chore(wren-ai-service): improve ai service

wren-ai-service/Justfile

@@ -65,4 +65,7 @@ prepare-wren-engine:
 		> tools/dev/etc/mdl/sample.json
 
 use-wren-ui-as-engine:
-	poetry run python -m src.force_update_config
+	poetry run python -m src.force_update_config
+
+run-sql mdl_path="" data_source="" sample_dataset="":
+	poetry run python demo/run_sql.py --mdl-path "{{mdl_path}}" --data-source "{{data_source}}" --sample-dataset "{{sample_dataset}}"

wren-ai-service/demo/run_sql.py

@@ -0,0 +1,79 @@
+import argparse
+import json
+
+from utils import get_data_from_wren_engine, rerun_wren_engine
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Execute SQL query against MDL manifest"
+    )
+
+    parser.add_argument(
+        "--mdl-path",
+        type=str,
+        required=True,
+        help="Path to MDL JSON file",
+    )
+
+    parser.add_argument(
+        "--data-source",
+        type=str,
+        default="bigquery",
+        choices=["bigquery", "duckdb"],
+        help="Data source (default: bigquery)",
+    )
+
+    parser.add_argument(
+        "--sample-dataset",
+        type=str,
+        default="ecommerce",
+        choices=["ecommerce", "hr", ""],
+        help="Sample dataset (default: ecommerce)",
+    )
+
+    args = parser.parse_args()
+
+    mdl_path = args.mdl_path
+    data_source = args.data_source
+    sample_dataset = args.sample_dataset
+
+    # Load MDL JSON file
+    try:
+        with open(mdl_path, "r") as f:
+            mdl_json = json.load(f)
+    except FileNotFoundError:
+        print(f"Error: MDL file not found at {mdl_path}")
+        return
+    except json.JSONDecodeError:
+        print(f"Error: Invalid JSON in MDL file {mdl_path}")
+        return
+
+    rerun_wren_engine(mdl_json, data_source, sample_dataset)
+
+    # Execute query
+    print("Enter SQL query (end with semicolon on a new line to execute, 'q' to quit):")
+    lines = []
+    while True:
+        line = input()
+        if line.strip() == "q":
+            break
+        if line.strip() == ";":
+            command = "\n".join(lines)
+            lines = []
+            try:
+                df = get_data_from_wren_engine(
+                    sql=command,
+                    dataset_type=data_source,
+                    manifest=mdl_json,
+                    limit=10,
+                )
+                print(f"\nExecution result:\n{df.to_string()}\n")
+            except Exception as e:
+                print(f"\nError executing query: {str(e)}")
+        else:
+            lines.append(line)
+
+
+if __name__ == "__main__":
+    main()

wren-ai-service/demo/utils.py

@@ -64,7 +64,7 @@ def _update_wren_engine_configs(configs: list[dict]):
     assert response.status_code == 200
 
 
-def rerun_wren_engine(mdl_json: Dict, dataset_type: str, dataset: str):
+def rerun_wren_engine(mdl_json: Dict, dataset_type: str, dataset: Optional[str] = None):
     assert dataset_type in DATA_SOURCES
 
     SOURCE = dataset_type
@@ -118,7 +118,6 @@ def get_mdl_json(database_name: str):
     return mdl_json
 
 
-@st.cache_data
 def get_data_from_wren_engine(
     sql: str,
     dataset_type: str,
@@ -139,7 +138,7 @@ def get_data_from_wren_engine(
             },
         )
 
-        assert response.status_code == 200, response.json()
+        assert response.status_code == 200, response.text
 
         data = response.json()
 
@@ -162,7 +161,7 @@ def get_data_from_wren_engine(
             },
         )
 
-        assert response.status_code == 200, response.json()
+        assert response.status_code == 200, response.text
 
         data = response.json()
 
@@ -803,7 +802,6 @@ def sql_regeneration(sql_regeneration_data: dict):
         return None
 
 
-@st.cache_data
 def fill_vega_lite_values(vega_lite_schema: dict, df: pd.DataFrame) -> dict:
     """Fill Vega-Lite schema values from pandas DataFrame based on x/y encodings.
 

wren-ai-service/src/pipelines/generation/utils/chart.py

@@ -10,7 +10,8 @@
 - Please generate vega-lite schema using v5 version, which is https://vega.github.io/schema/vega-lite/v5.json
 - Chart types: Bar chart, Line chart, Area chart, Pie chart, Stacked bar chart, Grouped bar chart
 - You can only use the chart types provided in the instructions
-- If you think the data is not suitable for visualization, you can return an empty string for the schema
+- If the sample data is not suitable for visualization, you must return an empty string for the schema
+- If the sample data is empty, you must return an empty string for the schema
 - The language for the chart and reasoning must be the same language provided by the user
 - Please use the current time provided by the user to generate the chart
 - In order to generate the grouped bar chart, you need to follow the given instructions: