制作一个简单的C语言词法分析程序

1.分析组成

C语言的程序中，有很单词多符号和保留字。一些单词符号还有对应的左线性文法。所以我们需要先做出一个单词字符表，给出对应的识别码，然后跟据对应的表格来写出程序

 

2.程序设计

程序主要有循环判断构成。不需推理即可产生的符号我们可以把它包装在函数中，返回值为对应的识别码即可。但是有线性文法的则需要单独的一遍推倒才可以得出词法分析结果。对于测试样例我们可以存储到txt文件中，使用循环读写可以更高效的测试和输出词法分析结果。最终的结果我们用二元式的形式来表示，存储在txt文件中

3.完整程序

#include 
#include 
#include 

int english(char ch) {
	if((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) return 1;
	else return 0;
}
int number(char ch) {
	if(ch >= '0' && ch <= '9') return 1;
	else return 0;
}
int reserved(char str[]) {
	if(strcmp(str, "void") == 0) return 3;
    else if(strcmp(str, "int") == 0) return 4;
    else if(strcmp(str, "float") == 0) return 5;
    else if(strcmp(str, "double") == 0) return 6;
    else if(strcmp(str, "if") == 0) return 7;
    else if(strcmp(str, "else") == 0) return 8;
    else if(strcmp(str, "for") == 0) return 9;
    else if(strcmp(str, "do") == 0) return 10;
    else if(strcmp(str, "while") == 0) return 11;
	else if(strcmp(str, "break") == 0) return 12;
    else if(strcmp(str, "return") == 0) return 13;
    else return 1;
}
int symbol(char ch) {
	if(ch == ';') return 14;
	else if(ch == ',') return 15;
	else if(ch == '(') return 16;
	else if(ch == ')') return 17;
	else if(ch == '{') return 18;
	else if(ch == '}') return 19;
	else if(ch == '[') return 20;
	else if(ch == ']') return 21;
	else if(ch == '%') return 22;
	else if(ch == '?') return 23;
	else if(ch == ':') return 24;
	else if(ch == '\'') return 25;
	else if(ch == '\"') return 26;
	else if(ch == '.') return 27;
	else return 0;
}

int main(){
	for(int i = 1; i <= 4; i++){
		char txt1[] = "test";
		char num[6];
		sprintf(num, "%d.txt", i);
		strcat(txt1, num);
		char txt2[] = "analyze";
		sprintf(num, "%d.txt", i);
		strcat(txt2, num);
		FILE *fp = fopen(txt1, "r");
		FILE *fw = fopen(txt2, "wt+");
		int flag = 0;

		char ch = fgetc(fp);	
		while(!feof(fp)) {
			char str[32];
			int j = 0;
			if(ch == ' ' || ch == '\t') {
				ch = fgetc(fp);
				continue;
			}
			else if(ch == '\n'){
				fprintf(fw, "\n");
				ch = fgetc(fp);
				continue;
			}
			else if(english(ch)) {
				str[j++] = ch;
				do{
					ch = fgetc(fp);
					str[j++] = ch;
				}while(english(ch)||number(ch));
				str[j-1] = '\0';
				int id = reserved(str);
				fprintf(fw, "(%d, %s) ", id, str);
			}
			else if(number(ch)) {
				str[j++] = ch;
				do{
					ch = fgetc(fp);
					str[j++] = ch;
				}while(number(ch));
				str[j-1] = '\0';
				fprintf(fw, "(2, %s) ", str);
			}
			else if(symbol(ch) != 0) {
				fprintf(fw, "(%d, %c) ", symbol(ch), ch);
				ch = fgetc(fp);
			}
			else if(ch == '>') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(29, >=) ");
					ch = fgetc(fp);
				}
				else if(ch == '>') {
					fprintf(fw, "(30, >>) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(28, >) ");
			}
			else if(ch == '<') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(32, <=) ");
					ch = fgetc(fp);
				}
				else if(ch == '<') {
					fprintf(fw, "(33, <<) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(31, <) ");
			}
			else if(ch == '!') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(35, !=) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(34, !) ");
			}
			else if(ch == '=') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(37, ==) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(36, =) ");
			}
			else if(ch == '/') {
				ch = fgetc(fp);
				if(ch == '*') {
					fprintf(fw, "(Start annotate, /*) ");
					do {
						ch = fgetc(fp);
						if(ch == '*') {
							ch = fgetc(fp);
							if(ch == '/') {
								fprintf(fw, "(End annotate, */) ");
								ch = fgetc(fp);
								break;
							}
						}
					}while(1);
				}
				else if(ch == '/') {
					fprintf(fw, "(annotate, //) ");
					do {
						ch = fgetc(fp);
					}while(ch != '\n');
					fprintf(fw, "\n");
					ch = fgetc(fp);
				}
				else if(ch == '=') {
					fprintf(fw, "(39, /=) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(38, /) ");
			}
			else if(ch == '&') {
				ch = fgetc(fp);
				if(ch == '&') {
					fprintf(fw, "(41, &&) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(40, &) ");
			}
			else if(ch == '|') {
				ch = fgetc(fp);
				if(ch == '|') {
					fprintf(fw, "(43, ||) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(42, |) ");
			}
			else if(ch == '+') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(46, +=) ");
					ch = fgetc(fp);
				}
				else if(ch == '+') {
					fprintf(fw, "(45 ++) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(44, +) ");
			}
			else if(ch == '-') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(49, -=) ");
					ch = fgetc(fp);
				}
				else if(ch == '-') {
					fprintf(fw, "(48, --) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(47, -) ");
			}
			else if(ch == '*') {
				ch = fgetc(fp);
				if(ch == '=') {
					fprintf(fw, "(51, *=) ");
					ch = fgetc(fp);
				}
				else fprintf(fw, "(50, *) ");
			}
			else if(ch == '\\') {
				ch = fgetc(fp);
				if(ch == 'n') {
					ch = fgetc(fp);
				}
			}
			else {
				fprintf(fw, "Undefined symbol!");
				printf("test%d: Undefined symbol!\n", i); 
				flag = 1;
				break;
			}
		}
		fclose(fp);
		fclose(fw);
		if(flag) continue;
		printf("test%d: Finish analyzing.\n", i);
	}
} 

