UBERF-8546 Refactor datalake routing (#7051)

Signed-off-by: Alexander Onnikov <Alexander.Onnikov@xored.com>

Author: aonnikov

workers/datalake/src/blob.ts

@@ -18,7 +18,7 @@ import postgres from 'postgres'
 import * as db from './db'
 import { toUUID } from './encodings'
 import { selectStorage } from './storage'
-import { type UUID } from './types'
+import { type BlobRequest, type WorkspaceRequest, type UUID } from './types'
 import { copyVideo, deleteVideo } from './video'
 
 const expires = 86400
@@ -39,13 +39,9 @@ export function getBlobURL (request: Request, workspace: string, name: string):
   return new URL(path, request.url).toString()
 }
 
-export async function handleBlobGet (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleBlobGet (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const sql = postgres(env.HYPERDRIVE.connectionString)
   const { bucket } = selectStorage(env, workspace)
 
@@ -82,13 +78,9 @@ export async function handleBlobGet (
   return response
 }
 
-export async function handleBlobHead (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleBlobHead (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const sql = postgres(env.HYPERDRIVE.connectionString)
   const { bucket } = selectStorage(env, workspace)
 
@@ -106,7 +98,9 @@ export async function handleBlobHead (
   return new Response(null, { headers, status: 200 })
 }
 
-export async function deleteBlob (env: Env, workspace: string, name: string): Promise<Response> {
+export async function handleBlobDelete (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const sql = postgres(env.HYPERDRIVE.connectionString)
 
   try {
@@ -120,13 +114,15 @@ export async function deleteBlob (env: Env, workspace: string, name: string): Pr
   }
 }
 
-export async function postBlobFormData (request: Request, env: Env, workspace: string): Promise<Response> {
+export async function handleUploadFormData (request: WorkspaceRequest, env: Env): Promise<Response> {
   const contentType = request.headers.get('Content-Type')
   if (contentType === null || !contentType.includes('multipart/form-data')) {
     console.error({ error: 'expected multipart/form-data' })
     return error(400, 'expected multipart/form-data')
   }
 
+  const { workspace } = request
+
   const sql = postgres(env.HYPERDRIVE.connectionString)
 
   let formData: FormData

workers/datalake/src/image.ts

@@ -14,15 +14,17 @@
 //
 
 import { getBlobURL } from './blob'
+import { type BlobRequest } from './types'
 
 const prefferedImageFormats = ['webp', 'avif', 'jpeg', 'png']
 
-export async function getImage (
-  request: Request,
-  workspace: string,
-  name: string,
-  transform: string
-): Promise<Response> {
+export async function handleImageGet (request: BlobRequest): Promise<Response> {
+  const {
+    workspace,
+    name,
+    params: { transform }
+  } = request
+
   const Accept = request.headers.get('Accept') ?? 'image/*'
   const image: Record<string, string> = {}
 

workers/datalake/src/index.ts

@@ -13,61 +13,90 @@
 // limitations under the License.
 //
 
-import { type IRequest, Router, error, html } from 'itty-router'
-import {
-  deleteBlob as handleBlobDelete,
-  handleBlobGet,
-  handleBlobHead,
-  postBlobFormData as handleUploadFormData
-} from './blob'
+import { WorkerEntrypoint } from 'cloudflare:workers'
+import { type IRequestStrict, type RequestHandler, Router, error, html } from 'itty-router'
+
+import { handleBlobDelete, handleBlobGet, handleBlobHead, handleUploadFormData } from './blob'
 import { cors } from './cors'
-import { getImage as handleImageGet } from './image'
-import { getVideoMeta as handleVideoMetaGet } from './video'
+import { handleImageGet } from './image'
+import { handleVideoMetaGet } from './video'
 import { handleSignAbort, handleSignComplete, handleSignCreate } from './sign'
+import { type BlobRequest, type WorkspaceRequest } from './types'
 
 const { preflight, corsify } = cors({
   maxAge: 86400
 })
 
