vaultik/internal/database/repository_edge_cases_test.go

//nolint:testpackage // inspects the unexported database connection
package database

import (
	"context"
	"database/sql"
	"fmt"
	"strings"
	"testing"
	"time"

	"sneak.berlin/go/vaultik/internal/types"
)

// fileEdgeCase describes one Create edge-case scenario.
type fileEdgeCase struct {
	name    string
	file    *File
	wantErr bool
	errMsg  string
}

// fileEdgeCases returns the Create edge-case table.
func fileEdgeCases() []fileEdgeCase {
	return []fileEdgeCase{
		{
			name: "empty path",
			file: &File{
				Path:  "",
				MTime: time.Now(),
				Size:  1024,
				Mode:  0644,
				UID:   1000,
				GID:   1000,
			},
			wantErr: false, // Empty strings are allowed, only NULL is not allowed
		},
		{
			name: "very long path",
			file: &File{
				Path:  types.FilePath("/" + strings.Repeat("a", 4096)),
				MTime: time.Now(),
				Size:  1024,
				Mode:  0644,
				UID:   1000,
				GID:   1000,
			},
			wantErr: false,
		},
		{
			name: "path with special characters",
			file: &File{
				//nolint:gosmopolitan // non-ASCII path is deliberate test data
				Path:  "/test/file with spaces and 特殊文字.txt",
				MTime: time.Now(),
				Size:  1024,
				Mode:  0644,
				UID:   1000,
				GID:   1000,
			},
			wantErr: false,
		},
		{
			name: "zero size file",
			file: &File{
				Path:  "/empty.txt",
				MTime: time.Now(),
				Size:  0,
				Mode:  0644,
				UID:   1000,
				GID:   1000,
			},
			wantErr: false,
		},
		{
			name: "symlink with target",
			file: &File{
				Path:       "/link",
				MTime:      time.Now(),
				Size:       0,
				Mode:       0777 | 0120000, // symlink mode
				UID:        1000,
				GID:        1000,
				LinkTarget: "/target",
			},
			wantErr: false,
		},
	}
}

// TestFileRepositoryEdgeCases tests edge cases for file repository
func TestFileRepositoryEdgeCases(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	t.Cleanup(cleanup)

	ctx := context.Background()
	repo := NewFileRepository(db)

	for i, tt := range fileEdgeCases() {
		t.Run(tt.name, func(t *testing.T) {
			t.Parallel()

			// Add a unique suffix to paths to avoid UNIQUE constraint violations
			if tt.file.Path != "" {
				tt.file.Path = types.FilePath(fmt.Sprintf("%s_%d_%d",
					tt.file.Path, i, time.Now().UnixNano()))
			}

			err := repo.Create(ctx, nil, tt.file)
			if (err != nil) != tt.wantErr {
				t.Errorf("Create() error = %v, wantErr %v", err, tt.wantErr)
			}

			if err != nil && tt.errMsg != "" && !strings.Contains(err.Error(), tt.errMsg) {
				t.Errorf("Create() error = %v, want error containing %q", err, tt.errMsg)
			}
		})
	}
}

// testDuplicateFilePaths exercises the UPSERT behavior for duplicate paths.
func testDuplicateFilePaths(t *testing.T, repos *Repositories) {
	t.Helper()

	ctx := context.Background()

	file1 := &File{
		Path:  "/duplicate.txt",
		MTime: time.Now(),
		Size:  1024,
		Mode:  0644,
		UID:   1000,
		GID:   1000,
	}
	file2 := &File{
		Path:  "/duplicate.txt", // Same path
		MTime: time.Now().Add(time.Hour),
		Size:  2048,
		Mode:  0644,
		UID:   1000,
		GID:   1000,
	}

	err := repos.Files.Create(ctx, nil, file1)
	if err != nil {
		t.Fatalf("failed to create file1: %v", err)
	}

	originalID := file1.ID

	// Create with same path should update the existing record (UPSERT behavior)
	err = repos.Files.Create(ctx, nil, file2)
	if err != nil {
		t.Fatalf("failed to create file2: %v", err)
	}

	// Verify the file was updated, not duplicated
	retrievedFile, err := repos.Files.GetByPath(ctx, "/duplicate.txt")
	if err != nil {
		t.Fatalf("failed to retrieve file: %v", err)
	}

