From 301746d49400c5a72f961939c2aa38a6fbfffcc3 Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 16 Aug 2022 22:17:30 +0800
Subject: [PATCH] Add /:id/mp3 router to redirect request to speech MP3 URL

---
 packages/api/src/routers/article_router.ts    | 46 +++++++++++++++++--
 .../svc/{textToSpeech.ts => speech.ts}        | 22 ++++++---
 packages/api/src/server.ts                    |  2 +
 3 files changed, 60 insertions(+), 10 deletions(-)
 rename packages/api/src/routers/svc/{textToSpeech.ts => speech.ts} (69%)

diff --git a/packages/api/src/routers/article_router.ts b/packages/api/src/routers/article_router.ts
index 5774483d6..836980cac 100644
--- a/packages/api/src/routers/article_router.ts
+++ b/packages/api/src/routers/article_router.ts
@@ -3,15 +3,20 @@
 /* eslint-disable @typescript-eslint/no-unsafe-assignment */
 /* eslint-disable @typescript-eslint/explicit-module-boundary-types */
 import express from 'express'
-import { CreateArticleErrorCode } from './../generated/graphql'
-import { isSiteBlockedForParse } from './../utils/blocked'
+import { CreateArticleErrorCode } from '../generated/graphql'
+import { isSiteBlockedForParse } from '../utils/blocked'
 import cors from 'cors'
-import { buildLogger } from './../utils/logger'
+import { buildLogger } from '../utils/logger'
 import { corsConfig } from '../utils/corsConfig'
 import { createPageSaveRequest } from '../services/create_page_save_request'
 import { initModels } from '../server'
 import { kx } from '../datalayer/knex_config'
 import { getClaimsByToken } from '../utils/auth'
+import * as jwt from 'jsonwebtoken'
+import { env } from '../env'
+import { Claims } from '../resolvers/types'
+import { getRepository } from '../entity/utils'
+import { Speech } from '../entity/speech'
 
 const logger = buildLogger('app.dispatch')
 
@@ -61,5 +66,40 @@ export function articleRouter() {
       articleSavingRequestId: result.id,
     })
   })
+
+  router.get(
+    '/:id/mp3',
+    cors<express.Request>(corsConfig),
+    async (req, res) => {
+      const id = req.params.id
+      const token = req.cookies?.auth || req.headers?.authorization
+      if (!token || !jwt.verify(token, env.server.jwtSecret)) {
+        return res.status(401).send({ errorCode: 'UNAUTHORIZED' })
+      }
+      const { uid } = jwt.decode(token) as Claims
+
+      logger.info('Get article speech in mp3 format', {
+        params: req.params,
+        labels: {
+          userId: uid,
+          source: 'GetArticleSpeechMp3',
+          articleId: id,
+        },
+      })
+
+      const speech = await getRepository(Speech).findOneBy({
+        elasticPageId: id,
+        user: { id: uid },
+      })
+
+      if (!speech) {
+        return res.status(404).send({ errorCode: 'NOT_FOUND' })
+      }
+
+      logger.info('Found speech mp3', { audioUrl: speech.audioUrl })
+      res.redirect(speech.audioUrl)
+    }
+  )
+
   return router
 }
diff --git a/packages/api/src/routers/svc/textToSpeech.ts b/packages/api/src/routers/svc/speech.ts
similarity index 69%
rename from packages/api/src/routers/svc/textToSpeech.ts
rename to packages/api/src/routers/svc/speech.ts
index 56b8e2a82..958750fc9 100644
--- a/packages/api/src/routers/svc/textToSpeech.ts
+++ b/packages/api/src/routers/svc/speech.ts
@@ -6,19 +6,20 @@ import { User } from '../../entity/user'
 import { getPageById } from '../../elastic/pages'
 import { synthesizeTextToSpeech } from '../../utils/textToSpeech'
 import { Speech } from '../../entity/speech'
+import { parseHTML } from 'linkedom'
 
-export function textToSpeechServiceRouter() {
+export function speechServiceRouter() {
   const router = express.Router()
 
   router.options('/', cors<express.Request>({ ...corsConfig, maxAge: 600 }))
+  // eslint-disable-next-line @typescript-eslint/no-misused-promises
   router.post('/', async (req, res) => {
-    const { userId, pageId, text } = req.body as {
+    const { userId, pageId } = req.body as {
       userId: string
       pageId: string
-      text: string
     }
 
-    if (!userId || !pageId || !text) {
+    if (!userId || !pageId) {
       return res.status(200).send('Invalid data')
     }
 
@@ -35,7 +36,12 @@ export function textToSpeechServiceRouter() {
       return res.status(200).send('Page not found')
     }
 
-    const audioAndSpeechMarks = await synthesizeTextToSpeech({
+    const text = parseHTML(page.content).document.textContent
+    if (!text) {
+      return res.status(200).send('Page has no text')
+    }
+
+    const speech = await synthesizeTextToSpeech({
       id: pageId,
       text,
       languageCode: page.language,
@@ -44,11 +50,13 @@ export function textToSpeechServiceRouter() {
 
     await getRepository(Speech).save({
       elasticPageId: pageId,
-      audioUrl: audioAndSpeechMarks.audioUrl,
-      speechMarks: JSON.stringify(audioAndSpeechMarks.speechMarks),
+      audioUrl: speech.audioUrl,
+      speechMarks: JSON.stringify(speech.speechMarks),
       user,
     })
 
     res.status(200).send('OK')
   })
+
+  return router
 }
diff --git a/packages/api/src/server.ts b/packages/api/src/server.ts
index 619a4fda5..1945b3e2c 100755
--- a/packages/api/src/server.ts
+++ b/packages/api/src/server.ts
@@ -45,6 +45,7 @@ import { uploadServiceRouter } from './routers/svc/upload'
 import rateLimit from 'express-rate-limit'
 import { webhooksServiceRouter } from './routers/svc/webhooks'
 import { integrationsServiceRouter } from './routers/svc/integrations'
+import { speechServiceRouter } from './routers/svc/speech'
 
 const PORT = process.env.PORT || 4000
 
@@ -119,6 +120,7 @@ export const createApp = (): {
   app.use('/svc/pubsub/integrations', integrationsServiceRouter())
   app.use('/svc/reminders', remindersServiceRouter())
   app.use('/svc/pdf-attachments', pdfAttachmentsRouter())
+  app.use('/svc/speech', speechServiceRouter())
 
   if (env.dev.isLocal) {
     app.use('/local/debug', localDebugRouter())