json.dumps() claims numpy.ndarray and numpy.bool_ are not serializable

[miscjunk]

BPO	18303
Nosy	@bitdancer, @ethanfurman

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2013-06-26.06:09:58.338>
created_at = <Date 2013-06-25.21:43:54.400>
labels = ['extension-modules', 'invalid', 'type-bug', 'library']
title = 'json.dumps() claims numpy.ndarray and numpy.bool_ are not serializable'
updated_at = <Date 2013-06-26.06:09:58.337>
user = 'https://bugs.python.org/miscjunk'

bugs.python.org fields:

activity = <Date 2013-06-26.06:09:58.337>
actor = 'ethan.furman'
assignee = 'none'
closed = True
closed_date = <Date 2013-06-26.06:09:58.338>
closer = 'ethan.furman'
components = ['Extension Modules', 'Library (Lib)']
creation = <Date 2013-06-25.21:43:54.400>
creator = 'miscjunk'
dependencies = []
files = []
hgrepos = []
issue_num = 18303
keywords = []
message_count = 5.0
messages = ['191887', '191889', '191890', '191892', '191895']
nosy_count = 3.0
nosy_names = ['r.david.murray', 'ethan.furman', 'miscjunk']
pr_nums = []
priority = 'normal'
resolution = 'not a bug'
stage = None
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue18303'
versions = ['Python 3.3']

[miscjunk]

When json.dumps() is called on a numpy.bool_ object, it crashes.

To recreate:
import numpy as np
import json
a = np.array([1,2,3,4,5,6])
a = a < 5
json.dumps(a) #crash
json.dumps(a[0]) #crash
json.dumps(a.tolist()) #this works!

Example of error output
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "..\python-3.3.2.amd64\lib\json\__init__.py", line 236, in dumps
    return _default_encoder.encode(obj)
  File "..\python-3.3.2.amd64\lib\json\encoder.py", line 191, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "..\python-3.3.2.amd64\lib\json\encoder.py", line 249, in iterencode
    return _iterencode(o, 0)
  File "..\python-3.3.2.amd64\lib\json\encoder.py", line 173, in default
    raise TypeError(repr(o) + " is not JSON serializable")
TypeError: False is not JSON serializable

[miscjunk]

numpy.ndarray is also not serializable by json.dumps()

import numpy as np
import json

a = np.array([1,2,3,4])
json.dumps(a) #crash
json.dumps(a[0]) #this works!

[bitdancer]

I would (naively) not expect either of these to be serializable. They are not base data types, which is all the json module handles unless you add extra handlers yourself.

'crash' means interpreter crash (segfault), by the way.

[miscjunk]

Thanks for the explanation. I suppose this should be posted to the numpy tracker then? Would it be possible for numpy to 'just work' with the json module? Or will the final resolution be to use a handler (the default= parameter in json.loads) ?

Thanks

[ethanfurman]

For the curious, here are all the tracebacks:

--> json.dumps(a)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.3/json/__init__.py", line 236, in dumps
    return _default_encoder.encode(obj)
  File "/usr/lib/python3.3/json/encoder.py", line 191, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "/usr/lib/python3.3/json/encoder.py", line 249, in iterencode
    return _iterencode(o, 0)
  File "/usr/lib/python3.3/json/encoder.py", line 173, in default
    raise TypeError(repr(o) + " is not JSON serializable")
TypeError: array([ True,  True,  True,  True, False, False], dtype=bool) is not JSON serializable


--> json.dumps(a[0])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.3/json/__init__.py", line 236, in dumps
    return _default_encoder.encode(obj)
  File "/usr/lib/python3.3/json/encoder.py", line 191, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "/usr/lib/python3.3/json/encoder.py", line 249, in iterencode
    return _iterencode(o, 0)
  File "/usr/lib/python3.3/json/encoder.py", line 173, in default
    raise TypeError(repr(o) + " is not JSON serializable")
TypeError: True is not JSON serializable

While the repr says 'True', the type is <class 'numpy.bool_'>.

and the success:

--> json.dumps(a.tolist()) #this works!
'[true, true, true, true, false, false]'

Summary
=======

No bug here, defined behavior. Raising the issue with NumPy won't help as this is not a bug... although perhaps they have a json handler already written somewhere? At any rate, yes, said handler would have to be specified as the 'default' parameter.