descrybe/apps/api/internal/processing/v1_process_item_test.go

package processing

import (
	"fmt"
	"strings"
	"testing"
)

func TestScoreV1ProcessCompletedItem_processedStructure(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "123",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "Sony Headphones",
		"name":                 "Sony Headphones",
		"description":          "Sony Headphones deliver clear audio with noise cancelling for daily commuting.",
		"meta_title":           "Sony | Sony Headphones",
		"meta_description":     "Sony Headphones deliver clear audio with noise cancelling for daily commuting.",
		"category":             "48",
		"category_name":        "Slušalke",
		"attributes":           map[string]any{"brand": "Sony", "color": "Black"},
		"main_image":           "https://example.com/a.jpg",
		"more_images":          []string{"https://example.com/b.jpg"},
		"eprel":                nil,
	}
	sc := ScoreV1ProcessCompletedItem(item, ScoreV1ProcessItemOptions{MappedCategory: "48"})
	if !sc.OK {
		t.Fatalf("expected OK, flags=%v score=%d", sc.FailFlags, sc.Score)
	}
	if sc.Score < 90 {
		t.Fatalf("score=%d want >= 90", sc.Score)
	}
}

func TestScoreV1ProcessCompletedItem_flagsGaps(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "999",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "Mount",
		"description":          nil,
		"meta_title":           nil,
		"meta_description":     nil,
		"category":             nil,
		"category_name":        nil,
		"attributes": map[string]any{
			"brand":    "X",
			"eprel_id": "1",
			":":        "true",
		},
		"eprel": "not-an-object",
	}
	sc := ScoreV1ProcessCompletedItem(item, ScoreV1ProcessItemOptions{MappedCategory: "28"})
	if sc.OK {
		t.Fatal("expected failures")
	}
	joined := strings.Join(sc.FailFlags, ",")
	for _, want := range []string{
		"missing_name",
		"description_empty_with_title",
		"meta_title_missing",
		"meta_description_missing",
		"category_missing_though_mapped",
		"eprel_invalid_shape",
	} {
		if !strings.Contains(joined, want) {
			t.Fatalf("missing flag %q in %v", want, sc.FailFlags)
		}
	}
}

func TestEnforceV1ProcessCompletedItem_fillsDescriptionMetaSanitizes(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "1",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "Fiksni stenski TV nosilec 32''-55'' MANHATTAN",
		"description":          "Fiksni stenski TV nosilec 32''-55'' MANHATTAN from Ostalo. Ready for retail listing.",
		"meta_title":           nil,
		"meta_description":     nil,
		"category":             "28",
		"category_name":        "Nosilci za TV",
		"attributes": map[string]any{
			"brand":         "Ostalo",
			"product_model": "460934",
			"eprel_label":   "A",
			"name":          "should-drop",
			":":             "true",
		},
		"eprel": nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "sl", nil)
	desc, _ := out["description"].(string)
	if desc == "" {
		t.Fatal("expected nonempty description")
	}
	if descriptionEchoesTitle(desc, fmt.Sprint(out["title"])) {
		t.Fatalf("EnforceV1 must replace weak/echo desc, got title-echo: %q", desc)
	}
	if containsWeakFillerPhrase(desc) {
		t.Fatalf("filler left in description: %q", desc)
	}
	if strings.TrimSpace(fmt.Sprint(out["meta_title"])) == "" || out["meta_title"] == nil {
		t.Fatalf("meta_title empty: %v", out["meta_title"])
	}
	if strings.TrimSpace(fmt.Sprint(out["meta_description"])) == "" || out["meta_description"] == nil {
		t.Fatalf("meta_description empty: %v", out["meta_description"])
	}
	if containsWeakFillerPhrase(fmt.Sprint(out["meta_description"])) {
		t.Fatalf("meta_description still filler: %v", out["meta_description"])
	}
	attrs, ok := out["attributes"].(map[string]any)
	if !ok {
		t.Fatalf("attributes type=%T", out["attributes"])
	}
	if _, bad := attrs["eprel_label"]; bad {
		t.Fatalf("eprel_* should be stripped: %v", attrs)
	}
	if _, bad := attrs[":"]; bad {
		t.Fatalf("invalid key kept: %v", attrs)
	}
	if _, bad := attrs["name"]; bad {
		t.Fatalf("reserved name kept: %v", attrs)
	}
	if out["name"] != out["title"] {
		t.Fatalf("name should mirror title: name=%v title=%v", out["name"], out["title"])
	}
	sc := ScoreV1ProcessCompletedItem(out, ScoreV1ProcessItemOptions{MappedCategory: "28"})
	if !sc.OK {
		t.Fatalf("enforced item should score OK, flags=%v", sc.FailFlags)
	}
}

