Add field to logger.Attachment that matches question response verbatim

[jnm]

From the OpenRosa Form Submission API specification:

Additional Parts

• Content Type: arbitrary
• Name: matches an appropriate element inside of the XForm Element's XML.

This means that if an XML submission contains:

<take_a_picture_of_your_shoes>Screenshot from 2022-03-30 12-12-56-12_13_3.png</take_a_picture_of_your_shoes>

…then the POST data should (eventually, because a single submission can be sent in multiple POSTs) contain a file with the identical name Screenshot from 2022-03-30 12-12-56-12_13_3.png. Great, all's well.

However, this file ends up being stored in the logger.Attachment model, where we and Django (see kobotoolbox/kpi#3659) do some transformations on the file name.

The result is that we do not have a good way to link the response to the take_a_picture_of_your_shoes question (as received in the XML but also stored in JSON), i.e. Screenshot from 2022-03-30 12-12-56-12_13_3.png, to its corresponding Attachment object and the media_file it contains.

We should add a new field on logger.Attachment that stores the verbatim question response. We could also take the opportunity to store the question XPath in the Attachment model as well (what do you think, @dorey?)

[noliveleger]

@jnm Or/and what about adding the question name in the _attachments list in Json response too?

[jnm]

@jnm Or/and what about adding the question name in the _attachments list in Json response too?

Aha, we were writing along the same lines 😄 I think perhaps we should use the XPath?

[Yann-J]

Hello here... Interested in following up here, since currently using the API to retrieve attachments is ambiguous because of this.
Any idea of a timeline for adding this explicit attachment<>question linkage?
Thx!