[Python-ideas] Serialization of CSV vs. JSON

[Philip Martin]

My bad. I think I need to clarify my objective. I definitely understand the
issues regarding serialization/deserialization on JSON, i.e. decimals as
strings, etc., and hooking in a default serializer function is easy enough.
I guess my question is more related to why the csv writer and DictWriter
don't provide similar functionality for serialization/deserialization
hooks? There seems to be a wide gap between reaching for a tool like pandas
were maybe too much auto-magical parsing and guessing happens, and wrapping
the functionality around the csv module IMO. I was curious to see if anyone
else had similar opinions, and if so, whether conversion around what
extended functionality would be most fruitful?

On Fri, Nov 2, 2018 at 11:28 AM Calvin Spealman <cspealma at redhat.com> wrote:

> First, this list is not appropriate. You should ask such a question in
> python-list.
>
> Second, JSON is a specific serialization format that explicitly rejects
> datetime objects in *all* the languages with JSON libraries. You can only
> use date objects in JSON if you control or understand both serialization
> and deserialization ends and have an agreed representation.
>
> On Fri, Nov 2, 2018 at 12:20 PM Philip Martin <philip.martin2007 at gmail.com>
> wrote:
>
>> Is there any reason why date, datetime, and UUID objects automatically
>> serialize to default strings using the csv module, but json.dumps throws an
>> error as a default? i.e.
>>
>> import csv
>> import json
>> import io
>> from datetime import date
>>
>> stream = io.StringIO()
>> writer = csv.writer(stream)
>> writer.writerow([date(2018, 11, 2)])
>> # versus
>> json.dumps(date(2018, 11, 2))
>>
>> _______________________________________________
>> Python-ideas mailing list
>> Python-ideas at python.org
>> https://mail.python.org/mailman/listinfo/python-ideas
>> Code of Conduct: http://python.org/psf/codeofconduct/
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-ideas/attachments/20181102/2f4bc5be/attachment.html>