def _element_to_bson(key, value, check_keys, opts):
"""Encode a single key, value pair."""
if not isinstance(key, string_type):
raise InvalidDocument("documents must have only string keys, "
"key was %r" % (key, ))
if check_keys:
if key.startswith("$"):
raise InvalidDocument("key %r must not start with '$'" % (key, ))
if "." in key:
raise InvalidDocument("key %r must not contain '.'" % (key, ))
name = _make_name(key)
return _name_value_to_bson(name, value, check_keys, opts)
def _make_name(string):
"""Make a 'C' string suitable for a BSON key."""
# Keys can only be text in python 3.
if "\x00" in string:
raise InvalidDocument("BSON keys / regex patterns must not "
"contain a NUL character")
return _utf_8_encode(string)[0] + b"\x00"
def _name_value_to_bson(name, value, check_keys, opts):
"""Encode a single name, value pair."""
# First see if the type is already cached. KeyError will only ever
# happen once per subtype.
try:
return _ENCODERS[type(value)](name, value, check_keys, opts)
except KeyError:
pass
# Second, fall back to trying _type_marker. This has to be done
# before the loop below since users could subclass one of our
# custom types that subclasses a python built-in (e.g. Binary)
marker = getattr(value, "_type_marker", None)
if isinstance(marker, int) and marker in _MARKERS:
func = _MARKERS[marker]
# Cache this type for faster subsequent lookup.
_ENCODERS[type(value)] = func
return func(name, value, check_keys, opts)
# If all else fails test each base type. This will only happen once for
# a subtype of a supported base type.
for base in _ENCODERS:
if isinstance(value, base):
func = _ENCODERS[base]
# Cache this type for faster subsequent lookup.
_ENCODERS[type(value)] = func
return func(name, value, check_keys, opts)
raise InvalidDocument("cannot convert value of type %s to bson" %
type(value))
def _make_c_string_check(string):
"""Make a 'C' string, checking for embedded NUL characters."""
if isinstance(string, bytes):
if b"\x00" in string:
raise InvalidDocument("BSON keys / regex patterns must not "
"contain a NUL character")
try:
_utf_8_decode(string, None, True)
return string + b"\x00"
except UnicodeError:
raise InvalidStringData("strings in documents must be valid "
"UTF-8: %r" % string)
else:
if "\x00" in string:
raise InvalidDocument("BSON keys / regex patterns must not "
"contain a NUL character")
return _utf_8_encode(string)[0] + b"\x00"