Signature::Coercible with user defined implicit casting (apache#14440)

* coerciblev2

Signed-off-by: Jay Zhan <[email protected]>

* repeat

Signed-off-by: Jay Zhan <[email protected]>

* fix possible types

* replace all coerciblev1

* cleanup

* remove specialize logic

* comment

* err msg

* ci escape

* rm coerciblev1

Signed-off-by: Jay Zhan <[email protected]>

* fmt

* rename

* rename

* refactor

* make default_casted_type private

* cleanup

* fmt

* integer

* rm binary for ascii

* rm unused

* conflit

* fmt

* Rename get_example_types, make method on TypeSignatureClass

* Move more logic into TypeSignatureClass

* fix docs

* 46

* enum

* fmt

* fmt

* doc

* upd doc

---------

Signed-off-by: Jay Zhan <[email protected]>
Co-authored-by: Andrew Lamb <[email protected]>

Author: jayzhan211

Cargo.lock

@@ -405,7 +405,7 @@ fn get_udf_args_and_return_types(
     udf: &Arc<ScalarUDF>,
 ) -> Result<Vec<(Vec<String>, Option<String>)>> {
     let signature = udf.signature();
-    let arg_types = signature.type_signature.get_possible_types();
+    let arg_types = signature.type_signature.get_example_types();
     if arg_types.is_empty() {
         Ok(vec![(vec![], None)])
     } else {
@@ -428,7 +428,7 @@ fn get_udaf_args_and_return_types(
     udaf: &Arc<AggregateUDF>,
 ) -> Result<Vec<(Vec<String>, Option<String>)>> {
     let signature = udaf.signature();
-    let arg_types = signature.type_signature.get_possible_types();
+    let arg_types = signature.type_signature.get_example_types();
     if arg_types.is_empty() {
         Ok(vec![(vec![], None)])
     } else {
@@ -452,7 +452,7 @@ fn get_udwf_args_and_return_types(
     udwf: &Arc<WindowUDF>,
 ) -> Result<Vec<(Vec<String>, Option<String>)>> {
     let signature = udwf.signature();
-    let arg_types = signature.type_signature.get_possible_types();
+    let arg_types = signature.type_signature.get_example_types();
     if arg_types.is_empty() {
         Ok(vec![(vec![], None)])
     } else {

datafusion/catalog/src/information_schema.rs

@@ -198,6 +198,11 @@ impl LogicalType for NativeType {
         TypeSignature::Native(self)
     }
 
+    /// Returns the default casted type for the given arrow type
+    ///
+    /// For types like String or Date, multiple arrow types mapped to the same logical type
+    /// If the given arrow type is one of them, we return the same type
+    /// Otherwise, we define the default casted type for the given arrow type
     fn default_cast_for(&self, origin: &DataType) -> Result<DataType> {
         use DataType::*;
 
@@ -226,6 +231,10 @@ impl LogicalType for NativeType {
             (Self::Decimal(p, s), _) if p <= &38 => Decimal128(*p, *s),
             (Self::Decimal(p, s), _) => Decimal256(*p, *s),
             (Self::Timestamp(tu, tz), _) => Timestamp(*tu, tz.clone()),
+            // If given type is Date, return the same type
+            (Self::Date, origin) if matches!(origin, Date32 | Date64) => {
+                origin.to_owned()
+            }
             (Self::Date, _) => Date32,
             (Self::Time(tu), _) => match tu {
                 TimeUnit::Second | TimeUnit::Millisecond => Time32(*tu),

datafusion/common/src/types/native.rs

@@ -39,5 +39,6 @@ path = "src/lib.rs"
 [dependencies]
 arrow = { workspace = true }
 datafusion-common = { workspace = true }
+indexmap = { workspace = true }
 itertools = { workspace = true }
 paste = "^1.0"