commit ea8b77c435bccd88c3082030159556dd3a78404c Author: Oleksandr Byelkin Date: Sun Aug 14 19:19:54 2022 +0200 Implementation of BERNOULLI for TABLESAMPLE diff --git a/mysql-test/main/sample_select.result b/mysql-test/main/sample_select.result new file mode 100644 index 00000000000..3e32a8f2113 --- /dev/null +++ b/mysql-test/main/sample_select.result @@ -0,0 +1,14 @@ +create table t1 (id int not null primary key, +val int); +select avg(id),count(id) from t1; +avg(id) count(id) +500.5000 1000 +# use second table, because count() do not work correctly +# with TABLESAMPLE +create table t2 (id int not null, +val int); +insert into t2 select * from t1 TABLESAMPLE BERNOULLI (50); +select abs(avg(id) - 500.5) < 500.5*0.1, abs(count(id) - 500) < 500*0.1 from t2; +abs(avg(id) - 500.5) < 500.5*0.1 abs(count(id) - 500) < 500*0.1 +1 1 +drop table t1, t2; diff --git a/mysql-test/main/sample_select.test b/mysql-test/main/sample_select.test new file mode 100644 index 00000000000..0b050ad6d3d --- /dev/null +++ b/mysql-test/main/sample_select.test @@ -0,0 +1,25 @@ +create table t1 (id int not null primary key, + val int); + +let $1=1000; + +--disable_query_log +while ($1) +{ + eval insert into t1 values ( $1 , 10); + dec $1; +} +--enable_query_log + +select avg(id),count(id) from t1; + +--echo # use second table, because count() do not work correctly +--echo # with TABLESAMPLE +create table t2 (id int not null, + val int); + +insert into t2 select * from t1 TABLESAMPLE BERNOULLI (50); + +select abs(avg(id) - 500.5) < 500.5*0.1, abs(count(id) - 500) < 500*0.1 from t2; + +drop table t1, t2; diff --git a/mysql-test/suite/perfschema/r/digest_view.result b/mysql-test/suite/perfschema/r/digest_view.result index 3fd7a090459..4c4fed39430 100644 --- a/mysql-test/suite/perfschema/r/digest_view.result +++ b/mysql-test/suite/perfschema/r/digest_view.result @@ -191,17 +191,17 @@ SELECT SCHEMA_NAME, DIGEST, DIGEST_TEXT, COUNT_STAR FROM performance_schema.events_statements_summary_by_digest ORDER BY DIGEST_TEXT; SCHEMA_NAME DIGEST DIGEST_TEXT COUNT_STAR -test 333a53e537d74bf74dd28c78ad5b23dd EXPLAIN SELECT * FROM `test` . `v1` 1 -test 1fb578da66e6583bae8e64061486f1b1 EXPLAIN SELECT * FROM `test` . `v1` WHERE `a` = ? 1 -test 923bca939a55826e231e1335016ba418 EXPLAIN SELECT * FROM `test` . `v1` WHERE `b` > ? 1 -test 6fa6d75432fd499d1c7d6f964c8310a2 EXPLAIN SELECT `a` , `b` FROM `test` . `v1` 1 -test 3421da1ec8ecb8cac97e12a0609f73cb EXPLAIN SELECT `b` , `a` FROM `test` . `v1` 1 -test c2f68fd41bfbf3acd52eb5e7306b0c00 SELECT * FROM `test` . `v1` 1 -test dd68812cbe4c1ed5a38921222981f8c5 SELECT * FROM `test` . `v1` WHERE `a` = ? 1 -test 7563bcc32c6d0d872c8d9f0bf7717e6a SELECT * FROM `test` . `v1` WHERE `b` > ? 1 -test 0817c53833dc6adbca581e8fe4c598c7 SELECT `a` , `b` FROM `test` . `v1` 1 -test a5f13903c70812ae08fa8c084e9cd503 SELECT `b` , `a` FROM `test` . `v1` 1 -test 66b14a14f2a42e1335dd28dfad8ea084 TRUNCATE TABLE `performance_schema` . `events_statements_summary_by_digest` 1 +test 0cd509bef4fcaec486c6fe40f5d0605d EXPLAIN SELECT * FROM `test` . `v1` 1 +test 78984efc9d252b92f78c0301adfb885f EXPLAIN SELECT * FROM `test` . `v1` WHERE `a` = ? 