Update app.py

app.py

@@ -4,63 +4,67 @@ import pandas as pd
 import duckdb
 import openai
 
-# 1) Load your OpenAI key from the Space’s Secrets
+# ─── 1) Load your OpenAI key from the Space’s Secrets ────────────────────────
 openai.api_key = os.getenv("OPENAI_API_KEY")
 if not openai.api_key:
     raise RuntimeError("Missing OPENAI_API_KEY secret in your Space settings")
 
-# 2) Load your CSV into DuckDB
+# ─── 2) Load your CSV into DuckDB ───────────────────────────────────────────
 df = pd.read_csv("synthetic_profit.csv")
 conn = duckdb.connect(":memory:")
 conn.register("sap", df)
 
-# 3) Build a one-line schema string for prompting
-schema = ", ".join(df.columns)
+# ─── 3) Build a one-line schema string for prompting ────────────────────────
+schema = ", ".join(df.columns)  # e.g. "Region,Product,FiscalYear, ..."
 
-# 4) Function to call OpenAI and get back a SQL string
+# ─── 4) Function to generate SQL via OpenAI’s new chat API ──────────────────
 def generate_sql(question: str) -> str:
     system = (
-        f"You are an expert SQL generator for DuckDB table `sap` with columns: {schema}. "
-        "Translate the user’s question into a valid SQL query. "
-        "Return ONLY the SQL, no explanation."
+        f"You are an expert SQL generator for a DuckDB table named `sap` "
+        f"with columns: {schema}. "
+        "Translate the user's question into a valid SQL query and return ONLY the SQL."
     )
-    resp = openai.ChatCompletion.create(
+    messages = [
+        {"role": "system", "content": system},
+        {"role": "user",   "content": question},
+    ]
+    resp = openai.chat.completions.create(
         model="gpt-3.5-turbo",
-        messages=[
-            {"role": "system",  "content": system},
-            {"role": "user",    "content": question}
-        ],
+        messages=messages,
         temperature=0.0,
         max_tokens=150,
     )
     sql = resp.choices[0].message.content.strip()
-    # Strip ``` if present
+    # strip ``` if the model wrapped it
     if sql.startswith("```") and sql.endswith("```"):
         sql = "\n".join(sql.splitlines()[1:-1])
     return sql
 
-# 5) The function Gradio will call
+# ─── 5) Core Q&A function: NL → SQL → execute → format ─────────────────────
 def answer_profitability(question: str) -> str:
-    # a) Generate the SQL
+    # a) generate SQL
     try:
         sql = generate_sql(question)
     except Exception as e:
         return f"❌ OpenAI error:\n{e}"
 
-    # b) Execute it in DuckDB
+    # b) execute it in DuckDB
     try:
-        result_df = conn.execute(sql).df()
+        df_out = conn.execute(sql).df()
     except Exception as e:
-        return f"❌ SQL error:\n{e}\n\nGenerated SQL:\n```sql\n{sql}\n```"
+        return (
+            f"❌ SQL error:\n{e}\n\n"
+            f"Generated SQL:\n```sql\n{sql}\n```"
+        )
 
-    # c) Format the output
-    if result_df.empty:
+    # c) format the result
+    if df_out.empty:
         return f"No results.\n\nSQL was:\n```sql\n{sql}\n```"
-    if result_df.shape == (1,1):
-        return str(result_df.iat[0,0])
-    return result_df.to_markdown(index=False)
+    if df_out.shape == (1,1):
+        return str(df_out.iat[0,0])
+    return df_out.to_markdown(index=False)
 
-# 6) Gradio interface—**note the explicit outputs=…** 
+# ─── 6) Gradio interface with explicit outputs ──────────────────────────────
 iface = gr.Interface(
     fn=answer_profitability,
     inputs=gr.Textbox(lines=2, placeholder="Ask a question…", label="Question"),