	// The file should have been updated with file2's data
	if retrievedFile.Size != 2048 {
		t.Errorf("expected size 2048, got %d", retrievedFile.Size)
	}

	// ID might be different due to the UPSERT
	if retrievedFile.ID != file2.ID {
		t.Logf("File ID changed from %s to %s during upsert",
			originalID, retrievedFile.ID)
	}
}

// testDuplicateFileChunks exercises idempotent file-chunk mapping creation.
func testDuplicateFileChunks(t *testing.T, repos *Repositories) {
	t.Helper()

	ctx := context.Background()

	file := &File{
		Path:  "/test-dup-fc.txt",
		MTime: time.Now(),
		Size:  1024,
		Mode:  0644,
		UID:   1000,
		GID:   1000,
	}

	err := repos.Files.Create(ctx, nil, file)
	if err != nil {
		t.Fatal(err)
	}

	chunk := &Chunk{
		ChunkHash: types.ChunkHash("test-chunk-dup"),
		Size:      1024,
	}

	err = repos.Chunks.Create(ctx, nil, chunk)
	if err != nil {
		t.Fatal(err)
	}

	fc := &FileChunk{
		FileID:    file.ID,
		Idx:       0,
		ChunkHash: chunk.ChunkHash,
	}

	err = repos.FileChunks.Create(ctx, nil, fc)
	if err != nil {
		t.Fatal(err)
	}

	// Creating the same mapping again should be idempotent
	err = repos.FileChunks.Create(ctx, nil, fc)
	if err != nil {
		t.Error("file-chunk creation should be idempotent")
	}
}

// TestDuplicateHandling tests handling of duplicate entries
func TestDuplicateHandling(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	t.Cleanup(cleanup)

	ctx := context.Background()
	repos := NewRepositories(db)

	// Test duplicate file paths - Create uses UPSERT logic
	t.Run("duplicate file paths", func(t *testing.T) {
		t.Parallel()
		testDuplicateFilePaths(t, repos)
	})

	// Test duplicate chunk hashes
	t.Run("duplicate chunk hashes", func(t *testing.T) {
		t.Parallel()

		chunk := &Chunk{
			ChunkHash: types.ChunkHash("duplicate-chunk"),
			Size:      1024,
		}

		err := repos.Chunks.Create(ctx, nil, chunk)
		if err != nil {
			t.Fatalf("failed to create chunk: %v", err)
		}

		// Creating the same chunk again should be idempotent (ON CONFLICT DO NOTHING)
		err = repos.Chunks.Create(ctx, nil, chunk)
		if err != nil {
			t.Errorf("duplicate chunk creation should be idempotent, got error: %v", err)
		}
	})

	// Test duplicate file-chunk mappings
	t.Run("duplicate file-chunk mappings", func(t *testing.T) {
		t.Parallel()
		testDuplicateFileChunks(t, repos)
	})
}

// TestNullHandling tests handling of NULL values
func TestNullHandling(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	t.Cleanup(cleanup)

	ctx := context.Background()
	repos := NewRepositories(db)

	// Test file with no link target
	t.Run("file without link target", func(t *testing.T) {
		t.Parallel()

		file := &File{
			Path:       "/regular.txt",
			MTime:      time.Now(),
			Size:       1024,
			Mode:       0644,
			UID:        1000,
			GID:        1000,
			LinkTarget: "", // Should be stored as NULL
		}

		err := repos.Files.Create(ctx, nil, file)
		if err != nil {
			t.Fatal(err)
		}

		retrieved, err := repos.Files.GetByID(ctx, file.ID)
		if err != nil {
			t.Fatal(err)
		}

		if retrieved.LinkTarget != "" {
			t.Errorf("expected empty link target, got %q", retrieved.LinkTarget)
		}
	})

	// Test snapshot with NULL completed_at
	t.Run("incomplete snapshot", func(t *testing.T) {
		t.Parallel()

		snapshot := &Snapshot{
			ID:          "incomplete-test",
			Hostname:    internalTestHost,
			StartedAt:   time.Now(),
			CompletedAt: nil, // Should remain NULL until completed
		}

		err := repos.Snapshots.Create(ctx, nil, snapshot)
		if err != nil {
			t.Fatal(err)
		}

		retrieved, err := repos.Snapshots.GetByID(ctx, snapshot.ID.String())
		if err != nil {
			t.Fatal(err)
		}

		if retrieved.CompletedAt != nil {
			t.Error("expected nil CompletedAt for incomplete snapshot")
		}
	})

