파일 업로드 방식을 Chunking 방식으로 변경

src/lib/server/db/file.ts

@@ -15,8 +15,6 @@ interface Directory {
   encName: Ciphertext;
 }
 
-export type NewDirectory = Omit<Directory, "id">;
-
 interface File {
   id: number;
   parentId: DirectoryId;
@@ -28,15 +26,13 @@ interface File {
   hskVersion: number | null;
   contentHmac: string | null;
   contentType: string;
-  encContentIv: string;
+  encContentIv: string | null;
   encContentHash: string;
   encName: Ciphertext;
   encCreatedAt: Ciphertext | null;
   encLastModifiedAt: Ciphertext;
 }
 
-export type NewFile = Omit<File, "id">;
-
 interface FileCategory {
   id: number;
   parentId: CategoryId;
@@ -46,7 +42,7 @@ interface FileCategory {
   encName: Ciphertext;
 }
 
-export const registerDirectory = async (params: NewDirectory) => {
+export const registerDirectory = async (params: Omit<Directory, "id">) => {
   await db.transaction().execute(async (trx) => {
     const mek = await trx
       .selectFrom("master_encryption_key")
@@ -214,69 +210,41 @@ export const unregisterDirectory = async (userId: number, directoryId: number) =
     });
 };
 
-export const registerFile = async (params: NewFile) => {
+export const registerFile = async (trx: typeof db, params: Omit<File, "id">) => {
   if ((params.hskVersion && !params.contentHmac) || (!params.hskVersion && params.contentHmac)) {
     throw new Error("Invalid arguments");
   }
 
-  return await db.transaction().execute(async (trx) => {
-    const mek = await trx
-      .selectFrom("master_encryption_key")
-      .select("version")
-      .where("user_id", "=", params.userId)
-      .where("state", "=", "active")
-      .limit(1)
-      .forUpdate()
-      .executeTakeFirst();
-    if (mek?.version !== params.mekVersion) {
-      throw new IntegrityError("Inactive MEK version");
-    }
-
-    if (params.hskVersion) {
-      const hsk = await trx
-        .selectFrom("hmac_secret_key")
-        .select("version")
-        .where("user_id", "=", params.userId)
-        .where("state", "=", "active")
-        .limit(1)
-        .forUpdate()
-        .executeTakeFirst();
-      if (hsk?.version !== params.hskVersion) {
-        throw new IntegrityError("Inactive HSK version");
-      }
-    }
-
-    const { fileId } = await trx
-      .insertInto("file")
-      .values({
-        parent_id: params.parentId !== "root" ? params.parentId : null,
-        user_id: params.userId,
-        path: params.path,
-        master_encryption_key_version: params.mekVersion,
-        encrypted_data_encryption_key: params.encDek,
-        data_encryption_key_version: params.dekVersion,
-        hmac_secret_key_version: params.hskVersion,
-        content_hmac: params.contentHmac,
-        content_type: params.contentType,
-        encrypted_content_iv: params.encContentIv,
-        encrypted_content_hash: params.encContentHash,
-        encrypted_name: params.encName,
-        encrypted_created_at: params.encCreatedAt,
-        encrypted_last_modified_at: params.encLastModifiedAt,
-      })
-      .returning("id as fileId")
-      .executeTakeFirstOrThrow();
-    await trx
-      .insertInto("file_log")
-      .values({
-        file_id: fileId,
-        timestamp: new Date(),
-        action: "create",
-        new_name: params.encName,
-      })
-      .execute();
-    return { id: fileId };
-  });
+  const { fileId } = await trx
+    .insertInto("file")
+    .values({
+      parent_id: params.parentId !== "root" ? params.parentId : null,
+      user_id: params.userId,
+      path: params.path,
+      master_encryption_key_version: params.mekVersion,
+      encrypted_data_encryption_key: params.encDek,
+      data_encryption_key_version: params.dekVersion,
+      hmac_secret_key_version: params.hskVersion,
+      content_hmac: params.contentHmac,
+      content_type: params.contentType,
+      encrypted_content_iv: params.encContentIv,
+      encrypted_content_hash: params.encContentHash,
+      encrypted_name: params.encName,
+      encrypted_created_at: params.encCreatedAt,
+      encrypted_last_modified_at: params.encLastModifiedAt,
+    })
+    .returning("id as fileId")
+    .executeTakeFirstOrThrow();
+  await trx
+    .insertInto("file_log")
+    .values({
+      file_id: fileId,
+      timestamp: new Date(),
+      action: "create",
+      new_name: params.encName,
+    })
+    .execute();
+  return { id: fileId };
 };
 
