fix

apps/api/internal/company/weak_desc.go

@@ -32,9 +32,9 @@ func IsWeakPriorEnhanceDescription(priorDesc string, titles ...string) bool {
 }
 
 // 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).
+// synthesizeDescriptionFromTitle / synthesizeDescriptionFromFormula / factualDescriptionIntro.
+// 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 ",
@@ -43,23 +43,63 @@ var heuristicSynthesizePhrases = []string{
 	"je izdelek v kategoriji",
 	"je izdelek znamke",
 	". ključne specifikacije:",
+	// factualDescriptionIntro (post-phrase-avoidance invent)
+	" — znamka ",
+	", znamka ",
+	" — katalogski izdelek",
+	" — from ",
+	" — catalog product",
 }
 
 // LooksLikeHeuristicSynthesize reports invent / formula-skeleton fallback copy.
 func LooksLikeHeuristicSynthesize(desc string) bool {
-	lower := strings.ToLower(desc)
+	plain := strings.ToLower(plainTextForWeakCheck(desc))
+	if plain == "" {
+		return false
+	}
 	for _, p := range heuristicSynthesizePhrases {
-		if p != "" && strings.Contains(lower, p) {
+		if p != "" && strings.Contains(plain, p) {
 			return true
 		}
 	}
 	// EN invent: "<title> is a <category> product from <brand>"
-	if strings.Contains(lower, " is a ") && strings.Contains(lower, " product from ") {
+	if strings.Contains(plain, " is a ") && strings.Contains(plain, " product from ") {
+		return true
+	}
+	// SL/CS short invent: "<title> je … znamk|kategor|produkt …"
+	if strings.Contains(plain, " je ") &&
+		(strings.Contains(plain, "znamk") ||
+			strings.Contains(plain, "kategor") ||
+			strings.Contains(plain, "produkt") ||
+			strings.Contains(plain, "televiz") ||
+			strings.Contains(plain, "monitor")) {
 		return true
 	}
 	return false
 }
 
+// plainTextForWeakCheck strips HTML tags so <p>thin invent</p> is judged on visible copy.
+func plainTextForWeakCheck(s string) string {
+	s = strings.TrimSpace(s)
+	if s == "" {
+		return ""
+	}
+	var b strings.Builder
+	b.Grow(len(s))
+	inTag := false
+	for _, r := range s {
+		switch {
+		case r == '<':
+			inTag = true
+		case r == '>':
+			inTag = false
+		case !inTag:
+			b.WriteRune(r)
+		}
+	}
+	return strings.TrimSpace(b.String())
+}
+
 // 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 {
@@ -67,22 +107,26 @@ func EnhanceHashSkipBlockReason(priorDesc string, titles ...string) string {
 	if priorDesc == "" || priorDesc == "<nil>" {
 		return "weak"
 	}
-	if len([]rune(priorDesc)) < minUsableProductDescRunes {
+	plain := plainTextForWeakCheck(priorDesc)
+	if plain == "" || plain == "<nil>" {
+		return "weak"
+	}
+	if len([]rune(plain)) < minUsableProductDescRunes {
 		return "weak"
 	}
 	for _, title := range titles {
-		if DescriptionEchoesTitle(priorDesc, title) {
+		if DescriptionEchoesTitle(plain, title) || DescriptionEchoesTitle(priorDesc, title) {
 			return "title-echo"
 		}
 	}
-	if ContainsWeakFillerPhrase(priorDesc) {
+	if ContainsWeakFillerPhrase(plain) || ContainsWeakFillerPhrase(priorDesc) {
 		return "weak"
 	}
 	if LooksLikeHeuristicSynthesize(priorDesc) {
 		return "synth"
 	}
-	if len([]rune(priorDesc)) <= shortBoilerplateDescRunes &&
-		!descriptionOverlapsProductFacts(priorDesc, titles...) {
+	if len([]rune(plain)) <= shortBoilerplateDescRunes &&
+		!descriptionOverlapsProductFacts(plain, titles...) {
 		return "weak"
 	}
 	return ""

apps/api/internal/company/weak_desc_test.go

@@ -13,6 +13,17 @@ func TestLooksLikeHeuristicSynthesize(t *testing.T) {
 	if !LooksLikeHeuristicSynthesize(sl) {
 		t.Fatalf("expected SL invent detected: %q", sl)
 	}
+	htmlInvent := "<p>QLED TV 55 je televizor znamke Samsung.</p>"
+	if !LooksLikeHeuristicSynthesize(htmlInvent) {
+		t.Fatalf("expected HTML-wrapped invent detected: %q", htmlInvent)
+	}
+	if !ShouldRefuseEnhanceHashSkip(htmlInvent, "QLED TV 55") {
+		t.Fatalf("HTML invent must block hash skip")
+	}
+	factual := "QLED TV 55 — Televizorji, znamka Samsung."
+	if !LooksLikeHeuristicSynthesize(factual) {
+		t.Fatalf("expected factualDescriptionIntro invent detected: %q", factual)
+	}
 	good := "Vogel's WALL 3245 is a sturdy TV mount with clear load ratings and install guidance for wall displays."
 	if LooksLikeHeuristicSynthesize(good) {
 		t.Fatalf("retail copy must not look like invent: %q", good)