Spaces:
Runtime error
Runtime error
Caleb Fahlgren
commited on
Commit
·
44cb622
1
Parent(s):
a00be78
create view and get ddl for parquet
Browse files
app.py
CHANGED
|
@@ -11,7 +11,9 @@ hf_api = HfApi()
|
|
| 11 |
conn = duckdb.connect()
|
| 12 |
|
| 13 |
|
| 14 |
-
def
|
|
|
|
|
|
|
| 15 |
response = requests.get(f"{BASE_DATASETS_SERVER_URL}/parquet?dataset={dataset_id}")
|
| 16 |
response.raise_for_status() # Check if the request was successful
|
| 17 |
|
|
@@ -21,10 +23,24 @@ def query_dataset(dataset_id: str, query: str) -> pd.DataFrame:
|
|
| 21 |
if not first_parquet_url:
|
| 22 |
raise ValueError("No valid URL found for the first parquet file.")
|
| 23 |
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
-
|
| 27 |
-
return df
|
| 28 |
|
| 29 |
|
| 30 |
with gr.Blocks() as demo:
|
|
@@ -39,11 +55,12 @@ with gr.Blocks() as demo:
|
|
| 39 |
|
| 40 |
btn = gr.Button("Ask 🪄")
|
| 41 |
df = gr.DataFrame(datatype="markdown")
|
|
|
|
| 42 |
|
| 43 |
btn.click(
|
| 44 |
-
|
| 45 |
-
inputs=[dataset_name
|
| 46 |
-
outputs=[
|
| 47 |
)
|
| 48 |
|
| 49 |
|
|
|
|
| 11 |
conn = duckdb.connect()
|
| 12 |
|
| 13 |
|
| 14 |
+
def get_dataset_ddl(dataset_id: str) -> pd.DataFrame:
|
| 15 |
+
view_name = "dataset_view"
|
| 16 |
+
|
| 17 |
response = requests.get(f"{BASE_DATASETS_SERVER_URL}/parquet?dataset={dataset_id}")
|
| 18 |
response.raise_for_status() # Check if the request was successful
|
| 19 |
|
|
|
|
| 23 |
if not first_parquet_url:
|
| 24 |
raise ValueError("No valid URL found for the first parquet file.")
|
| 25 |
|
| 26 |
+
conn.execute(
|
| 27 |
+
f"CREATE OR REPLACE VIEW {view_name} as SELECT * FROM read_parquet('{first_parquet_url}');"
|
| 28 |
+
)
|
| 29 |
+
dataset_ddl = conn.execute(f"PRAGMA table_info('{view_name}');").fetchall()
|
| 30 |
+
|
| 31 |
+
column_data_types = ",\n\t".join(
|
| 32 |
+
[f"{column[1]} {column[2]}" for column in dataset_ddl]
|
| 33 |
+
)
|
| 34 |
+
|
| 35 |
+
sql_ddl = """
|
| 36 |
+
CREATE TABLE {} (
|
| 37 |
+
{}
|
| 38 |
+
);
|
| 39 |
+
""".format(
|
| 40 |
+
view_name, column_data_types
|
| 41 |
+
)
|
| 42 |
|
| 43 |
+
return sql_ddl
|
|
|
|
| 44 |
|
| 45 |
|
| 46 |
with gr.Blocks() as demo:
|
|
|
|
| 55 |
|
| 56 |
btn = gr.Button("Ask 🪄")
|
| 57 |
df = gr.DataFrame(datatype="markdown")
|
| 58 |
+
ddl = gr.Text("")
|
| 59 |
|
| 60 |
btn.click(
|
| 61 |
+
get_dataset_ddl,
|
| 62 |
+
inputs=[dataset_name],
|
| 63 |
+
outputs=[ddl],
|
| 64 |
)
|
| 65 |
|
| 66 |
|