syntax = "proto3";
import "sarus_data_spec/protobuf/type.proto";
import "sarus_data_spec/protobuf/path.proto";
import "sarus_data_spec/protobuf/scalar.proto";

message Transform {
  // A dataset transform
  string uuid = 1; // e.g. RFC 4122 id used to refer to the transform
  string name = 2;
  string doc = 3;
  Spec spec = 4;
  map<string, string> properties = 5; // Other properties
  bool inversible = 6;
  bool schema_preserving = 7;

  // Definitions
  message Spec {
    oneof spec {
      Identity identity = 1;
      Variable variable = 2;
      Composed composed = 3;
      Project project = 4;
      Filter filter = 5;
      Shuffle shuffle = 6;
      Join join = 7;
      Cast cast = 8;
      Sample sample = 9;
      UserSettings user_settings = 10;
      Protect protect_dataset = 11;
      External external = 12; // np transforms, pd transforms,...
      Synthetic synthetic = 13;
      Transcode transcode = 14;
      InverseTranscode inverse_transcode = 15;
      GetItem get_item = 16;
      ProtectedPaths protected_paths = 17;
      AutomaticUserSettings automatic_user_settings = 18;
      PublicPaths public_paths = 19;
      AssignBudget assign_budget = 20;
      AutomaticBudget automatic_budget = 21;
      AttributesBudget attribute_budget = 22;
      SDBudget sd_budget = 23;
      DeriveSeed derive_seed = 24;
      GroupByPE group_by_pe = 25;
      SamplingRatios sampling_ratios = 26;
      SelectSql select_sql = 27;
      Extract extract = 28;
      RelationshipSpec relationship_spec = 29;
      DifferentiatedSample differentiated_sample = 30;
      ValidatedUserType validated_user_type=31;
      DPSelectSql dp_select_sql = 32;
    }
  }

  message External {
    bytes arguments = 1;
    bytes named_arguments = 2;
    OpIdentifier op_identifier = 3;

    message OpIdentifier{
      oneof op {
        Std std = 1;
        Pandas pandas = 2;
        Numpy numpy = 3;
        Tensorflow tensorflow = 4;
        Sklearn sklearn = 5;
        PandasProfiling pandas_profiling = 6;
        XGBoost xgboost = 7;
        Skopt skopt = 8;
        Imblearn imblearn = 9;
      }
    };

    message Std {
      string name = 1;
    }

    message Pandas {
      string name = 1;
    }

    message Numpy {
      string name = 1;
    }

    message Tensorflow {
      string name = 1;
    }

    message Sklearn {
      string name = 1;
    }

    message PandasProfiling {
      string name = 1;
    }

    message XGBoost {
      string name = 1;
    }

    message Skopt {
      string name = 1;
    }

    message Imblearn {
      string name = 1;
    }
  }

  message Identity {} // Does nothing

  message Variable {
    int32 position = 1;
    string name = 2;
  } // Numbered or named identity to use as input of the composed transform

  message Composed {
    string transform = 1; // Transform
    repeated string arguments = 2; // Arguments of the current transform are transforms
    map<string, string> named_arguments = 3;
  }

  message Project {
    sarus_data_spec.Type projection = 1; // This should be a 'supertype' the type the data can project into.
    // For product types (struct or tuple), this is a type with a subset of the fields
    // For map types, this is a map type with subset (subtype) of key and superset (supertype) of value
    // For union types, this is a type with more terms in the union
    // To start wit let's simply use this with structs
  }

  message Filter {
    sarus_data_spec.Type filter = 1; // This should be a 'subtype' the type the data can be retricted to.
    // For union types, this is a type with less terms in the union.
    // Optional types can for instance be filtered to non-optional
    // Value type and Predicate types can be used to restrict values
  }

  message Shuffle {
  }

  message Join {
    sarus_data_spec.Type on = 1; // This should be a common 'supertype' between tables.
  }

  message Cast {
    sarus_data_spec.Type type = 1; // Type to cast into.
  }

  message Sample {
    // Sample a dataset
    oneof proportion {
      double fraction = 1;
      int64 size = 2;
    }
    sarus_data_spec.Scalar seed = 3;
  }

  message SchemaInference {
    CastPolicy cast_policy = 1;
    enum CastPolicy {
      NONE = 0;
      MOST_LIKELY = 1;
    }
  }

  message GroupBy {
    string key = 1;
  }

  message Synthetic {
  }

  message UserSettings {
  }

  message AutomaticUserSettings {
    int64 max_categories = 1;
  }

  message Protect {
  }


  message Transcode {
  }

  message InverseTranscode {
  }
  message DifferentiatedSample{
    oneof proportion {
      double fraction = 1;
      int64 size = 2;
    }
    sarus_data_spec.Scalar seed = 3;
  }

  message ProtectedPaths {
  }

  message PublicPaths {
  }

  message GetItem {
    sarus_data_spec.Path path = 1;
  }

  message AssignBudget {
  }

  message AutomaticBudget {
  }

  message AttributesBudget {
  }

  message SDBudget {
  }

  message DeriveSeed {
    int64 random_integer = 1;
  }

  message GroupByPE {
  }

  message SamplingRatios {
  }

  message RelationshipSpec {
  }

  message SelectSql {
    oneof select {
      string query = 1;
      AliasedQueries aliased_queries = 2;
    }
    SQLDialect sql_dialect = 3;
  }

  message DPSelectSql {
    oneof select {
      string query = 1;
      AliasedQueries aliased_queries = 2;
    }
    SQLDialect sql_dialect = 3;
  }
  enum SQLDialect {
    NONE = 0;
    POSTGRES = 1;
    SQL_SERVER = 2;
    MY_SQL = 3;
    SQLLITE = 4;
    ORACLE = 5;
    BIG_QUERY = 6;
    REDSHIFT = 7;
    HIVE = 8;
  }


  message AliasedQueries {
    repeated AliasedQuery aliased_query = 1;
  }

  message AliasedQuery {
    sarus_data_spec.Path path = 1;
    string query = 2;
  }

  message Extract {
    int32 size = 1;
  }
   message ValidatedUserType {
  }
}