columns

type BucketColumn struct {
	SourceColumn *NumericColumn
	Boundaries   []int
}

type CategoryIDColumn struct {
	Key        string
	BucketSize int
	Delimiter  string
	Dtype      string
}

type ColumnSpec struct {
	ColumnName string
	IsSparse   bool
	Shape      []int
	DType      string
	Delimiter  string
	Vocabulary map[string]string // use a map to generate a list without duplication
	FeatureMap FeatureMap
}

type CrossColumn struct {
	Keys           []interface{}
	HashBucketSize int
}

type EmbeddingColumn struct {
	Key            string // only used when CategoryColumn = nil, feature derivation will fill up the details
	CategoryColumn interface{}
	Dimension      int
	Combiner       string
	Initializer    string
}

type FeatureColumn interface {
	// NOTE: submitters need to know the columnSpec when generating
	// feature_column code. And we maybe use one compound column's data to generate
	// multiple feature columns, so return a list of strings.
	GenerateCode(cs *ColumnSpec) ([]string, error)
	GetKey() string

	// FIXME(typhoonzero): remove delimiter, dtype shape from feature column
	// get these from column spec claused or by feature derivation.
	GetDelimiter() string
	GetDtype() string
	GetInputShape() string

	GetColumnType() int
}

type FeatureMap struct {
	Table     string
	Partition string
}

type NumericColumn struct {
	Key       string
	Shape     []int
	Delimiter string
	Dtype     string
}

type SequenceCategoryIDColumn struct {
	Key        string
	BucketSize int
	Delimiter  string
	Dtype      string
}