| Server IP : 37.220.80.31 / Your IP : 18.220.203.200 Web Server : Apache/2.4.52 (Ubuntu) System : Linux 3051455-guretool.twc1.net 5.15.0-107-generic #117-Ubuntu SMP Fri Apr 26 12:26:49 UTC 2024 x86_64 User : www-root ( 1010) PHP Version : 7.4.33 Disable Function : pcntl_alarm,pcntl_fork,pcntl_waitpid,pcntl_wait,pcntl_wifexited,pcntl_wifstopped,pcntl_wifsignaled,pcntl_wexitstatus,pcntl_wtermsig,pcntl_wstopsig,pcntl_signal,pcntl_signal_dispatch,pcntl_get_last_error,pcntl_strerror,pcntl_sigprocmask,pcntl_sigwaitinfo,pcntl_sigtimedwait,pcntl_exec,pcntl_getpriority,pcntl_setpriority, MySQL : OFF | cURL : ON | WGET : OFF | Perl : OFF | Python : OFF | Sudo : OFF | Pkexec : OFF Directory : /var/www/www-root/data/www/dev.artlot24.ru/bitrix/modules/seo/classes/general/ |
Upload File : |
<?
IncludeModuleLangFile(__FILE__);
use Bitrix\Main\Web\HttpClient;
class CSeoPageChecker
{
var $__site;
var $__url;
var $__lang;
var $__server_name;
var $__bCheckErrors = true;
var $__getter;
var $__result_headers;
var $__result_data;
var $__result_extended = array();
var $__result_meta = array('KEYWORDS' => '', 'DESCRIPTION' => '');
var $__result_errors = array();
var $__index;
var $__index_total_len;
var $pcre_backtrack_limit = false;
var $__qualifier_links_count = 100;
var $bError = false;
var $errorString = '';
var $bSearch = false;
public function __construct($site, $url, $get = true, $check_errors = true)
{
global $APPLICATION;
if (CModule::IncludeModule('search'))
$this->bSearch = true;
else
$APPLICATION->ThrowException(GetMessage('SEO_ERROR_NO_SEARCH')); // don't return false or set bError!
$this->__bCheckErrors = $check_errors;
$this->__site = $site;
$dbRes = CSite::GetByID($this->__site);
if ($arRes = $dbRes->Fetch())
{
$this->__lang = $arRes['LANGUAGE_ID'];
$this->__server_name = $arRes['SERVER_NAME'];
if ($this->__server_name == '')
$this->__server_name = COption::GetOptionString('main', 'server_name', '');
if ($this->__server_name <> '')
{
$this->__url = (CMain::IsHTTPS() ? "https://" : "http://")
.CBXPunycode::ToASCII($this->__server_name, $e)
.$url;
if(!$get || $this->GetHTTPData())
return true;
if($this->bError && $this->errorString <> '')
$APPLICATION->ThrowException($this->errorString);
return false;
}
else
{
$this->bError = true;
$APPLICATION->ThrowException(str_replace('#SITE_ID#', $this->__site, GetMessage('SEO_ERROR_NO_SERVER_NAME')));
return false;
}
}
return false;
}
function GetHTTPData()
{
$this->__getter = new HttpClient();
$this->__getter->setStreamTimeout(25);
$this->__getter->setRedirect(true);
if ($result = $this->__getter->get($this->__url))
{
$this->__result_data = $result;
$headers = $this->__getter->getHeaders()->toArray();
foreach ($headers as $header)
{
$currHeader = array();
foreach($header['values'] as $value)
$currHeader[] = $value;
$currHeader = implode(", ", $currHeader);
$this->__result_headers[$header["name"]] = $currHeader;
}
$this->_PrepareData();
unset($this->__getter);
$this->bError = false;
return true;
}
if($errors = $this->__getter->getError())
$this->errorString = implode(', ', $errors);
unset($this->__getter);
