upd

be/src/olap/rowset/segment_v2/segment_writer.cpp

@@ -23,6 +23,7 @@
 
 // IWYU pragma: no_include <opentelemetry/common/threadlocal.h>
 #include "cloud/config.h"
+#include "common/cast_set.h"
 #include "common/compiler_util.h" // IWYU pragma: keep
 #include "common/config.h"
 #include "common/logging.h" // LOG
@@ -74,7 +75,7 @@ namespace doris {
 namespace segment_v2 {
 
 using namespace ErrorCode;
-
+#include "common/compile_check_begin.h"
 const char* k_segment_magic = "D0R1";
 const uint32_t k_segment_magic_length = 4;
 
@@ -153,7 +154,7 @@ void SegmentWriter::init_column_meta(ColumnMetaPB* meta, uint32_t column_id,
  const TabletColumn& column, TabletSchemaSPtr tablet_schema) {
  meta->set_column_id(column_id);
  meta->set_type(int(column.type()));
- meta->set_length(column.length());
+ meta->set_length(cast_set<int>(column.length()));
  meta->set_encoding(DEFAULT_ENCODING);
  meta->set_compression(_opts.compression_type);
  meta->set_is_nullable(column.is_nullable());
@@ -180,7 +181,7 @@ void SegmentWriter::init_column_meta(ColumnMetaPB* meta, uint32_t column_id,
 
 Status SegmentWriter::init() {
  std::vector<uint32_t> column_ids;
- int column_cnt = _tablet_schema->num_columns();
+ auto column_cnt = _tablet_schema->num_columns();
  for (uint32_t i = 0; i < column_cnt; ++i) {
  column_ids.emplace_back(i);
  }
@@ -377,7 +378,7 @@ Status SegmentWriter::append_block_with_variant_subcolumns(vectorized::Block& da
  continue;
  }
  CHECK(entry->data.is_finalized());
- int current_column_id = column_id++;
+ auto current_column_id = cast_set<uint32_t>(column_id++);
  TabletColumn tablet_column = generate_column_info(entry);
  vectorized::schema_util::inherit_column_attributes(*parent_column, tablet_column,
  _flush_schema);
@@ -471,16 +472,16 @@ Status SegmentWriter::probe_key_for_mow(
  RowLocation loc;
  // save rowset shared ptr so this rowset wouldn't delete
  RowsetSharedPtr rowset;
- auto st = _tablet->lookup_row_key(key, _tablet_schema.get(), have_input_seq_column,
-  specified_rowsets, &loc, _mow_context->max_version,
-   segment_caches, &rowset);
+ auto st = _tablet->lookup_row_key(
+ key, _tablet_schema.get(), have_input_seq_column, specified_rowsets, &loc,
+ cast_set<uint32_t>(_mow_context->max_version), segment_caches, &rowset);
  if (st.is<KEY_NOT_FOUND>()) {
  if (_opts.rowset_ctx->partial_update_info->is_strict_mode) {
  ++stats.num_rows_filtered;
  // delete the invalid newly inserted row
  _mow_context->delete_bitmap->add(
  {_opts.rowset_ctx->rowset_id, _segment_id, DeleteBitmap::TEMP_VERSION_COMMON},
- segment_pos);
+ cast_set<uint32_t>(segment_pos));
  } else if (!have_delete_sign) {
  RETURN_IF_ERROR(
  _opts.rowset_ctx->partial_update_info->handle_non_strict_mode_not_found_error(
@@ -516,7 +517,7 @@ Status SegmentWriter::probe_key_for_mow(
  // for this row, we need to ensure that each column is aligned
  _mow_context->delete_bitmap->add(
  {_opts.rowset_ctx->rowset_id, _segment_id, DeleteBitmap::TEMP_VERSION_COMMON},
- segment_pos);
+ cast_set<uint32_t>(segment_pos));
  ++stats.num_rows_deleted;
  } else {
  _mow_context->delete_bitmap->add(
@@ -892,7 +893,7 @@ std::string SegmentWriter::_encode_keys(
 
 template <typename RowType>
 Status SegmentWriter::append_row(const RowType& row) {
- for (size_t cid = 0; cid < _column_writers.size(); ++cid) {
+ for (uint32_t cid = 0; cid < _column_writers.size(); ++cid) {
  auto cell = row.cell(cid);
  RETURN_IF_ERROR(_column_writers[cid]->append(cell));
  }

be/src/util/coding.h

@@ -108,7 +108,7 @@ inline uint128_t decode_fixed128_le(const uint8_t* buf) {
 }
 
 template <typename T>
-void put_fixed32_le(T* dst, uint32_t val) {
+void put_fixed32_le(T* dst, uint64_t val) {
  uint8_t buf[sizeof(val)];
  encode_fixed32_le(buf, val);
  dst->append((char*)buf, sizeof(buf));

