Update app.py

app.py

@@ -12,8 +12,8 @@ if 'audioop' not in sys.modules:
 import os
 import sqlite3
 import pandas as pd
+import json
 import gradio as gr
-from datetime import datetime
 from groq import Groq
 import requests
 
@@ -55,7 +55,47 @@ def init_db():
 
 init_db()
 
-# --- Core Query Logic ---
+# --- Common LLM API Request Orchestrator ---
+def call_llm(provider, api_key, model_choice, system_prompt, user_message, chat_history_format=None):
+    # Formulate standardized payload messages setup
+    messages = [{"role": "system", "content": system_prompt}]
+    
+    if chat_history_format:
+        messages.extend(chat_history_format)
+    else:
+        messages.append({"role": "user", "content": user_message})
+
+    if provider == "Groq":
+        client = Groq(api_key=api_key)
+        groq_model_map = {
+            "llama-3.1-8b-instant": "llama-3.1-8b-instant",
+            "llama-3.1-70b-versatile": "llama-3.1-70b-versatile",
+            "openai/gpt-oss-20b": "llama-3.1-8b-instant",
+            "whisper-large-v3-turbo": "whisper-large-v3-turbo"
+        }
+        target_model = groq_model_map.get(model_choice, "llama-3.1-8b-instant")
+        response = client.chat.completions.create(
+            model=target_model, messages=messages, temperature=0.0
+        )
+        return response.choices[0].message.content.strip()
+
+    elif provider == "OpenRouter":
+        headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
+        openrouter_model_map = {
+            "llama-3.1-8b-instant": "meta-llama/llama-3.1-8b-instruct",
+            "llama-3.1-70b-versatile": "meta-llama/llama-3.1-70b-instruct",
+            "openai/gpt-oss-20b": "openai/gpt-3.5-turbo",
+            "whisper-large-v3-turbo": "meta-llama/llama-3.1-8b-instruct"
+        }
+        target_model = openrouter_model_map.get(model_choice, "meta-llama/llama-3.1-8b-instruct")
+        payload = {"model": target_model, "messages": messages, "temperature": 0.0}
+        response = requests.post("https://openrouter.ai/api/v1/chat/completions", headers=headers, json=payload)
+        if response.status_code == 200:
+            return response.json()['choices'][0]['message']['content'].strip()
+        else:
+            raise Exception(f"OpenRouter Error: {response.text}")
+
+# --- Tab 1: Text-to-SQL Reader Desk Core Logic ---
 def execute_ai_query(provider, api_key, model_choice, user_question):
     if not api_key.strip():
         return None, "⚠️ Please enter a valid API Key to proceed."
@@ -70,117 +110,141 @@ def execute_ai_query(provider, api_key, model_choice, user_question):
         "Output raw text only."
     )
 
-    # Context map adjustment for audio transcription models vs LLM text interfaces
-    if "whisper" in model_choice.lower():
-        system_prompt += " Even though your engine architecture is optimized for transcription pipelines, execute this as text-to-SQL logic and return only the raw SQLite string."
-
     try:
-        sql_query = ""
-        
-        # Choice A: Groq API Gateway Routing
-        if provider == "Groq":
-            client = Groq(api_key=api_key)
-            
-            # Map clean names to Groq-supported technical production IDs
-            groq_model_map = {
-                "llama-3.1-8b-instant": "llama-3.1-8b-instant",
-                "llama-3.1-70b-versatile": "llama-3.1-70b-versatile",
-                "openai/gpt-oss-20b": "llama-3.1-8b-instant", # Fallback map if provider doesn't host open-source mock
-                "whisper-large-v3-turbo": "whisper-large-v3-turbo"
-            }
-            target_model = groq_model_map.get(model_choice, "llama-3.1-8b-instant")
-            
-            response = client.chat.completions.create(
-                model=target_model,
-                messages=[
-                    {"role": "system", "content": system_prompt},
-                    {"role": "user", "content": user_question}
-                ],
-                temperature=0.0
-            )
-            sql_query = response.choices[0].message.content.strip()
-
-        # Choice B: OpenRouter API Gateway Routing
-        elif provider == "OpenRouter":
-            headers = {
-                "Authorization": f"Bearer {api_key}",
-                "Content-Type": "application/json"
-            }
-            
-            # Map clean names to OpenRouter specific global endpoints
-            openrouter_model_map = {
-                "llama-3.1-8b-instant": "meta-llama/llama-3.1-8b-instruct",
-                "llama-3.1-70b-versatile": "meta-llama/llama-3.1-70b-instruct",
-                "openai/gpt-oss-20b": "openai/gpt-3.5-turbo", # Routed fallback target
-                "whisper-large-v3-turbo": "meta-llama/llama-3.1-8b-instruct" # Text fallback routing
-            }
-            target_model = openrouter_model_map.get(model_choice, "meta-llama/llama-3.1-8b-instruct")
-            
-            payload = {
-                "model": target_model,
-                "messages": [
-                    {"role": "system", "content": system_prompt},
-                    {"role": "user", "content": user_question}
-                ],
-                "temperature": 0.0
-            }
-            response = requests.post("https://openrouter.ai/api/v1/chat/completions", headers=headers, json=payload)
-            if response.status_code == 200:
-                sql_query = response.json()['choices'][0]['message']['content'].strip()
-            else:
-                return None, f"❌ OpenRouter Error: {response.text}"
-
-        # Clean formatting wrappers from LLM output
+        sql_query = call_llm(provider, api_key, model_choice, system_prompt, user_question)
         sql_query = sql_query.replace("```sql", "").replace("```", "").replace("`", "").strip()
 