func TestEnforceV1ProcessCompletedItem_skipsTerminalStatuses(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{"ean": "1", "status": "not_found", "error": "missing"}
	out := EnforceV1ProcessCompletedItem(item, "", nil)
	if _, ok := out["title"]; ok {
		t.Fatalf("should not invent fields for not_found: %v", out)
	}
}

func TestEnforceV1ProcessCompletedItem_categoryFromCallerPreserved(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "2",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "Cooker",
		"description":          "",
		"category":             "50",
		"category_name":        "Štedilniki",
		"attributes":           map[string]any{"brand": "Vox"},
		"eprel":                nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "", nil)
	if out["category"] != "50" {
		t.Fatalf("category=%v", out["category"])
	}
	if out["category_name"] != "Štedilniki" {
		t.Fatalf("category_name=%v", out["category_name"])
	}
	desc, _ := out["description"].(string)
	if desc == "" {
		t.Fatal("expected synthesized description when title set")
	}
	if descriptionEchoesTitle(desc, "Cooker") {
		t.Fatalf("synthesized description must not equal title: %q", desc)
	}
	sc := ScoreV1ProcessCompletedItem(out, ScoreV1ProcessItemOptions{MappedCategory: "50"})
	if !sc.OK {
		t.Fatalf("flags=%v", sc.FailFlags)
	}
}

func TestEnforceV1ProcessCompletedItem_replacesTitleEchoDescription(t *testing.T) {
	t.Parallel()
	title := "VOX EBR700"
	item := V1ProcessJobItem{
		"ean":                  "8806095210711",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                title,
		"description":          title,
		"category":             "50",
		"category_name":        "Štedilniki",
		"attributes":           map[string]any{"brand": "VOX"},
		"eprel":                nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "sl", nil)
	desc, _ := out["description"].(string)
	if desc == "" {
		t.Fatal("expected synthesized description")
	}
	if descriptionEchoesTitle(desc, title) {
		t.Fatalf("must not leave title-echo description: %q", desc)
	}
	if !strings.Contains(desc, "Štedilniki") && !strings.Contains(desc, "VOX") {
		t.Fatalf("expected factual synth with category/brand: %q", desc)
	}
}

func TestEnforceV1ProcessCompletedItem_refreshesPoisonedMetaTitle(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "8806095210711",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "VOX EBR700",
		"description":          "VOX EBR700 is a Štedilniki product from VOX.",
		"meta_title":           "VOX EBR700 | 50",
		"meta_description":     "Ready for retail listing.",
		"category":             "50",
		"category_name":        "Štedilniki",
		"attributes": map[string]any{
			"brand":  "VOX",
			"zavora": "junk",
		},
		"eprel": nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "sl", nil)
	mt := fmt.Sprint(out["meta_title"])
	if isPoisonedMetaTitle(mt) || strings.Contains(mt, "| 50") {
		t.Fatalf("poisoned meta_title preserved: %q", mt)
	}
	if !strings.Contains(mt, "Štedilniki") {
		t.Fatalf("expected CategoryName in meta_title: %q", mt)
	}
	md := fmt.Sprint(out["meta_description"])
	if md == "" || md == "<nil>" || out["meta_description"] == nil {
		t.Fatalf("meta_description empty after poisoned title refresh: %v", out["meta_description"])
	}
	if containsWeakFillerPhrase(md) || isWeakPriorEnhanceDescription(md, "VOX EBR700") {
		t.Fatalf("stub meta_description preserved with poisoned title: %q", md)
	}
	attrs, ok := out["attributes"].(map[string]any)
	if !ok {
		t.Fatalf("attributes type=%T", out["attributes"])
	}
	if _, bad := attrs["zavora"]; bad {
		t.Fatalf("feed junk zavora leaked: %v", attrs)
	}
	if _, ok := attrs["brand"]; !ok {
		t.Fatalf("brand should remain: %v", attrs)
	}
}

