[
https://issues.apache.org/jira/browse/DRILL-6199?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16399011#comment-16399011
]
ASF GitHub Bot commented on DRILL-6199:
---------------------------------------
Github user HanumathRao commented on a diff in the pull request:
https://github.com/apache/drill/pull/1152#discussion_r174558288
--- Diff:
exec/java-exec/src/main/java/org/apache/drill/exec/planner/logical/DrillFilterItemStarReWriterRule.java
---
@@ -54,83 +44,189 @@
import static
org.apache.drill.exec.planner.logical.FieldsReWriterUtil.FieldsReWriter;
/**
- * Rule will transform filter -> project -> scan call with item star
fields in filter
- * into project -> filter -> project -> scan where item star fields are
pushed into scan
- * and replaced with actual field references.
+ * Rule will transform item star fields in filter and replaced with actual
field references.
*
* This will help partition pruning and push down rules to detect fields
that can be pruned or push downed.
* Item star operator appears when sub-select or cte with star are used as
source.
*/
-public class DrillFilterItemStarReWriterRule extends RelOptRule {
+public class DrillFilterItemStarReWriterRule {
- public static final DrillFilterItemStarReWriterRule INSTANCE = new
DrillFilterItemStarReWriterRule(
- RelOptHelper.some(Filter.class, RelOptHelper.some(Project.class,
RelOptHelper.any( TableScan.class))),
- "DrillFilterItemStarReWriterRule");
+ public static final DrillFilterItemStarReWriterRule.ProjectOnScan
PROJECT_ON_SCAN = new ProjectOnScan(
+ RelOptHelper.some(DrillProjectRel.class,
RelOptHelper.any(DrillScanRel.class)),
+ "DrillFilterItemStarReWriterRule.ProjectOnScan");
- private DrillFilterItemStarReWriterRule(RelOptRuleOperand operand,
String id) {
- super(operand, id);
- }
+ public static final DrillFilterItemStarReWriterRule.FilterOnScan
FILTER_ON_SCAN = new FilterOnScan(
+ RelOptHelper.some(DrillFilterRel.class,
RelOptHelper.any(DrillScanRel.class)),
+ "DrillFilterItemStarReWriterRule.FilterOnScan");
- @Override
- public void onMatch(RelOptRuleCall call) {
- Filter filterRel = call.rel(0);
- Project projectRel = call.rel(1);
- TableScan scanRel = call.rel(2);
+ public static final DrillFilterItemStarReWriterRule.FilterOnProject
FILTER_ON_PROJECT = new FilterOnProject(
+ RelOptHelper.some(DrillFilterRel.class,
RelOptHelper.some(DrillProjectRel.class, RelOptHelper.any(DrillScanRel.class))),
+ "DrillFilterItemStarReWriterRule.FilterOnProject");
- ItemStarFieldsVisitor itemStarFieldsVisitor = new
ItemStarFieldsVisitor(filterRel.getRowType().getFieldNames());
- filterRel.getCondition().accept(itemStarFieldsVisitor);
- // there are no item fields, no need to proceed further
- if (!itemStarFieldsVisitor.hasItemStarFields()) {
- return;
+ private static class ProjectOnScan extends RelOptRule {
+
+ ProjectOnScan(RelOptRuleOperand operand, String id) {
+ super(operand, id);
}
- Map<String, DesiredField> itemStarFields =
itemStarFieldsVisitor.getItemStarFields();
+ @Override
+ public boolean matches(RelOptRuleCall call) {
+ DrillScanRel scan = call.rel(1);
+ return scan.getGroupScan() instanceof ParquetGroupScan &&
super.matches(call);
+ }
- // create new scan
- RelNode newScan = constructNewScan(scanRel, itemStarFields.keySet());
+ @Override
+ public void onMatch(RelOptRuleCall call) {
+ DrillProjectRel projectRel = call.rel(0);
+ DrillScanRel scanRel = call.rel(1);
+
+ ItemStarFieldsVisitor itemStarFieldsVisitor = new
ItemStarFieldsVisitor(scanRel.getRowType().getFieldNames());
+ List<RexNode> projects = projectRel.getProjects();
+ for (RexNode project : projects) {
+ project.accept(itemStarFieldsVisitor);
+ }
- // combine original and new projects
- List<RexNode> newProjects = new ArrayList<>(projectRel.getProjects());
+ Map<String, DesiredField> itemStarFields =
itemStarFieldsVisitor.getItemStarFields();
- // prepare node mapper to replace item star calls with new input field
references
- Map<RexNode, Integer> fieldMapper = new HashMap<>();
+ // if there are no item fields, no need to proceed further
+ if (itemStarFieldsVisitor.hasNoItemStarFields()) {
--- End diff --
Can this be moved to before getItemStarFields call?.
> Filter push down doesn't work with more than one nested subqueries
> ------------------------------------------------------------------
>
> Key: DRILL-6199
> URL: https://issues.apache.org/jira/browse/DRILL-6199
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.13.0
> Reporter: Anton Gozhiy
> Assignee: Arina Ielchiieva
> Priority: Major
> Fix For: 1.14.0
>
> Attachments: DRILL_6118_data_source.csv
>
>
> *Data set:*
> The data is generated used the attached file: *DRILL_6118_data_source.csv*
> Data gen commands:
> {code:sql}
> create table dfs.tmp.`DRILL_6118_parquet_partitioned_by_folders/d1` (c1, c2,
> c3, c4, c5) as select cast(columns[0] as int) c1, columns[1] c2, columns[2]
> c3, columns[3] c4, columns[4] c5 from dfs.tmp.`DRILL_6118_data_source.csv`
> where columns[0] in (1, 3);
> create table dfs.tmp.`DRILL_6118_parquet_partitioned_by_folders/d2` (c1, c2,
> c3, c4, c5) as select cast(columns[0] as int) c1, columns[1] c2, columns[2]
> c3, columns[3] c4, columns[4] c5 from dfs.tmp.`DRILL_6118_data_source.csv`
> where columns[0]=2;
> create table dfs.tmp.`DRILL_6118_parquet_partitioned_by_folders/d3` (c1, c2,
> c3, c4, c5) as select cast(columns[0] as int) c1, columns[1] c2, columns[2]
> c3, columns[3] c4, columns[4] c5 from dfs.tmp.`DRILL_6118_data_source.csv`
> where columns[0]>3;
> {code}
> *Steps:*
> # Execute the following query:
> {code:sql}
> explain plan for select * from (select * from (select * from
> dfs.tmp.`DRILL_6118_parquet_partitioned_by_folders`)) where c1<3
> {code}
> *Expected result:*
> numFiles=2, numRowGroups=2, only files from the folders d1 and d2 should be
> scanned.
> *Actual result:*
> Filter push down doesn't work:
> numFiles=3, numRowGroups=3, scanning from all files
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
