little-things/bencode2json

#!/usr/bin/env python3
try:
	from contextlib import nullcontext
except ImportError:
	# Python 3.6 and below
	# One-liner because I can.
	nullcontext = type("nc", (object,), {"__init__": (lambda self, v: setattr(self, "v", v)), "__enter__": (lambda self: self.v), "__exit__": (lambda self, *args, **kwargs: None)})
import json
import sys


def parse_dict(f):
	d = {}
	while True:
		c = f.read(1)
		if c == b'e':
			return d
		elif b'0' <= c <= b'9':
			key = parse_str(f, c)
			value = parse_any(f)
			d[key] = value
		else:
			raise ValueError(f'invalid bencode value: {c} instead of expected e or digit')


def parse_list(f):
	l = []
	while True:
		c = f.read(1)
		if c == b'e':
			return l
		l.append(parse_any(f, c))


def _parse_int_bytes(f, stopChar):
	# Read until encountering a non-decimal character; if it isn't stopChar, raise a ValueError
	v = b''
	# Python 3.8+ (walrus)
	#while b'0' <= (c := f.read(1)) <= b'9':
	#	v += c
	while True:
		c = f.read(1)
		if b'0' <= c <= b'9':
			v += c
		else:
			break
	if c == stopChar:
		return v
	raise ValueError(f'invalid bencode value: {c} instead of expected {stopChar}')


def parse_int(f):
	return int(_parse_int_bytes(f, b'e'))


def parse_str(f, c):
	v = _parse_int_bytes(f, b':')
	length = int(c + v)
	read = 0
	buf = []
	while read < length:
		d = f.read(length - read)
		buf.append(d)
		read += len(d)
	buf = b''.join(buf)
	try:
		return buf.decode('utf-8', 'strict')
	except UnicodeDecodeError:
		return buf.decode('iso-8859-1')


def parse_any(f, c = None):
	if c is None:
		c = f.read(1)
	if c == b'd':
		return parse_dict(f)
	elif c == b'l':
		return parse_list(f)
	elif c == b'i':
		return parse_int(f)
	elif b'0' <= c <= b'9':
		return parse_str(f, c)
	else:
		raise ValueError(f'invalid bencode value: {c} instead of expected d, l, i, or digit')


if sys.argv[1:]:
	c = open(sys.argv[1], 'rb')
else:
	c = nullcontext(sys.stdin.buffer)
with c as fp:
	print(json.dumps(parse_any(fp)))