fix some optimizer stats and regexp bugs

src/pl/sys_package/ob_dbms_stats.cpp

@@ -3580,7 +3580,7 @@ int ObDbmsStats::parse_table_info(ObExecContext &ctx,
       } else {/*do nothing*/}
     }
   }
-  if (OB_SUCC(ret) && table_schema != NULL) {
+  if (OB_SUCC(ret) && table_schema != NULL && !table_schema->is_view_table()) {
     param.table_id_ = table_schema->get_table_id();
     param.part_level_ = table_schema->get_part_level();
     if (OB_FAIL(set_param_global_part_id(ctx, param))) {
@@ -3629,7 +3629,7 @@ int ObDbmsStats::parse_table_info(ObExecContext &ctx,
     param.tenant_id_ = session->get_effective_tenant_id();
     param.is_temp_table_ = table_schema->is_tmp_table();
   }
-  if (OB_SUCC(ret) && table_schema != NULL) {
+  if (OB_SUCC(ret) && table_schema != NULL && !table_schema->is_view_table()) {
     param.table_id_ = table_schema->get_table_id();
     param.part_level_ = table_schema->get_part_level();
     if (OB_FAIL(set_param_global_part_id(ctx, param))) {
@@ -5094,7 +5094,8 @@ int ObDbmsStats::get_all_table_ids_in_database(ObExecContext &ctx,
             ret = OB_ERR_UNEXPECTED;
             LOG_WARN("get unexpected null", K(ret));
           } else if (!(table_schemas.at(i)->is_user_table() ||
-                       ObDbmsStatsUtils::is_stat_sys_table(table_schemas.at(i)->get_table_id()))) {
+                       ObDbmsStatsUtils::is_stat_sys_table(stat_param.tenant_id_,
+                                                           table_schemas.at(i)->get_table_id()))) {
             // only need following tables:
             // 1. user table
             // 2. valid sys table and real agent virtual table
@@ -5215,7 +5216,8 @@ int ObDbmsStats::get_need_statistics_tables(sql::ObExecContext &ctx,
             ret = OB_ERR_UNEXPECTED;
             LOG_WARN("get unexpected null", K(ret), K(table_schema));
           } else if (!(table_schema->is_user_table() ||
-                       ObDbmsStatsUtils::is_stat_sys_table(table_schema->get_table_id()))) {
+                       ObDbmsStatsUtils::is_stat_sys_table(tenant_id,
+                                                           table_schema->get_table_id()))) {
             // only gather statistics for following tables:
             // 1. user table
             // 2. valid sys table and real agent virtual table

src/share/stat/ob_dbms_stats_utils.cpp

@@ -186,10 +186,13 @@ int ObDbmsStatsUtils::check_table_read_write_valid(const uint64_t tenant_id, boo
   return ret;
 }
 
-bool ObDbmsStatsUtils::is_stat_sys_table(const int64_t table_id)
+bool ObDbmsStatsUtils::is_stat_sys_table(const uint64_t tenant_id, const int64_t table_id)
 {
   const uint64_t id = table_id;
-  return (is_sys_table(id) || share::is_oracle_mapping_real_virtual_table(id)) &&
+  return (is_sys_table(id) ||
+          (share::is_oracle_mapping_real_virtual_table(id) &&
+           (!is_restrict_access_virtual_table(id) || tenant_id == OB_SYS_TENANT_ID))
+         ) &&
          !(is_core_table(table_id) ||
            ObSysTableChecker::is_sys_table_index_tid(table_id) ||
            id == share::OB_ALL_TABLE_STAT_TID ||

src/share/stat/ob_dbms_stats_utils.h

@@ -61,7 +61,7 @@ public:
 
   static int check_table_read_write_valid(const uint64_t tenant_id, bool &is_valid);
 
