[
https://issues.apache.org/jira/browse/NIFI-14480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17945226#comment-17945226
]
Naman Arora commented on NIFI-14480:
------------------------------------
Thanks for the guidance [~exceptionfactory] — I think all of those are great
points. Here’s how I’d like to address them:
*1. Public Drools+JSON examples*
* You’re absolutely right — many enterprise Drools integrations aren’t
open-sourced, which limits the number of public examples. To make things
crystal clear for reviewers, I’ll include concise, self-contained code snippets
in the extension’s README that show exactly how a JSON payload can be
deserialized and processed with Drools.
Here’s a generic working snippet:
{code:java}
// Deserialize JSON to Map
ObjectMapper mapper = new ObjectMapper();
Map<String, Object> jsonPayload = mapper.readValue(inputStream, new
TypeReference<>(){});
// Insert into Drools session
KieSession kieSession = kieContainer.newKieSession();
kieSession.insert(jsonPayload);
kieSession.fireAllRules();
kieSession.dispose();
{code}
* Additionally, there are some useful open-source projects like [this
one|https://github.com/ityouknow/drools-examples] that showcase Drools usage
more broadly.
*2. Streaming vs. full parse*
* Under the hood, Drools rules execute against Java objects, so you’ll need to
deserialize the JSON (we’ll use Jackson). However, we can integrate this into a
streaming approach by using Jackson’s {{Streaming API}} ({{{}JsonParser{}}})
and constructing rule‐target objects incrementally, rather than loading the
entire JSON tree into memory first. I’ll include a proof‑of‑concept that reads
the incoming {{InputStream}} in chunks, maps fields on the fly, and fires rules
as data arrives.
*3. Dedicated NAR bundle*
* I agree that a separate {{nifi-drools-nar}} makes sense for clarity and
maintainability. I’ll set up the Maven modules accordingly so it can be
contributed as an extension without touching {{{}nifi-standard{}}}.
*4. Processor name*
* *DroolsTransformJSON* fits nicely with our existing pattern—happy to stick
with that.
Thanks again for your feedback
Best,
Naman
> Add DroolsProcessor for rule-based JSON processing with dynamic/static DRL
> support
> ----------------------------------------------------------------------------------
>
> Key: NIFI-14480
> URL: https://issues.apache.org/jira/browse/NIFI-14480
> Project: Apache NiFi
> Issue Type: New Feature
> Components: Extensions
> Reporter: Naman Arora
> Priority: Major
>
> This processor provides integration of the Drools Rule Engine with Apache
> NiFi. It allows for flexible rule-based processing of JSON payloads in two
> modes:
> 1. **Static DRL Mode**:
> - User provides an absolute path to a DRL file.
> - That rule is used for all incoming flowfiles.
> 2. **Dynamic DRL Mode**:
> - User specifies a base path and a key name.
> - Processor extracts the relative DRL path from the key inside the
> flowfile JSON payload.
> - Dynamically fetches the rule and executes it.
> Additional Features:
> - **File Watcher** service that watches static and dynamic paths for DRL file
> changes and reloads the KIE Container automatically.
> - **Retry Mechanism** for rule execution with configurable retry attempts.
> This processor supports advanced decisioning and rule separation, and helps
> enable dynamic business rule application inside NiFi dataflows.
> I would like to contribute this processor under the Apache License. I have
> signed and submitted the ICLA.
> Let me know the next steps or if any feedback is required. Thanks!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
