UpdateBen app.py

app.py

@@ -1,20 +1,20 @@
 import gradio as gr
 from transformers import pipeline
 
-# 🔁 In-memory memory store
+# In-memory memory store
 memory_db = {}
 
-# 🤖 Load Hugging Face text-to-text model (CPU-friendly)
+# Load the model
 interpreter = pipeline("text2text-generation", model="google/flan-t5-base")
 
-# 🧠 Core message interpretation function
+# Interpret a message with optional memory and return AI explanation + history
 def interpret_message(msg, name, closeness, traits, relation, past_events, feelings, use_memory):
     if not msg.strip() or not name.strip():
-        return "Please enter both a message and a person’s name."
+        return "Please enter both a message and a person’s name.", ""
 
     name = name.strip().title()
 
-    # Load memory if exists
+    # Initialize or update memory
     if use_memory:
         if name in memory_db:
             existing = memory_db[name]
@@ -29,19 +29,10 @@ def interpret_message(msg, name, closeness, traits, relation, past_events, feeli
                 "history": []
             }
 
-    # Create structured prompt
-    prompt = f"""New message from {name}: "{msg}"
+    # Construct rich prompt
+    prompt = f"""New message from {name}: \"{msg}\"\n\nKnown context:\n- Closeness: {closeness}\n- Personality: {traits}\n- Relationship: {relation}\n- Similar Past Events: {past_events}\n- Your Feelings: {feelings}\n\nBased on the message and context, explain what the person is *really thinking* or feeling but not saying directly.\nInfer their hidden intention, emotional state, and subtext. Then suggest how I should respond."""
 
-Known context:
-- Closeness: {closeness}
-- Personality: {traits}
-- Relationship: {relation}
-- Similar Past Events: {past_events}
-- Your Feelings: {feelings}
-
-Interpret the likely tone and intention. Suggest a possible meaning and how I could respond."""
-
-    # Update memory before model (to include prompt)
+    # Save to memory if applicable
     if use_memory:
         memory_db[name]["closeness"] = closeness
         memory_db[name]["personality"] = traits
@@ -52,16 +43,38 @@ Interpret the likely tone and intention. Suggest a possible meaning and how I co
             "interpretation": "TO BE FILLED"
         })
 
-    # Generate interpretation from model
-    result = interpreter(prompt, max_new_tokens=150)[0]["generated_text"]
+    # Run model
+    result = interpreter(prompt, max_new_tokens=200)[0]["generated_text"]
 
-    # Update memory with interpretation
     if use_memory:
         memory_db[name]["history"][-1]["interpretation"] = result
 
-    return f"🧠 Memory active for {name}.\n\n**Interpretation:**\n{result}"
+    # Format memory log
+    memory_log = ""
+    if use_memory and memory_db[name]["history"]:
+        memory_log = "\n\n".join(
+            [f"\u2022 \"{entry['message']}\"\n  \u2192 {entry['interpretation']}" for entry in memory_db[name]["history"]]
+        )
+
+    return result, memory_log
+
+# Chatbot advisor based on memory
+def memory_chat(user_input, person_name):
+    name = person_name.strip().title()
+    if name not in memory_db:
+        return "There's no memory for this person yet. Try interpreting a message first."
+
+    # Summarize memory history into prompt
+    history = memory_db[name]["history"]
+    past_text = "\n\n".join([f"- {h['message']}\n  -> {h['interpretation']}" for h in history])
+
+    chat_prompt = f"""You are a smart advisor trained on past conversations with {name}.\n
+Here is the message history:\n{past_text}\n\nUser asks: {user_input}\n\nGive a thoughtful answer using patterns or emotional tone from the past history."""
 
-# 🎛️ Gradio UI
+    response = interpreter(chat_prompt, max_new_tokens=200)[0]["generated_text"]
+    return response
+
+# Gradio UI
 with gr.Blocks(title="Textual Chemistry") as app:
     gr.Markdown("## 💬 Textual Chemistry\n_What do they really mean?_")
 
@@ -73,8 +86,7 @@ with gr.Blocks(title="Textual Chemistry") as app:
     with gr.Row():
         closeness = gr.Dropdown(
             choices=["Stranger", "Acquaintance", "Friend", "Close Friend", "Partner", "Ex", "Boss", "Teacher"],
-            label="Closeness Level"
-        )
+            label="Closeness Level")
         use_memory = gr.Checkbox(label="Use memory for this person?", value=True)
 
     sender_traits = gr.Textbox(label="Background on the Sender", placeholder="e.g., Sarcastic, hates small talk...")
@@ -82,13 +94,24 @@ with gr.Blocks(title="Textual Chemistry") as app:
     past_events = gr.Textbox(label="Similar Past Situations", placeholder="e.g., Last time I canceled they acted cold")
     user_feelings = gr.Textbox(label="Your Current Feelings", placeholder="e.g., I feel confused or anxious")
 
-    output = gr.Textbox(label="AI Interpretation", lines=10)
+    output = gr.Textbox(label="🧠 Interpretation", lines=6)
+    history_output = gr.Textbox(label="🧠 Memory Log for This Person", lines=10)
 
     submit_btn = gr.Button("Interpret Message")
     submit_btn.click(
         fn=interpret_message,
         inputs=[message, person_name, closeness, sender_traits, relationship, past_events, user_feelings, use_memory],
-        outputs=output
+        outputs=[output, history_output]
     )
 
+    gr.Markdown("---")
+    gr.Markdown("## 🧠 Memory Advisor Chatbot")
+
+    with gr.Row():
+        chatbot_input = gr.Textbox(label="Ask a follow-up question about this person")
+        chatbot_output = gr.Textbox(label="🤖 Advisor Response", lines=6)
+
+    chatbot_btn = gr.Button("Ask Memory Bot")
+    chatbot_btn.click(fn=memory_chat, inputs=[chatbot_input, person_name], outputs=chatbot_output)
+
 app.launch()