database/mariadb-columnstore-engine
1
0
Fork 0
mirror of https://github.com/mariadb-corporation/mariadb-columnstore-engine.git synced 2025-07-29 08:21:15 +03:00
Code Activity

MCOL-5328: use PCRE2 and JPCRE wrapper (#3137)

Browse Source 
PCRE2 for regexp functions in columnstore
This commit is contained in:
Leonid Fedorov
2024-03-14 18:39:29 +03:00
committed by GitHub
parent c01e1f4ed8
commit 5f40fb32d0
18 changed files with 6112 additions and 66 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
.drone.jsonnet
View File

@ -125,13 +125,13 @@ local testRun(platform) =
local testPreparation(platform) = local testPreparation(platform) =
local platform_map = { local platform_map = {
'centos:7': 'yum -y install epel-release && yum install -y git cppunit-devel cmake3 boost-devel snappy-devel', 'centos:7': 'yum -y install epel-release && yum install -y git cppunit-devel cmake3 boost-devel snappy-devel pcre2-devel',
'rockylinux:8': rockylinux8_build_deps + ' && dnf install -y git lz4 cppunit-devel cmake3 boost-devel snappy-devel', 'rockylinux:8': rockylinux8_build_deps + ' && dnf install -y git lz4 cppunit-devel cmake3 boost-devel snappy-devel pcre2-devel',
'rockylinux:9': rockylinux9_build_deps + ' && dnf install -y git lz4 cppunit-devel cmake3 boost-devel snappy-devel', 'rockylinux:9': rockylinux9_build_deps + ' && dnf install -y git lz4 cppunit-devel cmake3 boost-devel snappy-devel pcre2-devel',
'debian:11': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake', 'debian:11': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake libpcre2-dev',
'debian:12': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake', 'debian:12': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake libpcre2-dev',
'ubuntu:20.04': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake', 'ubuntu:20.04': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake libpcre2-dev',
'ubuntu:22.04': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake', 'ubuntu:22.04': 'apt update && apt install --yes git libboost-all-dev libcppunit-dev libsnappy-dev cmake libpcre2-dev',
}; };
platform_map[platform]; platform_map[platform];

47
build/bootstrap_mcs.sh
View File

@ -64,10 +64,12 @@ INSTALL_PREFIX="/usr/"
DATA_DIR="/var/lib/mysql/data" DATA_DIR="/var/lib/mysql/data"
CMAKE_BIN_NAME=cmake CMAKE_BIN_NAME=cmake
CTEST_BIN_NAME=ctest CTEST_BIN_NAME=ctest
CONFIG_DIR="/etc/my.cnf.d" RPM_CONFIG_DIR="/etc/my.cnf.d"
DEB_CONFIG_DIR="/etc/mysql/mariadb.conf.d"
CONFIG_DIR=$RPM_CONFIG_DIR
if [[ $OS = 'Ubuntu' || $OS = 'Debian' ]]; then if [[ $OS = 'Ubuntu' || $OS = 'Debian' ]]; then
CONFIG_DIR="/etc/mysql/mariadb.conf.d" CONFIG_DIR=$DEB_CONFIG_DIR
fi fi
export CLICOLOR_FORCE=1 export CLICOLOR_FORCE=1
@ -202,6 +204,8 @@ clean_old_installation()
rm -rf /var/run/mysqld rm -rf /var/run/mysqld
rm -rf $DATA_DIR rm -rf $DATA_DIR
rm -rf /etc/mysql rm -rf /etc/mysql
rm -rf /etc/my.cnf.d/columnstore.cnf
rm -rf /etc/mysql/mariadb.conf.d/columnstore.cnf
} }
build() build()
@ -442,25 +446,30 @@ fix_config_files()
systemctl daemon-reload systemctl daemon-reload
} }
make_dir()
{
mkdir -p $1
chown mysql:mysql $1
}
install() install()
{ {
message_split message_split
message "Installing MariaDB" message "Installing MariaDB"
disable_plugins_for_bootstrap disable_plugins_for_bootstrap
mkdir -p $REPORT_PATH make_dir $REPORT_PATH
chmod 777 $REPORT_PATH chmod 777 $REPORT_PATH
check_user_and_group check_user_and_group
mkdir -p /etc/my.cnf.d make_dir $CONFIG_DIR
bash -c 'echo "[client-server] echo "[client-server]
socket=/run/mysqld/mysqld.sock" > /etc/my.cnf.d/socket.cnf' socket=/run/mysqld/mysqld.sock" > $CONFIG_DIR/socket.cnf
mv $INSTALL_PREFIX/lib/mysql/plugin/ha_columnstore.so /tmp/ha_columnstore_1.so || mv $INSTALL_PREFIX/lib64/mysql/plugin/ha_columnstore.so /tmp/ha_columnstore_2.so mv $INSTALL_PREFIX/lib/mysql/plugin/ha_columnstore.so /tmp/ha_columnstore_1.so || mv $INSTALL_PREFIX/lib64/mysql/plugin/ha_columnstore.so /tmp/ha_columnstore_2.so
mkdir -p /var/lib/mysql make_dir /var/lib/mysql
chown mysql:mysql /var/lib/mysql
message "Running mysql_install_db" message "Running mysql_install_db"
sudo -u mysql mysql_install_db --rpm --user=mysql > /dev/null sudo -u mysql mysql_install_db --rpm --user=mysql > /dev/null
@ -468,7 +477,7 @@ socket=/run/mysqld/mysqld.sock" > /etc/my.cnf.d/socket.cnf'
enable_columnstore_back enable_columnstore_back
mkdir -p /etc/columnstore make_dir /etc/columnstore
cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/oam/etc/Columnstore.xml /etc/columnstore/Columnstore.xml cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/oam/etc/Columnstore.xml /etc/columnstore/Columnstore.xml
cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/storage-manager/storagemanager.cnf /etc/columnstore/storagemanager.cnf cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/storage-manager/storagemanager.cnf /etc/columnstore/storagemanager.cnf
@ -477,8 +486,8 @@ socket=/run/mysqld/mysqld.sock" > /etc/my.cnf.d/socket.cnf'
cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/oam/install_scripts/*.service /lib/systemd/system/ cp $MDB_SOURCE_PATH/storage/columnstore/columnstore/oam/install_scripts/*.service /lib/systemd/system/
if [[ "$OS" = 'Ubuntu' || "$OS" = 'Debian' ]]; then if [[ "$OS" = 'Ubuntu' || "$OS" = 'Debian' ]]; then
mkdir -p /usr/share/mysql make_dir /usr/share/mysql
mkdir -p /etc/mysql/ make_dir /etc/mysql/
cp $MDB_SOURCE_PATH/debian/additions/debian-start.inc.sh /usr/share/mysql/debian-start.inc.sh cp $MDB_SOURCE_PATH/debian/additions/debian-start.inc.sh /usr/share/mysql/debian-start.inc.sh
cp $MDB_SOURCE_PATH/debian/additions/debian-start /etc/mysql/debian-start cp $MDB_SOURCE_PATH/debian/additions/debian-start /etc/mysql/debian-start
> /etc/mysql/debian.cnf > /etc/mysql/debian.cnf
@ -496,15 +505,11 @@ socket=/run/mysqld/mysqld.sock" > /etc/my.cnf.d/socket.cnf'
cp -rp /etc/mysql/conf.d/* /etc/my.cnf.d cp -rp /etc/mysql/conf.d/* /etc/my.cnf.d
fi fi
mkdir -p /var/lib/columnstore/data1 make_dir /var/lib/columnstore/data1
mkdir -p /var/lib/columnstore/data1/systemFiles make_dir /var/lib/columnstore/data1/systemFiles
mkdir -p /var/lib/columnstore/data1/systemFiles/dbrm make_dir /var/lib/columnstore/data1/systemFiles/dbrm
mkdir -p /run/mysqld/ make_dir /run/mysqld/
make_dir $DATA_DIR
mkdir -p $DATA_DIR
chown -R mysql:mysql $DATA_DIR
chown -R mysql:mysql /var/lib/columnstore/
chown -R mysql:mysql /run/mysqld/
chmod +x $INSTALL_PREFIX/bin/mariadb* chmod +x $INSTALL_PREFIX/bin/mariadb*
@ -513,7 +518,7 @@ socket=/run/mysqld/mysqld.sock" > /etc/my.cnf.d/socket.cnf'
start_storage_manager_if_needed start_storage_manager_if_needed
message "Running columnstore-post-install" message "Running columnstore-post-install"
mkdir -p /var/lib/columnstore/local make_dir /var/lib/columnstore/local
columnstore-post-install --rpmmode=install columnstore-post-install --rpmmode=install
message "Running install_mcs_mysql" message "Running install_mcs_mysql"
install_mcs_mysql.sh install_mcs_mysql.sh

