extern/egroupware_official
1
0
Fork 0
mirror of https://github.com/EGroupware/egroupware.git synced 2024-11-22 16:03:47 +01:00
Code Issues Packages Projects Releases Wiki Activity

- Re-arrange HTML tag order to account for simplistic newline handling (broke nested tags)

Browse Source 
- If unable to load/parse content after trying to convert HTML styles into target document styles, fall back and try again, but strip all tags this time.
This commit is contained in:
Nathan Gray 2012-02-01 00:03:31 +00:00
parent 3ae6e76ff9
commit b8d9b748ac
1 changed files with 67 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

66
etemplate/inc/class.bo_merge.inc.php
View File

@ -71,6 +71,11 @@ abstract class bo_merge
'wrap' => 0, // Wrapping can break output
);
/**
* Parse HTML styles into target document style, if possible
*/
protected $parse_html_styles = true;
/**
* Constructor
*
@ -518,7 +523,20 @@ abstract class bo_merge
// Remove spans with no attributes, linebreaks inside them cause problems
'/<span>(.*?)<\/span>/' => '$1'
);
$content = preg_replace(array_keys($replace_tags),array_values($replace_tags),$content);
$content = preg_replace(array_keys($replace_tags),array_values($replace_tags),$content, -1, $count);
/*
In the case where you have something like <span><span></w:t><w:br/><w:t></span></span> (invalid - mismatched tags),
it takes multiple runs to get rid of both spans. So, loop.
OO.o files have not yet been shown to have this problem.
*/
$count = $i = 0;
do
{
$content = preg_replace('/<span>(.*?)<\/span>/','$1',$content, -1, $count);
$i++;
} while($count > 0 && $i < 10);
//echo $content;die();
$doc = new DOMDocument();
$xslt = new XSLTProcessor();
@ -530,13 +548,15 @@ abstract class bo_merge
// XSLT transform known tags
if($xslt)
{
try
{
// does NOT work with php 5.2.6: Catchable fatal error: argument 1 to transformToXml() must be of type DOMDocument
//$element = new SimpleXMLelement($content);
$element = new DOMDocument('1.0', 'utf-8');
$element->loadXML($content);
$result = $element->loadXML($content);
if(!$result)
{
throw new Exception('Unable to parse merged document for styles. Check warnings in log for details.');
}
$content = $xslt->transformToXml($element);
//echo $content;die();
@ -551,12 +571,6 @@ abstract class bo_merge
$doc->schemaValidate(*Schema (xsd) file*);
*/
}
catch (Exception $e)
{
error_log($e->getMessage());
// Failed...
}
}
}
/**
@ -804,7 +818,7 @@ abstract class bo_merge
// Tags we can replace with the target document's version
$replace_tags = array();
// only keep tags, if we have xsl extension available
if (class_exists(XSLTProcessor) && class_exists(DOMDocument))
if (class_exists(XSLTProcessor) && class_exists(DOMDocument) && $this->parse_html_styles)
{
switch($mimetype.$mso_application_progid)
{
@ -866,6 +880,18 @@ abstract class bo_merge
// replace </p> and <br /> with CRLF (remove <p> and CRLF)
$value = str_replace(array("\r","\n",'<p>','</p>','<br />'),array('','','',"\r\n","\r\n"),$value);
$value = strip_tags($value,implode('',$replace_tags));
// Change <tag>...\r\n</tag> to <tag>...</tag>\r\n or simplistic line break below will mangle it
// Loop to catch things like <b><span>Break:\r\n</span></b>
if($mso_application_progid)
{
$count = $i = 0;
do
{
$value = preg_replace('/<(b|strong|i|em|u|span)\b([^>]*?)>(.*?)'."\r\n".'<\/\1>/u', '<$1$2>$3</$1>'."\r\n",$value,-1,$count);
$i++;
} while($count > 0 && $i < 10); // Limit of 10 chosen arbitrarily just in case
}
}
// replace all control chars (C0+C1) but CR (\015), LF (\012) and TAB (\011) (eg. vertical tabulators) with space
// as they are not allowed in xml
@ -1150,10 +1176,26 @@ abstract class bo_merge
// Apply HTML formatting to target document, if possible
// check if we can use the XSL extension, to not give a fatal error and rendering whole merge-print non-functional
if (class_exists(XSLTProcessor) && class_exists(DOMDocument))
if (class_exists(XSLTProcessor) && class_exists(DOMDocument) && $this->parse_html_styles)
{
try
{
$this->apply_styles($merged, $mimetype);
}
catch (Exception $e)
{
// Error converting HTML styles over
error_log($e->getMessage());
error_log("Target document: $content_url, IDs: ". array2string($ids));
// Try again, but strip HTML so user gets something
$this->parse_html_styles = false;
if (!($merged =& $this->merge($content_url,$ids,$err,$mimetype,$fix)))
{
return $err;
}
}
}
if(!empty($name))
{
if(empty($ext))