Use jsonb instead of varchar for complex enums (close #31)

Author: istreeter

modules/common/src/main/scala/com/snowplowanalytics/snowplow/postgres/shredding/Type.scala

@@ -57,7 +57,7 @@ object Type {
   /** Derive a Postgres type, given JSON Schema */
   def getDataType(properties: Schema, suggestions: List[DataTypeSuggestion]): Type =
     suggestions match {
-      case Nil => Type.Varchar(DefaultVarcharSize) // Generic
+      case Nil => Type.Jsonb // Generic
       case suggestion :: tail =>
         suggestion(properties) match {
           case Some(format) => format
@@ -69,8 +69,7 @@ object Type {
   val complexEnumSuggestion: DataTypeSuggestion = properties =>
     properties.enum match {
       case Some(enums) if isComplexEnum(enums.value) =>
-        val longest = excludeNull(enums.value).map(_.noSpaces.length).maximumOption.getOrElse(16)
-        Some(Type.Varchar(longest))
+        Some(Type.Jsonb)
       case _ => None
     }
 
@@ -167,11 +166,6 @@ object Type {
           case Some(max) => Some(Type.Varchar(max))
           case None      => Some(Type.Varchar(DefaultVarcharSize))
         }
-      case (_, _, Some(e)) =>
-        e.value.map(jsonLength).maximumOption match {
-          case Some(max) => Some(Type.Varchar(max))
-          case None      => Some(Type.Varchar(DefaultVarcharSize))
-        }
       case _ => None
     }
   }
@@ -189,9 +183,6 @@ object Type {
     varcharSuggestion
   )
 
-  private def jsonLength(json: Json): Int =
-    json.fold(0, b => b.toString.length, _ => json.noSpaces.length, _.length, _ => json.noSpaces.length, _ => json.noSpaces.length)
-
   /**
     * Get set of types or enum as string excluding null
     *

modules/common/src/test/scala/com/snowplowanalytics/snowplow/postgres/shredding/TypeSpec.scala

@@ -95,7 +95,7 @@ class TypeSpec extends Specification {
 
     "return jsonb type for a complex enum field" >> {
       val properties = Schema(enum = Some(CommonProperties.Enum(List(Json.fromInt(12345), Json.fromString("y")))))
-      Type.getDataType(properties, Type.dataTypeSuggestions) must_== Type.Varchar(5)
+      Type.getDataType(properties, Type.dataTypeSuggestions) must_== Type.Jsonb
     }
 
   }

modules/common/src/test/scala/com/snowplowanalytics/snowplow/postgres/streaming/sinkspec.scala

@@ -98,7 +98,7 @@ class sinkspec extends Database {
         ColumnInfo("required_union", None, false, "jsonb", None),
         ColumnInfo("id", None, true, "uuid", None),
         ColumnInfo("nested.a", None, true, "double precision", None),
-        ColumnInfo("nested.b", None, true, "character varying", Some(4096)),
+        ColumnInfo("nested.b", None, true, "jsonb", None),
         ColumnInfo("some_array", None, true, "jsonb", None),
         ColumnInfo("nested.c", None, true, "bigint", None),
         ColumnInfo("some_date", None, true, "timestamp without time zone", None),