-        # Execute query securely against local SQLite engine
         conn = sqlite3.connect(DB_NAME)
         df = pd.read_sql_query(sql_query, conn)
         conn.close()
+        return df, f"⚡ **Generated SQL Query:** `{sql_query}`"
+    except Exception as e:
+        return None, f"❌ Execution Failed: {str(e)}"
 
-        return df, f"⚡ **Generated SQL Query via {model_choice}:** `{sql_query}`"
+# --- Tab 2: Conversation & Data Append Agent Core Logic ---
+def conversation_and_commit_agent(chat_history, provider, api_key, model_choice, user_msg):
+    if not api_key.strip():
+        chat_history.append({"role": "assistant", "content": "⚠️ Authentication missing. Please input your API Key on the left menu pane."})
+        return chat_history, ""
+    if not user_msg.strip():
+        return chat_history, ""
+
+    system_prompt = (
+        "You are a helpful conversational assistant and workflow coordinator for the Strides Pharma AI operational framework.\n"
+        "Your goal is twofold:\n"
+        "1. Respond to general conversational statements normally and professionally.\n"
+        "2. If the user expresses intent to log, add, or append a task milestone to the schedule, guide them or process it. "
+        "The required metrics are: Phase, Task_Name, Owner, Timeline, and Priority.\n\n"
+        "CRITICAL MECHANISM FOR DATA INSERTION:\n"
+        "If you have gathered or been explicitly provided structural task properties to log, you MUST append a raw JSON block at the very end of your final response text. "
+        "The format must exactly be: ||JSON_DATA: {\"phase\": \"...\", \"task_name\": \"...\", \"owner\": \"...\", \"timeline\": \"...\", \"priority\": \"...\"}||\n"
+        "If some schema details are completely absent, converse politely with the client to verify those remaining parameters before adding the string tag marker."
+    )
+
+    # Process conversational memory structure
+    formatted_history = []
+    for turn in chat_history:
+        formatted_history.append({"role": turn["role"], "content": turn["content"]})
+    formatted_history.append({"role": "user", "content": user_msg})
+
+    try:
+        # Fetch target assistant reply
+        raw_response = call_llm(provider, api_key, model_choice, system_prompt, "", chat_history_format=formatted_history)
+        
+        # Intercept database tags if present
+        cleaned_response = raw_response
+        database_committed_alert = ""
+        
+        if "||JSON_DATA:" in raw_response:
+            try:
+                parts = raw_response.split("||JSON_DATA:")
+                cleaned_response = parts[0].strip()
+                json_string = parts[1].replace("||", "").strip()
+                
+                # Extract dictionary properties cleanly
+                data_payload = json.loads(json_string)
+                
+                # Append to SQLite physical table
+                conn = sqlite3.connect(DB_NAME)
+                cursor = conn.cursor()
+                cursor.execute(
+                    "INSERT INTO project_tasks (phase, task_name, owner, timeline, priority) VALUES (?, ?, ?, ?, ?)",
+                    (data_payload.get('phase'), data_payload.get('task_name'), data_payload.get('owner'), data_payload.get('timeline'), data_payload.get('priority'))
+                )
+                conn.commit()
+                conn.close()
+                
+                database_committed_alert = f"\n\n⚙️ **[SYSTEM UPDATE]:** Successfully appended task '{data_payload.get('task_name')}' to the internal server database master record structure."
+            except Exception as inner_err:
+                database_committed_alert = f"\n\n⚠️ **[SYSTEM NOTICE]:** Captured structure request token, but insertion execution aborted due to tracking parsing discrepancies: {str(inner_err)}"
+
+        final_display_text = cleaned_response + database_committed_alert
+        
+        chat_history.append({"role": "user", "content": user_msg})
+        chat_history.append({"role": "assistant", "content": final_display_text})
+        return chat_history, ""
 
     except Exception as e:
-        return None, f"❌ Execution Failed: {str(e)}"
+        chat_history.append({"role": "user", "content": user_msg})
+        chat_history.append({"role": "assistant", "content": f"❌ Error communicating with LLM orchestration layout: {str(e)}"})
+        return chat_history, ""
 
