Merge branch 'main' into fix-redirect-audio

docs/contributing.md

@@ -45,7 +45,7 @@ The GCP credential is the JSON service key in stringified format.
 
 #### Install Docker
 
-The test suite ralso equires [Docker](https://www.docker.com/) and docker-compose to be installed on your laptop.  Docker is used to set up a network with all of the elements required to test the jambonz-feature-server in a black-box type of fashion.
+The test suite also requires [Docker](https://www.docker.com/) and docker-compose to be installed on your laptop.  Docker is used to set up a network with all of the elements required to test the jambonz-feature-server in a black-box type of fashion.
 
 Once you have docker installed, you can optionally make sure everything Docker-wise is working properly by running this command from the project folder:
 

lib/http-routes/schemas/create-call.js

@@ -116,12 +116,9 @@ const customSanitizeFunction = (value) => {
       /* trims characters at the beginning and at the end of a string */
       value = value.trim();
 
-      /* We don't escape URLs but verify them via new URL */
+      /* Verify strings including 'http' via new URL */
       if (value.includes('http')) {
         value = new URL(value).toString();
-      } else {
-        /* replaces <, >, &, ', " and / with their corresponding HTML entities */
-        value = escape(value);
       }
     }
   } catch (error) {

lib/middleware.js

@@ -52,10 +52,16 @@ module.exports = function(srf, logger) {
     let clientDb = null;
     if (req.has('X-Authenticated-User')) {
       req.locals.originatingUser = req.get('X-Authenticated-User');
+      let clientSettings;
       const arr = /^(.*)@(.*)/.exec(req.locals.originatingUser);
       if (arr) {
-        [clientDb] = await lookupClientByAccountAndUsername(account_sid, arr[1]);
+        [clientSettings] = await lookupClientByAccountAndUsername(account_sid, arr[1]);
       }
+      clientDb = await registrar.query(req.locals.originatingUser);
+      clientDb = {
+        ...clientDb,
+        ...clientSettings,
+      };
     }
 
     // check for call to application
@@ -160,7 +166,7 @@ module.exports = function(srf, logger) {
         };
         logger.info({callId, metadata, sdp}, 'successfully parsed SIPREC payload');
       } catch (err) {
-        logger.info({callId}, 'Error parsing multipart payload');
+        logger.info({err, callId}, 'Error parsing multipart payload');
         return res.send(503);
       }
     }

lib/session/adulting-call-session.js

@@ -1,4 +1,6 @@
 const CallSession = require('./call-session');
+const {CallStatus} = require('../utils/constants');
+const moment = require('moment');
 
 /**
  * @classdesc Subclass of CallSession.  Represents a CallSession
@@ -26,6 +28,7 @@ class AdultingCallSession extends CallSession {
       this._callReleased();
     });
     this.sd.emit('adulting');
+    this.on('callStatusChange', this._notifyCallStatusChange.bind(this));
   }
 
   get dlg() {
@@ -50,6 +53,18 @@ class AdultingCallSession extends CallSession {
   }
 
   _callerHungup() {
+    if (this.dlg.connectTime) {
+      const duration = moment().diff(this.dlg.connectTime, 'seconds');
+      this.rootSpan.setAttributes({'call.termination': 'hangup by caller'});
+      this.callInfo.callTerminationBy = 'caller';
+      this.emit('callStatusChange', {
+        callStatus: CallStatus.Completed,
+        duration
+      });
+    }
+    this.logger.info('InboundCallSession: caller hung up');
+    this._callReleased();
+    this.req.removeAllListeners('cancel');
   }
 }
 

lib/session/call-session.js

@@ -171,6 +171,18 @@ class CallSession extends Emitter {
     return this.application.notifier;
   }
 
