Update funcs.py

funcs.py

@@ -72,99 +72,98 @@ def generate_triggers_img(items):
     triggers_img = Image.open('triggeres.png')
     return triggers_img
 
-class session_processor:
-    session_conversation = []
-        
-    # Generate therapist responses and patient triggers
-    def get_doc_response_emotions(user_message, therapy_session_conversation):
-    
-        user_messages = []
-        user_messages.append(user_message)
-        emotion_set = detect_emotions(user_message)
-        print(emotion_set)
-    
-        emotions_msg = generate_triggers_img(emotion_set)
-        user_embedding = embed_model.encode(user_message, device='cuda' if torch.cuda.is_available() else 'cpu')
-    
-        similarities =[]
-        for v in dials_embeddings['embeddings']:
-          similarities.append(cosine_distance(user_embedding,v))
-    
-        top_match_index = similarities.index(max(similarities))
-        # doc_response = dials_embeddings.iloc[top_match_index+1]['Doctor']
-        doc_response = dials_embeddings.iloc[top_match_index]['Doctor']
-    
-        therapy_session_conversation.append(["User: "+user_message, "Therapist: "+doc_response])
-    
-        session_conversation.extend(["User: "+user_message, "Therapist: "+doc_response])
-    
-        print(f"User's message: {user_message}")
-        print(f"RAG Matching message: {dials_embeddings.iloc[top_match_index]['Patient']}")
-        # print(f"Therapist's response: {dials_embeddings.iloc[top_match_index+1]['Doctor']}\n\n")
-        print(f"Therapist's response: {dials_embeddings.iloc[top_match_index]['Doctor']}\n\n")
-    
-        return '', therapy_session_conversation, emotions_msg
-    
-    # Generate summarization and recommendations for teh session
-    def summarize_and_recommend():
-      session_time = str(datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
-      session_conversation_processed = session_conversation.copy()
-      session_conversation_processed.insert(0, "Session_time: "+session_time)
-      session_conversation_processed ='\n'.join(session_conversation_processed)
-      print("Session conversation:", session_conversation_processed)
-    
-      AI71_BASE_URL = "https://api.ai71.ai/v1/"
-    
-      client = OpenAI(
-          api_key=AI71_API_KEY,
-          base_url=AI71_BASE_URL,
-      )
-    
-      full_summary = ""
-      for chunk in AI71(AI71_API_KEY).chat.completions.create(
-          model="tiiuae/falcon-180b-chat",
-          messages=[
-              {"role": "system", "content": """You are an Expert Cognitive Behavioural Therapist and Precis writer.
-              Summarize the below user content <<<session_conversation_processed>>> into useful, ethical, relevant and realistic phrases with a format
-            Session Time:
-            Summary of the patient messages: #in two to four sentences
-            Summary of therapist messages: #in two to three sentences:
-            Summary of the whole session: # in two to three sentences. Ensure the entire session summary strictly does not exceed 100 tokens."""},
-              {"role": "user", "content": session_conversation_processed},
-          ],
-          stream=True,
-      ):
-          if chunk.choices[0].delta.content:
-              summary = chunk.choices[0].delta.content
-              # print("Chunk summary:", summary, sep="", end="", flush=True)
-              full_summary += summary
-      full_summary = full_summary.replace('User:', '').strip()
-      print("\n")
-      print("Full summary:", full_summary)
-    
-      full_recommendations = ""
-      for chunk in AI71(AI71_API_KEY).chat.completions.create(
-          model="tiiuae/falcon-180b-chat",
-          messages=[
-              {"role": "system", "content": """You are an expert Cognitive Behavioural Therapist.
-              Based on the full summary <<<full_summary>>> provide clinically valid, useful, appropriate action plan for the Patient as a bullted list.
-              The list shall contain both medical and non medical prescriptions, dos and donts. The format of response shall be in passive voice with proper tense.
-            - The patient is referred to........ #in one sentence
-            - The patient is advised to ........ #in one sentence
-            - The patient is refrained from........ #in one sentence
-            - It is suggested  that tha patient ........ #in one sentence
-            - Scheduled a follow-up session with the patient........#in one sentence
-              *Ensure the list contains NOT MORE THAN 7 points"""},
-              {"role": "user", "content": full_summary},
-          ],
-          stream=True,
-      ):
-          if chunk.choices[0].delta.content:
-              rec = chunk.choices[0].delta.content
-              # print("Chunk recommendation:", rec, sep="", end="", flush=True)
