pymongo 64位无符号整数

Question

I'm trying to insert a 64bit unsigned integer in mongodb using pymongo. The integer is the output of a CRC64 algorithm. I tried to following:

long(crc64(unicode(kw).encode('unicode-escape'))))

If I insert this into mongodb it starts to complain that only 64bit integers are supported by mongodb. Next I tried to convert it to a signed 64bit int like so:

ctypes.c_int64(crc64(unicode(kw).encode('unicode-escape')))).value

Which kind of works, mongodb stops complaining about the size of my int, but when I look at the data in mongodb I get this:

{
    "_id" : {
        "floatApprox" : -5307924876159732000,
        "top" : 3059119730,
        "bottom" : 2651469802 },
    "keyword" : "redacted",
    "normal_hash" : { 
        "floatApprox" : -671156942315906300,
        "top" : 4138701393,
        "bottom" : 549001936
    } 
}

What's going on here? Is there any way to put the 64bit int into the db as just an int (don't really care whether it's signed or unsigned.)

Answer 1

MongoDB uses BSON to store data, and the BSON spec says 64bit integer numbers are signed.

A sample session on a 64bit machine, 64bit mongo v2.0.1, python 2.6.5:

>>> num = long(9007199254740992)
>>> num
9007199254740992L
>>> bson = BSON.encode({"crc64":num})
>>> bson
'\x14\x00\x00\x00\x12crc64\x00\x00\x00\x00\x00\x00\x00 \x00\x00'
>>> bson_str = bson.decode()
>>> bson_str
{u'crc64': 9007199254740992}
>>> 

and running this script:

db.foo.save({"_id" : 1, "crc64" : long(9007199254740992)});

for doc in db.foo.find({"_id" : 1 }):
    crc = doc["crc64"]
    print("crc type: " + str(type(crc)))

prints:

crc type: <type 'int'>

and from the mongo shell:

> db.foo.findOne()
{ "_id" : 1, "crc64" : NumberLong("9007199254740992") }
>