-  static bool is_stat_sys_table(const int64_t table_id);
+  static bool is_stat_sys_table(const uint64_t tenant_id, const int64_t table_id);
 
   static int parse_granularity(const ObString &granularity,
                                bool &need_global,

src/share/stat/ob_stats_estimator.cpp

@@ -14,6 +14,7 @@
 #include "ob_stats_estimator.h"
 #include "share/stat/ob_dbms_stats_utils.h"
 #include "observer/ob_inner_sql_connection_pool.h"
+#include "sql/optimizer/ob_opt_selectivity.h"
 
 namespace oceanbase
 {
@@ -450,7 +451,7 @@ int ObStatsEstimator::copy_opt_stat(ObOptStat &src_opt_stat,
         dst_opt_stats.at(i).table_stat_->set_row_count(row_cnt);
         dst_opt_stats.at(i).table_stat_->set_avg_row_size(tmp_tab_stat->get_avg_row_size());
         dst_opt_stats.at(i).table_stat_->set_sample_size(tmp_tab_stat->get_row_count());
-        if (OB_FAIL(copy_col_stats(tmp_col_stats, dst_opt_stats.at(i).column_stats_))) {
+        if (OB_FAIL(copy_col_stats(tmp_tab_stat->get_row_count(), row_cnt, tmp_col_stats, dst_opt_stats.at(i).column_stats_))) {
           LOG_WARN("failed to copy col stat", K(ret));
         } else {/*do nothing*/}
       } else {/*do nothing*/}
@@ -463,7 +464,9 @@ int ObStatsEstimator::copy_opt_stat(ObOptStat &src_opt_stat,
   return ret;
 }
 
-int ObStatsEstimator::copy_col_stats(ObIArray<ObOptColumnStat *> &src_col_stats,
+int ObStatsEstimator::copy_col_stats(const int64_t cur_row_cnt,
+                                     const int64_t total_row_cnt,
+                                     ObIArray<ObOptColumnStat *> &src_col_stats,
                                      ObIArray<ObOptColumnStat *> &dst_col_stats)
 {
   int ret = OB_SUCCESS;
@@ -480,9 +483,9 @@ int ObStatsEstimator::copy_col_stats(ObIArray<ObOptColumnStat *> &src_col_stats,
         int64_t num_null = src_col_stats.at(i)->get_num_null();
         int64_t num_distinct = src_col_stats.at(i)->get_num_distinct();
         if (sample_value_ >= 0.000001 && sample_value_ < 100.0) {
+          num_distinct = ObOptSelectivity::scale_distinct(total_row_cnt, cur_row_cnt, num_distinct);
           num_not_null = static_cast<int64_t>(num_not_null * 100 / sample_value_);
           num_null = static_cast<int64_t>(num_null * 100 / sample_value_);
-          num_distinct = static_cast<int64_t>(num_distinct * 100 / sample_value_);
         }
         dst_col_stats.at(i)->set_max_value(src_col_stats.at(i)->get_max_value());
         dst_col_stats.at(i)->set_min_value(src_col_stats.at(i)->get_min_value());

src/share/stat/ob_stats_estimator.h

@@ -100,7 +100,9 @@ private:
   int copy_hybrid_hist_stat(ObOptStat &src_opt_stat,
                             ObIArray<ObOptStat> &dst_opt_stats);
 
-  int copy_col_stats(ObIArray<ObOptColumnStat *> &src_col_stats,
+  int copy_col_stats(const int64_t cur_row_cnt,
+                     const int64_t total_row_cnt,
+                     ObIArray<ObOptColumnStat *> &src_col_stats,
                      ObIArray<ObOptColumnStat *> &dst_col_stats);
 
 protected:

src/sql/engine/expr/ob_expr_regexp_context.cpp

@@ -782,7 +782,7 @@ int ObExprRegexContext::get_valid_replace_string(ObIAllocator &alloc,
     //oracle mode replace string '\1' <==> '$1' in mysql mode, we need extra convert.
     UErrorCode m_error_code = U_ZERO_ERROR;
     int32_t group_count = uregex_groupCount(regexp_engine_, &m_error_code);
-    MEMSET(u_replace, 0, origin_replace.length() * sizeof(UChar));
+    MEMSET(u_replace, 0, buf_len);
     if (OB_FAIL(check_icu_regexp_status(m_error_code))) {
       LOG_WARN("failed to check icu regexp status", K(ret), K(u_errorName(m_error_code)));
     } else {