2
dbcon/mysql/ha_mcs_execplan.cpp
View File

@ -4387,7 +4387,7 @@ ReturnedColumn* buildFunctionColumn(Item_func* ifp, gp_walk_info& gwi, bool& non
// A few functions use a different collation than that found in // A few functions use a different collation than that found in
// the base ifp class // the base ifp class
if (funcName == "locate" || funcName == "find_in_set" || funcName == "strcmp") if (funcName == "locate" || funcName == "find_in_set" || funcName == "strcmp" || funcName == "regexp_instr")
{ {
DTCollation dt; DTCollation dt;
ifp->Type_std_attributes::agg_arg_charsets_for_comparison(dt, ifp->func_name_cstring(), ifp->Type_std_attributes::agg_arg_charsets_for_comparison(dt, ifp->func_name_cstring(),

2
dmlproc/dmlprocessor.cpp
View File

@ -67,7 +67,7 @@ extern boost::condition_variable cond;
namespace namespace
{ {
const std::string myname = "DMLProc"; [[maybe_unused]] const std::string myname = "DMLProc";
} }
namespace dmlprocessor namespace dmlprocessor

522
mysql-test/columnstore/basic/r/mcs228_regexp_operator.result
View File

@ -42,6 +42,96 @@ t1_INT t1_INT REGEXP '-26'
-7299 0 -7299 0
103 0 103 0
9913 0 9913 0
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '99$') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '99$')
-7299 99
103
9913
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '\.99$') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '\.99$')
-7299 299
103
9913
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '03$') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '03$')
-7299
103 03
9913
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '\.+') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '\.+')
-7299 -7299
103 103
9913 9913
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '0?') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '0?')
-7299
103
9913
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '-26') FROM t1 ORDER BY 1;
t1_INT REGEXP_SUBSTR(t1_INT, '-26')
-7299
103
9913
SELECT t1_INT, REGEXP_INSTR(t1_INT, '99$') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '99$')
-7299 4
103 0
9913 0
SELECT t1_INT, REGEXP_INSTR(t1_INT, '\.99$') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '\.99$')
-7299 3
103 0
9913 0
SELECT t1_INT, REGEXP_INSTR(t1_INT, '03$') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '03$')
-7299 0
103 2
9913 0
SELECT t1_INT, REGEXP_INSTR(t1_INT, '\.+') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '\.+')
-7299 1
103 1
9913 1
SELECT t1_INT, REGEXP_INSTR(t1_INT, '0?') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '0?')
-7299 1
103 1
9913 1
SELECT t1_INT, REGEXP_INSTR(t1_INT, '-26') FROM t1 ORDER BY 1;
t1_INT REGEXP_INSTR(t1_INT, '-26')
-7299 0
103 0
9913 0
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '99$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '99$', 'TeddyBear')
-7299 -72TeddyBear
103 103
9913 9913
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '\.99$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '\.99$', 'TeddyBear')
-7299 -7TeddyBear
103 103
9913 9913
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '03$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '03$', 'TeddyBear')
-7299 -7299
103 1TeddyBear
9913 9913
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '\.+', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '\.+', 'TeddyBear')
-7299 TeddyBear
103 TeddyBear
9913 TeddyBear
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '0?', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '0?', 'TeddyBear')
-7299 TeddyBear-TeddyBear7TeddyBear2TeddyBear9TeddyBear9TeddyBear
103 TeddyBear1TeddyBearTeddyBear3TeddyBear
9913 TeddyBear9TeddyBear9TeddyBear1TeddyBear3TeddyBear
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '-26', 'TeddyBear') FROM t1 ORDER BY 1;
t1_INT REGEXP_REPLACE(t1_INT, '-26', 'TeddyBear')
-7299 -7299
103 103
9913 9913
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '99$' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '99$' FROM t1 ORDER BY 1;
t1_DECIMAL t1_DECIMAL REGEXP '99$' t1_DECIMAL t1_DECIMAL REGEXP '99$'
111.99000 0 111.99000 0
@ -72,6 +162,96 @@ t1_DECIMAL t1_DECIMAL REGEXP '-26'
111.99000 0 111.99000 0
1234.56990 0 1234.56990 0
98765.43210 0 98765.43210 0
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '99$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '99$')
111.99000
1234.56990
98765.43210
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '\.99$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '\.99$')
111.99000
1234.56990
98765.43210
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '03$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '03$')
111.99000
1234.56990
98765.43210
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '\.+') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '\.+')
111.99000 111.99000
1234.56990 1234.56990
98765.43210 98765.43210
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '0?') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '0?')
111.99000
1234.56990
98765.43210
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '-26') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_SUBSTR(t1_DECIMAL, '-26')
111.99000
1234.56990
98765.43210
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '99$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '99$')
111.99000 0
1234.56990 0
98765.43210 0
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '\.99$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '\.99$')
111.99000 0
1234.56990 0
98765.43210 0
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '03$') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '03$')
111.99000 0
1234.56990 0
98765.43210 0
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '\.+') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '\.+')
111.99000 1
1234.56990 1
98765.43210 1
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '0?') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '0?')
111.99000 1
1234.56990 1
98765.43210 1
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '-26') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_INSTR(t1_DECIMAL, '-26')
111.99000 0
1234.56990 0
98765.43210 0
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '99$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '99$', 'TeddyBear')
111.99000 111.99000
1234.56990 1234.56990
98765.43210 98765.43210
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '\.99$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '\.99$', 'TeddyBear')
111.99000 111.99000
1234.56990 1234.56990
98765.43210 98765.43210
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '03$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '03$', 'TeddyBear')
111.99000 111.99000
1234.56990 1234.56990
98765.43210 98765.43210
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '\.+', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '\.+', 'TeddyBear')
111.99000 TeddyBear
1234.56990 TeddyBear
98765.43210 TeddyBear
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '0?', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '0?', 'TeddyBear')
111.99000 TeddyBear1TeddyBear1TeddyBear1TeddyBear.TeddyBear9TeddyBear9TeddyBearTeddyBearTeddyBearTeddyBear
1234.56990 TeddyBear1TeddyBear2TeddyBear3TeddyBear4TeddyBear.TeddyBear5TeddyBear6TeddyBear9TeddyBear9TeddyBearTeddyBear
98765.43210 TeddyBear9TeddyBear8TeddyBear7TeddyBear6TeddyBear5TeddyBear.TeddyBear4TeddyBear3TeddyBear2TeddyBear1TeddyBearTeddyBear
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '-26', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DECIMAL REGEXP_REPLACE(t1_DECIMAL, '-26', 'TeddyBear')
111.99000 111.99000
1234.56990 1234.56990
98765.43210 98765.43210
SELECT t1_TEXT, t1_TEXT REGEXP 'oooo$' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'oooo$' FROM t1 ORDER BY 1;
t1_TEXT t1_TEXT REGEXP 'oooo$' t1_TEXT t1_TEXT REGEXP 'oooo$'
ooooo 1 ooooo 1
@ -112,6 +292,131 @@ t1_TEXT t1_TEXT REGEXP 'o?o'
ooooo 1 ooooo 1
ppppp 0 ppppp 0
qqqqq 0 qqqqq 0
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'oooo$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'oooo$')
ooooo oooo
ppppp
qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'q$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'q$')
ooooo
ppppp
qqqqq q
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'ppppp$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'ppppp$')
ooooo
ppppp ppppp
qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, '(ooo)+') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, '(ooo)+')
ooooo ooo
ppppp
qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, '(qqqqq)+') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, '(qqqqq)+')
ooooo
ppppp
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'p*') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'p*')
ooooo
ppppp ppppp
qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'qq+q') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'qq+q')
ooooo
ppppp
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'o?o') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_SUBSTR(t1_TEXT, 'o?o')
ooooo oo
ppppp
qqqqq
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'oooo$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'oooo$')
ooooo 2
ppppp 0
qqqqq 0
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'q$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'q$')
ooooo 0
ppppp 0
qqqqq 5
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'ppppp$') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'ppppp$')
ooooo 0
ppppp 1
qqqqq 0
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, '(ooo)+') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, '(ooo)+')
ooooo 1
ppppp 0
qqqqq 0
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, '(qqqqq)+') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, '(qqqqq)+')
ooooo 0
ppppp 0
qqqqq 1
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'p*') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'p*')
ooooo 1
ppppp 1
qqqqq 1
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'qq+q') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'qq+q')
ooooo 0
ppppp 0
qqqqq 1
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o?o') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_INSTR(t1_TEXT, 'o?o')
ooooo 1
ppppp 0
qqqqq 0
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'oooo$', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'oooo$', 'KittyCat')
ooooo oKittyCat
ppppp ppppp
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'q$', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'q$', 'KittyCat')
ooooo ooooo
ppppp ppppp
qqqqq qqqqKittyCat
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'ppppp$', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'ppppp$', 'KittyCat')
ooooo ooooo
ppppp KittyCat
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, '(ooo)+', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, '(ooo)+', 'KittyCat')
ooooo KittyCatoo
ppppp ppppp
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, '(qqqqq)+', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, '(qqqqq)+', 'KittyCat')
ooooo ooooo
ppppp ppppp
qqqqq KittyCat
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'p*', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'p*', 'KittyCat')
ooooo KittyCatoKittyCatoKittyCatoKittyCatoKittyCatoKittyCat
ppppp KittyCatKittyCat
qqqqq KittyCatqKittyCatqKittyCatqKittyCatqKittyCatqKittyCat
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'qq+q', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'qq+q', 'KittyCat')
ooooo ooooo
ppppp ppppp
qqqqq KittyCat
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'o?o', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'o?o', 'KittyCat')
ooooo KittyCatKittyCatKittyCat
ppppp ppppp
qqqqq qqqqq
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'q', 'KittyCat') FROM t1 ORDER BY 1;
t1_TEXT REGEXP_REPLACE(t1_TEXT, 'q', 'KittyCat')
ooooo ooooo
ppppp ppppp
qqqqq KittyCatKittyCatKittyCatKittyCatKittyCat
SELECT t1_DATE, t1_DATE REGEXP '(1997)+' FROM t1 ORDER BY 1; SELECT t1_DATE, t1_DATE REGEXP '(1997)+' FROM t1 ORDER BY 1;
t1_DATE t1_DATE REGEXP '(1997)+' t1_DATE t1_DATE REGEXP '(1997)+'
1997-12-12 1 1997-12-12 1
@ -132,6 +437,66 @@ t1_DATE t1_DATE REGEXP '(09-12-)+'
1997-12-12 0 1997-12-12 0
2001-01-01 0 2001-01-01 0
2009-12-11 1 2009-12-11 1
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(1997)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_SUBSTR(t1_DATE, '(1997)+')
1997-12-12 1997
2001-01-01
2009-12-11
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '01$') FROM t1 ORDER BY 1;
t1_DATE REGEXP_SUBSTR(t1_DATE, '01$')
1997-12-12
2001-01-01 01
2009-12-11
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(59)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_SUBSTR(t1_DATE, '(59)+')
1997-12-12
2001-01-01
2009-12-11
