This module is for you if you're tired of writing boilerplate that:
- builds a straightforward Python object from loaded JSON or similar dict-based data loading (e.g. CSV).
- checks that your input-loaded-JSON has all necessary attributes for your pipeline.
- checks that your input JSON has the right types.
Main intended usage is through the JSONclass
decorator, example below:
By default we don't check for anything, we just build the object as we received it.
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa', 'b': 'bb', 'c': 1}
>>> @JSONclass
... class Example:
... pass
...
>>> example = Example(data)
>>> example.a
'aa'
>>> example.b
'bb'
We want to ensure we have annotated parameters
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa', 'b': 'bb', 'c': 1}
>>> @JSONclass(annotations=True)
... class Example:
... a: str
... d: int
...
>>> try:
... example = Example(data)
... except KeyError:
... print("error - missing 'd'")
...
error - missing 'd'
>>> data['d'] = 1 # Let's fix the missing data
>>> example = Example(data) # No more error in loading.
We want to ensure we have only annotated parameters
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa', 'b': 'bb', 'c': 1}
>>> @JSONclass(annotations=True, annotations_strict=True)
... class Example:
... a: str
... b: int
...
>>> try:
... example = Example(data)
... except KeyError:
... print("error - extra 'c'")
...
error - extra 'c'
>>> del data['c'] # Let's remove unwanted data
>>> example = Example(data) # No more error in loading.
We want to ensure we have only annotated parameters and they are of annotated type.
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa', 'b': 'bb'}
>>> @JSONclass(annotations_strict=True, annotations_type=True)
... class Example:
... a: str
... b: int
...
>>> try:
... example = Example(data)
... except TypeError:
... print("error - b is not int")
...
error - b is not int
Default values are supported too.
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa'}
>>> @JSONclass(annotations_strict=True, annotations_type=True)
... class Example:
... a: str
... b: int = 1
...
>>> example = Example(data)
>>> example.b
1
A JSONclass can be converted back to a dict, even for recursive structures
>>> from jsonloader import JSONclass
>>> data = {'a': 'aa', 'b':2, 'c': {'foo': 'bar'}}
>>> @JSONclass(annotations_type=True, annotations=True)
... class Example:
... a: str
... b: int
...
>>> example = Example(data)
>>> assert dict(example) == data
# Recommendation: install jsonloader in your project virtualenv
# Should you not want to use virtualenv or equivalent, it's recommended to use
# '--user' pip option to avoid a system-level install.
pip3 install jsonloader
Github repository currently points to latest development version. Please
jump to latest released version tag if you intend to work on PyPI version.
For example git checkout tags/v0.4.3
.
# These commands assume virtualenv is installed
python3 -m virtualenv venv
. venv/bin/activate
# Actually install the deps
pip3 install -e '.[dev]'
# To setup the project's git hooks:
git config --local core.hooksPath hook
# From this repository top directory
pytest --doctest-modules
For example, leverage coverage
module:
coverage run -m pytest --doctest-modules
coverage html