duckdb 1.4.1 → 1.4.3-dev0.0

package/src/duckdb/third_party/parquet/parquet_types.cpp

@@ -1,5 +1,5 @@
 /**
- * Autogenerated by Thrift Compiler (0.21.0)
+ * Autogenerated by Thrift Compiler (0.22.0)
  *
  * DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING
  *  @generated
@@ -13,6 +13,14 @@
 
 namespace duckdb_parquet {
 
+template <class ENUM>
+static typename ENUM::type SafeEnumCast(const std::map<int, const char*> &values_to_names, const int &ecast) {
+  if (values_to_names.find(ecast) == values_to_names.end()) {
+    throw duckdb_apache::thrift::protocol::TProtocolException(duckdb_apache::thrift::protocol::TProtocolException::INVALID_DATA);
+  }
+  return static_cast<typename ENUM::type>(ecast);
+}
+
 int _kTypeValues[] = {
   Type::BOOLEAN,
   Type::INT32,
@@ -176,7 +184,14 @@ int _kConvertedTypeValues[] = {
    * the provided duration.  This duration of time is independent of any
    * particular timezone or date.
    */
-  ConvertedType::INTERVAL
+  ConvertedType::INTERVAL,
+  /**
+   * Non-standard NULL value
+   * 
+   * This was written by old writers - it is kept here for compatibility purposes.
+   * See https://github.com/duckdb/duckdb/pull/11774
+   */
+  ConvertedType::PARQUET_NULL
 };
 const char* _kConvertedTypeNames[] = {
   /**
@@ -300,9 +315,16 @@ const char* _kConvertedTypeNames[] = {
    * the provided duration.  This duration of time is independent of any
    * particular timezone or date.
    */
-  "INTERVAL"
+  "INTERVAL",
+  /**
+   * Non-standard NULL value
+   * 
+   * This was written by old writers - it is kept here for compatibility purposes.
+   * See https://github.com/duckdb/duckdb/pull/11774
+   */
+  "PARQUET_NULL"
 };
-const std::map<int, const char*> _ConvertedType_VALUES_TO_NAMES(::apache::thrift::TEnumIterator(22, _kConvertedTypeValues, _kConvertedTypeNames), ::apache::thrift::TEnumIterator(-1, nullptr, nullptr));
+const std::map<int, const char*> _ConvertedType_VALUES_TO_NAMES(::apache::thrift::TEnumIterator(23, _kConvertedTypeValues, _kConvertedTypeNames), ::apache::thrift::TEnumIterator(-1, nullptr, nullptr));
 
 std::ostream& operator<<(std::ostream& out, const ConvertedType::type& val) {
   std::map<int, const char*>::const_iterator it = _ConvertedType_VALUES_TO_NAMES.find(val);
@@ -3446,7 +3468,7 @@ GeographyType::~GeographyType() noexcept {
 
 GeographyType::GeographyType() noexcept
    : crs(),
-     algorithm(static_cast<EdgeInterpolationAlgorithm::type>(0)) {
+     algorithm(SafeEnumCast<EdgeInterpolationAlgorithm>(_EdgeInterpolationAlgorithm_VALUES_TO_NAMES, 0)) {
 }
 
 void GeographyType::__set_crs(const std::string& val) {
@@ -3498,7 +3520,7 @@ uint32_t GeographyType::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast114;
           xfer += iprot->readI32(ecast114);
-          this->algorithm = static_cast<EdgeInterpolationAlgorithm::type>(ecast114);
+          this->algorithm = SafeEnumCast<EdgeInterpolationAlgorithm>(_EdgeInterpolationAlgorithm_VALUES_TO_NAMES, ecast114);
           this->__isset.algorithm = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4067,12 +4089,12 @@ SchemaElement::~SchemaElement() noexcept {
 }
 
