duckdb 0.7.2-dev12.0 → 0.7.2-dev1244.0

package/src/duckdb/extension/parquet/column_reader.cpp

@@ -34,10 +34,69 @@ using duckdb_parquet::format::Encoding;
 using duckdb_parquet::format::PageType;
 using duckdb_parquet::format::Type;
 
-const uint32_t ParquetDecodeUtils::BITPACK_MASKS[] = {
-    0,       1,       3,        7,        15,       31,        63,        127,       255,        511,       1023,
-    2047,    4095,    8191,     16383,    32767,    65535,     131071,    262143,    524287,     1048575,   2097151,
-    4194303, 8388607, 16777215, 33554431, 67108863, 134217727, 268435455, 536870911, 1073741823, 2147483647};
+const uint64_t ParquetDecodeUtils::BITPACK_MASKS[] = {0,
+                                                      1,
+                                                      3,
+                                                      7,
+                                                      15,
+                                                      31,
+                                                      63,
+                                                      127,
+                                                      255,
+                                                      511,
+                                                      1023,
+                                                      2047,
+                                                      4095,
+                                                      8191,
+                                                      16383,
+                                                      32767,
+                                                      65535,
+                                                      131071,
+                                                      262143,
+                                                      524287,
+                                                      1048575,
+                                                      2097151,
+                                                      4194303,
+                                                      8388607,
+                                                      16777215,
+                                                      33554431,
+                                                      67108863,
+                                                      134217727,
+                                                      268435455,
+                                                      536870911,
+                                                      1073741823,
+                                                      2147483647,
+                                                      4294967295,
+                                                      8589934591,
+                                                      17179869183,
+                                                      34359738367,
+                                                      68719476735,
+                                                      137438953471,
+                                                      274877906943,
+                                                      549755813887,
+                                                      1099511627775,
+                                                      2199023255551,
+                                                      4398046511103,
+                                                      8796093022207,
+                                                      17592186044415,
+                                                      35184372088831,
+                                                      70368744177663,
+                                                      140737488355327,
+                                                      281474976710655,
+                                                      562949953421311,
+                                                      1125899906842623,
+                                                      2251799813685247,
+                                                      4503599627370495,
+                                                      9007199254740991,
+                                                      18014398509481983,
+                                                      36028797018963967,
+                                                      72057594037927935,
+                                                      144115188075855871,
+                                                      288230376151711743,
+                                                      576460752303423487,
+                                                      1152921504606846975,
+                                                      2305843009213693951,
+                                                      4611686018427387903};
 
 const uint8_t ParquetDecodeUtils::BITPACK_DLEN = 8;
 
@@ -449,13 +508,13 @@ idx_t ColumnReader::Read(uint64_t num_values, parquet_filter_t &filter, uint8_t
 			// TODO keep this in the state
 			auto read_buf = make_shared<ResizeableBuffer>();
 
-			switch (type.id()) {
-			case LogicalTypeId::INTEGER:
+			switch (type.InternalType()) {
+			case PhysicalType::INT32:
 				read_buf->resize(reader.allocator, sizeof(int32_t) * (read_now - null_count));
 				dbp_decoder->GetBatch<int32_t>(read_buf->ptr, read_now - null_count);
 
 				break;
-			case LogicalTypeId::BIGINT:
+			case PhysicalType::INT64:
 				read_buf->resize(reader.allocator, sizeof(int64_t) * (read_now - null_count));
 				dbp_decoder->GetBatch<int64_t>(read_buf->ptr, read_now - null_count);
 				break;
@@ -865,7 +924,7 @@ RowNumberColumnReader::RowNumberColumnReader(ParquetReader &reader, LogicalType
 
 unique_ptr<BaseStatistics> RowNumberColumnReader::Stats(idx_t row_group_idx_p,
                                                         const std::vector<ColumnChunk> &columns) {
-	auto stats = make_unique<NumericStatistics>(type, StatisticsType::LOCAL_STATS);
+	auto stats = NumericStats::CreateUnknown(type);
 	auto &row_groups = reader.GetFileMetadata()->row_groups;
 	D_ASSERT(row_group_idx_p < row_groups.size());
 	idx_t row_group_offset_min = 0;
@@ -873,11 +932,10 @@ unique_ptr<BaseStatistics> RowNumberColumnReader::Stats(idx_t row_group_idx_p,
 		row_group_offset_min += row_groups[i].num_rows;
 	}
 
