[Refactor] Remove wrapper_field (StarRocks#13191)

wrapper_field is only used by ColumnMapping, but it is not necessary to use wrapper_field actually.
Before default_value will be stored in wrapper_field, and convert wrapper_field to string, then to datum, then push the datum into the Column.
In the PR, I convert the input default value string into Datum directly.

be/src/storage/CMakeLists.txt

@@ -67,7 +67,6 @@ add_library(Storage STATIC
     types.cpp
     update_manager.cpp
     utils.cpp
-    wrapper_field.cpp
     compaction_utils.cpp
     rowset/array_column_iterator.cpp
     rowset/binary_plain_page.cpp

be/src/storage/column_mapping.h

@@ -21,25 +21,36 @@
 
 #pragma once
 
-namespace starrocks {
+#include <memory>
 
-class WrapperField;
+#include "column/datum.h"
+#include "types/bitmap_value.h"
+#include "types/hll.h"
+#include "util/json.h"
+#include "util/percentile_value.h"
 
-struct ColumnMapping {
-    ColumnMapping() {}
-    virtual ~ColumnMapping() = default;
+namespace starrocks {
 
+struct ColumnMapping {
     // <0: use default value
     // >=0: use origin column
     int32_t ref_column{-1};
 
     // The base reader selects part of the column
     // so it's different to ref_column
     int32_t ref_base_reader_column_index = -1;
-    // normally for default value. stores values for filters
-    WrapperField* default_value{nullptr};
     // materialize view transform function used in schema change
     std::string materialized_function;
+
+    // the following data is used by default_value_datum, because default_value_datum only
+    // have the reference. We need to keep the content has the same life cycle as the
+    // default_value_datum;
+    std::unique_ptr<HyperLogLog> default_hll;
+    std::unique_ptr<BitmapValue> default_bitmap;
+    std::unique_ptr<PercentileValue> default_percentile;
+    std::unique_ptr<JsonValue> default_json;
+
+    vectorized::Datum default_value_datum;
 };
 
 typedef std::vector<ColumnMapping> SchemaMapping;

be/src/storage/column_vector.h

@@ -32,6 +32,8 @@
 
 namespace starrocks {
 
+class Field;
+
 // struct that contains column data(null bitmap), data array in sub class.
 class ColumnVectorBatch {
 public:

be/src/storage/olap_common.h

@@ -405,10 +405,6 @@ struct HashOfVersion {
     }
 };
 
-class Field;
-class WrapperField;
-using KeyRange = std::pair<WrapperField*, WrapperField*>;
-
 // ReaderStatistics used to collect statistics when scan data from storage
 struct OlapReaderStatistics {
     int64_t create_segment_iter_ns = 0;

be/src/storage/rowset/column_reader.cpp

@@ -43,7 +43,6 @@
 #include "storage/rowset/zone_map_index.h"
 #include "storage/types.h"
 #include "storage/vectorized_column_predicate.h"
-#include "storage/wrapper_field.h"
 #include "util/compression/block_compression.h"
 #include "util/rle_encoding.h"
 

be/src/storage/rowset/column_reader.h

@@ -51,7 +51,6 @@ class ColumnBlock;
 class ColumnBlockView;
 class ColumnVectorBatch;
 class MemTracker;
-class WrapperField;
 
 namespace vectorized {
 class ColumnPredicate;

be/src/storage/schema_change_utils.cpp

@@ -4,7 +4,6 @@
 
 #include "column/datum_convert.h"
 #include "storage/chunk_helper.h"
-#include "storage/wrapper_field.h"
 #include "types/bitmap_value.h"
 #include "types/hll.h"
 #include "util/percentile_value.h"
@@ -16,9 +15,6 @@ ChunkChanger::ChunkChanger(const TabletSchema& tablet_schema) {
 }
 
 ChunkChanger::~ChunkChanger() {
-    for (auto it = _schema_mapping.begin(); it != _schema_mapping.end(); ++it) {
-        SAFE_DELETE(it->default_value);
-    }
     _schema_mapping.clear();
 }
 
