a basic interpreter. [Experimental]

README.md

Syntax:

Comments

# Comments

Variable declaration

var1 = "hello"

Accessing variable

print var1

Updating variable value

var1 = "Hello World"

Print statement

print "wth..!"

Swapping values

v1 = 1
v2 = 2
v1 = v2

interpreter.py

import colorama, re, argparse, sys

class PrintToken:
    ValueType: str
    Value: str

class VariableToken:
    VariableName: str
    ValueType: str
    Value: any

class Lexer:
    def __init__(self, code: str) -> None:
        self.code: str = code
        self.tokens: list = []
        self.chunk: str = ""
        self._lexer()
    
    def _lexer(self) -> None:
        i: int = 0

        while i < len(self.code):
            if self.code[i] == "\n":
                self.chunk = ""
                i += 1
                continue
            
            if self.code[i] == " ":
                i += 1
                continue

            if self.code[i] == "#":
                while i < len(self.code) and self.code[i] != "\n":
                    i += 1
                continue

            if self.chunk == "print" and (self.code[i-1] in (" ", "\n") or self.code[i] == "\""):
                while i < len(self.code) and self.code[i] in (" ", "\n"):
                    i += 1

                obj = PrintToken()

                if i < len(self.code) and self.code[i] == "\"":
                    i += 1
                    self.chunk = ""

                    while i < len(self.code) and self.code[i] != "\"":
                        self.chunk += self.code[i]
                        i += 1
                    
                    if i < len(self.code) and self.code[i] == "\"":
                        i += 1

                    obj.ValueType = str
                    obj.Value = self.chunk
                    self.tokens.append(obj)
                    self.chunk = ""
                    continue
                
                self.chunk = ""

                while i < len(self.code) and self.code[i] not in (" ", "\n"):
                    self.chunk += self.code[i]
                    i += 1

                if exists := [t for t in self.tokens if type(t) == VariableToken and t.VariableName == self.chunk]:
                    obj.ValueType = VariableToken
                    obj.Value = exists[0].Value

                elif _ := bool(re.match(r"^-?\d+$", self.chunk)):
                    obj.ValueType = int
                    obj.Value = int(self.chunk)

                elif _ := bool(re.match(r"-?(?:\d+\.\d+|\.\d+)", self.chunk)):
                    obj.ValueType = float
                    obj.Value = float(self.chunk)

                else:
                    obj.ValueType = NameError
                    obj.Value = f"'{self.chunk}' not defined"

                self.tokens.append(obj)
                self.chunk = ""
                continue

            if self.code[i] == "=":
                i += 1

                while i < len(self.code) and self.code[i] in (" ", "\n"):
                    i += 1
                
                obj = VariableToken()

                while len(self.chunk) and self.chunk[-1] == " ":
                    self.chunk = self.chunk[:-1]

                if exists := [t for t in self.tokens if type(t) == VariableToken and t.VariableName == self.chunk]:
                    obj = exists[0]
                else:
                    obj.VariableName = self.chunk

                while i < len(self.code) and self.code[i] in (" ", "\n"):
                    i += 1

                if i < len(self.code) and self.code[i] == "\"":
                    obj.ValueType = str
                    i += 1
                    val = ""
                    while i < len(self.code) and self.code[i] != "\"":
                        val += self.code[i]
                        i += 1

                    if i < len(self.code) and self.code[i] == "\"":
                        i += 1
                    
                    obj.Value = val
                    self.tokens.append(obj)
                    self.chunk = ""
                    continue

                self.chunk = ""

                while i < len(self.code) and self.code[i] not in (" ", "\n"):
                    self.chunk += self.code[i]
                    i += 1

                if _ := bool(re.match(r"^-?\d+$", self.chunk)):
                    obj.ValueType = int
                    obj.Value = int(self.chunk)
                    self.tokens.append(obj)
                    self.chunk = ""
                    continue

                if _ := bool(re.match(r"-?(?:\d+\.\d+|\.\d+)", self.chunk)):
                    obj.ValueType = float
                    obj.Value = float(self.chunk)
                    self.tokens.append(obj)
                    self.chunk = ""
                    continue
                
                if exists := [t for t in self.tokens if type(t) == VariableToken and t.VariableName == self.chunk]:
                    obj.ValueType = exists[0].ValueType
                    obj.Value = exists[0].Value
                    self.tokens.append(obj)
                    self.chunk = ""
                    continue
                
                obj.ValueType = NameError
                obj.Value = f"'{self.chunk}' not defined"
                self.tokens.append(obj)
                self.chunk = ""
                continue

            self.chunk += self.code[i]
            i += 1

def main() -> None:
    try:
        args = argparse.ArgumentParser()
        args.add_argument("source_file", help="Input source file")
        args = args.parse_args()

        with open(args.source_file, "r", encoding="utf-8") as source_file:
            code: str = source_file.read()

        lexer = Lexer(code)
        
        i: int = 0

        while i < len(lexer.tokens):
            token = lexer.tokens[i]
        
            if type(token) == PrintToken:
                if token.ValueType in (str, int, float):
                    print(token.Value)

                elif token.ValueType == VariableToken:
                    print(token.Value)

                elif token.ValueType == NameError:
                    print(colorama.Fore.RED, token.Value, colorama.Style.RESET_ALL, sep="")

            i += 1

    except FileNotFoundError:
        print(colorama.Fore.RED, "Error: Input file doesn't exists", colorama.Style.RESET_ALL, sep="")

if __name__ == "__main__":
    main()