-	stats->min = Value::BIGINT(row_group_offset_min);
-	stats->max = Value::BIGINT(row_group_offset_min + row_groups[row_group_idx_p].num_rows);
-
-	D_ASSERT(!stats->CanHaveNull() && stats->CanHaveNoNull());
-	return std::move(stats);
+	NumericStats::SetMin(stats, Value::BIGINT(row_group_offset_min));
+	NumericStats::SetMax(stats, Value::BIGINT(row_group_offset_min + row_groups[row_group_idx_p].num_rows));
+	stats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
+	return stats.ToUnique();
 }
 
 void RowNumberColumnReader::InitializeRead(idx_t row_group_idx_p, const std::vector<ColumnChunk> &columns,
@@ -1341,8 +1399,29 @@ unique_ptr<ColumnReader> ColumnReader::CreateReader(ParquetReader &reader, const
 		                                                                            file_idx_p, max_define, max_repeat);
 	case LogicalTypeId::TIME:
 	case LogicalTypeId::TIME_TZ:
-		return make_unique<CallbackColumnReader<int64_t, dtime_t, ParquetIntToTime>>(
-		    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+		if (schema_p.__isset.logicalType && schema_p.logicalType.__isset.TIME) {
+			if (schema_p.logicalType.TIME.unit.__isset.MILLIS) {
+				return make_unique<CallbackColumnReader<int32_t, dtime_t, ParquetIntToTimeMs>>(
+				    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+			} else if (schema_p.logicalType.TIME.unit.__isset.MICROS) {
+				return make_unique<CallbackColumnReader<int64_t, dtime_t, ParquetIntToTime>>(
+				    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+			} else if (schema_p.logicalType.TIME.unit.__isset.NANOS) {
+				return make_unique<CallbackColumnReader<int64_t, dtime_t, ParquetIntToTimeNs>>(
+				    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+			}
+		} else if (schema_p.__isset.converted_type) {
+			switch (schema_p.converted_type) {
+			case ConvertedType::TIME_MICROS:
+				return make_unique<CallbackColumnReader<int64_t, dtime_t, ParquetIntToTime>>(
+				    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+			case ConvertedType::TIME_MILLIS:
+				return make_unique<CallbackColumnReader<int32_t, dtime_t, ParquetIntToTimeMs>>(
+				    reader, type_p, schema_p, file_idx_p, max_define, max_repeat);
+			default:
+				break;
+			}
+		}
 	case LogicalTypeId::BLOB:
 	case LogicalTypeId::VARCHAR:
 		return make_unique<StringColumnReader>(reader, type_p, schema_p, file_idx_p, max_define, max_repeat);

package/src/duckdb/extension/parquet/column_writer.cpp

@@ -1994,7 +1994,6 @@ unique_ptr<ColumnWriter> ColumnWriter::CreateWriterRecursive(vector<duckdb_parqu
 			                                             max_define, can_have_nulls);
 		}
 	case LogicalTypeId::BLOB:
-	case LogicalTypeId::BIT:
 	case LogicalTypeId::VARCHAR:
 		return make_unique<StringColumnWriter>(writer, schema_idx, std::move(schema_path), max_repeat, max_define,
 		                                       can_have_nulls);

package/src/duckdb/extension/parquet/include/column_reader.hpp

@@ -18,8 +18,7 @@
 
 #include "duckdb.hpp"
 #ifndef DUCKDB_AMALGAMATION
-#include "duckdb/storage/statistics/string_statistics.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
+
 #include "duckdb/common/types/vector.hpp"
 #include "duckdb/common/types/string_type.hpp"
 #include "duckdb/common/types/chunk_collection.hpp"