-# --- Interface Layout ---
+# --- Interface Layout Configuration ---
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 💊 Strides Pharma AI Data Portal")
-    gr.Markdown("Transform conversational natural language directly into validated SQL operations.")
+    gr.Markdown("# 💊 Strides Pharma AI Production & Data Portal")
+    gr.Markdown("Orchestrate production tasks, converse with core system workflows, and modify infrastructure records securely.")
     
     with gr.Row():
+        # Universal Infrastructure Left Control Tower Pane
         with gr.Column(scale=1):
-            gr.Markdown("### 🔑 Authentication & LLM Setup")
+            gr.Markdown("### 🔑 Global Control Tower")
             provider_select = gr.Dropdown(choices=["Groq", "OpenRouter"], value="Groq", label="API Gateway Provider")
             token_input = gr.Textbox(label="User API Secret Key", type="password", placeholder="gsk_... or sk-or-...")
-            
-            # Updated Model Selection Field
             model_select = gr.Dropdown(
-                choices=[
-                    "llama-3.1-8b-instant",
-                    "llama-3.1-70b-versatile",
-                    "openai/gpt-oss-20b",
-                    "whisper-large-v3-turbo"
-                ], 
+                choices=["llama-3.1-8b-instant", "llama-3.1-70b-versatile", "openai/gpt-oss-20b", "whisper-large-v3-turbo"], 
                 value="llama-3.1-8b-instant", 
-                label="Target Language Model"
+                label="Target Engine Architecture"
             )
+            gr.Markdown("✨ **Global Platform Database State Schema:**\n- `phase` (Discovery, Data Prep, Modeling, Validation)\n- `task_name` (Structural workflow description string)\n- `owner` (Assigned personnel scientist name)\n- `timeline` (Expected operational delivery time frames)\n- `priority` (High, Medium, Low)")
             
-            gr.Markdown("💡 **Sample Queries to try:**\n- *Show me all tasks where priority is High*\n- *List tasks owned by M. Reddy*\n- *Count the number of tasks in each phase*")
-            
+        # Interactive Operational Tab Elements
         with gr.Column(scale=2):
-            gr.Markdown("### 🤖 Data Query Desk")
-            query_input = gr.Textbox(label="Ask your database a question:", placeholder="e.g., Show all tasks in the Validation phase")
-            submit_btn = gr.Button("Query Database Infrastructure", variant="primary")
-            
-            sql_status = gr.Markdown()
-            output_table = gr.DataFrame(label="Queried Records Dataframe Output")
-
-    submit_btn.click(
-        fn=execute_ai_query,
-        inputs=[provider_select, token_input, model_select, query_input],
-        outputs=[output_table, sql_status]
-    )
+            with gr.Tabs():
+                
+                # Tab A View: Operational Task Creator Conversational Chatbot
+                with gr.TabItem("🤖 Interactive Data Contributor Chatbot"):
+                    gr.Markdown("### Conversational Contributor Agent")
+                    gr.Markdown("Chat with this engine normally, or tell it to log a brand-new task assignment milestone directly into the active SQLite infrastructure.")
+                    
+                    chatbot_viewport = gr.Chatbot(type="messages", label="Operational History Workspace")
+                    chat_input = gr.Textbox(placeholder="Say hello, or submit task details to log (e.g., Log a task in Validation phase named Run final audit owned by M. Reddy for Week 8 with High priority)", label="Your Message")
+                    send_btn = gr.Button("Submit Message", variant="primary")
+                    
+                    send_btn.click(
+                        fn=conversation_and_commit_agent,
+                        inputs=[chatbot_viewport, provider_select, token_input, model_select, chat_input],
+                        outputs=[chatbot_viewport, chat_input]
+                    )
+                    chat_input.submit(
+                        fn=conversation_and_commit_agent,
+                        inputs=[chatbot_viewport, provider_select, token_input, model_select, chat_input],
+                        outputs=[chatbot_viewport, chat_input]
+                    )
+                
+                # Tab B View: Legacy Natural-Language-to-SQL Inquisitor Node
+                with gr.TabItem("🔎 SQL Inquisitor Desk"):
+                    gr.Markdown("### Natural Language SQL Query Engine")
+                    query_input = gr.Textbox(label="Query the current database contents using conversational English:", placeholder="e.g., Show me all records sorted by priority status")
+                    query_btn = gr.Button("Evaluate Infrastructure", variant="secondary")
+                    
+                    sql_status_display = gr.Markdown()
+                    output_data_table = gr.DataFrame(label="Queried Records Dataframe Live Output")
+                    
+                    query_btn.click(
+                        fn=execute_ai_query,
+                        inputs=[provider_select, token_input, model_select, query_input],
+                        outputs=[output_data_table, sql_status_display]
+                    )
 
 if __name__ == "__main__":
     demo.launch()