jk-shah opened a new issue, #41102:
URL: https://github.com/apache/beam/issues/41102
### What happened?
`beam.io.WriteToJson` (and Beam YAML `WriteToJson`) silently loses float
precision: values are rounded to 10 decimal places.
| input | written |
|---|---|
| `0.1234567890123` | `0.123456789` |
| `3.3333333333333335` | `3.3333333333` |
| `12345678901.5` | `12345678901.5` |
There is no warning, so a JSON export doesn't round-trip and differs from
the same data written with `WriteToParquet`, `WriteToCsv` or `json.dumps`.
Minimal reproducer (Beam YAML, Beam 2.75.0; `WriteToJson` on `master` @
`728ef1b4` has the same code path):
```yaml
pipeline:
type: chain
transforms:
- type: Create
config:
elements:
- {id: 1, x: 0.1234567890123}
- {id: 2, x: 3.3333333333333335}
- type: WriteToJson
config:
path: /tmp/c5
```
`python -m apache_beam.yaml.main --yaml_pipeline_file=c5.yaml` writes:
```
{"id":1,"x":0.123456789}
{"id":2,"x":3.3333333333}
```
### Cause
`WriteToJson` (`sdks/python/apache_beam/io/textio.py`) writes through the
DataFrame API, `apache_beam.dataframe.io.to_json` ->
`pandas.DataFrame.to_json`, and only forwards the caller's `**kwargs`. pandas'
`to_json` defaults to `double_precision=10`, so every float is rounded to 10
decimal places.
Options:
1. Default `double_precision=15` (pandas' maximum) when the caller doesn't
pass it. That fixes most cases but still isn't exact for 16-17 significant
digits.
2. Write JSON lines with `json.dumps` for `orient='records', lines=True`
(the YAML default), which round-trips floats exactly.
At minimum the docstring should mention the rounding and the
`double_precision` kwarg. Happy to send a PR for whichever approach maintainers
prefer.
### Issue Priority
Priority: 2 (default / most bugs should be filed as P2)
### Issue Components
- [x] Component: Python SDK
- [ ] Component: Java SDK
- [ ] Component: Go SDK
- [ ] Component: Typescript SDK
- [x] Component: IO connector
- [x] 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: Twister2 Runner
- [ ] Component: Hazelcast Jet Runner
- [ ] Component: Google Cloud Dataflow Runner
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
