talos.py

import re
import sys

def talos_lex_source_lines(source):
    return re.split("\n",source)

def talos_lex_source_tokens(source_line):
    return re.split("\s",source_line)
 
def convert_to_c(tokens):
    c_source_line = ""
    for token in tokens:
        if token == "int":
            c_source_line= c_source_line + token + " "
            continue
        if re.match(".*:",token):
            c_source_line= c_source_line + "(" + " "
            continue
        if re.match(".*[^:]",token):
            c_source_line= c_source_line + token + " "
            continue
    c_source_line = c_source_line + "\n"
        
    return c_source_line
  

if __name__ == '__main__':
    filename = sys.argv[1]
    file = open(filename)
    source = file.read()
    source_lines = talos_lex_source_lines(source)
    print("\nBeginning parsing source file: {}".format(filename))
    print("....................................")
  
    source_line_number = 0
    for source_line in source_lines:
        source_line_number = source_line_number + 1
    
        if source_line is not "":
            tokens = talos_lex_source_tokens(source_line)
            print("{} : {}".format(source_line_number,source_line))
            print("line {} has tokens: {}".format(source_line_number,str(tokens)))
            print("converted to c source code: {}".format(convert_to_c(tokens)))