Polish better decoder (Testing)

parquet/src/arrow/array_reader/byte_view_array.rs

@@ -432,7 +432,8 @@ impl ByteViewArrayDecoderDictionary {
         }
     }
 
-    /// Reads the next indexes from self.decoder
+    /// Reads the next `len` indexes from self.decoder
+    ///
     /// the indexes are assumed to be indexes into `dict`
     /// the output values are written to output
     ///
@@ -464,6 +465,8 @@ impl ByteViewArrayDecoderDictionary {
             }
         }
 
+        output.views.reserve(len);
+
         // Calculate the offset of the dictionary buffers in the output buffers
         // For example if the 2nd buffer in the dictionary is the 5th buffer in the output buffers,
         // then the base_buffer_idx is 5 - 2 = 3

parquet/src/arrow/arrow_reader/mod.rs

@@ -748,6 +748,7 @@ impl<T: ChunkReader + 'static> ReaderPageIterator<T> {
             .map(|i| i[rg_idx][self.column_idx].page_locations.clone());
         let total_rows = rg.num_rows() as usize;
         let reader = self.reader.clone();
+        // todo: add cache???
 
         SerializedPageReader::new(reader, column_chunk_metadata, total_rows, page_locations)?
             .add_crypto_context(