added transcribe functionality in server.py and Program test instance for it

Author: tyler232

PythonRpcServer/server.py

@@ -12,6 +12,9 @@
 from echo import EchoProvider
 from kaltura import KalturaProvider
 from mediaprovider import InvalidPlaylistInfoException
+from transcribe import transcribe_audio
+
+import json
 import hasher 
 import ffmpeg
 # import phrasehinter
@@ -44,12 +47,12 @@ class PythonServerServicer(ct_pb2_grpc.PythonServerServicer):
     # Transcribe it into a json string from the transcribe text
     # Make it returns a json string
     # change name to TranscribeRPC
-    def CaptionRPC(self, request, context):
-        #See CaptionRequest
-        print( f"CaptionRPC({request.logId};{request.refId};{request.filePath};{request.phraseHints};{request.courseHints};{request.outputLanguages})")
-        kalturaprovider = KalturaProvider()
-        result = LogWorker(f"CaptionRPC({request.filePath})", lambda: kalturaprovider.getCaptions(request.refId))
-        return  ct_pb2.JsonString(json = result)
+    # def CaptionRPC(self, request, context):
+    #     #See CaptionRequest
+    #     print( f"CaptionRPC({request.logId};{request.refId};{request.filePath};{request.phraseHints};{request.courseHints};{request.outputLanguages})")
+    #     kalturaprovider = KalturaProvider()
+    #     result = LogWorker(f"CaptionRPC({request.filePath})", lambda: kalturaprovider.getCaptions(request.refId))
+    #     return  ct_pb2.JsonString(json = result)
 
 
 
@@ -125,6 +128,23 @@ def ComputeFileHash(self, request, context):
     def GetMediaInfoRPC(self, request, context):
         result = LogWorker(f"GetMediaInfo({request.filePath})", lambda: ffmpeg.getMediaInfo(request.filePath))
         return  ct_pb2.JsonString(json = result)
+    
+    
+    def TranscribeAudioRPC(self, request, context):
+        print(f"TranscribeAudioRPC({request.logId};{request.filePath})")
+        try:
+            logging.info(f"Starting transcription for file: {request.filePath}")
+            transcription_result = LogWorker(
+                f"TranscribeAudioRPC({request.filePath})",
+                lambda: transcribe_audio(request.filePath)
+            )
+            logging.info(f"Transcription completed successfully for: {request.filePath}")
+            return ct_pb2.JsonString(json=json.dumps(transcription_result))
+
+        except Exception as e:
+            context.set_code(grpc.StatusCode.INTERNAL)
+            context.set_details(f"Transcription failed: {str(e)}")
+            return ct_pb2.JsonString(json=json.dumps({"error": str(e)}))
 
 def serve():
     print("Python RPC Server Starting")

PythonRpcServer/transcribe.py

@@ -9,7 +9,7 @@
 def transcribe_audio(media_filepath):
 
     if media_filepath == 'TEST-transcribe_example_result':
-        result_json_file = 'transcribe_example_result.json'
+        result_json_file = 'transcribe_exampleffmp_result.json'
         with open(result_json_file, 'r') as json_file:
             transcription_result = json.load(json_file)
         return transcription_result
@@ -20,7 +20,7 @@ def transcribe_audio(media_filepath):
 
     # Path to the output JSON file that Whisper will generate
     json_output_path = f"{media_filepath}.json"
-    if os.path.exists(media_filepath):
+    if os.path.exists(json_output_path):
         os.remove(json_output_path)
 
     # Command to run Whisper.cpp inside the container using the main executable

TaskEngine/Program.cs

@@ -133,21 +133,37 @@ static void runQueueAwakerForever() {
             _logger.LogInformation("Pausing {0} minutes before first periodicCheck", initialPauseInterval);
 
             // Thread.Sleep(initialPauseInterval);
-            Task.Delay(initialPauseInterval).Wait();
+            // Task.Delay(initialPauseInterval).Wait();
             // Check for new tasks every "timeInterval".
             // The periodic check will discover all undone tasks
             // TODO/REVIEW: However some tasks also publish the next items
             while (true)
             {
+                // try {
+                //     _logger.LogInformation("Periodic Check");
+                //     queueAwakerTask.Publish(new JObject
+                //     {
+                //         { "Type", TaskType.PeriodicCheck.ToString() }
+                //     });
+                // } catch (Exception e) {
+                //     _logger.LogError(e, "Error in Periodic Check");
+                // }
+
                 try {
-                    _logger.LogInformation("Periodic Check");
+                    var videoId = "ddceb720-a9d6-417d-b5ea-e94c6c0a86c6";
+                    _logger.LogInformation("Transcription Task Initiated");
                     queueAwakerTask.Publish(new JObject
                     {
-                        { "Type", TaskType.PeriodicCheck.ToString() }
+                        { "Type", TaskType.TranscribeVideo.ToString() },
+                        { "videoOrMediaId", videoId }
                     });
+
+                    _logger.LogInformation("Transcription Task Published Successfully");
                 } catch (Exception e) {
-                    _logger.LogError(e, "Error in Periodic Check");
+                    _logger.LogError(e, "Error in Transcription Task");
                 }
+
+
                 // Thread.Sleep(timeInterval);
                 Task.Delay(timeInterval).Wait();
                  _logger.LogInformation("Pausing {0} minutes before next periodicCheck", periodicCheck);