DynamicSizeRecarray

A dynamic, appandable version of Numpy's recarray. The goal is to have a recarray-like-object which can be appended to in a transparent and efficient way.

install

pip install dynamicsizerecarray

basic use

Initialize an empty DynamicSizeRecarray with a dtype.

import dynamicsizerecarray

dra = dynamicsizerecarray.DynamicSizeRecarray(
    dtype=[("hour", "u1"), ("minute", "u1"), ("temperature", "f8")]
)

len(dra)
0

Or initialize the DynamicSizeRecarray with an already existing recarray.

import numpy

rec = numpy.core.records.recarray(
    shape=1,
    dtype=[("hour", "u1"), ("minute", "u1"), ("temperature", "f8")],
)
rec["hour"][0] = 2
rec["minute"][0] = 13
rec["temperature"][0] = 20.123

dra = dynamicsizerecarray.DynamicSizeRecarray(recarray=rec)

len(dra)
1

After initializing, the DynamicSizeRecarray can be appended to dynamically. You can append a record, i.e. a dict.

dra.append_record({"hour": 3, "minute": 53, "temperature": 22.434})

len(dra)
2

Or you can append another recarray.

rec = numpy.core.records.recarray(
    shape=1,
    dtype=[("hour", "u1"), ("minute", "u1"), ("temperature", "f8")],
)
rec["hour"][0] = 13
rec["minute"][0] = 41
rec["temperature"][0] = 18.623

dra.append_recarray(rec)

len(dra)
3

When the dynamic appending is done, the DynamicSizeRecarray can be exported to a classic, and static recarray.

final = dra.to_recarray()

Further the DynamicSizeRecarray provides the properties shape and dtype, and also implements __gettitem__, __setitem__.

dra.shape                   # shape
(3, )

dra[0]                      # __gettitem__
(2, 13, 20.123)

dra[1] = (7, 25, 21.45)     # __setitem__

len(dra)                    # __len__
3

dra.dtype                   # exposes the internal recarray's dtype
dtype((numpy.record, [('hour', 'u1'), ('minute', 'u1'), ('temperature', '<f8')]))

wording

• record is a dict with keys (and values) matching the dtype of the DynamicSizeRecarray. (Wording is adopted from pandas).
• records is just a list of record s (Also adopted from pandas).
• recarray is short for np.core.records.recarray.