sunchao commented on a change in pull request #32753:
URL: https://github.com/apache/spark/pull/32753#discussion_r658990194
##########
File path:
sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/ParquetReadState.java
##########
@@ -17,13 +17,31 @@
package org.apache.spark.sql.execution.datasources.parquet;
+import java.util.ArrayList;
+import java.util.Iterator;
+import java.util.List;
+import java.util.PrimitiveIterator;
+
/**
* Helper class to store intermediate state while reading a Parquet column
chunk.
*/
final class ParquetReadState {
- /** Maximum definition level */
+ private static final RowRange MAX_ROW_RANGE = new RowRange(Long.MIN_VALUE,
Long.MAX_VALUE);
+ private static final RowRange MIN_ROW_RANGE = new RowRange(Long.MAX_VALUE,
Long.MIN_VALUE);
+
+ /** Iterator over all row ranges, only not-null if column index is present */
+ private final Iterator<RowRange> rowRanges;
Review comment:
No. The list of row ranges is associated with a Parquet row group. For
example, let's say you have two columns `c1:int` and `c2:bigint`, and the
following pages:
```
row index 0 500 1000 1500
-------------------------------
c1 (int) | | | |
-------------------------------
c2 (bigint) | | | | | | |
-------------------------------
0 250 500 750 1000 1250 1500
```
Suppose the query is `SELECT * FROM tbl WHERE c1 = 750 AND c2 = 1100`
This, when applied on `c1`, will produce row ranges `[500, 1000)`. When
applied on `c2`, will produce row ranges `[1000, 1250)`. These two will be
unioned into `[500, 1250)` and that is the row ranges for the whole row group.
--
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]
