Is json.loads() vulnerable to arbitrary code execution?

Solution 1:

Note that the below answer is relating to the default Python3.4 installation for Windows 10 64-bit. Also note that this answer only looks at the py scanner, not the c scanner.

For the source files see https://hg.python.org/cpython/file/tip/Lib/json or find them in your local python installation.

Research

See the reference implementation at the bottom of this post alongside this research

The parsing functions called by json.loads(s) are defined in \Lib\json\scanner.py:

parse_object = context.parse_object
parse_array = context.parse_array
parse_string = context.parse_string
parse_float = context.parse_float
parse_int = context.parse_int
parse_constant = context.parse_constant

with context being an instance of the JSONDecoder class which is defined in \Lib\json\decoder.py and uses the following parsers:

self.parse_float = parse_float or float
self.parse_int = parse_int or int
self.parse_constant = parse_constant or _CONSTANTS.__getitem__
self.parse_string = scanstring
self.parse_object = JSONObject
self.parse_array = JSONArray

From here we can look at each individual parser to determine whether or not it is susceptible to arbitrary code execution:


parse_float

This uses the default float function and so is safe.


parse_int

This uses the default int function and so is safe.


parse_constant

_CONSTANTS is defined within the same file as:

_CONSTANTS = {
    '-Infinity': NegInf,
    'Infinity': PosInf,
    'NaN': NaN,
}

and so a simple lookup is being performed, and so it is safe.


parse_string, JSONObject, JSONArray

As can be seen by looking at the implementations at the end of this post, the only external code that could be executed is:

From JSONObject:

  • object_pairs_hook
  • object_hook

From JSONArray:

  • scan_once


object_pairs_hook, object_hook

By default object_pairs_hook and object_hook are defined as None from the decoder initializer:

def __init__(self, object_hook=None, parse_float=None,
        parse_int=None, parse_constant=None, strict=True,
        object_pairs_hook=None)


scan_once

scan_once is defined as:

self.scan_once = scanner.make_scanner(self)

The source for which can be found in \Lib\json\scanner.py, from which we can see that it scan_once simply calls the appropriate parser for each part of the JSON object.


Conclusion

From the above and the reference implementation it can be seen that as long as the scanner used by the JSON decoder is the default, arbitrary code will not be executed, it is probably possible by using a custom decoder through the use of its __init__ parameters to instead make it execute arbitrary code, but save that I don't think so.


Implementation

BACKSLASH

BACKSLASH = {
    '"': '"', '\\': '\\', '/': '/',
    'b': '\b', 'f': '\f', 'n': '\n', 'r': '\r', 't': '\t',
}

STRINGCHUNK

STRINGCHUNK = re.compile(r'(.*?)(["\\\x00-\x1f])', FLAGS)

scanstring

def py_scanstring(s, end, strict=True,
        _b=BACKSLASH, _m=STRINGCHUNK.match):
    """Scan the string s for a JSON string. End is the index of the
    character in s after the quote that started the JSON string.
    Unescapes all valid JSON string escape sequences and raises ValueError
    on attempt to decode an invalid string. If strict is False then literal
    control characters are allowed in the string.

    Returns a tuple of the decoded string and the index of the character in s
    after the end quote."""
    chunks = []
    _append = chunks.append
    begin = end - 1
    while 1:
        chunk = _m(s, end)
        if chunk is None:
            raise ValueError(
                errmsg("Unterminated string starting at", s, begin))
        end = chunk.end()
        content, terminator = chunk.groups()
        # Content is contains zero or more unescaped string characters
        if content:
            _append(content)
        # Terminator is the end of string, a literal control character,
        # or a backslash denoting that an escape sequence follows
        if terminator == '"':
            break
        elif terminator != '\\':
            if strict:
                #msg = "Invalid control character %r at" % (terminator,)
                msg = "Invalid control character {0!r} at".format(terminator)
                raise ValueError(errmsg(msg, s, end))
            else:
                _append(terminator)
                continue
        try:
            esc = s[end]
        except IndexError:
            raise ValueError(
                errmsg("Unterminated string starting at", s, begin))
        # If not a unicode escape sequence, must be in the lookup table
        if esc != 'u':
            try:
                char = _b[esc]
            except KeyError:
                msg = "Invalid \\escape: {0!r}".format(esc)
                raise ValueError(errmsg(msg, s, end))
            end += 1
        else:
            uni = _decode_uXXXX(s, end)
            end += 5
            if 0xd800 <= uni <= 0xdbff and s[end:end + 2] == '\\u':
                uni2 = _decode_uXXXX(s, end + 1)
                if 0xdc00 <= uni2 <= 0xdfff:
                    uni = 0x10000 + (((uni - 0xd800) << 10) | (uni2 - 0xdc00))
                    end += 6
            char = chr(uni)
        _append(char)
    return ''.join(chunks), end

scanstring = c_scanstring or py_scanstring

WHITESPACE

WHITESPACE = re.compile(r'[ \t\n\r]*', FLAGS)

WHITESPACE_STR

WHITESPACE_STR = ' \t\n\r'

