Brand new html2bbcode. Some other changes in diaspora to bbcode handling

1 files changed, 271 insertions, 119 deletions

diff --git a/include/html2bbcode.php b/include/html2bbcode.php
index 8025c336b..1b4b5e99a 100755
--- a/include/html2bbcode.php
+++ b/include/html2bbcode.php
@@ -1,127 +1,279 @@
 <?php
+/*
+html2bbcode.php
+Converter for HTML to BBCode
+Made by: ike@piratenpartei.de
+Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
+					https://github.com/annando/Syncom
+*/
 
-/**
- * html2bbcode
- */
-
-
-function html2bbcode($s) {
-
-
-	// only keep newlines from source that are within pre tags
-
-	$s = stripnl_exceptinpre($s);
-
-
-	// Tags to Find
-
-	$htmltags = array(
-		'/\<pre\>(.*?)\<\/pre\>/is',
-		'/\<p(.*?)\>/is',
-		'/\<\/p\>/is',
-		'/\<b\>(.*?)\<\/b\>/is',
-		'/\<i\>(.*?)\<\/i\>/is',
-		'/\<u\>(.*?)\<\/u\>/is',
-		'/\<ul\>(.*?)\<\/ul\>/is',
-		'/\<li\>(.*?)\<\/li\>/is',
-		'/\<img(.*?)width: *([0-9]+)(.*?)height: *([0-9]+)(.*?)src=\"(.*?)\" (.*?)\>/is',
-		'/\<img(.*?)height: *([0-9]+)(.*?)width: *([0-9]+)(.*?)src=\"(.*?)\" (.*?)\>/is',
-		'/\<img(.*?)src=\"(.*?)\"(.*?)width: *([0-9]+)(.*?)height: *([0-9]+)(.*?)\>/is',
-		'/\<img(.*?)src=\"(.*?)\"(.*?)height: *([0-9]+)(.*?)width: *([0-9]+)(.*?)\>/is',
-		'/\<img(.*?) src=\"(.*?)\" (.*?)\>/is',
-		'/\<div(.*?)\>(.*?)\<\/div\>/is',
-		'/\<br(.*?)\>/is',
-		'/\<strong\>(.*?)\<\/strong\>/is',
-		'/\<a (.*?)href=\"(.*?)\"(.*?)\>(.*?)\<\/a\>/is',
-		'/\<code\>(.*?)\<\/code\>/is',
-		'/\<span style=\"color:(.*?)\"\>(.*?)\<\/span\>/is',
-		'/\<span style=\"font-size:(.*?)\"\>(.*?)\<\/span\>/is',
-		'/\<blockquote\>(.*?)\<\/blockquote\>/is',
-		'/\<video(.*?) src=\"(.*?)\" (.*?)\>(.*?)\<\/video\>/is',
-		'/\<audio(.*?) src=\"(.*?)\" (.*?)\>(.*?)\<\/audio\>/is',
-		'/\<iframe(.*?) src=\"(.*?)\" (.*?)\>(.*?)\<\/iframe\>/is',
-
-	);
-
-	// Replace with
-
-	$bbtags = array(
-		'[code]$1[/code]',
-		'',
-		"\n",
-		'[b]$1[/b]',
-		'[i]$1[/i]',
-		'[u]$1[/u]',
-		'[list]$1[/list]',
-		'[*]$1',
-		'[img=$2x$4]$6[/img]',
-		'[img=$4x$2]$6[/img]',
-		'[img=$4x$6]$2[/img]',
-		'[img=$6x$4]$2[/img]',
-		'[img]$2[/img]',
-		'$2',
-		"\n",
-		'[b]$1[/b]',
-		'[url=$2]$4[/url]',
-		'[code]$1[/code]',
-		'[color="$1"]$2[/color]',
-		'[size=$1]$2[/size]',
-		'[quote]$1[/quote]',
-		'[video]$1[/video]',
-		'[audio]$1[/audio]',
-		'[iframe]$1[/iframe]',
-	);
-
-	// Replace $htmltags in $text with $bbtags
-	$text = preg_replace ($htmltags, $bbtags, $s);
-
-	call_hooks('html2bbcode', $text);
-
-	// Strip all other HTML tags
-	$text = strip_tags($text);
-	return $text;
+function node2bbcode(&$doc, $oldnode, $attributes, $startbb, $endbb)
+{
+	do {
+		$done = node2bbcodesub(&$doc, $oldnode, $attributes, $startbb, $endbb);
+	} while ($done);
+}
+
+function node2bbcodesub(&$doc, $oldnode, $attributes, $startbb, $endbb)
+{
+	$savestart = str_replace('$', '%', $startbb);
+	$replace = false;
+
+	$xpath = new DomXPath($doc);
+
+	$list = $xpath->query("//".$oldnode);
+	foreach ($list as $oldNode) {
+
+		$attr = array();
+		if ($oldNode->attributes->length)
+			foreach ($oldNode->attributes as $attribute)
+				$attr[$attribute->name] = $attribute->value;
+
+		$replace = true;
+
+		$startbb = $savestart;
+
+		$i = 0;
+
+		foreach ($attributes as $attribute => $value) {
+
+			$startbb = str_replace('%'.++$i, '$1', $startbb);
+
+			if (strpos('*'.$startbb, '$1') > 0) {
+
+				if ($replace and (@$attr[$attribute] != '')) {
+
+					$startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
+
+					// If nothing could be changed
+					if ($count == 0)
+						$replace = false;
+				} else
+					$replace = false;
+			} else {
+				if (@$attr[$attribute] != $value)
+					$replace = false;
+			}
+		}
+
+		if ($replace) {
+			$StartCode = $oldNode->ownerDocument->createTextNode($startbb);
+			$EndCode = $oldNode->ownerDocument->createTextNode($endbb);
+
+			$oldNode->parentNode->insertBefore($StartCode, $oldNode);
 
