libnovel/scraper/internal/storage/hybrid_integration_test.go

//go:build integration

// Integration tests for HybridStore (PocketBase + MinIO) end-to-end.
//
// Run with:
//
//	MINIO_ENDPOINT=localhost:9000 \
//	POCKETBASE_URL=http://localhost:8090 \
//	go test -v -tags integration -timeout 120s \
//	    github.com/libnovel/scraper/internal/storage
package storage

import (
	"context"
	"fmt"
	"strings"
	"testing"
	"time"

	"github.com/libnovel/scraper/internal/scraper"
)

// newTestHybridStore constructs a HybridStore from environment variables.
// Skips the test if either MINIO_ENDPOINT or POCKETBASE_URL is unset.
func newTestHybridStore(t *testing.T) *HybridStore {
	t.Helper()
	if ep := envOr("MINIO_ENDPOINT", ""); ep == "" {
		t.Skip("MINIO_ENDPOINT not set — skipping HybridStore integration test")
	}
	if u := envOr("POCKETBASE_URL", ""); u == "" {
		t.Skip("POCKETBASE_URL not set — skipping HybridStore integration test")
	}

	pbCfg := PocketBaseConfig{
		BaseURL:       envOr("POCKETBASE_URL", "http://localhost:8090"),
		AdminEmail:    envOr("POCKETBASE_ADMIN_EMAIL", "admin@libnovel.local"),
		AdminPassword: envOr("POCKETBASE_ADMIN_PASSWORD", "changeme123"),
	}
	minioCfg := MinioConfig{
		Endpoint:       envOr("MINIO_ENDPOINT", "localhost:9000"),
		AccessKey:      envOr("MINIO_ACCESS_KEY", "admin"),
		SecretKey:      envOr("MINIO_SECRET_KEY", "changeme123"),
		UseSSL:         envOr("MINIO_USE_SSL", "false") == "true",
		BucketChapters: envOr("MINIO_BUCKET_CHAPTERS", "libnovel-chapters"),
		BucketAudio:    envOr("MINIO_BUCKET_AUDIO", "libnovel-audio"),
	}

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	hs, err := NewHybridStore(ctx, pbCfg, minioCfg)
	if err != nil {
		t.Fatalf("NewHybridStore: %v", err)
	}
	return hs
}

// ─── Tests ────────────────────────────────────────────────────────────────────

// TestHybridStore_WriteReadMetadata exercises WriteMetadata → ReadMetadata round-trip.
func TestHybridStore_WriteReadMetadata(t *testing.T) {
	hs := newTestHybridStore(t)
	slug := testSlug(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	t.Cleanup(func() {
		cleanCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
		defer cancel()
		_ = hs.pb.pb.deleteWhere(cleanCtx, "books", fmt.Sprintf(`slug="%s"`, slug))
	})

	meta := scraper.BookMeta{
		Slug:          slug,
		Title:         "Hybrid Store Test Novel",
		Author:        "Test Author",
		Cover:         "https://example.com/cover.jpg",
		Status:        "Ongoing",
		Genres:        []string{"Fantasy", "Action"},
		Summary:       "A novel for integration testing.",
		TotalChapters: 99,
		SourceURL:     fmt.Sprintf("https://example.com/book/%s", slug),
		Ranking:       5,
	}

	t.Run("WriteMetadata", func(t *testing.T) {
		if err := hs.WriteMetadata(ctx, meta); err != nil {
			t.Fatalf("WriteMetadata: %v", err)
		}
		t.Logf("wrote metadata for slug=%q", slug)
	})

	t.Run("ReadMetadata", func(t *testing.T) {
		got, found, err := hs.ReadMetadata(ctx, slug)
		if err != nil {
			t.Fatalf("ReadMetadata: %v", err)
		}
		if !found {
			t.Fatal("ReadMetadata: not found after WriteMetadata")
		}
		t.Logf("read: %+v", got)
		if got.Title != meta.Title {
			t.Errorf("Title = %q, want %q", got.Title, meta.Title)
		}
		if got.Author != meta.Author {
			t.Errorf("Author = %q, want %q", got.Author, meta.Author)
		}
		if got.TotalChapters != meta.TotalChapters {
			t.Errorf("TotalChapters = %d, want %d", got.TotalChapters, meta.TotalChapters)
		}
		if got.Ranking != meta.Ranking {
			t.Errorf("Ranking = %d, want %d", got.Ranking, meta.Ranking)
		}
	})