func TestIsPoisonedMetaTitle(t *testing.T) {
	t.Parallel()
	if !isPoisonedMetaTitle("VOX EBR700 | 50") {
		t.Fatal("expected poisoned")
	}
	if !isPoisonedMetaTitle("Name |  7") {
		t.Fatal("expected poisoned with extra spaces")
	}
	if isPoisonedMetaTitle("VOX | Štedilniki") {
		t.Fatal("display name suffix must not count as poisoned")
	}
	if isPoisonedMetaTitle("") {
		t.Fatal("empty not poisoned")
	}
	if !isPoisonedMetaTitle("Sony | short retail title; follow any Title formula constr…") {
		t.Fatal("expected prompt-leakage meta_title poisoned")
	}
	if !isPoisonedMetaTitle("ANKER | short retail title; follow any Title formula constraints that follow") {
		t.Fatal("expected Title formula leakage poisoned")
	}
}

func TestEnforceV1ProcessCompletedItem_refreshesPromptLeakageMeta(t *testing.T) {
	t.Parallel()
	item := V1ProcessJobItem{
		"ean":                  "4548736132597",
		"status":               "processed",
		"id":                   "dddddddd-dddd-dddd-dddd-dddddddddddd",
		"processed_product_id": "dddddddd-dddd-dddd-dddd-dddddddddddd",
		"raw_product_id":       "eeeeeeee-eeee-eeee-eeee-eeeeeeeeeeee",
		"title":                "SONY BT slušalke WH1000XM5S sive",
		"description":          "SONY BT slušalke WH1000XM5S sive je izdelek v kategoriji Slušalke znamke Sony.",
		"meta_title":           "Sony | short retail title; follow any Title formula constr…",
		"meta_description":     "prefer 1-3 factual paragraphs as ONE string; limited HTML () is allowed if needed",
		"category":             "48",
		"category_name":        "Slušalke",
		"attributes": map[string]any{
			"brand": "Sony",
		},
		"eprel": nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "sl", nil)
	mt := fmt.Sprint(out["meta_title"])
	if isPoisonedMetaTitle(mt) || strings.Contains(strings.ToLower(mt), "short retail title") {
		t.Fatalf("prompt-leakage meta_title preserved: %q", mt)
	}
	if !strings.Contains(mt, "Slušalke") && !strings.Contains(mt, "SONY") {
		t.Fatalf("expected synth meta_title with product/category: %q", mt)
	}
	md := fmt.Sprint(out["meta_description"])
	if md == "" || md == "<nil>" || out["meta_description"] == nil {
		t.Fatalf("meta_description empty after leakage refresh: %v", out["meta_description"])
	}
	if isPromptLeakageTitle(md) || strings.Contains(strings.ToLower(md), "prefer 1-3") {
		t.Fatalf("prompt-leakage meta_description preserved: %q", md)
	}
}

func TestEnforceV1ProcessCompletedItem_preservesFormulaHTML(t *testing.T) {
	t.Parallel()
	htmlDesc := `<h1>Anker Soundcore Space One Pro</h1><p>Zložljive ANC slušalke z bogatim zvokom.</p><ul><li>Bluetooth 5.3</li></ul>`
	item := V1ProcessJobItem{
		"ean":                  "1",
		"status":               "processed",
		"id":                   "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"processed_product_id": "bbbbbbbb-bbbb-bbbb-bbbb-bbbbbbbbbbbb",
		"raw_product_id":       "cccccccc-cccc-cccc-cccc-cccccccccccc",
		"title":                "Anker Soundcore Space One Pro",
		"description":          htmlDesc,
		"category":             "48",
		"category_name":        "Slušalke",
		"attributes":           map[string]any{"brand": "Anker"},
		"eprel":                nil,
	}
	out := EnforceV1ProcessCompletedItem(item, "sl", nil)
	desc, _ := out["description"].(string)
	for _, tag := range []string{"<h1>", "<p>", "<ul>"} {
		if !strings.Contains(desc, tag) {
			t.Fatalf("EnforceV1 must keep formula HTML %q, got %q", tag, desc)
		}
	}
	// meta_description must stay plain (no tags).
	md := fmt.Sprint(out["meta_description"])
	if strings.Contains(md, "<h1>") || strings.Contains(md, "<p>") {
		t.Fatalf("meta_description should be plain, got %q", md)
	}
}