        while(isDigit()){
            catToken() ;
            CHAR = fgetc(f) ;
        }
        retract(f) ;
        NUM = transNum() ;
        SYMBOL = INTSY ;
    }
    else if(isColon()){
        CHAR = fgetc(f) ;
        if(isEqu()){
            SYMBOL = ASSIGNSY ;
        }
        else{
            retract(f) ;
        }
        SYMBOL = COLONSY ;
    }
    else if(isPlus()){
        SYMBOL = PLUSSY ;
    }
    else if(isMinus()){
        SYMBOL = MINUSSY ;
    }
    else if(isStar()){
        SYMBOL = STARSY ;
    }
    else if(isLpar()){
        SYMBOL = LPARSY ;
    }
    else if(isRpar()){
        SYMBOL = RPARSY ;
    }
    else if(isComma()){
        SYMBOL = COMMASY ;
    }
    else if(isSemi()){
        SYMBOL = SEMISY ;
    }
    else if(isDivi()){
        CHAR = getchar() ;
        if(isStar()){
            do{
                do{
                    CHAR = fgetc(f) ;
                }while(!isStar()) ;
                do{
                    CHAR = fgetc(f) ;
                    if(isDivi()){
                        return 0 ;
                    }
                }while(isStar()) ;
            }while(!isStar()) ;
        }
        else{
            retract(f) ;
            SYMBOL = DIVISY ;
        }
    }
    else{
        error() ;
        return 0 ;
    }
    return SYMBOL ;
}




int main()
{
    FILE* f = fopen("code.txt", "r") ;
    if(!f){
        printf("No such file !\n") ;
        return 0 ;
    }
    while(!feof(f)){
        CHAR = fgetc(f) ;
        if(feof(f)){
            break ;
        }
        printf("%d\n", getsym(f)) ;
    }
    fclose(f) ;

    return 0;
}

对于样例：
code.txt

BEGIN begin 2333 hhhh  +  - __

测试结果如下：