JSONObject

def JSONObject(s_and_end, strict, scan_once, object_hook, object_pairs_hook,
               memo=None, _w=WHITESPACE.match, _ws=WHITESPACE_STR):
    s, end = s_and_end
    pairs = []
    pairs_append = pairs.append
    # Backwards compatibility
    if memo is None:
        memo = {}
    memo_get = memo.setdefault
    # Use a slice to prevent IndexError from being raised, the following
    # check will raise a more specific ValueError if the string is empty
    nextchar = s[end:end + 1]
    # Normally we expect nextchar == '"'
    if nextchar != '"':
        if nextchar in _ws:
            end = _w(s, end).end()
            nextchar = s[end:end + 1]
        # Trivial empty object
        if nextchar == '}':
            if object_pairs_hook is not None:
                result = object_pairs_hook(pairs)
                return result, end + 1
            pairs = {}
            if object_hook is not None:
                pairs = object_hook(pairs)
            return pairs, end + 1
        elif nextchar != '"':
            raise ValueError(errmsg(
                "Expecting property name enclosed in double quotes", s, end))
    end += 1
    while True:
        key, end = scanstring(s, end, strict)
        key = memo_get(key, key)
        # To skip some function call overhead we optimize the fast paths where
        # the JSON key separator is ": " or just ":".
        if s[end:end + 1] != ':':
            end = _w(s, end).end()
            if s[end:end + 1] != ':':
                raise ValueError(errmsg("Expecting ':' delimiter", s, end))
        end += 1

        try:
            if s[end] in _ws:
                end += 1
                if s[end] in _ws:
                    end = _w(s, end + 1).end()
        except IndexError:
            pass

        try:
            value, end = scan_once(s, end)
        except StopIteration as err:
            raise ValueError(errmsg("Expecting value", s, err.value)) from None
        pairs_append((key, value))
        try:
            nextchar = s[end]
            if nextchar in _ws:
                end = _w(s, end + 1).end()
                nextchar = s[end]
        except IndexError:
            nextchar = ''
        end += 1

        if nextchar == '}':
            break
        elif nextchar != ',':
            raise ValueError(errmsg("Expecting ',' delimiter", s, end - 1))
        end = _w(s, end).end()
        nextchar = s[end:end + 1]
        end += 1
        if nextchar != '"':
            raise ValueError(errmsg(
                "Expecting property name enclosed in double quotes", s, end - 1))
    if object_pairs_hook is not None:
        result = object_pairs_hook(pairs)
        return result, end
    pairs = dict(pairs)
    if object_hook is not None:
        pairs = object_hook(pairs)
    return pairs, end

JSONArray

def JSONArray(s_and_end, scan_once, _w=WHITESPACE.match, _ws=WHITESPACE_STR):
    s, end = s_and_end
    values = []
    nextchar = s[end:end + 1]
    if nextchar in _ws:
        end = _w(s, end + 1).end()
        nextchar = s[end:end + 1]
    # Look-ahead for trivial empty array
    if nextchar == ']':
        return values, end + 1
    _append = values.append
    while True:
        try:
            value, end = scan_once(s, end)
        except StopIteration as err:
            raise ValueError(errmsg("Expecting value", s, err.value)) from None
        _append(value)
        nextchar = s[end:end + 1]
        if nextchar in _ws:
            end = _w(s, end + 1).end()
            nextchar = s[end:end + 1]
        end += 1
        if nextchar == ']':
            break
        elif nextchar != ',':
            raise ValueError(errmsg("Expecting ',' delimiter", s, end - 1))
        try:
            if s[end] in _ws:
                end += 1
                if s[end] in _ws:
                    end = _w(s, end + 1).end()
        except IndexError:
            pass

    return values, end

scanner.make_scanner

def py_make_scanner(context):
    parse_object = context.parse_object
    parse_array = context.parse_array
    parse_string = context.parse_string
    match_number = NUMBER_RE.match
    strict = context.strict
    parse_float = context.parse_float
    parse_int = context.parse_int
    parse_constant = context.parse_constant
    object_hook = context.object_hook
    object_pairs_hook = context.object_pairs_hook
    memo = context.memo

    def _scan_once(string, idx):
        try:
            nextchar = string[idx]
        except IndexError:
            raise StopIteration(idx)

        if nextchar == '"':
            return parse_string(string, idx + 1, strict)
        elif nextchar == '{':
            return parse_object((string, idx + 1), strict,
                _scan_once, object_hook, object_pairs_hook, memo)
        elif nextchar == '[':
            return parse_array((string, idx + 1), _scan_once)
        elif nextchar == 'n' and string[idx:idx + 4] == 'null':
            return None, idx + 4
        elif nextchar == 't' and string[idx:idx + 4] == 'true':
            return True, idx + 4
        elif nextchar == 'f' and string[idx:idx + 5] == 'false':
            return False, idx + 5

        m = match_number(string, idx)
        if m is not None:
            integer, frac, exp = m.groups()
            if frac or exp:
                res = parse_float(integer + (frac or '') + (exp or ''))
            else:
                res = parse_int(integer)
            return res, m.end()
        elif nextchar == 'N' and string[idx:idx + 3] == 'NaN':
            return parse_constant('NaN'), idx + 3
        elif nextchar == 'I' and string[idx:idx + 8] == 'Infinity':
            return parse_constant('Infinity'), idx + 8
        elif nextchar == '-' and string[idx:idx + 9] == '-Infinity':
            return parse_constant('-Infinity'), idx + 9
        else:
            raise StopIteration(idx)

    def scan_once(string, idx):
        try:
            return _scan_once(string, idx)
        finally:
            memo.clear()

    return _scan_once

make_scanner = c_make_scanner or py_make_scanner