def _parse_canonical_dbref(doc):
"""Decode a JSON DBRef to bson.dbref.DBRef."""
for key in doc:
if key.startswith('$') and key not in _DBREF_KEYS:
# Other keys start with $, so dct cannot be parsed as a DBRef.
return doc
return DBRef(doc.pop('$ref'),
doc.pop('$id'),
database=doc.pop('$db', None),
**doc)
def _get_object(data, position, obj_end, opts):
"""Decode a BSON subdocument to opts.document_class or bson.dbref.DBRef."""
obj_size = _UNPACK_INT(data[position:position + 4])[0]
end = position + obj_size - 1
if data[end:position + obj_size] != b"\x00":
raise InvalidBSON("bad eoo")
if end >= obj_end:
raise InvalidBSON("invalid object length")
obj = _elements_to_dict(data, position + 4, end, opts)
position += obj_size
if "$ref" in obj:
return (DBRef(obj.pop("$ref"), obj.pop("$id", None),
obj.pop("$db", None), obj), position)
return obj, position
def _get_ref(data, position, obj_end, opts):
"""Decode (deprecated) BSON DBPointer to bson.dbref.DBRef."""
collection, position = _get_string(data, position, obj_end, opts)
oid, position = _get_oid(data, position, obj_end, opts)
return DBRef(collection, oid), position
def object_hook(dct):
if "$oid" in dct:
return ObjectId(str(dct["$oid"]))
if "$ref" in dct:
return DBRef(dct["$ref"], dct["$id"], dct.get("$db", None))
if "$date" in dct:
dtm = dct["$date"]
# mongoexport 2.6 and newer
if isinstance(dtm, string_type):
aware = datetime.datetime.strptime(
dtm[:23], "%Y-%m-%dT%H:%M:%S.%f").replace(tzinfo=utc)
offset = dtm[23:]
if not offset or offset == 'Z':
# UTC
return aware
else:
if len(offset) == 5:
# Offset from mongoexport is in format (+|-)HHMM
secs = (int(offset[1:3]) * 3600 + int(offset[3:]) * 60)
elif ':' in offset and len(offset) == 6:
# RFC-3339 format (+|-)HH:MM
hours, minutes = offset[1:].split(':')
secs = (int(hours) * 3600 + int(minutes) * 60)
else:
# Not RFC-3339 compliant or mongoexport output.
raise ValueError("invalid format for offset")
if offset[0] == "-":
secs *= -1
return aware - datetime.timedelta(seconds=secs)
# mongoexport 2.6 and newer, time before the epoch (SERVER-15275)
elif isinstance(dtm, collections.Mapping):
secs = float(dtm["$numberLong"]) / 1000.0
# mongoexport before 2.6
else:
secs = float(dtm) / 1000.0
return EPOCH_AWARE + datetime.timedelta(seconds=secs)
if "$regex" in dct:
flags = 0
# PyMongo always adds $options but some other tools may not.
for opt in dct.get("$options", ""):
flags |= _RE_OPT_TABLE.get(opt, 0)
return Regex(dct["$regex"], flags)
if "$minKey" in dct:
return MinKey()
if "$maxKey" in dct:
return MaxKey()
if "$binary" in dct:
if isinstance(dct["$type"], int):
dct["$type"] = "%02x" % dct["$type"]
subtype = int(dct["$type"], 16)
if subtype >= 0xffffff80: # Handle mongoexport values
subtype = int(dct["$type"][6:], 16)
return Binary(base64.b64decode(dct["$binary"].encode()), subtype)
if "$code" in dct:
return Code(dct["$code"], dct.get("$scope"))
if "$uuid" in dct:
return uuid.UUID(dct["$uuid"])
if "$undefined" in dct:
return None
if "$numberLong" in dct:
return Int64(dct["$numberLong"])
if "$timestamp" in dct:
tsp = dct["$timestamp"]
return Timestamp(tsp["t"], tsp["i"])
return dct