package/src/duckdb/extension/parquet/include/decode_utils.hpp

@@ -11,7 +11,7 @@ public:
 		return (n >> 1) ^ -(n & 1);
 	}
 
-	static const uint32_t BITPACK_MASKS[];
+	static const uint64_t BITPACK_MASKS[];
 	static const uint8_t BITPACK_DLEN;
 
 	template <typename T>
@@ -23,7 +23,7 @@ public:
 			bitpack_pos += width;
 			while (bitpack_pos > BITPACK_DLEN) {
 				buffer.inc(1);
-				val |= (buffer.get<uint8_t>() << (BITPACK_DLEN - (bitpack_pos - width))) & mask;
+				val |= (T(buffer.get<uint8_t>()) << T(BITPACK_DLEN - (bitpack_pos - width))) & mask;
 				bitpack_pos -= BITPACK_DLEN;
 			}
 			dest[i] = val;
@@ -37,9 +37,10 @@ public:
 		uint8_t shift = 0;
 		while (true) {
 			auto byte = buf.read<uint8_t>();
-			result |= (byte & 127) << shift;
-			if ((byte & 128) == 0)
+			result |= T(byte & 127) << shift;
+			if ((byte & 128) == 0) {
 				break;
+			}
 			shift += 7;
 			if (shift > sizeof(T) * 8) {
 				throw std::runtime_error("Varint-decoding found too large number");

package/src/duckdb/extension/parquet/include/generated_column_reader.hpp

@@ -29,17 +29,7 @@ public:
 	           Vector &result) override;
 
 	unique_ptr<BaseStatistics> Stats(idx_t row_group_idx_p, const std::vector<ColumnChunk> &columns) override {
-		switch (type.id()) {
-		case LogicalTypeId::VARCHAR: {
-			auto string_stats = make_unique<StringStatistics>(type, StatisticsType::LOCAL_STATS);
-			string string = constant.ToString();
-			string_stats->Update(string);
-			string_stats->max_string_length = string.length();
-			return std::move(string_stats);
-		}
-		default:
-			return nullptr;
-		}
+		return BaseStatistics::FromConstant(constant).ToUnique();
 	};
 
 	void InitializeRead(idx_t row_group_idx_p, const std::vector<ColumnChunk> &columns,

package/src/duckdb/extension/parquet/include/parquet_timestamp.hpp

@@ -16,13 +16,14 @@ struct Int96 {
 	uint32_t value[3];
 };
 
-int64_t ImpalaTimestampToNanoseconds(const Int96 &impala_timestamp);
 timestamp_t ImpalaTimestampToTimestamp(const Int96 &raw_ts);
 Int96 TimestampToImpalaTimestamp(timestamp_t &ts);
 timestamp_t ParquetTimestampMicrosToTimestamp(const int64_t &raw_ts);
 timestamp_t ParquetTimestampMsToTimestamp(const int64_t &raw_ts);
 timestamp_t ParquetTimestampNsToTimestamp(const int64_t &raw_ts);
 date_t ParquetIntToDate(const int32_t &raw_date);
+dtime_t ParquetIntToTimeMs(const int32_t &raw_time);
 dtime_t ParquetIntToTime(const int64_t &raw_time);
+dtime_t ParquetIntToTimeNs(const int64_t &raw_time);
 
 } // namespace duckdb

package/src/duckdb/extension/parquet/parquet-extension.cpp

@@ -122,7 +122,8 @@ struct ParquetWriteGlobalState : public GlobalFunctionData {
 };
 
 struct ParquetWriteLocalState : public LocalFunctionData {
-	explicit ParquetWriteLocalState(ClientContext &context, const vector<LogicalType> &types) : buffer(context, types) {
+	explicit ParquetWriteLocalState(ClientContext &context, const vector<LogicalType> &types)
+	    : buffer(Allocator::Get(context), types) {
 	}
 
