JSON serialization

Work in progress

This page is a stub. Paper-voice content arrives in a follow-up writing session.

tsecon.io.json provides lossless round-trip of Frequency / MIT / MITRange / TSeries / MVTSeries / Workspace to and from JSON.

Each composite carries a _type field that names the class so the decoder can rebuild it. Plain Python scalars (int, float, str, bool, None) and homogeneous lists pass through unwrapped — no _type overhead when it isn't needed.

import json

from tsecon import qq, TSeries
from tsecon.io.json import dumps, loads

t = TSeries(qq(2020, 1), [1.0, 2.0, 3.0, 4.0])
encoded = dumps(t)
roundtripped = loads(encoded)

assert roundtripped.equals(t)
print(json.dumps(json.loads(encoded), indent=2))

{
  "_type": "TSeries",
  "firstdate": {
    "_type": "MIT",
    "freq": {
      "name": "Quarterly",
      "end_month": 3
    },
    "value": 8080
  },
  "dtype": "float64",
  "values": [
    1.0,
    2.0,
    3.0,
    4.0
  ]
}

Why JSON, not pickle?

• Portable. Pickle is Python-specific; JSON is consumable by every language in the BoC pipeline.
• Inspectable. A failing round-trip is debugged by reading the JSON, not by running pickletools.
• Versionable. The _type discriminator is forward-compatible: a future version can rename a class and ship a migration shim without breaking on-disk artefacts.