Create new metadata fields that normalize units

[mccalluc]

When it was just for display, it didn't matter, but now that there are histograms, sorting, etc., we need numeric fields that are all on the same scale. Related to @john-conroy's search page work.

Not clear how to choose the best unit when their are multiple... Always choose the first or the last? or for that matter could map it to multiple scales...

[mccalluc]

Proposal:

from pint.unit import Unit

metadata = {
     'xyz_value': 1000, 'xyz_unit': 'ng',
     'abc_value': 60, 'abc_unit': 's'
}

target_units = ['g', 'kg', 'second', 'year'] # add or remove from this list as desired.

def _units_map(metadata):
    fields = metadata.keys()
    value_stems = set(f.replace('_value', '') for f in fields if f.endswith('_value'))
    unit_stems = set(f.replace('_unit', '') for f in fields if f.endswith('_unit'))
    stems = value_stems & unit_stems
    for stem in stems:
        q = metadata[f'{stem}_value'] * Unit(metadata[f'{stem}_unit'])
        for target_unit in target_units:
            if not q.units.is_compatible_with(target_unit):
                continue 
            q_conv = q.to(target_unit)
            
            metadata[f'{stem}_in_{q_conv.units}'] = float(f"{q_conv.magnitude:.3g}")
    return metadata

_units_map(metadata)

{'abc_in_second': 60.0,
 'abc_in_year': 1.9e-06,
 'abc_unit': 's',
 'abc_value': 60,
 'xyz_in_gram': 1e-06,
 'xyz_in_kilogram': 1e-09,
 'xyz_unit': 'ng',
 'xyz_value': 1000}