[
https://issues.apache.org/jira/browse/BEAM-12863?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Thomas Pocreau updated BEAM-12863:
----------------------------------
Description:
When using PubsubAvroToBigQuery Dataflow template, [I noticed this issue.
|https://github.com/GoogleCloudPlatform/DataflowTemplates/issues/287]
This seems to be related to the way
[toBeamRow|https://github.com/apache/beam/blob/v2.32.0/sdks/java/io/google-cloud-platform/src/main/java/org/apache/beam/sdk/io/gcp/bigquery/BigQueryUtils.java#L593]
is handling TableRow with List containing null values.
The error trace has this path (I only added the relevant ones):
a.
com.google.cloud.teleport.v2.transforms.BigQueryConverters$TableRowToGenericRecordFn.apply(BigQueryConverters.java:548)
b.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRow(BigQueryUtils.java:580)
c.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRowFieldValue(BigQueryUtils.java:593)
d.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamValue(BigQueryUtils.java:641)
On c., a validation of null "bqValue" objects is performed; however, it appears
that some of the elements are of List type (for BigQuery Record type);
therefore, the List object is validated to be non-null, but not the elements of
the List.
On d., the method is executed recursively to process all the elements of the
List object; however, it seems that some objects are Null so this method is
throwing the NullPointerExceptions.
toBeamValue method should probably not used toBeamValue recursively but
toBeamRowFieldValue instead.
was:
The error trace has this path (I only added the relevant ones):
a.
com.google.cloud.teleport.v2.transforms.BigQueryConverters$TableRowToGenericRecordFn.apply(BigQueryConverters.java:548)
b.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRow(BigQueryUtils.java:580)
c.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRowFieldValue(BigQueryUtils.java:593)
d.
[org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamValue(BigQueryUtils.java:641)
On c., a validation of null "bqValue" objects is performed; however, it appears
that some of the elements are of List type (for BigQuery Record type);
therefore, the List object is validated to be non-null, but not the elements of
the List.
On d., the method is executed recursively to process all the elements of the
List object; however, it seems that some objects are Null so this method is
throwing the NullPointerExceptions.
toBeamValue method should probably not used toBeamValue recursively but
toBeamRowFieldValue instead.
> BigQueryUtils doesn't process List or Map with nullables
> --------------------------------------------------------
>
> Key: BEAM-12863
> URL: https://issues.apache.org/jira/browse/BEAM-12863
> Project: Beam
> Issue Type: Bug
> Components: io-java-gcp
> Affects Versions: 2.32.0
> Reporter: Thomas Pocreau
> Assignee: Matthew Ouyang
> Priority: P3
>
> When using PubsubAvroToBigQuery Dataflow template, [I noticed this issue.
> |https://github.com/GoogleCloudPlatform/DataflowTemplates/issues/287]
> This seems to be related to the way
> [toBeamRow|https://github.com/apache/beam/blob/v2.32.0/sdks/java/io/google-cloud-platform/src/main/java/org/apache/beam/sdk/io/gcp/bigquery/BigQueryUtils.java#L593]
> is handling TableRow with List containing null values.
> The error trace has this path (I only added the relevant ones):
> a.
> com.google.cloud.teleport.v2.transforms.BigQueryConverters$TableRowToGenericRecordFn.apply(BigQueryConverters.java:548)
>
> b.
> [org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRow(BigQueryUtils.java:580)
>
> c.
> [org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamRowFieldValue(BigQueryUtils.java:593)
> d.
> [org.apache.beam.sdk.io|https://org.apache.beam.sdk.io/].gcp.bigquery.BigQueryUtils.toBeamValue(BigQueryUtils.java:641)
> On c., a validation of null "bqValue" objects is performed; however, it
> appears that some of the elements are of List type (for BigQuery Record
> type); therefore, the List object is validated to be non-null, but not the
> elements of the List.
> On d., the method is executed recursively to process all the elements of the
> List object; however, it seems that some objects are Null so this method is
> throwing the NullPointerExceptions.
> toBeamValue method should probably not used toBeamValue recursively but
> toBeamRowFieldValue instead.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
