Source code is available at github.
The code is licensed MIT. See the included LICENSE file for the exact terms.
JSONstreams is a package that attempts to making writing JSON in a streaming format easier. In contrast to the core json module, this package doesn't require building a complete tree of dicts and lists before writing, instead it provides a straightforward way to to write a JSON document without building the whole data structure ahead of time.
JSONstreams considers there to be two basic types, the JSON array and the JSON object, which correspond to Python's list and dict respectively, and can encode any types that the json.JSONEncoder can, or can use an subclass to handle additional types.
The interface is designed to be context manger centric. The Stream class, and the Array and Object classes returned by the subarray and subobject methods (respectively), can be used as context managers or not, but use as context managers are recommended to ensure that each container is closed properly.
A simple example looks like this
import jsonstreams
with jsonstreams.Stream(jsonstreams.Type.OBJECT, filename='foo') as s:
s.write('foo', 'bar')
with s.subobject('a') as a:
a.write('foo', 1)
a.write('bar', 2)
s.write('bar', 'foo')Writing into a closed group will raise an exception, which should not be handled, this exception is always an error in programming and should be corrected.
It is possible to write any value that the encoder (json.JSONEncoder by default) can encode, so iterating over lists or dictionaries to write them in is unnecessary:
import jsonstreams
mylist = list(range(10))
mydict = {a: b for a in range(10) for b in 'abcdefghij'}
with jsonstreams.Stream(jsonstreams.Type.OBJECT, filename='foo') as s:
s.write('list', mylist)
s.write('dict', mydict)