+			if ($oldNode->hasChildNodes()) {
+				foreach ($oldNode->childNodes as $child) {
+					$newNode = $child->cloneNode(true);
+					$oldNode->parentNode->insertBefore($newNode, $oldNode);
+				}
+			}
+
+			$oldNode->parentNode->insertBefore($EndCode, $oldNode);
+			$oldNode->parentNode->removeChild($oldNode);
+		}
+	}
+	return($replace);
 }
 
-function stripnl_exceptinpre($string)
+function deletenode(&$doc, $node)
 {
-    // First, check for <pre> tag
-    if(strpos($string, '<pre>') === false)
-    {
-        return str_replace("\n","", $string);
-    }
-
-    // If there is a <pre>, we have to split by line
-    // and manually replace the linebreaks
-
-    $strArr=explode("\n", $string);
-
-    $output="";
-    $preFound=false;
-
-    // Loop over each line
-    foreach($strArr as $line)
-    {    // See if the line has a <pre>. If it does, set $preFound to true
-        if(strpos($line, "<pre>") !== false)
-        {
-            $preFound=true;
-        }
-        elseif(strpos($line, "</pre>") !== false)
-        {
-            $preFound=false;
-        }
-       
-        // If we are in a pre tag, add line and also add \n, else add the line without \n
-        if($preFound)
-        {
-            $output .= $line . "\n";
-        }
-        else
-        {
-            $output .= $line ;
-        }
-    }
-
-    return $output;
+	$xpath = new DomXPath($doc);
+	$list = $xpath->query("//".$node);
+	foreach ($list as $child)
+		$child->parentNode->removeChild($child);
 }
 
