[
https://issues.apache.org/jira/browse/NIFI-11959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17763770#comment-17763770
]
Daniel Stieglitz edited comment on NIFI-11959 at 9/11/23 6:35 PM:
------------------------------------------------------------------
[~mattyb149] I was able to trace this bug to code you had added for NIFI-4957
lines 488-495 in JoltTransformRecord and lines 396-402 in JoltTransformJSON
{code:java}
private String readTransform(final PropertyValue propertyValue) {
final ResourceReference resourceReference = propertyValue.asResource();
try (final BufferedReader reader = new BufferedReader(new
InputStreamReader(resourceReference.read()))) {
return reader.lines().collect(Collectors.joining());
} catch (final IOException e) {
throw new UncheckedIOException("Read JOLT Transform failed", e);
}
}{code}
Per [this article|https://www.baeldung.com/convert-input-stream-to-string]
'reader.lines()' uses readLine underneath the hood which thereby strips the new
lines. For single line comments ('//') that seems to prevent the JSON from
being parsed even if the parser is set to ignore comments. Multiline comments
('/**/') are unaffected. The fix for this would be to reinsert the new lines
with
{code:java}
return reader.lines().collect(Collectors.joining(System.lineSeparator()));{code}
I just wanted to double check with you first if there was any other reason not
to have the new lines in the spec. Please advise. Thanks!
was (Author: JIRAUSER294662):
[~mattyb149] I was able to trace this bug to code you had added for NIFI-4957
lines 488-495 in JoltTransformRecord and lines 396-402 in JoltTransformJSON
{code:java}
private String readTransform(final PropertyValue propertyValue) {
final ResourceReference resourceReference = propertyValue.asResource();
try (final BufferedReader reader = new BufferedReader(new
InputStreamReader(resourceReference.read()))) {
return reader.lines().collect(Collectors.joining());
} catch (final IOException e) {
throw new UncheckedIOException("Read JOLT Transform failed", e);
}
}{code}
Per [this article|https://www.baeldung.com/convert-input-stream-to-string]
'reader.lines()' uses readLine underneath the hood which thereby strips the new
lines. For single line comments ('//') that seems to prevent the JSON from
being parsed even if the parser is set to ignore comments. Multiline comments
('/**/') are unaffected. The fix for this would to reinsert the new lines with
{code:java}
return reader.lines().collect(Collectors.joining(System.lineSeparator()));{code}
I just wanted to double check with you first if there was any other reason not
to have the new lines in the spec. Please advise. Thanks!
> Jolt spec in JoltTransformJson processor no longer supporting text comments
> ---------------------------------------------------------------------------
>
> Key: NIFI-11959
> URL: https://issues.apache.org/jira/browse/NIFI-11959
> Project: Apache NiFi
> Issue Type: Bug
> Affects Versions: 1.23.0
> Reporter: Mahieddine Cherif
> Assignee: Daniel Stieglitz
> Priority: Major
> Attachments: Screenshot 2023-08-17 at 09.24.44-1.png, jolt-issue.json
>
>
> Since NiFi 1.23.0, it seems like comments are no longer supported in Jolt
> Spec (JoltTransformJson/JoltTransformRecord processors)
> This however seems something that was already supported at least the most
> wide spread jolt engine [https://jolt-demo.appspot.com/#inception]
> *Reproducer :*
> We can easily reproduce it by simply putting a jolt with a comment using
> either JoltTransformRecord or JoltTransformJson processors as both are
> impacted.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