+  /**
+   * syntheizer
+   */
+
+  get synthesizer() {
+    return this._synthesizer;
+  }
+
+  set synthesizer(synth) {
+    this._synthesizer = synth;
+  }
+
   /**
    * default vendor to use for speech synthesis if not provided in the app
    */
@@ -253,6 +265,16 @@ class CallSession extends Emitter {
   set fallbackSpeechRecognizerVendor(vendor) {
     this.application.fallback_speech_recognizer_vendor = vendor;
   }
+  /**
+   * recognizer
+   */
+  get recognizer() {
+    return this._recognizer;
+  }
+
+  set recognizer(rec) {
+    this._recognizer = rec;
+  }
   /**
    * default vendor to use for speech recognition if not provided in the app
    */
@@ -834,7 +856,8 @@ class CallSession extends Emitter {
         } else if ('elevenlabs' === vendor) {
           return {
             api_key: credential.api_key,
-            model_id: credential.model_id
+            model_id: credential.model_id,
+            options: credential.options
           };
         } else if ('assemblyai' === vendor) {
           return {
@@ -919,7 +942,8 @@ class CallSession extends Emitter {
       if (0 === this.tasks.length &&
         this.requestor instanceof WsRequestor &&
         !this.requestor.closedGracefully &&
-        !this.callGone
+        !this.callGone &&
+        !this.disableWaitCommand
       ) {
         try {
           await this._awaitCommandsOrHangup();
@@ -1187,6 +1211,38 @@ class CallSession extends Emitter {
     }
   }
 
+  /**
+   * perform live call control - send RFC 2833 DTMF
+   * @param {obj} opts
+   * @param {string} opts.dtmf.digit - DTMF digit
+   * @param {string} opts.dtmf.duration - Optional, Duration
+   */
+  async _lccDtmf(opts, callSid) {
+    const {dtmf} = opts;
+    const {digit, duration = 250} = dtmf;
+    if (!this.hasStableDialog) {
+      this.logger.info('CallSession:_lccDtmf - invalid command as we do not have a stable call');
+      return;
+    }
+    try {
+      const dlg = callSid === this.callSid ? this.dlg : this.currentTask.dlg;
+      const res = await dlg.request({
+        method: 'INFO',
+        headers: {
+          'Content-Type': 'application/dtmf',
+          'X-Reason': 'Dtmf'
+        },
+        body: `Signal=${digit}
+Duration=${duration} `
+      });
+      this.logger.debug({res}, `CallSession:_lccDtmf
+ got response to INFO DTMF digit=${digit} and duration=${duration}`);
+      return res;
+    } catch (err) {
+      this.logger.error({err}, 'CallSession:_lccDtmf - error sending INFO RFC 2833 DTMF');
+    }
+  }
+
   /**
    * perform live call control -- whisper to one party or the other on a call
    * @param {array} opts - array of play or say tasks
@@ -1270,6 +1326,8 @@ class CallSession extends Emitter {
     else if (opts.sip_request) {
       const res = await this._lccSipRequest(opts, callSid);
       return {status: res.status, reason: res.reason};
+    } else if (opts.dtmf) {
+      await this._lccDtmf(opts, callSid);
     }
     else if (opts.record) {
       await this.notifyRecordOptions(opts.record);
@@ -1461,6 +1519,13 @@ class CallSession extends Emitter {
           });
         break;
 
+      case 'dtmf':
+        this._lccDtmf(data, call_sid)
+          .catch((err) => {
+            this.logger.info({err, data}, `CallSession:_onCommand - error sending RFC 2833 DTMF ${data}`);
+          });
+        break;
+
       default:
         this.logger.info(`CallSession:_onCommand - invalid command ${command}`);
     }

lib/session/confirm-call-session.js

@@ -23,6 +23,7 @@ class ConfirmCallSession extends CallSession {
     });
     this.dlg = dlg;
     this.ep = ep;
+    this.disableWaitCommand = true;
   }
 