	t.Run("MetadataMtime", func(t *testing.T) {
		mtime := hs.MetadataMtime(ctx, slug)
		if mtime == 0 {
			t.Error("MetadataMtime returned 0")
		}
		t.Logf("mtime: %d (%s)", mtime, time.Unix(mtime, 0))
	})

	t.Run("ReadMetadata_NotFound", func(t *testing.T) {
		_, found, err := hs.ReadMetadata(ctx, "this-slug-does-not-exist-xyz")
		if err != nil {
			t.Fatalf("ReadMetadata (miss): %v", err)
		}
		if found {
			t.Error("ReadMetadata returned found=true for a non-existent slug")
		}
	})
}

// TestHybridStore_WriteReadChapter exercises WriteChapter (MinIO blob + PocketBase
// index), ReadChapter, CountChapters, and ListChapters.
func TestHybridStore_WriteReadChapter(t *testing.T) {
	hs := newTestHybridStore(t)
	slug := testSlug(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	t.Cleanup(func() {
		cleanCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
		defer cancel()
		_ = hs.pb.pb.deleteWhere(cleanCtx, "chapters_idx", fmt.Sprintf(`slug="%s"`, slug))
		// MinIO objects are not cleaned up — they use the test slug as prefix
		// and are effectively isolated.
	})

	chapters := []scraper.Chapter{
		{
			Ref:  scraper.ChapterRef{Number: 1, Title: "Chapter 1: The Beginning", Volume: 0},
			Text: "The first chapter text with enough content to be meaningful for a real novel chapter.",
		},
		{
			Ref:  scraper.ChapterRef{Number: 2, Title: "Chapter 2: Rising Action", Volume: 0},
			Text: "The second chapter text continues the story from where the first left off.",
		},
		{
			Ref:  scraper.ChapterRef{Number: 3, Title: "Chapter 3: Climax", Volume: 0},
			Text: "The third chapter text reaches the peak of tension and conflict.",
		},
	}

	t.Run("WriteChapter", func(t *testing.T) {
		for _, ch := range chapters {
			if err := hs.WriteChapter(ctx, slug, ch); err != nil {
				t.Fatalf("WriteChapter(%d): %v", ch.Ref.Number, err)
			}
			t.Logf("wrote chapter %d", ch.Ref.Number)
		}
	})

	t.Run("ChapterExists", func(t *testing.T) {
		for _, ch := range chapters {
			if !hs.ChapterExists(ctx, slug, ch.Ref) {
				t.Errorf("ChapterExists(chapter %d) = false after WriteChapter", ch.Ref.Number)
			}
		}
		missing := scraper.ChapterRef{Number: 999, Volume: 0}
		if hs.ChapterExists(ctx, slug, missing) {
			t.Error("ChapterExists(999) = true for a chapter that was never written")
		}
	})

	t.Run("ReadChapter", func(t *testing.T) {
		for _, ch := range chapters {
			got, err := hs.ReadChapter(ctx, slug, ch.Ref.Number)
			if err != nil {
				t.Fatalf("ReadChapter(%d): %v", ch.Ref.Number, err)
			}
			// WriteChapter prepends "# <title>\n\n" and appends "\n".
			expectedPrefix := "# " + ch.Ref.Title
			if !strings.HasPrefix(got, expectedPrefix) {
				t.Errorf("chapter %d: content doesn't start with expected header\ngot:  %q\nwant prefix: %q",
					ch.Ref.Number, got[:min(len(got), 80)], expectedPrefix)
			}
			if !strings.Contains(got, ch.Text) {
				t.Errorf("chapter %d: content doesn't contain original text", ch.Ref.Number)
			}
			t.Logf("chapter %d: %d bytes", ch.Ref.Number, len(got))
		}
	})

	t.Run("CountChapters", func(t *testing.T) {
		count := hs.CountChapters(ctx, slug)
		if count != len(chapters) {
			t.Errorf("CountChapters = %d, want %d", count, len(chapters))
		}
	})

