xinmi/internal/application/service/wiki_real_windows_replay_test.go

package service

import (
	"encoding/json"
	"os"
	"testing"

	"github.com/stretchr/testify/require"
)

// Replays the fifteen real candidate payloads captured from one document's
// reading stage, in order, to reproduce the slug conflict offline.
func TestWikiRealWindowsReplay(t *testing.T) {
	path := os.Getenv("WIKI_REPLAY_FILE")
	if path == "" {
		t.Skip("WIKI_REPLAY_FILE not set")
	}
	raw, err := os.ReadFile(path)
	require.NoError(t, err)
	var texts []string
	require.NoError(t, json.Unmarshal(raw, &texts))

	merged := combinedExtraction{}
	for i, text := range texts {
		var part combinedExtraction
		require.NoError(t, json.Unmarshal([]byte(text), &part), "window %d", i)
		merged.Entities = append(merged.Entities, part.Entities...)
		merged.Concepts = append(merged.Concepts, part.Concepts...)
		t.Logf("window %2d: part E=%3d C=%3d | accumulated E=%3d C=%3d",
			i+1, len(part.Entities), len(part.Concepts), len(merged.Entities), len(merged.Concepts))
	}

	out, err := normalizeWikiFreshCandidates(merged)
	t.Logf("normalized: E=%d C=%d err=%v", len(out.Entities), len(out.Concepts), err)
	require.NoError(t, err, "normalization rejected the real candidate set")

	require.NoError(t, validateWikiSourceIdentities(out), "identity validation rejected the normalized set")
	slugs := map[string]string{}
	for _, item := range append(append([]extractedItem{}, out.Entities...), out.Concepts...) {
		if other, ok := slugs[item.Slug]; ok && other != item.Name {
			t.Fatalf("slug %q still shared by %q and %q", item.Slug, other, item.Name)
		}
		slugs[item.Slug] = item.Name
	}
	t.Logf("replay complete: %d identities, no collisions", len(slugs))
}