Conversation
|
Missing test for the headline feature. Tests cover static pipeline, field extract/strip, nested prune, JSON Array, suppress-nulls, and Update. There is no NDJSON FlowFile with two lines and two different pipeline names in one bulk. That is the behavior the PR text sells. No Create case. Docs say Index and Create; tests only use default Index plus Update. Cheap to add. PutElasticsearchRecord is unchanged. Fine if the JIRA is Json-only; say so in the PR so nobody treats it as “ES ingest pipelines are done.” -- |
Add Pipeline, Pipeline Field, and Retain Pipeline Field properties, mirroring the existing Index configuration. Pipeline sets a static ingest pipeline for every document, while Pipeline Field resolves the pipeline per document from the payload and honors the Field Path Mode property, so the value can come from a top-level field or a nested "/"-delimited path. When the field is absent or the property is blank, the static Pipeline value is used as the fallback. The resolved value is written to the bulk action header for Index and Create operations, so records within a single FlowFile can be routed through different ingest pipelines in one _bulk request. Retain Pipeline Field controls whether the field is stripped from the document body before indexing; the field is stripped whenever requested, including for operations the pipeline itself does not apply to, so routing metadata is never indexed. Generalize the index name extraction helpers to resolveFieldValue and extractFieldValue since they are now shared with the pipeline resolution.
75a78b2 to
0693884
Compare
|
Added two new tests (testDifferentPipelinesPerDocumentInSingleBulk/testPipelineAppliedToCreateOperation) and added a clarifying line to the description. |
2 participants
This change is limited to PutElasticsearchJson. PutElasticsearchRecord is intentionally unchanged;
Add Pipeline, Pipeline Field, and Retain Pipeline Field properties, mirroring the existing Index configuration. Pipeline sets a static ingest pipeline for every document, while Pipeline Field resolves the pipeline per document from the payload and honors the Field Path Mode property, so the value can come from a top-level field or a nested "/"-delimited path. When the field is absent or the property is blank, the static Pipeline value is used as the fallback.
The resolved value is written to the bulk action header for Index and Create operations, so records within a single FlowFile can be routed through different ingest pipelines in one _bulk request. Retain Pipeline Field controls whether the field is stripped from the document body before indexing; the field is stripped whenever requested, including for operations the pipeline itself does not apply to, so routing metadata is never indexed.
Generalize the index name extraction helpers to resolveFieldValue and extractFieldValue since they are now shared with the pipeline resolution.
Summary
NIFI-16247
Tracking
Please complete the following tracking steps prior to pull request creation.
Issue Tracking
Pull Request Tracking
NIFI-00000NIFI-00000VerifiedstatusPull Request Formatting
mainbranchVerification
Please indicate the verification steps performed prior to pull request creation.
Build
./mvnw clean install -P contrib-checkLicensing
LICENSEandNOTICEfilesDocumentation