[Bug]: [Python] ReadFromJson stores nested JSON objects and arrays as Python repr strings

Open
  • Is something wrong?
#41,098 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
3/5
Estimated time
Half a day
Newbie friendliness
70/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
pandas, python
Domain
data-engineering

Research direction

Start in sdks/python/apache_beam/dataframe/io.py, in ReadViaPandas.expand, where object-dtype columns are cast with astype(pd.StringDtype()). Run the reproducer from the issue to see the nested dict and list values come out as Python repr. Done when nested objects and arrays read from JSON are stored as valid JSON strings, and a regression test in textio_test.JsonTest covers this.

Written by the indexing model from the issue text.

Description

awaiting triage bug io P2 yaml
What happened?

beam.io.ReadFromJson (and therefore Beam YAML ReadFromJson) reads nested JSON objects and arrays as strings holding their Python repr instead of JSON. For the line {"id": 1, "user": {"name": "a"}, "tags": ["x", "y"]} the element is id=1, user="{'name': 'a'}", tags="['x', 'y']". The single-quoted repr is not valid JSON, so downstream steps can't recover the nested values, and the schema reports the field as Optional[str].

Minimal reproducer:

import json, os, tempfile
import apache_beam as beam

with tempfile.TemporaryDirectory() as d:
  path = os.path.join(d, 'nested.jsonl')
  with open(path, 'w') as f:
    f.write('{"id": 1, "user": {"name": "a"}, "tags": ["x", "y"]}\n')
  with beam.Pipeline() as p:
    (p
     | beam.io.ReadFromJson(path)
     | beam.Map(lambda r: print(repr(r.user), json.loads(r.user))))
# prints "{'name': 'a'}" and then fails with json.decoder.JSONDecodeError
Cause

ReadFromJson delegates to apache_beam.dataframe.io.ReadViaPandas('json', ...). With the default objects_as_strings=True, ReadViaPandas.expand (sdks/python/apache_beam/dataframe/io.py) casts every object-dtype column with df[col].astype(pd.StringDtype()). For JSON input, pandas keeps nested objects and arrays as dict / list values in object-dtype columns, so the cast stores str(dict).

Encoding dict / list values with json.dumps before the cast (JSON format only) keeps the column a string but makes it valid JSON. I have a small fix with a regression test in textio_test.JsonTest and can open a PR.

Issue Priority

Priority: 2 (default / most bugs should be filed as P2)

Issue Components
  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam YAML
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Infrastructure
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Prism Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner
Dominant language
Java
Stars
8.7k
Forks
4.8k
Avg merge
2d 9h
Merged PRs (30d)
368

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from apache/beam

All issues in apache/beam

Similar issues

More Java issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.