mirror of
https://github.com/FirebirdSQL/firebird-qa.git
synced 2025-01-22 21:43:06 +01:00
90 lines
2.5 KiB
Python
90 lines
2.5 KiB
Python
#coding:utf-8
|
|
#
|
|
# id: bugs.core_6542
|
|
# title: Implementation of SUBSTRING for UTF8 character set is inefficient
|
|
# decription:
|
|
# Confirmed issues on 4.0.0.2422 and 3.0.8.33345: ratio is about 16x ... 17x.
|
|
# Checked on 4.0.0.2425 and 3.0.8.33349: ratio is about 1.05 ... 1.07.
|
|
# Decided to use threshold = 1.15 for check performance ratio.
|
|
#
|
|
# tracker_id: CORE-6542
|
|
# min_versions: ['3.0.8']
|
|
# versions: 3.0.8
|
|
# qmid:
|
|
|
|
import pytest
|
|
from firebird.qa import db_factory, isql_act, Action
|
|
|
|
# version: 3.0.8
|
|
# resources: None
|
|
|
|
substitutions_1 = []
|
|
|
|
init_script_1 = """"""
|
|
|
|
db_1 = db_factory(sql_dialect=3, init=init_script_1)
|
|
|
|
test_script_1 = """
|
|
set term ^;
|
|
execute block as
|
|
declare str1 varchar(8000) character set unicode_fss;
|
|
declare str2 varchar(10) character set unicode_fss;
|
|
declare n int = 100000;
|
|
declare t0 timestamp;
|
|
declare t1 timestamp;
|
|
begin
|
|
str1 = LPAD('abcd', 8000, '--');
|
|
t0 = 'now';
|
|
while (n > 0) do
|
|
begin
|
|
str2 = SUBSTRING(str1 from 1 FOR 10);
|
|
n = n - 1;
|
|
end
|
|
t1 = 'now';
|
|
rdb$set_context('USER_SESSION', 'TIME_UFSS', datediff(millisecond from t0 to t1) );
|
|
end
|
|
^
|
|
execute block as
|
|
declare str1 varchar(8000) character set utf8;
|
|
declare str2 varchar(10) character set utf8;
|
|
declare n int = 100000;
|
|
declare t0 timestamp;
|
|
declare t1 timestamp;
|
|
begin
|
|
str1 = LPAD('abcd', 8000, '--');
|
|
t0 = 'now';
|
|
while (n > 0) do
|
|
begin
|
|
str2 = SUBSTRING(str1 from 1 FOR 10);
|
|
n = n - 1;
|
|
end
|
|
t1 = 'now';
|
|
rdb$set_context('USER_SESSION', 'TIME_UTF8', datediff(millisecond from t0 to t1) );
|
|
end
|
|
^
|
|
set term ;^
|
|
|
|
set list on;
|
|
select iif( r < threshold, 'acceptable.', 'POOR: ' || r || ' - exceeds threshold = ' || threshold) as "UTF8_substring_performance:"
|
|
from (
|
|
select
|
|
1.000 * cast( rdb$get_context('USER_SESSION', 'TIME_UTF8') as int) / cast( rdb$get_context('USER_SESSION', 'TIME_UFSS') as int) as r
|
|
,1.15 as threshold
|
|
from rdb$database
|
|
);
|
|
|
|
"""
|
|
|
|
act_1 = isql_act('db_1', test_script_1, substitutions=substitutions_1)
|
|
|
|
expected_stdout_1 = """
|
|
UTF8_substring_performance: acceptable.
|
|
"""
|
|
|
|
@pytest.mark.version('>=3.0.8')
|
|
def test_1(act_1: Action):
|
|
act_1.expected_stdout = expected_stdout_1
|
|
act_1.execute()
|
|
assert act_1.clean_expected_stdout == act_1.clean_stdout
|
|
|