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(09-12-)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_SUBSTR(t1_DATE, '(09-12-)+')
1997-12-12
2001-01-01
2009-12-11 09-12-
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(1997)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_INSTR(t1_DATE, '(1997)+')
1997-12-12 1
2001-01-01 0
2009-12-11 0
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '01$') FROM t1 ORDER BY 1;
t1_DATE REGEXP_INSTR(t1_DATE, '01$')
1997-12-12 0
2001-01-01 9
2009-12-11 0
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(59)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_INSTR(t1_DATE, '(59)+')
1997-12-12 0
2001-01-01 0
2009-12-11 0
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(09-12-)+') FROM t1 ORDER BY 1;
t1_DATE REGEXP_INSTR(t1_DATE, '(09-12-)+')
1997-12-12 0
2001-01-01 0
2009-12-11 3
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(1997)+', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DATE REGEXP_REPLACE(t1_DATE, '(1997)+', 'TeddyBear')
1997-12-12 TeddyBear-12-12
2001-01-01 2001-01-01
2009-12-11 2009-12-11
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '01$', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DATE REGEXP_REPLACE(t1_DATE, '01$', 'TeddyBear')
1997-12-12 1997-12-12
2001-01-01 2001-01-TeddyBear
2009-12-11 2009-12-11
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(59)+', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DATE REGEXP_REPLACE(t1_DATE, '(59)+', 'TeddyBear')
1997-12-12 1997-12-12
2001-01-01 2001-01-01
2009-12-11 2009-12-11
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(09-12-)+', 'TeddyBear') FROM t1 ORDER BY 1;
t1_DATE REGEXP_REPLACE(t1_DATE, '(09-12-)+', 'TeddyBear')
1997-12-12 1997-12-12
2001-01-01 2001-01-01
2009-12-11 20TeddyBear11
SELECT t1_TIME, t1_TIME REGEXP '(59)+' FROM t1 ORDER BY 1; SELECT t1_TIME, t1_TIME REGEXP '(59)+' FROM t1 ORDER BY 1;
t1_TIME t1_TIME REGEXP '(59)+' t1_TIME t1_TIME REGEXP '(59)+'
01:08:59 1 01:08:59 1
@ -142,4 +507,161 @@ t1_TIME t1_TIME REGEXP '22$'
01:08:59 0 01:08:59 0
22:12:02 0 22:12:02 0
23:59:59 0 23:59:59 0
SELECT t1_TIME, REGEXP_SUBSTR(t1_TIME, '(59)+') FROM t1 ORDER BY 1;
t1_TIME REGEXP_SUBSTR(t1_TIME, '(59)+')
01:08:59 59
22:12:02
23:59:59 59
SELECT t1_TIME, REGEXP_SUBSTR(t1_TIME, '22$') FROM t1 ORDER BY 1;
t1_TIME REGEXP_SUBSTR(t1_TIME, '22$')
01:08:59
22:12:02
23:59:59
SELECT t1_TIME, REGEXP_INSTR(t1_TIME, '(59)+') FROM t1 ORDER BY 1;
t1_TIME REGEXP_INSTR(t1_TIME, '(59)+')
01:08:59 7
22:12:02 0
23:59:59 4
SELECT t1_TIME, REGEXP_INSTR(t1_TIME, '22$') FROM t1 ORDER BY 1;
t1_TIME REGEXP_INSTR(t1_TIME, '22$')
01:08:59 0
22:12:02 0
23:59:59 0
SELECT t1_TIME, REGEXP_REPLACE(t1_TIME, '(59)+', 'KittyCat') FROM t1 ORDER BY 1;
t1_TIME REGEXP_REPLACE(t1_TIME, '(59)+', 'KittyCat')
01:08:59 01:08:KittyCat
22:12:02 22:12:02
23:59:59 23:KittyCat:KittyCat
SELECT t1_TIME, REGEXP_REPLACE(t1_TIME, '22$', 'KittyCat') FROM t1 ORDER BY 1;
t1_TIME REGEXP_REPLACE(t1_TIME, '22$', 'KittyCat')
01:08:59 01:08:59
22:12:02 22:12:02
23:59:59 23:59:59
SET character_set_connection = 'utf8';
SET NAMES utf8mb3;
CREATE TABLE t2 (hello text) DEFAULT CHARSET=utf8 engine columnstore;
INSERT INTO t2 values('こんにちは');
INSERT INTO t2 values('привет');
INSERT INTO t2 values('Γεια');
INSERT INTO t2 values('სალამი');
SELECT hello, hello regexp 'ん.ち' FROM t2;
hello hello regexp 'ん.ち'
こんにちは 0
привет 0
Γεια 0
სალამი 0
SELECT hello, hello regexp 'и.е' FROM t2;
hello hello regexp 'и.е'
こんにちは 0
привет 0
Γεια 0
სალამი 0
SELECT hello, hello regexp 'ε.α' FROM t2;
hello hello regexp 'ε.α'
こんにちは 0
привет 0
Γεια 0
სალამი 0
SELECT hello, hello regexp 'ა.ა' FROM t2;
hello hello regexp 'ა.ა'
こんにちは 0
привет 0
Γεια 0
სალამი 0
SELECT hello, regexp_substr(hello, 'ん.ち') FROM t2;
hello regexp_substr(hello, 'ん.ち')
こんにちは んにち
привет
Γεια
სალამი
SELECT hello, regexp_substr(hello, 'и.е') FROM t2;
hello regexp_substr(hello, 'и.е')
こんにちは
привет иве
Γεια
სალამი
SELECT hello, regexp_substr(hello, 'ε.α') FROM t2;
hello regexp_substr(hello, 'ε.α')
こんにちは
привет
Γεια εια
სალამი
SELECT hello, regexp_substr(hello, 'ა.ა') FROM t2;
hello regexp_substr(hello, 'ა.ა')
こんにちは
привет
Γεια
სალამი ალა
SELECT hello, regexp_instr(hello, 'ん.ち') FROM t2;
hello regexp_instr(hello, 'ん.ち')
こんにちは 2
привет 0
Γεια 0
სალამი 0
SELECT hello, regexp_instr(hello, 'и.е') FROM t2;
hello regexp_instr(hello, 'и.е')
こんにちは 0
привет 3
Γεια 0
სალამი 0
SELECT hello, regexp_instr(hello, 'ε.α') FROM t2;
hello regexp_instr(hello, 'ε.α')
こんにちは 0
привет 0
Γεια 2
სალამი 0
SELECT hello, regexp_instr(hello, 'ა.ა') FROM t2;
hello regexp_instr(hello, 'ა.ა')
こんにちは 0
привет 0
Γεια 0
სალამი 2
SELECT hello, regexp_replace(hello, 'ん.ち', 'Достоевский') FROM t2;
hello regexp_replace(hello, 'ん.ち', 'Достоевский')
こんにちは こДостоевскийは
привет привет
Γεια Γεια
სალამი სალამი
SELECT hello, regexp_replace(hello, 'и.е', 'Достоевский') FROM t2;
hello regexp_replace(hello, 'и.е', 'Достоевский')
こんにちは こんにちは
привет прДостоевскийт
Γεια Γεια
სალამი სალამი
SELECT hello, regexp_replace(hello, 'ε.α', 'Достоевский') FROM t2;
hello regexp_replace(hello, 'ε.α', 'Достоевский')
こんにちは こんにちは
привет привет
Γεια ΓДостоевский
სალამი სალამი
SELECT hello, regexp_replace(hello, 'ა.ა', 'Достоевский') FROM t2;
hello regexp_replace(hello, 'ა.ა', 'Достоевский')
こんにちは こんにちは
привет привет
Γεια Γεια
სალამი სДостоевскийმი
SHOW VARIABLES LIKE 'character_set%';
Variable_name Value
character_set_client utf8mb3
character_set_connection utf8mb3
character_set_database utf8mb3
character_set_filesystem binary
character_set_results utf8mb3
character_set_server utf8mb3
character_set_system utf8mb3
character_sets_dir /usr/share/mysql/charsets/
CREATE TABLE tw(hello text) DEFAULT CHARSET=cp1251 ENGINE COLUMNSTORE;
INSERT INTO tw values(convert('привет' USING cp1251));
SELECT hello, regexp_instr(hello, convert('и.е' USING cp1251)) FROM tw;
hello regexp_instr(hello, convert('и.е' USING cp1251))
привет 3
SELECT hello, convert(regexp_substr(hello, convert('и.е' USING cp1251)) using utf8) FROM tw;
hello convert(regexp_substr(hello, convert('и.е' USING cp1251)) using utf8)
привет иве
SELECT hello, convert(regexp_replace(hello, convert('и.е' USING cp1251), convert('Достоевкий' USING cp1251)) using utf8) FROM tw;
hello convert(regexp_replace(hello, convert('и.е' USING cp1251), convert('Достоевкий' USING cp1251)) using utf8)
привет привет
SELECT hello, hello regexp convert('и.е' USING cp1251) FROM tw;
hello hello regexp convert('и.е' USING cp1251)
привет 1
DROP DATABASE mcs228_db; DROP DATABASE mcs228_db;

