[GitHub] [iceberg] rdblue commented on a change in pull request #2358: Core: Add predicate push down for Partitions table

[GitBox <gi...@apache.org>]

rdblue commented on a change in pull request #2358:
URL: https://github.com/apache/iceberg/pull/2358#discussion_r635674026



##########
File path: core/src/main/java/org/apache/iceberg/PartitionsTable.java
##########
@@ -98,21 +105,46 @@ private DataTask task(TableScan scan) {
     return StaticDataTask.Row.of(partition.key, partition.recordCount, partition.fileCount);
   }
 
-  private static Iterable<Partition> partitions(Table table, Long snapshotId) {
+  private Iterable<Partition> partitions(StaticTableScan scan) {
+    CloseableIterable<FileScanTask> tasks = planTasks(scan);
+
     PartitionMap partitions = new PartitionMap(table.spec().partitionType());
-    TableScan scan = table.newScan();
+    for (FileScanTask task : tasks) {
+      partitions.get(task.file().partition()).update(task.file());
+    }
+    return partitions.all();
+  }
 
-    if (snapshotId != null) {
-      scan = scan.useSnapshot(snapshotId);
+  @VisibleForTesting
+  CloseableIterable<FileScanTask> planTasks(StaticTableScan scan) {
+    boolean caseSensitive = scan.isCaseSensitive();
+    boolean ignoreResiduals = scan.shouldIgnoreResiduals();
+    long snapshotId = scan.snapshot().snapshotId();
+
+    ManifestEvaluator eval = ManifestEvaluator.forPartitionFilter(
+        Projections.metadata(table.spec(), caseSensitive).project(scan.filter()),

Review comment:
       @szehon-ho, what I'm suggesting is that we can use the existing projections to do this rewrite work instead of creating a new projection. Using an existing projection with a different partition spec would be easier because you'd get the benefit you're talking about without adding a new form of projection that doesn't quite meet the definition of a projection.




-- 
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org