Bug #74935 [Com]: mb_str*pos functions do not take into consideration mb_internal_encoding

Edit report at https://bugs.php.net/bug.php?id=74935&edit=1 ID: 74935 Comment by: spam2 at rhsoft dot net Reported by: andxfiles at gmail dot com Summary: mb_str*pos functions do not take into consideration mb_internal_encoding Status: Closed Type: Bug Package: mbstring related Operating System: All PHP Version: Irrelevant Assigned To: nikic Block user comment: N Private report: N New Comment: you need master and not 7.1.9-dev when on the developer mailing list is a discussion ongoing if the mbstring refactoring could be included in 7.2 Previous Comments: ------------------------------------------------------------------------ [2017-07-24 06:51:05] andxfiles at gmail dot com Indeed this report is the same as https://bugs.php.net/bug.php?id=74933 so you can definitely close one of the two. Here are the results for the script with the new patches of master which unfortunately do not show any change. mb_stripos with mb_internal_encoding is still as much slow as before. The OS is: Windows Server 2012 Standard 64bit, php is x86(32bit). But the same results come out on Windows 7 64bit with php x86(32bit). @nicic did you try it under Linux or Windows ? 7.1.1 mb_strlen with parameter UTF8: 290.4 milliseconds mb_stripos with parameter UTF8: 3250.2 milliseconds mb_substr with parameter UTF8: 295.9 milliseconds mb_strlen with mb_internal_encoding UTF8: 15.1 milliseconds mb_stripos with mb_internal_encoding UTF8: 3292.2 milliseconds mb_substr with mb_internal_encoding UTF8: 21.9 milliseconds 7.1.9-dev (master) mb_strlen with parameter UTF8: 290.2 milliseconds mb_stripos with parameter UTF8: 3245.3 milliseconds mb_substr with parameter UTF8: 300.3 milliseconds mb_strlen with mb_internal_encoding UTF8: 16.1 milliseconds mb_stripos with mb_internal_encoding UTF8: 3240.4 milliseconds mb_substr with mb_internal_encoding UTF8: 23.8 milliseconds ------------------------------------------------------------------------ [2017-07-23 11:00:11] nikic@php.net A number of performance improvements for mb_stripos have been implemented. Output of your test code on PHP 7.1: mb_strlen with parameter UTF8: 38.2 milliseconds mb_stripos with parameter UTF8: 705.8 milliseconds mb_substr with parameter UTF8: 42.3 milliseconds mb_strlen with mb_internal_encoding UTF8: 15.1 milliseconds mb_stripos with mb_internal_encoding UTF8: 663.6 milliseconds mb_substr with mb_internal_encoding UTF8: 20.9 milliseconds Output on master: mb_strlen with parameter UTF8: 12.4 milliseconds mb_stripos with parameter UTF8: 83.4 milliseconds mb_substr with parameter UTF8: 19.5 milliseconds mb_strlen with mb_internal_encoding UTF8: 11.7 milliseconds mb_stripos with mb_internal_encoding UTF8: 82.4 milliseconds mb_substr with mb_internal_encoding UTF8: 17.8 milliseconds For your particular example this gives an approximately 8x improvement. Whether these changes will be in PHP 7.2 is yet to be seen. ------------------------------------------------------------------------ [2017-07-17 08:50:38] andxfiles at gmail dot com Description: ------------ when calling mb_str*pos (stripos, strrpos etc.) functions using mb_internal_encoding before their call instead of passing the encoding as a parameter, those functions do not use the encoding provided, thus being way slower. As one can see in the actual results below, when using mb_internal_encoding, all functions become way faster except for mb_stripos. Test script: --------------- <?php $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_strlen("fdsfdssdfoifjosdifjosdifjosdij:ά", "UTF-8"); } $finishtime = microtime(true); echo "mb_strlen with parameter UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_stripos("fdsfdssdfoifjosdifjosdifjosdij:ά", "α", 0, "UTF-8"); } $finishtime = microtime(true); echo "mb_stripos with parameter UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_substr("fdsfdssdfoifjosdifjosdifjosdij:ά", $i, 1, "UTF-8"); } $finishtime = microtime(true); echo "mb_substr with parameter UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; mb_internal_encoding("UTF-8"); $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_strlen("fdsfdssdfoifjosdifjosdifjosdij:ά"); } $finishtime = microtime(true); echo "mb_strlen with mb_internal_encoding UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_stripos("fdsfdssdfoifjosdifjosdifjosdij:ά", "α", 0); } $finishtime = microtime(true); echo "mb_stripos with mb_internal_encoding UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; $starttime = microtime(true); for ($i=0; $i<100000; $i++) { $a = mb_substr("fdsfdssdfoifjosdifjosdifjosdij:ά", $i, 1); } $finishtime = microtime(true); echo "mb_substr with mb_internal_encoding UTF8: " . number_format($finishtime - $starttime, 4)*1000 ." milliseconds<br/>"; ?> Actual result: -------------- mb_strlen with parameter UTF8: 103.7 milliseconds mb_stripos with parameter UTF8: 1450.7 milliseconds mb_substr with parameter UTF8: 81.6 milliseconds mb_strlen with mb_internal_encoding UTF8: 39.2 milliseconds mb_stripos with mb_internal_encoding UTF8: 1435.3 milliseconds mb_substr with mb_internal_encoding UTF8: 48.5 milliseconds ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=74935&edit=1