1 +test 5a70d7b33fff3b10abe27f89e86153e2 EXPLAIN SELECT * FROM `test` . `v1` WHERE `b` > ? 1 +test 0e7ac33d57dba3db9619605f3e5beab8 EXPLAIN SELECT `a` , `b` FROM `test` . `v1` 1 +test 94e84b1d8e23bb5072e90b93e819f90e EXPLAIN SELECT `b` , `a` FROM `test` . `v1` 1 +test 4bee0754c6f0dd9b07c7924d8a16a316 SELECT * FROM `test` . `v1` 1 +test d17fa9790594cd1f8df5314c95f7f7bf SELECT * FROM `test` . `v1` WHERE `a` = ? 1 +test dd83248bed00a08e37ec370f4219d938 SELECT * FROM `test` . `v1` WHERE `b` > ? 1 +test 72144fe1e02106cfdf84ca3817f60f03 SELECT `a` , `b` FROM `test` . `v1` 1 +test b605d56234b6b28974e8ec9562a544a2 SELECT `b` , `a` FROM `test` . `v1` 1 +test 6a416f83fde8bae670e01cd1b29b1c18 TRUNCATE TABLE `performance_schema` . `events_statements_summary_by_digest` 1 DROP TABLE test.v1; CREATE VIEW test.v1 AS SELECT * FROM test.t1; EXPLAIN SELECT * from test.v1; @@ -248,19 +248,19 @@ SELECT SCHEMA_NAME, DIGEST, DIGEST_TEXT, COUNT_STAR FROM performance_schema.events_statements_summary_by_digest ORDER BY DIGEST_TEXT; SCHEMA_NAME DIGEST DIGEST_TEXT COUNT_STAR -test 98531b331031b84ddfbb2de8b601a704 CREATE VIEW `test` . `v1` AS SELECT * FROM `test` . `t1` 1 -test 5352d7d117e97fecd312e354e9e290ce DROP TABLE `test` . `v1` 1 -test 333a53e537d74bf74dd28c78ad5b23dd EXPLAIN SELECT * FROM `test` . `v1` 2 -test 1fb578da66e6583bae8e64061486f1b1 EXPLAIN SELECT * FROM `test` . `v1` WHERE `a` = ? 2 -test 923bca939a55826e231e1335016ba418 EXPLAIN SELECT * FROM `test` . `v1` WHERE `b` > ? 2 -test 6fa6d75432fd499d1c7d6f964c8310a2 EXPLAIN SELECT `a` , `b` FROM `test` . `v1` 2 -test 3421da1ec8ecb8cac97e12a0609f73cb EXPLAIN SELECT `b` , `a` FROM `test` . `v1` 2 -test c2f68fd41bfbf3acd52eb5e7306b0c00 SELECT * FROM `test` . `v1` 2 -test dd68812cbe4c1ed5a38921222981f8c5 SELECT * FROM `test` . `v1` WHERE `a` = ? 2 -test 7563bcc32c6d0d872c8d9f0bf7717e6a SELECT * FROM `test` . `v1` WHERE `b` > ? 2 -test 21c19dd7ef5b894f3e32d0585cb3007f SELECT SCHEMA_NAME , `DIGEST` , `DIGEST_TEXT` , `COUNT_STAR` FROM `performance_schema` . `events_statements_summary_by_digest` ORDER BY `DIGEST_TEXT` 1 -test 0817c53833dc6adbca581e8fe4c598c7 SELECT `a` , `b` FROM `test` . `v1` 2 -test a5f13903c70812ae08fa8c084e9cd503 SELECT `b` , `a` FROM `test` . `v1` 2 -test 66b14a14f2a42e1335dd28dfad8ea084 TRUNCATE TABLE `performance_schema` . `events_statements_summary_by_digest` 1 +test 81ba956e271ccc6710100f66fbe89a95 CREATE VIEW `test` . `v1` AS SELECT * FROM `test` . `t1` 1 +test 741e23323033aa1a27d484e31273d1f7 DROP TABLE `test` . `v1` 1 +test 0cd509bef4fcaec486c6fe40f5d0605d EXPLAIN SELECT * FROM `test` . `v1` 2 +test 78984efc9d252b92f78c0301adfb885f EXPLAIN SELECT * FROM `test` . `v1` WHERE `a` = ? 2 +test 5a70d7b33fff3b10abe27f89e86153e2 EXPLAIN SELECT * FROM `test` . `v1` WHERE `b` > ? 