fix: cast thrift encoded values to ints to preven issues with 32 systems

Author: norberttech

src/lib/parquet/src/Flow/Parquet/ParquetFile/Metadata.php

@@ -24,8 +24,8 @@ public static function fromThrift(FileMetaData $thrift, Options $options) : self
         return new self(
             Schema::fromThrift($thrift->schema),
             RowGroups::fromThrift($thrift->row_groups, $options),
-            $thrift->num_rows,
-            $thrift->version,
+            (int) $thrift->num_rows,
+            (int) $thrift->version,
             $thrift->created_by
         );
     }

src/lib/parquet/src/Flow/Parquet/ParquetFile/Page/Header/DataPageHeader.php

@@ -22,7 +22,7 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\DataPageHeader $thri
             Encodings::from($thrift->encoding),
             Encodings::from($thrift->repetition_level_encoding),
             Encodings::from($thrift->definition_level_encoding),
-            $thrift->num_values
+            (int) $thrift->num_values
         );
     }
 

src/lib/parquet/src/Flow/Parquet/ParquetFile/Page/Header/DataPageHeaderV2.php

@@ -25,12 +25,12 @@ public function __construct(
     public static function fromThrift(\Flow\Parquet\ThriftModel\DataPageHeaderV2 $thrift, Options $options) : self
     {
         return new self(
-            $thrift->num_values,
-            $thrift->num_nulls,
-            $thrift->num_rows,
+            (int) $thrift->num_values,
+            (int) $thrift->num_nulls,
+            (int) $thrift->num_rows,
             Encodings::from($thrift->encoding),
-            $thrift->definition_levels_byte_length,
-            $thrift->repetition_levels_byte_length,
+            (int) $thrift->definition_levels_byte_length,
+            (int) $thrift->repetition_levels_byte_length,
             /** @phpstan-ignore-next-line */
             $thrift->is_compressed ?? null,
             $thrift->statistics ? Statistics::fromThrift($thrift->statistics) : null,

src/lib/parquet/src/Flow/Parquet/ParquetFile/Page/Header/DictionaryPageHeader.php

@@ -18,7 +18,7 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\DictionaryPageHeader
     {
         return new self(
             Encodings::from($thrift->encoding),
-            $thrift->num_values
+            (int) $thrift->num_values
         );
     }
 

src/lib/parquet/src/Flow/Parquet/ParquetFile/Page/PageHeader.php

@@ -24,8 +24,8 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\PageHeader $thrift,
     {
         return new self(
             Type::from($thrift->type),
-            $thrift->compressed_page_size,
-            $thrift->uncompressed_page_size,
+            (int) $thrift->compressed_page_size,
+            (int) $thrift->uncompressed_page_size,
             $thrift->data_page_header !== null ? DataPageHeader::fromThrift($thrift->data_page_header) : null,
             $thrift->data_page_header_v2 !== null ? DataPageHeaderV2::fromThrift($thrift->data_page_header_v2, $options) : null,
             $thrift->dictionary_page_header !== null ? DictionaryPageHeader::fromThrift($thrift->dictionary_page_header) : null

src/lib/parquet/src/Flow/Parquet/ParquetFile/RowGroup.php

@@ -25,7 +25,7 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\RowGroup $thrift, Op
     {
         return new self(
             \array_map(static fn (\Flow\Parquet\ThriftModel\ColumnChunk $columnChunk) => ColumnChunk::fromThrift($columnChunk, $options), $thrift->columns),
-            $thrift->num_rows
+            (int) $thrift->num_rows
         );
     }
 

src/lib/parquet/src/Flow/Parquet/ParquetFile/RowGroup/ColumnChunk.php

@@ -46,15 +46,15 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\ColumnChunk $thrift,
         return new self(
             PhysicalType::from($thrift->meta_data->type),
             Compressions::from($thrift->meta_data->codec),
-            $thrift->meta_data->num_values,
-            $thrift->file_offset,
+            (int) $thrift->meta_data->num_values,
+            (int) $thrift->file_offset,
             $thrift->meta_data->path_in_schema,
             \array_map(static fn ($encoding) => Encodings::from($encoding), $thrift->meta_data->encodings),
-            $thrift->meta_data->total_compressed_size,
-            $thrift->meta_data->total_uncompressed_size,
-            $thrift->meta_data->dictionary_page_offset,
-            $thrift->meta_data->data_page_offset,
-            $thrift->meta_data->index_page_offset,
+            (int) $thrift->meta_data->total_compressed_size,
+            (int) $thrift->meta_data->total_uncompressed_size,
+            $thrift->meta_data->dictionary_page_offset !== null ? (int) $thrift->meta_data->dictionary_page_offset : null,
+            $thrift->meta_data->data_page_offset !== null ? (int) $thrift->meta_data->data_page_offset : null,
+            $thrift->meta_data->index_page_offset !== null ? (int) $thrift->meta_data->index_page_offset : null,
             $thrift->meta_data->statistics ? Statistics::fromThrift($thrift->meta_data->statistics) : null,
             $options
         );

src/lib/parquet/src/Flow/Parquet/ParquetFile/Schema/FlatColumn.php

@@ -95,9 +95,9 @@ public static function fromThrift(SchemaElement $thrift) : self
             $thrift->converted_type === null ? null : ConvertedType::from($thrift->converted_type),
             $thrift->logicalType === null ? null : LogicalType::fromThrift($thrift->logicalType),
             $thrift->repetition_type === null ? null : Repetition::from($thrift->repetition_type),
-            $thrift->precision,
-            $thrift->scale,
-            $thrift->type_length,
+            $thrift->precision !== null ? (int) $thrift->precision : null,
+            $thrift->scale !== null ? (int) $thrift->scale : null,
+            $thrift->type_length !== null ? (int) $thrift->type_length : null,
         );
     }
 

src/lib/parquet/src/Flow/Parquet/ParquetFile/Schema/LogicalType/Decimal.php

@@ -17,8 +17,8 @@ public function __construct(
     public static function fromThrift(DecimalType $thrift) : self
     {
         return new self(
-            $thrift->scale,
-            $thrift->precision
+            (int) $thrift->scale,
+            (int) $thrift->precision
         );
     }
 

src/lib/parquet/src/Flow/Parquet/ParquetFile/Statistics.php

@@ -24,8 +24,8 @@ public static function fromThrift(\Flow\Parquet\ThriftModel\Statistics $thrift)
         return new self(
             $thrift->max,
             $thrift->min,
-            $thrift->null_count,
-            $thrift->distinct_count,
+            $thrift->null_count !== null ? (int) $thrift->null_count : null,
+            $thrift->distinct_count !== null ? (int) $thrift->distinct_count : null,
             $thrift->max_value,
             $thrift->min_value,
             $thrift->is_max_value_exact,