wip

2025-12-19 03:37:49 +00:00 · 2024-04-30 15:45:10 +07:00
parent 0d1cd37097
commit 68a7f2b0d4
4 changed files with 39 additions and 38 deletions
--- a/index.js
+++ b/index.js
@@ -13,10 +13,10 @@ module.exports = (opts, logger) => {
    getTtsSize: require('./lib/get-tts-size').bind(null, client, logger),
    purgeTtsCache: require('./lib/purge-tts-cache').bind(null, client, logger),
    addFileToCache: require('./lib/add-file-to-cache').bind(null, client, logger),
-    synthAudio: require('./lib/synth-audio').bind(null, client, logger),
+    synthAudio: require('./lib/synth-audio').bind(null, client, createHash, retrieveHash, logger),
    getNuanceAccessToken: require('./lib/get-nuance-access-token').bind(null, client, logger),
    getIbmAccessToken: require('./lib/get-ibm-access-token').bind(null, client, logger),
    getAwsAuthToken: require('./lib/get-aws-sts-token').bind(null, logger, createHash, retrieveHash),
-    getTtsVoices: require('./lib/get-tts-voices').bind(null, client, logger),
+    getTtsVoices: require('./lib/get-tts-voices').bind(null, client, createHash, retrieveHash, logger),
  };
 };
--- a/lib/get-aws-sts-token.js
+++ b/lib/get-aws-sts-token.js
@@ -1,4 +1,4 @@
-const { STSClient, GetSessionTokenCommand } = require('@aws-sdk/client-sts');
+const { STSClient, GetSessionTokenCommand, AssumeRoleCommand } = require('@aws-sdk/client-sts');
 const {makeAwsKey, noopLogger} = require('./utils');
 const debug = require('debug')('jambonz:speech-utils');
 const EXPIRY = 3600;
@@ -6,30 +6,41 @@ const EXPIRY = 3600;
 async function getAwsAuthToken(
  logger,
  createHash, retrieveHash,
-  awsAccessKeyId, awsSecretAccessKey, awsRegion) {
+  awsAccessKeyId, awsSecretAccessKey, awsRegion, roleArn = null) {
  logger = logger || noopLogger;
  try {
-    const key = makeAwsKey(awsAccessKeyId);
+    const key = makeAwsKey(roleArn || awsAccessKeyId);
    const obj = await retrieveHash(key);
    if (obj) return {...obj, servedFromCache: true};

-    /* access token not found in cache, so generate it using STS */
-    const stsClient = new STSClient({
-      region: awsRegion,
-      credentials: {
-        accessKeyId: awsAccessKeyId,
-        secretAccessKey: awsSecretAccessKey,
-      }
-    });
-    const command = new GetSessionTokenCommand({DurationSeconds: EXPIRY});
-    const data = await stsClient.send(command);
+    let data;
+    if (roleArn) {
+      const stsClient = new STSClient({ region: awsRegion});
+      const roleToAssume = { RoleArn: roleArn, RoleSessionName: 'Jambonz_Speech', DurationSeconds: EXPIRY};
+      const command = new AssumeRoleCommand(roleToAssume);
+
+      const response = await stsClient.send(command);
+      data = response;
+    } else {
+      /* access token not found in cache, so generate it using STS */
+      const stsClient = new STSClient({
+        region: awsRegion,
+        credentials: {
+          accessKeyId: awsAccessKeyId,
+          secretAccessKey: awsSecretAccessKey,
+        }
+      });
+      const command = new GetSessionTokenCommand({DurationSeconds: EXPIRY});
+      data = await stsClient.send(command);
+    }

    const credentials = {
      accessKeyId: data.Credentials.AccessKeyId,
      secretAccessKey: data.Credentials.SecretAccessKey,
-      securityToken: data.Credentials.SessionToken
+      sessionToken: data.Credentials.SessionToken
    };

+    console.log(credentials, 'xquanluu');
    /* expire 10 minutes before the hour, so we don't lose the use of it during a call */
    createHash(key, credentials, EXPIRY - 600)
      .catch((err) => logger.error(err, `Error saving hash for key ${key}`));
--- a/lib/get-tts-voices.js
+++ b/lib/get-tts-voices.js
@@ -6,7 +6,7 @@ const TextToSpeechV1 = require('ibm-watson/text-to-speech/v1');
 const { IamAuthenticator } = require('ibm-watson/auth');
 const ttsGoogle = require('@google-cloud/text-to-speech');
 const { PollyClient, DescribeVoicesCommand } = require('@aws-sdk/client-polly');
-const { STSClient, AssumeRoleCommand } = require('@aws-sdk/client-sts');
+const getAwsAuthToken = require('./get-aws-sts-token');

 const getIbmVoices = async(client, logger, credentials) => {
  const {tts_region, tts_api_key} = credentials;
@@ -88,7 +88,7 @@ const getGoogleVoices = async(_client, logger, credentials) => {
  return await client.listVoices();
 };

