A Python script to convert Hasura APIs from 🐍snake_case to 🐪camelCase using aliases through Hasura's metadata API + introspection.

You can run the script as a copy using Google Colab here: https://colab.research.google.com/drive/1LYajKSAqg6FEruDfWSVN4RB1JfpWdH9A

Please read and test with your setup first before running on your project.

1. Finds all fields with _aggregate at the end
2. De-camels any fields which already are camelCased (ie. if there's a field like alreadyCamelCase_Aggregate)
   • This step is optional
3. Explode the words in snake_case
4. Convert the last word of the phrase to plural + singular using inflect
5. Re-join the words in camelCase
6. Build the payload
   • If you look for the custom_root_fields JSON build steps you can modify those to customize your naming convention
7. Send to the /v1/metadata API

Example for => Table Name : active_user
---------------------------------
Before                   : After
active_user              : activeUsers
active_user_by_pk        : activeUser
active_user_aggregagte   : activeUsersAggregate
active_user_insert       : activeUsersInsert
active_user_insert_one   : activeUserInsert
active_user_update       : activeUsersUpdate
active_user_update_by_pk : activeUserUpdate
active_user_delete       : activeUsersDelete
active_user_delete_by_pk : activeUserDelete

This version is built for the 2.0+ pg_set_table_customization API which supports multiple data sources (which this script isn't currently handling, it relies on the default database - it would probably need something like the note below where we'd get table names via runSQL) [https://hasura.io/docs/latest/graphql/core/api-reference/metadata-api/table-view.html#pg-set-table-customization]

To back port it to 1.3.4, pretty much all the logic would be the same minus some minor tweaks to the JSON payload which is being sent using set_table_customization instead [https://hasura.io/docs/latest/graphql/core/api-reference/schema-metadata-api/table-view.html#set-table-customization].

For versions around 1.3.3 where this feature was introduced you can also back port to the older set_table_custom_fields which should be a similarily small tweak to the payload being sent [https://hasura.io/docs/latest/graphql/core/api-reference/schema-metadata-api/table-view.html#set-table-custom-fields-deprecated]

This was created pretty quickly so feel free to let me know if there are any improvements you see.

The one edge case that I can imagine is if someone has already aliased their root field with a new name which also has _aggregate in it. This would fail for that field since the /v1/metdata API is expecting a table name with the reference (and the alias would be different from the table name).

A fix for this would be to use the runSQL API (https://hasura.io/docs/latest/graphql/core/api-reference/schema-metadata-api/run-sql.html) to check that all fields matched database tables / views beforehand -- it seemed like a very infrequent edge-case at the time, I may update that as a later update.