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

package processing

import (
	"context"
	"encoding/json"
	"strings"
	"testing"

	"github.com/descrybe/descrybe-v2/apps/api/internal/aiprompts"
	"github.com/descrybe/descrybe-v2/apps/api/internal/company"
)

// Regression: enhance used to stamp processed_* onto name/description, so Review
// always showed Matched — and formula invent ("Ključne lastnosti") overwrote the
// feed Original. Original must stay supplier; enriched must change.
func TestRunSteps_enhancePreservesOriginalVsEnriched(t *testing.T) {
	t.Parallel()
	prompt := aiprompts.SectionTitleStart + "\n" +
		`Napiši novo ime izdelka po formuli: tip znamka model barva.` + "\n" +
		aiprompts.SectionTitleEnd + "\n\n" +
		aiprompts.SectionDescriptionStart + "\n" +
		`<H2>{Naslov}</H2><p>{Odstavek}</p><ul><li>{Spec}</li></ul>` + "\n" +
		aiprompts.SectionDescriptionEnd

	feedName := "BRUNNER zložljiv stol za kampiranje ONE SHOT 0404164N.C20"
	feedDesc := "Predstavljajte si eleganten in izjemno lahek stol, zasnovan posebej za umetniške direktorje."
	newName := "zložljiv stol Brunner ONE SHOT siv"
	newDesc := "<h2>Lahek kamp stol</h2><p>" + strings.Repeat("Udobje na poti. ", 15) +
		"</p><ul><li>width: 0.44m</li><li>weight: 2.88kg</li></ul>"

	reply, _ := json.Marshal(map[string]any{
		"name":             newName,
		"description":      newDesc,
		"meta_title":       newName,
		"meta_description": "Brunner ONE SHOT zložljiv kamp stol",
		"attrs":            map[string]any{"brand": "BRUNNER"},
	})
	e := &Engine{
		Completer: stubCompleter{fn: func(_, _ string) (Completion, error) {
			return Completion{Text: string(reply), TotalTokens: 20, Model: "mock"}, nil
		}},
		Vector: NoopVectorCategorizer{},
	}
	descTpl := aiprompts.EffectiveDescriptionTemplateAny(nil, prompt)
	out, err := e.RunSteps(context.Background(), "co-preserve", ProductInput{
		GTIN:                  "8022068075495",
		Name:                  feedName,
		Description:           feedDesc,
		Mapped:                map[string]any{"name": feedName, "description": feedDesc, "brand": "BRUNNER", "category": "stolcki"},
		Language:              "sl",
		CategoryEnhancePrompt: prompt,
		DescriptionTemplate:   descTpl,
		CategoryNamesByUID:    map[string]string{"stolcki": "Stolčki"},
		CategoryUniqueID:      "stolcki",
		PriorCategory:         "stolcki",
	}, "enhance_only", nil, StepPolicy{AllowAI: true})
	if err != nil {
		t.Fatal(err)
	}
	if out.Name != feedName {
		t.Fatalf("Original name must stay feed, got %q", out.Name)
	}
	if out.Description != feedDesc {
		t.Fatalf("Original description must stay feed, got %q", out.Description)
	}
	if out.ProcessedName != newName {
		t.Fatalf("ProcessedName=%q want %q", out.ProcessedName, newName)
	}
	if !strings.Contains(out.ProcessedDescription, "<h2>") {
		t.Fatalf("ProcessedDescription missing formula HTML: %s", out.ProcessedDescription)
	}
	if out.ProcessedName == out.Name {
		t.Fatal("Review must show name Changed (enriched ≠ original)")
	}
	if out.ProcessedDescription == out.Description {
		t.Fatal("Review must show description Changed (enriched ≠ original)")
	}
}

func TestRunSteps_formulaInventDoesNotOverwriteOriginal(t *testing.T) {
	t.Parallel()
	prompt := aiprompts.SectionDescriptionStart + "\n" +
		`<H2>{Naslov}</H2><p>{Odstavek}</p><ul><li>{Spec}</li></ul>` + "\n" +
		aiprompts.SectionDescriptionEnd
	feedName := "BRUNNER ONE SHOT"
	feedDesc := "Predstavljajte si eleganten in izjemno lahek stol za umetniške direktorje."
	// Completer echoes supplier — formula gate synthesizes processed HTML.
	reply, _ := json.Marshal(map[string]any{
		"name":        feedName,
		"description": feedDesc,
	})
	e := &Engine{
		Completer: stubCompleter{fn: func(_, _ string) (Completion, error) {
			return Completion{Text: string(reply), TotalTokens: 8, Model: "mock"}, nil
		}},
		Vector: NoopVectorCategorizer{},
	}
	descTpl := aiprompts.EffectiveDescriptionTemplateAny(nil, prompt)
	out, err := e.RunSteps(context.Background(), "co-invent", ProductInput{
		Name:                  feedName,
		Description:           feedDesc,
		Mapped:                map[string]any{"name": feedName, "description": feedDesc, "brand": "BRUNNER", "width": "0.44m"},
		Language:              "sl",
		CategoryEnhancePrompt: prompt,
		DescriptionTemplate:   descTpl,
	}, "enhance_only", nil, StepPolicy{AllowAI: true})
	if err != nil {
		t.Fatal(err)
	}
	if out.Description != feedDesc {
		t.Fatalf("invent must not overwrite Original description, got %q", out.Description)
	}
	if !strings.Contains(strings.ToLower(out.ProcessedDescription), "ključne lastnosti") &&
		!company.LooksLikeHeuristicSynthesize(out.ProcessedDescription) &&
		!descriptionSatisfiesFormula(out.ProcessedDescription, descTpl) {
		t.Fatalf("expected processed formula/synth HTML, got %q", out.ProcessedDescription)
	}
	if out.ProcessedDescription == out.Description {
		t.Fatal("Original and Enriched must differ when invent fills processed only")
	}
}

func TestEnhanceHashForceReason_descCopyPaste(t *testing.T) {
	t.Parallel()
	long := "Supplier copy about the widget with enough characters to clear the weak-description floor for hash skip checks."
	in := ProductInput{
		Name:                      "Widget Pro Retail Title",
		Description:               long,
		PriorProcessedName:        "Widget Pro Retail Title Enriched",
		PriorProcessedDescription: long,
		DescriptionTemplate: map[string]any{
			"sections": []any{
				map[string]any{"type": "h2", "instructions": "x"},
				map[string]any{"type": "p", "instructions": "y"},
			},
		},
	}
	if got := enhanceHashForceReason(in); got != "formula-mismatch" {
		t.Fatalf("got %q want formula-mismatch", got)
	}
	in.PriorProcessedDescription = "<h2>A</h2><p>" + long + "</p>"
	in.Description = "<h2>A</h2><p>" + long + "</p>"
	if got := enhanceHashForceReason(in); got != "desc-copy-paste" {
		t.Fatalf("got %q want desc-copy-paste", got)
	}
}