c语言实现 PASCAL 子集的词法分析程序

以下是用 C 语言实现 PASCAL 子集的词法分析程序的示例代码：

#include <stdio.h>
#include <ctype.h>
#include <string.h>

#define MAX_LENGTH 100

char input[MAX_LENGTH];
int pos = 0;

typedef enum {
    NONE,
    ID,
    NUMBER,
    PLUS,
    MINUS,
    MUL,
    DIV,
    LPAREN,
    RPAREN,
    SEMICOLON,
    ASSIGN,
    END
} TokenType;

typedef struct {
    TokenType type;
    char value[MAX_LENGTH];
} Token;

void advance() {
    pos++;
}

char peek() {
    return input[pos];
}

Token get_next_token() {
    Token token;
    token.type = NONE;
    memset(token.value, 0, sizeof(token.value));

    char c = peek();
    while (c != '\0') {
        if (isspace(c)) {
            advance();
            c = peek();
            continue;
        }

        if (isalpha(c)) {
            int i = 0;
            while (isalnum(c)) {
                token.value[i] = c;
                i++;
                advance();
                c = peek();
            }
            token.type = ID;
            return token;
        }

        if (isdigit(c)) {
            int i = 0;
            while (isdigit(c)) {
                token.value[i] = c;
                i++;
                advance();
                c = peek();
            }
            token.type = NUMBER;
            return token;
        }

        if (c == '+') {
            token.type = PLUS;
            advance();
            return token;
        }
        
        if (c == '-') {
            token.type = MINUS;
            advance();
            return token;
        }

        if (c == '*') {
            token.type = MUL;
            advance();
            return token;
        }

        if (c == '/') {
            token.type = DIV;
            advance();
            return token;
        }

        if (c == '(') {
            token.type = LPAREN;
            advance();
            return token;
        }

        if (c == ')') {
            token.type = RPAREN;
            advance();
            return token;
        }

        if (c == ';') {
            token.type = SEMICOLON;
            advance();
            return token;
        }

        if (c == ':') {
            advance();
            c = peek();
            if (c == '=') {
                token.type = ASSIGN;
                token.value[0] = ':';
                token.value[1] = '=';
                advance();
                return token;
            }
        }

        printf("Invalid character: %c\n", c);
        token.type = END;
        return token;
    }

    token.type = END;
    return token;
}

int main() {
    printf("Enter Pascal code:\n");
    fgets(input, MAX_LENGTH, stdin);

    Token token = get_next_token();
    while (token.type != END) {
        printf("Token: %s\n", token.value);
        token = get_next_token();
    }

    return 0;
}

该程序实现了对 Pascal 子集中的 ID、数字、运算符、括号、分号和赋值符号进行识别。在 `main` 函数中，用户可以输入 Pascal 代码，并且使用 `get_next_token` 函数来逐个获取每个词法单元的类型和值。