Feature/vm detection (#137)

* initial changes for amd

* wip

* fix bug where transcripts were discarded

* a bit of refactoring, and adding support for avmd in config verb

* bug fixes

lib/tasks/config.js

@@ -50,12 +50,19 @@ class TaskConfig extends Task {
       const s = `{${v},${l}}`;
       phrase.push(`set recognizer${s}`);
     }
+    if (this.data.amd) phrase.push('enable amd');
     return  `${this.name}{${phrase.join(',')}`;
   }
 
   async exec(cs) {
     await super.exec(cs);
 
+    if (this.data.amd) {
+      this.startAmd = cs.startAmd;
+      this.stopAmd = cs.stopAmd;
+      this.on('amd', this._onAmdEvent.bind(this, cs));
+    }
+
     if (this.hasSynthesizer) {
       cs.speechSynthesisVendor = this.synthesizer.vendor !== 'default'
         ? this.synthesizer.vendor
@@ -107,6 +114,12 @@ class TaskConfig extends Task {
   async kill(cs) {
     super.kill(cs);
   }
+
+  _onAmdEvent(cs, evt) {
+    this.logger.info({evt}, 'Config:_onAmdEvent');
+    const {actionHook} = this.data.amd;
+    this.performHook(cs, actionHook, evt);
+  }
 }
 
 module.exports = TaskConfig;

lib/tasks/dial.js

@@ -134,7 +134,10 @@ class TaskDial extends Task {
   get name() { return TaskName.Dial; }
 
   get canReleaseMedia() {
-    return !process.env.ANCHOR_MEDIA_ALWAYS && !this.listenTask && !this.transcribeTask;
+    return !process.env.ANCHOR_MEDIA_ALWAYS &&
+      !this.listenTask &&
+      !this.transcribeTask &&
+      !this.startAmd;
   }
 
   get summary() {
@@ -158,6 +161,11 @@ class TaskDial extends Task {
   async exec(cs) {
     await super.exec(cs);
     try {
+      if (this.data.amd) {
+        this.startAmd = cs.startAmd;
+        this.stopAmd = cs.stopAmd;
+        this.on('amd', this._onAmdEvent.bind(this, cs));
+      }
       if (cs.direction === CallDirection.Inbound) {
         await this._initializeInbound(cs);
       }
@@ -181,6 +189,11 @@ class TaskDial extends Task {
 
   async kill(cs, reason) {
     super.kill(cs);
+    try {
+      if (this.ep && this.ep.amd) this.stopAmd(this.ep, this);
+    } catch (err) {
+      this.logger.error({err}, 'DialTask:kill - error stopping answering machine detectin');
+    }
     if (this.dialMusic && this.epOther) {
       this.epOther.api('uuid_break', this.epOther.uuid)
         .catch((err) => this.logger.info(err, 'Error killing dialMusic'));
@@ -226,7 +239,7 @@ class TaskDial extends Task {
       this.logger.debug('Dial:whisper executing tasks');
       while (tasks.length && !cs.callGone) {
         const task = tasks.shift();
-        const {span, ctx} = this.startChildSpan(`whisper:${this.sayTask.summary}`);
+        const {span, ctx} = this.startChildSpan(`whisper:${task.summary}`);
         task.span = span;
         task.ctx = ctx;
         await task.exec(cs, callSid === this.callSid ? this.ep : this.epOther);
@@ -571,6 +584,7 @@ class TaskDial extends Task {
    * - save the dialog and endpoint
    * - clock the start time of the call,
    * - start a max call length timer (optionally)
+   * - start answering machine detection (optionally)
    * - launch any nested tasks
    * - and establish a handler to clean up if the called party hangs up
    */
@@ -613,6 +627,13 @@ class TaskDial extends Task {
 
     if (this.transcribeTask) this.transcribeTask.exec(cs, this.epOther, this.ep);
     if (this.listenTask) this.listenTask.exec(cs, this.epOther);
+    if (this.startAmd) {
+      try {
+        this.startAmd(cs, this.ep, this, this.data.amd);
+      } catch (err) {
+        this.logger.info({err}, 'Dial:_selectSingleDial - Error calling startAmd');
+      }
+    }
 