 export const getAllFilesByParent = async (userId: number, parentId: DirectoryId) => {

src/lib/server/db/index.ts

@@ -5,6 +5,7 @@ export * as HskRepo from "./hsk";
 export * as MediaRepo from "./media";
 export * as MekRepo from "./mek";
 export * as SessionRepo from "./session";
+export * as UploadRepo from "./upload";
 export * as UserRepo from "./user";
 
 export * from "./error";

src/lib/server/db/migrations/1768062380-AddChunkedUpload.ts

@@ -0,0 +1,50 @@
+import { Kysely, sql } from "kysely";
+
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+export const up = async (db: Kysely<any>) => {
+  // file.ts
+  await db.schema
+    .alterTable("file")
+    .alterColumn("encrypted_content_iv", (col) => col.dropNotNull())
+    .execute();
+
+  // upload.ts
+  await db.schema
+    .createTable("upload_session")
+    .addColumn("id", "uuid", (col) => col.primaryKey().defaultTo(sql`gen_random_uuid()`))
+    .addColumn("user_id", "integer", (col) => col.references("user.id").notNull())
+    .addColumn("total_chunks", "integer", (col) => col.notNull())
+    .addColumn("uploaded_chunks", sql`integer[]`, (col) => col.notNull().defaultTo(sql`'{}'`))
+    .addColumn("expires_at", "timestamp(3)", (col) => col.notNull())
+    .addColumn("parent_id", "integer", (col) => col.references("directory.id"))
+    .addColumn("master_encryption_key_version", "integer", (col) => col.notNull())
+    .addColumn("encrypted_data_encryption_key", "text", (col) => col.notNull())
+    .addColumn("data_encryption_key_version", "timestamp(3)", (col) => col.notNull())
+    .addColumn("hmac_secret_key_version", "integer")
+    .addColumn("content_type", "text", (col) => col.notNull())
+    .addColumn("encrypted_name", "json", (col) => col.notNull())
+    .addColumn("encrypted_created_at", "json")
+    .addColumn("encrypted_last_modified_at", "json", (col) => col.notNull())
+    .addForeignKeyConstraint(
+      "upload_session_fk01",
+      ["user_id", "master_encryption_key_version"],
+      "master_encryption_key",
+      ["user_id", "version"],
+    )
+    .addForeignKeyConstraint(
+      "upload_session_fk02",
+      ["user_id", "hmac_secret_key_version"],
+      "hmac_secret_key",
+      ["user_id", "version"],
+    )
+    .execute();
+};
+
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
+export const down = async (db: Kysely<any>) => {
+  await db.schema.dropTable("upload_session").execute();
+  await db.schema
+    .alterTable("file")
+    .alterColumn("encrypted_content_iv", (col) => col.setNotNull())
+    .execute();
+};

src/lib/server/db/migrations/index.ts

@@ -1,9 +1,11 @@
 import * as Initial1737357000 from "./1737357000-Initial";
 import * as AddFileCategory1737422340 from "./1737422340-AddFileCategory";
 import * as AddThumbnail1738409340 from "./1738409340-AddThumbnail";
+import * as AddChunkedUpload1768062380 from "./1768062380-AddChunkedUpload";
 
 export default {
   "1737357000-Initial": Initial1737357000,
   "1737422340-AddFileCategory": AddFileCategory1737422340,
   "1738409340-AddThumbnail": AddThumbnail1738409340,
+  "1768062380-AddChunkedUpload": AddChunkedUpload1768062380,
 };

src/lib/server/db/schema/file.ts

@@ -30,7 +30,7 @@ interface FileTable {
   hmac_secret_key_version: number | null;
   content_hmac: string | null; // Base64
   content_type: string;
-  encrypted_content_iv: string; // Base64
+  encrypted_content_iv: string | null; // Base64
   encrypted_content_hash: string; // Base64
   encrypted_name: Ciphertext;
   encrypted_created_at: Ciphertext | null;

src/lib/server/db/schema/index.ts

@@ -5,6 +5,7 @@ export * from "./hsk";
 export * from "./media";
 export * from "./mek";
 export * from "./session";
+export * from "./upload";
 export * from "./user";
 export * from "./util";
 