	// Test blob with NULL uploaded_ts
	t.Run("blob not uploaded", func(t *testing.T) {
		t.Parallel()
		verifyBlobNullUploadTS(ctx, t, repos)
	})
}

// verifyBlobNullUploadTS checks that a blob created without an upload
// timestamp round-trips with UploadedTS nil.
func verifyBlobNullUploadTS(
	ctx context.Context, t *testing.T, repos *Repositories,
) {
	t.Helper()

	blob := &Blob{
		ID:         types.NewBlobID(),
		Hash:       types.BlobHash("test-hash"),
		CreatedTS:  time.Now(),
		UploadedTS: nil, // Not uploaded yet
	}

	err := repos.Blobs.Create(ctx, nil, blob)
	if err != nil {
		t.Fatal(err)
	}

	retrieved, err := repos.Blobs.GetByID(ctx, blob.ID.String())
	if err != nil {
		t.Fatal(err)
	}

	if retrieved.UploadedTS != nil {
		t.Error("expected nil UploadedTS for non-uploaded blob")
	}
}

// createLargeDatasetFiles creates fileCount files and adds every other
// one to the snapshot, in one transaction as a backup writes them.
func createLargeDatasetFiles(
	t *testing.T,
	repos *Repositories,
	snapshotID string,
	fileCount int,
) {
	t.Helper()

	start := time.Now()

	err := repos.WithTx(context.Background(),
		func(ctx context.Context, tx *sql.Tx) error {
			for i := range fileCount {
				file := &File{
					Path:  types.FilePath(fmt.Sprintf("/large/file%05d.txt", i)),
					MTime: time.Now(),
					Size:  int64(i * 1024),
					Mode:  0644,
					UID:   uint32(1000 + (i % 10)),
					GID:   uint32(1000 + (i % 10)),
				}

				err := repos.Files.Create(ctx, tx, file)
				if err != nil {
					return fmt.Errorf("creating file %d: %w", i, err)
				}

				// Add half to snapshot
				if i%2 == 0 {
					err = repos.Snapshots.AddFileByID(ctx, tx, snapshotID, file.ID)
					if err != nil {
						return err
					}
				}
			}

			return nil
		})
	if err != nil {
		t.Fatal(err)
	}

	t.Logf("Created %d files in %v", fileCount, time.Since(start))
}

// TestLargeDatasets tests operations with large amounts of data
//
//nolint:tparallel // subtests share one database and are order-dependent
func TestLargeDatasets(t *testing.T) {
	t.Parallel()

	if testing.Short() {
		t.Skip("skipping large dataset test in short mode")
	}

	db, cleanup := setupTestDB(t)
	defer cleanup()

	ctx := context.Background()
	repos := NewRepositories(db)

	// Create a snapshot
	snapshot := &Snapshot{
		ID:        "large-dataset-test",
		Hostname:  internalTestHost,
		StartedAt: time.Now(),
	}

	err := repos.Snapshots.Create(ctx, nil, snapshot)
	if err != nil {
		t.Fatal(err)
	}

	// Create many files
	const fileCount = 1000

	//nolint:paralleltest // phases share one database and are order-dependent
	t.Run("create many files", func(t *testing.T) {
		createLargeDatasetFiles(t, repos, snapshot.ID.String(), fileCount)
	})

	// Test ListUnderPath performance
	//nolint:paralleltest // phases share one database and are order-dependent
	t.Run("list under path performance", func(t *testing.T) {
		start := time.Now()

		files, err := repos.Files.ListUnderPath(ctx, "/large/")
		if err != nil {
			t.Fatal(err)
		}

		if len(files) != fileCount {
			t.Errorf("expected %d files, got %d", fileCount, len(files))
		}

		t.Logf("Listed %d files in %v", len(files), time.Since(start))
	})

	// Test orphaned cleanup performance
	//nolint:paralleltest // phases share one database and are order-dependent
	t.Run("orphaned cleanup performance", func(t *testing.T) {
		start := time.Now()

		err := repos.Files.DeleteOrphaned(ctx)
		if err != nil {
			t.Fatal(err)
		}

		t.Logf("Cleaned up orphaned files in %v", time.Since(start))

