Merge commit 'f8c623fe045d70a87eac8dc8620b74ff73be56d5' into chunchun…

…/update-df-apr-week-4-3

datafusion/core/src/physical_optimizer/join_selection.rs

@@ -882,28 +882,6 @@ mod tests_statistical {
         (big, medium, small)
     }
 
-    pub(crate) fn crosscheck_plans(plan: Arc<dyn ExecutionPlan>) -> Result<()> {
-        let subrules: Vec<Box<PipelineFixerSubrule>> = vec![
-            Box::new(hash_join_convert_symmetric_subrule),
-            Box::new(hash_join_swap_subrule),
-        ];
-        let new_plan = plan
-            .transform_up(|p| apply_subrules(p, &subrules, &ConfigOptions::new()))
-            .data()?;
-        // TODO: End state payloads will be checked here.
-        let config = ConfigOptions::new().optimizer;
-        let collect_left_threshold = config.hash_join_single_partition_threshold;
-        let collect_threshold_num_rows = config.hash_join_single_partition_threshold_rows;
-        let _ = new_plan.transform_up(|plan| {
-            statistical_join_selection_subrule(
-                plan,
-                collect_left_threshold,
-                collect_threshold_num_rows,
-            )
-        })?;
-        Ok(())
-    }
-
     #[tokio::test]
     async fn test_join_with_swap() {
         let (big, small) = create_big_and_small();
@@ -958,7 +936,6 @@ mod tests_statistical {
             swapped_join.right().statistics().unwrap().total_byte_size,
             Precision::Inexact(2097152)
         );
-        crosscheck_plans(join.clone()).unwrap();
     }
 
     #[tokio::test]
@@ -1001,7 +978,6 @@ mod tests_statistical {
             swapped_join.right().statistics().unwrap().total_byte_size,
             Precision::Inexact(2097152)
         );
-        crosscheck_plans(join.clone()).unwrap();
     }
 
     #[tokio::test]
@@ -1055,7 +1031,6 @@ mod tests_statistical {
                 Precision::Inexact(2097152)
             );
             assert_eq!(original_schema, swapped_join.schema());
-            crosscheck_plans(join).unwrap();
         }
     }
 
@@ -1078,7 +1053,6 @@ mod tests_statistical {
                 "\n\nexpected:\n\n{:#?}\nactual:\n\n{:#?}\n\n",
                 expected_lines, actual_lines
             );
-            crosscheck_plans(plan).unwrap();
         };
     }
 
@@ -1180,7 +1154,6 @@ mod tests_statistical {
             swapped_join.right().statistics().unwrap().total_byte_size,
             Precision::Inexact(2097152)
         );
-        crosscheck_plans(join).unwrap();
     }
 
     #[rstest(
@@ -1249,7 +1222,6 @@ mod tests_statistical {
             swapped_join.right().statistics().unwrap().total_byte_size,
             Precision::Inexact(2097152)
         );
-        crosscheck_plans(join.clone()).unwrap();
     }
 
     #[rstest(
@@ -1311,7 +1283,6 @@ mod tests_statistical {
             swapped_join.right().statistics().unwrap().total_byte_size,
             Precision::Inexact(2097152)
         );
-        crosscheck_plans(join.clone()).unwrap();
     }
 
     #[tokio::test]
@@ -1523,7 +1494,6 @@ mod tests_statistical {
 
             assert_eq!(*swapped_join.partition_mode(), expected_mode);
         }
-        crosscheck_plans(join).unwrap();
     }
 }
 
@@ -1568,8 +1538,6 @@ mod util_tests {
 
 #[cfg(test)]
 mod hash_join_tests {
-
-    use self::tests_statistical::crosscheck_plans;
     use super::*;
     use crate::physical_optimizer::test_utils::SourceType;
     use crate::test_util::UnboundedExec;
@@ -2000,7 +1968,6 @@ mod hash_join_tests {
                 )
             );
         };
-        crosscheck_plans(plan).unwrap();
         Ok(())
     }
 }

datafusion/expr/src/built_in_window_function.rs

@@ -71,7 +71,7 @@ pub enum BuiltInWindowFunction {
 }
 
 impl BuiltInWindowFunction {
-    fn name(&self) -> &str {
+    pub fn name(&self) -> &str {
         use BuiltInWindowFunction::*;
         match self {
             RowNumber => "ROW_NUMBER",

datafusion/expr/src/expr.rs

@@ -669,6 +669,16 @@ impl WindowFunctionDefinition {
             WindowFunctionDefinition::WindowUDF(fun) => fun.signature().clone(),
         }
     }
+
+    /// Function's name for display
+    pub fn name(&self) -> &str {
+        match self {
+            WindowFunctionDefinition::BuiltInWindowFunction(fun) => fun.name(),
+            WindowFunctionDefinition::WindowUDF(fun) => fun.name(),
+            WindowFunctionDefinition::AggregateFunction(fun) => fun.name(),
+            WindowFunctionDefinition::AggregateUDF(fun) => fun.name(),
+        }
+    }
 }
 
 impl fmt::Display for WindowFunctionDefinition {

datafusion/functions/src/string/ascii.rs

@@ -47,6 +47,13 @@ pub fn ascii<T: OffsetSizeTrait>(args: &[ArrayRef]) -> Result<ArrayRef> {
 pub struct AsciiFunc {
     signature: Signature,
 }
+
+impl Default for AsciiFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl AsciiFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/bit_length.rs

