fixes

apps/api/internal/processing/specs.go

@@ -13,6 +13,8 @@ var (
 	specsHTMLTagRe = regexp.MustCompile(`(?is)<[^>]+>`)
 	csvLikeRe      = regexp.MustCompile(`(?m)^\s*([^:=\n\r]{1,120})\s*[:=]\s*(.+?)\s*$`)
 	bulletLineRe   = regexp.MustCompile(`(?m)^\s*[-•*]\s*([^:=\n\r]{1,120})\s*[:=]\s*(.+?)\s*$`)
+	// GTIN/EAN/UPC masquerading as product_model (feed mapping mistakes).
+	barcodeLikeModelRe = regexp.MustCompile(`^\d{8,14}$`)
 )
 
 // Caps for locale/spec parsing — unbounded FindAll on huge CDATA can OOM the worker.
@@ -173,6 +175,73 @@ func SanitizeV1ProcessAttributesAllowed(attrs map[string]any, allowed map[string
 	return FilterAttributesByAllowed(sanitizeProductAttributes(attrs, true), allowed)
 }
 
+// AttrsForEnhance prepares attributes for AI enhance prompts / hashes.
+// Always drops reserved and invalid keys. When allowed is non-nil (including empty),
+// keeps only coreCharacteristicAttrKeys plus the allowlist (category_attributes).
+// When allowed is nil, returns sanitized attrs unchanged (unit-test fallback).
+func AttrsForEnhance(attrs map[string]any, allowed map[string]struct{}) map[string]any {
+	cleaned := sanitizeProductAttributes(attrs, true)
+	if allowed == nil {
+		return cleaned
+	}
+	cleaned = MapAttrsOntoAllowedKeys(cleaned, allowed)
+	return FilterAttributesByAllowed(cleaned, allowed)
+}
+
+// ensureEnergyClassFromEPREL copies eprel_energy_class → energy_class when the
+// core energy_class key is empty so AttrsForEnhance (which dropEPREL strips
+// eprel_*) still surfaces the label class to the model.
+func ensureEnergyClassFromEPREL(attrs map[string]any) {
+	if attrs == nil {
+		return
+	}
+	if v, ok := attrs["energy_class"]; ok {
+		if s := strings.TrimSpace(fmt.Sprint(v)); s != "" && s != "<nil>" {
+			return
+		}
+	}
+	for _, k := range []string{"eprel_energy_class", "eprelEnergyClass"} {
+		if v, ok := attrs[k]; ok {
+			s := strings.TrimSpace(fmt.Sprint(v))
+			if s != "" && s != "<nil>" {
+				attrs["energy_class"] = s
+				return
+			}
+		}
+	}
+}
+
+// AttrsForPersist prepares attributes for DB storage (attributes / processed_attributes).
+// Same category allowlist semantics as AttrsForEnhance, but uses SanitizeProductAttributes
+// (keeps eprel_* for poll extractEPRELFromAttrs) then FilterAttributesByAllowed.
+// Feed/spec keys are remapped onto category_attributes (formula) keys when possible so
+// values under near-miss labels (velikost-zaslona → diagonala_zaslona) are kept.
+// When allowed is nil, returns sanitized attrs unchanged (unit-test fallback).
+func AttrsForPersist(attrs map[string]any, allowed map[string]struct{}) map[string]any {
+	cleaned := SanitizeProductAttributes(attrs)
+	if allowed == nil {
+		return cleaned
+	}
+	cleaned = MapAttrsOntoAllowedKeys(cleaned, allowed)
+	filtered := FilterAttributesByAllowed(cleaned, allowed)
+	// Preserve eprel_* that SanitizeProductAttributes kept (not on category allowlists).
+	for k, v := range cleaned {
+		if strings.HasPrefix(strings.ToLower(strings.TrimSpace(k)), "eprel") {
+			filtered[k] = v
+		}
+	}
+	return filtered
+}
+
+// enhanceAllowedAttrKeys prefers category_attributes for the resolved category;
+// falls back to AllowedAttrKeys (nil = sanitize-only for unit tests).
+func enhanceAllowedAttrKeys(in ProductInput, categoryUID string) map[string]struct{} {
+	if in.CategoryAttrKeys != nil {
+		return allowedAttrKeysFromSets(in.CategoryAttrKeys, categoryUID)
+	}
+	return in.AllowedAttrKeys
+}
+
 // FilterAttributesByAllowed keeps coreCharacteristicAttrKeys plus keys present in
 // allowed (after canonicalizeAttrKey). When allowed is nil, returns attrs unchanged.
 func FilterAttributesByAllowed(attrs map[string]any, allowed map[string]struct{}) map[string]any {
@@ -216,6 +285,31 @@ func sanitizeProductAttributes(attrs map[string]any, dropEPREL bool) map[string]
 		if dropEPREL && strings.HasPrefix(strings.ToLower(key), "eprel") {
 			continue
 		}
+		// Keep nested eprel object for storage / extractEPRELFromAttrs (stringify would drop maps).
+		if !dropEPREL && strings.EqualFold(key, "eprel") {
+			if m, ok := v.(map[string]any); ok && len(m) > 0 {
+				nested := make(map[string]any, len(m))
+				for nk, nv := range m {
+					nk = strings.TrimSpace(nk)
+					if nk == "" || nv == nil {
+						continue
+					}
+					if s, ok := nv.(string); ok {
+						s = strings.TrimSpace(SanitizeOutput(s))
+						if s == "" {
+							continue
+						}
+						nested[nk] = s
+						continue
+					}
+					nested[nk] = nv
+				}
+				if len(nested) > 0 {
+					out["eprel"] = nested
+				}
+			}
+			continue
+		}
 		s := stringifySpecValue(v)
 		if s == "" || s == "<nil>" {
 			continue
@@ -231,6 +325,10 @@ func sanitizeProductAttributes(attrs map[string]any, dropEPREL bool) map[string]
 		if isDimensionKey(canon) && isZeroishString(s) {
 			continue
 		}
+		// Drop barcode-as-model (e.g. mapped product_model = GTIN).
+		if canon == "product_model" && barcodeLikeModelRe.MatchString(strings.TrimSpace(s)) {
+			continue
+		}
 		if _, exists := out[canon]; exists && (canon != key) {
 			continue
 		}
@@ -257,6 +355,8 @@ func canonicalizeAttrKey(k string) string {
 		return "product_model"
 	case "energijskirazred", "energyclass":
 		return "energy_class"
+	case "eprelid", "eprel":
+		return "eprel_id"
 	default:
 		return compact
 	}
@@ -459,4 +559,4 @@ func attributeKeyFromLabel(label string) string {
 		return ""
 	}
 	return slug
-}
+}