-const getAwsVoices = async(_client, logger, credentials) => {
+const getAwsVoices = async(_client, createHash, retrieveHash, logger, credentials) => {
  try {
    const {region, accessKeyId, secretAccessKey, roleArn} = credentials;
    let client = null;
@@ -101,15 +101,9 @@ const getAwsVoices = async(_client, logger, credentials) => {
        }
      });
    } else if (roleArn) {
-      const stsClient = new STSClient({ region});
-      const roleToAssume = { RoleArn: roleArn, RoleSessionName: 'Jambonz Speech' };
-      const command = new AssumeRoleCommand(roleToAssume);
-
-      const response = await stsClient.send(command);
-      const assumedRoleCreds = response.Credentials;
      client = new PollyClient({
        region,
-        credentials: assumedRoleCreds,
+        credentials: await getAwsAuthToken(logger, createHash, retrieveHash, null, null, region, roleArn),
      });
    } else {
      client = new PollyClient();
@@ -139,7 +133,7 @@ const getAwsVoices = async(_client, logger, credentials) => {
 * @returns object containing filepath to an mp3 file in the /tmp folder containing
 * the synthesized audio, and a variable indicating whether it was served from cache
 */
-async function getTtsVoices(client, logger, {vendor, credentials}) {
+async function getTtsVoices(client, createHash, retrieveHash, logger, {vendor, credentials}) {
  logger = logger || noopLogger;

  assert.ok(['nuance', 'ibm', 'google', 'aws', 'polly'].includes(vendor),
@@ -154,7 +148,7 @@ async function getTtsVoices(client, logger, {vendor, credentials}) {
      return getGoogleVoices(client, logger, credentials);
    case 'aws':
    case 'polly':
-      return getAwsVoices(client, logger, credentials);
+      return getAwsVoices(client, createHash, retrieveHash, logger, credentials);
    default:
      break;
  }
--- a/lib/synth-audio.js
+++ b/lib/synth-audio.js
@@ -3,7 +3,6 @@ const fs = require('fs');
 const bent = require('bent');
 const ttsGoogle = require('@google-cloud/text-to-speech');
 const { PollyClient, SynthesizeSpeechCommand } = require('@aws-sdk/client-polly');
-const { STSClient, AssumeRoleCommand } = require('@aws-sdk/client-sts');

 const sdk = require('microsoft-cognitiveservices-speech-sdk');
 const TextToSpeechV1 = require('ibm-watson/text-to-speech/v1');
@@ -40,6 +39,7 @@ const debug = require('debug')('jambonz:realtimedb-helpers');
 const EXPIRES = (process.env.JAMBONES_TTS_CACHE_DURATION_MINS || 4 * 60) * 60; // cache tts for 4 hours
 const TMP_FOLDER = '/tmp';
 const OpenAI = require('openai');
+const getAwsAuthToken = require('./get-aws-sts-token');


 const trimTrailingSilence = (buffer) => {
@@ -76,7 +76,7 @@ const trimTrailingSilence = (buffer) => {
 * @returns object containing filepath to an mp3 file in the /tmp folder containing
 * the synthesized audio, and a variable indicating whether it was served from cache
 */
-async function synthAudio(client, logger, stats, { account_sid,
+async function synthAudio(client, createHash, retrieveHash, logger, stats, { account_sid,
  vendor, language, voice, gender, text, engine, salt, model, credentials, deploymentId,
  disableTtsCache, renderForCaching, disableTtsStreaming, options
 }) {
@@ -188,7 +188,8 @@ async function synthAudio(client, logger, stats, { account_sid,
      case 'aws':
      case 'polly':
        vendorLabel = 'aws';
-        audioBuffer = await synthPolly(logger, {credentials, stats, language, voice, text, engine});
+        audioBuffer = await synthPolly(createHash, retrieveHash, logger,
+          {credentials, stats, language, voice, text, engine});
        break;
      case 'azure':
      case 'microsoft':
@@ -264,7 +265,8 @@ async function synthAudio(client, logger, stats, { account_sid,
  });
 }

-const synthPolly = async(logger, {credentials, stats, language, voice, engine, text}) => {
+const synthPolly = async(createHash, retrieveHash, logger,
+  {credentials, stats, language, voice, engine, text}) => {
  try {
    const {region, accessKeyId, secretAccessKey, roleArn} = credentials;
    let polly;
@@ -277,15 +279,9 @@ const synthPolly = async(logger, {credentials, stats, language, voice, engine, t
        }
      });
    } else if (roleArn) {
-      const stsClient = new STSClient({ region});
-      const roleToAssume = { RoleArn: roleArn, RoleSessionName: 'Jambonz_Speech' };
-      const command = new AssumeRoleCommand(roleToAssume);
-
-      const response = await stsClient.send(command);
-      const assumedRoleCreds = response.Credentials;
      polly = new PollyClient({
        region,
-        credentials: assumedRoleCreds,
+        credentials: await getAwsAuthToken(logger, createHash, retrieveHash, null, null, region, roleArn),
      });
    } else {
      // AWS RoleArn assigned to Instance profile