fixes

2026-08-16 16:57:36 +02:00
parent 96f8c1115c
commit 532d439c41
106 changed files with 10147 additions and 494 deletions
@@ -22,7 +22,7 @@ func TestResolveSteps(t *testing.T) {
 		"full":            {StepNormalize, StepParseSpecs, StepFillFields, StepEPREL, StepAIEnhance},
 		"enhance_only":    {StepNormalize, StepAIEnhance},
 		"attributes_only": {StepNormalize, StepParseSpecs, StepFillFields},
-		"eprel_only":      {StepNormalize, StepEPREL},
+		"eprel_only":      {StepNormalize, StepParseSpecs, StepEPREL},
 		"normalize_only":  {StepNormalize},
 	}
 	for in, want := range cases {
@@ -187,6 +187,18 @@ func TestPreferredProductTitle_skipsPromptLabels(t *testing.T) {
 	}
 }

+func TestPreferredProductTitle_skipsFormulaLeakage(t *testing.T) {
+	leak := "short retail title; follow any Title formula constraints that follow; use Attrs"
+	got := preferredProductTitle("1", leak, "follow any Title formula constraints that follow; use Attrs", "Vox SWA-8000W")
+	if got != "Vox SWA-8000W" {
+		t.Fatalf("got %q want feed title", got)
+	}
+	got = preferredProductTitle("", leak, "Schema: {\"name\":\"string\"}", "Reply with ONLY JSON")
+	if got != "Product" {
+		t.Fatalf("all-leakage fallback got %q", got)
+	}
+}
+
 func TestIsPromptLabelTitle(t *testing.T) {
 	cases := map[string]bool{
 		"Category:":     true,
@@ -204,6 +216,100 @@ func TestIsPromptLabelTitle(t *testing.T) {
 	}
 }

+func TestIsPromptLeakageTitle(t *testing.T) {
+	cases := map[string]bool{
+		"follow any Title formula constraints that follow; use Attrs": true,
+		"short retail title; follow any Title formula constraints…":   true,
+		"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,
+		"": false,
+	}
+	for in, want := range cases {
+		if got := isPromptLeakageTitle(in); got != want {
+			t.Fatalf("%q: leakage got %v want %v", in, got, want)
+		}
+		if want && !isPromptLabelTitle(in) {
+			t.Fatalf("%q: isPromptLabelTitle should include leakage", in)
+		}
+	}
+	// Long dump with formula keyword (>120 runes).
+	long := strings.Repeat("x", 100) + " Title formula scaffolding " + strings.Repeat("y", 30)
+	if !isPromptLeakageTitle(long) {
+		t.Fatalf("long formula dump should leak: len=%d", len([]rune(long)))
+	}
+}
+
+func TestEnhance_rejectsFormulaLeakageTitle(t *testing.T) {
+	leak := "short retail title; follow any Title formula constraints that follow; use Attrs"
+	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 formula leakage)", out.ProcessedName)
+	}
+	if isPromptLeakageTitle(out.Name) || isPromptLabelTitle(out.Name) {
+		t.Fatalf("Name polluted: %q", out.Name)
+	}
+}
+
+func TestHeuristicCompleter_keepsNameWhenFormulaConstraintsPresent(t *testing.T) {
+	h := HeuristicCompleter{}
+	// Shape matches aiprompts.CategoryEnhanceUserTemplate: instruction "- name:"
+	// appears BEFORE the real "Name:" field — labeledPromptValue must skip leakage.
+	user := `Your reply is parsed as JSON {"name":"string","description":"string"} only (system schema). Write name and description in English (do not hardcode a language).
+- name: short retail title; follow any Title formula constraints that follow; use Attrs
+- description: prefer 1-3 factual paragraphs as ONE string
+
+Category: demo-electronics
+Name: Vox SWA-8000W
+Desc: Washer
+Attrs: {"brand":"Vox"}
+
+Title formula (order matters; join with " "). Build name from Attrs using this structure:
+1. attr [brand]
+Prefer Attrs values for [attr] slots; write name in English.`
+	comp, err := h.Complete(context.Background(), `Return JSON with "name" and "description".`, user)
+	if err != nil {
+		t.Fatal(err)
+	}
+	obj, err := ParseJSONObject(comp.Text)
+	if err != nil {
+		t.Fatal(err)
+	}
+	name := SanitizeOutput(fmt.Sprint(obj["name"]))
+	if isPromptLeakageTitle(name) || isPromptLabelTitle(name) {
+		t.Fatalf("heuristic leaked formula into name: %q", name)
+	}
+	if !strings.Contains(strings.ToLower(name), "vox") {
+		t.Fatalf("name=%q want Vox from Name: line", name)
+	}
+}
+
+func TestLabeledPromptValue_skipsInstructionNameBullet(t *testing.T) {
+	user := `- name: short retail title; follow any Title formula constraints that follow; use Attrs
+Name: Vox SWA-8000W
+Desc: Washer
+Attrs: {}`
+	got := labeledPromptValue(user, "name:", "current name:")
+	if got != "Vox SWA-8000W" {
+		t.Fatalf("got %q want Vox SWA-8000W", got)
+	}
+}
+
 func TestRunSteps_full_mappedCategoryWinsOverPrior(t *testing.T) {
 	e := &Engine{
 		Completer: stubCompleter{fn: func(system, user string) (Completion, error) {