(
&self,
reader: Box<dyn FileRead>,
file_size: u64,
read_fields: &[DataField],
// TODO: support predicate pushdown for ORC (stripe pruning + row-level filtering
| 32 | #[async_trait] |
| 33 | impl FormatFileReader for OrcFormatReader { |
| 34 | async fn read_batch_stream( |
| 35 | &self, |
| 36 | reader: Box<dyn FileRead>, |
| 37 | file_size: u64, |
| 38 | read_fields: &[DataField], |
| 39 | // TODO: support predicate pushdown for ORC (stripe pruning + row-level filtering) |
| 40 | _predicates: Option<&FilePredicates>, |
| 41 | batch_size: Option<usize>, |
| 42 | row_selection: Option<Vec<RowRange>>, |
| 43 | ) -> crate::Result<ArrowRecordBatchStream> { |
| 44 | let orc_reader = OrcFileReader::new(file_size, reader); |
| 45 | |
| 46 | let builder = ArrowReaderBuilder::try_new_async(orc_reader) |
| 47 | .await |
| 48 | .map_err(|e| Error::UnexpectedError { |
| 49 | message: format!("Failed to open ORC file: {e}"), |
| 50 | source: Some(Box::new(e)), |
| 51 | })?; |
| 52 | |
| 53 | let projected_names: Vec<&str> = read_fields.iter().map(|f| f.name()).collect(); |
| 54 | let projection = |
| 55 | ProjectionMask::named_roots(builder.file_metadata().root_data_type(), &projected_names); |
| 56 | |
| 57 | let mut builder = builder.with_projection(projection); |
| 58 | |
| 59 | if let Some(size) = batch_size { |
| 60 | builder = builder.with_batch_size(size); |
| 61 | } |
| 62 | |
| 63 | if let Some(ref ranges) = row_selection { |
| 64 | let total_rows: usize = builder |
| 65 | .file_metadata() |
| 66 | .stripe_metadatas() |
| 67 | .iter() |
| 68 | .map(|s| s.number_of_rows() as usize) |
| 69 | .sum(); |
| 70 | let selection = build_range_row_selection(total_rows, ranges); |
| 71 | builder = builder.with_row_selection(selection); |
| 72 | } |
| 73 | |
| 74 | let stream = builder.build_async(); |
| 75 | Ok(stream |
| 76 | .map(|r| { |
| 77 | r.map_err(|e| Error::UnexpectedError { |
| 78 | message: format!("ORC read error: {e}"), |
| 79 | source: Some(Box::new(e)), |
| 80 | }) |
| 81 | }) |
| 82 | .boxed()) |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | // --------------------------------------------------------------------------- |
nothing calls this directly
no test coverage detected