$this->bError = true;
return false;
}
function __prepareText($text)
{
$res = array();
if ($this->bSearch)
$res = stemming(CSearch::KillTags($text), $this->__lang);
else
$res = array();
return $res;
}
function _PrepareData()
{
if($this->pcre_backtrack_limit === false)
$this->pcre_backtrack_limit = intval(ini_get("pcre.backtrack_limit"));
$text_len = function_exists('mb_strlen')? mb_strlen($this->__result_data, 'latin1') : mb_strlen($this->__result_data);
$text_len++;
if($this->pcre_backtrack_limit > 0 && $this->pcre_backtrack_limit < $text_len)
{
@ini_set("pcre.backtrack_limit", $text_len);
$this->pcre_backtrack_limit = intval(ini_get("pcre.backtrack_limit"));
}
if($this->__bCheckErrors && $this->pcre_backtrack_limit > 0 && $this->pcre_backtrack_limit < $text_len)
{
$this->__result_errors[] = array(
'CODE' => 'SEO_PCRE',
'TYPE' => 'NOTE',
'DETAIL' => array(
'#PCRE_BACKTRACK_LIMIT#' => $this->pcre_backtrack_limit,
'#TEXT_LEN#' => $text_len,
)
);
}
$this->__index = array('TOTAL' => array(), 'BOLD' => array(), 'ITALIC' => array(), 'LINK' => array(), 'DESCRIPTION' => array(), 'KEYWORDS' => array());
// replace all images on their not empty ALT or TITLE attributes
$this->__result_data = preg_replace('/<img[^>]*(alt|title)=\"([^\"]*)\".*?>/is', '\\2', $this->__result_data);
if ($this->__bCheckErrors && ($img_cnt = preg_match('/<img.*?>/is', $this->__result_data)))
{
$this->__result_errors[] = array(
'CODE' => 'SEO_IMG_NO_ALT',
'TYPE' => 'NOTE',
'DETAIL' => array(
'#COUNT#' => $img_cnt
)
);
}
// get full words index
$this->__index['TOTAL'] = $this->__prepareText($this->__result_data);
// get bold words index
$arRes = array();
if(preg_match_all("/<(b|strong)>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['BOLD'] = $arRes[0];
$this->__index['BOLD'] = $this->__prepareText(implode(" ", $arRes[2]));
}
// get italic words index
if(preg_match_all("/<(i|em)>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['ITALIC'] = $arRes[0];
$this->__index['ITALIC'] = $this->__prepareText(implode(" ", $arRes[2]));
}
// get noindex tags
if(preg_match_all("/<(noindex)>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['NOINDEX'] = $arRes[0];
$this->__index['NOINDEX'] = $this->__prepareText(implode(" ", $arRes[2]));
}
// get link words index
if(preg_match_all("/<(a) ([^>]*)>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['LINK'] = $arRes[0];
$this->__index['LINK'] = $this->__prepareText(implode(" ", $arRes[3]));
$this->__result_extended['NOFOLLOW'] = array();
$this->__result_extended['LINK_EXTERNAL'] = array();
$this->__index['LINK_EXTERNAL'] = array();
foreach ($arRes[2] as $key => $attrs)
{
if (false !== mb_strpos($attrs, 'rel="nofollow"'))
$this->__result_extended['NOFOLLOW'][] = $arRes[0][$key];
if (false !== ($pos = mb_strpos($attrs, 'href="')))
{
$pos1 = mb_strpos($attrs, '"', $pos + 6);
$url = mb_substr($attrs, $pos, $pos1 - $pos);
if ($this->IsOuterUrl($url))
{
$this->__index['LINK_EXTERNAL'] = array_merge($this->__index['LINK_EXTERNAL'], $this->__prepareText($arRes[3][$key]));
$this->__result_extended['LINK_EXTERNAL'][] = $arRes[0][$key];
