engrcao / pyjsoncpp

Python Bindings to JsonCpp

PyJsonCpp: Python bindings for JsonCpp

The JsonCpp project provides an excellent in-memory JSON data structure as well as string writers and parsers. Here are Python bindings for JsonCpp using Cython.

Installation

Installing PyJsonCpp from source is easy. First, download or clone this repository. Then run the setup commands from the base directory:

cd pyne/
python setup.py install --user

These bindings require on Cython v0.17+. The tests rely on nose.

Usage Example

The Value object is the main interface for Values may be converted to and from regular Python types. These have the normal behavior for their type.

>>> from jsoncpp import Value, Reader, FastWriter, StyledWriter

>>> v = Value({'name': 'Terry Jones', 'age': 42.0})
>>> v['quest'] = "To find the grail."
>>> v.keys()
['age', 'name', 'quest']
>>> v['name']
'Terry Jones'

>>> v = Value([1, 2, 5, 3])
>>> v[1:-1]
[2, 5]
>>> v[1:-1] = [42, 65]
>>> v
[1, 42, 65, 3]

>>> v = Value("No one expects the Spanish Inquisition!!!!")
>>> len(v)
42

The Python Value class provides a view into the underlying C++ class. This allows you to create several views into the same data. For example, start with the following nested dictionary:

# make a nested dict and a view into a top-level item
>>> v = Value({'a': {'b': 14}})
>>> view_a = v['a']
>>> view_a
{"b":14}

# add an item to the view
>>> view_a['c'] = 16
>>> view_a
{"b":14,"c":16}

# this item is present in the original value
>>> v
{"a":{"b":14,"c":16}}

Furthermore, there is a Reader class for converting JSON strings or files into Value instances. There are also two writer classes, FastWriter and StyledWriter, for converting Value instances into compact and human-readable strings respectively. For example:

>>> v = Value({'hello': 1})
>>> fw = FastWriter()
>>> fw.write(v)
'{"hello":1}\n'

>>> sw = StyledWriter()
>>> print sw.write(v)
{
   "hello" : 1
}

>>> r = Reader()
>>> new_v = r.parse('{"hello":1}\n')
>>> isinstance(new_v, Value)
True