Trying to avoid parser problems.

This commit is contained in:
Michael Vogel 2015-02-24 20:39:02 +01:00
parent 59fa1c9453
commit b827824faf
1 changed files with 5 additions and 18 deletions

View File

@ -17,23 +17,11 @@ function diaspora2bb($s) {
$s = html_entity_decode($s,ENT_COMPAT,'UTF-8');
// Simply remove cr.
// Remove CR to avoid problems with following code
//$s = str_replace("\r","",$s);
/*
// The parser has problems with unbalanced HTML elements
$doc = new DOMDocument();
$doc->preserveWhiteSpace = false;
$s = mb_convert_encoding($s, 'HTML-ENTITIES', "UTF-8");
$doctype = '<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">';
$encoding = '<?xml encoding="UTF-8">';
@$doc->loadHTML($encoding.$doctype."<html><body>".$s."</body></html>");
$doc->encoding = 'UTF-8';
$s = $doc->saveHTML();
$s = str_replace(array("<html><body>", "</body></html>", $doctype, $encoding), array("", "", "", ""), $s);
*/
// The parser has problems with unbalanced html elements
$s = str_replace(array("<br/>", "</p>", "<p>", '<p dir="ltr">'),array("<br />", "<br />", "<br />", "<br />"),$s);
// The parser cannot handle paragraphs correctly
$s = str_replace(array("</p>", "<p>", '<p dir="ltr">'),array("<br>", "<br>", "<br>"),$s);
// Escaping the hash tags
$s = preg_replace('/\#([^\s\#])/','&#35;$1',$s);
@ -116,8 +104,7 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
$a->save_timestamp($stamp1, "parser");
// Libertree has a problem with escaped hashtags - Diaspora doesn't seem to.
if (!$fordiaspora)
// Libertree has a problem with escaped hashtags.
$Text = str_replace(array('\#'), array('#'), $Text);
// Remove any leading or trailing whitespace, as this will mess up