Create ListingTableConfig which includes file format and schema inference

ballista/rust/core/src/serde/logical_plan/from_proto.rs

@@ -27,7 +27,7 @@ use datafusion::datasource::file_format::avro::AvroFormat;
 use datafusion::datasource::file_format::csv::CsvFormat;
 use datafusion::datasource::file_format::parquet::ParquetFormat;
 use datafusion::datasource::file_format::FileFormat;
-use datafusion::datasource::listing::{ListingOptions, ListingTable};
+use datafusion::datasource::listing::{ListingOptions, ListingTable, ListingTableConfig};
 use datafusion::datasource::object_store::local::LocalFileSystem;
 use datafusion::datasource::object_store::{FileMeta, SizedFile};
 use datafusion::logical_plan::window_frames::{

ballista/rust/core/src/serde/logical_plan/mod.rs

@@ -27,7 +27,7 @@ use datafusion::datasource::file_format::avro::AvroFormat;
 use datafusion::datasource::file_format::csv::CsvFormat;
 use datafusion::datasource::file_format::parquet::ParquetFormat;
 use datafusion::datasource::file_format::FileFormat;
-use datafusion::datasource::listing::{ListingOptions, ListingTable};
+use datafusion::datasource::listing::{ListingOptions, ListingTable, ListingTableConfig};
 use datafusion::datasource::object_store::local::LocalFileSystem;
 use datafusion::logical_plan::plan::Extension;
 use datafusion::logical_plan::plan::{
@@ -241,12 +241,11 @@ impl AsLogicalPlan for LogicalPlanNode {
                     scan.path.as_str()
                 );
 
-                let provider = ListingTable::new(
-                    object_store,
-                    scan.path.clone(),
-                    Arc::new(schema),
-                    options,
-                );
+                let config = ListingTableConfig::new(object_store, scan.path.as_str())
+                    .with_listing_options(options)
+                    .with_schema(Arc::new(schema));
+
+                let provider = ListingTable::try_new(config)?;
 
                 LogicalPlanBuilder::scan_with_filters(
                     &scan.table_name,

benchmarks/src/bin/tpch.rs

@@ -51,7 +51,7 @@ use datafusion::{
 use datafusion::{
     arrow::util::pretty,
     datasource::{
-        listing::{ListingOptions, ListingTable},
+        listing::{ListingOptions, ListingTable, ListingTableConfig},
         object_store::local::LocalFileSystem,
     },
 };
@@ -724,12 +724,11 @@ fn get_table(
         table_partition_cols: vec![],
     };
 
-    Ok(Arc::new(ListingTable::new(
-        Arc::new(LocalFileSystem {}),
-        path,
-        schema,
-        options,
-    )))
+    let config = ListingTableConfig::new(Arc::new(LocalFileSystem {}), path)
+        .with_listing_options(options)
+        .with_schema(schema);
+
+    Ok(Arc::new(ListingTable::try_new(config)?))
 }
 
 fn get_schema(table: &str) -> Schema {
@@ -1389,12 +1388,13 @@ mod tests {
                     .has_header(false)
                     .file_extension(".tbl");
                 let listing_options = options.to_listing_options(1);
-                let provider = ListingTable::new(
+                let config = ListingTableConfig::new(
                     Arc::new(LocalFileSystem {}),
-                    format!("{}/{}.tbl", tpch_data_path, table),
-                    Arc::new(schema),
-                    listing_options,
-                );
+                    tpch_data_path.clone(),
+                )
+                .with_listing_options(listing_options)
+                .with_schema(Arc::new(schema));
+                let provider = ListingTable::try_new(config)?;
                 ctx.register_table(table, Arc::new(provider))?;
             }
 

datafusion/benches/sort_limit_query_sql.rs

@@ -19,7 +19,7 @@
 extern crate criterion;
 use criterion::Criterion;
 use datafusion::datasource::file_format::csv::CsvFormat;
-use datafusion::datasource::listing::{ListingOptions, ListingTable};
+use datafusion::datasource::listing::{ListingOptions, ListingTable, ListingTableConfig};
 use datafusion::datasource::object_store::local::LocalFileSystem;
 
 use parking_lot::Mutex;
@@ -63,14 +63,16 @@ fn create_context() -> Arc<Mutex<ExecutionContext>> {
 
     let testdata = datafusion::test_util::arrow_test_data();
 
+    let path = format!("{}/csv/aggregate_test_100.csv", testdata);
+
     // create CSV data source
     let listing_options = ListingOptions::new(Arc::new(CsvFormat::default()));
-    let csv = ListingTable::new(
-        Arc::new(LocalFileSystem {}),
-        format!("{}/csv/aggregate_test_100.csv", testdata),
-        schema,
-        listing_options,
-    );
+
+    let config = ListingTableConfig::new(Arc::new(LocalFileSystem {}), &path)
+        .with_listing_options(listing_options)
+        .with_schema(schema);
+
+    let csv = async { ListingTable::try_new(config).unwrap() };
 
     let rt = Runtime::new().unwrap();
 
@@ -85,7 +87,7 @@ fn create_context() -> Arc<Mutex<ExecutionContext>> {
         ctx.state.lock().config.target_partitions = 1;
         let runtime = ctx.state.lock().runtime_env.clone();
 
-        let mem_table = MemTable::load(Arc::new(csv), Some(partitions), runtime)
+        let mem_table = MemTable::load(Arc::new(csv.await), Some(partitions), runtime)
             .await
             .unwrap();
         ctx.register_table("aggregate_test_100", Arc::new(mem_table))

datafusion/src/datasource/listing/mod.rs

@@ -21,4 +21,4 @@
 mod helpers;
 mod table;
 
-pub use table::{ListingOptions, ListingTable};
+pub use table::{ListingOptions, ListingTable, ListingTableConfig};