-export default {
-  async fetch (request, env, ctx): Promise<Response> {
-    const router = Router<IRequest>({
-      before: [preflight],
-      finally: [corsify]
-    })
+const router = Router<IRequestStrict, [Env, ExecutionContext], Response>({
+  before: [preflight],
+  finally: [corsify]
+})
+
+const withWorkspace: RequestHandler<WorkspaceRequest> = (request: WorkspaceRequest) => {
+  if (request.params.workspace === undefined || request.params.workspace === '') {
+    return error(400, 'Missing workspace')
+  }
+  request.workspace = decodeURIComponent(request.params.workspace)
+}
+
+const withBlob: RequestHandler<BlobRequest> = (request: BlobRequest) => {
+  if (request.params.name === undefined || request.params.name === '') {
+    return error(400, 'Missing blob name')
+  }
+  request.workspace = decodeURIComponent(request.params.name)
+}
+
+router
+  .get('/blob/:workspace/:name', withBlob, handleBlobGet)
+  .head('/blob/:workspace/:name', withBlob, handleBlobHead)
+  .delete('/blob/:workspace/:name', withBlob, handleBlobDelete)
+  // Image
+  .get('/image/:transform/:workspace/:name', withBlob, handleImageGet)
+  // Video
+  .get('/video/:workspace/:name/meta', withBlob, handleVideoMetaGet)
+  // Form Data
+  .post('/upload/form-data/:workspace', withWorkspace, handleUploadFormData)
+  // Signed URL
+  .post('/upload/signed-url/:workspace/:name', withBlob, handleSignCreate)
+  .put('/upload/signed-url/:workspace/:name', withBlob, handleSignComplete)
+  .delete('/upload/signed-url/:workspace/:name', withBlob, handleSignAbort)
+  .all('/', () =>
+    html(
+      `Huly&reg; Datalake&trade; <a href="https://huly.io">https://huly.io</a>
+      &copy; 2024 <a href="https://hulylabs.com">Huly Labs</a>`
+    )
+  )
+  .all('*', () => error(404))
+
+export default class DatalakeWorker extends WorkerEntrypoint<Env> {
+  async fetch (request: Request): Promise<Response> {
+    return await router.fetch(request, this.env, this.ctx).catch(error)
+  }
+
+  async getBlob (workspace: string, name: string): Promise<ArrayBuffer> {
+    const request = new Request(`https://datalake/blob/${workspace}/${name}`)
+    const response = await router.fetch(request)
+
+    if (!response.ok) {
+      console.error({ error: 'datalake error: ' + response.statusText, workspace, name })
+      throw new Error(`Failed to fetch blob: ${response.statusText}`)
+    }
+
+    return await response.arrayBuffer()
+  }
+
+  async putBlob (workspace: string, name: string, data: ArrayBuffer | Blob | string, type: string): Promise<void> {
+    const request = new Request(`https://datalake/upload/form-data/${workspace}`)
+
+    const body = new FormData()
+    const blob = new Blob([data], { type })
+    body.set('file', blob, name)
 
-    router
-      .get('/blob/:workspace/:name', ({ params }) => handleBlobGet(request, env, ctx, params.workspace, params.name))
-      .head('/blob/:workspace/:name', ({ params }) => handleBlobHead(request, env, ctx, params.workspace, params.name))
-      .delete('/blob/:workspace/:name', ({ params }) => handleBlobDelete(env, params.workspace, params.name))
-      // Image
-      .get('/image/:transform/:workspace/:name', ({ params }) =>
-        handleImageGet(request, params.workspace, params.name, params.transform)
-      )
-      // Video
-      .get('/video/:workspace/:name/meta', ({ params }) =>
-        handleVideoMetaGet(request, env, ctx, params.workspace, params.name)
-      )
-      // Form Data
-      .post('/upload/form-data/:workspace', ({ params }) => handleUploadFormData(request, env, params.workspace))
-      // Signed URL
-      .post('/upload/signed-url/:workspace/:name', ({ params }) =>
-        handleSignCreate(request, env, ctx, params.workspace, params.name)
-      )
-      .put('/upload/signed-url/:workspace/:name', ({ params }) =>
-        handleSignComplete(request, env, ctx, params.workspace, params.name)
-      )
-      .delete('/upload/signed-url/:workspace/:name', ({ params }) =>
-        handleSignAbort(request, env, ctx, params.workspace, params.name)
-      )
-      .all('/', () =>
-        html(
-          `Huly&reg; Datalake&trade; <a href="https://huly.io">https://huly.io</a>
-          &copy; 2024 <a href="https://hulylabs.com">Huly Labs</a>`
-        )
-      )
-      .all('*', () => error(404))
+    const response = await router.fetch(request, { method: 'POST', body })
 