 4.测试运行

其中一个测试样例：（剩余的大家可以自主编写） 

void test1() {
    int a = 5, b = 10;
    if (a > b) {
        printf("a is greater than b");
    } else if (a < b) {
        printf("b is greater than a");
    } else {
        printf("a and b are equal");
    }

    for (int i = 0; i < 5; i++) {
        printf("%d ", i);
    }
    printf("\n");

    int arr[5] = {1, 2, 3, 4, 5};
    int sum = 0;
    for (int i = 0; i < 5; i++) {
        sum += arr[i];
    }
    printf("The sum of the array is: %d\n", sum);

    return 0;
}

 运行结果:

(3, void) (1, test1) (16, () (17, )) (18, {) 
(4, int) (1, a) (36, =) (2, 5) (15, ,) (1, b) (36, =) (2, 10) (14, ;) 
(7, if) (16, () (1, a) (28, >) (1, b) (17, )) (18, {) 
(1, printf) (16, () (26, ") (1, a) (1, is) (1, greater) (1, than) (1, b) (26, ") (17, )) (14, ;) 
(19, }) (8, else) (7, if) (16, () (1, a) (31, <) (1, b) (17, )) (18, {) 
(1, printf) (16, () (26, ") (1, b) (1, is) (1, greater) (1, than) (1, a) (26, ") (17, )) (14, ;) 
(19, }) (8, else) (18, {) 
(1, printf) (16, () (26, ") (1, a) (1, and) (1, b) (1, are) (1, equal) (26, ") (17, )) (14, ;) 
(19, }) 

(9, for) (16, () (4, int) (1, i) (36, =) (2, 0) (14, ;) (1, i) (31, <) (2, 5) (14, ;) (1, i) (45 ++) (17, )) (18, {) 
(1, printf) (16, () (26, ") (22, %) (1, d) (26, ") (15, ,) (1, i) (17, )) (14, ;) 
(19, }) 
(1, printf) (16, () (26, ") (26, ") (17, )) (14, ;) 

(4, int) (1, arr) (20, [) (2, 5) (21, ]) (36, =) (18, {) (2, 1) (15, ,) (2, 2) (15, ,) (2, 3) (15, ,) (2, 4) (15, ,) (2, 5) (19, }) (14, ;) 
(4, int) (1, sum) (36, =) (2, 0) (14, ;) 
(9, for) (16, () (4, int) (1, i) (36, =) (2, 0) (14, ;) (1, i) (31, <) (2, 5) (14, ;) (1, i) (45 ++) (17, )) (18, {) 
(1, sum) (46, +=) (1, arr) (20, [) (1, i) (21, ]) (14, ;) 
(19, }) 
(1, printf) (16, () (26, ") (1, The) (1, sum) (1, of) (1, the) (1, array) (1, is) (24, :) (22, %) (1, d) (26, ") (15, ,) (1, sum) (17, )) (14, ;) 

(13, return) (2, 0) (14, ;) 
(19, })