}
}
}
if ($this->__bCheckErrors && count($arRes[0]) > $this->__qualifier_links_count)
{
$this->__result_errors[] = array(
'CODE' => 'SEO_LINKS_COUNT',
'TYPE' => 'NOTE',
'DETAIL' => array(
'#COUNT#' => count($arRes[0]),
'#COUNT_EXTERNAL#' => count($this->__result_extended['LINK_EXTERNAL']),
'#QUALIFIER#' => $this->__qualifier_links_count,
)
);
}
}
// get meta description words index
if(preg_match('/<meta.*?name=\"description\".*?content=\"([^\"]+)\"[^>]*>/i', $this->__result_data, $arRes))
{
$this->__result_meta['DESCRIPTION'] = $arRes[1];
$this->__result_extended['META_DESCRIPTION'] = $arRes[0];
$this->__index['DESCRIPTION'] = $this->__prepareText($this->__result_meta['DESCRIPTION']);
}
else
{
$this->__result_errors[] = array(
'CODE' => 'SEO_META_NO_DESCRIPTION',
'TYPE' => 'NOTE',
'DETAIL' => array()
);
}
// get meta keywords words index
if(preg_match('/<meta.*?name=\"keywords\".*?content=\"([^\"]+)\"[^>]*>/i', $this->__result_data, $arRes))
{
$this->__result_meta['KEYWORDS'] = $arRes[1];
$this->__result_extended['META_KEYWORDS'] = $arRes[0];
$this->__index['KEYWORDS'] = $this->__prepareText($this->__result_meta['KEYWORDS']);
}
else
{
$this->__result_errors[] = array(
'CODE' => 'SEO_META_NO_KEYWORDS',
'TYPE' => 'NOTE',
'DETAIL' => array()
);
}
// get titles words index
if(preg_match("/<(title)>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['TITLE'] = $arRes[0];
$this->__index['TITLE'] = $this->__prepareText($arRes[2]);
}
if(preg_match_all("/<(h[\d]{1}).*?>.*?<\\/\\1>/is", $this->__result_data, $arRes))
{
$this->__result_extended['H'] = $arRes[0];
}
if(preg_match_all("/<(h1).*?>(.*?)<\\/\\1>/is", $this->__result_data, $arRes))
{
if ($this->__bCheckErrors && count($arRes[0]) > 1)
{
$this->__result_errors[] = array(
'CODE' => 'SEO_H1_UNIQUE',
'TYPE' => 'NOTE',
'DETAIL' => array(
'#COUNT#' => count($arRes[0]),
'#VALUES#' => htmlspecialcharsbx('"'.implode('", "', $arRes[2]).'"'),
)
);
}
$this->__index['H1'] = $this->__prepareText(implode(" ", $arRes[2]));
}
elseif ($this->__bCheckErrors)
{
$this->__result_errors[] = array(
'CODE' => 'SEO_H1_ABSENT',
'TYPE' => 'NOTE',
'DETAIL' => array()
);
}
if ($this->__bCheckErrors)
{
foreach(GetModuleEvents('seo', 'onPageCheck', true) as $arEvent)
{
if (!ExecuteModuleEventEx($arEvent, array(
array(
'URL' => $this->__url,
'LANG' => $this->__lang,
'SERVER_NAME' => $this->__server_name,
'SITE' => $this->__site,
),
array(
'HEADERS' => $this->__result_headers,
'BODY' => $this->__result_data,
),
$this->__result_meta,
$this->__index,
)) && ($ex = $GLOBALS['APPLICATION']->GetException()))
{
$this->__result_errors[] = array(
'CODE' => $ex->GetId(),
'TYPE' => 'NOTE',
'TEXT' => $ex->GetString(),
);
}
}
}
}
function _GetContrast($word)
{
if (null == $this->__index_total_len)
$this->__index_total_len = array_sum($this->__index['TOTAL']);
$logDocLength = log($this->__index_total_len < 20 ? 20 : $this->__index_total_len);
$count = intval($this->__index['TOTAL'][$word]);
return log($count+1)/$logDocLength;
}
function GetStatistics()
{
if (!is_array($this->__index))
return false;