2 +test 0e7ac33d57dba3db9619605f3e5beab8 EXPLAIN SELECT `a` , `b` FROM `test` . `v1` 2 +test 94e84b1d8e23bb5072e90b93e819f90e EXPLAIN SELECT `b` , `a` FROM `test` . `v1` 2 +test 4bee0754c6f0dd9b07c7924d8a16a316 SELECT * FROM `test` . `v1` 2 +test d17fa9790594cd1f8df5314c95f7f7bf SELECT * FROM `test` . `v1` WHERE `a` = ? 2 +test dd83248bed00a08e37ec370f4219d938 SELECT * FROM `test` . `v1` WHERE `b` > ? 2 +test 37b6bbf54848f2a0bfff4c3cf142d50a SELECT SCHEMA_NAME , `DIGEST` , `DIGEST_TEXT` , `COUNT_STAR` FROM `performance_schema` . `events_statements_summary_by_digest` ORDER BY `DIGEST_TEXT` 1 +test 72144fe1e02106cfdf84ca3817f60f03 SELECT `a` , `b` FROM `test` . `v1` 2 +test b605d56234b6b28974e8ec9562a544a2 SELECT `b` , `a` FROM `test` . `v1` 2 +test 6a416f83fde8bae670e01cd1b29b1c18 TRUNCATE TABLE `performance_schema` . `events_statements_summary_by_digest` 1 DROP VIEW test.v1; DROP TABLE test.t1; diff --git a/mysql-test/suite/perfschema/r/start_server_low_digest_sql_length.result b/mysql-test/suite/perfschema/r/start_server_low_digest_sql_length.result index ff554155f33..4bbe49f4be3 100644 --- a/mysql-test/suite/perfschema/r/start_server_low_digest_sql_length.result +++ b/mysql-test/suite/perfschema/r/start_server_low_digest_sql_length.result @@ -8,5 +8,5 @@ SELECT 1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1 #################################### SELECT event_name, digest, digest_text, sql_text FROM events_statements_history_long; event_name digest digest_text sql_text -statement/sql/select 96a6bb95edaf100857f09c968aca354b SELECT ? + ? + SELECT ... -statement/sql/truncate c1d647bb870f2c63c22a16707bb8aee3 TRUNCATE TABLE truncat... +statement/sql/select 157d58c66754c972235fd9535db86f79 SELECT ? + ? + SELECT ... +statement/sql/truncate a8b0b5b2e2ff71e80a50828553a70e9a TRUNCATE TABLE truncat... diff --git a/sql/lex.h b/sql/lex.h index ae70537c73b..a81f2b50bb0 100644 --- a/sql/lex.h +++ b/sql/lex.h @@ -87,6 +87,7 @@ SYMBOL symbols[] = { { "BACKUP", SYM(BACKUP_SYM)}, { "BEFORE", SYM(BEFORE_SYM)}, { "BEGIN", SYM(BEGIN_MARIADB_SYM)}, + { "BERNOULLI", SYM(BERNOULLI_SYM)}, { "BETWEEN", SYM(BETWEEN_SYM)}, { "BIGINT", SYM(BIGINT)}, { "BINARY", SYM(BINARY)}, diff --git a/sql/records.cc b/sql/records.cc index 9eadf788585..e4cae4696c4 100644 --- a/sql/records.cc +++ b/sql/records.cc @@ -34,6 +34,7 @@ #include "sql_class.h" // THD #include "sql_base.h" #include "sql_sort.h" // SORT_ADDON_FIELD +#include "sql_statistics.h" static int rr_quick(READ_RECORD *info); int rr_sequential(READ_RECORD *info); @@ -324,10 +325,9 @@ bool init_read_record(READ_RECORD *info,THD *thd, TABLE *table, } else if (table->tablesample) { - double fract= table->tablesample->val_real() / 100.0; - info->sample_counter= (ha_rows)(table->file->records() * fract + 0.5); + info->tablesample= table->tablesample; info->read_record_func= rr_sequential_sample; - if (table->file->ha_sample_init()) + if (table->tablesample->init(thd, table)) DBUG_RETURN(1); } else @@ -535,17 +535,10 @@ int rr_sequential(READ_RECORD *info) int rr_sequential_sample(READ_RECORD *info) { - int tmp; - if (!info->sample_counter) - return -1; // End of file - - info->sample_counter--; - while ((tmp= info->table->file->ha_sample_next(info->record()))) - { - tmp= rr_handle_error(info, tmp); - break; - } - return tmp; + int rc= info->tablesample->next(info->record()); + if (rc) + rc= rr_handle_error(info, rc); + return rc; } diff --git a/sql/records.h b/sql/records.h index 579e0e2bf3a..9bbfbbea1a8 100644 --- a/sql/records.h +++ b/sql/records.h @@ -61,7 +61,7 @@ struct READ_RECORD SQL_SELECT *select; uint ref_length, reclength, rec_cache_size, error_offset; - ha_rows sample_counter; + Statistic_collector *tablesample; /** Counting records when reading result from filesort(). Used when filesort leaves the result in the filesort buffer. diff --git a/sql/sql_statistics.cc b/sql/sql_statistics.cc index f2a38e27722..ccee3c8cb08 100644 --- a/sql/sql_statistics.cc +++ b/sql/sql_statistics.cc @@ -2672,6 +2672,70 @@ int collect_statistics_for_index(THD *thd, TABLE *table, uint index) } +bool System_staticitc_collector::init(THD *thd __attribute__((unused)), + TABLE *table) +{ + file= table->file; + limit= (ha_rows)(file->records() * sample_fraction + 0.5); + if (file->ha_sample_init()) + return (TRUE); + return (FALSE); +} + +int System_staticitc_collector::next(uchar *record) +{ + int rc= 0; + if (!limit) + return HA_ERR_END_OF_FILE; + + if (!(rc= file->ha_sample_next(record))) + limit --; + return rc; +} + +void System_staticitc_collector::close() +{ + if (file) + { + file->ha_sample_end(); + file= 0; + } +} + +bool Bernoulli_staticitc_collector::init(THD *thd, TABLE *table) +{ + this->thd= thd; + this->file= table->file; + this->sample_fraction= sample_fraction; + if (file->ha_rnd_init(TRUE)) + return (TRUE); + return (FALSE); +} + +int Bernoulli_staticitc_collector::next(uchar *record) +{ + int rc; + while (!(rc= file->ha_rnd_next(record))) + { + if (thd->killed) + return 0; + + if (thd_rnd(thd) <= sample_fraction) + return 0; + } + return rc; +} + +void Bernoulli_staticitc_collector::close() +{ + if (file) + { + file->ha_rnd_end(); + file= 0; + } +} + + /** @brief Collect statistical data for a table @@ -2763,73 +2827,42 @@ int collect_statistics_for_table(THD *thd, TABLE *table) restore_record(table, s->default_values); - if(file->ha_table_flags() & HA_NATIVE_SAMPLING) - { - if(!(rc= file->ha_sample_init())) { - DEBUG_SYNC(table->in_use, "statistics_collection_start"); - rows = file->records() * (thd->variables.sample_percentage / 100); - - for (ulonglong i= 0; i < rows; ++i) - { - rc = file->ha_sample_next(table->record[0]); - if (thd->killed) - break; - - if (rc) - break; - for (field_ptr= table->field; *field_ptr; field_ptr++) - { - table_field= *field_ptr; - if (!table_field->collected_stats) - continue; - if ((rc= table_field->collected_stats->add())) - break; - } - if (rc) - break; - } - file->ha_sample_end(); - } - rc= (rc == 0 && !thd->killed) ? 