fix

apps/api/internal/processing/specs.go

@@ -44,8 +44,16 @@ func parseSpecsInto(dst map[string]any, v any) {
 		// Already structured attributes / grouped specs
 		if looksLikeAttrMap(t) {
 			for k, val := range t {
+				lk := strings.ToLower(strings.TrimSpace(k))
+				if isReservedProductKey(lk) || isInvalidAttributeKey(k) {
+					continue
+				}
 				if s := stringifySpecValue(val); s != "" {
-					dst[SanitizeOutput(k)] = s
+					key := SanitizeOutput(k)
+					if key == "" || isInvalidAttributeKey(key) {
+						continue
+					}
+					dst[key] = s
 				} else if nested, ok := val.(map[string]any); ok {
 					parseSpecsInto(dst, nested)
 				}
@@ -58,7 +66,7 @@ func parseSpecsInto(dst map[string]any, v any) {
 				parseSpecsInto(dst, val)
 				continue
 			}
-			if s := stringifySpecValue(val); s != "" && !isReservedProductKey(lk) {
+			if s := stringifySpecValue(val); s != "" && !isReservedProductKey(lk) && !isInvalidAttributeKey(k) {
 				dst[SanitizeOutput(k)] = s
 			}
 		}
@@ -105,15 +113,110 @@ func looksLikeAttrMap(m map[string]any) bool {
 }
 
 func isReservedProductKey(k string) bool {
-	switch k {
-	case "name", "title", "description", "gtin", "ean", "brand", "category",
-		"price", "image", "stock", "eprel_id", "specifications", "raw", "mapped":
+	switch strings.ToLower(strings.TrimSpace(k)) {
+	case "name", "title", "description", "gtin", "ean", "category", "category_unique_id",
+		"price", "purchaseprice", "purchase_price", "sellingprice", "selling_price",
+		"image", "main_image", "mainimage", "moreimages", "more_images", "images",
+		"image_url", "imageurl", "image_link", "imagelink", "additional_image_urls",
+		"additional_image_link", "videourl", "video_url",
+		"stock", "stockstatus", "stock_status", "availability",
+		"id", "sku", "officiallink", "official_link", "service",
+		"specifications", "specs", "specification",
+		"raw", "mapped", "search":
 		return true
 	default:
 		return false
 	}
 }
 
+// isInvalidAttributeKey rejects garbage keys from bad feed specs (e.g. ":").
+func isInvalidAttributeKey(k string) bool {
+	k = strings.TrimSpace(k)
+	if k == "" || k == ":" || k == "=" || k == "-" || k == "_" {
+		return true
+	}
+	// Must contain at least one letter after sanitize.
+	hasLetter := false
+	for _, r := range strings.ToLower(k) {
+		if r >= 'a' && r <= 'z' {
+			hasLetter = true
+			break
+		}
+	}
+	return !hasLetter
+}
+
+// SanitizeProductAttributes keeps characteristic attrs for API/storage and drops
+// core product fields that belong on the V1 item root (title, description, images, …).
+func SanitizeProductAttributes(attrs map[string]any) map[string]any {
+	return sanitizeProductAttributes(attrs, false)
+}
+
+// SanitizeV1ProcessAttributes is the stricter poll projection: also drops eprel_*
+// keys (those are exposed on item.eprel).
+func SanitizeV1ProcessAttributes(attrs map[string]any) map[string]any {
+	return sanitizeProductAttributes(attrs, true)
+}
+
+func sanitizeProductAttributes(attrs map[string]any, dropEPREL bool) map[string]any {
+	if len(attrs) == 0 {
+		return map[string]any{}
+	}
+	out := make(map[string]any, len(attrs))
+	for k, v := range attrs {
+		key := strings.TrimSpace(SanitizeOutput(k))
+		if key == "" || isReservedProductKey(key) || isInvalidAttributeKey(key) {
+			continue
+		}
+		if dropEPREL && strings.HasPrefix(strings.ToLower(key), "eprel") {
+			continue
+		}
+		s := stringifySpecValue(v)
+		if s == "" || s == "<nil>" {
+			continue
+		}
+		// Prefer canonical dimension/model keys when aliases collide.
+		canon := canonicalizeAttrKey(key)
+		if canon == "" || isReservedProductKey(canon) || isInvalidAttributeKey(canon) {
+			continue
+		}
+		if dropEPREL && strings.HasPrefix(strings.ToLower(canon), "eprel") {
+			continue
+		}
+		if isDimensionKey(canon) && isZeroishString(s) {
+			continue
+		}
+		if _, exists := out[canon]; exists && (canon != key) {
+			continue
+		}
+		out[canon] = SanitizeOutput(s)
+	}
+	return out
+}
+
+func canonicalizeAttrKey(k string) string {
+	compact := strings.ToLower(strings.TrimSpace(k))
+	compact = strings.ReplaceAll(compact, "-", "_")
+	compact = strings.ReplaceAll(compact, " ", "_")
+	noUnderscore := strings.ReplaceAll(compact, "_", "")
+	switch noUnderscore {
+	case "netwidth", "width", "sirina":
+		return "width"
+	case "netheight", "height", "visina":
+		return "height"
+	case "netdepth", "depth", "globina":
+		return "depth"
+	case "netmass", "weight", "mass", "teza":
+		return "weight"
+	case "productmodel", "model":
+		return "product_model"
+	case "energijskirazred", "energyclass":
+		return "energy_class"
+	default:
+		return compact
+	}
+}
+
 func parseHTMLSpecs(dst map[string]any, s string) {
 	s = capSpecInput(s)
 	matches := htmlLiRe.FindAllStringSubmatch(s, maxSpecPairs)
@@ -126,8 +229,9 @@ func parseHTMLSpecs(dst map[string]any, s string) {
 			continue
 		}
 		key, val := splitLabelValue(text)
-		if key != "" && val != "" {
-			dst[attributeKeyFromLabel(key)] = SanitizeOutput(val)
+		attrKey := attributeKeyFromLabel(key)
+		if attrKey != "" && val != "" && !isReservedProductKey(attrKey) && !isInvalidAttributeKey(attrKey) {
+			dst[attrKey] = SanitizeOutput(val)
 		}
 	}
 	if len(matches) == 0 {
@@ -198,10 +302,10 @@ func parseCSVLikeSpecs(dst map[string]any, s string) {
 			if len(m) < 3 {
 				continue
 			}
-			key := strings.TrimSpace(m[1])
+			key := attributeKeyFromLabel(strings.TrimSpace(m[1]))
 			val := strings.TrimSpace(m[2])
-			if key != "" && val != "" {
-				dst[attributeKeyFromLabel(key)] = SanitizeOutput(val)
+			if key != "" && val != "" && !isReservedProductKey(key) && !isInvalidAttributeKey(key) {
+				dst[key] = SanitizeOutput(val)
 			}
 		}
 	}