     /* if we can release the media back to the SBC, do so now */
     if (this.canReleaseMedia) this._releaseMedia(cs, sd);
@@ -659,6 +680,12 @@ class TaskDial extends Task {
     this.logger.info({sdp}, 'Dial:handleReinviteAfterMediaReleased - sent reinvite to B leg');
     res.send(200, {body: sdp});
   }
+
+  _onAmdEvent(cs, evt) {
+    this.logger.info({evt}, 'Dial:_onAmdEvent');
+    const {actionHook} = this.data.amd;
+    this.performHook(cs, actionHook, evt);
+  }
 }
 
 module.exports = TaskDial;

lib/tasks/gather.js

@@ -273,6 +273,7 @@ class TaskGather extends Task {
     }
 
     if ('google' === this.vendor) {
+      this.bugname = 'google_trancribe';
       if (this.sttCredentials) opts.GOOGLE_APPLICATION_CREDENTIALS = JSON.stringify(this.sttCredentials.credentials);
       [
         ['enhancedModel', 'GOOGLE_SPEECH_USE_ENHANCED'],
@@ -308,6 +309,7 @@ class TaskGather extends Task {
       ep.addCustomEventListener(GoogleTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
     }
     else if (['aws', 'polly'].includes(this.vendor)) {
+      this.bugname = 'aws_trancribe';
       if (this.vocabularyName) opts.AWS_VOCABULARY_NAME = this.vocabularyName;
       if (this.vocabularyFilterName) {
         opts.AWS_VOCABULARY_NAME = this.vocabularyFilterName;
@@ -324,6 +326,7 @@ class TaskGather extends Task {
       ep.addCustomEventListener(AwsTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
     }
     else if ('microsoft' === this.vendor) {
+      this.bugname = 'azure_trancribe';
       if (this.sttCredentials) {
         Object.assign(opts, {
           'AZURE_SUBSCRIPTION_KEY': this.sttCredentials.api_key,
@@ -355,12 +358,14 @@ class TaskGather extends Task {
     this.logger.debug({
       vendor: this.vendor,
       locale: this.language,
-      interim: this.interim
+      interim: this.interim,
+      bugname: this.bugname
     }, 'Gather:_startTranscribing');
     ep.startTranscription({
       vendor: this.vendor,
       locale: this.language,
       interim: this.interim,
+      bugname: this.bugname,
     }).catch((err) => {
       const {writeAlerts, AlertType} = this.cs.srf.locals;
       this.logger.error(err, 'TaskGather:_startTranscribing error');
@@ -440,7 +445,11 @@ class TaskGather extends Task {
     }
   }
 
-  _onTranscription(cs, ep, evt) {
+  _onTranscription(cs, ep, evt, fsEvent) {
+    // make sure this is not a transcript from answering machine detection
+    const bugname = fsEvent.getHeader('media-bugname');
+    if (bugname && this.bugname !== bugname) return;
+
     if ('aws' === this.vendor && Array.isArray(evt) && evt.length > 0) evt = evt[0];
     if ('microsoft' === this.vendor) {
       const final = evt.RecognitionStatus === 'Success';

lib/tasks/specs.json

@@ -37,7 +37,8 @@
       "synthesizer": "#synthesizer",
       "recognizer": "#recognizer",
       "bargeIn": "#bargeIn",
-      "record": "#recordOptions"
+      "record": "#recordOptions",
+      "amd": "#amd"
     },
     "required": []
   },
@@ -172,7 +173,8 @@
       "timeLimit": "number",
       "timeout": "number",
       "proxy": "string",
-      "transcribe": "#transcribe"
+      "transcribe": "#transcribe",
+      "amd": "#amd"
     },
     "required": [
       "target"
@@ -517,5 +519,22 @@
     "required": [
       "enable"
     ]
+  },
+  "amd": {
+    "properties": {
+      "actionHook": "object|string",
+      "thresholdWordCount": "number",
+      "timers": "#amdTimers",
+      "recognizer": "#recognizer"
+    },
+    "required": [
+      "actionHook"
+    ]
+  },
+  "amdTimers": {
+    "noSpeechTimeoutMs": "number",
+    "decisionTimeoutMs": "number",
+    "toneTimeoutMs": "number",
+    "greetingCompletionTimeoutMs": "number"
   }
 }

