sitemapfile.php

<?php
namespace Bitrix\Seo;
 
use Bitrix\Main\IO\Path;
use Bitrix\Main\IO\File;
use Bitrix\Main\SiteTable;
use Bitrix\Main\Text\Converter;
 
class SitemapFile
    extends File
{
    const XML_HEADER = '<?xml version="1.0" encoding="UTF-8"?>';
 
    const FILE_HEADER = '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">';
    const FILE_FOOTER = '</urlset>';
 
    const ENTRY_TPL = '<url><loc>%s</loc><lastmod>%s</lastmod></url>';
    const ENTRY_TPL_SEARCH = '<url><loc>%s</loc>';
 
    const XPATH_URL = '/urlset/url';
 
    const MAX_SIZE = 5000000;
 
    const FILE_EXT = '.xml';
    const FILE_PART_SUFFIX = '.part';
 
    protected $documentRoot;
    protected $settings = array();
    protected $parser = false;
 
    protected $siteRoot = '';
    protected $partFile = '';
    protected $partList = array();
    protected $part = 0;
    protected $partChanged = false;
    protected $footerClosed = false;
 
    protected $urlToSearch = '';
    protected $urlFound = false;
 
    public function __construct($fileName, $settings)
    {
        $this->settings = array(
            'SITE_ID' => $settings['SITE_ID'],
            'PROTOCOL' => $settings['PROTOCOL'] == 'https' ? 'https' : 'http',
            'DOMAIN' => $settings['DOMAIN'],
        );
 
        $site = SiteTable::getRow(array("filter" => array("LID" => $this->settings['SITE_ID'])));
 
        $this->documentRoot = SiteTable::getDocumentRoot($this->settings['SITE_ID']);
        $this->footerClosed = false;
 
        $this->siteRoot = Path::combine(
            $this->documentRoot,
            $site['DIR']
        );
        
        $fileName = $this->prepareFileName($fileName);
        $this->partFile = $this->partFile ?: $fileName;
        $this->pathPhysical = null; // hack for object reconstuct during file splitting
        parent::__construct($this->siteRoot.'/'.$fileName, $this->settings['SITE_ID']);
        $this->partChanged = $this->isExists() && !$this->isSplitNeeded();
    }
    
    protected function prepareFileName($fileName)
    {
        // normalize slashes
        $fileName = Path::normalize($fileName);
        if (mb_substr($fileName, -mb_strlen(self::FILE_EXT)) != self::FILE_EXT)
        {
            $fileName .= self::FILE_EXT;
        }
        
        // convert words delimiter, google dont't like '_''
        $fileName = str_replace('_', '-', $fileName);
        
        return $fileName;
    }
    
    protected function reInit($fileName)
    {
        $this->__construct($fileName, $this->settings);
    }
 
    public function addHeader()
    {
        $this->partChanged = true;
        $this->putContents(self::XML_HEADER.self::FILE_HEADER);
    }
 
    protected function isSplitNeeded()
    {
        return $this->isExists() && $this->getSize() >= self::MAX_SIZE;
    }
 
    public function addEntry($entry)
    {
        if ($this->isSplitNeeded())
        {
            $this->split();
            $this->addEntry($entry);
        }
        else
        {
            if (!$this->partChanged)
            {
                $this->addHeader();
            }
 
            $this->putContents(
                sprintf(
                    self::ENTRY_TPL,
                    Converter::getXmlConverter()->encode($entry['XML_LOC']),
                    Converter::getXmlConverter()->encode($entry['XML_LASTMOD'])
                ), self::APPEND
            );
        }
    }
 
    public function split()
    {
        if($this->partChanged)
        {
            $this->addFooter();
        }
 
        $this->partList[] = $this->getName();
        $this->part++;
 
        $fileName = $this->partFile;
        $fileName = mb_substr($fileName, 0, -mb_strlen(self::FILE_EXT)).self::FILE_PART_SUFFIX.$this->part.mb_substr($fileName, -mb_strlen(self::FILE_EXT));
 
        $this->reInit($fileName);
 
        $this->partChanged = $this->isExists() && !$this->isSplitNeeded();
 
        return $fileName;
    }
 
    public function getNameList()
    {
        return $this->isCurrentPartNotEmpty() ? array_merge($this->partList, array($this->getName())) : $this->partList;
    }
    
    public function getPathDirectory()
    {
//      normalize slashes
        $siteRoot = Path::normalize($this->siteRoot);
        $fileName = $this->getName();
        $path = Path::normalize($this->path);
        
