Update app.py

app.py

@@ -5,31 +5,29 @@ from groq import Groq
 
 load_dotenv()
 api = os.getenv("GROQ_API_KEY")
+client = Groq(api_key=api)
 
+### --- TAB 1: SQL Generator --- ###
 def create_prompt(user_query, table_metadata):
     system_prompt = """
 You are a SQL query generator for a single relational table.
 You must strictly follow the metadata and never guess or invent column names.
-
 Instructions:
 - Use only the table and columns listed in the metadata.
 - Never generate queries with columns not present in the metadata.
 - If a column like 'gender' is not present, do not mention it.
 - Do not hallucinate values or table names. Use provided structure only.
 - Output valid SQL (DuckDB-compatible), single line, no comments or explanations.
-
 Input:
 User Query: {user_query}
 Table Metadata:
 {table_metadata}
-
 Output:
 A single valid SQL SELECT statement using only metadata-provided columns.
     """
     return system_prompt.strip(), f"User Query: {user_query}\nTable Metadata: {table_metadata}"
 
 def generate_output(system_prompt, user_prompt):
-    client = Groq(api_key=api)
     chat_completion = client.chat.completions.create(
         messages=[
             {"role": "system", "content": system_prompt},
@@ -46,7 +44,39 @@ def response(payload):
     system_prompt, user_prompt = create_prompt(user_query, table_metadata)
     return generate_output(system_prompt, user_prompt)
 
-demo = gr.Interface(
+
+### --- TAB 2: SQL Output Explanation --- ###
+def explain_output_prompt(sql_query, query_result):
+    system_prompt = """
+You are an assistant that explains the meaning of SQL query results in plain language.
+You should take into account the SQL query used and the resulting output.
+Avoid assumptions. Focus on summarizing what the data reveals.
+"""
+    user_prompt = f"""
+SQL Query:
+{sql_query}
+
+Query Result:
+{query_result}
+
+Explanation:
+"""
+    return system_prompt.strip(), user_prompt.strip()
+
+def explain_sql_output(sql_query, query_result):
+    system_prompt, user_prompt = explain_output_prompt(sql_query, query_result)
+    chat_completion = client.chat.completions.create(
+        messages=[
+            {"role": "system", "content": system_prompt},
+            {"role": "user", "content": user_prompt}
+        ],
+        model="llama3-70b-8192"
+    )
+    return chat_completion.choices[0].message.content.strip()
+
+
+### --- Gradio Interface --- ###
+tab1 = gr.Interface(
     fn=response,
     inputs=gr.JSON(label="Input JSON (question, schema)"),
     outputs="text",
@@ -54,4 +84,18 @@ demo = gr.Interface(
     description="Input: question & table metadata. Output: SQL using dynamic schema."
 )
 
-demo.launch()
+tab2 = gr.Interface(
+    fn=explain_sql_output,
+    inputs=[
+        gr.Textbox(label="SQL Query"),
+        gr.Textbox(label="SQL Output (Raw JSON or Table Result)")
+    ],
+    outputs="text",
+    title="Explain SQL Result (Groq + LLaMA3)",
+    description="Input a SQL query and its result. Get an AI-generated explanation."
+)
+
+demo = gr.TabbedInterface([tab1, tab2], ["SQL Generator", "Explain Output"])
+
+if __name__ == '__main__':
+    demo.launch()