mysqlimagereplacer/exec.php

<?php
/**
* mySQL image replacer - https://gitlab.com/zlax/mysqlimagereplacer
* adopted for php8.2
* v.0.3 beta - uder DWTWL license https://soundragon.su/license/license.html
* exec script
*
* !MAKE BACKUP OF YOUR SQL-TABLE BEFORE YOU TRY THIS SCRIPT!
*/
ignore_user_abort(true);
ini_set('max_execution_time', '6000');
ini_set('max_input_time', '6000');

function SaveStatus ($status,$raws,$updated,$failed,$deleted,$execdir) {
    $csvstatusndata[]=$raws;
    $csvstatusndata[]=$updated;
    $csvstatusndata[]=$failed;
    $csvstatusndata[]=$deleted;
    if ($status=="started") {
        $fhandle=fopen($execdir."/outdata.csv","w");
        fputcsv($fhandle,$csvstatusndata);
        fclose($fhandle);
    } else
        if ($status=="finished") unlink ($execdir."/outdata.csv");
}

function DownloadFopen($URL) {
    if (file_exists(rawurldecode(urldecode(pathinfo($URL)['basename'])))) {
        $fileprefix=mt_rand(1000000,9999999);
        file_put_contents($fileprefix.rawurldecode(urldecode(pathinfo($URL)['basename'])), file_get_contents($URL));
        if (md5_file($fileprefix.rawurldecode(urldecode(pathinfo($URL)['basename'])))==md5_file(rawurldecode(urldecode(pathinfo($URL)['basename'])))) {
            unlink ($fileprefix.rawurldecode(urldecode(pathinfo($URL)['basename'])));
            $result=rawurldecode(urldecode(pathinfo($URL)['basename']));
        } else {
            if (exif_imagetype($fileprefix.rawurldecodeurldecode((pathinfo($URL)['basename'])))>0) $result=$fileprefix.rawurldecode(urldecode(pathinfo($URL)['basename']));
            else {
                unlink ($fileprefix.rawurldecode(urldecode(pathinfo($URL)['basename'])));
                $result=false;
            }
        }
    } else {
        file_put_contents(rawurldecode(urldecode(pathinfo($URL)['basename'])), file_get_contents($URL));
        if (exif_imagetype(rawurldecode(urldecode(pathinfo($URL)['basename'])))>0) $result=rawurldecode(urldecode(pathinfo($URL)['basename']));
        else {
            unlink (rawurldecode(urldecode(pathinfo($URL)['basename'])));
            $result=false;
        }
    }
    return $result;    
}

$execdir=__DIR__;
$fhandle=fopen("indata.csv","r") or $logmessage="emptyindata";
$indatacsv=fgetcsv($fhandle, 1024, ",");
fclose($fhandle);
$dbhost=$indatacsv[0];
$dbuser=$indatacsv[1];
$dbpass=$indatacsv[2];
$dbname=$indatacsv[3];
$dbtable=$indatacsv[4];
$dbcolumn=$indatacsv[5];
$servername=$indatacsv[6];
$fixbroken=$indatacsv[7];
$dirname=$indatacsv[8];
$subdir=$indatacsv[9];
$downloadtype=$indatacsv[10];
$htmlprefix=$indatacsv[11];
$fixalready=$indatacsv[12];
$fixundownload=$indatacsv[13];
$correctimagenumber=$indatacsv[14];
$filtcolumn=$indatacsv[15];
$filtvalue=$indatacsv[16];
$ahrefparents=$indatacsv[17];
$webarchived=explode (",", $indatacsv[18]);
$fromnum=$indatacsv[19];

unlink ("indata.csv");

if (isset($logmessage) && $logmessage=="emptyindata")
    echo "Can't read data file, please check write pirivlegies for ",$execdir," directory";
chdir("../".$dirname);
if ($subdir)
    chdir ($subdir);
$dbconnection=mysqli_connect($dbhost,$dbuser,$dbpass,$dbname);
mysqli_set_charset($dbconnection,'utf8');
$resulttable=mysqli_query($dbconnection,"SELECT * FROM ".$dbtable);
$totalstatus="started";
$updatedurls=0;
$failedtoupdate=0;
$deletedurls=0;
$rawsedit=0;
SaveStatus ($totalstatus,$rawsedit,$updatedurls,$failedtoupdate,$deletedurls,$execdir);
$filteredpostnumber=0;
$postnumber=0;
if (!$fromnum) $fromnum=0;
while($row=mysqli_fetch_assoc($resulttable)){
    if ($postnumber >= $fromnum) {
        if (!empty($filtcolumn) && !empty($filtvalue)) {
            if ($row[$filtcolumn] == $filtvalue) {
                if (!empty($row[$dbcolumn]))
                    $originalcontent=$row[$dbcolumn];
                    $filteredpostnumber++;
                }
            } else { if (!empty($row[$dbcolumn]))
                $originalcontent=$row[$dbcolumn];
                $filteredpostnumber++;
            }
        if (isset($originalcontent)) {
            $newcontent=$originalcontent;
            if ($fixbroken) {
                $contentpost=new DOMDocument();
                $xmlEncodding = '<?xml version="1.0" encoding="UTF-8"?>';
                $contentpost->loadHTML($xmlEncodding.$originalcontent, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
                $imageTags=$contentpost->getElementsByTagName('img');
                for ($i = $imageTags->length; --$i >= 0; ) { // http://php.net/manual/class.domnodelist.php#83390
                    $tag = $imageTags->item($i);
                    $imgsrctag=$tag->getAttribute('src');
                    $parsedimgsrctag=parse_url($imgsrctag);
                    if (!((!empty($parsedimgsrctag['host'])&&!empty($parsedimgsrctag['path']))&&(($parsedimgsrctag['scheme']=="http")||($parsedimgsrctag ['scheme']=="https")))) {
                        $tag->parentNode->removeChild($tag);
                        $deletedurls++;
                    }
                }
                $newcontent=str_replace($xmlEncodding,"",$contentpost->saveHTML());
            }
            if ($correctimagenumber>0) {
                $contentpost=new DOMDocument();
                $contentpost->loadHTML($newcontent);
                $imageTags=$contentpost->getElementsByTagName('img');
                for ($i = $imageTags->length; --$i >= 0; ) { // http://php.net/manual/class.domnodelist.php#83390
                    $tag = $imageTags->item($i);
                    if ($ahrefparents) {
                        $aofimg=$contentpost->importNode($tag->parentNode, true);
                        if (isset($aofimg)) {
                            $ahreofimg=$aofimg->getAttribute('href');
                            if ($ahreofimg=="") $ahreofimg=$tag->getAttribute('src');
                            $imgsrctag=$ahreofimg;
                        }
                    } else $imgsrctag=$tag->getAttribute('src');
                    $parsedimgsrctag=parse_url($imgsrctag);
                    if ((!empty($parsedimgsrctag['host'])&&!empty($parsedimgsrctag['path']))&&(($parsedimgsrctag['scheme']=="http")||($parsedimgsrctag ['scheme']=="https"))&&(!($parsedimgsrctag['host']==$servername))) {
                        if ($webarchived) {
                            $parsedimgsrctagarch=parse_url($tag->getAttribute('src'));
                            if (in_array($parsedimgsrctagarch['host'], $webarchived)) {
                                $webarchivelink="https://web.archive.org/web/20150101id_/".$tag->getAttribute('src');
                                $newimgsrc=DownloadFopen($webarchivelink);
                            } else
                                $newimgsrc=DownloadFopen($imgsrctag);
                        } else
                            $newimgsrc=DownloadFopen($imgsrctag);
                        if ($newimgsrc) {
                            if ($htmlprefix=="http") {
                                $srcbase="http://".$servername."/".$dirname."/";
                            } else if ($htmlprefix=="https") {
                                $srcbase="https://".$servername."/".$dirname."/";
                            } else {
                                $srcbase="/".$dirname."/";
                            }
                            if ($subdir) $srcbase=$srcbase.rawurlencode($subdir)."/";
                            $newimgsrc=$srcbase.$newimgsrc;
                            $newcontent=str_replace($tag->getAttribute('src'),$newimgsrc,$newcontent);
                            if ($ahrefparents)
                                $newcontent=str_replace($imgsrctag,$newimgsrc,$newcontent);
                            unset($newimgsrc);
                            $updatedurls++;
                            SaveStatus ($totalstatus,$rawsedit,$updatedurls,$failedtoupdate,$deletedurls,$execdir);
                        } else if ($fixundownload==true) {
                            $tag->parentNode->removeChild($tag);
                            $deletedurls++;
                            SaveStatus ($totalstatus,$rawsedit,$updatedurls,$failedtoupdate,$deletedurls,$execdir);
                        } else $failedtoupdate++;
                    }
                }
            }
            $stmt = mysqli_prepare($dbconnection, "UPDATE `".$dbtable."` SET `".$dbcolumn."` = ? WHERE `".$dbtable."`.`".$dbcolumn."`=?");
            mysqli_stmt_bind_param($stmt, 'ss', $newcontent, $originalcontent);
            mysqli_stmt_execute($stmt);
            $rawsedit++;
            SaveStatus ($totalstatus,$rawsedit,$updatedurls,$failedtoupdate,$deletedurls,$execdir);
        }
    }
    $postnumber++;
}
$totalstatus="finished";
SaveStatus ($totalstatus,$rawsedit,$updatedurls,$failedtoupdate,$deletedurls,$execdir);
echo $updatedurls," urls was downloaded and updated, ",$failedtoupdate," urls was failed to update","<br>\n";
if ($fixbroken or $fixundownload) echo $deletedurls, " tags was deleted<br>\n";

?>