Update app.py

app.py

@@ -3,6 +3,7 @@ import gradio as gr
 from openai import OpenAI
 import json
 import requests
+import datetime
 
 openai_api_key = os.getenv("OPENROUTER_API_KEY")
 openai_base_url = os.getenv("OPENAI_BASE_URL")
@@ -82,12 +83,12 @@ def think(request):
         #messages = [{"role": "user", "content": content}]
 
         #if reasoning != "":
-        #    messages.append({"role": "assistant", "content": "<think>\n" + reasoning + "</think>\n"})
+        #    messages.append({"role": "assistant", "content": "\n"})
         payload = {
             "model": model,
             "messages": messages,
             "include_reasoning": include_reasoning,
-            "stop": "</think>"
+            ""
         }
 
         return requests.post(url, headers=headers, data=json.dumps(payload))
@@ -111,40 +112,7 @@ def chat_with_openai(message: str, history: list, temperature: float, max_new_to
     """
     Call the OpenAI ChatCompletion endpoint using the new client and yield streaming responses.
     
-    Implements <think> logic and retries if the full response is blank.
-
-    Args:
-        message (str): The latest user message.
-        history (list): Conversation history as a list of (user, assistant) tuples.
-        temperature (float): Sampling temperature.
-        max_new_tokens (int): Maximum tokens to generate.
-
-    Yields:
-        str: Partial cumulative output from the assistant.
-    """
-
-    conversation = []
-
-    if (not history and message.startswith("Start a talk therapy session with me.")) or \
-       any(user_msg.startswith("Start a talk therapy session with me.") for user_msg, _ in history):
-        fast_mode = True    
-
-    if not history:
-        # Initialize with system prompt and assistant confirmation.
-        conversation.append({"role": "system", "content": SYSTEM_PROMPT})
-        conversation.append({"role": "assistant", "content": "Understood! I will act as the user's healthcare provider..."})
-
-    for user_msg, assistant_msg in history:
-        conversation.append({"role": "user", "content": user_msg})
-        conversation.append({"role": "assistant", "content": assistant_msg})
-
-    conversation.append({"role": "user", "content": message})
-
-    if not fast_mode:
-        # Indicate that the assistant is thinking.
-        yield "HealthAssistant is Thinking! Please wait, your response will output shortly. This may take 10-30 seconds...\n\n"
-        think_result = think(conversation)
-        conversation.append({"role": "assistant", "content": "<think>\n" + think_result + "\n</think> I will now respond to the user's message:\n\n"})
+    Implementscontent": "I will now respond to the user's message:\n\n"})
     else:
         yield "HealthAssistant is Thinking! Please wait, your response will output shortly...\n\n"
 
@@ -203,7 +171,54 @@ def chat_with_openai(message: str, history: list, temperature: float, max_new_to
         modified_full_response = apply_replacements(full_response)
         history.append((message, modified_full_response))
 
-
+def export_chat(history):
+    """Export chat history as a JSONL file in OpenAI messages format."""
+    if not history:
+        return None
+    
+    messages = []
+    # Add a blank system message (to maintain format compatibility, but without revealing the actual system prompt)
+    messages.append({"role": "system", "content": ""})
+    
+    # Convert history to messages format
+    for user_msg, assistant_msg in history:
+        messages.append({"role": "user", "content": user_msg})
+        messages.append({"role": "assistant", "content": assistant_msg})
+    
+    # Convert to JSONL format (each line is a JSON object)
+    jsonl_content = "\n".join([json.dumps(msg) for msg in messages])
+    
+    # Create a file for download
+    timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+    file_name = f"chat_export_{timestamp}.jsonl"
+    
+    return (file_name, jsonl_content)
+
+def import_chat(file):
+    """Import chat history from a JSONL file."""
+    try:
+        content = file.decode('utf-8')
+        lines = [line.strip() for line in content.split('\n') if line.strip()]
+        
+        messages = [json.loads(line) for line in lines]
+        
+        new_history = []
+        i = 0
+        
+        # Skip system message if it's the first one (we don't use imported system prompts)
+        if messages and messages[0]["role"] == "system":
+            i = 1
+        
+        while i < len(messages) - 1:
+            if messages[i]["role"] == "user" and messages[i+1]["role"] == "assistant":
+                new_history.append((messages[i]["content"], messages[i+1]["content"]))
+                i += 2
+            else:
+                i += 1
+        
+        return new_history
+    except Exception as e:
+        raise gr.Error(f"Error importing chat: {str(e)}")
 
 # Create the Chatbot component.
 chatbot = gr.Chatbot(height=450, placeholder=PLACEHOLDER, label='HealthAssistant')
@@ -215,7 +230,7 @@ with gr.Blocks(css=css) as demo:
     # Add the checkbox directly to the layout
     fast_mode_checkbox = gr.Checkbox(label="Fast Mode (Skips Reasoning. Provides Immediate, Less Accurate Responses.) RECOMMENDED FOR TALK THERAPY.", value=False)
 
-    gr.ChatInterface(
+    chat_interface = gr.ChatInterface(
         fn=chat_with_openai,
         chatbot=chatbot,
         fill_height=True,
@@ -233,6 +248,15 @@ with gr.Blocks(css=css) as demo:
         ],
         cache_examples=False,
     )
+    
+    # Add export and import buttons
+    with gr.Row():
+        export_btn = gr.Button("Export Chat")
+        import_btn = gr.UploadButton("Import Chat", file_types=[".jsonl"])
+    
+    # Connect buttons to functions
+    export_btn.click(fn=export_chat, inputs=chatbot, outputs=gr.File())
+    import_btn.upload(fn=import_chat, inputs=[import_btn], outputs=chatbot)
 
     gr.Markdown(LICENSE)