Download all latest submission attachments for a given form_id

download_attachments.py

from pyodk.client import Client
import json
from functools import reduce
import urllib.parse
import os.path

def get_attachment_fields():
    schema_r = client.get(f"projects/{client.config.central.default_project_id}/forms/{FORM_ID}/fields?odata=true")
    schema = schema_r.json()

    return [field for field in schema if field['binary']]

# https://stackoverflow.com/a/52260663/137744
def recursive_get(data_dict, path):
    try:
        return reduce(dict.get, path, data_dict)
    except TypeError:
        return None

def save_all_attachments(form_id, path="media"):
    attachment_fields = get_attachment_fields()

    for submission in data:
        submission_id = submission["__id"]

        submission_url = f"projects/{client.config.central.default_project_id}/forms/{FORM_ID}/submissions/{submission_id}"

        for field in attachment_fields:
            attachment_path = field["path"].lstrip("/").split("/")
            attachment_filename = recursive_get(submission, attachment_path)

            if attachment_filename is not None:
                attachment_r = client.get(f"{submission_url}/attachments/{urllib.parse.quote(attachment_filename)}")
                if attachment_r.status_code == 200:
                    attachment_path = os.path.join(path, attachment_filename)
                    os.makedirs(os.path.dirname(attachment_path), exist_ok=True)
                    with open(attachment_path, "wb") as f:
                        f.write(attachment_r.content)
                else:
                    print(f"{attachment_r.status_code} for {attachment_filename} in {submission_id}")
                    attachment_r.raise_for_status()

FORM_ID = "all-widgets"

client = Client().open()
data = client.submissions.get_table(form_id=FORM_ID)['value']                    
                    
save_all_attachments(FORM_ID)