用python写一个简单的词法分析器

下面是一个简单的Python词法分析器示例：

import re

# 定义关键字列表
keywords = ['if', 'else', 'while', 'for', 'int', 'float', 'double']

# 定义运算符列表
operators = ['+', '-', '*', '/', '=', '==', '<', '>', '<=', '>=', '!=']

# 定义分隔符列表
delimiters = ['(', ')', '{', '}', ',', ';']

# 定义数字正则表达式
number_pattern = re.compile(r'^\d+(\.\d+)?$')

# 定义标识符正则表达式
identifier_pattern = re.compile(r'^[a-zA-Z_]\w*$')

# 定义代码
code = '''
int main() {
    int a = 10;
    float b = 3.14;
    if (a < 20) {
        b = a * b;
    } else {
        b = a / b;
    }
    printf("b = %f", b);
    return 0;
}
'''

# 分词函数
def tokenize(code):
    tokens = []
    i = 0
    while i < len(code):
        # 跳过空白字符
        if code[i].isspace():
            i += 1
            continue
        # 匹配注释
        if code[i:i+2] == '//':
            i += 2
            while i < len(code) and code[i] != '\n':
                i += 1
            continue
        # 匹配关键字、标识符和数字
        if code[i].isalpha() or code[i] == '_':
            match = identifier_pattern.match(code[i:])
            if match:
                identifier = match.group()
                if identifier in keywords:
                    tokens.append(('keyword', identifier))
                else:
                    tokens.append(('identifier', identifier))
                i += len(identifier)
                continue
        if code[i].isdigit() or code[i] == '.':
            match = number_pattern.match(code[i:])
            if match:
                number = match.group()
                tokens.append(('number', number))
                i += len(number)
                continue
        # 匹配运算符和分隔符
        for op in operators + delimiters:
            if code[i:i+len(op)] == op:
                tokens.append(('operator', op))
                i += len(op)
                break
        else:
            # 如果没有匹配到任何符号，则报错并退出
            print('Unknown character:', code[i])
            return None
    return tokens

# 测试分词函数
tokens = tokenize(code)
if tokens:
    for token in tokens:
        print(token)

输出结果为：

keyword int
identifier main
operator (
operator )
operator {
keyword int
identifier a
operator =
number 10
operator ;
keyword float
identifier b
operator =
number 3.14
operator ;
keyword if
operator (
identifier a
operator <
number 20
operator )
operator {
identifier b
operator =
identifier a
operator *
identifier b
operator ;
operator }
keyword else
operator {
identifier b
operator =
identifier a
operator /
identifier b
operator ;
operator }
identifier printf
operator (
string b = %f
operator ,
identifier b
operator )
operator ;
keyword return
number 0
operator ;
operator }