	t.Run("ListChapters", func(t *testing.T) {
		infos, err := hs.ListChapters(ctx, slug)
		if err != nil {
			t.Fatalf("ListChapters: %v", err)
		}
		if len(infos) != len(chapters) {
			t.Errorf("ListChapters returned %d entries, want %d", len(infos), len(chapters))
		}
		for i, info := range infos {
			t.Logf("infos[%d]: number=%d title=%q date=%q", i, info.Number, info.Title, info.Date)
		}
		// Verify sorted order.
		for i := 1; i < len(infos); i++ {
			if infos[i].Number <= infos[i-1].Number {
				t.Errorf("ListChapters not sorted: infos[%d].Number=%d <= infos[%d].Number=%d",
					i, infos[i].Number, i-1, infos[i-1].Number)
			}
		}
	})
}

// TestHybridStore_WriteReadRanking exercises WriteRanking → ReadRankingItems
// round-trip.
func TestHybridStore_WriteReadRanking(t *testing.T) {
	hs := newTestHybridStore(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	items := []RankingItem{
		{Rank: 1, Slug: "top-novel", Title: "Top Novel", Author: "Author A", Status: "Ongoing", SourceURL: "https://example.com/book/top-novel"},
		{Rank: 2, Slug: "second-novel", Title: "Second Novel", Author: "Author B", Genres: []string{"Action"}, Status: "Completed"},
		{Rank: 3, Slug: "third-novel", Title: "Third Novel"},
	}

	t.Run("WriteRanking", func(t *testing.T) {
		if err := hs.WriteRanking(ctx, items); err != nil {
			t.Fatalf("WriteRanking: %v", err)
		}
		t.Logf("wrote %d ranking items", len(items))
	})

	t.Run("ReadRankingItems", func(t *testing.T) {
		got, err := hs.ReadRankingItems(ctx)
		if err != nil {
			t.Fatalf("ReadRankingItems: %v", err)
		}
		if len(got) != len(items) {
			t.Errorf("ReadRankingItems returned %d items, want %d", len(got), len(items))
		}
		for i, item := range got {
			t.Logf("items[%d]: rank=%d slug=%q title=%q", i, item.Rank, item.Slug, item.Title)
		}
		if len(got) > 0 {
			if got[0].Rank != 1 {
				t.Errorf("items[0].Rank = %d, want 1", got[0].Rank)
			}
			if got[0].Title != "Top Novel" {
				t.Errorf("items[0].Title = %q, want %q", got[0].Title, "Top Novel")
			}
		}
	})

	t.Run("RankingFileInfo", func(t *testing.T) {
		fi, err := hs.RankingFileInfo(ctx)
		if err != nil {
			t.Fatalf("RankingFileInfo: %v", err)
		}
		if fi == nil {
			t.Fatal("RankingFileInfo returned nil")
		}
		if fi.ModTime().IsZero() {
			t.Error("RankingFileInfo.ModTime() is zero")
		}
		t.Logf("ranking file info: modtime=%s", fi.ModTime())
	})
}

// TestHybridStore_Progress exercises SetProgress → GetProgress → AllProgress →
// DeleteProgress via the HybridStore.
func TestHybridStore_Progress(t *testing.T) {
	hs := newTestHybridStore(t)
	slug := testSlug(t)
	const sessionID = "hybrid-test-session-abc"

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	t.Cleanup(func() {
		cleanCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
		defer cancel()
		_ = hs.pb.pb.deleteWhere(cleanCtx, "progress",
			fmt.Sprintf(`session_id="%s"`, sessionID))
	})

	p := ReadingProgress{Slug: slug, Chapter: 7, UpdatedAt: time.Now()}

	t.Run("SetProgress", func(t *testing.T) {
		if err := hs.SetProgress(ctx, sessionID, p); err != nil {
			t.Fatalf("SetProgress: %v", err)
		}
	})

	t.Run("GetProgress", func(t *testing.T) {
		got, ok := hs.GetProgress(ctx, sessionID, slug)
		if !ok {
			t.Fatal("GetProgress: not found after SetProgress")
		}
		if got.Chapter != 7 {
			t.Errorf("Chapter = %d, want 7", got.Chapter)
		}
		if got.Slug != slug {
			t.Errorf("Slug = %q, want %q", got.Slug, slug)
		}
		t.Logf("progress: chapter=%d slug=%q updated=%s", got.Chapter, got.Slug, got.UpdatedAt)
	})

