update API

apache · Jun 11, 2024 · ade44df · ade44df
1 parent ac38bc4
commit ade44df
Showing 1 changed file with 20 additions and 15 deletions.
diff --git a/datafusion-examples/examples/advanced_parquet_index.rs b/datafusion-examples/examples/advanced_parquet_index.rs
@@ -26,7 +26,7 @@ use async_trait::async_trait;
 use bytes::Bytes;
 use datafusion::datasource::listing::PartitionedFile;
 use datafusion::datasource::physical_plan::parquet::{
-    ParquetAccessPlan, RequestedStatistics, RowGroupAccess, StatisticsConverter,
+    ParquetAccessPlan, RowGroupAccess, StatisticsConverter,
 };
 use datafusion::datasource::physical_plan::{
     parquet::ParquetFileReaderFactory, FileMeta, FileScanConfig, ParquetExec,
@@ -712,20 +712,25 @@ impl ParquetRowGroupMetadataIndexBuilder {
         let num_row_groups = metadata.num_row_groups();
 
         // Extract the min/max values for each row group from the statistics
-        // TODO make an API that permits appending a row group at a time
-        let row_counts = StatisticsConverter::row_counts(metadata)?;
-        let value_column_mins =
-            StatisticsConverter::try_new("value", RequestedStatistics::Min, schema)?
-                .extract(metadata)?;
-        let value_column_maxes =
-            StatisticsConverter::try_new("value", RequestedStatistics::Max, schema)?
-                .extract(metadata)?;
-        let tag_column_mins =
-            StatisticsConverter::try_new("tag", RequestedStatistics::Min, schema)?
-                .extract(metadata)?;
-        let tag_column_maxes =
-            StatisticsConverter::try_new("tag", RequestedStatistics::Max, schema)?
-                .extract(metadata)?;
+        let row_group_meta = metadata.row_groups();
+
+        let row_counts = StatisticsConverter::row_group_row_counts(row_group_meta)?;
+
+        let value_converter = StatisticsConverter::try_new(
+            "value",
+            schema,
+            metadata.file_metadata().schema_descr(),
+        )?;
+        let value_column_mins = value_converter.row_group_mins(row_group_meta)?;
+        let value_column_maxes = value_converter.row_group_maxes(row_group_meta)?;
+
+        let tag_converter = StatisticsConverter::try_new(
+            "tag",
+            schema,
+            metadata.file_metadata().schema_descr(),
+        )?;
+        let tag_column_mins = tag_converter.row_group_mins(row_group_meta)?;
+        let tag_column_maxes = tag_converter.row_group_maxes(row_group_meta)?;
 
         // sanity check the statistics
         assert_eq!(row_counts.len(), num_row_groups);