This helper looks good, but it might make sense to move it into its own small util file (e.g., snakeToTitleCase.ts) so it can be reused wherever needed.

You may also want to add a guard to handle fieldName or name being null or undefined, or update the type signature to NonNullable<string> so the function guarantees it always receives a valid string. Adding a fallback return value or throwing a clear error will make this more robust and help avoid silent failures. Below is a code suggestion that includes a safety guard and a helpful function name change.

Finally, the transformation logic is straightforward, but you could make it slightly safer and more concise by trimming input and ignoring extra underscores (e.g., .filter(Boolean), which removes an additional space if the field name has two or more underscores), or by handling unexpected characters.

Yeah I think I will just remove this helper. It seems best to keep the field name on the table matching the field name on the csv so there is no confusion. And it reduces the need for text manipulation.

Is it possible that Google will change this link in the future, rendering it dead?

There is a chance it would change, but this seems like a straightforward way for now. But open to other ideas.

After digging deeper into this, it seems that Google keeps their links quite stable. The FILE_ID is unlikely to change even if you modify the file's name, location, or permissions. If you want to prevent users from accidentally editing the file, you could update the link to a read-only format, like this: https://docs.google.com/spreadsheets/d/FILE_ID/view. Other than that, this approach looks good.

Cypress intercept misreads multipart payload

The upload flow posts a FormData (see handleSubmit creating a File and appending it to FormData), but the new Cypress test assumes req.body is a CSV string and immediately casts it to string. For multipart requests req.body is a FormData/Blob, so the include assertions will run against "[object FormData]" and always fail even when the CSV is sent correctly. This makes the added test a permanent false negative. Consider extracting the blob from req.body (e.g., req.body.get('file') and reading it) before asserting on its contents.

Useful? React with 👍 / 👎.

I originally tried this approach, but the Cypress interceptor does not allow the .get('file'), the submission is just sent as a large string in the body in the Cypress environment, from what I can tell.