Wavyzz/dolibarr
2
0
Fork 1
mirror of https://github.com/Dolibarr/dolibarr.git synced 2025-12-06 01:28:19 +01:00
Code Issues Packages Projects Releases Wiki Activity

Try fix to solve pb with xml tag with MAIN_RESTRICTHTML_ONLY_VALID_HTML

Browse Source
This commit is contained in:
ldestailleur
2025-08-06 11:55:44 +02:00
parent 3dbf9ad759
commit aa376fe333
1 changed files with 15 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
htdocs/core/lib/functions.lib.php
View File

@@ -8541,7 +8541,8 @@ function dol_string_onlythesehtmlattributes($stringtoclean, $allowed_attributes
}
if (class_exists('DOMDocument') && !empty($stringtoclean)) {
$stringtoclean = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body>'.$stringtoclean.'</body></html>';
//$stringtoclean = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body>'.$stringtoclean.'</body></html>';
$stringtoclean = '<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body>'.$stringtoclean.'</body></html>';
// Warning: loadHTML does not support HTML5 on old libxml versions.
$dom = new DOMDocument('', 'UTF-8');
@@ -8597,7 +8598,7 @@ function dol_string_onlythesehtmlattributes($stringtoclean, $allowed_attributes
$return = $dom->saveHTML(); // This may add a LF at end of lines, so we will trim later
//$return = '<html><body>aaaa</p>bb<p>ssdd</p>'."\n<p>aaa</p>aa<p>bb</p>";
$return = preg_replace('/^'.preg_quote('<?xml encoding="UTF-8">', '/').'/', '', $return);
//$return = preg_replace('/^'.preg_quote('<?xml encoding="UTF-8">', '/').'/', '', $return);
$return = preg_replace('/^'.preg_quote('<html><head><', '/').'[^<>]*'.preg_quote('></head><body>', '/').'/', '', $return);
$return = preg_replace('/'.preg_quote('</body></html>', '/').'$/', '', trim($return));
@@ -8767,14 +8768,19 @@ function dol_htmlwithnojs($stringtoencode, $nouseofiframesandbox = 0, $check = '
}
$dom = new DOMDocument();
// Add a trick to solve pb with text without parent tag
// like '<h1>Foo</h1><p>bar</p>' that wrongly ends up, without the trick, with '<h1>Foo<p>bar</p></h1>'
// like 'abc' that wrongly ends up, without the trick, with '<p>abc</p>'
// Add a trick '<div class="tricktoremove">' to solve pb with text without parent tag
// like '<h1>Foo</h1><p>bar</p>' that wrongly ends up, without the trick, with '<h1>Foo<p>bar</p></h1>'
// like 'abc' that wrongly ends up, without the trick, with '<p>abc</p>'
// Add also a trick <html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"> to solve utf8 lost.
// I don't know what the xml encoding is the trick for
if (dol_textishtml($out)) {
$out = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.$out.'</div></body></html>';
//$out = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.$out.'</div></body></html>';
$out = '<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.$out.'</div></body></html>';
//$out = '<html><head><meta charset="utf-8"></head><body><div class="tricktoremove">'.$out.'</div></body></html>';
} else {
$out = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.dol_nl2br($out).'</div></body></html>';
//$out = '<?xml encoding="UTF-8"><html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.dol_nl2br($out).'</div></body></html>';
$out = '<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body><div class="tricktoremove">'.dol_nl2br($out).'</div></body></html>';
//$out = '<html><head><meta charset="utf-8"></head><body><div class="tricktoremove">'.dol_nl2br($out).'</div></body></html>';
}
$dom->loadHTML($out, LIBXML_HTML_NODEFDTD | LIBXML_ERR_NONE | LIBXML_HTML_NOIMPLIED | LIBXML_NONET | LIBXML_NOWARNING | LIBXML_NOERROR | LIBXML_NOXMLDECL);
@@ -8784,7 +8790,7 @@ function dol_htmlwithnojs($stringtoencode, $nouseofiframesandbox = 0, $check = '
$out = trim($dom->saveHTML());
// Remove the trick added to solve pb with text in utf8 and text without parent tag
$out = preg_replace('/^'.preg_quote('<?xml encoding="UTF-8">', '/').'/', '', $out);
//$out = preg_replace('/^'.preg_quote('<?xml encoding="UTF-8">', '/').'/', '', $out);
$out = preg_replace('/^'.preg_quote('<html><head><', '/').'[^<>]+'.preg_quote('></head><body><div class="tricktoremove">', '/').'/', '', $out);
$out = preg_replace('/'.preg_quote('</div></body></html>', '/').'$/', '', trim($out));
// $out = preg_replace('/^<\?xml encoding="UTF-8"><div class="tricktoremove">/', '', $out);