src/lib/server/db/schema/upload.ts

@@ -0,0 +1,26 @@
+import type { Generated } from "kysely";
+import type { Ciphertext } from "./util";
+
+interface UploadSessionTable {
+  id: Generated<string>;
+  user_id: number;
+  total_chunks: number;
+  uploaded_chunks: Generated<number[]>;
+  expires_at: Date;
+
+  parent_id: number | null;
+  master_encryption_key_version: number;
+  encrypted_data_encryption_key: string; // Base64
+  data_encryption_key_version: Date;
+  hmac_secret_key_version: number | null;
+  content_type: string;
+  encrypted_name: Ciphertext;
+  encrypted_created_at: Ciphertext | null;
+  encrypted_last_modified_at: Ciphertext;
+}
+
+declare module "./index" {
+  interface Database {
+    upload_session: UploadSessionTable;
+  }
+}

src/lib/server/db/upload.ts

@@ -0,0 +1,122 @@
+import { sql } from "kysely";
+import { IntegrityError } from "./error";
+import db from "./kysely";
+import type { Ciphertext } from "./schema";
+
+interface UploadSession {
+  id: string;
+  userId: number;
+  totalChunks: number;
+  uploadedChunks: number[];
+  expiresAt: Date;
+
+  parentId: DirectoryId;
+  mekVersion: number;
+  encDek: string;
+  dekVersion: Date;
+  hskVersion: number | null;
+  contentType: string;
+  encName: Ciphertext;
+  encCreatedAt: Ciphertext | null;
+  encLastModifiedAt: Ciphertext;
+}
+
+export const createUploadSession = async (params: Omit<UploadSession, "id" | "uploadedChunks">) => {
+  return await db.transaction().execute(async (trx) => {
+    const mek = await trx
+      .selectFrom("master_encryption_key")
+      .select("version")
+      .where("user_id", "=", params.userId)
+      .where("state", "=", "active")
+      .limit(1)
+      .forUpdate()
+      .executeTakeFirst();
+    if (mek?.version !== params.mekVersion) {
+      throw new IntegrityError("Inactive MEK version");
+    }
+
+    if (params.hskVersion) {
+      const hsk = await trx
+        .selectFrom("hmac_secret_key")
+        .select("version")
+        .where("user_id", "=", params.userId)
+        .where("state", "=", "active")
+        .limit(1)
+        .forUpdate()
+        .executeTakeFirst();
+      if (hsk?.version !== params.hskVersion) {
+        throw new IntegrityError("Inactive HSK version");
+      }
+    }
+
+    const { sessionId } = await trx
+      .insertInto("upload_session")
+      .values({
+        user_id: params.userId,
+        total_chunks: params.totalChunks,
+        expires_at: params.expiresAt,
+        parent_id: params.parentId !== "root" ? params.parentId : null,
+        master_encryption_key_version: params.mekVersion,
+        encrypted_data_encryption_key: params.encDek,
+        data_encryption_key_version: params.dekVersion,
+        hmac_secret_key_version: params.hskVersion,
+        content_type: params.contentType,
+        encrypted_name: params.encName,
+        encrypted_created_at: params.encCreatedAt,
+        encrypted_last_modified_at: params.encLastModifiedAt,
+      })
+      .returning("id as sessionId")
+      .executeTakeFirstOrThrow();
+    return { id: sessionId };
+  });
+};
+
+export const getUploadSession = async (sessionId: string, userId: number) => {
+  const session = await db
+    .selectFrom("upload_session")
+    .selectAll()
+    .where("id", "=", sessionId)
+    .where("user_id", "=", userId)
+    .where("expires_at", ">", new Date())
+    .limit(1)
+    .executeTakeFirst();
+  return session
+    ? ({
+        id: session.id,
+        userId: session.user_id,
+        totalChunks: session.total_chunks,
+        uploadedChunks: session.uploaded_chunks,
+        expiresAt: session.expires_at,
+        parentId: session.parent_id ?? "root",
+        mekVersion: session.master_encryption_key_version,
+        encDek: session.encrypted_data_encryption_key,
+        dekVersion: session.data_encryption_key_version,
+        hskVersion: session.hmac_secret_key_version,
+        contentType: session.content_type,
+        encName: session.encrypted_name,
+        encCreatedAt: session.encrypted_created_at,
+        encLastModifiedAt: session.encrypted_last_modified_at,
+      } satisfies UploadSession)
+    : null;
+};
+
+export const markChunkAsUploaded = async (sessionId: string, chunkIndex: number) => {
+  await db
+    .updateTable("upload_session")
+    .set({ uploaded_chunks: sql`array_append(uploaded_chunks, ${chunkIndex})` })
+    .where("id", "=", sessionId)
+    .execute();
+};
+
+export const deleteUploadSession = async (trx: typeof db, sessionId: string) => {
+  await trx.deleteFrom("upload_session").where("id", "=", sessionId).execute();
+};
+
+export const cleanupExpiredUploadSessions = async () => {
+  const sessions = await db
+    .deleteFrom("upload_session")
+    .where("expires_at", "<", new Date())
+    .returning("id")
+    .execute();
+  return sessions.map(({ id }) => id);
+};