 	ColumnDataCollection buffer;
@@ -354,6 +355,7 @@ public:
 		return_types.assign(union_col_types.begin(), union_col_types.end());
 		result->SetInitialReader(result->union_readers[0]);
 		D_ASSERT(names.size() == return_types.size());
+		result->types = union_col_types;
 
 		return std::move(result);
 	}
@@ -368,7 +370,10 @@ public:
 		if (!config.options.enable_external_access) {
 			throw PermissionException("Scanning Parquet files is disabled through configuration");
 		}
-		auto file_name = input.inputs[0].GetValue<string>();
+		if (input.inputs[0].IsNull()) {
+			throw ParserException("Parquet reader cannot take NULL list as parameter");
+		}
+		auto file_name = StringValue::Get(input.inputs[0]);
 		ParquetOptions parquet_options(context);
 		for (auto &kv : input.named_parameters) {
 			auto loption = StringUtil::Lower(kv.first);
@@ -395,10 +400,16 @@ public:
 		if (!config.options.enable_external_access) {
 			throw PermissionException("Scanning Parquet files is disabled through configuration");
 		}
+		if (input.inputs[0].IsNull()) {
+			throw ParserException("Parquet reader cannot take NULL list as parameter");
+		}
 		FileSystem &fs = FileSystem::GetFileSystem(context);
 		vector<string> files;
 		for (auto &val : ListValue::GetChildren(input.inputs[0])) {
-			auto glob_files = ParquetGlob(fs, val.ToString(), context);
+			if (val.IsNull()) {
+				throw ParserException("Parquet reader cannot take NULL input as parameter");
+			}
+			auto glob_files = ParquetGlob(fs, StringValue::Get(val), context);
 			files.insert(files.end(), glob_files.begin(), glob_files.end());
 		}
 		if (files.empty()) {

package/src/duckdb/extension/parquet/parquet_reader.cpp

@@ -200,6 +200,11 @@ LogicalType ParquetReader::DeriveLogicalType(const SchemaElement &s_ele, bool bi
 				throw IOException("UTF8 converted type can only be set for Type::(FIXED_LEN_)BYTE_ARRAY");
 			}
 		case ConvertedType::TIME_MILLIS:
+			if (s_ele.type == Type::INT32) {
+				return LogicalType::TIME;
+			} else {
+				throw IOException("TIME_MILLIS converted type can only be set for value of Type::INT32");
+			}
 		case ConvertedType::TIME_MICROS:
 			if (s_ele.type == Type::INT64) {
 				return LogicalType::TIME;
@@ -320,7 +325,7 @@ unique_ptr<ColumnReader> ParquetReader::CreateReaderRecursive(const FileMetaData
 			                                     std::move(struct_reader));
 		}
 		if (child_types.size() > 1 || (!is_list && !is_map && !is_repeated)) {
-			result_type = LogicalType::STRUCT(std::move(child_types));
+			result_type = LogicalType::STRUCT(child_types);
 			result = make_unique<StructColumnReader>(*this, result_type, s_ele, this_idx, max_define, max_repeat,
 			                                         std::move(child_readers));
 		} else {

package/src/duckdb/extension/parquet/parquet_statistics.cpp

@@ -1,13 +1,11 @@
 #include "parquet_statistics.hpp"
 #include "parquet_decimal_utils.hpp"
 #include "parquet_timestamp.hpp"
-
 #include "duckdb.hpp"
 #ifndef DUCKDB_AMALGAMATION
 #include "duckdb/common/types/blob.hpp"
 #include "duckdb/common/types/value.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
-#include "duckdb/storage/statistics/string_statistics.hpp"
+#include "duckdb/common/types/time.hpp"
 #endif
 
 namespace duckdb {
@@ -18,27 +16,29 @@ using duckdb_parquet::format::Type;
 static unique_ptr<BaseStatistics> CreateNumericStats(const LogicalType &type,
                                                      const duckdb_parquet::format::SchemaElement &schema_ele,
                                                      const duckdb_parquet::format::Statistics &parquet_stats) {
-	auto stats = make_unique<NumericStatistics>(type, StatisticsType::LOCAL_STATS);
+	auto stats = NumericStats::CreateUnknown(type);
 
