sumjson

type ArraySummaryNode struct {
	Freq int `json:"freq"`
}

type BoolSummaryNode struct {
	Freq      int `json:"freq"`
	TrueFreq  int `json:"trues"`
	FalseFreq int `json:"falses"`
}

type BucketRange struct {
	From float64 `json:"from"`
	To   float64 `json:"to"`
	Freq int     `json:"freq"`
}

type Key struct {
	Name string `json:"key"`
	Freq int    `json:"freq"`
}

type Node struct {
	Key      string  `json:"key,omitempty"`
	Children []*Node `json:"objects,omitempty"`
	Elems    []*Node `json:"arrays,omitempty"`

	ScalarNumbers []float64 `json:"numbers,omitempty"`
	ScalarStrings []string  `json:"strings,omitempty"`
	ScalarBools   []bool    `json:"bools,omitempty"`
	ScalarNulls   int       `json:"nulls,omitempty"`

	// stats
	Freq       int          `json:"freq"`
	Summarized *SummaryNode `json:"summary,omitempty"`
}

type NullSummaryNode struct {
	Freq int `json:"freq"`
}

type NumberSummaryNode struct {
	Freq    int     `json:"freq"`
	Unique  int     `json:"unique"`
	AllInts bool    `json:"all_ints"`
	Min     float64 `json:"min"`
	Max     float64 `json:"max"`

	Distribution []BucketRange `json:"distribution"`
}

type ObjectSummaryNode struct {
	Keys []Key `json:"keys"`
}

type Reporter interface {
	ObjectRead(from, to, total int)
	Summarized(done, total int)
}

type StringSample struct {
	Value string `json:"val"`
	Freq  int    `json:"freq"`
}

type StringSummaryNode struct {
	Freq   int            `json:"freq"`
	Unique int            `json:"unique"`
	MinLen int            `json:"min_len"`
	MaxLen int            `json:"max_len"`
	Top    []StringSample `json:"top"`
}

type Summary struct {
	Root *Node `json:"root"`
}

type SummaryNode struct {
	Objects *ObjectSummaryNode `json:"object,omitempty"`
	Arrays  *ArraySummaryNode  `json:"array,omitempty"`
	Numbers *NumberSummaryNode `json:"number,omitempty"`
	Strings *StringSummaryNode `json:"string,omitempty"`
	Bools   *BoolSummaryNode   `json:"bool,omitempty"`
	Nulls   *NullSummaryNode   `json:"null,omitempty"`
}