   /**

lib/tasks/config.js

@@ -109,6 +109,7 @@ class TaskConfig extends Task {
     });
 
     if (this.hasSynthesizer) {
+      cs.synthesizer = this.synthesizer;
       cs.speechSynthesisVendor = this.synthesizer.vendor !== 'default'
         ? this.synthesizer.vendor
         : cs.speechSynthesisVendor;
@@ -138,6 +139,7 @@ class TaskConfig extends Task {
       this.logger.info({synthesizer: this.synthesizer}, 'Config: updated synthesizer');
     }
     if (this.hasRecognizer) {
+      cs.recognizer = this.recognizer;
       cs.speechRecognizerVendor = this.recognizer.vendor !== 'default'
         ? this.recognizer.vendor
         : cs.speechRecognizerVendor;

lib/tasks/dial.js

@@ -17,7 +17,7 @@ const dbUtils = require('../utils/db-utils');
 const debug = require('debug')('jambonz:feature-server');
 const {parseUri} = require('drachtio-srf');
 const {ANCHOR_MEDIA_ALWAYS, JAMBONZ_DISABLE_DIAL_PAI_HEADER} = require('../config');
-const { isOnhold } = require('../utils/sdp-utils');
+const { isOnhold, isOpusFirst } = require('../utils/sdp-utils');
 const { normalizeJambones } = require('@jambonz/verb-specifications');
 
 function parseDtmfOptions(logger, dtmfCapture) {
@@ -488,7 +488,8 @@ class TaskDial extends Task {
       headers: this.headers,
       proxy: `sip:${sbcAddress}`,
       callingNumber: this.callerId || req.callingNumber,
-      ...(this.callerName && {callingName: this.callerName})
+      ...(this.callerName && {callingName: this.callerName}),
+      opusFirst: isOpusFirst(this.cs.ep.remote.sdp)
     };
 
     const t = this.target.find((t) => t.type === 'teams');
@@ -790,9 +791,11 @@ class TaskDial extends Task {
     assert(cs.ep && sd.ep);
 
     try {
+      // Wait until we got new SDP from B leg to ofter to A Leg
       const aLegSdp = cs.ep.remote.sdp;
+      await sd.releaseMediaToSBC(aLegSdp, cs.ep.local.sdp);
       const bLegSdp = sd.dlg.remote.sdp;
-      await Promise.all[sd.releaseMediaToSBC(aLegSdp, cs.ep.local.sdp), cs.releaseMediaToSBC(bLegSdp)];
+      await cs.releaseMediaToSBC(bLegSdp);
       this.epOther = null;
       this.logger.info('Dial:_releaseMedia - successfully released media from freewitch');
     } catch (err) {

lib/tasks/gather.js

@@ -103,7 +103,7 @@ class TaskGather extends SttTask {
 
   async exec(cs, {ep}) {
     this.logger.debug({options: this.data}, 'Gather:exec');
-    await super.exec(cs);
+    await super.exec(cs, {ep});
     const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, cs.srf);
 
