Support binary / largebinary arguments to string functions

Author: alamb

datafusion/expr/src/built_in_function.rs

@@ -1523,12 +1523,17 @@ impl FromStr for BuiltinScalarFunction {
     }
 }
 
+/// Creates a function that returns a type based on the type of string to the
+/// first argument
+///
+/// If the input type is `LargeUtf8` or `LargeBinary` the return type is `largeUtf8Type`,
+/// If the input type is `Utf8` or `Binary` the return type is `utf8Type`,
 macro_rules! make_utf8_to_return_type {
     ($FUNC:ident, $largeUtf8Type:expr, $utf8Type:expr) => {
         fn $FUNC(arg_type: &DataType, name: &str) -> Result<DataType> {
             Ok(match arg_type {
-                DataType::LargeUtf8 => $largeUtf8Type,
-                DataType::Utf8 => $utf8Type,
+                DataType::LargeUtf8 | DataType::LargeBinary => $largeUtf8Type,
+                DataType::Utf8 | DataType::Binary => $utf8Type,
                 DataType::Null => DataType::Null,
                 DataType::Dictionary(_, value_type) => match **value_type {
                     DataType::LargeUtf8 => $largeUtf8Type,
@@ -1554,7 +1559,9 @@ macro_rules! make_utf8_to_return_type {
     };
 }
 
+/// `utf8_to_str_type` returns either a Utf8 or LargeUtf8 based on the input type size.
 make_utf8_to_return_type!(utf8_to_str_type, DataType::LargeUtf8, DataType::Utf8);
+/// `utf8_to_str_type` returns either a Int32 or Int64 based on the input type size.
 make_utf8_to_return_type!(utf8_to_int_type, DataType::Int64, DataType::Int32);
 
 fn utf8_or_binary_to_binary_type(arg_type: &DataType, name: &str) -> Result<DataType> {

datafusion/sqllogictest/test_files/binary.slt

@@ -226,21 +226,29 @@ SELECT largebinary FROM t where largebinary LIKE '%F';
 
 
 # character_length function
-# https://github.com/apache/arrow-datafusion/issues/7344
-query error DataFusion error: Error during planning: The "character_length" function can only accept strings, but got Binary\.
+query TITI
 SELECT
   cast(binary as varchar) as str,
   character_length(binary) as binary_len,
   cast(largebinary as varchar) as large_str,
   character_length(binary) as largebinary_len
 from t;
+----
+Foo 3 Foo 3
+NULL NULL NULL NULL
+Bar 3 Bar 3
+FooBar 6 FooBar 6
 
 # regexp_replace
-# https://github.com/apache/arrow-datafusion/issues/7345
-query error DataFusion error: Error during planning: The "regexp_replace" function can only accept strings, but got Binary\.
+query TTTT
 SELECT
   cast(binary as varchar) as str,
   regexp_replace(binary, 'F', 'f') as binary_replaced,
   cast(largebinary as varchar) as large_str,
   regexp_replace(largebinary, 'F', 'f') as large_binary_replaced
 from t;
+----
+Foo foo Foo foo
+NULL NULL NULL NULL
+Bar Bar Bar Bar
+FooBar fooBar FooBar fooBar