pyOdk does not fetch submission attachments.

[BritoAlv]

I have a form and submissions of it on my odk server instance, when using pyOdk it does not fetch the attachments from my submissions, I checked that attachments exists on odk central. Can someone confirm that this is a bug or a problem with my setup?

[yanokwa]

It's most likely a problem with your setup, so I am closing this issue.

Please describe your setup in detail on the community forum to get further assistance.

[lognaturel]

I don't think we currently have a method specifically for downloading submission attachments so you would have to use HTTP verb methods. Describing what you'd ideally like to achieve on the forum as @yanokwa would be a great next step, thank you!

[BritoAlv]

I can't post on your forum, so I'm replying here what I wrote.

1. What is the issue? Please be detailed.

pyOdk does not fetch attachments from submissions.

2. What steps can we take to reproduce this issue?

In your odk server instance setup this form :

form.xlsx

As you can see its only purpose is to receive images.

Submit this form.

Now with pyOdk there is the following API or wrapper for communicating with the server via python:

from pyodk import Client

client = Client()
project_id = 3
form_id = "form"
submission_id = "uuid:f686c5f5-7edc-4798-967f-80cc1e6180d1"

print(
    client.submissions.get(
        instance_id=submission_id,
        form_id=form_id,
        project_id=project_id,
    )
)

The output is this:

instanceId='uuid:f686c5f5-7edc-4798-967f-80cc1e6180d1' submitterId=61 createdAt=datetime.datetime(2026, 2, 3, 15, 50, 18, 682000, tzinfo=TzInfo(UTC)) deviceId=None reviewState=None userAgent='Enketo/7.5.1 Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/142.0.0.0 Safari/537.36' instanceName=None updatedAt=None attachments=None

As you can see there are no attachments, even though they should exist.

I’m using these versions of ODK Central:

versions:
cf374ca368e93182124396701bd76770430ecf2f (v2025.4.1-1-gcf374ca)
 bd1be46611410f2c1f441b154f4d625b27e093c5 client (v2025.4.0)
 97981e9694f392e5b3df90a2d28c6f7d3aefef17 server (v2025.4.0)

3. What have you tried to fix the issue?

I debugged the pyOdk code and it seems it’s not calling anything for fetching the submission’s attachments.

If the intent or goal of the attachments field in the Submission object from pyOdk is to obtain the attachments of the submissions, seems like it does not work.

Anyways, using the HTTP verbose methods, I was able to fetch the attachments without problems.

@lognaturel If there is no method for downloading the attachments, then it maybe misleading the Attachments field on the sumbission object returned by pyODK. By reading the code, I may guess that the code for obtaining the attachments is there, in the class SubmissionAttachmentService.

@yanokwa I reproduced the issue in other form as I described and still get None.

Thanks.

[lognaturel]

Oh no, I'm sorry the forum didn't work for you! Please let me know what happened so we can troubleshoot that as well. The reason we try to have this kind of conversation on the forum is that it brings together more users who can chime in with their ideas on what would be most helpful.

Central generally provides different endpoints for accessing a resource, its metadata, and its attachments. That's not likely a very natural way for end users to think about it. Currently pyodk's get methods get metadata only and I'm not sure how useful that really is. My naive expectation not having recollection of the details would be that attachments would be an array of either the attachments' filenames or resource id. It would be helpful to document Submission.

I'm glad that for now you were able to use the raw HTTP methods to meet your needs. Can you share a bit more about your end goal? E.g. are you building a local replica of data over time? Pulling just once? That will help us make sure we can meet that need better. Thanks!

[BritoAlv]

Thanks for your time.

The issue with the forum was my fault, I created a post about the problem that was too vague, so I decided to delete it immediately after posting. Shortly after, I received a notification that my account was put on hold. I would like to move our conversation back there.

My end goal is to perform some statistical analysis on python, which is why I need to fetch the data. The HTTP methods can be used, but it's much simpler to use a wrapper for communicating with the API; this seems to be the purpose of pyOdk.

I've inspected the code, and now understand the following:

As you mentioned the get method, only fetches metadata. While it has the field Attachments there, it isn't utilized because the ODK Central response for that query does't include attachment information.

I originally thought the get method would fetch information for a specific submission via its ID, but it appears only metadata is retrieved. Based on my understanding now, I should use the get_table method and then (after receiving the response) filter for the specific submission data I need.

[lognaturel]

Shortly after, I received a notification that my account was put on hold

We've been drowning in spam recently so our automatic filters are tuned aggressively. I've un-silenced you so feel free to post again there with some ideas on what your ideal workflow for getting submission attachments would look like and we can get more of the community involved. I'm sure you're not the only one with this need.

I should use the get_table method and then (after receiving the response) filter for the specific submission data I need

Yes, getting the data in table format would generally be the recommended approach for analysis. We've been adding "native" functionality iteratively as we get explicit requests and haven't explored fetching submission attachments yet. You may have seen there is support for creating them.

Downloading submission attachments present unique challenges because they're binary data and possibly quite large. It would be useful to know more about what your next steps with them would be. For example, are you embedding them in a dashboard that will be published somewhere? Generating one-time PDF reports? ...? Will you download all the data once at the end of data collection or are you planning to do live reporting as the data comes in? We'll have to consider whether to save the files to disk or hold them in memory for the user to do something with and whether to batch fetch them or give the user control in making each request. Good topics for the forum, see you there!