 	// for reasons unknown to science, Parquet defines *both* `min` and `min_value` as well as `max` and
 	// `max_value`. All are optional. such elegance.
+	Value min;
+	Value max;
 	if (parquet_stats.__isset.min) {
-		stats->min = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.min).DefaultCastAs(type);
+		min = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.min).DefaultCastAs(type);
 	} else if (parquet_stats.__isset.min_value) {
-		stats->min =
-		    ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.min_value).DefaultCastAs(type);
+		min = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.min_value).DefaultCastAs(type);
 	} else {
-		stats->min = Value(type);
+		min = Value(type);
 	}
 	if (parquet_stats.__isset.max) {
-		stats->max = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.max).DefaultCastAs(type);
+		max = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.max).DefaultCastAs(type);
 	} else if (parquet_stats.__isset.max_value) {
-		stats->max =
-		    ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.max_value).DefaultCastAs(type);
+		max = ParquetStatisticsUtils::ConvertValue(type, schema_ele, parquet_stats.max_value).DefaultCastAs(type);
 	} else {
-		stats->max = Value(type);
+		max = Value(type);
 	}
-	return std::move(stats);
+	NumericStats::SetMin(stats, min);
+	NumericStats::SetMax(stats, max);
+	return stats.ToUnique();
 }
 
 Value ParquetStatisticsUtils::ConvertValue(const LogicalType &type,
@@ -154,11 +154,31 @@ Value ParquetStatisticsUtils::ConvertValue(const LogicalType &type,
 		return Value::DATE(date_t(Load<int32_t>((data_ptr_t)stats.c_str())));
 	case LogicalTypeId::TIME:
 	case LogicalTypeId::TIME_TZ: {
-		if (stats.size() != sizeof(int64_t)) {
+		int64_t val;
+		if (stats.size() == sizeof(int32_t)) {
+			val = Load<int32_t>((data_ptr_t)stats.c_str());
+		} else if (stats.size() == sizeof(int64_t)) {
+			val = Load<int64_t>((data_ptr_t)stats.c_str());
+		} else {
 			throw InternalException("Incorrect stats size for type TIME");
 		}
-		auto time = dtime_t(Load<int64_t>((data_ptr_t)stats.c_str()));
-		return Value::TIME(time);
+		if (schema_ele.__isset.logicalType && schema_ele.logicalType.__isset.TIME) {
+			// logical type
+			if (schema_ele.logicalType.TIME.unit.__isset.MILLIS) {
+				return Value::TIME(Time::FromTimeMs(val));
+			} else if (schema_ele.logicalType.TIME.unit.__isset.NANOS) {
+				return Value::TIME(Time::FromTimeNs(val));
+			} else if (schema_ele.logicalType.TIME.unit.__isset.MICROS) {
+				return Value::TIME(dtime_t(val));
+			} else {
+				throw InternalException("Time logicalType is set but unit is not defined");
+			}
+		}
+		if (schema_ele.converted_type == duckdb_parquet::format::ConvertedType::TIME_MILLIS) {
+			return Value::TIME(Time::FromTimeMs(val));
+		} else {
+			return Value::TIME(dtime_t(val));
+		}
 	}
 	case LogicalTypeId::TIMESTAMP:
 	case LogicalTypeId::TIMESTAMP_TZ: {
@@ -228,24 +248,24 @@ unique_ptr<BaseStatistics> ParquetStatisticsUtils::TransformColumnStatistics(con
 		row_group_stats = CreateNumericStats(type, s_ele, parquet_stats);
 		break;
 	case LogicalTypeId::VARCHAR: {
-		auto string_stats = make_unique<StringStatistics>(type, StatisticsType::LOCAL_STATS);
+		auto string_stats = StringStats::CreateEmpty(type);
 		if (parquet_stats.__isset.min) {
-			string_stats->Update(parquet_stats.min);
+			StringStats::Update(string_stats, parquet_stats.min);
 		} else if (parquet_stats.__isset.min_value) {
-			string_stats->Update(parquet_stats.min_value);
+			StringStats::Update(string_stats, parquet_stats.min_value);
 		} else {
 			return nullptr;
 		}
 		if (parquet_stats.__isset.max) {
-			string_stats->Update(parquet_stats.max);
+			StringStats::Update(string_stats, parquet_stats.max);
 		} else if (parquet_stats.__isset.max_value) {
-			string_stats->Update(parquet_stats.max_value);
+			StringStats::Update(string_stats, parquet_stats.max_value);
 		} else {
 			return nullptr;
 		}
-		string_stats->has_unicode = true; // we dont know better
-		string_stats->max_string_length = NumericLimits<uint32_t>::Maximum();
-		row_group_stats = std::move(string_stats);
+		StringStats::SetContainsUnicode(string_stats);
+		StringStats::ResetMaxStringLength(string_stats);
+		row_group_stats = string_stats.ToUnique();
 		break;
 	}
 	default:
@@ -254,21 +274,14 @@ unique_ptr<BaseStatistics> ParquetStatisticsUtils::TransformColumnStatistics(con
 	} // end of type switch
 
