`from_ini()` returns a `str` instead of a `dict`

[d-chris]

Python version
Python 3.11.0

Package version
python-benedict==0.30.0

Current behavior (bug description)

from benedict import benedict

original = benedict({
    'section1': {'key1': 'value1'},
    'sectionA': {'keyA': 'valueA', 'keyB': 'valueB', 'keyC': {'subkeyC': 'subvalueC'}},
})

readback = benedict.from_ini(original.to_ini())


print('original vs readback')

path = 'sectionA.keyC'

print(original.get(path), type(original.get(path)))
print('-- vs --')
print(readback.get(path), type(readback.get(path)))

output

original vs readback
{'subkeyC': 'subvalueC'} <class 'benedict.dicts.benedict'>
-- vs --
{'subkeyC': 'subvalueC'} <class 'str'>

Expected behavior
value type of the keyC is a dict not a string

[fabiocaccamo]

@d-chris the problem is that the ini format doesn't support nested data structures, for this reason everything beyond the first level of depth gets stringified.

[d-chris]

damn it, thought so.

i didn't inspect your package very thoroughly but i tried to use get_dict() but it always returns the default-value

print(readback.get_dict(path), type(readback.get_dict(path)))

output

{} <class 'benedict.dicts.benedict'>

maybe i can use the json.loads() method somehow, to convert the str back into a dict

[fabiocaccamo]

@d-chris I could add a new option to the ini serializer for serialize nested values in json format.

The only doubt about this approach is the decoding: how can we know if a string value is a json string that needs to be decoded or not?!

[d-chris]

as far as i understand ini files:

[section]
key=value

i guess the encoder doesn't need any changes?

or what will happen when you always use in the INISerializer something like the following

pseudo-code

#encode() - write values into section of an ini
writeline(f"{key}={json.dumps(value)}"

#decode() - read lines/values from a section of an ini
section_value={}

for line in section:
  key, value = split(line) # e.g. use regex r"^(?P<key>\w+)\=(?P<value>.*?$"
  section_value[key] = json.loads(value)

[fabiocaccamo]

@d-chris I check if this is backward-compatible.

[d-chris]

@d-chris I check if this is backward-compatible.

thx

for me your to_ini() is ok, only from_ini() needs to return a dict

ini-file

[avr]
setup = {'bitclock': '125kHz', 'chiperase': True}
fuses = {'bitclock': '1.8MHz', 'lock': '0xFC'}

[m8]
parent = avr
mcu = m8
setup = {'lfuse': '0xE4'}

python-dict

ini = benedict({
    'avr': {
        'setup': {'bitclock': '125kHz', 'chiperase': True},
        'fuses': {'bitclock': '1.8MHz', 'lock': '0xFC'}
    },
    'm8': {
        'parent': 'avr',
        'mcu': 'm8',
        'setup': {'lfuse': '0xE4'},
    }
}

[d-chris]

i guess the encoder doesn't need any changes?

seems that encoder also must be changed

>>>  json.dumps({'key':True})
'{"key": true}'

in the 'str' key's and values must be in double-quotes that json.loads works. and True must be lower-case in the ini-file

so the encoder must be also modified

[d-chris]

via inheritance it seems to work

import benedict as bd
import json


class benedict(bd.benedict):
    def to_ini(self, *args, **kwargs):
        def super_class(*vargs):
            # return  bd.benedict(*vargs)
            return self.__class__.__bases__[0](*vargs)

        data = dict()

        for section, keys in self.dict().items():
            data[section] = dict()
            for key, value in keys.items():
                data[section][key] = json.dumps(value)

        return super_class(data).to_ini(*args, **kwargs)

    @classmethod
    def from_ini(cls, s, **kwargs):
        ini = super().from_ini(s, **kwargs)

        data = dict()

        for section, keys in ini.dict().items():
            data[section] = dict()
            for key, value in keys.items():
                data[section][key] = json.loads(value)

        return cls(data)


if __name__ == '__main__':

    original = benedict({
        'avr': {
            'setup': {'bitclock': '125kHz', 'chiperase': True},
            'fuses': {'bitclock': '1.8MHz', 'lock': '0xFC'}
        },
        'm8': {
            'parent': 'avr',
            'mcu': 'm8',
            'setup': {'lfuse': '0xE4'},
        }
    })

    content = original.to_ini()

    readback = benedict.from_ini(content)

    print(original.avr.setup.chiperase, readback.avr.setup.chiperase)