PARQUET-136: In case of all nulls in a binary column, statistics object read from file metadata is empty, and should return true for all nulls check for the column

Yash Datta · Yash Datta · commit 974a22b79829 · 2015-01-07T18:27:30.000+05:30
diff --git a/parquet-hadoop/src/main/java/parquet/filter2/statisticslevel/StatisticsFilter.java b/parquet-hadoop/src/main/java/parquet/filter2/statisticslevel/StatisticsFilter.java
@@ -67,8 +67,9 @@ private ColumnChunkMetaData getColumnChunk(ColumnPath columnPath) {
   }
 
   // is this column chunk composed entirely of nulls?
+  // in case of all nulls, the stats object read from file metadata is empty
   private boolean isAllNulls(ColumnChunkMetaData column) {
-    return column.getStatistics().getNumNulls() == column.getValueCount();
+    return (column.getStatistics().isEmpty()) || (column.getStatistics().getNumNulls() == column.getValueCount());
   }
 
   // are there any nulls in this column chunk?

Original file line number	Diff line number	Diff line change
`@@ -67,8 +67,9 @@ private ColumnChunkMetaData getColumnChunk(ColumnPath columnPath) {`
`67`	`67`	`}`
`68`	`68`
`69`	`69`	`// is this column chunk composed entirely of nulls?`
	`70`	`+ // in case of all nulls, the stats object read from file metadata is empty`
`70`	`71`	`private boolean isAllNulls(ColumnChunkMetaData column) {`
`71`		`- return column.getStatistics().getNumNulls() == column.getValueCount();`
	`72`	`+ return (column.getStatistics().isEmpty()) \|\| (column.getStatistics().getNumNulls() == column.getValueCount());`
`72`	`73`	`}`
`73`	`74`
`74`	`75`	`// are there any nulls in this column chunk?`