add data preview for data curation app

wren-ai-service/eval/data_curation/app.py

@@ -5,6 +5,7 @@
 from datetime import datetime
 
 import orjson
+import pandas as pd
 import streamlit as st
 import tomlkit
 from openai import AsyncClient
@@ -72,6 +73,7 @@ def on_click_generate_question_sql_pairs(llm_client: AsyncClient):
             st.session_state["llm_model"],
             st.session_state["mdl_json"],
             st.session_state["custom_instructions_for_llm"],
+            st.session_state["data_source"],
         )
     )
 
@@ -289,9 +291,14 @@ def on_click_remove_candidate_dataset_button(i: int):
                         on_change=on_change_sql,
                         args=(i, f"sql_{i}"),
                     )
-
                     if st.session_state["llm_question_sql_pairs"][i]["is_valid"]:
                         st.success("SQL is valid")
+                        st.dataframe(
+                            pd.DataFrame(
+                                question_sql_pair["data"]["data"],
+                                columns=question_sql_pair["data"]["columns"],
+                            )
+                        )
                     else:
                         st.error(
                             f"SQL is invalid: {st.session_state["llm_question_sql_pairs"][i]["error"]}"

wren-ai-service/eval/data_curation/utils.py

@@ -305,6 +305,7 @@ async def get_question_sql_pairs(
     llm_model: str,
     mdl_json: dict,
     custom_instructions: str,
+    data_source: str,
     num_pairs: int = 10,
 ) -> list[dict]:
     messages = [
@@ -359,9 +360,12 @@ async def get_question_sql_pairs(
         question_sql_pairs = await get_validated_question_sql_pairs(results)
         sqls = [question_sql_pair["sql"] for question_sql_pair in question_sql_pairs]
         contexts = await get_contexts_from_sqls(sqls)
+        sqls_data = await get_data_from_wren_engine(sqls, data_source)
         return [
-            {**quesiton_sql_pair, "context": context}
-            for quesiton_sql_pair, context in zip(question_sql_pairs, contexts)
+            {**quesiton_sql_pair, "context": context, "data": sql_data}
+            for quesiton_sql_pair, context, sql_data in zip(
+                question_sql_pairs, contexts, sqls_data
+            )
         ]
     except Exception as e:
         st.error(f"Error generating question-sql-pairs: {e}")
@@ -376,6 +380,39 @@ def prettify_sql(sql: str) -> str:
     )
 
 
+async def get_data_from_wren_engine(sqls: List[str], data_source: str):
+    assert data_source in DATA_SOURCES, f"Invalid data source: {data_source}"
+
+    async def _get_data(sql: str, data_source: str):
+        async with aiohttp.request(
+            "POST",
+            f"{WREN_IBIS_ENDPOINT}/v2/connector/{data_source}/query",
+            json={
+                "sql": add_quotes(sql),
+                "manifestStr": base64.b64encode(
+                    orjson.dumps(st.session_state["mdl_json"])
+                ).decode(),
+                "connectionInfo": st.session_state["connection_info"],
+            },
+            timeout=aiohttp.ClientTimeout(total=60),
+        ) as response:
+            if response.status != 200:
+                return {"data": [], "columns": []}
+
+            data = await response.json()
+            column_names = [f"{i}_{col}" for i, col in enumerate(data["columns"])]
+
+            return {"data": data["data"], "columns": column_names}
+
+    async with aiohttp.ClientSession():
+        tasks = []
+        for sql in sqls:
+            task = asyncio.ensure_future(_get_data(sql, data_source))
+            tasks.append(task)
+
+        return await asyncio.gather(*tasks)
+
+
 @st.cache_data
 def get_eval_dataset_in_toml_string(mdl: dict, dataset: list) -> str:
     doc = tomlkit.document()