5
mysql-test/columnstore/basic/r/mcs244_regexp_instr_function.result
View File

@ -13,5 +13,8 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o') FROM t1 ORDER BY 1; SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o') FROM t1 ORDER BY 1;
ERROR 42000: The storage engine for the table doesn't support MCS-1001: Function 'regexp_instr' isn't supported. t1_TEXT REGEXP_INSTR(t1_TEXT, 'o')
ooooo 1
ppppp 0
qqqqq 0
DROP DATABASE mcs244_db; DROP DATABASE mcs244_db;

5
mysql-test/columnstore/basic/r/mcs245_regexp_replace_function.result
View File

@ -13,5 +13,8 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
SELECT REGEXP_REPLACE(t1_TEXT, 'o', 'X') FROM t1 ORDER BY 1; SELECT REGEXP_REPLACE(t1_TEXT, 'o', 'X') FROM t1 ORDER BY 1;
ERROR 42000: The storage engine for the table doesn't support MCS-1001: Function 'regexp_replace' isn't supported. REGEXP_REPLACE(t1_TEXT, 'o', 'X')
ppppp
qqqqq
XXXXX
DROP DATABASE mcs245_db; DROP DATABASE mcs245_db;

5
mysql-test/columnstore/basic/r/mcs246_regexp_substr_function.result
View File

@ -13,5 +13,8 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
SELECT REGEXP_SUBSTR(t1_TEXT, '[a-z]+') FROM t1 ORDER BY 1; SELECT REGEXP_SUBSTR(t1_TEXT, '[a-z]+') FROM t1 ORDER BY 1;
ERROR 42000: The storage engine for the table doesn't support MCS-1001: Function 'regexp_substr' isn't supported. REGEXP_SUBSTR(t1_TEXT, '[a-z]+')
ooooo
ppppp
qqqqq
DROP DATABASE mcs246_db; DROP DATABASE mcs246_db;

