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

package processing

import (
	"regexp"
	"strings"
)

var (
	brandPrefixRe = regexp.MustCompile(`(?i)^([A-Za-z][A-Za-z0-9&.\-]{1,40})\b`)
	dimTripleRe   = regexp.MustCompile(`(?i)(\d+(?:[.,]\d+)?)\s*[x×]\s*(\d+(?:[.,]\d+)?)\s*[x×]\s*(\d+(?:[.,]\d+)?)`)
	dimPairRe     = regexp.MustCompile(`(?i)(\d+(?:[.,]\d+)?)\s*[x×]\s*(\d+(?:[.,]\d+)?)`)
	weightRe      = regexp.MustCompile(`(?i)(\d+(?:[.,]\d+)?)\s*(kg|g|lb|oz)\b`)
)

// FillMissingFields derives sensible standard fields from name/attrs when absent.
func FillMissingFields(mapped map[string]any, attrs map[string]any) map[string]any {
	out := make(map[string]any, len(mapped)+8)
	for k, v := range mapped {
		out[k] = v
	}
	name := stringFromAny(out["name"])
	if name == "" {
		name = stringFromAny(out["title"])
	}

	if stringFromAny(out["brand"]) == "" {
		if b := stringFromAny(attrs["brand"]); b != "" {
			out["brand"] = b
		} else if b := inferBrand(name); b != "" {
			out["brand"] = b
		}
	}

	if stringFromAny(out["gtin"]) == "" {
		if g := stringFromAny(out["ean"]); g != "" {
			out["gtin"] = g
		}
	}

	blob := name + " " + stringFromAny(out["description"])
	for _, m := range []map[string]any{attrs, out} {
		for _, k := range []string{"dimensions", "size", "dimension"} {
			blob += " " + stringFromAny(m[k])
		}
	}

	if stringFromAny(out["width"]) == "" || stringFromAny(out["height"]) == "" || stringFromAny(out["depth"]) == "" {
		if w, h, d, ok := parseDimensions(blob); ok {
			if stringFromAny(out["width"]) == "" {
				out["width"] = w
			}
			if stringFromAny(out["height"]) == "" {
				out["height"] = h
			}
			if stringFromAny(out["depth"]) == "" && d != "" {
				out["depth"] = d
			}
		}
	}

	if stringFromAny(out["weight"]) == "" {
		if w := stringFromAny(attrs["weight"]); w != "" {
			out["weight"] = w
		} else if w, ok := parseWeight(blob); ok {
			out["weight"] = w
		}
	}

	if c := categoryUniqueIDFromAny(out["category"]); c != "" {
		out["category"] = c
	} else if c := categoryUniqueIDFromAny(attrs["category"]); c != "" {
		out["category"] = c
	} else if c := categoryUniqueIDFromAny(out["category_unique_id"]); c != "" {
		out["category"] = c
	}

	if stringFromAny(out["stock_status"]) == "" {
		if s := stringFromAny(out["stock"]); s != "" {
			out["stock_status"] = normalizeStockStatusLabel(s)
		}
	} else {
		out["stock_status"] = normalizeStockStatusLabel(stringFromAny(out["stock_status"]))
	}

	return out
}

func inferBrand(name string) string {
	name = strings.TrimSpace(name)
	if name == "" {
		return ""
	}
	m := brandPrefixRe.FindStringSubmatch(name)
	if len(m) < 2 {
		return ""
	}
	b := strings.TrimSpace(m[1])
	// Skip generic leading words
	switch strings.ToLower(b) {
	case "the", "new", "set", "pack", "pair", "product", "item":
		return ""
	}
	return SanitizeOutput(b)
}

func parseDimensions(blob string) (w, h, d string, ok bool) {
	if m := dimTripleRe.FindStringSubmatch(blob); len(m) == 4 {
		return normalizeNum(m[1]), normalizeNum(m[2]), normalizeNum(m[3]), true
	}
	if m := dimPairRe.FindStringSubmatch(blob); len(m) == 3 {
		return normalizeNum(m[1]), normalizeNum(m[2]), "", true
	}
	return "", "", "", false
}

func parseWeight(blob string) (string, bool) {
	m := weightRe.FindStringSubmatch(blob)
	if len(m) < 3 {
		return "", false
	}
	return normalizeNum(m[1]) + " " + strings.ToLower(m[2]), true
}

func normalizeNum(s string) string {
	return strings.ReplaceAll(strings.TrimSpace(s), ",", ".")
}

func normalizeStockStatusLabel(s string) string {
	if mapped := MapStockStatus(s); mapped != "" {
		return mapped
	}
	s = strings.ToLower(strings.TrimSpace(s))
	switch {
	case s == "" || s == "0" || strings.Contains(s, "out"):
		return "out_of_stock"
	case strings.Contains(s, "pre"):
		return "preorder"
	case strings.Contains(s, "back"):
		return "backorder"
	default:
		return "in_stock"
	}
}

func stringFromAny(v any) string {
	// Flatten arrays / #text wrappers; callers that need HTML-stripped descriptions
	// should use PlainDescriptionFromAny (normalize + V1 enforce do that).
	return coerceScalarText(v)
}