be/src/vec/jsonb/serialize.cpp

@@ -45,7 +45,7 @@
 namespace doris::vectorized {
 
 void JsonbSerializeUtil::block_to_jsonb(const TabletSchema& schema, const Block& block,
- ColumnString& dst, int num_cols,
+ ColumnString& dst, size_t num_cols,
  const DataTypeSerDeSPtrs& serdes,
  const std::unordered_set<int32_t>& row_store_cids) {
  auto num_rows = block.rows();

be/src/vec/jsonb/serialize.h

@@ -39,7 +39,7 @@ class JsonbSerializeUtil {
  // encode partial columns into jsonb
  // empty row_store_cids means encode full schema columns for compability
  static void block_to_jsonb(const TabletSchema& schema, const Block& block, ColumnString& dst,
- int num_cols, const DataTypeSerDeSPtrs& serdes,
+ size_t num_cols, const DataTypeSerDeSPtrs& serdes,
  const std::unordered_set<int32_t>& row_store_cids);
  // batch rows
  static void jsonb_to_block(const DataTypeSerDeSPtrs& serdes, const ColumnString& jsonb_column,

be/src/vec/runtime/vdatetime_value.cpp

@@ -42,6 +42,8 @@
 
 namespace doris {
 
+#include "common/compile_check_begin.h"
+
 static const char* s_ab_month_name[] = {"", "Jan", "Feb", "Mar", "Apr", "May", "Jun",
  "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", nullptr};
 

be/src/vec/runtime/vdatetime_value.h

@@ -32,6 +32,7 @@
 #include <type_traits>
 #include <utility>
 
+#include "common/cast_set.h"
 #include "util/hash_util.hpp"
 #include "util/time_lut.h"
 #include "util/timezone_utils.h"
@@ -47,7 +48,7 @@ class DataTypeDateTimeV2;
 } // namespace doris::vectorized
 
 namespace doris {
-
+#include "common/compile_check_begin.h"
 enum TimeUnit {
  MICROSECOND,
  MILLISECOND,
@@ -297,15 +298,17 @@ class VecDateTimeValue { // Now this type is a temp solution with little changes
  _type = TIME_DATETIME;
  uint64_t date = datetime / 1000000;
  uint64_t time = datetime % 1000000;
-
- auto [year, month, day, hour, minute, second] = std::tuple {0, 0, 0, 0, 0, 0};
- year = date / 10000;
+ // uint32_t year, uint32_t month, uint32_t day, uint32_t hour,uint32_t minute, uint32_t second
+ auto [year, month, day, hour, minute, second] =
+ std::tuple<uint32_t, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t> {0, 0, 0,
+ 0, 0, 0};
+ cast_set(year, date / 10000);
  date %= 10000;
- month = date / 100;
+ cast_set(month, date / 100);
  day = date % 100;
- hour = time / 10000;
+ cast_set(hour, time / 10000);
  time %= 10000;
- minute = time / 100;
+ cast_set(minute, time / 100);
  second = time % 100;
 
  return check_range_and_set_time(year, month, day, hour, minute, second, _type);
@@ -320,14 +323,16 @@ class VecDateTimeValue { // Now this type is a temp solution with little changes
  bool from_olap_date(uint64_t date) {
  _neg = 0;
  _type = TIME_DATE;
-
- auto [year, month, day, hour, minute, second] = std::tuple {0, 0, 0, 0, 0, 0};
+ // uint32_t year, uint32_t month, uint32_t day, uint32_t hour,uint32_t minute, uint32_t second
+ auto [year, month, day, hour, minute, second] =
+ std::tuple<uint32_t, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t> {0, 0, 0,
+ 0, 0, 0};
 
  day = date & 0x1f;
  date >>= 5;
  month = date & 0x0f;
  date >>= 4;
- year = date;
+ cast_set(year, date);
 
  return check_range_and_set_time(year, month, day, hour, minute, second, _type);
  }
@@ -623,7 +628,7 @@ class VecDateTimeValue { // Now this type is a temp solution with little changes
 
  uint32_t hash(int seed) const { return HashUtil::hash(this, sizeof(*this), seed); }
 
- int day_of_year() const { return daynr() - calc_daynr(_year, 1, 1) + 1; }
+ int64_t day_of_year() const { return daynr() - calc_daynr(_year, 1, 1) + 1; }
 