134
mysql-test/columnstore/basic/t/mcs228_regexp_operator.test
View File

@ -30,6 +30,27 @@ SELECT t1_INT, t1_INT REGEXP '\.+' FROM t1 ORDER BY 1;
SELECT t1_INT, t1_INT REGEXP '0?' FROM t1 ORDER BY 1; SELECT t1_INT, t1_INT REGEXP '0?' FROM t1 ORDER BY 1;
SELECT t1_INT, t1_INT REGEXP '-26' FROM t1 ORDER BY 1; SELECT t1_INT, t1_INT REGEXP '-26' FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '99$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '\.99$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '03$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '\.+') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '0?') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_SUBSTR(t1_INT, '-26') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '99$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '\.99$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '03$') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '\.+') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '0?') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_INSTR(t1_INT, '-26') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '99$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '\.99$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '03$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '\.+', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '0?', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_INT, REGEXP_REPLACE(t1_INT, '-26', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '99$' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '99$' FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '\.99$' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '\.99$' FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '03$' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '03$' FROM t1 ORDER BY 1;
@ -37,6 +58,28 @@ SELECT t1_DECIMAL, t1_DECIMAL REGEXP '\.+' FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '0?' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '0?' FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, t1_DECIMAL REGEXP '-26' FROM t1 ORDER BY 1; SELECT t1_DECIMAL, t1_DECIMAL REGEXP '-26' FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '99$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '\.99$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '03$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '\.+') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '0?') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_SUBSTR(t1_DECIMAL, '-26') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '99$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '\.99$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '03$') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '\.+') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '0?') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_INSTR(t1_DECIMAL, '-26') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '99$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '\.99$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '03$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '\.+', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '0?', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DECIMAL, REGEXP_REPLACE(t1_DECIMAL, '-26', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_TEXT, t1_TEXT REGEXP 'oooo$' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'oooo$' FROM t1 ORDER BY 1;
SELECT t1_TEXT, t1_TEXT REGEXP 'q$' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'q$' FROM t1 ORDER BY 1;
SELECT t1_TEXT, t1_TEXT REGEXP 'ppppp$' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'ppppp$' FROM t1 ORDER BY 1;
@ -46,13 +89,104 @@ SELECT t1_TEXT, t1_TEXT REGEXP 'p*' FROM t1 ORDER BY 1;
SELECT t1_TEXT, t1_TEXT REGEXP 'qq+q' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'qq+q' FROM t1 ORDER BY 1;
SELECT t1_TEXT, t1_TEXT REGEXP 'o?o' FROM t1 ORDER BY 1; SELECT t1_TEXT, t1_TEXT REGEXP 'o?o' FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'oooo$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'q$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'ppppp$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, '(ooo)+') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, '(qqqqq)+') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'p*') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'qq+q') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_SUBSTR(t1_TEXT, 'o?o') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'oooo$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'q$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'ppppp$') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, '(ooo)+') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, '(qqqqq)+') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'p*') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'qq+q') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o?o') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'oooo$', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'q$', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'ppppp$', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, '(ooo)+', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, '(qqqqq)+', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'p*', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'qq+q', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'o?o', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TEXT, REGEXP_REPLACE(t1_TEXT, 'q', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_DATE, t1_DATE REGEXP '(1997)+' FROM t1 ORDER BY 1; SELECT t1_DATE, t1_DATE REGEXP '(1997)+' FROM t1 ORDER BY 1;
SELECT t1_DATE, t1_DATE REGEXP '01$' FROM t1 ORDER BY 1; SELECT t1_DATE, t1_DATE REGEXP '01$' FROM t1 ORDER BY 1;
SELECT t1_DATE, t1_DATE REGEXP '(59)+' FROM t1 ORDER BY 1; SELECT t1_DATE, t1_DATE REGEXP '(59)+' FROM t1 ORDER BY 1;
SELECT t1_DATE, t1_DATE REGEXP '(09-12-)+' FROM t1 ORDER BY 1; SELECT t1_DATE, t1_DATE REGEXP '(09-12-)+' FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(1997)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '01$') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(59)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_SUBSTR(t1_DATE, '(09-12-)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(1997)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '01$') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(59)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_INSTR(t1_DATE, '(09-12-)+') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(1997)+', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '01$', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(59)+', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_DATE, REGEXP_REPLACE(t1_DATE, '(09-12-)+', 'TeddyBear') FROM t1 ORDER BY 1;
SELECT t1_TIME, t1_TIME REGEXP '(59)+' FROM t1 ORDER BY 1; SELECT t1_TIME, t1_TIME REGEXP '(59)+' FROM t1 ORDER BY 1;
SELECT t1_TIME, t1_TIME REGEXP '22$' FROM t1 ORDER BY 1; SELECT t1_TIME, t1_TIME REGEXP '22$' FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_SUBSTR(t1_TIME, '(59)+') FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_SUBSTR(t1_TIME, '22$') FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_INSTR(t1_TIME, '(59)+') FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_INSTR(t1_TIME, '22$') FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_REPLACE(t1_TIME, '(59)+', 'KittyCat') FROM t1 ORDER BY 1;
SELECT t1_TIME, REGEXP_REPLACE(t1_TIME, '22$', 'KittyCat') FROM t1 ORDER BY 1;
SET character_set_connection = 'utf8';
SET NAMES utf8mb3;
CREATE TABLE t2 (hello text) DEFAULT CHARSET=utf8 engine columnstore;
INSERT INTO t2 values('こんにちは');
INSERT INTO t2 values('привет');
INSERT INTO t2 values('Γεια');
INSERT INTO t2 values('სალამი');
SELECT hello, hello regexp 'ん.ち' FROM t2;
SELECT hello, hello regexp 'и.е' FROM t2;
SELECT hello, hello regexp 'ε.α' FROM t2;
SELECT hello, hello regexp 'ა.ა' FROM t2;
SELECT hello, regexp_substr(hello, 'ん.ち') FROM t2;
SELECT hello, regexp_substr(hello, 'и.е') FROM t2;
SELECT hello, regexp_substr(hello, 'ε.α') FROM t2;
SELECT hello, regexp_substr(hello, 'ა.ა') FROM t2;
SELECT hello, regexp_instr(hello, 'ん.ち') FROM t2;
SELECT hello, regexp_instr(hello, 'и.е') FROM t2;
SELECT hello, regexp_instr(hello, 'ε.α') FROM t2;
SELECT hello, regexp_instr(hello, 'ა.ა') FROM t2;
SELECT hello, regexp_replace(hello, 'ん.ち', 'Достоевский') FROM t2;
SELECT hello, regexp_replace(hello, 'и.е', 'Достоевский') FROM t2;
SELECT hello, regexp_replace(hello, 'ε.α', 'Достоевский') FROM t2;
SELECT hello, regexp_replace(hello, 'ა.ა', 'Достоевский') FROM t2;
SHOW VARIABLES LIKE 'character_set%';
CREATE TABLE tw(hello text) DEFAULT CHARSET=cp1251 ENGINE COLUMNSTORE;
INSERT INTO tw values(convert('привет' USING cp1251));
SELECT hello, regexp_instr(hello, convert('и.е' USING cp1251)) FROM tw;
SELECT hello, convert(regexp_substr(hello, convert('и.е' USING cp1251)) using utf8) FROM tw;
SELECT hello, convert(regexp_replace(hello, convert('и.е' USING cp1251), convert('Достоевкий' USING cp1251)) using utf8) FROM tw;
SELECT hello, hello regexp convert('и.е' USING cp1251) FROM tw;
# Clean UP # Clean UP
DROP DATABASE mcs228_db; DROP DATABASE mcs228_db;

1
mysql-test/columnstore/basic/t/mcs244_regexp_instr_function.test
View File

@ -23,7 +23,6 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
--error ER_CHECK_NOT_IMPLEMENTED
SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o') FROM t1 ORDER BY 1; SELECT t1_TEXT, REGEXP_INSTR(t1_TEXT, 'o') FROM t1 ORDER BY 1;
# Clean UP # Clean UP

1
mysql-test/columnstore/basic/t/mcs245_regexp_replace_function.test
View File

@ -23,7 +23,6 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
--error ER_CHECK_NOT_IMPLEMENTED
SELECT REGEXP_REPLACE(t1_TEXT, 'o', 'X') FROM t1 ORDER BY 1; SELECT REGEXP_REPLACE(t1_TEXT, 'o', 'X') FROM t1 ORDER BY 1;
# Clean UP # Clean UP

