lexer.py

lexer.py

import re
from enum import Enum

class Lexer_state(Enum):
    s = 1
    word = 2
    real = 3
    number = 4
    string = 5

class Lexer:

    def __init__(self):
        self.identify = re.compile("[a-zA-Z0-9_\.-]+")

    def lexer(self,text):
        state = Lexer_state.s
        token = ""

        while text != "":
            if state == Lexer_state.s:
                if text[0] in (' ', '\t', '\n'):
                    text = text[1:]
                elif text[0] == '"':
                    text = text[1:]
                    state = Lexer_state.string
                elif text[0].isdigit():
                    token += text[0]
                    text = text[1:]
                    state = Lexer_state.number
                elif self.identify.match(text[0]):
                    state = Lexer_state.word
                    token += text[0]
                    text = text[1:]
                else:
                    print("[Error]Not valid charatect")
                    break

            elif state == Lexer_state.string:
                if self.identify.match(text[0]):
                    token += text[0]
                    text = text[1:]
                elif text[0] == '"':
                    text = text[1:]
                    yield token
                    token = ""
                    state = Lexer_state.s
                else:
                    break

            elif state == Lexer_state.number:
                if text[0].isdigit():
                    print(text[0])
                    token += text[0]
                    text = text[1:]
                elif text[0] == '.':
                    token += text[0]
                    text = text[1:]
                    state = Lexer_state.real
                else:
                    state = Lexer_state.s
                    yield int(token)
                    token = ""

            elif state == Lexer_state.real:
                if text[0].isdigit():
                    token += text[0]
                    text = text[1:]
                else:
                    yield float(token)
                    token = ""

            elif state == Lexer_state.word:
                if self.identify.match(text[0]):
                    token += text[0]
                    text = text[1:]
                else:
                    state = Lexer_state.s
                    token = ""
                    yield token

    def evaluate(self):

        def operation(stack,op):
            a = stack.pop()
            b = stack.pop()
            stack.append(op(a,b))

        def performBinaryOp(op):
                def perform(stack):
                       a = stack.pop()
                       b = stack.pop()
                       stack.append(op(a, b))
                return perform

        stack = []
        wordList = {
                    "PRINT": lambda x: print(x),
                    "DUP": lambda x: x.append(x[-1]),
                    "+":performBinaryOp(lambda x,y: x + y),
                    "/": [operation, lambda x,y: x // y],
                    "-": [operation, lambda x,z: x - y],
                    #"slovo": []
                }
        while True:
            for token in self.lexer(input('>')):
                print(token)
                try:
                    if isinstance(token, str):
                        wordList[token](stack)
                    elif isinstance(token, int):
                        stack.append(token)
                except KeyError:
                    pass

if __name__ == "__main__":
    l = Lexer()
    l.evaluate()