Add option to keep null values in generated options

Despite the big effort made in pyecharts, it normal to find ECharts options unavailable using pyecharts classes. This is where the _option attribute becomes helpful (is there a better way?). Sometimes an ECharts option needs to be set to null as in the example below when setting a candlestick chart.

itemStyle: {
  normal: {
    color: '#06B800',
    color0: '#FA0000',
    borderColor: null,
    borderColor0: null
    }
}

When adding these keys to _option, the keys whose value is None are removes from the generated options because of the following code in the Base class:

@property
def options(self):
    return utils.remove_key_with_none_value(self._option)

Also this makes it harder when there’s a lot of work needs to be done with _option as the removed keys need to be recreated first. For example, in the example above itemStyle needed to be created first, then the dictionary value assigned ({'normal' : { 'color': ...} }) assigned instead of accessing individual keys.

Is there a way to handle this? if not, I suggest adding an option to preserve null values in the generated options such as:

bar = Bar(remove_none_options=True)

[chfw]

OK. Just checked the options documentation, if pyecharts omits both options, borderColor defaults to #c23531 and borderColor0 defaults to #314656.

So I would propose that pyecharts provides ‘NULL’ definition to signal that the key should be present and with value ‘null’ in js/html. In this way, remove_key_with_none_value() will ignore NULLed keys.

[YasserHassan]

Great. What do you think the best value that NOT_SET should take? I think NumPy’s way of using nanis good if there’s no other safe value in mind. As show in numpy.nan’s documentation, it is “IEEE 754 floating point representation of Not a Number (NaN)”. In this case, is_not_set() could be the equivalent of NumPy’s isnan() to test for unassigned options.