bencode decoder

bdecode.py

import sys
import json

def bdecode(bencoded_bytes):
    def decode_int(b, i):
        j = b.find(b'e', i)
        return int(b[i:j]), j + 1
        
    def decode_bytes(b, i):
        colon = b.find(b':', i)
        length = int(b[i:colon])
        start = colon + 1
        return b[start:start+length], start+length
        
    def decode_list(b, i):
        items = []
        i += 1
        while b[i:i+1] != b'e':
            value, i = decode_element(b, i)
            items.append(value)
        return items, i + 1

    def decode_dict(b, i):
        items = {}
        i += 1
        while b[i:i+1] != b'e':
            key, i = decode_bytes(b, i)
            value, i = decode_element(b, i)
            items[key] = value
        return items, i + 1

    def decode_element(b, i):
        if b[i:i+1] == b'i':
            return decode_int(b, i+1)
        elif b[i:i+1] == b'l':
            return decode_list(b, i)
        elif b[i:i+1] == b'd':
            return decode_dict(b, i)
        elif b[i:i+1].isdigit():
            return decode_bytes(b, i)
        else:
            raise ValueError("Invalid bencode format")
            
    decoded_data, _ = decode_element(bencoded_bytes, 0)
    return decoded_data

def convert_bytes_to_str(data):
    """Recursively converts bytes keys and values in a dict/list to UTF-8 strings."""
    if isinstance(data, bytes):
        try:
            # Assume UTF-8 encoding for standard torrent metadata (filenames, etc.)
            return data.decode('utf-8')
        except UnicodeDecodeError:
            # Handle non-UTF-8 data gracefully (maybe represent as hex or just skip)
            return data.hex() 
    elif isinstance(data, list):
        return [convert_bytes_to_str(item) for item in data]
    elif isinstance(data, dict):
        new_dict = {}
        for key, value in data.items():
            # Convert both the key (which is currently a bytes object) 
            # and the value recursively
            new_key = convert_bytes_to_str(key)
            new_value = convert_bytes_to_str(value)
            new_dict[new_key] = new_value
        return new_dict
    else:
        # Handle integers or other types
        return data

raw_bencode_response = sys.stdin.buffer.read()

decoded_info = bdecode(raw_bencode_response)

# Convert all bytes to strings recursively
decoded_info_str = convert_bytes_to_str(decoded_info)

# Safely remove the "pieces" key now that keys are strings
if "pieces" in decoded_info_str:
    del decoded_info_str["pieces"]

# Dump the fully stringified dictionary to stdout as JSON
print(json.dumps(decoded_info_str, indent=2))