-    return await router.fetch(request).catch(error)
+    if (!response.ok) {
+      console.error({ error: 'datalake error: ' + response.statusText, workspace, name })
+      throw new Error(`Failed to fetch blob: ${response.statusText}`)
+    }
   }
-} satisfies ExportedHandler<Env>
+}

workers/datalake/src/sign.ts

@@ -17,7 +17,7 @@ import { AwsClient } from 'aws4fetch'
 import { error } from 'itty-router'
 
 import { handleBlobUploaded } from './blob'
-import { type UUID } from './types'
+import { type BlobRequest, type UUID } from './types'
 import { selectStorage, type Storage } from './storage'
 
 const S3_SIGNED_LINK_TTL = 3600
@@ -39,13 +39,8 @@ function getS3Client (storage: Storage): AwsClient {
   })
 }
 
-export async function handleSignCreate (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleSignCreate (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
   const storage = selectStorage(env, workspace)
   const accountId = env.R2_ACCOUNT_ID
 
@@ -78,13 +73,9 @@ export async function handleSignCreate (
   return new Response(signed.url, { status: 200, headers })
 }
 
-export async function handleSignComplete (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleSignComplete (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const { bucket } = selectStorage(env, workspace)
   const key = signBlobKey(workspace, name)
 
@@ -117,13 +108,9 @@ export async function handleSignComplete (
   return new Response(null, { status: 201 })
 }
 
-export async function handleSignAbort (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleSignAbort (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const key = signBlobKey(workspace, name)
 
   // Check if the blob has been uploaded

workers/datalake/src/types.ts

@@ -13,10 +13,21 @@
 // limitations under the License.
 //
 
+import { type IRequestStrict } from 'itty-router'
+
 export type Location = 'weur' | 'eeur' | 'wnam' | 'enam' | 'apac'
 
 export type UUID = string & { __uuid: true }
 
+export type WorkspaceRequest = {
+  workspace: string
+} & IRequestStrict
+
+export type BlobRequest = {
+  workspace: string
+  name: string
+} & IRequestStrict
+
 export interface CloudflareResponse {
   success: boolean
   errors: any

workers/datalake/src/video.ts

@@ -15,7 +15,7 @@
 
 import { error, json } from 'itty-router'
 
-import { type CloudflareResponse, type StreamUploadResponse } from './types'
+import { type BlobRequest, type CloudflareResponse, type StreamUploadResponse } from './types'
 
 export type StreamUploadState = 'ready' | 'error' | 'inprogress' | 'queued' | 'downloading' | 'pendingupload'
 
@@ -42,13 +42,9 @@ function streamBlobKey (workspace: string, name: string): string {
   return `v/${workspace}/${name}`
 }
 
-export async function getVideoMeta (
-  request: Request,
-  env: Env,
-  ctx: ExecutionContext,
-  workspace: string,
-  name: string
-): Promise<Response> {
+export async function handleVideoMetaGet (request: BlobRequest, env: Env, ctx: ExecutionContext): Promise<Response> {
+  const { workspace, name } = request
+
   const key = streamBlobKey(workspace, name)
 
   const streamInfo = await env.datalake_blobs.get<StreamBlobInfo>(key, { type: 'json' })