[
https://issues.apache.org/jira/browse/DRILL-7313?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16878815#comment-16878815
]
ASF GitHub Bot commented on DRILL-7313:
---------------------------------------
vvysotskyi commented on pull request #1819: DRILL-7313: Use Hive schema for
MaprDB native reader when field was empty
URL: https://github.com/apache/drill/pull/1819#discussion_r299908023
##########
File path:
exec/java-exec/src/main/java/org/apache/drill/exec/vector/complex/fn/JsonReaderUtils.java
##########
@@ -33,25 +38,48 @@ public static void
ensureAtLeastOneField(BaseWriter.ComplexWriter writer,
boolean allTextMode,
List<BaseWriter.ListWriter>
emptyArrayWriters) {
- List<BaseWriter.MapWriter> writerList = Lists.newArrayList();
- List<PathSegment> fieldPathList = Lists.newArrayList();
+ ensureAtLeastOneField(writer, columns, null /* schema */, allTextMode,
emptyArrayWriters);
+ }
+
+ public static void ensureAtLeastOneField(BaseWriter.ComplexWriter writer,
+ Collection<SchemaPath> columns,
+ TupleMetadata schema,
+ boolean allTextMode,
+ List<BaseWriter.ListWriter>
emptyArrayWriters) {
+
+ List<BaseWriter.MapWriter> writerList = new ArrayList<>();
+ List<PathSegment> fieldPathList = new ArrayList<>();
+ List<TypeProtos.MajorType> types = new ArrayList<>();
BitSet emptyStatus = new BitSet(columns.size());
- int i = 0;
+ int fieldIndex = 0;
// first pass: collect which fields are empty
- for (SchemaPath sp : columns) {
- PathSegment fieldPath = sp.getRootSegment();
+ for (SchemaPath schemaPath : columns) {
+ PathSegment fieldPath = schemaPath.getRootSegment();
BaseWriter.MapWriter fieldWriter = writer.rootAsMap();
+ TupleMetadata columnMetadata = schema;
while (fieldPath.getChild() != null && !fieldPath.getChild().isArray()) {
- fieldWriter = fieldWriter.map(fieldPath.getNameSegment().getPath());
+ String name = fieldPath.getNameSegment().getPath();
+ if (columnMetadata != null) {
+ ColumnMetadata metadata = columnMetadata.metadata(name);
+ columnMetadata = metadata != null ? metadata.mapSchema() : null;
+ }
+ fieldWriter = fieldWriter.map(name);
fieldPath = fieldPath.getChild();
}
writerList.add(fieldWriter);
fieldPathList.add(fieldPath);
+ // for the case when field is absent in the schema, use INT type
+ TypeProtos.MajorType majorType =
Types.optional(TypeProtos.MinorType.INT);
Review comment:
Good idea, I have added it in the new commit.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> Use Hive schema for MaprDB native reader when field was empty
> -------------------------------------------------------------
>
> Key: DRILL-7313
> URL: https://issues.apache.org/jira/browse/DRILL-7313
> Project: Apache Drill
> Issue Type: Task
> Affects Versions: 1.16.0
> Reporter: Volodymyr Vysotskyi
> Assignee: Volodymyr Vysotskyi
> Priority: Major
> Fix For: 1.17.0
>
>
> Currently, when an external Hive MaprDB table is queried using hive plugin
> with enabled {{store.hive.maprdb_json.optimize_scan_with_native_reader}},
> some queries may fail due to soft schema change, though Hive knows actual
> data types.
> For example, when we have a table with several fields, and one of them has
> only several non-null values, queries with grouping by such field will fail
> due to schema change.
> The goal of this Jira is to allow using types from Hive when a non-existing
> field is created, so it will allow avoiding such issues.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
