gene-bordegaray commented on code in PR #22657:
URL: https://github.com/apache/datafusion/pull/22657#discussion_r3401722649


##########
datafusion/catalog-listing/src/table.rs:
##########
@@ -561,24 +583,27 @@ impl TableProvider for ListingTable {
         };
 
         let file_source = self.create_file_source();
+        let mut scan_config_builder =
+            FileScanConfigBuilder::new(object_store_url, file_source)
+                .with_file_groups(partitioned_file_lists)
+                .with_constraints(self.constraints.clone())
+                .with_statistics(statistics)
+                .with_projection_indices(projection)?
+                .with_limit(limit)
+                .with_output_ordering(output_ordering)
+                .with_output_partitioning(declared_output_partitioning)
+                .with_expr_adapter(self.expr_adapter_factory.clone());
+        if partitioned_by_file_group {
+            scan_config_builder =
+                scan_config_builder.with_partitioned_by_file_group(true);
+        }
+        let scan_config = scan_config_builder.build();

Review Comment:
   Addressed. The builder path is now direct: scan planning computes the 
physical output_partitioning once and passes it through 
FileScanConfigBuilder::with_output_partitioning(...).



##########
datafusion/catalog-listing/src/table.rs:
##########
@@ -747,27 +805,26 @@ impl ListingTable {
         // hash repartitioning for aggregates and joins on partition columns.
         let threshold = 
ctx.config_options().optimizer.preserve_file_partitions;
 
-        let (file_groups, grouped_by_partition) = if threshold > 0
-            && !self.options.table_partition_cols.is_empty()
-        {
-            let grouped =
-                
file_group.group_by_partition_values(self.options.target_partitions);
+        let (mut file_groups, grouped_by_partition) = if 
preserve_partition_count {
+            (file_group.split_files(target_partitions), false)
+        } else if threshold > 0 && 
!self.options.table_partition_cols.is_empty() {
+            let grouped = 
file_group.group_by_partition_values(target_partitions);
             if grouped.len() >= threshold {
                 (grouped, true)
             } else {
                 let all_files: Vec<_> =
                     grouped.into_iter().flat_map(|g| g.into_inner()).collect();
                 (
-                    
FileGroup::new(all_files).split_files(self.options.target_partitions),
+                    FileGroup::new(all_files).split_files(target_partitions),
                     false,
                 )
             }
         } else {
-            (
-                file_group.split_files(self.options.target_partitions),
-                false,
-            )
+            (file_group.split_files(target_partitions), false)
         };
+        if preserve_partition_count && !file_groups.is_empty() {
+            file_groups.resize_with(target_partitions, || 
FileGroup::new(vec![]));
+        }
 

Review Comment:
   Addressed by removing the preserve_partition_count flag/path. The grouping 
behavior is now derived from whether output_partitioning is declared.



-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to