feat: add columns using pyarrow schema (#2284)

This commit is contained in:
Lei Xu
2025-03-28 08:51:50 -07:00
committed by GitHub
parent c321cccc12
commit f52d05d3fa
4 changed files with 75 additions and 8 deletions

View File

@@ -1,9 +1,11 @@
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright The LanceDB Authors
use std::{collections::HashMap, sync::Arc};
use arrow::{
datatypes::DataType,
datatypes::{DataType, Schema},
ffi_stream::ArrowArrayStreamReader,
pyarrow::{FromPyArrow, ToPyArrow},
pyarrow::{FromPyArrow, PyArrowType, ToPyArrow},
};
use lancedb::table::{
AddDataMode, ColumnAlteration, Duration, NewColumnTransform, OptimizeAction, OptimizeOptions,
@@ -16,7 +18,6 @@ use pyo3::{
Bound, FromPyObject, PyAny, PyRef, PyResult, Python,
};
use pyo3_async_runtimes::tokio::future_into_py;
use std::collections::HashMap;
use crate::{
error::PythonErrorExt,
@@ -444,6 +445,20 @@ impl Table {
})
}
pub fn add_columns_with_schema(
self_: PyRef<'_, Self>,
schema: PyArrowType<Schema>,
) -> PyResult<Bound<'_, PyAny>> {
let arrow_schema = &schema.0;
let transform = NewColumnTransform::AllNulls(Arc::new(arrow_schema.clone()));
let inner = self_.inner_ref()?.clone();
future_into_py(self_.py(), async move {
inner.add_columns(transform, None).await.infer_error()?;
Ok(())
})
}
pub fn alter_columns<'a>(
self_: PyRef<'a, Self>,
alterations: Vec<Bound<PyDict>>,