ewhauser
diff --git a/‎internal/gen.go‎
Lines changed: 93 additions & 9 deletions b/‎internal/gen.go‎
Lines changed: 93 additions & 9 deletions
@@ -53,6 +53,8 @@ type Field struct {
 	Name    string
 	Type    pyType
 	Comment string
+	// EmbedFields contains the embedded fields that require scanning.
+	EmbedFields []Field
 }
 
 type Struct struct {
@@ -105,14 +107,42 @@ func (v QueryValue) RowNode(rowVar string) *pyast.Node {
 	call := &pyast.Call{
 		Func: v.Annotation(),
 	}
-	for i, f := range v.Struct.Fields {
-		call.Keywords = append(call.Keywords, &pyast.Keyword{
-			Arg: f.Name,
-			Value: subscriptNode(
+	rowIndex := 0 // We need to keep track of the index in the row variable.
+	for _, f := range v.Struct.Fields {
+
+		var valueNode *pyast.Node
+		// Check if we are using sqlc.embed, if so we need to create a new object.
+		if len(f.EmbedFields) > 0 {
+			// We keep this separate so we can easily add all arguments.
+			embed_call := &pyast.Call{Func: f.Type.Annotation()}
+
+			// Now add all field Initializers for the embedded model that index into the original row.
+			for i, embedField := range f.EmbedFields {
+				embed_call.Keywords = append(embed_call.Keywords, &pyast.Keyword{
+					Arg: embedField.Name,
+					Value: subscriptNode(
+						rowVar,
+						constantInt(rowIndex+i),
+					),
+				})
+			}
+
+			valueNode = &pyast.Node{
+				Node: &pyast.Node_Call{
+					Call: embed_call,
+				},
+			}
+
+			rowIndex += len(f.EmbedFields)
+		} else {
+			valueNode = subscriptNode(
 				rowVar,
-				constantInt(i),
-			),
-		})
+				constantInt(rowIndex),
+			)
+			rowIndex++
+		}
+
+		call.Keywords = append(call.Keywords, &pyast.Keyword{Arg: f.Name, Value: valueNode})
 	}
 	return &pyast.Node{
 		Node: &pyast.Node_Call{
@@ -319,6 +349,47 @@ func paramName(p *plugin.Parameter) string {
 type pyColumn struct {
 	id int32
 	*plugin.Column
+	embed *pyEmbed
+}
+
+type pyEmbed struct {
+	modelType string
+	modelName string
+	fields    []Field
+}
+
+// Taken from https://github.com/sqlc-dev/sqlc/blob/8c59fbb9938a0bad3d9971fc2c10ea1f83cc1d0b/internal/codegen/golang/result.go#L123-L126
+// look through all the structs and attempt to find a matching one to embed
+// We need the name of the struct and its field names.
+func newGoEmbed(embed *plugin.Identifier, structs []Struct, defaultSchema string) *pyEmbed {
+	if embed == nil {
+		return nil
+	}
+
+	for _, s := range structs {
+		embedSchema := defaultSchema
+		if embed.Schema != "" {
+			embedSchema = embed.Schema
+		}
+
+		// compare the other attributes
+		if embed.Catalog != s.Table.Catalog || embed.Name != s.Table.Name || embedSchema != s.Table.Schema {
+			continue
+		}
+
+		fields := make([]Field, len(s.Fields))
+		for i, f := range s.Fields {
+			fields[i] = f
+		}
+
+		return &pyEmbed{
+			modelType: s.Name,
+			modelName: s.Name,
+			fields:    fields,
+		}
+	}
+
+	return nil
 }
 
 func columnsToStruct(req *plugin.GenerateRequest, name string, columns []pyColumn) *Struct {
@@ -342,10 +413,22 @@ func columnsToStruct(req *plugin.GenerateRequest, name string, columns []pyColum
 		if suffix > 0 {
 			fieldName = fmt.Sprintf("%s_%d", fieldName, suffix)
 		}
-		gs.Fields = append(gs.Fields, Field{
+
+		f := Field{
 			Name: fieldName,
 			Type: makePyType(req, c.Column),
-		})
+		}
+
+		if c.embed != nil {
+			f.Type = pyType{
+				InnerType: "models." + modelName(c.embed.modelType, req.Settings),
+				IsArray:   false,
+				IsNull:    false,
+			}
+			f.EmbedFields = c.embed.fields
+		}
+
+		gs.Fields = append(gs.Fields, f)
 		seen[colName]++
 	}
 	return &gs
@@ -459,6 +542,7 @@ func buildQueries(conf Config, req *plugin.GenerateRequest, structs []Struct) ([
 					columns = append(columns, pyColumn{
 						id:     int32(i),
 						Column: c,
+						embed:  newGoEmbed(c.EmbedTable, structs, req.Catalog.DefaultSchema),
 					})
 				}
 				gs = columnsToStruct(req, query.Name+"Row", columns)