s-to-json

s2json.py

import json
import re
import sys

def tokenize(s_expr):
    """
    Extracts tokens from an S-expression string.
    Matches:
      1. Strings enclosed in double quotes (handling escaped characters)
      2. Parentheses '(' and ')'
      3. Any other contiguous sequence of non-whitespace, non-parenthesis characters
    """
    token_pattern = r'"(?:\\.|[^"\\])*"|[()]|[^\s()]+'
    return re.findall(token_pattern, s_expr)

def parse(tokens):
    """
    Parses a list of tokens into a nested Python list structure.
    """
    stack = [[]]
    for token in tokens:
        if token == '(':
            new_list = []
            stack[-1].append(new_list)
            stack.append(new_list)
        elif token == ')':
            if len(stack) > 1:
                stack.pop()
            else:
                raise ValueError("Parsing error: Too many closing parentheses.")
        else:
            stack[-1].append(token)
            
    if len(stack) != 1:
        raise ValueError("Parsing error: Missing closing parentheses.")
        
    return stack[0][0] if stack[0] else []

def s_expr_to_json(s_expr, indent=2):
    """
    Converts a raw S-expression string to a JSON string.
    """
    tokens = tokenize(s_expr)
    parsed_data = parse(tokens)
    return json.dumps(parsed_data, indent=indent)

# Main

wat_content = open(sys.argv[1]).read()
json_output = s_expr_to_json(wat_content)
print(json_output)