        $directory = str_replace(array($siteRoot, $fileName), array('',''), $path);
        
        return ltrim($directory, '/');
    }
 
    public function isNotEmpty()
    {
        return (count($this->partList) > 0) || $this->isCurrentPartNotEmpty();
    }
 
    public function isCurrentPartNotEmpty()
    {
        if($this->isExists())
        {
            $c = $this->getContents();
            return $c <> '' && $c != self::XML_HEADER.self::FILE_HEADER;
        }
 
        return false;
    }
 
    public function appendEntry($entry)
    {
        if($this->isSplitNeeded())
        {
            $this->split();
            $this->appendEntry($entry);
        }
        else
        {
            if(!$this->partChanged)
            {
                $this->addHeader();
                $offset = $this->getSize();
            }
            else
            {
                $offset = $this->getSize() - mb_strlen(self::FILE_FOOTER);
            }
 
            $fd = $this->open('r+');
 
            fseek($fd, $offset);
            fwrite($fd, sprintf(
                self::ENTRY_TPL,
                Converter::getXmlConverter()->encode($entry['XML_LOC']),
                Converter::getXmlConverter()->encode($entry['XML_LASTMOD'])
            ).self::FILE_FOOTER);
            fclose($fd);
            
            $this->footerClosed = true;
        }
    }
 
    public function removeEntry($url)
    {
        $fileName = $this->partFile;
        $e = [];
        $url = $this->settings['PROTOCOL'] . '://' . \CBXPunycode::toASCII($this->settings['DOMAIN'], $e) . $url;
        $pattern = sprintf(self::ENTRY_TPL_SEARCH, $url);
 
        while($this->isExists())
        {
            $c = $this->getContents();
            $p = mb_strpos($c, $pattern);
            unset($c);
 
            if($p !== false)
            {
                $fd = $this->open('r+');
 
                fseek($fd, intval($p));
                fwrite($fd, str_repeat(" ", mb_strlen(sprintf(
                    self::ENTRY_TPL,
                    Converter::getXmlConverter()->encode($url),
                    Converter::getXmlConverter()->encode(date('c'))
                ))));
                fclose($fd);
                break;
            }
 
            if(!$this->isSplitNeeded())
            {
                break;
            }
            else
            {
                $this->part++;
                $fileName = mb_substr($fileName, 0, -mb_strlen(self::FILE_EXT)).self::FILE_PART_SUFFIX.$this->part.mb_substr($fileName, -mb_strlen(self::FILE_EXT));
                $this->reInit($fileName);
            }
        }
        
        return $fileName;
    }
 
    public function addFileEntry(File $f)
    {
        if($f->isExists() && !$f->isSystem())
        {
            $e = [];
            $this->addEntry(array(
                'XML_LOC' => $this->settings['PROTOCOL'].'://'.\CBXPunycode::toASCII($this->settings['DOMAIN'], $e).$this->getFileUrl($f),
                'XML_LASTMOD' => date('c', $f->getModificationTime()),
            ));
        }
    }
 
    public function addIBlockEntry($url, $modifiedDate)
    {
        $e = [];
        $this->addEntry(array(
            'XML_LOC' => $this->settings['PROTOCOL'].'://'.\CBXPunycode::toASCII($this->settings['DOMAIN'], $e).$url,
            'XML_LASTMOD' => date('c', $modifiedDate - \CTimeZone::getOffset()),
        ));
    }
 
    public function appendIBlockEntry($url, $modifiedDate)
    {
        if($this->isExists())
        {
            $e = [];
            $this->appendEntry(array(
                'XML_LOC' => $this->settings['PROTOCOL'].'://'.\CBXPunycode::toASCII($this->settings['DOMAIN'], $e).$url,
                'XML_LASTMOD' => date('c', $modifiedDate - \CTimeZone::getOffset()),
            ));
        }
        else
        {
            $this->addHeader();
            $this->addIBlockEntry($url, $modifiedDate);
            $this->addFooter();
        }
    }
 
    public function addFooter()
    {
        $this->putContents(self::FILE_FOOTER, self::APPEND);
        $this->footerClosed = true;
    }
 
    public function getSiteRoot()
    {
        return $this->siteRoot;
    }
 
    public function getUrl()
    {
        $e = [];
        return $this->settings['PROTOCOL'].'://'.\CBXPunycode::toASCII($this->settings['DOMAIN'], $e).$this->getFileUrl($this);
    }
 
    public function parse()
    {
        if(!$this->parser)
        {
            if($this->isExists())
            {
                $this->parser = new \CDataXML();
                $this->parser->loadString($this->getContents());
            }
        }
 
        return $this->parser;
    }
 
    protected function getFileUrl(File $f)
    {
        static $indexNames;
        if(!is_array($indexNames))
        {
            $indexNames = GetDirIndexArray();
        }
        
        $documentRoot  = Path::normalize($this->documentRoot);
        $path = '/';
        if (mb_substr($this->path, 0, mb_strlen($documentRoot)) === $documentRoot)
        {
            $path = '/'.mb_substr($f->getPath(), mb_strlen($documentRoot));
        }
 
        $path = Path::convertLogicalToUri($path);
 
        $path = in_array($f->getName(), $indexNames)
            ? str_replace('/'.$f->getName(), '/', $path)
            : $path;
 
        return '/'.ltrim($path, '/');
    }
}