 	// null count is generic
-	if (row_group_stats) {
-		if (column_chunk.meta_data.type == duckdb_parquet::format::Type::FLOAT ||
-		    column_chunk.meta_data.type == duckdb_parquet::format::Type::DOUBLE) {
-			// floats/doubles can have infinity, which can become NULL
-			row_group_stats->validity_stats = make_unique<ValidityStatistics>(true);
-		} else if (parquet_stats.__isset.null_count) {
-			row_group_stats->validity_stats = make_unique<ValidityStatistics>(parquet_stats.null_count != 0);
-		} else {
-			row_group_stats->validity_stats = make_unique<ValidityStatistics>(true);
-		}
-	} else {
+	if (!row_group_stats) {
 		// if stats are missing from any row group we know squat
 		return nullptr;
 	}
-
+	row_group_stats->Set(StatsInfo::CAN_HAVE_NULL_AND_VALID_VALUES);
+	if (parquet_stats.__isset.null_count && parquet_stats.null_count == 0) {
+		row_group_stats->Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
+	}
 	return row_group_stats;
 }
 

package/src/duckdb/extension/parquet/parquet_timestamp.cpp

@@ -12,24 +12,26 @@ namespace duckdb {
 // surely they are joking
 static constexpr int64_t JULIAN_TO_UNIX_EPOCH_DAYS = 2440588LL;
 static constexpr int64_t MILLISECONDS_PER_DAY = 86400000LL;
-static constexpr int64_t NANOSECONDS_PER_DAY = MILLISECONDS_PER_DAY * 1000LL * 1000LL;
+static constexpr int64_t MICROSECONDS_PER_DAY = MILLISECONDS_PER_DAY * 1000LL;
+static constexpr int64_t NANOSECONDS_PER_MICRO = 1000LL;
 
-int64_t ImpalaTimestampToNanoseconds(const Int96 &impala_timestamp) {
+static int64_t ImpalaTimestampToMicroseconds(const Int96 &impala_timestamp) {
 	int64_t days_since_epoch = impala_timestamp.value[2] - JULIAN_TO_UNIX_EPOCH_DAYS;
 	auto nanoseconds = Load<int64_t>((data_ptr_t)impala_timestamp.value);
-	return days_since_epoch * NANOSECONDS_PER_DAY + nanoseconds;
+	auto microseconds = nanoseconds / NANOSECONDS_PER_MICRO;
+	return days_since_epoch * MICROSECONDS_PER_DAY + microseconds;
 }
 
 timestamp_t ImpalaTimestampToTimestamp(const Int96 &raw_ts) {
-	auto impala_ns = ImpalaTimestampToNanoseconds(raw_ts);
-	return Timestamp::FromEpochNanoSeconds(impala_ns);
+	auto impala_us = ImpalaTimestampToMicroseconds(raw_ts);
+	return Timestamp::FromEpochMicroSeconds(impala_us);
 }
 
 Int96 TimestampToImpalaTimestamp(timestamp_t &ts) {
 	int32_t hour, min, sec, msec;
 	Time::Convert(Timestamp::GetTime(ts), hour, min, sec, msec);
 	uint64_t ms_since_midnight = hour * 60 * 60 * 1000 + min * 60 * 1000 + sec * 1000 + msec;
-	auto days_since_epoch = Date::Epoch(Timestamp::GetDate(ts)) / (24 * 60 * 60);
+	auto days_since_epoch = Date::Epoch(Timestamp::GetDate(ts)) / int64_t(24 * 60 * 60);
 	// first two uint32 in Int96 are nanoseconds since midnights
 	// last uint32 is number of days since year 4713 BC ("Julian date")
 	Int96 impala_ts;
@@ -52,8 +54,16 @@ date_t ParquetIntToDate(const int32_t &raw_date) {
 	return date_t(raw_date);
 }
 
