python-bsonjs

Info:	See github for the latest source.
Author:	Shane Harvey <[email protected]>

About

A fast BSON to MongoDB Extended JSON converter for Python that uses libbson.

Installation

python-bsonjs can be installed with pip:

$ python -m pip install python-bsonjs

Examples

>>> import bsonjs
>>> bson_bytes = bsonjs.loads('{"hello": "world"}')
>>> bson_bytes
'\x16\x00\x00\x00\x02hello\x00\x06\x00\x00\x00world\x00\x00'
>>> bsonjs.dumps(bson_bytes)
'{ "hello" : "world" }'

Using bsonjs with pymongo to insert a RawBSONDocument.

>>> import bsonjs
>>> from pymongo import MongoClient
>>> from bson.raw_bson import RawBSONDocument
>>> client = MongoClient("localhost", 27017, document_class=RawBSONDocument)
>>> db = client.test
>>> bson_bytes = bsonjs.loads('{"_id": 1, "x": 2}')
>>> bson_bytes
'\x15\x00\x00\x00\x10_id\x00\x01\x00\x00\x00\x10x\x00\x02\x00\x00\x00\x00'
>>> result = db.test.insert_one(RawBSONDocument(bson_bytes))
>>> result.inserted_id  # NOTE: inserted_id is None
>>> result.acknowledged
True
>>> raw_doc = db.test.find_one({'x': 2})
>>> raw_doc.raw == bson_bytes
True
>>> bsonjs.dumps(raw_doc.raw)
'{ "_id" : 1, "x" : 2 }'

Speed

bsonjs is roughly 10-15x faster than PyMongo's json_util at decoding BSON to JSON and encoding JSON to BSON. See benchmark.py:

$ python benchmark.py
Timing: bsonjs.dumps(b)
10000 loops, best of 3: 0.110911846161
Timing: json_util.dumps(bson.BSON(b).decode())
10000 loops, best of 3: 1.46571397781
bsonjs is 13.22x faster than json_util

Timing: bsonjs.loads(j)
10000 loops, best of 3: 0.0628039836884
Timing: bson.BSON().encode(json_util.loads(j))
10000 loops, best of 3: 0.683200120926
bsonjs is 11.72x faster than json_util

Limitations

Top Level Arrays

Because libbson does not distinguish between top level arrays and top level documents, neither does python-bsonjs. This means that if you give dumps or dump a top level array it will give you back a dictionary. Below are two examples of this behavior

>>> import bson
>>> from bson import json_util
>>> import bsonjs
>>> bson.decode(bsonjs.loads(json_util.dumps(["a", "b", "c"])))
{'0': 'a', '1': 'b', '2': 'c'}
>>> bson.decode(bsonjs.loads(json_util.dumps([])))
{}

One potential solution to this problem is to wrap your list in a dictionary, like so

>>> list = ["a", "b", "c"]
>>> dict = {"data": list}
>>> wrapped = bson.decode(bsonjs.loads(json_util.dumps(dict)))
{'data': ['a', 'b', 'c']}
>>> wrapped["data"]
['a', 'b', 'c']

Installing From Source

python-bsonjs supports CPython 3.8+.

Compiler

You must build python-bsonjs separately for each version of Python. On Windows this means you must use the same C compiler your Python version was built with.

• Windows build requires Microsoft Visual Studio 2015

Source

You can download the source using git:

$ git clone https://github.com/mongodb-labs/python-bsonjs.git

Install

Once you have the source properly downloaded, build and install the package:

$ pip install -v .

Test

To run the test suite:

$ python -m pytest