dict_list_to_recarray.py

dict_list_to_recarray.py

#!/usr/bin/python
import numpy as np

def dict_list_to_recarray(items):
    """
    @brief Take a loose list of dictionaries, and turn them into a recarray with
    fields defined as 'object' for clarity
    @param items Iterable collection of dictionaries
    @author Eric Cousineau <eacousineau@gmail.com>
    @note xref: http://docs.scipy.org/doc/numpy/reference/generated/numpy.core.records.fromrecords.html
    """
    # Collect the unique fields
    fields_raw = []
    for item in items:
        fields_raw += item.keys()
    fields = set(fields_raw)
    # Get value sets, replacing empty fields with 'None'
    rec_list = []
    for item in items:
        values = [item.get(field, None) for field in fields]
        rec_list.append(values)
    # Construct the np.recarray
    return np.core.records.fromrecords(rec_list, names=','.join(fields))

if __name__ == '__main__':
    raw = [
        {'name': 'Billy', 'has_money': True, 'age': 45},
        {'name': 'Tommy', 'has_money': False, 'age': 32, 'extra': 'Something wicked'},
        {'name': 'Al', 'has_money': True, 'age': 31}
        ]
    rec = dict_list_to_recarray(raw)
    
    print "Names: {}".format(', '.join(rec.name))
    print "Mean Age: {}".format(np.mean(rec.age))
    print "Money Havers: {} / {}".format(np.sum(rec.has_money), len(rec))
    print "Extras: {}".format(', '.join(map(str, rec.extra)))
    
    """
    Names: Billy, Tommy, Al
    Mean Age: 36.0
    Money Havers: 2 / 3
    Extras: None, Something wicked, None
    """