avro: handle empty projection and keep schema metadata normalized

Author: getChan

datafusion/datasource-avro/src/mod.rs

@@ -46,7 +46,9 @@ pub fn read_avro_schema_from_reader<R: Read>(
     // that are stricter than DataFusion's table schema handling needs for inferred schemas.
     // Drop metadata from inferred schemas so runtime batches and inferred table schemas
     // compare consistently without requiring strict Avro metadata identity.
-    Ok(strip_metadata_from_schema(avro_reader.schema().as_ref().clone()))
+    Ok(strip_metadata_from_schema(
+        avro_reader.schema().as_ref().clone(),
+    ))
 }
 
 fn strip_metadata_from_schema(schema: Schema) -> Schema {
@@ -86,10 +88,9 @@ fn strip_metadata_from_data_type(data_type: &DataType) -> DataType {
             Arc::new(strip_metadata_from_field(field.as_ref())),
             *size,
         ),
-        DataType::Map(field, sorted) => DataType::Map(
-            Arc::new(strip_metadata_from_field(field.as_ref())),
-            *sorted,
-        ),
+        DataType::Map(field, sorted) => {
+            DataType::Map(Arc::new(strip_metadata_from_field(field.as_ref())), *sorted)
+        }
         _ => data_type.clone(),
     }
 }

datafusion/datasource-avro/src/source.rs

@@ -54,11 +54,15 @@ impl AvroSource {
     }
 
     fn open<R: std::io::BufRead>(&self, reader: R) -> Result<Reader<R>> {
-        ReaderBuilder::new()
-            .with_batch_size(self.batch_size.expect("Batch size must set before open"))
-            .with_projection(self.projection.file_indices.clone())
-            .build(reader)
-            .map_err(Into::into)
+        let mut builder = ReaderBuilder::new()
+            .with_batch_size(self.batch_size.expect("Batch size must set before open"));
+
+        // Avoid pushing an empty projection into arrow-avro.
+        if !self.projection.file_indices.is_empty() {
+            builder = builder.with_projection(self.projection.file_indices.clone());
+        }
+
+        builder.build(reader).map_err(Into::into)
     }
 }