src/lib/server/loadenv.ts

@@ -26,4 +26,5 @@ export default {
   },
   libraryPath: env.LIBRARY_PATH || "library",
   thumbnailsPath: env.THUMBNAILS_PATH || "thumbnails",
+  uploadsPath: env.UPLOADS_PATH || "uploads",
 };

src/lib/server/modules/filesystem.ts

@@ -1,4 +1,7 @@
 import { unlink } from "fs/promises";
+import env from "$lib/server/loadenv";
+
+export const getChunkDirectoryPath = (sessionId: string) => `${env.uploadsPath}/${sessionId}`;
 
 export const safeUnlink = async (path: string | null | undefined) => {
   if (path) {

src/lib/server/schemas/file.ts

@@ -1,36 +1,7 @@
-import mime from "mime";
 import { z } from "zod";
-import { directoryIdSchema } from "./directory";
 
 export const fileThumbnailUploadRequest = z.object({
   dekVersion: z.iso.datetime(),
   contentIv: z.base64().nonempty(),
 });
 export type FileThumbnailUploadRequest = z.input<typeof fileThumbnailUploadRequest>;
-
-export const fileUploadRequest = z.object({
-  parent: directoryIdSchema,
-  mekVersion: z.int().positive(),
-  dek: z.base64().nonempty(),
-  dekVersion: z.iso.datetime(),
-  hskVersion: z.int().positive(),
-  contentHmac: z.base64().nonempty(),
-  contentType: z
-    .string()
-    .trim()
-    .nonempty()
-    .refine((value) => mime.getExtension(value) !== null), // MIME type
-  contentIv: z.base64().nonempty(),
-  name: z.base64().nonempty(),
-  nameIv: z.base64().nonempty(),
-  createdAt: z.base64().nonempty().optional(),
-  createdAtIv: z.base64().nonempty().optional(),
-  lastModifiedAt: z.base64().nonempty(),
-  lastModifiedAtIv: z.base64().nonempty(),
-});
-export type FileUploadRequest = z.input<typeof fileUploadRequest>;
-
-export const fileUploadResponse = z.object({
-  file: z.int().positive(),
-});
-export type FileUploadResponse = z.output<typeof fileUploadResponse>;

src/lib/server/services/file.ts

@@ -6,17 +6,20 @@ import { dirname } from "path";
 import { Readable } from "stream";
 import { pipeline } from "stream/promises";
 import { v4 as uuidv4 } from "uuid";
-import { FileRepo, MediaRepo, IntegrityError } from "$lib/server/db";
+import { CHUNK_SIZE, ENCRYPTION_OVERHEAD } from "$lib/constants";
+import { FileRepo, MediaRepo, UploadRepo, IntegrityError } from "$lib/server/db";
 import env from "$lib/server/loadenv";
-import { safeUnlink } from "$lib/server/modules/filesystem";
+import { getChunkDirectoryPath, safeUnlink } from "$lib/server/modules/filesystem";
+
+const uploadLocks = new Set<string>();
 
 const createEncContentStream = async (
   path: string,
-  iv: Buffer,
+  iv?: Buffer,
   range?: { start?: number; end?: number },
 ) => {
   const { size: fileSize } = await stat(path);
-  const ivSize = iv.byteLength;
+  const ivSize = iv?.byteLength ?? 0;
   const totalSize = fileSize + ivSize;
 
   const start = range?.start ?? 0;
@@ -30,7 +33,7 @@ const createEncContentStream = async (
       Readable.from(
         (async function* () {
           if (start < ivSize) {
-            yield iv.subarray(start, Math.min(end + 1, ivSize));
+            yield iv!.subarray(start, Math.min(end + 1, ivSize));
           }
           if (end >= ivSize) {
             yield* createReadStream(path, {
@@ -55,7 +58,11 @@ export const getFileStream = async (
     error(404, "Invalid file id");
   }
 