 SchemaElement::SchemaElement() noexcept
-   : type(static_cast<Type::type>(0)),
+   : type(SafeEnumCast<Type>(_Type_VALUES_TO_NAMES, 0)),
      type_length(0),
-     repetition_type(static_cast<FieldRepetitionType::type>(0)),
+     repetition_type(SafeEnumCast<FieldRepetitionType>(_FieldRepetitionType_VALUES_TO_NAMES, 0)),
      name(),
      num_children(0),
-     converted_type(static_cast<ConvertedType::type>(0)),
+     converted_type(SafeEnumCast<ConvertedType>(_ConvertedType_VALUES_TO_NAMES, 0)),
      scale(0),
      precision(0),
      field_id(0) {
@@ -4159,7 +4181,7 @@ uint32_t SchemaElement::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast123;
           xfer += iprot->readI32(ecast123);
-          this->type = static_cast<Type::type>(ecast123);
+          this->type = SafeEnumCast<Type>(_Type_VALUES_TO_NAMES, ecast123);
           this->__isset.type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4177,7 +4199,7 @@ uint32_t SchemaElement::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast124;
           xfer += iprot->readI32(ecast124);
-          this->repetition_type = static_cast<FieldRepetitionType::type>(ecast124);
+          this->repetition_type = SafeEnumCast<FieldRepetitionType>(_FieldRepetitionType_VALUES_TO_NAMES, ecast124);
           this->__isset.repetition_type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4203,7 +4225,7 @@ uint32_t SchemaElement::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast125;
           xfer += iprot->readI32(ecast125);
-          this->converted_type = static_cast<ConvertedType::type>(ecast125);
+          this->converted_type = SafeEnumCast<ConvertedType>(_ConvertedType_VALUES_TO_NAMES, ecast125);
           this->__isset.converted_type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4405,9 +4427,9 @@ DataPageHeader::~DataPageHeader() noexcept {
 
 DataPageHeader::DataPageHeader() noexcept
    : num_values(0),
-     encoding(static_cast<Encoding::type>(0)),
-     definition_level_encoding(static_cast<Encoding::type>(0)),
-     repetition_level_encoding(static_cast<Encoding::type>(0)) {
+     encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)),
+     definition_level_encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)),
+     repetition_level_encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)) {
 }
 
 void DataPageHeader::__set_num_values(const int32_t val) {
@@ -4474,7 +4496,7 @@ uint32_t DataPageHeader::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast130;
           xfer += iprot->readI32(ecast130);
-          this->encoding = static_cast<Encoding::type>(ecast130);
+          this->encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast130);
           isset_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4484,7 +4506,7 @@ uint32_t DataPageHeader::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast131;
           xfer += iprot->readI32(ecast131);
-          this->definition_level_encoding = static_cast<Encoding::type>(ecast131);
+          this->definition_level_encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast131);
           isset_definition_level_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4494,7 +4516,7 @@ uint32_t DataPageHeader::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast132;
           xfer += iprot->readI32(ecast132);
-          this->repetition_level_encoding = static_cast<Encoding::type>(ecast132);
+          this->repetition_level_encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast132);
           isset_repetition_level_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4697,7 +4719,7 @@ DictionaryPageHeader::~DictionaryPageHeader() noexcept {
 
 DictionaryPageHeader::DictionaryPageHeader() noexcept
    : num_values(0),
-     encoding(static_cast<Encoding::type>(0)),
+     encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)),
      is_sorted(0) {
 }
 
@@ -4755,7 +4777,7 @@ uint32_t DictionaryPageHeader::read(::apache::thrift::protocol::TProtocol* iprot
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast141;
           xfer += iprot->readI32(ecast141);
-          this->encoding = static_cast<Encoding::type>(ecast141);
+          this->encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast141);
           isset_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -4859,7 +4881,7 @@ DataPageHeaderV2::DataPageHeaderV2() noexcept
    : num_values(0),
      num_nulls(0),
      num_rows(0),
-     encoding(static_cast<Encoding::type>(0)),
+     encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)),
      definition_levels_byte_length(0),
      repetition_levels_byte_length(0),
      is_compressed(true) {
@@ -4960,7 +4982,7 @@ uint32_t DataPageHeaderV2::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast146;
           xfer += iprot->readI32(ecast146);
-          this->encoding = static_cast<Encoding::type>(ecast146);
+          this->encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast146);
           isset_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -5867,7 +5889,7 @@ PageHeader::~PageHeader() noexcept {
 }
 
 PageHeader::PageHeader() noexcept
-   : type(static_cast<PageType::type>(0)),
+   : type(SafeEnumCast<PageType>(_PageType_VALUES_TO_NAMES, 0)),
      uncompressed_page_size(0),
      compressed_page_size(0),
      crc(0) {
@@ -5944,7 +5966,7 @@ uint32_t PageHeader::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast179;
           xfer += iprot->readI32(ecast179);
-          this->type = static_cast<PageType::type>(ecast179);
+          this->type = SafeEnumCast<PageType>(_PageType_VALUES_TO_NAMES, ecast179);
           isset_type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -6435,8 +6457,8 @@ PageEncodingStats::~PageEncodingStats() noexcept {
 }
 
 PageEncodingStats::PageEncodingStats() noexcept
-   : page_type(static_cast<PageType::type>(0)),
-     encoding(static_cast<Encoding::type>(0)),
+   : page_type(SafeEnumCast<PageType>(_PageType_VALUES_TO_NAMES, 0)),
+     encoding(SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, 0)),
      count(0) {
 }
 
