[jira] [Commented] (NIFI-4272) ReplaceText processor does not properly iterate multiple replacement values when EL is used

["Otto Fowler (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-4272?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16493546#comment-16493546 ] 

Otto Fowler commented on NIFI-4272:
-----------------------------------

I have a test to reproduce this issue, and I understand why it is happening.

The EL get's evaluated before the regex, so the regex ends up NOT having the first capture group $1 in it, but instead after the EL has NAME, thus no replacement down the line.

Without el the replacement ( String.replaceAll(REGEX)) regex has new$1 in it so it works.

I am looking at how to resolve this, I think I have a methodology, but I don't think it can be implemented as things currently stand in the frameworks.

 

Basically, we need to evaluate the EL AFTER we do the regex replacement.

 
{code:java}
@Test
public void testWithELInReplacement() throws IOException {
    final TestRunner runner = getRunner();
    runner.setProperty(ReplaceText.SEARCH_VALUE, "\"([a-z]+)\":\"(\\w+)\"");
    runner.setProperty(ReplaceText.REPLACEMENT_VALUE, "${'$1':toUpper()}:$2");
    //runner.setProperty(ReplaceText.REPLACEMENT_VALUE, "\"new$1\":\"$2\"");
    runner.enqueue("{\"name\":\"Smith\",\"middle\":\"nifi\",\"firstname\":\"John\"}");
    runner.run();

    runner.assertAllFlowFilesTransferred(ReplaceText.REL_SUCCESS, 1);
    final MockFlowFile out = runner.getFlowFilesForRelationship(ReplaceText.REL_SUCCESS).get(0);
    out.assertContentEquals("{\"NAME\":\"Smith\",\"MIDDLE\":\"nifi\",\"FIRSTNAME\":\"John\"}");

}
{code}

> ReplaceText processor does not properly iterate multiple replacement values when EL is used
> -------------------------------------------------------------------------------------------
>
>                 Key: NIFI-4272
>                 URL: https://issues.apache.org/jira/browse/NIFI-4272
>             Project: Apache NiFi
>          Issue Type: Bug
>          Components: Core Framework
>    Affects Versions: 1.1.0, 1.2.0, 1.3.0
>            Reporter: Matthew Clarke
>            Priority: Major
>
> I am using the ReplaceText processor to take a string input (example:   {"name":"Smith","middle":"nifi","firstname":"John"} ) and change all the filed names to all uppercase.
> Using above input as an example, I expect output like {"NAME":"Smith","MIDDLE":"nifi","FIRSTNAME":"John"}
> I expect I should be able to do this with ReplaceText processor; however, I see some unexpected behavior:
> -------
> Test 1:  (uses EL in the replacement value property)
> Search value:  \"([a-z]+?)\":\"(.+?)\"
> Replacement Value: \"*${'$1':toUpper()}*":\"$2\"
> Result: {"NAME":"Smith","NAME":"nifi","NAME":"John"}
> -------
> Test 2:  (Does not use EL in the replacement Value property)
> Search value:  \"([a-z]+?)\":\"(.+?)\"
> Replacement Value: \"new$1":\"$2\"
> Result: {"newname":"Smith","newmiddle":"nifi","newfirstname":"John"}
> --------
> As you can see if I use a NiFi expression Language statement in the Replacement Value property it no longer iterates as expect through the various $1 captured values. It repeatedly uses the EL result from the first EL evaluation in every iteration while $2 correctly iterates through the search values.



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