-  return createEncContentStream(file.path, Buffer.from(file.encContentIv, "base64"), range);
+  return createEncContentStream(
+    file.path,
+    file.encContentIv ? Buffer.from(file.encContentIv, "base64") : undefined,
+    range,
+  );
 };
 
 export const getFileThumbnailStream = async (
@@ -110,56 +117,70 @@ export const uploadFileThumbnail = async (
   }
 };
 
-export const uploadFile = async (
-  params: Omit<FileRepo.NewFile, "path" | "encContentHash">,
-  encContentStream: Readable,
-  encContentHash: Promise<string>,
+export const uploadChunk = async (
+  userId: number,
+  sessionId: string,
+  chunkIndex: number,
+  encChunkStream: Readable,
+  encChunkHash: string,
 ) => {
-  const oneDayAgo = new Date(Date.now() - 24 * 60 * 60 * 1000);
-  const oneMinuteLater = new Date(Date.now() + 60 * 1000);
-  if (params.dekVersion <= oneDayAgo || params.dekVersion >= oneMinuteLater) {
-    error(400, "Invalid DEK version");
+  const lockKey = `${sessionId}/${chunkIndex}`;
+  if (uploadLocks.has(lockKey)) {
+    error(409, "Chunk already uploaded"); // TODO: Message
+  } else {
+    uploadLocks.add(lockKey);
   }
 
-  const path = `${env.libraryPath}/${params.userId}/${uuidv4()}`;
-  await mkdir(dirname(path), { recursive: true });
+  const filePath = `${getChunkDirectoryPath(sessionId)}/${chunkIndex}`;
 
   try {
-    const hashStream = createHash("sha256");
-    const [, hash] = await Promise.all([
-      pipeline(
-        encContentStream,
-        async function* (source) {
-          for await (const chunk of source) {
-            hashStream.update(chunk);
-            yield chunk;
-          }
-        },
-        createWriteStream(path, { flags: "wx", mode: 0o600 }),
-      ),
-      encContentHash,
-    ]);
-    if (hashStream.digest("base64") !== hash) {
-      throw new Error("Invalid checksum");
+    const session = await UploadRepo.getUploadSession(sessionId, userId);
+    if (!session) {
+      error(404, "Invalid upload id");
+    } else if (chunkIndex >= session.totalChunks) {
+      error(400, "Invalid chunk index");
+    } else if (session.uploadedChunks.includes(chunkIndex)) {
+      error(409, "Chunk already uploaded");
     }
 
-    const { id: fileId } = await FileRepo.registerFile({
-      ...params,
-      path,
-      encContentHash: hash,
-    });
-    return { fileId };
-  } catch (e) {
-    await safeUnlink(path);
+    const isLastChunk = chunkIndex === session.totalChunks - 1;
 
-    if (e instanceof IntegrityError && e.message === "Inactive MEK version") {
-      error(400, "Invalid MEK version");
+    let writtenBytes = 0;
+    const hashStream = createHash("sha256");
+    const writeStream = createWriteStream(filePath, { flags: "wx", mode: 0o600 });
+
+    for await (const chunk of encChunkStream) {
+      writtenBytes += chunk.length;
+      hashStream.update(chunk);
+      writeStream.write(chunk);
+    }
+
+    await new Promise<void>((resolve, reject) => {
+      writeStream.end((e: any) => (e ? reject(e) : resolve()));
+    });
+
+    if (hashStream.digest("base64") !== encChunkHash) {
+      throw new Error("Invalid checksum");
     } else if (
+      (!isLastChunk && writtenBytes !== CHUNK_SIZE + ENCRYPTION_OVERHEAD) ||
+      (isLastChunk &&
+        (writtenBytes <= ENCRYPTION_OVERHEAD || writtenBytes > CHUNK_SIZE + ENCRYPTION_OVERHEAD))
+    ) {
+      throw new Error("Invalid chunk size");
+    }
+
+    await UploadRepo.markChunkAsUploaded(sessionId, chunkIndex);
+  } catch (e) {
+    await safeUnlink(filePath);
+
+    if (
       e instanceof Error &&
-      (e.message === "Invalid request body" || e.message === "Invalid checksum")
+      (e.message === "Invalid checksum" || e.message === "Invalid chunk size")
     ) {
       error(400, "Invalid request body");
     }
     throw e;
+  } finally {
+    uploadLocks.delete(lockKey);
   }
 };