if (null == $this->__index_total_len)
$this->__index_total_len = array_sum($this->__index['TOTAL']);
return array(
'URL' => $this->__url,
'TOTAL_LENGTH' => function_exists('mb_strlen')? mb_strlen($this->__result_data, 'latin1') : mb_strlen($this->__result_data),
'TOTAL_WORDS_COUNT' => $this->__index_total_len ? $this->__index_total_len : '-',
'UNIQUE_WORDS_COUNT' => $this->__index_total_len ? count($this->__index['TOTAL']) : '-',
'META_KEYWORDS' => $this->__result_meta['KEYWORDS'],
'META_DESCRIPTION' => $this->__result_meta['DESCRIPTION'],
);
}
function GetURL()
{
return $this->__url;
}
function CheckKeyword($keyword, $bStemmed = false)
{
if (!is_array($this->__index))
return false;
if (is_array($keyword))
{
$arResult = array();
foreach ($keyword as $key => $word)
{
$arResult[$key] = $this->CheckKeyword($bStemmed ? $key : $word, $bStemmed);
}
return $arResult;
}
if (!$bStemmed && $this->bSearch)
$keyword = stemming($keyword, $this->__lang);
if (is_array($keyword))
return $this->CheckKeyword($keyword, true);
$arResult = array(
'TOTAL' => intval($this->__index['TOTAL'][$keyword]),
'BOLD' => intval($this->__index['BOLD'][$keyword]),
'ITALIC' => intval($this->__index['ITALIC'][$keyword]),
'LINK' => intval($this->__index['LINK'][$keyword]),
'LINK_EXTERNAL' => intval($this->__index['LINK_EXTERNAL'][$keyword]),
'DESCRIPTION' => intval($this->__index['DESCRIPTION'][$keyword]),
'KEYWORDS' => intval($this->__index['KEYWORDS'][$keyword]),
'TITLE' => intval($this->__index['TITLE'][$keyword]),
'H1' => intval($this->__index['H1'][$keyword]),
'CONTRAST' => $this->_GetContrast($keyword),
);
return $arResult;
}
function GetExtendedData()
{
return array_merge(array('HEADERS' => $this->__result_headers), $this->__result_extended);
}
function GetErrors()
{
$arResult = false;
if (count($this->__result_errors) > 0)
{
$arResult = array();
foreach ($this->__result_errors as $arError)
{
$arResult[] = array(
'CODE' => $arError['CODE'],
'TYPE' => $arError['TYPE'],
'TEXT' => isset($arError['TEXT']) ? $arError['TEXT'] : str_replace(array_keys($arError['DETAIL']), array_values($arError['DETAIL']), GetMessage($arError['CODE'].'_ERROR')),
);
}
}
return $arResult;
}
function IsOuterUrl($url)
{
if (strncmp($url, '#', 1) === 0) return false;
if (strncmp($url, 'mailto:', 7) === 0) return false;
if (strncmp($url, 'javascript:', 11) === 0) return false;
$pos = mb_strpos($url, '://');
if ($pos === false) return false;
static $arDomainNames = null;
if (null == $arDomainNames)
{
$arDomainNames = array($_SERVER['SERVER_NAME']);
$dbRes = CSite::GetList('sort', 'asc', array('ACTIVE' => 'Y'));
while ($arSite = $dbRes->Fetch())
{
if ($arSite['DOMAINS'])
$arDomainNames = array_merge($arDomainNames, explode("\r\n", $arSite['DOMAINS']));
}
$arDomainNames = array_values(array_unique($arDomainNames));
}
$url = mb_substr($url, $pos + 3);
$pos = mb_strpos($url, '/');
if ($pos === false)
{
$pos = mb_strlen($url);
}
$domain = mb_substr($url, 0, $pos);
if (mb_substr($domain, 0, 4) == 'www.')
{
$domain = mb_substr($domain, 4);
}
if ($domain)
return !in_array($domain, $arDomainNames);
return false;
}
}
?>