1
mysql-test/columnstore/basic/t/mcs246_regexp_substr_function.test
View File

@ -23,7 +23,6 @@ INSERT INTO t1 VALUES(103, 1234.5699, repeat('o', 5), '1997-12-12', '22:12:02');
INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59'); INSERT INTO t1 VALUES(-7299, 111.99, repeat('p', 5), '2001-1-1', '23:59:59');
INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59'); INSERT INTO t1 VALUES(9913, 98765.4321, repeat('q', 5), '09-12-11', '01:08:59');
--error ER_CHECK_NOT_IMPLEMENTED
SELECT REGEXP_SUBSTR(t1_TEXT, '[a-z]+') FROM t1 ORDER BY 1; SELECT REGEXP_SUBSTR(t1_TEXT, '[a-z]+') FROM t1 ORDER BY 1;
# Clean UP # Clean UP

17
utils/common/collation.h
View File

@ -146,6 +146,23 @@ class Charset
Charset(CHARSET_INFO* cs = nullptr) : mCharset(cs ? cs : &my_charset_bin) Charset(CHARSET_INFO* cs = nullptr) : mCharset(cs ? cs : &my_charset_bin)
{ {
} }
bool operator==(const Charset& rhs)
{
return rhs.getCharset().cs_name.str == getCharset().cs_name.str;
}
std::string convert(const std::string& from, const datatypes::Charset& fromCs) const
{
std::string result;
uint dummy_errors;
result.resize(from.size() * getCharset().mbmaxlen);
size_t resultingSize = my_convert(const_cast<char*>(result.c_str()), result.size(), &getCharset(), from.c_str(),
from.size(), &fromCs.getCharset(), &dummy_errors);
result.resize(resultingSize);
return result;
}
Charset(uint32_t charsetNumber); Charset(uint32_t charsetNumber);
void setCharset(uint32_t charsetNumber); void setCharset(uint32_t charsetNumber);
CHARSET_INFO& getCharset() const CHARSET_INFO& getCharset() const

2
utils/funcexp/CMakeLists.txt
View File

@ -150,7 +150,7 @@ add_library(funcexp SHARED ${funcexp_LIB_SRCS})
add_dependencies(funcexp loggingcpp) add_dependencies(funcexp loggingcpp)
target_link_libraries(funcexp ${NETSNMP_LIBRARIES} pron) target_link_libraries(funcexp ${NETSNMP_LIBRARIES} pron pcre2-8)
install(TARGETS funcexp DESTINATION ${ENGINE_LIBDIR} COMPONENT columnstore-engine) install(TARGETS funcexp DESTINATION ${ENGINE_LIBDIR} COMPONENT columnstore-engine)

207
utils/funcexp/func_regexp.cpp
View File

@ -25,14 +25,10 @@
#include <string> #include <string>
using namespace std; using namespace std;
#ifdef __linux__ #include "utils/pcre2/jpcre2.hpp"
#include <regex.h>
#else
#include <regex>
using namespace boost;
#endif
#include "functor_bool.h" #include "functor_bool.h"
#include "functor_str.h"
#include "functioncolumn.h" #include "functioncolumn.h"
#include "predicateoperator.h" #include "predicateoperator.h"
#include "constantcolumn.h" #include "constantcolumn.h"
@ -43,12 +39,58 @@ using namespace execplan;
#include "errorcodes.h" #include "errorcodes.h"
#include "idberrorinfo.h" #include "idberrorinfo.h"
#include "errorids.h" #include "errorids.h"
using namespace logging; using namespace logging;
namespace namespace
{ {
inline bool getBool(rowgroup::Row& row, funcexp::FunctionParm& pm, bool& isNull,
CalpontSystemCatalog::ColType& ct, long timeZone) using jp = jpcre2::select<char>;
struct PCREOptions
{
PCREOptions(execplan::CalpontSystemCatalog::ColType& ct);
datatypes::Charset dataCharset = my_charset_utf8mb3_general_ci;
datatypes::Charset libraryCharset = my_charset_utf8mb3_general_ci;
jpcre2::Uint flags = 0;
bool conversionIsNeeded = false;
};
PCREOptions::PCREOptions(execplan::CalpontSystemCatalog::ColType& ct)
{
datatypes::Charset cs = ct.getCharset();
// TODO use system variable instead if hardcode default_regex_flags_pcre(_current_thd());
// PCRE2_DOTALL | PCRE2_DUPNAMES | PCRE2_EXTENDED | PCRE2_EXTENDED_MORE | PCRE2_MULTILINE | PCRE2_UNGREEDY;
jpcre2::Uint defaultFlags = 0;
flags = (cs != &my_charset_bin ? (PCRE2_UTF | PCRE2_UCP) : 0) |
((cs.getCharset().state & (MY_CS_BINSORT | MY_CS_CSSORT)) ? 0 : PCRE2_CASELESS) | defaultFlags;
// Convert text data to utf-8.
dataCharset = cs;
libraryCharset = cs == my_charset_bin ? my_charset_bin : my_charset_utf8mb3_general_ci;
}
struct RegExpParams
{
std::string expression;
std::string pattern;
RegExpParams& CharsetFix(const PCREOptions options)
{
if (options.conversionIsNeeded)
return *this;
expression = options.libraryCharset.convert(expression, options.dataCharset);
pattern = options.libraryCharset.convert(pattern, options.dataCharset);
return *this;
}
};
inline RegExpParams getEpressionAndPattern(rowgroup::Row& row, funcexp::FunctionParm& pm, bool& isNull,
CalpontSystemCatalog::ColType& ct, long timeZone)
{ {
string expr; string expr;
string pattern; string pattern;
@ -212,39 +254,148 @@ inline bool getBool(rowgroup::Row& row, funcexp::FunctionParm& pm, bool& isNull,
} }
} }
#ifdef __linux__ return RegExpParams{expr, pattern};
regex_t re;
regcomp(&re, pattern.c_str(), REG_EXTENDED | REG_NOSUB);
int res = regexec(&re, expr.c_str(), 0, NULL, 0);
regfree(&re);
if (res == 0)
return true;
else
return false;
#else
std::regex pat(pattern.c_str());
return std::regex_search(expr.c_str(), pat);
#endif
} }
} // namespace } // namespace
namespace funcexp namespace funcexp
{ {
CalpontSystemCatalog::ColType Func_regexp_replace::operationType(FunctionParm& fp,
CalpontSystemCatalog::ColType& resultType)
{
// operation type is not used by this functor
return fp[0]->data()->resultType();
}
CalpontSystemCatalog::ColType Func_regexp_substr::operationType(FunctionParm& fp,
CalpontSystemCatalog::ColType& resultType)
{
// operation type is not used by this functor
return fp[0]->data()->resultType();
}
CalpontSystemCatalog::ColType Func_regexp_instr::operationType(FunctionParm& fp,
CalpontSystemCatalog::ColType& resultType)
{
// operation type is not used by this functor
return fp[0]->data()->resultType();
}
CalpontSystemCatalog::ColType Func_regexp::operationType(FunctionParm& fp, CalpontSystemCatalog::ColType Func_regexp::operationType(FunctionParm& fp,
CalpontSystemCatalog::ColType& resultType) CalpontSystemCatalog::ColType& resultType)
{ {
return resultType; return resultType;
} }
bool Func_regexp::getBoolVal(rowgroup::Row& row, FunctionParm& pm, bool& isNull, /*
returns the string subject with all occurrences of the regular expression pattern replaced by
the string replace. If no occurrences are found, then subject is returned as is.
https://mariadb.com/kb/en/regexp_replace/
*/
std::string Func_regexp_replace::getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& ct)
{
RegExpParams param = getEpressionAndPattern(row, fp, isNull, ct, ct.getTimeZone());
if (isNull)
return std::string{};
const auto& replaceWith = fp[2]->data()->getStrVal(row, isNull);
if (replaceWith.isNull())
return param.expression;
PCREOptions options(ct);
param.CharsetFix(options);
jp::Regex re(param.pattern, options.flags);
const auto& replaceWithStr = replaceWith.unsafeStringRef();
if (options.conversionIsNeeded)
{
const auto& convertedReplaceToken = options.libraryCharset.convert(replaceWithStr, options.dataCharset);
return re.replace(param.expression, convertedReplaceToken, "g");
}
return re.replace(param.expression, replaceWithStr, "g");
}
/*
Returns the part of the string subject that matches the regular expression pattern, or an empty string if
pattern was not found. https://mariadb.com/kb/en/regexp_substr/
*/
std::string Func_regexp_substr::getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& ct)
{
RegExpParams param = getEpressionAndPattern(row, fp, isNull, ct, ct.getTimeZone());
if (isNull)
return std::string{};
PCREOptions options(ct);
param.CharsetFix(options);
jp::Regex re(param.pattern, options.flags);
jp::RegexMatch rm(&re);
jp::VecNum vec_num;
size_t count = rm.setSubject(param.expression).setNumberedSubstringVector(&vec_num).match();
if (count == 0)
return std::string{};
return vec_num[0][0];
}
/*
Returns the position of the first occurrence of the regular expression pattern in the string subject, or 0
if pattern was not found. https://mariadb.com/kb/en/regexp_instr/
*/
std::string Func_regexp_instr::getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& ct)
{
RegExpParams param = getEpressionAndPattern(row, fp, isNull, ct, ct.getTimeZone());
if (isNull)
return std::string{};
PCREOptions options(ct);
param.CharsetFix(options);
jp::Regex re(param.pattern, options.flags);
jp::RegexMatch rm(&re);
jpcre2::VecOff vec_soff;
size_t count = rm.setSubject(param.expression).setMatchStartOffsetVector(&vec_soff).match();
if (count == 0)
return "0";
size_t offset = vec_soff[0];
size_t charNumber = options.libraryCharset.getCharset().numchars(param.expression.c_str(),
param.expression.c_str() + offset);
return std::to_string(charNumber + 1);
}
/*
https://mariadb.com/kb/en/regexp/
*/
bool Func_regexp::getBoolVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
CalpontSystemCatalog::ColType& ct) CalpontSystemCatalog::ColType& ct)
{ {
return getBool(row, pm, isNull, ct, ct.getTimeZone()) && !isNull; RegExpParams param = getEpressionAndPattern(row, fp, isNull, ct, ct.getTimeZone());
if (isNull)
return false;
PCREOptions options(ct);
param.CharsetFix(options);
jp::Regex re(param.pattern, options.flags);
return re.match(param.expression);
} }
} // namespace funcexp } // namespace funcexp

