seaweedfs/weed/data/columnar.proto

syntax = "proto3";

package columnar_pb;

option go_package = "github.com/seaweedfs/seaweedfs/weed/data/columnar_pb";

message FileId {
  uint32 volume_id = 1;
  uint64 file_key = 2;
  fixed32 cookie = 3;
}

enum LogicalType {
  Uint8 = 0;
  Uint16 = 1;
  Float32 = 4;
}

message ColumnUint16 {
  uint32 base = 1;
  uint32 min = 3;
  uint32 max = 4;
}

message ColumnUint32 {
  uint32 base = 1;
  uint32 min = 3;
  uint32 max = 4;
}

message ColumnFloat32 {
  uint32 min = 3;
  uint32 max = 4;
}

message ColumnSplit {
  // The ids of the fields/columns in this file
  int32 field_id = 1;
  FileId file_id = 2;
  int64 row_offset = 3;
  int32 row_count = 4;

  oneof storage_type {
    ColumnUint16  meta_uint16 = 8;
    ColumnUint32  meta_uint32 = 9;
    ColumnFloat32 meta_float32 = 10;
  }
}

message Snapshot {
  // All fields of the dataset, including the nested fields.
  repeated Field fields = 1;

  repeated string data_files = 2;

  // Snapshot version number.
  uint64 version = 3;

}

message DataFile {
  repeated int32 field_ids = 1;
  repeated RowGroup row_groups = 2;
}

message RowGroup {
  int64 row_offset = 1;
  int32 row_count = 2;
  repeated ColumnSplit column_splits = 3;
}

// Field metadata for a column.
message Field {
  enum Type {
    PARENT = 0;
    REPEATED = 1;
    LEAF = 2;
  }
  Type type = 1;

  // Fully qualified name.
  string name = 2;
  /// Field Id.
  int32 id = 3;
  /// Parent Field ID. If not set, this is a top-level column.
  int32 parent_id = 4;

  // Logical types, support parameterized Arrow Type.
  LogicalType logical_type = 5;
  // If this field is nullable.
  bool nullable = 6;
}


message AnyValue {
  oneof value {
    bytes bytes_value = 1;
    bool bool_value = 2;
    uint64 int64_value = 3;
    uint32 int32_value = 4;
    double double_value = 5;
  }
}