chore: Use Builder API to create FileScanConfig

Signed-off-by: Andrew Lamb <[email protected]>

Author: alamb

crates/core/src/delta_datafusion/mod.rs

@@ -648,25 +648,24 @@ impl<'a> DeltaScanBuilder<'a> {
             ..Default::default()
         };
 
-        let mut exec_plan_builder = ParquetExecBuilder::new(FileScanConfig {
-            object_store_url: self.log_store.object_store_url(),
-            file_schema,
-            // If all files were filtered out, we still need to emit at least one partition to
-            // pass datafusion sanity checks.
-            //
-            // See https://github.com/apache/datafusion/issues/11322
-            file_groups: if file_groups.is_empty() {
-                vec![vec![]]
-            } else {
-                file_groups.into_values().collect()
-            },
-            constraints: Constraints::default(),
-            statistics: stats,
-            projection: self.projection.cloned(),
-            limit: self.limit,
-            table_partition_cols,
-            output_ordering: vec![],
-        })
+        let mut exec_plan_builder = ParquetExecBuilder::new(
+            FileScanConfig::new(self.log_store.object_store_url(), file_schema)
+                .with_file_groups(
+                    // If all files were filtered out, we still need to emit at least one partition to
+                    // pass datafusion sanity checks.
+                    //
+                    // See https://github.com/apache/datafusion/issues/11322
+                    if file_groups.is_empty() {
+                        vec![vec![]]
+                    } else {
+                        file_groups.into_values().collect()
+                    },
+                )
+                .with_statistics(stats)
+                .with_projection(self.projection.cloned())
+                .with_limit(self.limit)
+                .with_table_partition_cols(table_partition_cols),
+        )
         .with_schema_adapter_factory(Arc::new(DeltaSchemaAdapterFactory {}))
         .with_table_parquet_options(parquet_options);
 

crates/core/src/operations/load_cdf.rs

@@ -21,7 +21,7 @@ use datafusion::datasource::file_format::FileFormat;
 use datafusion::datasource::physical_plan::FileScanConfig;
 use datafusion::execution::SessionState;
 use datafusion::prelude::SessionContext;
-use datafusion_common::{Constraints, ScalarValue, Statistics};
+use datafusion_common::ScalarValue;
 use datafusion_physical_expr::{expressions, PhysicalExpr};
 use datafusion_physical_plan::projection::ProjectionExec;
 use datafusion_physical_plan::union::UnionExec;
@@ -378,53 +378,32 @@ impl CdfLoadBuilder {
         let cdc_scan = ParquetFormat::new()
             .create_physical_plan(
                 session_sate,
-                FileScanConfig {
-                    object_store_url: self.log_store.object_store_url(),
-                    file_schema: cdc_file_schema.clone(),
-                    file_groups: cdc_file_groups.into_values().collect(),
-                    constraints: Constraints::default(),
-                    statistics: Statistics::new_unknown(&cdc_file_schema),
-                    projection: None,
-                    limit: None,
-                    table_partition_cols: cdc_partition_cols,
-                    output_ordering: vec![],
-                },
+                FileScanConfig::new(self.log_store.object_store_url(), cdc_file_schema)
+                    .with_file_groups(cdc_file_groups.into_values().collect())
+                    .with_table_partition_cols(cdc_partition_cols),
                 filters,
             )
             .await?;
 
         let add_scan = ParquetFormat::new()
             .create_physical_plan(
                 session_sate,
-                FileScanConfig {
-                    object_store_url: self.log_store.object_store_url(),
-                    file_schema: add_remove_file_schema.clone(),
-                    file_groups: add_file_groups.into_values().collect(),
-                    constraints: Constraints::default(),
-                    statistics: Statistics::new_unknown(&add_remove_file_schema.clone()),
-                    projection: None,
-                    limit: None,
-                    table_partition_cols: add_remove_partition_cols.clone(),
-                    output_ordering: vec![],
-                },
+                FileScanConfig::new(
+                    self.log_store.object_store_url(),
+                    add_remove_file_schema.clone(),
+                )
+                .with_file_groups(add_file_groups.into_values().collect())
+                .with_table_partition_cols(add_remove_partition_cols.clone()),
                 filters,
             )
             .await?;
 
         let remove_scan = ParquetFormat::new()
             .create_physical_plan(
                 session_sate,
-                FileScanConfig {
-                    object_store_url: self.log_store.object_store_url(),
-                    file_schema: add_remove_file_schema.clone(),
-                    file_groups: remove_file_groups.into_values().collect(),
-                    constraints: Constraints::default(),
-                    statistics: Statistics::new_unknown(&add_remove_file_schema),
-                    projection: None,
-                    limit: None,
-                    table_partition_cols: add_remove_partition_cols,
-                    output_ordering: vec![],
-                },
+                FileScanConfig::new(self.log_store.object_store_url(), add_remove_file_schema)
+                    .with_file_groups(remove_file_groups.into_values().collect())
+                    .with_table_partition_cols(add_remove_partition_cols),
                 filters,
             )
             .await?;
@@ -435,7 +414,7 @@ impl CdfLoadBuilder {
             Arc::new(UnionExec::new(vec![cdc_scan, add_scan, remove_scan]));
 
         // We project the union in the order of the input_schema + cdc cols at the end
-        // This is to ensure the DeltaCdfTableProvider uses the correct schema consturction.
+        // This is to ensure the DeltaCdfTableProvider uses the correct schema construction.
         let mut fields = schema.fields().to_vec();
         for f in ADD_PARTITION_SCHEMA.clone() {
             fields.push(f.into());