  // TODO(zhaochun): local time ???
  static VecDateTimeValue local_time();
@@ -786,31 +791,33 @@ class DateV2Value {
  void set_microsecond(uint32_t microsecond);
 
  bool from_olap_date(uint64_t date) {
- auto [year, month, day] = std::tuple {0, 0, 0};
+ //uint16_t year, uint8_t month, uint8_t day
+ auto [year, month, day] = std::tuple<uint16_t, uint8_t, uint8_t> {0, 0, 0};
 
  day = date & 0x1f;
  date >>= 5;
  month = date & 0x0f;
  date >>= 4;
- year = date;
+ cast_set(year, date);
 
  return check_range_and_set_time(year, month, day, 0, 0, 0, 0);
  }
 
  bool from_olap_datetime(uint64_t datetime) {
  uint64_t date = datetime / 1000000;
  uint64_t time = datetime % 1000000;
-
- auto [year, month, day, hour, minute, second] = std::tuple {0, 0, 0, 0, 0, 0};
- year = date / 10000;
+ //uint16_t year, uint8_t month, uint8_t day,uint8_t hour, uint8_t minute, uint8_t second, uint32_t microsecond
+ auto [year, month, day, hour, minute, second] =
+ std::tuple<uint16_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t> {0, 0, 0,
+ 0, 0, 0};
+ cast_set(year, date / 10000);
  date %= 10000;
- month = date / 100;
- day = date % 100;
- hour = time / 10000;
+ cast_set(month, date / 100);
+ cast_set(day, date % 100);
+ cast_set(hour, time / 10000);
  time %= 10000;
- minute = time / 100;
- second = time % 100;
-
+ cast_set(minute, time / 100);
+ cast_set(second, time % 100);
  return check_range_and_set_time(year, month, day, hour, minute, second, 0);
  }
 
@@ -1116,7 +1123,7 @@ class DateV2Value {
 
  uint32_t hash(int seed) const { return HashUtil::hash(this, sizeof(*this), seed); }
 
- int day_of_year() const { return daynr() - calc_daynr(this->year(), 1, 1) + 1; }
+ int64_t day_of_year() const { return daynr() - calc_daynr(this->year(), 1, 1) + 1; }
 
  std::string debug_string() const {
  char buf[64];
@@ -1294,7 +1301,7 @@ class DateV2Value {
 
  void format_datetime(uint32_t* date_v, bool* carry_bits) const;
 
- void set_int_val(uint64_t val) { this->int_val_ = val; }
+ void set_int_val(underlying_value val) { this->int_val_ = val; }
 
 private:
  static uint8_t calc_week(const uint32_t& day_nr, const uint16_t& year, const uint8_t& month,
@@ -1369,7 +1376,7 @@ int64_t datetime_diff(const VecDateTimeValue& ts_value1, const VecDateTimeValue&
  return month;
  }
  case WEEK: {
- int day = ts_value2.daynr() - ts_value1.daynr();
+ int64_t day = ts_value2.daynr() - ts_value1.daynr();
  if (day > 0) {
  day -= ts_value2.time_part_diff(ts_value1) < 0;
  } else if (day < 0) {
@@ -1378,7 +1385,7 @@ int64_t datetime_diff(const VecDateTimeValue& ts_value1, const VecDateTimeValue&
  return day / 7;
  }
  case DAY: {
- int day = ts_value2.daynr() - ts_value1.daynr();
+ int64_t day = ts_value2.daynr() - ts_value1.daynr();
  if (day > 0) {
  day -= ts_value2.time_part_diff(ts_value1) < 0;
  } else if (day < 0) {
@@ -1493,7 +1500,7 @@ int64_t datetime_diff(const DateV2Value<T0>& ts_value1, const DateV2Value<T1>& t
  return month;
  }
  case WEEK: {
- int day = ts_value2.daynr() - ts_value1.daynr();
+ int64_t day = ts_value2.daynr() - ts_value1.daynr();
  int64_t ms_diff = ts_value2.time_part_diff_microsecond(ts_value1);
  if (day > 0 && ms_diff < 0) {
  day--;
@@ -1503,7 +1510,7 @@ int64_t datetime_diff(const DateV2Value<T0>& ts_value1, const DateV2Value<T1>& t
  return day / 7;
  }
  case DAY: {
- int day = ts_value2.daynr() - ts_value1.daynr();
+ int64_t day = ts_value2.daynr() - ts_value1.daynr();
  int64_t ms_diff = ts_value2.time_part_diff_microsecond(ts_value1);
  if (day > 0 && ms_diff < 0) {
  day--;
@@ -1634,3 +1641,5 @@ struct std::hash<doris::DateV2Value<doris::DateTimeV2ValueType>> {
  return doris::hash_value(v);
  }
 };
+
+#include "common/compile_check_end.h"