jdata.encode not working for np.complex64 scalar

[tyu-mit]

For scalar values of np.complex64 type, encoder does not process it correctly.

Example:

import numpy as np
import jdata as jd

x = {"a": np.complex64(1+1j)}
enc_x = jd.encode(x)
print(enc_x)

Expected behavior:

{'a': {'_ArrayType_': 'single',
  '_ArraySize_': 1,
  '_ArrayIsComplex_': True,
  '_ArrayData_': [1.0, 1.0]}}

Actual behavior:

{'a': (1+1j)}

Suggested change:

Add condition isinstance(d, np.complex64): to jdata.py

[fangq]

@tyu-mit, sorry for the delay, the above patch should fix the reported issue. this is the output on my side

import jdata as jd
import numpy as np
x = {"a": np.complex64(1+1j)}
enc_x = jd.encode(x)
jd.decode(enc_x)
  {'a': (1+1j)}

print(enc_x)
  {'a': {'_ArrayType_': 'single', '_ArraySize_': [1], '_ArrayIsComplex_': True, '_ArrayData_': [[1.0], [1.0]]}}