cleaner api

Author: alamb

datafusion-examples/examples/csv_opener.rs

@@ -17,7 +17,6 @@
 
 use std::{sync::Arc, vec};
 
-use datafusion::common::Statistics;
 use datafusion::{
     assert_batches_eq,
     datasource::{
@@ -58,11 +57,11 @@ async fn main() -> Result<()> {
 
     let path = std::path::Path::new(&path).canonicalize()?;
 
-    let mut scan_config =
+    let scan_config =
         FileScanConfig::new(ObjectStoreUrl::local_filesystem(), schema.clone())
-            .with_projection(vec![12, 0])
-            .with_limit(Some(5));
-    scan_config.add_file(PartitionedFile::new(path.display().to_string(), 10));
+            .with_projection(Some(vec![12, 0]))
+            .with_limit(Some(5))
+            .with_file(PartitionedFile::new(path.display().to_string(), 10));
 
     let result =
         FileStream::new(&scan_config, 0, opener, &ExecutionPlanMetricsSet::new())

datafusion-examples/examples/json_opener.rs

@@ -29,7 +29,6 @@ use datafusion::{
     error::Result,
     physical_plan::metrics::ExecutionPlanMetricsSet,
 };
-use datafusion_common::Statistics;
 
 use futures::StreamExt;
 use object_store::ObjectStore;
@@ -61,12 +60,11 @@ async fn main() -> Result<()> {
         Arc::new(object_store),
     );
 
-    let mut scan_config =
+    let scan_config =
         FileScanConfig::new(ObjectStoreUrl::local_filesystem(), schema.clone())
             .with_projection(Some(vec![1, 0]))
-            .with_limit(5);
-
-    scan_config.add_file(PartitionedFile::new(path.to_string(), 10));
+            .with_limit(Some(5))
+            .with_file(PartitionedFile::new(path.to_string(), 10));
 
     let result =
         FileStream::new(&scan_config, 0, opener, &ExecutionPlanMetricsSet::new())

datafusion/core/src/datasource/physical_plan/file_scan_config.rs

@@ -187,6 +187,22 @@ impl FileScanConfig {
         self
     }
 
+    /// Add a file as a single group
+    ///
+    /// See [Self::file_groups] for more information.
+    pub fn with_file(mut self, file: PartitionedFile) -> Self {
+        self.add_file(file);
+        self
+    }
+
+    /// Add a new file group
+    ///
+    /// See [Self::file_groups] for more information
+    pub fn with_file_group(mut self, file_group: Vec<PartitionedFile>) -> Self {
+        self.add_file_group(file_group);
+        self
+    }
+
     /// Set the partitioning columns of the files
     pub fn with_table_partition_cols(mut self, table_partition_cols: Vec<Field>) -> Self {
         self.table_partition_cols = table_partition_cols;