PHP: template performance

What am I talking about?

Oh yeah, I wanted to tell you about a little test that I spent at leisure. The fact is that I love reinventing bicycles (don't kick me for this innocent hobby). Therefore, when I was a poachy programmer, I thought about the implementation of the template engine (and whoever did not think, let him throw a stone at me).
If you want to compare the performance of famous template engines - sorry, another time
The topic of this article is to study the performance of some particular cases of using simple templates.

Actually what was analyzed

I think many people know what templates are, who do not know, read Wikipedia , templates are a way of representing a view (View)
Actually there are three approaches to this business (not counting third-party developments and modules like XML / XSLT):

1. Connecting ordinary PHP files where the template is described
   Reading, parsing and executing a template in some sample language
   Point 2 but with caching of the template code compiled in PHP
   
   

   Testing sources

   
   

   Simple connection

   
   

    function file_included () {
        $ vars ['some'] = 'some';
        include 'included.php';
    }
   

   
   

   Simple require

   
   

    function file_required () {
        $ vars ['some'] = 'some';
        require 'included.php';
    }
   

   
   

   Reading and executing a file

   
   

    function file_evaled () {
        $ vars ['some'] = 'some';
        $ content = file_get_contents ('included.php');
        eval ('?>'. $ content);
    }
   

   
   

   Parsing regexp

   
   

    function file_tpl_preg () {
        $ vars ['some'] = 'some';
        $ content = file_get_contents ('included.tpl');
        $ content = preg_replace ('# \ {(\ w +) \} #', '<? = $ vars [\' \ 1 \ ']?>', $ content);
        eval ('?>'. $ content);
    }
   

   
   

   Parsing string replacements

   
   

    function file_tpl_strtr () {
        $ vars ['some'] = 'some';
        $ content = file_get_contents ('included.tpl');
        $ content = strtr ($ content, array (
            '{' => '<?  $ vars [\ '',
            '}' => '\']?> ',
        ))
        eval ('?>'. $ content);
    }
   

   
   

   Regexp with caching time change check

   
   

    function file_tpl_preg_cashe () {
        $ file = 'included';
        if (filemtime ($ file. '. tpl.php') <filemtime ($ file. '. tpl')) {
            $ content = file_get_contents ('included.tpl');
            $ content = preg_replace ('# \ {(\ w +) \} #', '<? = $ vars [\' \ 1 \ ']?>', $ content);
            file_put_contents ($ file. '. tpl.php', $ content);
        }
        $ vars ['some'] = 'some';
        include $ file. '. tpl.php';
    }
   

   
   

   Regexp with caching simple existence check

   
   

    function file_tpl_manual_cashe () {
        $ file = 'included';
        if (! file_exists ($ file. '. tpl1.php')) {
            $ content = file_get_contents ('included.tpl');
            $ content = preg_replace ('# \ {(\ w +) \} #', '<? = $ vars [\' \ 1 \ ']?>', $ content);
            file_put_contents ($ file. '. tpl1.php', $ content);
        }
        $ vars ['some'] = 'some';
        include $ file. '. tpl1.php';
    }
   

   
   

   results

   
   Functions were simply called in a loop, file system access caching is disabled.
   

   Type of test Time % of speedy	file_included0.00021430100.00%	file_required0.00021500100.33%	file_evaled0.00022488104.94%	file_tpl_preg0.00024208112.96%	file_tpl_strtr0.00022802106.40%	file_tpl_preg_cashe0.00024305113.42%	file_tpl_manual_cashe0.00023100107.79% I will explain the results: file_included - simple PHP template incl, fastest file_required - the same connection, only in profile file_evaled - separately read and execute PHP template, a bit slower file_tpl_preg - parsing the regexp template, 10-12 percent slower than the inclus file_tpl_strtr - parsing with quick replacements in lines, an inflexible approach, but only 5-6 percent slower file_tpl_preg_cashe - parsing regexpom and caching with automatic check for cache aging. The speed is almost the same as the same without caching, it looks like filemtime is a very slow function file_tpl_manual_cashe - caching only with checking the existence of the cache (for parsing, you need to delete the old cache). Pretty good results, only 6-7 percent slower than the inclusion. ') Draw your own conclusions :)