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

package processing

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

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

type stubEprel struct {
	enabled bool
	data    *eprel.Data
	err     error
	calls   int
	lastID  string
}

func (s *stubEprel) Enabled() bool { return s.enabled }

func (s *stubEprel) Fetch(_ context.Context, id string) (*eprel.Data, error) {
	s.calls++
	s.lastID = id
	return s.data, s.err
}

func TestRunSteps_EPREL_mergesAttributes(t *testing.T) {
	st := &stubEprel{
		enabled: true,
		data: &eprel.Data{
			ID:          "246834",
			Label:       "https://eprel.ec.europa.eu/api/product/246834/labels?format=png",
			PDF:         "https://eprel.ec.europa.eu/fiches/x.pdf",
			EnergyClass: "C",
			EnergyScale: "A-G",
		},
	}
	e := &Engine{EPREL: st, Completer: HeuristicCompleter{}, Vector: NoopVectorCategorizer{}}
	out, err := e.RunSteps(context.Background(), "co", ProductInput{
		Mapped: map[string]any{"name": "Fridge"},
		Raw:    map[string]any{"EPRELID": "246834"},
	}, "eprel_only", nil, StepPolicy{AllowAI: true, AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	if st.calls != 1 || st.lastID != "246834" {
		t.Fatalf("calls=%d id=%q", st.calls, st.lastID)
	}
	if out.ProcessedAttributes["eprel_id"] != "246834" {
		t.Fatalf("attrs=%v", out.ProcessedAttributes)
	}
	if out.ProcessedAttributes["eprel_energy_class"] != "C" {
		t.Fatalf("class missing: %v", out.ProcessedAttributes)
	}
	if out.ProcessedAttributes["energy_class"] != "C" {
		t.Fatalf("energy_class not promoted: %v", out.ProcessedAttributes)
	}
	if out.EPREL["id"] != "246834" || out.EPREL["label"] == "" {
		t.Fatalf("out.EPREL=%v", out.EPREL)
	}
}

func TestRunSteps_EPREL_disabledOrMissingID(t *testing.T) {
	e := &Engine{EPREL: eprel.Disabled{}, Completer: HeuristicCompleter{}, Vector: NoopVectorCategorizer{}}
	out, err := e.RunSteps(context.Background(), "co", ProductInput{
		Raw: map[string]any{"EPRELID": "1"},
	}, "eprel_only", nil, StepPolicy{AllowAI: true, AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	// Disabled enricher still records the discovered id; it must not fetch remote data.
	if out.ProcessedAttributes["eprel_energy_class"] != nil {
		t.Fatal("should not fetch energy class when disabled")
	}

	st := &stubEprel{enabled: true}
	e.EPREL = st
	_, err = e.RunSteps(context.Background(), "co", ProductInput{}, "eprel_only", nil, StepPolicy{AllowAI: true, AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	if st.calls != 0 {
		t.Fatal("should not call fetch without id")
	}
}

func TestRunSteps_EPREL_readsIDFromParsedAttrs(t *testing.T) {
	st := &stubEprel{
		enabled: true,
		data: &eprel.Data{
			ID:          "999001",
			Label:       "https://eprel.ec.europa.eu/api/product/999001/labels?format=png",
			PDF:         "https://eprel.ec.europa.eu/fiches/y.pdf",
			EnergyClass: "B",
			EnergyScale: "A-G",
		},
	}
	e := &Engine{EPREL: st, Completer: HeuristicCompleter{}, Vector: NoopVectorCategorizer{}}
	out, err := e.RunSteps(context.Background(), "co", ProductInput{
		Mapped: map[string]any{
			"name": "Washer",
			"specifications": map[string]any{
				"eprel_id": "999001",
				"brand":    "Samsung",
			},
		},
	}, "eprel_only", nil, StepPolicy{AllowAI: false, AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	if st.calls != 1 || st.lastID != "999001" {
		t.Fatalf("calls=%d id=%q", st.calls, st.lastID)
	}
	if out.EPREL["id"] != "999001" || out.EPREL["energy_class"] != "B" {
		t.Fatalf("eprel=%v", out.EPREL)
	}
	if out.ProcessedAttributes["energy_class"] != "B" {
		t.Fatalf("energy_class not promoted: %v", out.ProcessedAttributes)
	}
	if out.ProcessedAttributes["eprel_id"] != "999001" {
		t.Fatalf("eprel_id missing: %v", out.ProcessedAttributes)
	}
	// Nested eprel object must survive sanitize/persist for V1 extract.
	nested, _ := out.ProcessedAttributes["eprel"].(map[string]any)
	if nested == nil || nested["label"] == nil {
		t.Fatalf("nested eprel missing: %v", out.ProcessedAttributes["eprel"])
	}
	extracted := extractEPRELFromAttrs(out.ProcessedAttributes)
	em, ok := extracted.(map[string]any)
	if !ok || em["id"] != "999001" || em["label"] == nil || em["pdf"] == nil || em["energy_class"] != "B" {
		t.Fatalf("V1 extract=%v", extracted)
	}
}

