update

Author: kaka11chen

be/src/vec/exec/format/parquet/byte_array_dict_decoder.cpp

@@ -140,19 +140,11 @@ Status ByteArrayDictDecoder::_decode_values(MutableColumnPtr& doris_column, Data
             break;
         }
         case ColumnSelectVector::FILTERED_CONTENT: {
-            // In lazy materialization, keep filtered rows (fill with dict data to maintain row count)
-            std::vector<StringRef> string_values;
-            string_values.reserve(run_length);
-            for (size_t i = 0; i < run_length; ++i) {
-                string_values.emplace_back(_dict_items[_indexes[dict_index++]]);
-            }
-            doris_column->insert_many_strings_overflow(string_values.data(), run_length,
-                                                       _max_value_length);
+            dict_index += run_length;
             break;
         }
         case ColumnSelectVector::FILTERED_NULL: {
-            // In lazy materialization, keep filtered null rows (fill with defaults to maintain row count)
-            doris_column->insert_many_defaults(run_length);
+            // do nothing
             break;
         }
         }

be/src/vec/exec/format/parquet/parquet_common.h

@@ -154,19 +154,22 @@ class ColumnSelectVector {
 
             _num_filtered = num_values - num_read;
 
-            // Fill null_map for ALL rows (including filtered ones) to maintain row count consistency
-            if (null_map != nullptr) {
+            if (null_map != nullptr && num_read > 0) {
                 NullMap& map_data_column = *null_map;
                 auto null_map_index = map_data_column.size();
-                map_data_column.resize(null_map_index +
-                                       num_values); // Resize to num_values, not num_read
-
-                // Fill null map for all rows based on _data_map
-                for (i = 0; i < num_values; ++i) {
-                    if (_data_map[i] == CONTENT || _data_map[i] == FILTERED_CONTENT) {
-                        map_data_column[null_map_index++] = (UInt8) false;
-                    } else { // NULL_DATA or FILTERED_NULL
-                        map_data_column[null_map_index++] = (UInt8) true;
+                map_data_column.resize(null_map_index + num_read);
+
+                if (_num_nulls == 0) {
+                    memset(map_data_column.data() + null_map_index, 0, num_read);
+                } else if (_num_nulls == num_values) {
+                    memset(map_data_column.data() + null_map_index, 1, num_read);
+                } else {
+                    for (i = 0; i < num_values; ++i) {
+                        if (_data_map[i] == CONTENT) {
+                            map_data_column[null_map_index++] = (UInt8) false;
+                        } else if (_data_map[i] == NULL_DATA) {
+                            map_data_column[null_map_index++] = (UInt8) true;
+                        }
                     }
                 }
             }