0 : 1; - } + Statistic_collector *collector; + if (file->ha_table_flags() & HA_NATIVE_SAMPLING) + collector= + new (thd->mem_root) System_staticitc_collector(sample_fraction); else - { + collector= + new (thd->mem_root) Bernoulli_staticitc_collector(sample_fraction); - /* Perform a full table scan to collect statistics on 'table's columns */ - if (!(rc= file->ha_rnd_init(TRUE))) - { - DEBUG_SYNC(table->in_use, "statistics_collection_start"); + if (collector->init(thd, table)) + DBUG_RETURN(1); + DEBUG_SYNC(table->in_use, "statistics_collection_start"); - while ((rc= file->ha_rnd_next(table->record[0])) != HA_ERR_END_OF_FILE) - { - if (thd->killed) - break; + while ((rc= collector->next(table->record[0])) != HA_ERR_END_OF_FILE) + { + if (thd->killed) + break; - if (rc) - break; + if (rc) + break; - if (thd_rnd(thd) <= sample_fraction) - { - for (field_ptr= table->field; *field_ptr; field_ptr++) - { - table_field= *field_ptr; - if (!table_field->collected_stats) - continue; - if ((rc= table_field->collected_stats->add())) - break; - } - if (rc) - break; - rows++; - } - } - file->ha_rnd_end(); + for (field_ptr= table->field; *field_ptr; field_ptr++) + { + table_field= *field_ptr; + if (!table_field->collected_stats) + continue; + if ((rc= table_field->collected_stats->add())) + break; } - rc= (rc == HA_ERR_END_OF_FILE && !thd->killed) ? 0 : 1; + if (rc) + break; + rows++; } + collector->close(); + rc= (rc == HA_ERR_END_OF_FILE && !thd->killed) ? 0 : 1; - - /* + /* Calculate values for all statistical characteristics on columns and and for each field f of 'table' save them in the write_stat structure from the Field object for f. diff --git a/sql/sql_statistics.h b/sql/sql_statistics.h index c0df15ea4ad..7a938304eca 100644 --- a/sql/sql_statistics.h +++ b/sql/sql_statistics.h @@ -620,4 +620,48 @@ class Index_statistics }; +class Statistic_collector: public Sql_alloc +{ + protected: + handler *file; + double sample_fraction; + + public: + Statistic_collector(double fraction): + file(0), + sample_fraction(fraction) + {} + virtual ~Statistic_collector() + {} + + + virtual bool init(THD *thd, TABLE *table)= 0; + virtual int next(uchar *record)= 0; + virtual void close()= 0; +}; + +class System_staticitc_collector: public Statistic_collector +{ + ha_rows limit; + public: + System_staticitc_collector(double fraction): + Statistic_collector(fraction) + {} + bool init(THD *thd, TABLE *table) override; + int next(uchar *record) override; + void close() override; +}; + +class Bernoulli_staticitc_collector: public Statistic_collector +{ + THD *thd; + public: + Bernoulli_staticitc_collector(double fraction): + Statistic_collector(fraction) + {} + bool init(THD *thd, TABLE *table) override; + int next(uchar *record) override; + void close() override; +}; + #endif /* SQL_STATISTICS_H */ diff --git a/sql/sql_yacc.yy b/sql/sql_yacc.yy index 2ad1a02170e..2b1e940de0c 100644 --- a/sql/sql_yacc.yy +++ b/sql/sql_yacc.yy @@ -69,6 +69,7 @@ #include "my_base.h" #include "sql_type_json.h" #include "json_table.h" +#include "sql_statistics.h" /* this is to get the bison compilation windows warnings out */ #ifdef _MSC_VER @@ -329,6 +330,7 @@ void _CONCAT_UNDERSCORED(turn_parser_debug_on,yyparse)() enum Column_definition::enum_column_versioning vers_column_versioning; enum plsql_cursor_attr_t plsql_cursor_attr; privilege_t privilege; + Statistic_collector *sample_collector; } %{ @@ -755,6 +757,7 @@ bool my_yyoverflow(short **a, YYSTYPE **b, size_t *yystacksize); %token BACKUP_SYM %token BEGIN_MARIADB_SYM /* SQL-2003-R, PLSQL-R */ %token BEGIN_ORACLE_SYM /* SQL-2003-R, PLSQL-R */ +%token BERNOULLI_SYM %token BINLOG_SYM %token BIT_SYM /* MYSQL-FUNC */ %token BLOCK_SYM @@ -1461,7 +1464,7 @@ bool my_yyoverflow(short **a, YYSTYPE **b, size_t *yystacksize); ws_level_flag_desc ws_level_flag_reverse ws_level_flags opt_ws_levels ws_level_list ws_level_list_item ws_level_number ws_level_range ws_level_list_or_range bool - field_options last_field_options + field_options last_field_options system_or_bernoully %type ulonglong_num real_ulonglong_num @@ -1511,7 +1514,10 @@ bool my_yyoverflow(short **a, YYSTYPE **b, size_t *yystacksize); %type param_marker %type - NUM_literal opt_table_sample + NUM_literal + +%type + opt_table_sample %type text_literal @@ -11754,13 +11760,26 @@ join_table_parens: } ; +system_or_bernoully: + SYSTEM { $$= 1; } + | BERNOULLI_SYM { $$= 0; } + ; + /* psergey */ opt_table_sample: - /* empty */ { $$=0; } - | TABLESAMPLE SYSTEM '(' NUM_literal ')' - { - $$=$4; - } + /* empty */ + { + $$= 0; + } + | TABLESAMPLE system_or_bernoully '(' NUM_literal ')' + { + if ($2) + $$= new (thd->mem_root) + System_staticitc_collector($4->val_real()/100.0); + else + $$= new (thd->mem_root) + Bernoulli_staticitc_collector($4->val_real()/100.0); + } ; table_primary_ident: @@ -11778,7 +11797,9 @@ table_primary_ident: if ($3) $$->vers_conditions= Lex->vers_conditions; if ($6) + { $$->tablesample= $6; + } } ; @@ -15728,6 +15749,7 @@ keyword_sp_var_and_label: | AUTO_SYM | AVG_ROW_LENGTH | AVG_SYM + | BERNOULLI_SYM | BLOCK_SYM | BODY_MARIADB_SYM | BTREE_SYM diff --git a/sql/table.h b/sql/table.h index c944ac076cb..35751fd058c 100644 --- a/sql/table.h +++ b/sql/table.h @@ -65,6 +65,7 @@ class Pushdown_derived; struct Name_resolution_context; class Table_function_json_table; class Open_table_context; +class Statistic_collector; /* Used to identify NESTED_JOIN structures within a join (applicable only to @@ -1516,7 +1517,7 @@ struct TABLE /* used in RBR Triggers */ bool master_had_triggers; #endif - Item *tablesample; + Statistic_collector *tablesample; REGINFO reginfo; /* field connections */ MEM_ROOT mem_root; /** @@ -2267,7 +2268,7 @@ struct TABLE_LIST *last_ptr= &next_global; for_insert_data= insert_data; } - Item *tablesample; + Statistic_collector *tablesample; /* List of tables local to a subquery (used by SQL_I_List). Considers