[Octave-patch-tracker] [patch #9980] JSON encoder and decoder, alternative to object2json

[Philip Nienhuis]

Update of patch #9980 (project octave):

                  Status:               Need Info => In Progress            

    _______________________________________________________

Follow-up Comment #21:

Just curious and for some more evaluation, in Octave-7.0.0 I tried fromJSON
and jsonencode on some random json file in my Firefox and Seamonkey profiles
(lots of json files there),
For example:

>> file = 'broadcast-listeners.json'
file = broadcast-listeners.json
>> fid = fopen (file);
txt = fread (fid, Inf, "*char")';
fclose (fid);
>> txt
txt =
{"version":1,"listeners":{"remote-settings/monitor_changes":{"version":"\"1603569466778\"","sourceInfo":{"moduleURI":"resource://services-settings/remote-settings.js","symbolName":"remoteSettingsBroadcastHandler"}}}}

>> fromJSON (txt)
ans =
  scalar structure containing the fields:
    version = 1
    listeners =
      scalar structure containing the fields:
        remote-settings/monitor_changes =
          scalar structure containing the fields:
            version: 1x17 sq_string
            sourceInfo: 1x1 scalar struct
>> ans.listeners.('remote-settings/monitor_changes').version
ans = \"1603569466778\"

>> jsondecode (txt)
ans =
  scalar structure containing the fields:
    version = 1
    listeners =
      scalar structure containing the fields:
        remote_settings_monitor_changes =
          scalar structure containing the fields:
            version: 1x15 sq_string
            sourceInfo: 1x1 scalar struct
>> ans.listeners.remote_settings_monitor_changes.version
ans = "1603569466778"


Clearly Octave-7.0.0's jsondecode adapts structfield names and maybe struct
field contents, I suppose for easier handling and maybe Matlab compatibility,
while fromJSON delivers them 1:1. Yet fromJSON's output string with the
escaped backslashes looks suspicious.
Again I have no idea if these results are right or wrong; opinions, anyone?

I'm inclined to think that it's a good thing that users can (also) get "raw"
data from JSON files (provided they're correct). But that's me, being used to
digging around in data files I tend to prefer to stay as close as possible to
raw data before processing them my own way.

In another json file in my seamonkey profile I got

>> file = 'sessionstore-4.json';
:
>> jsondecode (txt);
## no errors
>> fromJSON (txt);
error: regexprep: the input string is invalid UTF-8
error: called from
    fromJSON at line 61 column 7


... but I suppose that's because the URL's inside can have varying encodings
(peeking in the file I saw a lot of gunk).
Octave-7.0.0's jsondecode seems more resilient.

    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/patch/?9980>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/