+dtime_t ParquetIntToTimeMs(const int32_t &raw_time) {
+	return Time::FromTimeMs(raw_time);
+}
+
 dtime_t ParquetIntToTime(const int64_t &raw_time) {
 	return dtime_t(raw_time);
 }
 
+dtime_t ParquetIntToTimeNs(const int64_t &raw_time) {
+	return Time::FromTimeNs(raw_time);
+}
+
 } // namespace duckdb

package/src/duckdb/src/catalog/catalog.cpp

@@ -63,20 +63,28 @@ Catalog &Catalog::GetSystemCatalog(ClientContext &context) {
 	return Catalog::GetSystemCatalog(*context.db);
 }
 
-Catalog &Catalog::GetCatalog(ClientContext &context, const string &catalog_name) {
+optional_ptr<Catalog> Catalog::GetCatalogEntry(ClientContext &context, const string &catalog_name) {
 	auto &db_manager = DatabaseManager::Get(context);
 	if (catalog_name == TEMP_CATALOG) {
-		return ClientData::Get(context).temporary_objects->GetCatalog();
+		return &ClientData::Get(context).temporary_objects->GetCatalog();
 	}
 	if (catalog_name == SYSTEM_CATALOG) {
-		return GetSystemCatalog(context);
+		return &GetSystemCatalog(context);
 	}
 	auto entry = db_manager.GetDatabase(
 	    context, IsInvalidCatalog(catalog_name) ? DatabaseManager::GetDefaultDatabase(context) : catalog_name);
 	if (!entry) {
+		return nullptr;
+	}
+	return &entry->GetCatalog();
+}
+
+Catalog &Catalog::GetCatalog(ClientContext &context, const string &catalog_name) {
+	auto catalog = Catalog::GetCatalogEntry(context, catalog_name);
+	if (!catalog) {
 		throw BinderException("Catalog \"%s\" does not exist!", catalog_name);
 	}
-	return entry->GetCatalog();
+	return *catalog;
 }
 
 //===--------------------------------------------------------------------===//
@@ -576,7 +584,15 @@ CatalogEntry *Catalog::GetEntry(ClientContext &context, CatalogType type, const
 	vector<CatalogLookup> lookups;
 	lookups.reserve(entries.size());
 	for (auto &entry : entries) {
-		lookups.emplace_back(Catalog::GetCatalog(context, entry.catalog), entry.schema);
+		if (if_exists_p) {
+			auto catalog_entry = Catalog::GetCatalogEntry(context, entry.catalog);
+			if (!catalog_entry) {
+				return nullptr;
+			}
+			lookups.emplace_back(*catalog_entry, entry.schema);
+		} else {
+			lookups.emplace_back(Catalog::GetCatalog(context, entry.catalog), entry.schema);
+		}
 	}
 	auto result = LookupEntry(context, lookups, type, name, if_exists_p, error_context);
 	if (!result.Found()) {
@@ -625,6 +641,19 @@ vector<SchemaCatalogEntry *> Catalog::GetSchemas(ClientContext &context) {
 	return schemas;
 }
 
+bool Catalog::TypeExists(ClientContext &context, const string &catalog_name, const string &schema, const string &name) {
+	CatalogEntry *entry;
+	entry = GetEntry(context, CatalogType::TYPE_ENTRY, catalog_name, schema, name, true);
+	if (!entry) {
+		// look in the system catalog
+		entry = GetEntry(context, CatalogType::TYPE_ENTRY, SYSTEM_CATALOG, schema, name, true);
+		if (!entry) {
+			return false;
+		}
+	}
+	return true;
+}
+
 vector<SchemaCatalogEntry *> Catalog::GetSchemas(ClientContext &context, const string &catalog_name) {
 	vector<Catalog *> catalogs;
 	if (IsInvalidCatalog(catalog_name)) {

