[jira] [Commented] (NIFI-5998) EvaluateRecordPath Processor

["Bryan Bende (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-5998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16768476#comment-16768476 ] 

Bryan Bende commented on NIFI-5998:
-----------------------------------

Most of the other evaluate processors are working on a flow file where the content is a single root document.

Since the record approach promotes bundling many records into a single flow file to improve performance, does it really make sense to evaluate record path against many records?

Imagine you had 1k records in a flow file, lets say csv records with fields (a, b, c) and then you tried to evaluate /a, the results are 1k values for /a, so you don't want to put these in flow file attributes. You could possibly write it to the content of the flow file, but why not just use QueryRecord and write a SQL statement to select field a, or use a ConvertRecord processor with a write schema that only has field a.

 

> EvaluateRecordPath Processor
> ----------------------------
>
>                 Key: NIFI-5998
>                 URL: https://issues.apache.org/jira/browse/NIFI-5998
>             Project: Apache NiFi
>          Issue Type: New Feature
>    Affects Versions: 1.8.0
>            Reporter: Nimrod Avni
>            Priority: Minor
>
> We should have a EvaluateRecrodPath processor, which accepts a record reader as a property.
> similarly to the EvaluateJsonPath and EvaluateXpath we should have some of their properties like:
>  * Destination: flowfile-content or flowfile-attributes
>  * Return type: return only the value of the record path or the entire structure (xml,json,csv), should have auto-detect,structure,value
>  * Path not found behavior: ignore,warn or fail
>  * Null Value Representation: empty string or the string 'null' 
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)