Fix error in reading nested parquet structs (#1015)

jorgecarleitao · May 27, 2022 · 7cc874f · 7cc874f
1 parent bbe7209
commit 7cc874f
Show file tree

Hide file tree

Showing 9 changed files with 162 additions and 136 deletions.
diff --git a/src/io/parquet/read/deserialize/binary/mod.rs b/src/io/parquet/read/deserialize/binary/mod.rs
@@ -23,7 +23,7 @@ pub use dictionary::DictIter;
 /// Converts [`DataPages`] to an [`Iterator`] of [`Array`]
 pub fn iter_to_arrays_nested<'a, O, A, I>(
     iter: I,
-    init: InitNested,
+    init: Vec<InitNested>,
     data_type: DataType,
     chunk_size: usize,
 ) -> NestedArrayIter<'a>

diff --git a/src/io/parquet/read/deserialize/binary/nested.rs b/src/io/parquet/read/deserialize/binary/nested.rs
@@ -142,15 +142,15 @@ impl<'a, O: Offset> utils::Decoder<'a> for BinaryDecoder<O> {
 pub struct ArrayIterator<O: Offset, A: TraitBinaryArray<O>, I: DataPages> {
     iter: I,
     data_type: DataType,
-    init: InitNested,
+    init: Vec<InitNested>,
     items: VecDeque<(Binary<O>, MutableBitmap)>,
     nested: VecDeque<NestedState>,
     chunk_size: usize,
     phantom_a: std::marker::PhantomData<A>,
 }
 
 impl<O: Offset, A: TraitBinaryArray<O>, I: DataPages> ArrayIterator<O, A, I> {
-    pub fn new(iter: I, init: InitNested, data_type: DataType, chunk_size: usize) -> Self {
+    pub fn new(iter: I, init: Vec<InitNested>, data_type: DataType, chunk_size: usize) -> Self {
         Self {
             iter,
             data_type,

diff --git a/src/io/parquet/read/deserialize/boolean/mod.rs b/src/io/parquet/read/deserialize/boolean/mod.rs
@@ -16,7 +16,7 @@ pub use self::basic::Iter;
 /// Converts [`DataPages`] to an [`Iterator`] of [`Array`]
 pub fn iter_to_arrays_nested<'a, I: 'a>(
     iter: I,
-    init: InitNested,
+    init: Vec<InitNested>,
     chunk_size: usize,
 ) -> NestedArrayIter<'a>
 where

diff --git a/src/io/parquet/read/deserialize/boolean/nested.rs b/src/io/parquet/read/deserialize/boolean/nested.rs
@@ -113,15 +113,15 @@ impl<'a> Decoder<'a> for BooleanDecoder {
 #[derive(Debug)]
 pub struct ArrayIterator<I: DataPages> {
     iter: I,
-    init: InitNested,
+    init: Vec<InitNested>,
     // invariant: items.len() == nested.len()
     items: VecDeque<(MutableBitmap, MutableBitmap)>,
     nested: VecDeque<NestedState>,
     chunk_size: usize,
 }
 
 impl<I: DataPages> ArrayIterator<I> {
-    pub fn new(iter: I, init: InitNested, chunk_size: usize) -> Self {
+    pub fn new(iter: I, init: Vec<InitNested>, chunk_size: usize) -> Self {
         Self {
             iter,
             init,