descrybe/apps/api/internal/processing/enrich.go

package processing

import (
	"regexp"
	"strconv"
	"strings"
)

var (
	enrichCDATAWrapRe = regexp.MustCompile(`(?is)^\s*<!\[CDATA\[(.*)\]\]>\s*$`)
	enrichMassValueRe = regexp.MustCompile(`(?i)^\s*([0-9]+(?:[.,][0-9]+)?)\s*([a-zµμ]+)?\s*$`)
	enrichNbspRe      = regexp.MustCompile(`(?i)&nbsp;|&#160;`)
	enrichDimKeyRe    = regexp.MustCompile(`(?i)^(net)?(width|height|depth|length|dimension)s?$`)
	enrichZeroNumRe   = regexp.MustCompile(`^\s*0+(?:[.,]0+)?\s*$`)
	enrichHTMLTagRe   = regexp.MustCompile(`(?is)<[^>]*>`)
)

// Availability values normalized from vendor stockStatus text.
const (
	AvailabilityInStock    = "in_stock"
	AvailabilityOutOfStock = "out_of_stock"
	AvailabilityPreorder   = "preorder"
	AvailabilityBackorder  = "backorder"
	AvailabilityLimited    = "limited"
)

// EnrichMapped returns a processing-time copy of mapped feed fields with
// derived fills and cleanup. Sync/map must keep raw values unchanged.
//
// Complements NormalizeMapped (alias flatten / zero dims) with Janus-style rules:
// gtin←EAN, title←name, strip empty CDATA/HTML, parse netMass+unit,
// stockStatus→availability enum. Does not invent empty EPRELID/mainImage.
func EnrichMapped(mapped map[string]any) map[string]any {
	if mapped == nil {
		return map[string]any{}
	}
	out := make(map[string]any, len(mapped)+4)
	for k, v := range mapped {
		out[k] = v
	}

	stripEmptyMarkup(out)
	dropZeroDimensions(out)
	deriveGTIN(out)
	deriveTitle(out)
	parseNetMass(out)
	applyAvailabilityFromStock(out)
	return out
}

func stripEmptyMarkup(m map[string]any) {
	for k, v := range m {
		s, ok := enrichAsString(v)
		if !ok {
			continue
		}
		cleaned := cleanMarkupValue(s)
		if cleaned == "" {
			delete(m, k)
			continue
		}
		if cleaned != s {
			m[k] = cleaned
		}
	}
}

func cleanMarkupValue(s string) string {
	s = strings.TrimSpace(s)
	if s == "" {
		return ""
	}
	if sub := enrichCDATAWrapRe.FindStringSubmatch(s); len(sub) == 2 {
		s = strings.TrimSpace(sub[1])
	}
	if strings.EqualFold(s, "<![CDATA[]]>") || strings.EqualFold(s, "<![CDATA[ ]]>") {
		return ""
	}
	plain := enrichNbspRe.ReplaceAllString(s, " ")
	// Reuse specs package htmlTagRe via stripping through a local call pattern:
	// htmlTagRe lives in specs.go — strip tags with a dedicated helper.
	plain = stripHTMLTags(plain)
	plain = strings.Join(strings.Fields(plain), " ")
	if plain == "" {
		return ""
	}
	return strings.TrimSpace(s)
}

func stripHTMLTags(s string) string {
	return enrichHTMLTagRe.ReplaceAllString(s, " ")
}

func dropZeroDimensions(m map[string]any) {
	for k, v := range m {
		leaf := enrichLeafKey(k)
		if !enrichDimKeyRe.MatchString(leaf) {
			continue
		}
		if enrichIsZeroValue(v) {
			delete(m, k)
		}
	}
}

func deriveGTIN(m map[string]any) {
	if hasNonEmptyEnrich(m, "gtin", "GTIN") {
		return
	}
	for _, key := range []string{"ean", "EAN", "upc", "UPC", "barcode", "Barcode"} {
		if s, ok := enrichAsString(m[key]); ok && s != "" {
			m["gtin"] = s
			return
		}
	}
}

func deriveTitle(m map[string]any) {
	if hasNonEmptyEnrich(m, "title", "Title") {
		return
	}
	for _, key := range []string{"name", "Name", "product_name", "productName", "ProductName"} {
		if s, ok := enrichAsString(m[key]); ok && s != "" {
			m["title"] = s
			return
		}
	}
}

func parseNetMass(m map[string]any) {
	var raw any
	var srcKey string
	for _, key := range []string{"netMass", "net_mass", "NetMass", "weight", "Weight", "mass"} {
		if v, ok := m[key]; ok {
			raw = v
			srcKey = key
			break
		}
	}
	if raw == nil {
		return
	}
	s, ok := enrichAsString(raw)
	if !ok || s == "" {
		return
	}
	sub := enrichMassValueRe.FindStringSubmatch(s)
	if len(sub) < 2 {
		return
	}
	num := strings.ReplaceAll(sub[1], ",", ".")
	f, err := strconv.ParseFloat(num, 64)
	if err != nil {
		return
	}
	if f == 0 {
		delete(m, srcKey)
		return
	}
	unit := ""
	if len(sub) >= 3 {
		unit = normalizeMassUnit(sub[2])
	}
	m["net_mass"] = f
	if unit != "" {
		m["net_mass_unit"] = unit
	}
	if unit != "" {
		m["weight"] = strings.TrimSpace(num + " " + unit)
	} else {
		m["weight"] = num
	}
}