     if (cs.hasGlobalSttHints && !this.maskGlobalSttHints) {
@@ -141,59 +141,6 @@ class TaskGather extends SttTask {
       this.interim = true;
       this.logger.debug('Gather:exec - early hints match enabled');
     }
-
-    this.ep = ep;
-    if ('default' === this.vendor || !this.vendor) {
-      this.vendor = cs.speechRecognizerVendor;
-      if (this.data.recognizer) this.data.recognizer.vendor = this.vendor;
-    }
-    if ('default' === this.language || !this.language) {
-      this.language = cs.speechRecognizerLanguage;
-      if (this.data.recognizer) this.data.recognizer.language = this.language;
-    }
-    if ('default' === this.label || !this.label) {
-      this.label = cs.speechRecognizerLabel;
-      if (this.data.recognizer) this.data.recognizer.label = this.label;
-    }
-    // Fallback options
-    if ('default' === this.fallbackVendor || !this.fallbackVendor) {
-      this.fallbackVendor = cs.fallbackSpeechRecognizerVendor;
-      if (this.data.recognizer) this.data.recognizer.fallbackVendor = this.fallbackVendor;
-    }
-    if ('default' === this.fallbackLanguage || !this.fallbackLanguage) {
-      this.fallbackLanguage = cs.fallbackSpeechRecognizerLanguage;
-      if (this.data.recognizer) this.data.recognizer.fallbackLanguage = this.fallbackLanguage;
-    }
-    if ('default' === this.fallbackLabel || !this.fallbackLabel) {
-      this.fallbackLabel = cs.fallbackSpeechRecognizerLabel;
-      if (this.data.recognizer) this.data.recognizer.fallbackLabel = this.fallbackLabel;
-    }
-    if (!this.data.recognizer.vendor) {
-      this.data.recognizer.vendor = this.vendor;
-    }
-    if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
-      // By default, application saves cobalt model in language
-      this.data.recognizer.model = cs.speechRecognizerLanguage;
-    }
-
-    if (this.needsStt && !this.sttCredentials) {
-      try {
-        this.sttCredentials = await this._initSpeechCredentials(cs, this.vendor, this.label);
-      } catch (error) {
-        if (this.fallbackVendor && this.isHandledByPrimaryProvider) {
-          await this._fallback();
-        } else {
-          throw error;
-        }
-      }
-    }
-
-    /* when using cobalt model is required */
-    if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
-      this.notifyError({ msg: 'ASR error', details:'Cobalt requires a model to be specified'});
-      throw new Error('Cobalt requires a model to be specified');
-    }
-
     const startListening = async(cs, ep) => {
       this._startTimer();
       if (this.isContinuousAsr && 0 === this.timeout) this._startAsrTimer();
@@ -319,6 +266,13 @@ class TaskGather extends SttTask {
       this._resolve('dtmf-terminator-key');
     }
     else if (this.input.includes('digits')) {
+      if (this.digitBuffer.length === 0 && this.needsStt) {
+        // DTMF is higher priority than STT.
+        this.removeSpeechListeners(ep);
+        ep.stopTranscription({vendor: this.vendor})
+          .catch((err) => this.logger.error({err},
+            ` Received DTMF, Error stopping transcription for vendor ${this.vendor}`));
+      }
       this.digitBuffer += evt.dtmf;
       const len = this.digitBuffer.length;
       if (len === this.numDigits || len === this.maxDigits) {
@@ -528,7 +482,9 @@ class TaskGather extends SttTask {
     this._clearTimer();
     this._timeoutTimer = setTimeout(() => {
       if (this.isContinuousAsr) this._startAsrTimer();
-      else if (this.interDigitTimeout <= 0 || this.digitBuffer.length < this.minDigits || this.needsStt) {
+      else if (this.interDigitTimeout <= 0 ||
+        this.digitBuffer.length < this.minDigits ||
+        this.needsStt && this.digitBuffer.length === 0) {
         this._resolve(this.digitBuffer.length >= this.minDigits ? 'dtmf-num-digits' : 'timeout');
       }
     }, this.timeout);
@@ -634,7 +590,8 @@ class TaskGather extends SttTask {
       return;
     }
 
-    evt = this.normalizeTranscription(evt, this.vendor, 1, this.language, this.shortUtterance);
+    evt = this.normalizeTranscription(evt, this.vendor, 1, this.language,
+      this.shortUtterance, this.data.recognizer.punctuation);
     if (evt.alternatives.length === 0) {
       this.logger.info({evt}, 'TaskGather:_onTranscription - got empty transcript, continue listening');
       return;

lib/tasks/say.js

@@ -86,6 +86,13 @@ class TaskSay extends Task {
       credentials.api_key = this.options.apiKey || credentials.apiKey;
       credentials.region = this.options.region || credentials.region;
       voice = this.options.voice || voice;
+    } else if (vendor === 'elevenlabs') {
+      credentials = credentials || {};
+      credentials.model_id = this.options.model_id || credentials.model_id;
+      credentials.voice_settings = this.options.voice_settings || {};
+      credentials.optimize_streaming_latency = this.options.optimize_streaming_latency
+      || credentials.optimize_streaming_latency;
+      voice = this.options.voice_id || voice;
     }
 
     this.logger.info({vendor, language, voice, model}, 'TaskSay:exec');
@@ -127,6 +134,7 @@ class TaskSay extends Task {
             model,
             salt,
             credentials,
+            options: this.options,
             disableTtsCache : this.disableTtsCache
           });
           this.logger.debug(`file ${filePath}, served from cache ${servedFromCache}`);