Solving parsing problems with the markdown library and html elements.

7 years ago · d4b0a9ea2a
2 changed files with 23 additions and 2 deletions
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@ -18,7 +18,22 @@ function diaspora2bb($s) {
 	$s = html_entity_decode($s,ENT_COMPAT,'UTF-8');

 	// Simply remove cr.
-	$s = str_replace("\r","",$s);
+	//$s = str_replace("\r","",$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);

 	// Escaping the hash tags
 	$s = preg_replace('/\#([^\s\#])/','&#35;$1',$s);
@ -53,6 +68,8 @@ function diaspora2bb($s) {

 function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {

+	$a = get_app();
+
 	$OriginalText = $Text;

 	// Since Diaspora is creating a summary for links, this function removes them before posting
@ -92,9 +109,13 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
 	// Maybe we should make this newline at every time before a quote.
 	$Text = str_replace(array("</a><blockquote>"), array("</a><br><blockquote>"), $Text);

+	$stamp1 = microtime(true);
+
 	// Now convert HTML to Markdown
 	$Text = new HTML_To_Markdown($Text);

+	$a->save_timestamp($stamp1, "parser");
+
 	// Libertree has a problem with escaped hashtags - Diaspora doesn't seem to.
 	if (!$fordiaspora)
 		$Text = str_replace(array('\#'), array('#'), $Text);
--- a/library/markdown.php
+++ b/library/markdown.php
@ -17,7 +17,7 @@ function Markdown($text) {
 	# Read file and pass content through the Markdown parser
 	$html = MarkdownExtra::defaultTransform($text);

-	$a->save_timestamp($stamp1, "rendering");
+	$a->save_timestamp($stamp1, "parser");

 	return $html;
 }