	t.Run("AllProgress", func(t *testing.T) {
		all, err := hs.AllProgress(ctx, sessionID)
		if err != nil {
			t.Fatalf("AllProgress: %v", err)
		}
		found := false
		for _, item := range all {
			if item.Slug == slug {
				found = true
			}
		}
		if !found {
			t.Errorf("AllProgress did not contain slug %q (total=%d)", slug, len(all))
		}
	})

	t.Run("DeleteProgress", func(t *testing.T) {
		if err := hs.DeleteProgress(ctx, sessionID, slug); err != nil {
			t.Fatalf("DeleteProgress: %v", err)
		}
		_, ok := hs.GetProgress(ctx, sessionID, slug)
		if ok {
			t.Error("GetProgress returned ok=true after DeleteProgress")
		}
	})
}

// TestHybridStore_PresignChapter writes a chapter to MinIO via HybridStore,
// then calls PresignChapter and verifies a non-empty URL is returned.
func TestHybridStore_PresignChapter(t *testing.T) {
	hs := newTestHybridStore(t)
	slug := testSlug(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	ch := scraper.Chapter{
		Ref:  scraper.ChapterRef{Number: 1, Title: "Chapter 1: Presign Test", Volume: 0},
		Text: "Text for the presign chapter test.",
	}

	if err := hs.WriteChapter(ctx, slug, ch); err != nil {
		t.Fatalf("WriteChapter: %v", err)
	}

	url, err := hs.PresignChapter(ctx, slug, 1, 10*time.Minute)
	if err != nil {
		t.Fatalf("PresignChapter: %v", err)
	}
	if url == "" {
		t.Fatal("PresignChapter returned empty URL")
	}
	if !strings.HasPrefix(url, "http") {
		t.Errorf("PresignChapter URL does not start with http: %q", url)
	}
	t.Logf("presigned chapter URL: %s", url)
}

// TestHybridStore_PresignAudio puts a fake audio blob into MinIO via the
// underlying MinioClient and verifies PresignAudio returns a valid URL.
func TestHybridStore_PresignAudio(t *testing.T) {
	hs := newTestHybridStore(t)
	slug := testSlug(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	key := hs.AudioObjectKey(slug, 1, "af_bella", 1.0)
	fakeAudio := []byte("ID3\x03\x00\x00\x00\x00\x00\x00hybrid-presign-audio-test")

	if err := hs.minio.PutAudio(ctx, key, fakeAudio); err != nil {
		t.Fatalf("PutAudio: %v", err)
	}

	url, err := hs.PresignAudio(ctx, key, 10*time.Minute)
	if err != nil {
		t.Fatalf("PresignAudio: %v", err)
	}
	if url == "" {
		t.Fatal("PresignAudio returned empty URL")
	}
	if !strings.HasPrefix(url, "http") {
		t.Errorf("PresignAudio URL does not start with http: %q", url)
	}
	t.Logf("presigned audio URL: %s", url)
}

// TestHybridStore_AudioCache exercises SetAudioCache → GetAudioCache via HybridStore.
func TestHybridStore_AudioCache(t *testing.T) {
	hs := newTestHybridStore(t)

	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	cacheKey := fmt.Sprintf("hybrid-audio-test-%d", time.Now().UnixMilli())
	const filename = "speech_hybrid123.mp3"

	t.Cleanup(func() {
		cleanCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
		defer cancel()
		_ = hs.pb.pb.deleteWhere(cleanCtx, "audio_cache",
			fmt.Sprintf(`cache_key="%s"`, cacheKey))
	})

	if err := hs.SetAudioCache(ctx, cacheKey, filename); err != nil {
		t.Fatalf("SetAudioCache: %v", err)
	}

	got, ok := hs.GetAudioCache(ctx, cacheKey)
	if !ok {
		t.Fatal("GetAudioCache returned ok=false after SetAudioCache")
	}
	if got != filename {
		t.Errorf("filename = %q, want %q", got, filename)
	}
	t.Logf("audio cache: cacheKey=%q filename=%q", cacheKey, got)
}

// ─── helpers ──────────────────────────────────────────────────────────────────

func min(a, b int) int {
	if a < b {
		return a
	}
	return b
}