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

package processing

import (
	"strings"
	"testing"
)

func TestIsWeakPriorEnhanceDescription_fillerPhrases(t *testing.T) {
	title := "Vogel's WALL 3245 TV Wall Mount"
	cases := []struct {
		desc string
		want bool
	}{
		{"Ready for retail listing with extra padding text here.", true},
		{title + " in the Mounts category. Ready for retail listing.", true},
		{"Product description placeholder text that is long enough to pass length.", true},
		{"A product in the Mounts category based on available specifications.", true},
		{"Product with available specifications ready for retail listing.", true},
		{"Generic one-liner without product tokens and enough length!!!!!", true},
		{
			"Vogel's WALL 3245 is a TV Mounts product from Vogel's. Key specs: width 45 cm, max_load 40 kg.",
			false,
		},
		{
			"This durable retail mount includes mounting hardware, load ratings, and install guidance for wall displays.",
			false,
		},
	}
	for _, tc := range cases {
		got := isWeakPriorEnhanceDescription(tc.desc, title)
		if got != tc.want {
			t.Fatalf("desc=%q got weak=%v want %v", tc.desc, got, tc.want)
		}
	}
	// Long non-filler copy that does not share tokens with this title is still usable
	// once past the short-boilerplate window (hash-skip quality floor).
	longUnrelated := strings.Repeat("Detailed retail copy covering materials finishes warranty and care. ", 3)
	if isWeakPriorEnhanceDescription(longUnrelated, title) {
		t.Fatalf("long non-filler unrelated copy should not be weak solely for token miss: %q", longUnrelated)
	}
	if !isWeakPriorEnhanceDescription("ok", title) {
		t.Fatal("too-short must be weak")
	}
	if !isWeakPriorEnhanceDescription(title, title) {
		t.Fatal("title-echo must be weak")
	}
}

func TestSynthesizeProductDescription_formulaHTML(t *testing.T) {
	t.Parallel()
	title := "GIGABYTE GS27QC"
	attrs := map[string]any{"brand": "GIGABYTE", "width": "61 cm"}
	tpl := map[string]any{
		"sections": []any{
			map[string]any{"type": "h1", "instructions": "Main heading"},
			map[string]any{"type": "p", "instructions": "Intro"},
			map[string]any{"type": "ul", "instructions": "Specs"},
		},
	}
	got := synthesizeProductDescription(title, "Gaming monitorji", "sl", attrs, tpl)
	if !strings.Contains(got, "<h1>") || !strings.Contains(got, "<p>") || !strings.Contains(got, "<ul>") {
		t.Fatalf("expected HTML skeleton from formula, got %q", got)
	}
	if !strings.Contains(got, title) {
		t.Fatalf("expected title in formula synth: %q", got)
	}
	plain := synthesizeProductDescription(title, "Gaming monitorji", "sl", attrs, nil)
	if strings.Contains(plain, "<h1>") {
		t.Fatalf("nil formula should stay plain: %q", plain)
	}
}

func TestSynthesizeDescriptionFromTitle_brandModelCategory(t *testing.T) {
	title := "Vogel's WALL 3245 TV Wall Mount"
	attrs := map[string]any{
		"brand":         "Vogel's",
		"product_model": "WALL 3245",
		"width":         "45 cm",
		"max_load":      "40 kg",
	}
	en := synthesizeDescriptionFromTitle(title, "TV Mounts", "en", attrs)
	if en == "" {
		t.Fatal("expected nonempty English factual fallback")
	}
	if strings.Contains(strings.ToLower(en), "ready for retail listing") {
		t.Fatalf("must not emit retail filler: %q", en)
	}
	if strings.Contains(strings.ToLower(en), "is a ") && strings.Contains(strings.ToLower(en), " product from ") {
		t.Fatalf("must not emit invent boilerplate: %q", en)
	}
	for _, need := range []string{"Vogel", "TV Mounts", "45 cm", "40 kg"} {
		if !strings.Contains(en, need) {
			t.Fatalf("English fallback missing %q: %q", need, en)
		}
	}
	if isWeakPriorEnhanceDescription(en, title) {
		t.Fatalf("factual fallback must not be weak: %q", en)
	}

	sl := synthesizeDescriptionFromTitle(title, "TV Mounts", "sl", attrs)
	if sl == "" || !strings.Contains(sl, "znamka") {
		t.Fatalf("expected Slovenian factual fallback, got %q", sl)
	}
	if strings.Contains(strings.ToLower(sl), "je izdelek v kategoriji") {
		t.Fatalf("must not emit invent boilerplate: %q", sl)
	}
	if strings.Contains(strings.ToLower(sl), "ready for retail listing") {
		t.Fatalf("SL fallback must not emit EN filler: %q", sl)
	}
	if isWeakPriorEnhanceDescription(sl, title) {
		t.Fatalf("SL factual fallback must not be weak: %q", sl)
	}
}

func TestInventHeuristicDescription_fromBrandModelCategory(t *testing.T) {
	title := "Vogel's WALL 3245 TV Wall Mount"
	user := ProductEnhanceUser("TV Mounts", title, "", map[string]any{
		"brand":         "Vogel's",
		"product_model": "WALL 3245",
		"width":         "45 cm",
		"max_load":      "40 kg",
	})
	system := "Write name and description in English. Return JSON with \"name\" and \"description\"."
	got := inventHeuristicDescription(system, user, title)
	if got == "" {
		t.Fatal("expected invent output")
	}
	t.Logf("TV mount invent sample: %s", got)
	if strings.Contains(strings.ToLower(got), "ready for retail listing") {
		t.Fatalf("invent must not return retail filler: %q", got)
	}
	for _, need := range []string{"Vogel", "TV Mounts", "45 cm"} {
		if !strings.Contains(got, need) {
			t.Fatalf("invent missing %q: %q", need, got)
		}
	}
	if isWeakPriorEnhanceDescription(got, title) {
		t.Fatalf("invent output must not be classified weak: %q", got)
	}

	slSystem := "Write name and description in Slovenian. Return JSON with \"name\"."
	sl := inventHeuristicDescription(slSystem, user, title)
	if !strings.Contains(sl, "znamka") {
		t.Fatalf("expected SL factual invent from system language, got %q", sl)
	}
}