nan.go

package mapping

import "reflect"

// based on 'na_values' from:
// https://pandas.pydata.org/pandas-docs/stable/generated/pandas.read_csv.html
var nanMap = map[string]struct{}{
	"":         {},
	"#N/A":     {},
	"#N/A N/A": {},
	"#NA":      {},
	"-1.#IND":  {},
	"-1.#QNAN": {},
	"-NaN":     {},
	"-nan":     {},
	"1.#IND":   {},
	"1.#QNAN":  {},
	"N/A":      {},
	"NA":       {},
	"NULL":     {},
	"n/a":      {},
	"null":     {},
}

// MapNaN returns a Map function that can be used with csvutil.Decoder to recognize
// an extended list of NaN values. In addition to the default NaNs list it is
// possible to supply a list of custom ones.
//
// NaN Mapping works when v is a float64, float32, or any floatX alias.
// It is not tied up with any column.
//
// The default NaN formats provided with this function are as follows:
//	""
//	"#N/A"
//	"#N/A N/A"
//	"#NA"
//	"-1.#IND"
//	"-1.#QNAN"
//	"-NaN"
//	"-nan"
//	"1.#IND"
//	"1.#QNAN"
//	"N/A"
//	"NA"
//	"NULL"
//	"n/a"
//	"null"
//
// The above list was based on default pandas na_values:
//
// https://pandas.pydata.org/pandas-docs/stable/generated/pandas.read_csv.html
func MapNaN(nans ...string) func(string, string, interface{}) string {
	m := nanMap

	if len(nans) > 0 {
		m = make(map[string]struct{}, len(nanMap)+len(nans))
		for k, v := range nanMap {
			m[k] = v
		}
		for _, v := range nans {
			m[v] = struct{}{}
		}
	}

	return func(s, _ string, v interface{}) string {
		return nan(m, s, v)
	}
}

func nan(m map[string]struct{}, field string, v interface{}) string {
	if k := reflect.TypeOf(v).Kind(); k != reflect.Float64 && k != reflect.Float32 {
		return field
	}
	if _, ok := m[field]; ok {
		return "NaN"
	}
	return field
}