@@ -6486,7 +6508,7 @@ uint32_t PageEncodingStats::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast192;
           xfer += iprot->readI32(ecast192);
-          this->page_type = static_cast<PageType::type>(ecast192);
+          this->page_type = SafeEnumCast<PageType>(_PageType_VALUES_TO_NAMES, ecast192);
           isset_page_type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -6496,7 +6518,7 @@ uint32_t PageEncodingStats::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast193;
           xfer += iprot->readI32(ecast193);
-          this->encoding = static_cast<Encoding::type>(ecast193);
+          this->encoding = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast193);
           isset_encoding = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -6593,8 +6615,8 @@ ColumnMetaData::~ColumnMetaData() noexcept {
 }
 
 ColumnMetaData::ColumnMetaData() noexcept
-   : type(static_cast<Type::type>(0)),
-     codec(static_cast<CompressionCodec::type>(0)),
+   : type(SafeEnumCast<Type>(_Type_VALUES_TO_NAMES, 0)),
+     codec(SafeEnumCast<CompressionCodec>(_CompressionCodec_VALUES_TO_NAMES, 0)),
      num_values(0),
      total_uncompressed_size(0),
      total_compressed_size(0),
@@ -6721,7 +6743,7 @@ uint32_t ColumnMetaData::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast198;
           xfer += iprot->readI32(ecast198);
-          this->type = static_cast<Type::type>(ecast198);
+          this->type = SafeEnumCast<Type>(_Type_VALUES_TO_NAMES, ecast198);
           isset_type = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -6740,7 +6762,7 @@ uint32_t ColumnMetaData::read(::apache::thrift::protocol::TProtocol* iprot) {
             {
               int32_t ecast204;
               xfer += iprot->readI32(ecast204);
-              this->encodings[_i203] = static_cast<Encoding::type>(ecast204);
+              this->encodings[_i203] = SafeEnumCast<Encoding>(_Encoding_VALUES_TO_NAMES, ecast204);
             }
             xfer += iprot->readListEnd();
           }
@@ -6773,7 +6795,7 @@ uint32_t ColumnMetaData::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast210;
           xfer += iprot->readI32(ecast210);
-          this->codec = static_cast<CompressionCodec::type>(ecast210);
+          this->codec = SafeEnumCast<CompressionCodec>(_CompressionCodec_VALUES_TO_NAMES, ecast210);
           isset_codec = true;
         } else {
           xfer += iprot->skip(ftype);
@@ -8651,7 +8673,7 @@ ColumnIndex::~ColumnIndex() noexcept {
 }
 
 ColumnIndex::ColumnIndex() noexcept
-   : boundary_order(static_cast<BoundaryOrder::type>(0)) {
+   : boundary_order(SafeEnumCast<BoundaryOrder>(_BoundaryOrder_VALUES_TO_NAMES, 0)) {
 }
 
 void ColumnIndex::__set_null_pages(const duckdb::vector<bool> & val) {
@@ -8780,7 +8802,7 @@ uint32_t ColumnIndex::read(::apache::thrift::protocol::TProtocol* iprot) {
         if (ftype == ::apache::thrift::protocol::T_I32) {
           int32_t ecast310;
           xfer += iprot->readI32(ecast310);
-          this->boundary_order = static_cast<BoundaryOrder::type>(ecast310);
+          this->boundary_order = SafeEnumCast<BoundaryOrder>(_BoundaryOrder_VALUES_TO_NAMES, ecast310);
           isset_boundary_order = true;
         } else {
           xfer += iprot->skip(ftype);

package/src/duckdb/third_party/parquet/parquet_types.h

@@ -1,5 +1,5 @@
 /**
- * Autogenerated by Thrift Compiler (0.21.0)
+ * Autogenerated by Thrift Compiler (0.22.0)
  *
  * DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING
  *  @generated
@@ -178,7 +178,14 @@ struct ConvertedType {
      * the provided duration.  This duration of time is independent of any
      * particular timezone or date.
      */
-    INTERVAL = 21
+    INTERVAL = 21,
+    /**
+     * Non-standard NULL value
+     * 
+     * This was written by old writers - it is kept here for compatibility purposes.
+     * See https://github.com/duckdb/duckdb/pull/11774
+     */
+    PARQUET_NULL = 24
   };
 };
 

package/src/duckdb/ub_src_common_types_row.cpp

@@ -1,5 +1,3 @@
-#include "src/common/types/row/block_iterator.cpp"
-
 #include "src/common/types/row/partitioned_tuple_data.cpp"
 
 #include "src/common/types/row/row_data_collection.cpp"