Update vendored DuckDB sources to 7f5cc3d

duckdb · Feb 11, 2025 · fa69ecb · fa69ecb
1 parent 7f5cc3d
commit fa69ecb
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 9 deletions.
diff --git a/src/duckdb/extension/parquet/column_reader.cpp b/src/duckdb/extension/parquet/column_reader.cpp
@@ -286,7 +286,7 @@ void ColumnReader::PrepareRead(optional_ptr<const TableFilter> filter) {
 		if (dictionary_size < 0) {
 			throw std::runtime_error("Invalid dictionary page header (num_values < 0)");
 		}
-		dictionary_decoder.InitializeDictionary(dictionary_size, filter);
+		dictionary_decoder.InitializeDictionary(dictionary_size, filter, HasDefines());
 		break;
 	}
 	default:

diff --git a/src/duckdb/extension/parquet/decoder/dictionary_decoder.cpp b/src/duckdb/extension/parquet/decoder/dictionary_decoder.cpp
@@ -12,11 +12,13 @@ DictionaryDecoder::DictionaryDecoder(ColumnReader &reader)
       dictionary_selection_vector(STANDARD_VECTOR_SIZE), dictionary_size(0) {
 }
 
-void DictionaryDecoder::InitializeDictionary(idx_t new_dictionary_size, optional_ptr<const TableFilter> filter) {
+void DictionaryDecoder::InitializeDictionary(idx_t new_dictionary_size, optional_ptr<const TableFilter> filter,
+                                             bool has_defines) {
 	auto old_dict_size = dictionary_size;
 	dictionary_size = new_dictionary_size;
 	filter_result.reset();
 	filter_count = 0;
+	can_have_nulls = has_defines;
 	// we use the first value in the dictionary to keep a NULL
 	if (!dictionary) {
 		dictionary = make_uniq<Vector>(reader.type, dictionary_size + 1);
@@ -27,7 +29,9 @@ void DictionaryDecoder::InitializeDictionary(idx_t new_dictionary_size, optional
 	// we use the last entry as a NULL, dictionary vectors don't have a separate validity mask
 	auto &dict_validity = FlatVector::Validity(*dictionary);
 	dict_validity.Reset(dictionary_size + 1);
-	dict_validity.SetInvalid(dictionary_size);
+	if (can_have_nulls) {
+		dict_validity.SetInvalid(dictionary_size);
+	}
 	reader.Plain(reader.block, nullptr, dictionary_size, 0, *dictionary);
 
 	if (filter && CanFilter(*filter)) {
@@ -73,6 +77,7 @@ void DictionaryDecoder::ConvertDictToSelVec(uint32_t *offsets, const SelectionVe
 idx_t DictionaryDecoder::GetValidValues(uint8_t *defines, idx_t read_count, idx_t result_offset) {
 	idx_t valid_count = read_count;
 	if (defines) {
+		D_ASSERT(can_have_nulls);
 		valid_count = 0;
 		for (idx_t i = 0; i < read_count; i++) {
 			valid_sel.set_index(valid_count, i);
@@ -104,10 +109,10 @@ idx_t DictionaryDecoder::Read(uint8_t *defines, idx_t read_count, Vector &result
 		ConvertDictToSelVec(reinterpret_cast<uint32_t *>(offset_buffer.ptr), valid_sel, valid_count);
 	}
 #ifdef DEBUG
-	dictionary_selection_vector.Verify(read_count, dictionary_size + 1);
+	dictionary_selection_vector.Verify(read_count, dictionary_size + can_have_nulls);
 #endif
 	if (result_offset == 0) {
-		result.Dictionary(*dictionary, dictionary_size + 1, dictionary_selection_vector, read_count);
+		result.Dictionary(*dictionary, dictionary_size + can_have_nulls, dictionary_selection_vector, read_count);
 		DictionaryVector::SetDictionaryId(result, dictionary_id);
 		D_ASSERT(result.GetVectorType() == VectorType::DICTIONARY_VECTOR);
 	} else {

diff --git a/src/duckdb/extension/parquet/include/decoder/dictionary_decoder.hpp b/src/duckdb/extension/parquet/include/decoder/dictionary_decoder.hpp
@@ -20,7 +20,7 @@ class DictionaryDecoder {
 	explicit DictionaryDecoder(ColumnReader &reader);
 
 public:
-	void InitializeDictionary(idx_t dictionary_size, optional_ptr<const TableFilter> filter);
+	void InitializeDictionary(idx_t dictionary_size, optional_ptr<const TableFilter> filter, bool has_defines);
 	void InitializePage();
 	idx_t Read(uint8_t *defines, idx_t read_count, Vector &result, idx_t result_offset);
 	void Skip(uint8_t *defines, idx_t skip_count);
@@ -48,6 +48,7 @@ class DictionaryDecoder {
 	unique_ptr<Vector> dictionary;
 	unsafe_unique_array<bool> filter_result;
 	idx_t filter_count;
+	bool can_have_nulls;
 	string dictionary_id;
 };
 

diff --git a/src/duckdb/src/function/table/version/pragma_version.cpp b/src/duckdb/src/function/table/version/pragma_version.cpp
@@ -1,5 +1,5 @@
 #ifndef DUCKDB_PATCH_VERSION
-#define DUCKDB_PATCH_VERSION "1-dev152"
+#define DUCKDB_PATCH_VERSION "1-dev163"
 #endif
 #ifndef DUCKDB_MINOR_VERSION
 #define DUCKDB_MINOR_VERSION 2
@@ -8,10 +8,10 @@
 #define DUCKDB_MAJOR_VERSION 1
 #endif
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "v1.2.1-dev152"
+#define DUCKDB_VERSION "v1.2.1-dev163"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "83d07cb683"
+#define DUCKDB_SOURCE_ID "f317a112e0"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"