fix

2026-08-16 21:35:48 +02:00
parent 106f602232
commit 389608ea56
28 changed files with 2491 additions and 115 deletions
@@ -223,8 +223,12 @@ func TestIsPromptLeakageTitle(t *testing.T) {
 		"write name in {{language}}":                                  true,
 		"Schema: {\"name\":\"string\",\"description\":\"string\"}":    true,
 		"Reply with ONLY JSON (no markdown)":                          true,
-		"Vox SWA-8000W dishwasher":                                    false,
-		"Acme Widget Pro":                                             false,
+		"Emit description as ONE HTML string covering each section in order": true,
+		"Description formula (REQUIRED — overrides any shorter \"1-2 sentences\" rule)": true,
+		"when a Description formula follows, emit ONE HTML string": true,
+		"never copy name as description":                           true,
+		"Vox SWA-8000W dishwasher":                                 false,
+		"Acme Widget Pro":                                          false,
 		"": false,
 	}
 	for in, want := range cases {
@@ -242,6 +246,119 @@ func TestIsPromptLeakageTitle(t *testing.T) {
 	}
 }

+func TestIsUnusableCategoryValue_rejectsNameAndFormula(t *testing.T) {
+	t.Parallel()
+	title := "NOSILEC W53070 81-140CM 180ST VOGELS"
+	if !isUnusableCategoryValue(title, title) {
+		t.Fatal("product title must not be usable as category")
+	}
+	if !isUnusableCategoryValue("Title formula (order matters; join with \" \")", title) {
+		t.Fatal("title formula must not be usable as category")
+	}
+	if !isUnusableCategoryValue("Emit description as ONE HTML string covering each section", title) {
+		t.Fatal("description formula must not be usable as category")
+	}
+	if !isUnusableCategoryValue("Category:", title) {
+		t.Fatal("prompt label must not be usable as category")
+	}
+	if isUnusableCategoryValue("28", title) {
+		t.Fatal("taxonomy unique_id must remain usable")
+	}
+	if isUnusableCategoryValue("TV mounts", title) {
+		t.Fatal("real category display name must remain usable")
+	}
+}
+
+func TestScrubCategoryPollution_clearsNameAsCategory(t *testing.T) {
+	t.Parallel()
+	title := "Vogel's WALL 3245 TV Wall Mount"
+	out := StepResult{
+		Name:          title,
+		ProcessedName: title,
+		Category:      title,
+		CategoryName:  title,
+		FieldSources:  map[string]any{"category": "vector"},
+	}
+	scrubCategoryPollution(&out, nil, nil)
+	if out.Category != "" || out.CategoryName != "" {
+		t.Fatalf("expected category cleared, got Category=%q CategoryName=%q", out.Category, out.CategoryName)
+	}
+	if src, _ := out.FieldSources["category"].(string); src != "cleared_invalid" {
+		t.Fatalf("field_sources.category=%v want cleared_invalid", out.FieldSources["category"])
+	}
+}
+
+func TestEnhance_rejectsDescriptionFormulaLeakageTitle(t *testing.T) {
+	leak := "Emit description as ONE HTML string covering each section in order in Slovenian"
+	e := &Engine{
+		Completer: stubCompleter{fn: func(system, user string) (Completion, error) {
+			return Completion{Text: `{"name":"` + leak + `","description":"A solid washer."}`, TotalTokens: 2}, nil
+		}},
+		Vector: NoopVectorCategorizer{},
+	}
+	out, err := e.RunSteps(context.Background(), "co", ProductInput{
+		GTIN:   "8712285326882",
+		Name:   "Vox SWA-8000W",
+		Mapped: map[string]any{"name": "Vox SWA-8000W", "description": "Washer", "category": "demo-electronics"},
+	}, "enhance_only", nil, StepPolicy{AllowAI: true})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if out.ProcessedName != "Vox SWA-8000W" {
+		t.Fatalf("ProcessedName=%q want Vox SWA-8000W (reject description formula leakage)", out.ProcessedName)
+	}
+}
+
+func TestRunSteps_rejectsProductNameAsCategory(t *testing.T) {
+	title := "NOSILEC W53070 81-140CM 180ST VOGELS"
+	e := &Engine{
+		Completer: stubCompleter{fn: func(system, user string) (Completion, error) {
+			return Completion{Text: `{"name":"` + title + `","description":"TV wall mount for large screens."}`, TotalTokens: 2}, nil
+		}},
+		Vector: NoopVectorCategorizer{},
+	}
+	out, err := e.RunSteps(context.Background(), "co", ProductInput{
+		GTIN:          "8712285326882",
+		Name:          title,
+		PriorCategory: title,
+		Mapped:        map[string]any{"name": title, "description": title, "category": title},
+	}, "enhance_only", nil, StepPolicy{AllowAI: true})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if out.Category == title || strings.EqualFold(out.Category, out.ProcessedName) {
+		t.Fatalf("Category=%q must not equal product title", out.Category)
+	}
+	if categoryDisplayLabel(out) == title {
+		t.Fatalf("display category leaked product title: %q", categoryDisplayLabel(out))
+	}
+}
+
+func TestRunSteps_rejectsFormulaAsCategory(t *testing.T) {
+	leak := "Title formula (order matters; join with \" \"). Build name from Attrs"
+	e := &Engine{
+		Completer: stubCompleter{fn: func(system, user string) (Completion, error) {
+			return Completion{Text: `{"name":"Vox SWA-8000W","description":"Washer."}`, TotalTokens: 2}, nil
+		}},
+		Vector: NoopVectorCategorizer{},
+	}
+	out, err := e.RunSteps(context.Background(), "co", ProductInput{
+		GTIN:          "8712285326882",
+		Name:          "Vox SWA-8000W",
+		PriorCategory: leak,
+		Mapped:        map[string]any{"name": "Vox SWA-8000W", "description": "Washer", "category": leak},
+	}, "enhance_only", nil, StepPolicy{AllowAI: true})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if out.Category != "" {
+		t.Fatalf("Category=%q want empty (formula rejected)", out.Category)
+	}
+	if categoryDisplayLabel(out) != "" {
+		t.Fatalf("display=%q want empty", categoryDisplayLabel(out))
+	}
+}
+
 func TestEnhance_rejectsFormulaLeakageTitle(t *testing.T) {
 	leak := "short retail title; follow any Title formula constraints that follow; use Attrs"
 	e := &Engine{