Cleanup

src/io/parquet/read/deserialize/binary/nested.rs

@@ -3,15 +3,12 @@ use std::collections::VecDeque;
 use parquet2::{encoding::Encoding, page::DataPage, schema::Repetition};
 
 use crate::{
-    array::Offset,
-    bitmap::MutableBitmap,
-    datatypes::DataType,
-    error::Result,
+    array::Offset, bitmap::MutableBitmap, datatypes::DataType, error::Result,
     io::parquet::read::DataPages,
 };
 
-use super::super::utils::MaybeNext;
 use super::super::nested_utils::*;
+use super::super::utils::MaybeNext;
 use super::utils::Binary;
 use super::{
     super::utils,

src/io/parquet/read/file.rs

@@ -14,14 +14,13 @@ use super::{infer_schema, read_metadata, FileMetaData, RowGroupDeserializer, Row
 
 type GroupFilter = Arc<dyn Fn(usize, &RowGroupMetaData) -> bool>;
 
-/// An iterator of [`Chunk`] coming from row groups of a paquet file.
+/// An iterator of [`Chunk`]s coming from row groups of a parquet file.
 ///
-/// This can be thought of flatten chain of [`Iterator<Item=Chunk>`] - each row group is sequentially
+/// This can be thought of a flatten chain of [`Iterator<Item=Chunk>`] - each row group is sequentially
 /// mapped to an [`Iterator<Item=Chunk>`] and each iterator is iterated upon until either the limit
 /// or the last iterator ends.
-///
 /// # Implementation
-/// Note that because
+/// This iterator mixes IO-bounded and CPU-bounded operations.
 pub struct FileReader<R: Read + Seek> {
     row_groups: RowGroupReader<R>,
     metadata: FileMetaData,

src/io/parquet/read/mod.rs

@@ -7,12 +7,9 @@ mod row_group;
 pub mod schema;
 pub mod statistics;
 
-use std::{
-    io::{Read, Seek},
-    sync::Arc,
-};
-
 use futures::{AsyncRead, AsyncSeek};
+
+// re-exports of parquet2's relevant APIs
 pub use parquet2::{
     error::ParquetError,
     fallible_streaming_iterator,
@@ -32,21 +29,25 @@ pub use parquet2::{
     FallibleStreamingIterator,
 };
 
-use crate::{array::Array, error::Result};
-
 pub use deserialize::{column_iter_to_arrays, get_page_iterator};
 pub use file::{FileReader, RowGroupReader};
 pub use row_group::*;
 pub(crate) use schema::is_type_nullable;
 pub use schema::{infer_schema, FileMetaData};
 
-//use simple::nested_utils::{InitNested, NestedArrayIter, NestedState};
+use std::{
+    io::{Read, Seek},
+    sync::Arc,
+};
+
+use crate::{array::Array, error::Result};
 
 /// Trait describing a [`FallibleStreamingIterator`] of [`DataPage`]
 pub trait DataPages:
     FallibleStreamingIterator<Item = DataPage, Error = ParquetError> + Send + Sync
 {
 }
+
 impl<I: FallibleStreamingIterator<Item = DataPage, Error = ParquetError> + Send + Sync> DataPages
     for I
 {

src/io/parquet/read/row_group.rs

@@ -87,7 +87,7 @@ impl Iterator for RowGroupDeserializer {
     }
 }
 
-/// Returns all the parquet columns associated to `field_name`.
+/// Returns all [`ColumnChunkMetaData`] associated to `field_name`.
 /// For non-nested parquet types, this returns a single column
 pub(super) fn get_field_columns<'a>(
     columns: &'a [ColumnChunkMetaData],