WeKnora deployment source with the reviewed wiki identity, summary and recovery fixes

internal/application/service/wiki_identity_targets_test.go

@@ -0,0 +1,132 @@
+package service
+
+import (
+	"context"
+	"testing"
+
+	"github.com/Tencent/WeKnora/internal/types"
+	"github.com/Tencent/WeKnora/internal/types/interfaces"
+	"github.com/stretchr/testify/require"
+)
+
+// fakeWikiPages serves the pages a knowledge base already owns, so identity
+// resolution can be exercised without a database.
+type fakeWikiPages struct {
+	interfaces.WikiPageService
+	bySlug map[string]*types.WikiPageLite
+}
+
+func (f fakeWikiPages) ListBySlugs(ctx context.Context, kbID string, slugs []string) (map[string]*types.WikiPageLite, error) {
+	out := make(map[string]*types.WikiPageLite, len(slugs))
+	for _, slug := range slugs {
+		if page := f.bySlug[slug]; page != nil {
+			out[slug] = page
+		}
+	}
+	return out, nil
+}
+
+func (f fakeWikiPages) FindPagesByNormalizedTitles(ctx context.Context, kbID, pageType string, identities []string) ([]*types.WikiPageLite, error) {
+	wanted := make(map[string]bool, len(identities))
+	for _, identity := range identities {
+		wanted[identity] = true
+	}
+	var out []*types.WikiPageLite
+	for _, page := range f.bySlug {
+		if page.PageType == pageType && wanted[normalizeWikiIdentityTitle(page.Title)] {
+			out = append(out, page)
+		}
+	}
+	return out, nil
+}
+
+func pageTitle(entityType, slug, title string) *types.WikiPageLite {
+	return &types.WikiPageLite{Slug: slug, Title: title, PageType: entityType}
+}
+
+// A model-written spelling variant must resolve onto the page that already owns the
+// slug instead of ending the document: window title "G 点" against page "G点".
+func TestWikiIdentityTargetsAdoptTheOwningPagesSpelling(t *testing.T) {
+	service := &wikiIngestService{wikiService: fakeWikiPages{bySlug: map[string]*types.WikiPageLite{
+		"entity/g-dian": pageTitle(types.WikiPageTypeEntity, "entity/g-dian", "G点"),
+	}}}
+	result := combinedExtraction{Entities: []extractedItem{{
+		Name: "G 点", Slug: "entity/g-dian", Description: "d", Details: "x",
+	}}}
+
+	out, dropped, err := service.resolveWikiIdentityTargets(context.Background(), "kb", result)
+	require.NoError(t, err)
+	require.Equal(t, 0, dropped, "a spelling-only difference is not a drop")
+	require.Len(t, out.Entities, 1)
+	require.Equal(t, "G点", out.Entities[0].Name, "the page keeps owning its canonical title")
+	require.Equal(t, "entity/g-dian", out.Entities[0].Slug)
+}
+
+// A different identity claiming an owned slug is dropped and counted, and the rest
+// of the document continues instead of failing.
+func TestWikiIdentityTargetsDropTheIntruderAndKeepTheRest(t *testing.T) {
+	service := &wikiIngestService{wikiService: fakeWikiPages{bySlug: map[string]*types.WikiPageLite{
+		"entity/du-lei-si": pageTitle(types.WikiPageTypeEntity, "entity/du-lei-si", "杜蕾斯"),
+	}}}
+	result := combinedExtraction{
+		Entities: []extractedItem{
+			{Name: "Durex", Slug: "entity/du-lei-si", Description: "d", Details: "x"},
+			{Name: "跳蛋", Slug: "entity/tiao-dan", Description: "d", Details: "x"},
+		},
+	}
+
+	out, dropped, err := service.resolveWikiIdentityTargets(context.Background(), "kb", result)
+	require.NoError(t, err)
+	require.Equal(t, 1, dropped)
+	require.Len(t, out.Entities, 1)
+	require.Equal(t, "跳蛋", out.Entities[0].Name)
+	require.Equal(t, 1, out.CandidatePruning.SlugConflicts, "the drop is counted for audit")
+}
+
+// A slug owned by another page type cannot be taken over silently.
+func TestWikiIdentityTargetsRejectACrossTypeSlug(t *testing.T) {
+	service := &wikiIngestService{wikiService: fakeWikiPages{bySlug: map[string]*types.WikiPageLite{
+		"entity/sm": pageTitle(types.WikiPageTypeConcept, "entity/sm", "SM"),
+	}}}
+	result := combinedExtraction{
+		Entities: []extractedItem{
+			{Name: "SM", Slug: "entity/sm", Description: "d", Details: "x"},
+			{Name: "跳蛋", Slug: "entity/tiao-dan", Description: "d", Details: "x"},
+		},
+	}
+
+	out, dropped, err := service.resolveWikiIdentityTargets(context.Background(), "kb", result)
+	require.NoError(t, err)
+	require.Equal(t, 1, dropped)
+	require.Len(t, out.Entities, 1)
+	require.Equal(t, "跳蛋", out.Entities[0].Name)
+}
+
+// A set that loses every candidate still fails closed rather than publishing nothing
+// under the pretence of a successful document.
+func TestWikiIdentityTargetsFailClosedWhenEverythingCollides(t *testing.T) {
+	service := &wikiIngestService{wikiService: fakeWikiPages{bySlug: map[string]*types.WikiPageLite{
+		"entity/sm": pageTitle(types.WikiPageTypeConcept, "entity/sm", "SM"),
+	}}}
+	result := combinedExtraction{Entities: []extractedItem{
+		{Name: "SM", Slug: "entity/sm", Description: "d", Details: "x"},
+	}}
+
+	_, _, err := service.resolveWikiIdentityTargets(context.Background(), "kb", result)
+	require.Error(t, err)
+	require.Contains(t, err.Error(), "wiki_identity_slug_conflict")
+}
+
+// Without a wiki repository the resolution is a pass-through, so Lite mode keeps
+// working and reports no drops.
+func TestWikiIdentityTargetsPassThroughWithoutRepository(t *testing.T) {
+	service := &wikiIngestService{}
+	result := combinedExtraction{Entities: []extractedItem{
+		{Name: "跳蛋", Slug: "entity/tiao-dan", Description: "d", Details: "x"},
+	}}
+
+	out, dropped, err := service.resolveWikiIdentityTargets(context.Background(), "kb", result)
+	require.NoError(t, err)
+	require.Equal(t, 0, dropped)
+	require.Len(t, out.Entities, 1)
+}