package/src/duckdb/src/catalog/catalog_entry/duck_schema_entry.cpp

@@ -145,6 +145,10 @@ CatalogEntry *DuckSchemaEntry::CreateFunction(CatalogTransaction transaction, Cr
 		function = make_unique_base<StandardEntry, ScalarFunctionCatalogEntry>(catalog, this,
 		                                                                       (CreateScalarFunctionInfo *)info);
 		break;
+	case CatalogType::TABLE_FUNCTION_ENTRY:
+		function =
+		    make_unique_base<StandardEntry, TableFunctionCatalogEntry>(catalog, this, (CreateTableFunctionInfo *)info);
+		break;
 	case CatalogType::MACRO_ENTRY:
 		// create a macro function
 		function = make_unique_base<StandardEntry, ScalarMacroCatalogEntry>(catalog, this, (CreateMacroInfo *)info);

package/src/duckdb/src/catalog/catalog_entry/duck_table_entry.cpp

@@ -682,27 +682,8 @@ void DuckTableEntry::SetAsRoot() {
 	storage->info->table = name;
 }
 
-void DuckTableEntry::CommitAlter(AlterInfo &info) {
-	D_ASSERT(info.type == AlterType::ALTER_TABLE);
-	auto &alter_table = (AlterTableInfo &)info;
-	string column_name;
-	switch (alter_table.alter_table_type) {
-	case AlterTableType::REMOVE_COLUMN: {
-		auto &remove_info = (RemoveColumnInfo &)alter_table;
-		column_name = remove_info.removed_column;
-		break;
-	}
-	case AlterTableType::ALTER_COLUMN_TYPE: {
-		auto &change_info = (ChangeColumnTypeInfo &)alter_table;
-		column_name = change_info.column_name;
-		break;
-	}
-	default:
-		break;
-	}
-	if (column_name.empty()) {
-		return;
-	}
+void DuckTableEntry::CommitAlter(string &column_name) {
+	D_ASSERT(!column_name.empty());
 	idx_t removed_index = DConstants::INVALID_INDEX;
 	for (auto &col : columns.Logical()) {
 		if (col.Name() == column_name) {

package/src/duckdb/src/catalog/catalog_entry/scalar_function_catalog_entry.cpp

@@ -1,5 +1,5 @@
 #include "duckdb/catalog/catalog_entry/scalar_function_catalog_entry.hpp"
-#include "duckdb/parser/parsed_data/alter_function_info.hpp"
+#include "duckdb/parser/parsed_data/alter_scalar_function_info.hpp"
 
 namespace duckdb {
 
@@ -9,14 +9,15 @@ ScalarFunctionCatalogEntry::ScalarFunctionCatalogEntry(Catalog *catalog, SchemaC
 }
 
 unique_ptr<CatalogEntry> ScalarFunctionCatalogEntry::AlterEntry(ClientContext &context, AlterInfo *info) {
-	if (info->type != AlterType::ALTER_FUNCTION) {
+	if (info->type != AlterType::ALTER_SCALAR_FUNCTION) {
 		throw InternalException("Attempting to alter ScalarFunctionCatalogEntry with unsupported alter type");
 	}
-	auto &function_info = (AlterFunctionInfo &)*info;
-	if (function_info.alter_function_type != AlterFunctionType::ADD_FUNCTION_OVERLOADS) {
-		throw InternalException("Attempting to alter ScalarFunctionCatalogEntry with unsupported alter function type");
+	auto &function_info = (AlterScalarFunctionInfo &)*info;
+	if (function_info.alter_scalar_function_type != AlterScalarFunctionType::ADD_FUNCTION_OVERLOADS) {
+		throw InternalException(
+		    "Attempting to alter ScalarFunctionCatalogEntry with unsupported alter scalar function type");
 	}
-	auto &add_overloads = (AddFunctionOverloadInfo &)function_info;
+	auto &add_overloads = (AddScalarFunctionOverloadInfo &)function_info;
 
 	ScalarFunctionSet new_set = functions;
 	if (!new_set.MergeFunctionSet(add_overloads.new_overloads)) {