Github user ajantha-bhat commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/1995#discussion_r170412437
--- Diff:
hadoop/src/main/java/org/apache/carbondata/hadoop/api/CarbonTableInputFormat.java
---
@@ -353,7 +353,53 @@ public AbsoluteTableIdentifier
getAbsoluteTableIdentifier(Configuration configur
List<String> validSegments = segments.getValidSegments();
streamSegments = segments.getStreamSegments();
if (validSegments.size() == 0) {
- return getSplitsOfStreaming(job, identifier, streamSegments);
+ if (streamSegments.size() != 0) {
+ return getSplitsOfStreaming(job, identifier, streamSegments);
+ }
+ // check for externalTable segment (Segment_null)
+ {
+ // process and resolve the expression
+ Expression filter = getFilterPredicates(job.getConfiguration());
+ TableProvider tableProvider = new
SingleTableProvider(carbonTable);
+ // this will be null in case of corrupt schema file.
+ PartitionInfo partitionInfo =
carbonTable.getPartitionInfo(carbonTable.getTableName());
+ CarbonInputFormatUtil.processFilterExpression(filter,
carbonTable, null, null);
+
+ // prune partitions for filter query on partition table
--- End diff --
This code changes are not required. Because we already handle this logic in
**CarbonFileInputFormat**
This logic is to get the splits file level. But this is a table level
reader.
There should not be any change in this class for this Requirement.
---