-              full_recommendations += rec
-      full_recommendations = full_recommendations.replace('User:', '').strip()
-      print("\n")
-      print("Full recommendations:", full_recommendations)
-      session_conversation=[]
-      return full_summary, full_recommendations
+# Generate therapist responses and patient triggers
+def get_doc_response_emotions(user_message, therapy_session_conversation):
+    global session_conversation
+
+    user_messages = []
+    user_messages.append(user_message)
+    emotion_set = detect_emotions(user_message)
+    print(emotion_set)
+
+    emotions_msg = generate_triggers_img(emotion_set)
+    user_embedding = embed_model.encode(user_message, device='cuda' if torch.cuda.is_available() else 'cpu')
+
+    similarities =[]
+    for v in dials_embeddings['embeddings']:
+      similarities.append(cosine_distance(user_embedding,v))
+
+    top_match_index = similarities.index(max(similarities))
+    # doc_response = dials_embeddings.iloc[top_match_index+1]['Doctor']
+    doc_response = dials_embeddings.iloc[top_match_index]['Doctor']
+
+    therapy_session_conversation.append(["User: "+user_message, "Therapist: "+doc_response])
+
+    session_conversation.extend(["User: "+user_message, "Therapist: "+doc_response])
+
+    print(f"User's message: {user_message}")
+    print(f"RAG Matching message: {dials_embeddings.iloc[top_match_index]['Patient']}")
+    # print(f"Therapist's response: {dials_embeddings.iloc[top_match_index+1]['Doctor']}\n\n")
+    print(f"Therapist's response: {dials_embeddings.iloc[top_match_index]['Doctor']}\n\n")
+
+    return '', therapy_session_conversation, emotions_msg
+
+# Generate summarization and recommendations for teh session
+def summarize_and_recommend():
+  global session_conversation
+  session_time = str(datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
+  session_conversation_processed = session_conversation.copy()
+  session_conversation_processed.insert(0, "Session_time: "+session_time)
+  session_conversation_processed ='\n'.join(session_conversation_processed)
+  print("Session conversation:", session_conversation_processed)
+
+  AI71_BASE_URL = "https://api.ai71.ai/v1/"
+
+  client = OpenAI(
+      api_key=AI71_API_KEY,
+      base_url=AI71_BASE_URL,
+  )
+
+  full_summary = ""
+  for chunk in AI71(AI71_API_KEY).chat.completions.create(
+      model="tiiuae/falcon-180b-chat",
+      messages=[
+          {"role": "system", "content": """You are an Expert Cognitive Behavioural Therapist and Precis writer.
+          Summarize the below user content <<<session_conversation_processed>>> into useful, ethical, relevant and realistic phrases with a format
+        Session Time:
+        Summary of the patient messages: #in two to four sentences
+        Summary of therapist messages: #in two to three sentences:
+        Summary of the whole session: # in two to three sentences. Ensure the entire session summary strictly does not exceed 100 tokens."""},
+          {"role": "user", "content": session_conversation_processed},
+      ],
+      stream=True,
+  ):
+      if chunk.choices[0].delta.content:
+          summary = chunk.choices[0].delta.content
+          # print("Chunk summary:", summary, sep="", end="", flush=True)
+          full_summary += summary
+  full_summary = full_summary.replace('User:', '').strip()
+  print("\n")
+  print("Full summary:", full_summary)
+
+  full_recommendations = ""
+  for chunk in AI71(AI71_API_KEY).chat.completions.create(
+      model="tiiuae/falcon-180b-chat",
+      messages=[
+          {"role": "system", "content": """You are an expert Cognitive Behavioural Therapist.
+          Based on the full summary <<<full_summary>>> provide clinically valid, useful, appropriate action plan for the Patient as a bullted list.
+          The list shall contain both medical and non medical prescriptions, dos and donts. The format of response shall be in passive voice with proper tense.
+        - The patient is referred to........ #in one sentence
+        - The patient is advised to ........ #in one sentence
+        - The patient is refrained from........ #in one sentence
+        - It is suggested  that tha patient ........ #in one sentence
+        - Scheduled a follow-up session with the patient........#in one sentence
+          *Ensure the list contains NOT MORE THAN 7 points"""},
+          {"role": "user", "content": full_summary},
+      ],
+      stream=True,
+  ):
+      if chunk.choices[0].delta.content:
+          rec = chunk.choices[0].delta.content
+          # print("Chunk recommendation:", rec, sep="", end="", flush=True)
+          full_recommendations += rec
+  full_recommendations = full_recommendations.replace('User:', '').strip()
+  print("\n")
+  print("Full recommendations:", full_recommendations)
+  session_conversation=[]
+  return full_summary, full_recommendations