Fix: Preserve float columns in JSON loader when values are integer-like (e.g. 0.0, 1.0) #7635
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ke (e.g. 0.0, 1.0)
This PR fixes a bug in the JSON loader where columns containing float values like `[0.0, 1.0, 2.0]` were being implicitly coerced to `int`, due to pandas or Arrow type inference.
This caused issues downstream in statistics computation (e.g., dataset-viewer) where such columns were incorrectly labeled as `"int"` instead of `"float"`.
### 🔍 What was happening:
When the JSON loader falls back to `pandas_read_json()` (after `pa.read_json()` fails), pandas/Arrow can coerce float values to integers if all values are integer-like (e.g., `0.0 == 0`).
### ✅ What this PR does:
- Adds a check in the fallback path of `_generate_tables()`
- Ensures that columns made entirely of floats are preserved as `"float64"` even if they are integer-like (e.g. `0.0`, `1.0`)
- This prevents loss of float semantics when creating the Arrow table
### 🧪 Reproducible Example:
```json
[{"col": 0.0}, {"col": 1.0}, {"col": 2.0}]
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR fixes a bug in the JSON loader where columns containing float values like
[0.0, 1.0, 2.0]were being implicitly coerced toint, due to pandas or Arrow type inference.This caused issues downstream in statistics computation (e.g., dataset-viewer) where such columns were incorrectly labeled as
"int"instead of"float".🔍 What was happening:
When the JSON loader falls back to
pandas_read_json()(afterpa.read_json()fails), pandas/Arrow can coerce float values to integers if all values are integer-like (e.g.,0.0 == 0).✅ What this PR does:
_generate_tables()"float64"even if they are integer-like (e.g.0.0,1.0)🧪 Reproducible Example:
[{"col": 0.0}, {"col": 1.0}, {"col": 2.0}]Previously loaded as:
intNow correctly loaded as:
floatFixes #6937