minion/lexer.py

#!/usr/bin/env python3
import json
import sys

def find(needle: str, haystack: str) -> bool:
    return haystack.find(needle) != -1

source: list  = list()
program = './prog.mn'
if len(sys.argv) > 1:
    program = sys.argv[1]

with open(program, 'r') as prog:
    source = prog.readlines()

num_lines = len(source)
tokens = list()
for lineno in range(1, num_lines + 1):
    line = source[lineno - 1].strip('\n')
    split_on_colon = line.split(':')
    colon_lhs = split_on_colon[0]
    split_on_equals = line.split('=')
    equals_lhs = split_on_equals[0] 
    if find(':', line):
        colon_rhs = ':'.join(split_on_colon[1:])
        colon_rhs_split_on_equals = colon_rhs.split('=')
        colon_equals_lhs = colon_rhs_split_on_equals[0]
        type = {"variant": "type", "value": colon_lhs.strip(), "line_number": lineno}
        tokens.append(type)
        operator = {"variant": "operator", "value": ':', "line_number": lineno}
        tokens.append(operator)
        signifier = {"variant": "signifier", "value": colon_equals_lhs.strip(), "line_number": lineno}
        if colon_lhs.strip() == 'func':
            split_on_lparen = colon_equals_lhs.strip().split('(')
            name = split_on_lparen[0]
            raw_params = list(map(str.strip,'('.join(split_on_lparen[1:]).split(',')))
            params = list()
            for raw in raw_params:
                raw_split_on_colon = raw.split(':')
                raw_split_on_rparen = raw_split_on_colon[1].split(')')
                ptype = raw_split_on_colon[0].strip()
                pname = raw_split_on_rparen[0].strip()
                params.append({"type": ptype, "name": pname})
                signifier = {"variant": "function_declaration", "value": name, "params": params, "line_number": lineno}
        tokens.append(signifier)
        if find('=', line):
            colon_equals_rhs = '='.join(colon_rhs_split_on_equals[1:])
            operator = {"variant": "operator", "value": '=', "line_number": lineno}
            tokens.append(operator)
            expr = colon_equals_rhs.strip()
            expression = {"variant": "expression", "value": expr, "line_number": lineno}
            if expr == '{':
                expression = {"variant": "start_block", "value": expr, "line_number": lineno}
            tokens.append(expression)
    elif find('=', line):
        equals_rhs = '='.join(split_on_equals[1:])
        signifier = {"variant": "signifier", "value": equals_lhs.strip(), "line_number": lineno}
        tokens.append(signifier)
        operator = {"variant": "operator", "value": '=', "line_number": lineno}
        tokens.append(operator)
        expr = equals_rhs.strip()
        expression = {"variant": "expression", "value": expr, "line_number": lineno}
        if expr == '{':
            expression = {"variant": "start_block", "value": expr, "line_number": lineno}
        tokens.append(expression)
    elif find('}',line):
        expression = {"variant": "end_block", "value": '}', "line_number": lineno}
        tokens.append(expression)
    elif find('(', line):
        split_on_lparen = line.split('(')
        name = split_on_lparen[0].strip()
        without_name = '('.join(split_on_lparen[1:])
        split_on_rparen = without_name.split(')')
        params_str = ')'.join(split_on_rparen[:-1])
        params = list(map(str.strip,params_str.split(',')))
        if name not in ['if', 'while']:
            function_call = {"variant": "function_call", "value": name, "params": params, "line_number": lineno}
            tokens.append(function_call)
        elif name == "if":
            conditional = {"variant": "conditional", "value": name, "condition": ','.join(params), "line_number": lineno}
            tokens.append(conditional)
            expression = {"variant": "start_block", "value": "{", "line_number": lineno}
            tokens.append(expression)
        else:
            loop = {"variant": "loop", "value": name, "condition": ','.join(params),  "line_number": lineno}
            tokens.append(loop)
            expression = {"variant": "start_block", "value": "{", "line_number": lineno}
            tokens.append(expression)
    else:
        if line.strip() == '{':
            expression = {"variant": "start_block", "value": '{', "line_number": lineno}
            tokens.append(expression)
        elif line != "":
            error = {"variant": "syntax_error", "line": line, "line_number": lineno}
            tokens.append(error)
print(json.dumps(tokens))