Dereku/snappymail
1
0
Fork 0
mirror of https://github.com/the-djmaze/snappymail.git synced 2026-09-05 15:37:03 +03:00
Code Issues Projects Releases Packages Wiki Activity Actions

Find links in html messages

Browse source
This commit is contained in:
RainLoop Team 2014-02-06 15:56:55 +04:00 • committed by RayMan
parent d50c068a3d
commit 041bd1ae1b
6 changed files with 388 additions and 335 deletions
Download patch file Download diff file Expand all files Collapse all files

114
rainloop/v/0.0.0/app/libraries/MailSo/Base/HtmlUtils.php
View file

@ -8,6 +8,8 @@ namespace MailSo\Base;
*/
class HtmlUtils
{
static $KOS = '@@_KOS_@@';
/**
* @access private
*/
@ -261,14 +263,106 @@ class HtmlUtils
return \implode(';', $aOutStyles);
}
/**
* @param \DOMDocument $oDom
*/
public static function FindLinksInDOM(&$oDom)
{
$aNodes = $oDom->getElementsByTagName('*');
foreach ($aNodes as /* @var $oElement \DOMElement */ $oElement)
{
$sTagNameLower = \strtolower($oElement->tagName);
$sParentTagNameLower = isset($oElement->parentNode) && isset($oElement->parentNode->tagName) ?
\strtolower($oElement->parentNode->tagName) : '';
if (!\in_array($sTagNameLower, array('html', 'meta', 'head', 'style', 'script', 'img', 'button', 'input', 'textarea', 'a')) &&
'a' !== $sParentTagNameLower && $oElement->childNodes && 0 < $oElement->childNodes->length)
{
$oSubItem = null;
$aTextNodes = array();
$iIndex = $oElement->childNodes->length - 1;
while ($iIndex > -1)
{
$oSubItem = $oElement->childNodes->item($iIndex);
if ($oSubItem && XML_TEXT_NODE === $oSubItem->nodeType)
{
$aTextNodes[] = $oSubItem;
}
$iIndex--;
}
unset($oSubItem);
foreach ($aTextNodes as $oTextNode)
{
if ($oTextNode && 0 < \strlen($oTextNode->wholeText)/* && \preg_match('/http[s]?:\/\//i', $oTextNode->wholeText)*/)
{
$sText = \MailSo\Base\LinkFinder::NewInstance()
->Text($oTextNode->wholeText)
->UseDefaultWrappers(true)
->CompileText()
;
$oSubDom = \MailSo\Base\HtmlUtils::GetDomFromText('<html><body>'.$sText.'</body></html>');
if ($oSubDom)
{
$oBodyNodes = $oSubDom->getElementsByTagName('body');
if ($oBodyNodes && 0 < $oBodyNodes->length)
{
$oBodyChildNodes = $oBodyNodes->item(0)->childNodes;
if ($oBodyChildNodes && $oBodyChildNodes->length)
{
for ($iIndex = 0, $iLen = $oBodyChildNodes->length; $iIndex < $iLen; $iIndex++)
{
$oSubItem = $oBodyChildNodes->item($iIndex);
if ($oSubItem)
{
if (XML_ELEMENT_NODE === $oSubItem->nodeType &&
'a' === \strtolower($oSubItem->tagName))
{
$oLink = $oDom->createElement('a',
\str_replace(':', \MailSo\Base\HtmlUtils::$KOS, \htmlspecialchars($oSubItem->nodeValue)));
$sHref = $oSubItem->getAttribute('href');
if ($sHref)
{
$oLink->setAttribute('href', $sHref);
}
$oElement->insertBefore($oLink, $oTextNode);
}
else
{
$oElement->insertBefore($oDom->importNode($oSubItem), $oTextNode);
}
}
}
$oElement->removeChild($oTextNode);
}
}
unset($oBodyNodes);
}
unset($oSubDom, $sText);
}
}
}
}
unset($aNodes);
}
/**
* @param string $sHtml
* @param bool $bDoNotReplaceExternalUrl = false
* @param bool $bFindLinksInHtml = false
*
* @return string
*/
public static function ClearHtmlSimple($sHtml, $bDoNotReplaceExternalUrl = false)
public static function ClearHtmlSimple($sHtml, $bDoNotReplaceExternalUrl = false, $bFindLinksInHtml = false)
{
$bHasExternals = false;
$aFoundCIDs = array();
@ -276,7 +370,7 @@ class HtmlUtils
$aFoundedContentLocationUrls = array();
return \MailSo\Base\HtmlUtils::ClearHtml($sHtml, $bHasExternals, $aFoundCIDs,
$aContentLocationUrls, $aFoundedContentLocationUrls, $bDoNotReplaceExternalUrl);
$aContentLocationUrls, $aFoundedContentLocationUrls, $bDoNotReplaceExternalUrl, $bFindLinksInHtml);
}
/**
@ -286,11 +380,13 @@ class HtmlUtils
* @param array $aContentLocationUrls = array()
* @param array $aFoundedContentLocationUrls = array()
* @param bool $bDoNotReplaceExternalUrl = false
* @param bool $bFindLinksInHtml = false
*
* @return string
*/
public static function ClearHtml($sHtml, &$bHasExternals = false, &$aFoundCIDs = array(),
$aContentLocationUrls = array(), &$aFoundedContentLocationUrls = array(), $bDoNotReplaceExternalUrl = false)
$aContentLocationUrls = array(), &$aFoundedContentLocationUrls = array(),
$bDoNotReplaceExternalUrl = false, $bFindLinksInHtml = false)
{
$sHtml = null === $sHtml ? '' : (string) $sHtml;
$sHtml = \trim($sHtml);
@ -313,11 +409,16 @@ class HtmlUtils
if ($oDom)
{
if ($bFindLinksInHtml)
{
\MailSo\Base\HtmlUtils::FindLinksInDOM($oDom);
}
$aNodes = $oDom->getElementsByTagName('*');
foreach ($aNodes as /* @var $oElement \DOMElement */ $oElement)
{
$sTagNameLower = \strtolower($oElement->tagName);
// convert body attributes to styles
if ('body' === $sTagNameLower)
{
@ -493,6 +594,8 @@ class HtmlUtils
$sResult = '<div data-x-div-type="body" '.$sBodyAttrs.'>'.$sResult.'</div>';
$sResult = '<div data-x-div-type="html" '.$sHtmlAttrs.'>'.$sResult.'</div>';
$sResult = \str_replace(\MailSo\Base\HtmlUtils::$KOS, ':', $sResult);
return \trim($sResult);
}
@ -658,8 +761,7 @@ class HtmlUtils
$sText = \MailSo\Base\LinkFinder::NewInstance()
->Text($sText)
->UseDefaultWrappers($bLinksWithTargetBlank)
// ->CompileText(true, false);
->CompileText(true, true);
->CompileText();
$sText = \str_replace("\r", '', $sText);