lib/tasks/task.js

@@ -171,12 +171,13 @@ class Task extends Emitter {
   }
 
   async performHook(cs, hook, results) {
+    const params = results ? Object.assign(cs.callInfo.toJSON(), results) : cs.callInfo.toJSON();
     const span = this.startSpan('verb:hook', {'hook.url': hook});
     const b3 = this.getTracingPropagation('b3', span);
     const httpHeaders = b3 && {b3};
-    span.setAttributes({'http.body': JSON.stringify(results)});
+    span.setAttributes({'http.body': JSON.stringify(params)});
     try {
-      const json = await cs.requestor.request('verb:hook', hook, results, httpHeaders);
+      const json = await cs.requestor.request('verb:hook', hook, params, httpHeaders);
       span.setAttributes({'http.statusCode': 200});
       span.end();
       if (json && Array.isArray(json)) {

lib/tasks/transcribe.js

@@ -102,7 +102,7 @@ class TaskTranscribe extends Task {
 
   async kill(cs) {
     super.kill(cs);
-    if (this.ep.connected) {
+    if (this.ep?.connected) {
       this.ep.stopTranscription({vendor: this.vendor})
         .catch((err) => this.logger.info(err, 'Error TaskTranscribe:kill'));
 
@@ -140,6 +140,7 @@ class TaskTranscribe extends Task {
     ep.addCustomEventListener(AzureTranscriptionEvents.NoSpeechDetected, this._onNoAudio.bind(this, cs, ep, channel));
 
     if (this.vendor === 'google') {
+      this.bugname = 'google_trancribe';
       if (this.sttCredentials) opts.GOOGLE_APPLICATION_CREDENTIALS = JSON.stringify(this.sttCredentials.credentials);
       [
         ['enhancedModel', 'GOOGLE_SPEECH_USE_ENHANCED'],
@@ -175,6 +176,7 @@ class TaskTranscribe extends Task {
         .catch((err) => this.logger.info(err, 'TaskTranscribe:_startTranscribing with google'));
     }
     else if (this.vendor === 'aws') {
+      this.bugname = 'aws_trancribe';
       [
         ['diarization', 'AWS_SHOW_SPEAKER_LABEL'],
         ['identifyChannels', 'AWS_ENABLE_CHANNEL_IDENTIFICATION']
@@ -206,6 +208,7 @@ class TaskTranscribe extends Task {
         .catch((err) => this.logger.info(err, 'TaskTranscribe:_startTranscribing with aws'));
     }
     else if (this.vendor === 'microsoft') {
+      this.bugname = 'azure_trancribe';
       Object.assign(opts, {
         'AZURE_SUBSCRIPTION_KEY': this.sttCredentials.api_key,
         'AZURE_REGION': this.sttCredentials.region
@@ -231,11 +234,16 @@ class TaskTranscribe extends Task {
       vendor: this.vendor,
       interim: this.interim ? true : false,
       locale: this.language,
-      channels: /*this.separateRecognitionPerChannel ? 2 : */ 1
+      channels: /*this.separateRecognitionPerChannel ? 2 : */ 1,
+      bugname: this.bugname
     });
   }
 
-  _onTranscription(cs, ep, channel, evt) {
+  _onTranscription(cs, ep, channel, evt, fsEvent) {
+    // make sure this is not a transcript from answering machine detection
+    const bugname = fsEvent.getHeader('media-bugname');
+    if (bugname && this.bugname !== bugname) return;
+
     this.logger.debug({evt, channel}, 'TaskTranscribe:_onTranscription');
     if ('aws' === this.vendor && Array.isArray(evt) && evt.length > 0) evt = evt[0];
     if ('microsoft' === this.vendor) {
@@ -249,7 +257,7 @@ class TaskTranscribe extends Task {
       }) :
         [
           {
-            transcript: evt.DisplayText
+            transcript: evt.DisplayText || evt.Text
           }
         ];