fix

apps/api/internal/company/weak_desc.go

@@ -25,25 +25,98 @@ var weakFillerPhrases = []string{
 
 // IsWeakPriorEnhanceDescription reports empty, too-short, title-echo, known filler,
 // or short boilerplate without overlapping tokens from title/fact sources.
+// Heuristic synthesize (invent) is intentionally excluded — use ShouldRefuseEnhanceHashSkip.
 func IsWeakPriorEnhanceDescription(priorDesc string, titles ...string) bool {
-	priorDesc = strings.TrimSpace(priorDesc)
-	if priorDesc == "" || priorDesc == "<nil>" {
-		return true
-	}
-	if len([]rune(priorDesc)) < minUsableProductDescRunes {
-		return true
-	}
-	for _, title := range titles {
-		if DescriptionEchoesTitle(priorDesc, title) {
+	reason := EnhanceHashSkipBlockReason(priorDesc, titles...)
+	return reason == "weak" || reason == "title-echo"
+}
+
+// heuristicSynthesizePhrases are distinctive invent / formula-skeleton snippets from
+// synthesizeDescriptionFromTitle / synthesizeDescriptionFromFormula. These may look
+// "strong" enough to pass IsWeakPriorEnhanceDescription but must never hash-skip
+// enhance (would leave fallback copy forever on reprocess).
+var heuristicSynthesizePhrases = []string{
+	"is a catalog product with the known attributes",
+	"is listed in the ",
+	". key specs:",
+	"je katalogski izdelek z znanimi atributi",
+	"je izdelek v kategoriji",
+	"je izdelek znamke",
+	". ključne specifikacije:",
+}
+
+// LooksLikeHeuristicSynthesize reports invent / formula-skeleton fallback copy.
+func LooksLikeHeuristicSynthesize(desc string) bool {
+	lower := strings.ToLower(desc)
+	for _, p := range heuristicSynthesizePhrases {
+		if p != "" && strings.Contains(lower, p) {
 			return true
 		}
 	}
-	if ContainsWeakFillerPhrase(priorDesc) {
+	// EN invent: "<title> is a <category> product from <brand>"
+	if strings.Contains(lower, " is a ") && strings.Contains(lower, " product from ") {
 		return true
 	}
+	return false
+}
+
+// EnhanceHashSkipBlockReason returns a stable reason when prior description must
+// not hash-skip the enhance LLM: "weak", "title-echo", "synth", or "".
+func EnhanceHashSkipBlockReason(priorDesc string, titles ...string) string {
+	priorDesc = strings.TrimSpace(priorDesc)
+	if priorDesc == "" || priorDesc == "<nil>" {
+		return "weak"
+	}
+	if len([]rune(priorDesc)) < minUsableProductDescRunes {
+		return "weak"
+	}
+	for _, title := range titles {
+		if DescriptionEchoesTitle(priorDesc, title) {
+			return "title-echo"
+		}
+	}
+	if ContainsWeakFillerPhrase(priorDesc) {
+		return "weak"
+	}
+	if LooksLikeHeuristicSynthesize(priorDesc) {
+		return "synth"
+	}
 	if len([]rune(priorDesc)) <= shortBoilerplateDescRunes &&
 		!descriptionOverlapsProductFacts(priorDesc, titles...) {
-		return true
+		return "weak"
+	}
+	return ""
+}
+
+// ShouldRefuseEnhanceHashSkip is true when a stored enhance_input_hash must be
+// ignored / cleared (weak, title-echo, or heuristic synthesize).
+func ShouldRefuseEnhanceHashSkip(priorDesc string, titles ...string) bool {
+	return EnhanceHashSkipBlockReason(priorDesc, titles...) != ""
+}
+
+// DescriptionMissingFormulaHTMLTags is true when sectionTypes require HTML tags
+// (h1/h2/h3/h4, p, ul) that are absent from desc — formula-mismatch for skip/clear.
+func DescriptionMissingFormulaHTMLTags(desc string, sectionTypes []string) bool {
+	if len(sectionTypes) == 0 {
+		return false
+	}
+	lower := strings.ToLower(desc)
+	for _, raw := range sectionTypes {
+		typ := strings.ToLower(strings.TrimSpace(raw))
+		switch typ {
+		case "h1", "h2", "h3", "h4":
+			if !strings.Contains(lower, "<"+typ) {
+				return true
+			}
+		case "ul":
+			if !strings.Contains(lower, "<ul") {
+				return true
+			}
+		case "p", "":
+			if !strings.Contains(lower, "<p") {
+				return true
+			}
+		}
 	}
 	return false
 }