Update dialogue_generator.py

dialogue_generator.py

@@ -5,11 +5,14 @@ from typing import Optional, List, Dict
 import google.generativeai as genai
 from google.ai import generativelanguage as glm
 import instructor
+from dotenv import load_dotenv
 from models.chesspiece import ChessPiece
 from personality_repository import ChessRepository, WhitePiece
 from elevenlabs_texttospeech import text_to_speech
 import pathlib
+import re
 
+load_dotenv()
 
 # Endpoint Input
 class TextDialogueInput(BaseModel):
@@ -110,7 +113,7 @@ def create_image_dicts(encoded_images: List[str]) -> List[Dict[str, any]]:
 
 def create_audio_dict(encoded_audio: str) -> List[Dict[str, any]]:
     audio_data = base64.b64decode(encoded_audio)
-    audio_blob = glm.Blob(mime_type="audio/aac", data=audio_data)
+    audio_blob = glm.Blob(mime_type="audio/opus", data=audio_data)
     audio_dict = {
         "role": "user",
         "content": audio_blob
@@ -118,6 +121,13 @@ def create_audio_dict(encoded_audio: str) -> List[Dict[str, any]]:
 
     return audio_dict
 
+def extract_base64(data_uri: str) -> str:
+    """Extract the base64 part from a Data URI."""
+    match = re.match(r'data:.*?;base64,(.+)', data_uri)
+    if match:
+        return match.group(1)
+    return data_uri  # Return as-is if it doesn't match the pattern
+
 def process_dialogue(input_content: str, piece: ChessPiece, language: str, is_audio: bool = False, images: List[str] = None) -> DialogueOutput:
     system_prompt = generate_system_prompt(piece, language)
     messages = [
@@ -146,7 +156,7 @@ def process_dialogue(input_content: str, piece: ChessPiece, language: str, is_au
         chess_piece_llm_generation = llm_response
         player_voice_attributes = None
 
-    print(input_content)
+    # print(input_content)
     print('='*50)
     print(llm_response)
     audio_file = generate_audio(text_response, piece)
@@ -182,14 +192,17 @@ def audio_to_base64(audio_path: str) -> str:
 def generate_response_from_audio(dialogue: AudioDialogueInput) -> DialogueOutput:
     piece = get_chess_piece(dialogue.IdPieza)
     
-    # Convert audio to base64 string
-    base64_audio = audio_to_base64(dialogue.AudioData)
-    print(dialogue.AudioData)
-    print(f"Length of base64 string: {len(base64_audio)}")
+    print("AudioData received (first 100 characters):")
+    print(dialogue.AudioData[:100])
+    
+    # Extract the base64 part from the Data URI
+    base64_audio = extract_base64(dialogue.AudioData)
+    
+    print("Extracted base64 (first 100 characters):")
+    print(base64_audio[:100])
     
     return process_dialogue(base64_audio, piece, dialogue.Language, is_audio=True)
- 
-        
+
 def generate_response_from_webcam(dialogue: WebcamDialogueInput) -> DialogueOutput:
     # Convert audio to base64 string
     base64_audio = audio_to_base64(dialogue.AudioData)