Implementing Dynamic Tag with Attribute Coercion via Python Sidecar & XSLT
Rady
Hello dcm4che Community,
I wanted to share a solution I’ve recently implemented for dcm4chee-arc-light that addresses a common challenge: Dynamic, database-driven tag modification during outgoing transfers.
The Challenge:
In our environment, we needed to modify the Retrieve AE Title (0008,0054) and Accession Number (0008,0050) on-the-fly when data is moved via C-MOVE to a destination. This logic depends on an external MySQL database, and we required the changes to be transient (in-memory only) to ensure the original data in our S3-backed storage remained untouched.
The Solution:
I’ve developed a "Sidecar Bridge" pattern that allows dcm4chee to maintain its enterprise-scale performance while gaining the agility of external scripting:
Attribute Coercion: A standard rule in the Archive Device set to C_STORE_RQ and role SCP.
XSLT Script: A namespace-agnostic stylesheet that uses the XSLT document() function to make real-time HTTP calls.
Python FastAPI Bridge: A lightweight service that queries the MySQL database and returns the required tags as XML.
Performance Optimization: We implemented TTL (Time-To-Live) Caching on the Python side to ensure that multi-image studies don't overwhelm the database, resulting in near-zero latency for subsequent instances.
The Implementation Journey:
It took about a week of deep-diving into the archive’s internal XML representation and WildFly’s logging subsystems to get this production-ready. I found that the "secret" to success was building a local XSLT testing environment in Python to simulate dcm4chee’s behavior, which allowed for rapid iteration without needing to trigger real moves or restart the container.
I have documented the full technical breakdown, including:
The FastAPI code with async database handling.
The XSLT stylesheet designed for both production and local testing.
The WildFly logging configuration required to see xsl:message output in the server.log.
The Testing Suite I used to validate the flow before deployment.
You can find the comprehensive guide and all code samples on my blog:
👉 https://radyworks.wordpress.com/
I’m sharing this in hopes it helps anyone looking to extend dcm4chee’s routing capabilities with external business logic.