You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2022/07/07 06:20:55 UTC

[GitHub] [flink-table-store] JingsongLi commented on a diff in pull request #202: [FLINK-28110] Support projection pushdown for Hive readers

JingsongLi commented on code in PR #202:
URL: https://github.com/apache/flink-table-store/pull/202#discussion_r915498253


##########
flink-table-store-hive/flink-table-store-hive-connector/src/main/java/org/apache/flink/table/store/mapred/TableStoreRecordReader.java:
##########
@@ -36,24 +41,44 @@ public class TableStoreRecordReader implements RecordReader<Void, RowDataContain
     private final RecordReaderIterator<RowData> iterator;
     private final long splitLength;
 
+    @Nullable private final ProjectedRowData reusedProjectedRow;
+
     private float progress;
 
     public TableStoreRecordReader(
-            org.apache.flink.table.store.file.utils.RecordReader<RowData> wrapped,
-            long splitLength) {
-        this.iterator = new RecordReaderIterator<>(wrapped);
-        this.splitLength = splitLength;
+            TableRead read,
+            TableStoreInputSplit split,
+            List<String> columnNames,
+            List<String> selectedColumns)

Review Comment:
   Add documentation here: Still output the original schema of the Row, the role of `selectedColumns` is to read out the required fields, other fields to fill Nulls.



##########
flink-table-store-hive/flink-table-store-hive-connector/src/main/java/org/apache/flink/table/store/mapred/TableStoreRecordReader.java:
##########
@@ -36,24 +41,44 @@ public class TableStoreRecordReader implements RecordReader<Void, RowDataContain
     private final RecordReaderIterator<RowData> iterator;
     private final long splitLength;
 
+    @Nullable private final ProjectedRowData reusedProjectedRow;
+
     private float progress;
 
     public TableStoreRecordReader(
-            org.apache.flink.table.store.file.utils.RecordReader<RowData> wrapped,
-            long splitLength) {
-        this.iterator = new RecordReaderIterator<>(wrapped);
-        this.splitLength = splitLength;
+            TableRead read,
+            TableStoreInputSplit split,
+            List<String> columnNames,
+            List<String> selectedColumns)
+            throws IOException {
+        if (columnNames.equals(selectedColumns)) {
+            reusedProjectedRow = null;
+        } else {
+            read.withProjection(selectedColumns.stream().mapToInt(columnNames::indexOf).toArray());
+            reusedProjectedRow =
+                    ProjectedRowData.from(
+                            columnNames.stream().mapToInt(selectedColumns::indexOf).toArray());
+        }
+
+        this.iterator = new RecordReaderIterator<>(read.createReader(split.split()));
+        this.splitLength = split.getLength();
         this.progress = 0;
     }
 
     @Override
     public boolean next(Void key, RowDataContainer value) throws IOException {
         RowData rowData = iterator.next();
+
         if (rowData == null) {
             progress = 1;
             return false;
         } else {
-            value.set(rowData);
+            if (reusedProjectedRow != null) {
+                reusedProjectedRow.replaceRow(rowData);
+                value.set(reusedProjectedRow);

Review Comment:
   `value.set(reusedProjectedRow.replaceRow(rowData));`



-- 
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: issues-unsubscribe@flink.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org