clean

Author: alamb

datafusion/datasource-parquet/src/opener.rs

@@ -935,7 +935,8 @@ impl RowGroupsPrunedParquetOpen {
                 .collect();
 
             for idx in self.row_groups.row_group_indexes() {
-                let mut column_sbbf = HashMap::with_capacity(parquet_columns.len());
+                let mut row_group_filters =
+                    BloomFilterStatistics::with_capacity(parquet_columns.len());
                 for (column_name, column_idx, physical_type) in &parquet_columns {
                     let bf: Sbbf = match builder
                         .get_row_group_column_bloom_filter(idx, *column_idx)
@@ -949,10 +950,9 @@ impl RowGroupsPrunedParquetOpen {
                             continue;
                         }
                     };
-                    column_sbbf.insert(column_name.clone(), (bf, *physical_type));
+                    row_group_filters.insert(column_name, bf, *physical_type);
                 }
-                row_group_bloom_filters[idx] =
-                    BloomFilterStatistics::from_column_sbbf(column_sbbf);
+                row_group_bloom_filters[idx] = row_group_filters;
             }
         }
 

datafusion/datasource-parquet/src/row_group_filter.rs

@@ -434,10 +434,16 @@ impl BloomFilterStatistics {
         Default::default()
     }
 
-    /// Create [`BloomFilterStatistics`] for a single Parquet row group from the
-    /// loaded bloom filters keyed by predicate column name.
-    pub(crate) fn from_column_sbbf(column_sbbf: HashMap<String, (Sbbf, Type)>) -> Self {
-        Self { column_sbbf }
+    /// Create an empty [`BloomFilterStatistics`] with the specified capacity
+    pub(crate) fn with_capacity(capacity: usize) -> Self {
+        Self {
+            column_sbbf: HashMap::with_capacity(capacity),
+        }
+    }
+
+    /// Add a Bloom filter and type for the specified column
+    pub(crate) fn insert(&mut self, column: impl Into<String>, sbbf: Sbbf, ty: Type) {
+        self.column_sbbf.insert(column.into(), (sbbf, ty));
     }
 
     /// Helper function for checking if [`Sbbf`] filter contains [`ScalarValue`].