Feature auto infer orient as table or split in read_json

[chandra-teajunkie]

• closes ENH: automatically detect orient= for read_json if not supplied #52713
• [Tests added and passed]
• All [code checks passed]
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

Example:

This example demonstrates how to use the updated functionality of read_json to correctly handle the new orient='table' format.

Saving DataFrame to JSON with orient='table':

import pandas as pd

# Create a sample DataFrame
df = pd.DataFrame({"A": [1, 2, 3]})

df.to_json('test.json', indent=1, orient='table')
read = pd.read_json('test.json')
print(read)

   A
0  1
1  2
2  3

Saving DataFrame to JSON with orient='split':

df = pd.DataFrame({"A": [1, 2, 3]})

df.to_json('test.json', indent=1, orient='split')
read = pd.read_json('test.json')
print(read)

   A
0  1
1  2
2  3

[WillAyd]

I am generally -1 on this feature, especially if it will require fully loading the document twice

[WillAyd]

So this is loading the entire JSON document twice now? Isn't that going to at least double the runtime?

[chandra-teajunkie]

Hi @WillAyd

Thanks a lot for your feedback. Much appreciated.
Yeah I did not fully consider the performance aspect of this.

So, if there is no orient explicitly mentioned by the user, this will read the document one extra time as you have mentioned.

But, I couldn't think of any other way to validate the schema of the json file to automatically infer an appropriate orient.

[mroeschke]

I am also generally -1 on this feature. We've tried to move toward explicit behaviors over time, and inferencing goes against explicitness.

[chandra-teajunkie]

Hi @mroeschke
Oh alright.

[WillAyd]

Thanks for the interest in a PR @chandra-teajunkie , but from the discussion I don't think this is one we will move forward with

[chandra-teajunkie]

Thanks for the feedbacks!