[chore](build) Use include-what-you-use to optimize includes (PART III) (#18958)

Currently, there are some useless includes in the codebase. We can use a tool named include-what-you-use to optimize these includes. By using a strict include-what-you-use policy, we can get lots of benefits from it.

be/src/vec/data_types/serde/data_type_array_serde.cpp

@@ -17,9 +17,14 @@
 
 #include "data_type_array_serde.h"
 
+#include "util/jsonb_document.h"
+#include "vec/columns/column.h"
+#include "vec/common/string_ref.h"
+
 namespace doris {
 
 namespace vectorized {
+class Arena;
 
 void DataTypeArraySerDe::write_one_cell_to_jsonb(const IColumn& column, JsonbWriter& result,
                                                  Arena* mem_pool, int32_t col_id,

be/src/vec/data_types/serde/data_type_array_serde.h

@@ -18,17 +18,21 @@
 #pragma once
 
 #include <glog/logging.h>
+#include <stdint.h>
 
 #include <ostream>
 
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeArraySerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_bitmap_serde.cpp

@@ -17,9 +17,15 @@
 
 #include "data_type_bitmap_serde.h"
 
+#include <gen_cpp/types.pb.h>
+
+#include <string>
+
+#include "util/bitmap_value.h"
+#include "util/jsonb_document.h"
 #include "vec/columns/column_complex.h"
 #include "vec/common/arena.h"
-#include "vec/data_types/data_type.h"
+#include "vec/common/assert_cast.h"
 
 namespace doris {
 

be/src/vec/data_types/serde/data_type_bitmap_serde.h

@@ -17,14 +17,19 @@
 
 #pragma once
 
+#include <stdint.h>
+
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeBitMapSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_decimal_serde.h

@@ -18,11 +18,18 @@
 #pragma once
 
 #include <gen_cpp/types.pb.h>
+#include <glog/logging.h>
 #include <stddef.h>
+#include <stdint.h>
+
+#include <ostream>
+#include <string>
 
 #include "common/status.h"
 #include "data_type_serde.h"
 #include "olap/olap_common.h"
+#include "util/jsonb_document.h"
+#include "util/jsonb_writer.h"
 #include "vec/columns/column.h"
 #include "vec/common/string_ref.h"
 #include "vec/core/types.h"
@@ -32,6 +39,7 @@ namespace doris {
 namespace vectorized {
 template <typename T>
 class ColumnDecimal;
+class Arena;
 
 template <typename T>
 class DataTypeDecimalSerDe : public DataTypeSerDe {

be/src/vec/data_types/serde/data_type_fixedlengthobject_serde.h

@@ -18,17 +18,21 @@
 #pragma once
 
 #include <glog/logging.h>
+#include <stdint.h>
 
 #include <ostream>
 
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeFixedLengthObjectSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_hll_serde.cpp

@@ -24,6 +24,7 @@
 #include <string>
 
 #include "olap/hll.h"
+#include "util/jsonb_document.h"
 #include "util/slice.h"
 #include "vec/columns/column_complex.h"
 #include "vec/common/arena.h"

be/src/vec/data_types/serde/data_type_hll_serde.h

@@ -17,14 +17,19 @@
 
 #pragma once
 
+#include <stdint.h>
+
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeHLLSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_map_serde.cpp

@@ -17,8 +17,14 @@
 
 #include "data_type_map_serde.h"
 
+#include "util/jsonb_document.h"
+#include "vec/columns/column.h"
+#include "vec/common/string_ref.h"
+
 namespace doris {
 namespace vectorized {
+class Arena;
+
 void DataTypeMapSerDe::read_one_cell_from_jsonb(IColumn& column, const JsonbValue* arg) const {
     auto blob = static_cast<const JsonbBlobVal*>(arg);
     column.deserialize_and_insert_from_arena(blob->getBlob());

be/src/vec/data_types/serde/data_type_map_serde.h

@@ -18,17 +18,21 @@
 #pragma once
 
 #include <glog/logging.h>
+#include <stdint.h>
 
 #include <ostream>
 
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeMapSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_nullable_serde.cpp

@@ -23,6 +23,7 @@
 #include <boost/iterator/iterator_facade.hpp>
 #include <memory>
 
+#include "util/jsonb_document.h"
 #include "vec/columns/column.h"
 #include "vec/columns/column_nullable.h"
 #include "vec/columns/column_vector.h"
@@ -33,6 +34,7 @@
 namespace doris {
 
 namespace vectorized {
+class Arena;
 
 Status DataTypeNullableSerDe::write_column_to_pb(const IColumn& column, PValues& result, int start,
                                                  int end) const {

be/src/vec/data_types/serde/data_type_nullable_serde.h

@@ -17,14 +17,19 @@
 
 #pragma once
 
+#include <stdint.h>
+
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeNullableSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_number_serde.h

@@ -18,19 +18,28 @@
 #pragma once
 
 #include <gen_cpp/types.pb.h>
+#include <glog/logging.h>
 #include <stddef.h>
+#include <stdint.h>
+
+#include <ostream>
+#include <string>
 
