chore: rebased, better error messaging in slt

Author: mesejo

datafusion/functions/src/unicode/substr.rs

@@ -81,8 +81,13 @@ impl ScalarUDFImpl for SubstrFunc {
     }
 
     fn coerce_types(&self, arg_types: &[DataType]) -> Result<Vec<DataType>> {
-        if ![DataType::LargeUtf8, DataType::Utf8View, DataType::Utf8]
-            .contains(&arg_types[0])
+        if ![
+            DataType::LargeUtf8,
+            DataType::Utf8View,
+            DataType::Utf8,
+            DataType::Null,
+        ]
+        .contains(&arg_types[0])
         {
             return plan_err!(
                 "The first argument of the {} function can only be a string, but got {:?}.",
@@ -91,7 +96,7 @@ impl ScalarUDFImpl for SubstrFunc {
             );
         }
 
-        if ![DataType::Int64, DataType::Int32].contains(&arg_types[1]) {
+        if ![DataType::Int64, DataType::Int32, DataType::Null].contains(&arg_types[1]) {
             return plan_err!(
                 "The second argument of the {} function can only be an integer, but got {:?}.",
                 self.name(),
@@ -100,7 +105,7 @@ impl ScalarUDFImpl for SubstrFunc {
         }
 
         if arg_types.len() == 3
-            && ![DataType::Int64, DataType::Int32].contains(&arg_types[2])
+            && ![DataType::Int64, DataType::Int32, DataType::Null].contains(&arg_types[2])
         {
             return plan_err!(
                 "The third argument of the {} function can only be an integer, but got {:?}.",
@@ -109,11 +114,17 @@ impl ScalarUDFImpl for SubstrFunc {
             );
         }
 
+        let first_data_type = if arg_types[0] == DataType::Null {
+            DataType::Utf8
+        } else {
+            arg_types[0].clone()
+        };
+
         if arg_types.len() == 2 {
-            Ok(vec![arg_types[0].to_owned(), DataType::Int64])
+            Ok(vec![first_data_type.to_owned(), DataType::Int64])
         } else {
             Ok(vec![
-                arg_types[0].to_owned(),
+                first_data_type.to_owned(),
                 DataType::Int64,
                 DataType::Int64,
             ])

datafusion/sqllogictest/test_files/encoding.slt

@@ -28,13 +28,13 @@ CREATE TABLE test(
 ;
 
 # errors
-query error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+query error 1st argument should be Utf8 or Binary or Null, got Int64
 select encode(12, 'hex');
 
 query error DataFusion error: Error during planning: There is no built\-in encoding named 'non_encoding', currently supported encodings are: base64, hex
 select encode(bin_field, 'non_encoding') from test;
 
-query error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+query error 1st argument should be Utf8 or Binary or Null, got Int64
 select decode(12, 'hex');
 
 query error DataFusion error: Error during planning: There is no built\-in encoding named 'non_encoding', currently supported encodings are: base64, hex

datafusion/sqllogictest/test_files/functions.slt

@@ -497,10 +497,10 @@ SELECT substr('alphabet', 3, CAST(NULL AS int))
 ----
 NULL
 
-statement error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+statement error The first argument of the substr function can only be a string, but got Int64
 SELECT substr(1, 3)
 
-statement error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+statement error The first argument of the substr function can only be a string, but got Int64
 SELECT substr(1, 3, 4)
 
 query T

datafusion/sqllogictest/test_files/scalar.slt

@@ -1907,7 +1907,7 @@ select position('' in '')
 1
 
 
-query error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+query error POSITION function can only accept strings
 select position(1 in 1)
 
 

datafusion/sqllogictest/test_files/string_functions/substr/substr_literal.slt.part

@@ -123,10 +123,10 @@ SELECT substr('Hello🌏世界', 5, 3)
 ----
 o🌏世
 
-statement error The SUBSTR function can only accept strings, but got Int64.
+statement error The first argument of the substr function can only be a string, but got Int64
 SELECT substr(1, 3)
 
-statement error The SUBSTR function can only accept strings, but got Int64.
+statement error The first argument of the substr function can only be a string, but got Int64
 SELECT substr(1, 3, 4)
 
 statement error Execution error: negative substring length not allowed

datafusion/sqllogictest/test_files/timestamps.slt

@@ -2858,7 +2858,7 @@ statement error
 select to_local_time('2024-04-01T00:00:20Z'::timestamp, 'some string');
 
 # invalid argument data type
-statement error DataFusion error: Error during planning: Execution error: User\-defined coercion failed with Plan
+statement error The to_local_time function can only accept Timestamp as the arg got Utf8
 select to_local_time('2024-04-01T00:00:20Z');
 
 # invalid timezone