pub async fn fetch_rows(
results: &[SearchResult],
store: Arc<dyn Store>,
vector_column: &str,
dim: u32,
) -> AilakeResult<RecordBatch>Expand description
Fetch full row data for a slice of search results.
Groups results by Parquet file, reads each file once, extracts the matching rows
via arrow_select::take, then concatenates everything back in original top-k order
with a _distance: Float32 column appended.
Use this immediately after search() to retrieve the actual text / metadata
columns (e.g. chunk_text, document_title) alongside the distance scores.