Data preparation script for ЛЯМ-search module used at memorici.de

update.php

<?php
/*+++ CONFIG +++*/
        $ramfs = '/mnt/ramdisk/';
        $dbdir = '/home/sweater/coding/scripting/lyam/data/';
        $url   = 'http://forum.sc2tv.ru/archive/';
        $file  = 'index.php?t-14423';

        $page_pattern = '|<a .*?>(?<pagenum>\d*)</a>|';
/*--- CONFIG ---*/

        $ramdir = $ramfs.'data/';
        $cpfile = $dbdir.'current.dat';
        $pages  = $dbdir.'cache.dat';

        function fpc($x,$y) {return file_put_contents($y,$x);}
        function fgc($x)    {return file_get_contents($x);   }

        function last($a)   {return $a[(count($a)-1)];}

        function get_last_page() {
                global $page_pattern, $url, $file;
                $matches = array();

                preg_match_all($page_pattern, fgc($url.$file.'.html'), $matches);
                return last($matches['pagenum']);
        }

        function current_page() {
                global $cpfile;
                return (int)trim(fgc($cpfile), "\n ");
        }

        function linearize($html) {
                return 
                        str_replace('<hr />', '<hr />'."\n", 
                                str_replace("\n", '', $html)
                        )
                ;
        }

        function update() {
                global $url, $file, $ramdir, $dbdir, $pages, $cpfile;

                $cp = current_page();
                $np = get_last_page();
                echo "<<INFO>> cp: $cp, np: $np\n\n\n";
                for ($i = $cp; $i <= $np; ++$i) {
                        echo "<<GOGO>> $i\n";
                        $i_name = $file.'-p-'.$i.'.html';
                        unlink($dbdir.$i_name);

                        shell_exec('wget '.$url.$file.'-p-'.$i.'.html -P '.$dbdir);
                        fpc(linearize(fgc($dbdir.$i_name)), $dbdir.$i_name);

                        shell_exec('cp -rvut '.$ramdir.' '.$dbdir.$i_name);
                        fpc($ramdir.$i_name."\n".fgc($pages), $pages);
                        echo "<<DONE>> $i_name\n";
                }
                fpc ($np, $cpfile);
        }

        die(update());