func normalizeMassUnit(u string) string {
	u = strings.ToLower(strings.TrimSpace(u))
	u = strings.ReplaceAll(u, "μ", "u")
	u = strings.ReplaceAll(u, "µ", "u")
	switch u {
	case "kg", "kilogram", "kilograms":
		return "kg"
	case "g", "gram", "grams":
		return "g"
	case "mg", "milligram", "milligrams":
		return "mg"
	case "lb", "lbs", "pound", "pounds":
		return "lb"
	case "oz", "ounce", "ounces":
		return "oz"
	case "t", "ton", "tonne", "tonnes":
		return "t"
	case "ug", "mcg":
		return "ug"
	default:
		return u
	}
}

func applyAvailabilityFromStock(m map[string]any) {
	var raw any
	for _, key := range []string{"stockStatus", "stock_status", "StockStatus", "availability", "Availability"} {
		if v, ok := m[key]; ok {
			raw = v
			break
		}
	}
	if raw == nil {
		return
	}
	s, ok := enrichAsString(raw)
	if !ok || s == "" {
		return
	}
	if avail := MapStockStatus(s); avail != "" {
		m["availability"] = avail
	}
}

// MapStockStatus maps vendor stock text onto a stable availability enum.
func MapStockStatus(s string) string {
	n := normalizeStockToken(s)
	if n == "" {
		return ""
	}
	switch {
	case n == "instock" || n == "in_stock" || n == "available" || n == "nazalogi" ||
		n == "naskladiscu" || n == "auflager" || n == "yes" || n == "1" || n == "true":
		return AvailabilityInStock
	case n == "outofstock" || n == "out_of_stock" || n == "unavailable" || n == "ninazalogi" ||
		n == "soldout" || n == "no" || n == "0" || n == "false":
		return AvailabilityOutOfStock
	case strings.Contains(n, "preorder") || strings.Contains(n, "pre_order"):
		return AvailabilityPreorder
	case strings.Contains(n, "backorder") || strings.Contains(n, "back_order"):
		return AvailabilityBackorder
	case strings.Contains(n, "limited") || n == "lowstock" || n == "low_stock":
		return AvailabilityLimited
	case strings.Contains(n, "instock") || strings.Contains(n, "in_stock") || strings.Contains(n, "available"):
		return AvailabilityInStock
	case strings.Contains(n, "outofstock") || strings.Contains(n, "out_of_stock"):
		return AvailabilityOutOfStock
	default:
		return ""
	}
}

func normalizeStockToken(s string) string {
	s = strings.ToLower(strings.TrimSpace(s))
	s = strings.ReplaceAll(s, "-", "")
	s = strings.ReplaceAll(s, " ", "")
	s = strings.ReplaceAll(s, "_", "")
	replacer := strings.NewReplacer(
		"č", "c", "ć", "c", "š", "s", "ž", "z", "đ", "d",
		"ä", "a", "ö", "o", "ü", "u", "ß", "ss",
	)
	return replacer.Replace(s)
}

func hasNonEmptyEnrich(m map[string]any, keys ...string) bool {
	for _, k := range keys {
		if s, ok := enrichAsString(m[k]); ok && s != "" {
			return true
		}
	}
	return false
}

func enrichAsString(v any) (string, bool) {
	switch t := v.(type) {
	case string:
		return strings.TrimSpace(t), true
	case float64:
		if t == float64(int64(t)) {
			return strconv.FormatInt(int64(t), 10), true
		}
		return strconv.FormatFloat(t, 'f', -1, 64), true
	case float32:
		return strconv.FormatFloat(float64(t), 'f', -1, 64), true
	case int:
		return strconv.Itoa(t), true
	case int64:
		return strconv.FormatInt(t, 10), true
	default:
		return "", false
	}
}

func enrichIsZeroValue(v any) bool {
	switch t := v.(type) {
	case nil:
		return true
	case string:
		return enrichZeroNumRe.MatchString(t) || strings.TrimSpace(t) == "" || isZeroishString(t)
	case float64:
		return t == 0
	case float32:
		return t == 0
	case int:
		return t == 0
	case int64:
		return t == 0
	case int32:
		return t == 0
	default:
		if s, ok := enrichAsString(v); ok {
			return enrichZeroNumRe.MatchString(s) || isZeroishString(s)
		}
		return false
	}
}

func enrichLeafKey(k string) string {
	k = strings.TrimSpace(k)
	if i := strings.LastIndex(k, "/"); i >= 0 {
		k = k[i+1:]
	}
	return k
}