57
rainloop/v/0.0.0/app/libraries/MailSo/Base/LinkFinder.php
View file

@ -126,13 +126,8 @@ class LinkFinder
*/
public function UseDefaultWrappers($bAddTargetBlank = false)
{
$this->fLinkWrapper = function ($sLink, $bShortLink = false) use ($bAddTargetBlank) {
if ($bShortLink && \in_array(\strtolower($sLink), array('asp.net', 'vb.net', 'mailbee.net')))
{
return $sLink;
}
$this->fLinkWrapper = function ($sLink) use ($bAddTargetBlank) {
$sNameLink = $sLink;
if (!\preg_match('/^[a-z]{3,5}\:\/\//i', \ltrim($sLink)))
{
@ -151,11 +146,10 @@ class LinkFinder
/**
* @param bool $bUseHtmlSpecialChars = true
* @param bool $bFindShortLinks = true
*
* @return string
*/
public function CompileText($bUseHtmlSpecialChars = true, $bFindShortLinks = true)
public function CompileText($bUseHtmlSpecialChars = true)
{
$sText = \substr($this->sText, 0, $this->iOptimizationLimit);
$sSubText = \substr($this->sText, $this->iOptimizationLimit);
@ -171,11 +165,6 @@ class LinkFinder
$sText = $this->findMails($sText, $this->fMailWrapper);
}
if ($bFindShortLinks && null !== $this->fLinkWrapper && \is_callable($this->fLinkWrapper))
{
$sText = $this->findShortLinks($sText, $this->fLinkWrapper);
}
$sResult = '';
if ($bUseHtmlSpecialChars)
{
@ -259,46 +248,6 @@ class LinkFinder
return $sText;
}
/**
* @param string $sText
* @param mixed $fWrapper
*
* @return string
*/
private function findShortLinks($sText, $fWrapper)
{
$sPattern = '/([a-z0-9-\.]+\.(?:com|org|net|ru))([^a-z0-9-\.])/i';
$aPrepearPlainStringUrls = $this->aPrepearPlainStringUrls;
$sText = \preg_replace_callback($sPattern, function ($aMatch) use ($fWrapper, &$aPrepearPlainStringUrls) {
if (\is_array($aMatch) && 2 < \count($aMatch) && isset($aMatch[1]) && 0 < \strlen($aMatch[1]))
{
$sLinkWithWrap = \call_user_func_array($fWrapper, array($aMatch[1], true));
if (\is_string($sLinkWithWrap))
{
$aPrepearPlainStringUrls[] = \stripslashes($sLinkWithWrap);
return \MailSo\Base\LinkFinder::OPEN_LINK.
(\count($aPrepearPlainStringUrls) - 1).
\MailSo\Base\LinkFinder::CLOSE_LINK.
$aMatch[2];
}
return $aMatch[0];
}
return '';
}, $sText);
if (0 < \count($aPrepearPlainStringUrls))
{
$this->aPrepearPlainStringUrls = $aPrepearPlainStringUrls;
}
return $sText;
}
/**
* @param string $sText
* @param mixed $fWrapper

14
rainloop/v/0.0.0/app/libraries/MailSo/Base/Utils.php
View file

@ -740,7 +740,7 @@ class Utils
{
$sResult = '';
$sContentType = \strtolower($sContentType);
if (0 === strpos($sContentType, 'image/'))
if (0 === \strpos($sContentType, 'image/'))
{
$sResult = 'image';
}
@ -847,7 +847,7 @@ class Utils
*/
public static function Php2js($mInput)
{
return \json_encode($mInput, defined('JSON_UNESCAPED_UNICODE') ? JSON_UNESCAPED_UNICODE : 0);
return \json_encode($mInput, \defined('JSON_UNESCAPED_UNICODE') ? JSON_UNESCAPED_UNICODE : 0);
// if (\is_null($mInput))
// {
@ -918,25 +918,25 @@ class Utils
*/
public static function RecRmDir($sDir)
{
if (@is_dir($sDir))
if (@\is_dir($sDir))
{
$aObjects = scandir($sDir);
$aObjects = \scandir($sDir);
foreach ($aObjects as $sObject)
{
if ('.' !== $sObject && '..' !== $sObject)
{
if ('dir' === filetype($sDir.'/'.$sObject))
if ('dir' === \filetype($sDir.'/'.$sObject))
{
self::RecRmDir($sDir.'/'.$sObject);
}
else
{
@unlink($sDir.'/'.$sObject);
@\unlink($sDir.'/'.$sObject);
}
}
}
return @rmdir($sDir);
return @\rmdir($sDir);
}
return false;