@@ -321,48 +317,8 @@ bool ChunkChanger::change_chunk(ChunkPtr& base_chunk, ChunkPtr& new_chunk, const
             }
         } else {
             ColumnPtr& new_col = new_chunk->get_column_by_index(i);
-            Datum dst_datum;
-            if (_schema_mapping[i].default_value->is_null()) {
-                dst_datum.set_null();
-                COLUMN_APPEND_DATUM();
-            } else {
-                Field new_field =
-                        ChunkHelper::convert_field_to_format_v2(i, new_tablet_meta->tablet_schema().column(i));
-                const FieldType field_type = new_field.type()->type();
-                std::string tmp = _schema_mapping[i].default_value->to_string();
-                if (field_type == OLAP_FIELD_TYPE_HLL || field_type == OLAP_FIELD_TYPE_OBJECT ||
-                    field_type == OLAP_FIELD_TYPE_PERCENTILE) {
-                    switch (field_type) {
-                    case OLAP_FIELD_TYPE_HLL: {
-                        HyperLogLog hll(tmp);
-                        dst_datum.set_hyperloglog(&hll);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    case OLAP_FIELD_TYPE_OBJECT: {
-                        BitmapValue bitmap(tmp);
-                        dst_datum.set_bitmap(&bitmap);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    case OLAP_FIELD_TYPE_PERCENTILE: {
-                        PercentileValue percentile(tmp);
-                        dst_datum.set_percentile(&percentile);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    default:
-                        LOG(WARNING) << "the column type is wrong. column_type: " << field_type_to_string(field_type);
-                        return false;
-                    }
-                } else {
-                    Status st = datum_from_string(new_field.type().get(), &dst_datum, tmp, nullptr);
-                    if (!st.ok()) {
-                        LOG(WARNING) << "create datum from string failed: status=" << st;
-                        return false;
-                    }
-                    COLUMN_APPEND_DATUM();
-                }
+            for (size_t row_index = 0; row_index < base_chunk->num_rows(); ++row_index) {
+                new_col->append_datum(_schema_mapping[i].default_value_datum);
             }
         }
     }
@@ -463,47 +419,8 @@ bool ChunkChanger::change_chunk_v2(ChunkPtr& base_chunk, ChunkPtr& new_chunk, co
             }
         } else {
             ColumnPtr& new_col = new_chunk->get_column_by_index(i);
-            Datum dst_datum;
-            if (_schema_mapping[i].default_value->is_null()) {
-                dst_datum.set_null();
-                COLUMN_APPEND_DATUM();
-            } else {
-                TypeInfoPtr new_type_info = new_schema.field(i)->type();
-                const FieldType field_type = new_type_info->type();
-                std::string tmp = _schema_mapping[i].default_value->to_string();
-                if (field_type == OLAP_FIELD_TYPE_HLL || field_type == OLAP_FIELD_TYPE_OBJECT ||
-                    field_type == OLAP_FIELD_TYPE_PERCENTILE) {
-                    switch (field_type) {
-                    case OLAP_FIELD_TYPE_HLL: {
-                        HyperLogLog hll(tmp);
-                        dst_datum.set_hyperloglog(&hll);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    case OLAP_FIELD_TYPE_OBJECT: {
-                        BitmapValue bitmap(tmp);
-                        dst_datum.set_bitmap(&bitmap);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    case OLAP_FIELD_TYPE_PERCENTILE: {
-                        PercentileValue percentile(tmp);
-                        dst_datum.set_percentile(&percentile);
-                        COLUMN_APPEND_DATUM();
-                        break;
-                    }
-                    default:
-                        LOG(WARNING) << "the column type is wrong. column_type: " << field_type_to_string(field_type);
-                        return false;
-                    }
-                } else {
-                    Status st = datum_from_string(new_type_info.get(), &dst_datum, tmp, nullptr);
-                    if (!st.ok()) {
-                        LOG(WARNING) << "create datum from string failed: status=" << st;
-                        return false;
-                    }
-                    COLUMN_APPEND_DATUM();
-                }
+            for (size_t row_index = 0; row_index < base_chunk->num_rows(); ++row_index) {
+                new_col->append_datum(_schema_mapping[i].default_value_datum);
             }
         }
     }
@@ -703,16 +620,36 @@ Status SchemaChangeUtils::parse_request(const TabletSchema& base_schema, const T
 
 Status SchemaChangeUtils::init_column_mapping(ColumnMapping* column_mapping, const TabletColumn& column_schema,
                                               const std::string& value) {
-    column_mapping->default_value = WrapperField::create(column_schema);
-
-    if (column_mapping->default_value == nullptr) {
-        return Status::InternalError("malloc error");
-    }
-
     if (column_schema.is_nullable() && value.length() == 0) {
-        column_mapping->default_value->set_null();
+        column_mapping->default_value_datum.set_null();
     } else {
-        column_mapping->default_value->from_string(value);
+        auto field_type = column_schema.type();
+        auto type_info = get_type_info(column_schema);
+
+        switch (field_type) {
+        case OLAP_FIELD_TYPE_HLL: {
+            column_mapping->default_hll = std::make_unique<HyperLogLog>(value);
+            column_mapping->default_value_datum.set_hyperloglog(column_mapping->default_hll.get());
+            break;
+        }
+        case OLAP_FIELD_TYPE_OBJECT: {
+            column_mapping->default_bitmap = std::make_unique<BitmapValue>(value);
+            column_mapping->default_value_datum.set_bitmap(column_mapping->default_bitmap.get());
+            break;
+        }
+        case OLAP_FIELD_TYPE_PERCENTILE: {
+            column_mapping->default_percentile = std::make_unique<PercentileValue>(value);
+            column_mapping->default_value_datum.set_percentile(column_mapping->default_percentile.get());
+            break;
+        }
+        case OLAP_FIELD_TYPE_JSON: {
+            column_mapping->default_json = std::make_unique<JsonValue>(value);
+            column_mapping->default_value_datum.set_json(column_mapping->default_json.get());
+            break;
+        }
+        default:
+            return datum_from_string(type_info.get(), &column_mapping->default_value_datum, value, nullptr);
+        }
     }
 
     return Status::OK();

be/src/storage/schema_change_utils.h

@@ -24,7 +24,7 @@ class ChunkChanger {
 
     ColumnMapping* get_mutable_column_mapping(size_t column_index);
 
-    SchemaMapping get_schema_mapping() const { return _schema_mapping; }
+    const SchemaMapping& get_schema_mapping() const { return _schema_mapping; }
 
     std::vector<ColumnId>* get_mutable_selected_column_indexs() { return &_selected_column_indexs; }
 

be/src/storage/tablet_updates.cpp

@@ -36,7 +36,6 @@
 #include "storage/types.h"
 #include "storage/update_compaction_state.h"
 #include "storage/update_manager.h"
-#include "storage/wrapper_field.h"
 #include "util/defer_op.h"
 #include "util/pretty_printer.h"
 #include "util/scoped_cleanup.h"