func TestRunSteps_EPREL_skipsPlaceholderID(t *testing.T) {
	st := &stubEprel{enabled: true}
	e := &Engine{EPREL: st, Completer: HeuristicCompleter{}, Vector: NoopVectorCategorizer{}}
	_, err := e.RunSteps(context.Background(), "co", ProductInput{
		Mapped: map[string]any{"eprel_id": "0000"},
	}, "eprel_only", nil, StepPolicy{AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	if st.calls != 0 {
		t.Fatalf("placeholder id must not fetch, calls=%d", st.calls)
	}
}

func TestExtractEPRELFromAttrs_flatAndNested(t *testing.T) {
	got := extractEPRELFromAttrs(map[string]any{
		"eprel_id":           "1632113",
		"eprel_label":        "https://eprel.ec.europa.eu/api/product/1632113/labels?format=png",
		"eprel_pdf_url":      "https://eprel.ec.europa.eu/fiches/x.pdf",
		"eprel_energy_class": "A",
		"brand":              "Samsung",
	})
	m, ok := got.(map[string]any)
	if !ok {
		t.Fatalf("type=%T", got)
	}
	if m["id"] != "1632113" || m["label"] == nil || m["pdf"] == nil || m["energy_class"] != "A" {
		t.Fatalf("got=%v", m)
	}
	// Nested legacy aliases must fold onto public keys.
	got2 := extractEPRELFromAttrs(map[string]any{
		"eprel": map[string]any{
			"eprel_id":           "999",
			"eprel_label":        "https://eprel.example/label",
			"eprel_pdf":          "https://eprel.example/pdf",
			"eprel_energy_class": "B",
		},
	})
	m2, ok := got2.(map[string]any)
	if !ok || m2["id"] != "999" || m2["energy_class"] != "B" || m2["label"] == nil || m2["pdf"] == nil {
		t.Fatalf("nested aliases: %v", got2)
	}
	// V1 poll must strip eprel_* from attributes after extract.
	clean := SanitizeV1ProcessAttributesAllowed(map[string]any{
		"eprel_id":           "1632113",
		"eprel_label":        "https://eprel.example/label",
		"brand":              "Samsung",
		"eprel_energy_class": "A",
	}, map[string]struct{}{})
	if _, ok := clean["eprel_id"]; ok {
		t.Fatalf("eprel leaked into attributes: %v", clean)
	}
	if clean["brand"] != "Samsung" {
		t.Fatalf("brand missing: %v", clean)
	}
}

func TestRunSteps_EPREL_liveFetchIfNetwork(t *testing.T) {
	if testing.Short() {
		t.Skip("short")
	}
	client := eprel.NewClient(eprel.Options{Enabled: true, Timeout: 20 * time.Second})
	if !client.Enabled() {
		t.Fatal("client disabled")
	}
	e := &Engine{EPREL: client, Completer: HeuristicCompleter{}, Vector: NoopVectorCategorizer{}}
	out, err := e.RunSteps(context.Background(), "co", ProductInput{
		Mapped: map[string]any{
			"name":     "Samsung WW90CGC04DAELE",
			"eprel_id": "1632113",
		},
	}, "eprel_only", nil, StepPolicy{AllowEPREL: true})
	if err != nil {
		t.Fatal(err)
	}
	if out.EPREL["id"] != "1632113" {
		t.Fatalf("eprel=%v", out.EPREL)
	}
	if out.EPREL["label"] == nil || out.EPREL["label"] == "" {
		t.Fatalf("missing label: %v", out.EPREL)
	}
	if out.EPREL["pdf"] == nil || out.EPREL["pdf"] == "" {
		t.Fatalf("missing pdf: %v", out.EPREL)
	}
	if out.EPREL["energy_class"] == nil || out.EPREL["energy_class"] == "" {
		t.Fatalf("missing energy_class: %v", out.EPREL)
	}
	if out.ProcessedAttributes["eprel_id"] != "1632113" {
		t.Fatalf("attrs=%v", out.ProcessedAttributes)
	}
	eprelVal := extractEPRELFromAttrs(out.ProcessedAttributes)
	m, ok := eprelVal.(map[string]any)
	if !ok || m["id"] != "1632113" {
		t.Fatalf("extract=%v", eprelVal)
	}
	if m["label"] == nil || m["label"] == "" || m["pdf"] == nil || m["pdf"] == "" || m["energy_class"] == nil || m["energy_class"] == "" {
		t.Fatalf("extract missing fields: %v", m)
	}
	clean := SanitizeV1ProcessAttributesAllowed(out.ProcessedAttributes, map[string]struct{}{})
	for k := range clean {
		if strings.HasPrefix(strings.ToLower(k), "eprel") {
			t.Fatalf("eprel leaked into V1 attributes: %v", clean)
		}
	}
	sample, _ := json.MarshalIndent(map[string]any{
		"eprel":      m,
		"attributes": clean,
		"step_eprel": out.EPREL,
	}, "", "  ")
	t.Logf("sample eprel JSON:\n%s", sample)
}