@@ -31,6 +31,12 @@ pub struct BitLengthFunc {
     signature: Signature,
 }
 
+impl Default for BitLengthFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl BitLengthFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/btrim.rs

@@ -41,6 +41,12 @@ pub struct BTrimFunc {
     aliases: Vec<String>,
 }
 
+impl Default for BTrimFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl BTrimFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/chr.rs

@@ -65,6 +65,12 @@ pub struct ChrFunc {
     signature: Signature,
 }
 
+impl Default for ChrFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl ChrFunc {
     pub fn new() -> Self {
         Self {

datafusion/functions/src/string/levenshtein.rs

@@ -34,6 +34,12 @@ pub struct LevenshteinFunc {
     signature: Signature,
 }
 
+impl Default for LevenshteinFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl LevenshteinFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/lower.rs

@@ -31,6 +31,12 @@ pub struct LowerFunc {
     signature: Signature,
 }
 
+impl Default for LowerFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl LowerFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/ltrim.rs

@@ -40,6 +40,12 @@ pub struct LtrimFunc {
     signature: Signature,
 }
 
+impl Default for LtrimFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl LtrimFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/mod.rs

@@ -21,28 +21,28 @@ use std::sync::Arc;
 
 use datafusion_expr::ScalarUDF;
 
-mod ascii;
-mod bit_length;
-mod btrim;
-mod chr;
-mod common;
-mod concat;
-mod concat_ws;
-mod ends_with;
-mod initcap;
-mod levenshtein;
-mod lower;
-mod ltrim;
-mod octet_length;
-mod overlay;
-mod repeat;
-mod replace;
-mod rtrim;
-mod split_part;
-mod starts_with;
-mod to_hex;
-mod upper;
-mod uuid;
+pub mod ascii;
+pub mod bit_length;
+pub mod btrim;
+pub mod chr;
+pub mod common;
+pub mod concat;
+pub mod concat_ws;
+pub mod ends_with;
+pub mod initcap;
+pub mod levenshtein;
+pub mod lower;
+pub mod ltrim;
+pub mod octet_length;
+pub mod overlay;
+pub mod repeat;
+pub mod replace;
+pub mod rtrim;
+pub mod split_part;
+pub mod starts_with;
+pub mod to_hex;
+pub mod upper;
+pub mod uuid;
 
 // create UDFs
 make_udf_function!(ascii::AsciiFunc, ASCII, ascii);

datafusion/functions/src/string/octet_length.rs

@@ -31,6 +31,12 @@ pub struct OctetLengthFunc {
     signature: Signature,
 }
 
+impl Default for OctetLengthFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl OctetLengthFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/overlay.rs

@@ -34,6 +34,12 @@ pub struct OverlayFunc {
     signature: Signature,
 }
 
+impl Default for OverlayFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl OverlayFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/repeat.rs

@@ -34,6 +34,12 @@ pub struct RepeatFunc {
     signature: Signature,
 }
 
+impl Default for RepeatFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl RepeatFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/replace.rs

@@ -34,6 +34,12 @@ pub struct ReplaceFunc {
     signature: Signature,
 }
 
+impl Default for ReplaceFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl ReplaceFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/rtrim.rs

@@ -40,6 +40,12 @@ pub struct RtrimFunc {
     signature: Signature,
 }
 
+impl Default for RtrimFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl RtrimFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/split_part.rs

@@ -34,6 +34,12 @@ pub struct SplitPartFunc {
     signature: Signature,
 }
 
+impl Default for SplitPartFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl SplitPartFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/starts_with.rs

@@ -43,6 +43,13 @@ pub fn starts_with<T: OffsetSizeTrait>(args: &[ArrayRef]) -> Result<ArrayRef> {
 pub struct StartsWithFunc {
     signature: Signature,
 }
+
+impl Default for StartsWithFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl StartsWithFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/to_hex.rs

@@ -63,6 +63,13 @@ where
 pub struct ToHexFunc {
     signature: Signature,
 }
+
+impl Default for ToHexFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl ToHexFunc {
     pub fn new() -> Self {
         use DataType::*;

datafusion/functions/src/string/upper.rs

@@ -28,6 +28,12 @@ pub struct UpperFunc {
     signature: Signature,
 }
 
+impl Default for UpperFunc {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl UpperFunc {
     pub fn new() -> Self {
         use DataType::*;