		// Verify correct number remain
		files, err := repos.Files.ListUnderPath(ctx, "/large/")
		if err != nil {
			t.Fatal(err)
		}

		if len(files) != fileCount/2 {
			t.Errorf("expected %d files after cleanup, got %d",
				fileCount/2, len(files))
		}
	})
}

// TestErrorPropagation tests that errors are properly propagated
func TestErrorPropagation(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	t.Cleanup(cleanup)

	ctx := context.Background()
	repos := NewRepositories(db)

	// Test GetByID with non-existent ID
	t.Run("GetByID non-existent", func(t *testing.T) {
		t.Parallel()

		file, err := repos.Files.GetByID(ctx, types.NewFileID())
		if err != nil {
			t.Errorf("GetByID should not return error for non-existent ID, got: %v", err)
		}

		if file != nil {
			t.Error("expected nil file for non-existent ID")
		}
	})

	// Test GetByPath with non-existent path
	t.Run("GetByPath non-existent", func(t *testing.T) {
		t.Parallel()

		file, err := repos.Files.GetByPath(ctx, "/non/existent/path.txt")
		if err != nil {
			t.Errorf("GetByPath should not return error for non-existent path, got: %v",
				err)
		}

		if file != nil {
			t.Error("expected nil file for non-existent path")
		}
	})

	// Test invalid foreign key reference
	t.Run("invalid foreign key", func(t *testing.T) {
		t.Parallel()

		fc := &FileChunk{
			FileID:    types.NewFileID(),
			Idx:       0,
			ChunkHash: types.ChunkHash("some-chunk"),
		}

		err := repos.FileChunks.Create(ctx, nil, fc)
		if err == nil {
			t.Error("expected error for invalid foreign key")
		}

		if !strings.Contains(err.Error(), "FOREIGN KEY") {
			t.Errorf("expected foreign key error, got: %v", err)
		}
	})
}

// TestQueryInjection tests that the system is safe from SQL injection
func TestQueryInjection(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	t.Cleanup(cleanup)

	ctx := context.Background()
	repos := NewRepositories(db)

	// Test various injection attempts
	injectionTests := []string{
		"'; DROP TABLE files; --",
		"' OR '1'='1",
		"'; DELETE FROM files WHERE '1'='1'; --",
		`test'); DROP TABLE files; --`,
	}

	for _, injection := range injectionTests {
		t.Run("injection attempt", func(t *testing.T) {
			t.Parallel()

			// Try injection in file path
			file := &File{
				Path:  types.FilePath(injection),
				MTime: time.Now(),
				Size:  1024,
				Mode:  0644,
				UID:   1000,
				GID:   1000,
			}
			_ = repos.Files.Create(ctx, nil, file)
			// Should either succeed (treating as normal string) or fail with constraint
			// but should NOT execute the injected SQL

			// Verify tables still exist
			var count int

			err := db.conn.QueryRowContext(ctx, countFilesQuery).Scan(&count)
			if err != nil {
				t.Fatal("files table was damaged by injection")
			}
		})
	}
}

// TestTimezoneHandling tests that times are properly handled in UTC
func TestTimezoneHandling(t *testing.T) {
	t.Parallel()

	db, cleanup := setupTestDB(t)
	defer cleanup()

	ctx := context.Background()
	repos := NewRepositories(db)

	// Create file with specific timezone
	loc, err := time.LoadLocation("America/New_York")
	if err != nil {
		t.Skip("timezone not available")
	}

	nyTime := time.Now().In(loc)
	file := &File{
		Path:  "/timezone-test.txt",
		MTime: nyTime,
		Size:  1024,
		Mode:  0644,
		UID:   1000,
		GID:   1000,
	}

	err = repos.Files.Create(ctx, nil, file)
	if err != nil {
		t.Fatal(err)
	}

	// Retrieve and verify times are in UTC
	retrieved, err := repos.Files.GetByID(ctx, file.ID)
	if err != nil {
		t.Fatal(err)
	}

	// Check that times are equivalent (same instant)
	if !retrieved.MTime.Equal(nyTime) {
		t.Error("time was not preserved correctly")
	}

	// Check that retrieved time is in UTC
	if retrieved.MTime.Location() != time.UTC {
		t.Error("retrieved time is not in UTC")
	}
}