Structured Dataset with generic format should be castable to Flyte Sc…

…hema (#536) * Structured Dataset with generic format should be castable to Schema Signed-off-by: Kevin Su <[email protected]> * Structured Dataset with generic format should be castable to Schema Signed-off-by: Kevin Su <[email protected]> * update Signed-off-by: Kevin Su <[email protected]> --------- Signed-off-by: Kevin Su <[email protected]>
flyteorg · Mar 8, 2023 · 48c37de · 48c37de
1 parent 328a41c
commit 48c37de
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 2 deletions.
diff --git a/pkg/compiler/validators/typing.go b/pkg/compiler/validators/typing.go
@@ -115,7 +115,7 @@ func (t schemaTypeChecker) CastsFrom(upstreamType *flyte.LiteralType) bool {
 	}
 
 	// Flyte Schema can only be serialized to parquet
-	if !strings.EqualFold(structuredDatasetType.Format, "parquet") {
+	if len(structuredDatasetType.Format) != 0 && !strings.EqualFold(structuredDatasetType.Format, "parquet") {
 		return false
 	}
 
@@ -147,7 +147,8 @@ func (t structuredDatasetChecker) CastsFrom(upstreamType *flyte.LiteralType) boo
 	}
 	if schemaType != nil {
 		// Flyte Schema can only be serialized to parquet
-		if !strings.EqualFold(t.literalType.GetStructuredDatasetType().Format, "parquet") {
+		format := t.literalType.GetStructuredDatasetType().Format
+		if len(format) != 0 && !strings.EqualFold(format, "parquet") {
 			return false
 		}
 		return structuredDatasetCastFromSchema(schemaType, t.literalType.GetStructuredDatasetType())

diff --git a/pkg/compiler/validators/typing_test.go b/pkg/compiler/validators/typing_test.go
@@ -579,6 +579,14 @@ func TestSchemaCasting(t *testing.T) {
 			},
 		},
 	}
+	genericStructuredDataset := &core.LiteralType{
+		Type: &core.LiteralType_StructuredDatasetType{
+			StructuredDatasetType: &core.StructuredDatasetType{
+				Columns: []*core.StructuredDatasetType_DatasetColumn{},
+				Format:  "",
+			},
+		},
+	}
 	subsetIntegerSchema := &core.LiteralType{
 		Type: &core.LiteralType_Schema{
 			Schema: &core.SchemaType{
@@ -657,6 +665,16 @@ func TestSchemaCasting(t *testing.T) {
 		assert.True(t, castable, "Schema(a=Integer, b=Float) should be castable to Schema(a=Integer)")
 	})
 
+	t.Run("GenericToSubsetTypedSchema", func(t *testing.T) {
+		castable := AreTypesCastable(genericStructuredDataset, subsetIntegerSchema)
+		assert.True(t, castable, "StructuredDataset() with generic format should be castable to Schema(a=Integer)")
+	})
+
+	t.Run("SubsetTypedSchemaToGeneric", func(t *testing.T) {
+		castable := AreTypesCastable(subsetIntegerSchema, genericStructuredDataset)
+		assert.True(t, castable, "Schema(a=Integer) should be castable to StructuredDataset() with generic format")
+	})
+
 	t.Run("SupersetStructuredToSubsetTypedSchema", func(t *testing.T) {
 		castable := AreTypesCastable(supersetStructuredDataset, subsetIntegerSchema)
 		assert.True(t, castable, "StructuredDataset(a=Integer, b=Float) should be castable to Schema(a=Integer)")