5
utils/funcexp/funcexp.cpp
View File

@ -215,7 +215,10 @@ FuncExp::FuncExp()
fFuncMap["quarter"] = new Func_quarter(); fFuncMap["quarter"] = new Func_quarter();
fFuncMap["radians"] = new Func_radians(); // dlh fFuncMap["radians"] = new Func_radians(); // dlh
fFuncMap["rand"] = new Func_rand(); fFuncMap["rand"] = new Func_rand();
fFuncMap["regexp"] = new Func_regexp(); // dlh fFuncMap["regexp"] = new Func_regexp();
fFuncMap["regexp_instr"] = new Func_regexp_instr();
fFuncMap["regexp_replace"] = new Func_regexp_replace();
fFuncMap["regexp_substr"] = new Func_regexp_substr(); // dlh
fFuncMap["repeat"] = new Func_repeat(); // dlh fFuncMap["repeat"] = new Func_repeat(); // dlh
fFuncMap["replace"] = new Func_replace(); // dlh fFuncMap["replace"] = new Func_replace(); // dlh
fFuncMap["replace_oracle"] = new Func_replace_oracle(); // dlh fFuncMap["replace_oracle"] = new Func_replace_oracle(); // dlh

55
utils/funcexp/functor_str.h
View File

@ -470,6 +470,61 @@ class Func_replace : public Func_Str
execplan::CalpontSystemCatalog::ColType& op_ct); execplan::CalpontSystemCatalog::ColType& op_ct);
}; };
class Func_regexp_replace : public Func_Str
{
public:
Func_regexp_replace() : Func_Str("regexp_replace")
{
}
virtual ~Func_regexp_replace()
{
}
execplan::CalpontSystemCatalog::ColType operationType(FunctionParm& fp,
execplan::CalpontSystemCatalog::ColType& resultType);
std::string getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& op_ct);
};
class Func_regexp_instr : public Func_Str
{
public:
Func_regexp_instr() : Func_Str("regexp_instr")
{
}
virtual ~Func_regexp_instr()
{
}
execplan::CalpontSystemCatalog::ColType operationType(FunctionParm& fp,
execplan::CalpontSystemCatalog::ColType& resultType);
std::string getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& op_ct);
};
class Func_regexp_substr : public Func_Str
{
public:
Func_regexp_substr() : Func_Str("regexp_substr")
{
}
virtual ~Func_regexp_substr()
{
}
execplan::CalpontSystemCatalog::ColType operationType(FunctionParm& fp,
execplan::CalpontSystemCatalog::ColType& resultType);
std::string getStrVal(rowgroup::Row& row, FunctionParm& fp, bool& isNull,
execplan::CalpontSystemCatalog::ColType& op_ct);
};
class Func_replace_oracle : public Func_Str class Func_replace_oracle : public Func_Str
{ {
public: public:

5153
utils/pcre2/jpcre2.hpp Normal file
View File

File diff suppressed because it is too large Load Diff