 #include "common/status.h"
 #include "data_type_serde.h"
 #include "olap/olap_common.h"
+#include "util/jsonb_document.h"
+#include "util/jsonb_writer.h"
 #include "vec/columns/column.h"
 #include "vec/columns/column_vector.h"
 #include "vec/common/string_ref.h"
 #include "vec/core/types.h"
 
 namespace doris {
+class JsonbOutStream;
 
 namespace vectorized {
+class Arena;
 
 template <typename T>
 class DataTypeNumberSerDe : public DataTypeSerDe {

be/src/vec/data_types/serde/data_type_object_serde.h

@@ -18,17 +18,21 @@
 #pragma once
 
 #include <glog/logging.h>
+#include <stdint.h>
 
 #include <ostream>
 
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeObjectSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_quantilestate_serde.h

@@ -17,9 +17,20 @@
 
 #pragma once
 
+#include <gen_cpp/types.pb.h>
+#include <stddef.h>
+#include <stdint.h>
+
+#include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_document.h"
+#include "util/jsonb_writer.h"
+#include "util/quantile_state.h"
+#include "util/slice.h"
+#include "vec/columns/column.h"
 #include "vec/columns/column_complex.h"
 #include "vec/common/arena.h"
+#include "vec/common/string_ref.h"
 
 namespace doris {
 

be/src/vec/data_types/serde/data_type_serde.h

@@ -17,18 +17,21 @@
 
 #pragma once
 
+#include <stdint.h>
+
 #include <memory>
 #include <vector>
 
 #include "common/status.h"
 #include "util/jsonb_writer.h"
-#include "vec/columns/column.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 // Deserialize means read from different file format or memory format,
 // for example read from arrow, read from parquet.
 // Serialize means write the column cell or the total column into another

be/src/vec/data_types/serde/data_type_string_serde.cpp

@@ -17,15 +17,18 @@
 
 #include "data_type_string_serde.h"
 
+#include <assert.h>
 #include <gen_cpp/types.pb.h>
 #include <stddef.h>
 
+#include "util/jsonb_document.h"
 #include "vec/columns/column.h"
 #include "vec/columns/column_string.h"
 #include "vec/common/string_ref.h"
 
 namespace doris {
 namespace vectorized {
+class Arena;
 
 Status DataTypeStringSerDe::write_column_to_pb(const IColumn& column, PValues& result, int start,
                                                int end) const {

be/src/vec/data_types/serde/data_type_string_serde.h

@@ -17,14 +17,19 @@
 
 #pragma once
 
+#include <stdint.h>
+
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeStringSerDe : public DataTypeSerDe {
 public:

be/src/vec/data_types/serde/data_type_struct_serde.cpp

@@ -16,9 +16,16 @@
 // under the License.
 
 #include "data_type_struct_serde.h"
+
+#include "util/jsonb_document.h"
+#include "vec/columns/column.h"
+#include "vec/common/string_ref.h"
+
 namespace doris {
 
 namespace vectorized {
+class Arena;
+
 void DataTypeStructSerDe::write_one_cell_to_jsonb(const IColumn& column, JsonbWriter& result,
                                                   Arena* mem_pool, int32_t col_id,
                                                   int row_num) const {

be/src/vec/data_types/serde/data_type_struct_serde.h

@@ -18,17 +18,21 @@
 #pragma once
 
 #include <glog/logging.h>
+#include <stdint.h>
 
 #include <ostream>
 
 #include "common/status.h"
 #include "data_type_serde.h"
+#include "util/jsonb_writer.h"
 
 namespace doris {
 class PValues;
+class JsonbValue;
 
 namespace vectorized {
 class IColumn;
+class Arena;
 
 class DataTypeStructSerDe : public DataTypeSerDe {
 public:

be/src/vec/functions/uuid.cpp

@@ -18,8 +18,8 @@
 #include <glog/logging.h>
 #include <stddef.h>
 
-#include <boost/uuid/random_generator.hpp>
-#include <boost/uuid/uuid_io.hpp>
+// IWYU pragma: no_include <boost/uuid/random_generator.hpp>
+// IWYU pragma: no_include <boost/uuid/uuid_io.hpp>
 #include <memory>
 #include <string>
 #include <utility>

be/src/vec/jsonb/serialize.cpp

@@ -18,34 +18,23 @@
 #include "vec/jsonb/serialize.h"
 
 #include <assert.h>
-#include <glog/logging.h>
-#include <stdint.h>
 
 #include <algorithm>
-#include <ostream>
+#include <memory>
 #include <vector>
 
-#include "olap/hll.h"
-#include "olap/olap_common.h"
 #include "olap/tablet_schema.h"
-#include "runtime/define_primitive_type.h"
 #include "runtime/descriptors.h"
-#include "runtime/primitive_type.h"
-#include "runtime/types.h"
-#include "util/bitmap_value.h"
 #include "util/jsonb_document.h"
 #include "util/jsonb_stream.h"
 #include "util/jsonb_writer.h"
-#include "util/slice.h"
 #include "vec/columns/column.h"
 #include "vec/columns/column_string.h"
 #include "vec/common/arena.h"
 #include "vec/common/string_ref.h"
 #include "vec/core/column_with_type_and_name.h"
-#include "vec/core/field.h"
-#include "vec/core/types.h"
 #include "vec/data_types/data_type.h"
-#include "vec/runtime/vdatetime_value.h"
+#include "vec/data_types/serde/data_type_serde.h"
 
 namespace doris::vectorized {
 

be/src/vec/jsonb/serialize.h

@@ -18,7 +18,6 @@
 #pragma once
 #include <stddef.h>
 
-#include "runtime/descriptors.h"
 #include "vec/core/block.h"
 
 namespace doris {