+function html2bbcode($message)
+{
+
+	//$file = tempnam("/tmp/", "html");
+	//file_put_contents($file, $message);
+
+	$message = str_replace("\r", "", $message);
+
+	$message = str_replace(array(
+					"<li><p>",
+					"</p></li>"),
+				array(
+					"<li>",
+					"</li>"),
+				$message);
+
+	// remove namespaces
+	$message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
+	$message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
+
+	$doc = new DOMDocument();
+	$doc->preserveWhiteSpace = false;
+
+	$message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
+
+	@$doc->loadHTML($message);
+
+	deletenode($doc, 'style');
+	deletenode($doc, 'head');
+	deletenode($doc, 'title');
+	deletenode($doc, 'meta');
+	deletenode($doc, 'xml');
+	deletenode($doc, 'removeme');
+
+	$xpath = new DomXPath($doc);
+	$list = $xpath->query("//pre");
+	foreach ($list as $node)
+		$node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
+
+	$message = $doc->saveHTML();
+	$message = str_replace(array("\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"), array("<", ">", "<br>", " ", ""), $message);
+	$message = preg_replace('= [\s]*=i', " ", $message);
+	@$doc->loadHTML($message);
+
+	node2bbcode($doc, 'html', array(), "", "");
+	node2bbcode($doc, 'body', array(), "", "");
+
+	// Outlook-Quote - Variant 1
+	node2bbcode($doc, 'p', array('class'=>'MsoNormal', 'style'=>'margin-left:35.4pt'), '[quote]', '[/quote]');
+
+	// Outlook-Quote - Variant 2
+	node2bbcode($doc, 'div', array('style'=>'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'), '[quote]', '[/quote]');
+
+	// MyBB-Stuff
+	node2bbcode($doc, 'span', array('style'=>'text-decoration: underline;'), '[u]', '[/u]');
+	node2bbcode($doc, 'span', array('style'=>'font-style: italic;'), '[i]', '[/i]');
+	node2bbcode($doc, 'span', array('style'=>'font-weight: bold;'), '[b]', '[/b]');
+
+	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
+	node2bbcode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
+	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
+	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
+	node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
+	node2bbcode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
+	node2bbcode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
+
+	node2bbcode($doc, 'span', array('style'=>'/.*color:\s*(.+?)[,;].*/'), '[color="$1"]', '[/color]');
+	node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)/'), '[size=$1]', '[/size]');
+
+	//node2bbcode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
+	//node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
+	//node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
+	//node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
+
+	node2bbcode($doc, 'strong', array(), '[b]', '[/b]');
+	node2bbcode($doc, 'em', array(), '[i]', '[/i]');
+	node2bbcode($doc, 'b', array(), '[b]', '[/b]');
+	node2bbcode($doc, 'i', array(), '[i]', '[/i]');
+	node2bbcode($doc, 'u', array(), '[u]', '[/u]');
+
+	node2bbcode($doc, 'big', array(), "[size=large]", "[/size]");
+	node2bbcode($doc, 'small', array(), "[size=small]", "[/size]");
+
+	node2bbcode($doc, 'blockquote', array(), '[quote]', '[/quote]');
+
+	node2bbcode($doc, 'br', array(), "\n", '');
+
+	node2bbcode($doc, 'p', array('class'=>'MsoNormal'), "\n", "");
+	node2bbcode($doc, 'div', array('class'=>'MsoNormal'), "\r", "");
+
+	node2bbcode($doc, 'span', array(), "", "");
+
+	node2bbcode($doc, 'span', array(), "", "");
+	node2bbcode($doc, 'pre', array(), "", "");
+	node2bbcode($doc, 'div', array(), "\r", "\r");
+	node2bbcode($doc, 'p', array(), "\n", "\n");
+
+	node2bbcode($doc, 'ul', array(), "[list]", "[/list]");
+	node2bbcode($doc, 'ol', array(), "[list=1]", "[/list]");
+	node2bbcode($doc, 'li', array(), "[*]", "");
+
+	node2bbcode($doc, 'hr', array(), "[hr]", "");
+
+	node2bbcode($doc, 'table', array(), "", "");
+	node2bbcode($doc, 'tr', array(), "\n", "");
+	node2bbcode($doc, 'td', array(), "\t", "");
+
+	node2bbcode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
+	node2bbcode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
+	node2bbcode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
+	node2bbcode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
+	node2bbcode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
+	node2bbcode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
+
+	node2bbcode($doc, 'a', array('href'=>'/(.+)/'), '[url=$1]', '[/url]');
+
+	node2bbcode($doc, 'img', array('src'=>'/(.+)/', 'width'=>'/(\d+)/', 'height'=>'/(\d+)/'), '[img$2x$3]$1', '[/img]');
+	node2bbcode($doc, 'img', array('src'=>'/(.+)/'), '[img]$1', '[/img]');
+
+
+	node2bbcode($doc, 'video', array('src'=>'/(.+)/'), '[video]$1', '[/video]');
+	node2bbcode($doc, 'audio', array('src'=>'/(.+)/'), '[audio]$1', '[/audio]');
+	node2bbcode($doc, 'iframe', array('src'=>'/(.+)/'), '[iframe]$1', '[/iframe]');
+
+	node2bbcode($doc, 'code', array(), '[code]$1', '[/code]');
+
+	$message = $doc->saveHTML();
+
+	// I'm removing something really disturbing
+	// Don't know exactly what it is
+	$message = str_replace(chr(194).chr(160), ' ', $message);
+
+	$message = str_replace("&nbsp;", " ", $message);
+
+	// removing multiple DIVs
+	$message = preg_replace('=\r *\r=i', "\n", $message);
+	$message = str_replace("\r", "\n", $message);
+
+	call_hooks('html2bbcode', $message);
+
+	$message = strip_tags($message);
+
+	$message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
+
+	$message = str_replace(array("<"), array("&lt;"), $message);
+
+	// remove quotes if they don't make sense
+	$message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
+
+	$message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
+	$message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
+
+	do {
+		$oldmessage = $message;
+		$message = str_replace("\n \n", "\n\n", $message);
+	} while ($oldmessage != $message);
+
+	do {
+		$oldmessage = $message;
+		$message = str_replace("\n\n\n", "\n\n", $message);
+	} while ($oldmessage != $message);
+
+	$message = str_replace(array(
+					"[/size]\n\n",
+					"\n\n[hr]\n",
+					"\n[hr]\n\n",
+					"\n\n[list",
+					"[/list]\n\n",
+					"\n[/list]",
+					"[list]\n",
+					"[list=1]\n",
+					"\n\n[*]"),
+				array(
+					"[/size]\n",
+					"\n[hr]\n",
+					"\n[hr]\n",
+					"\n[list",
+					"[/list]\n",
+					"[/list]",
+					"[list]",
+					"[list=1]",
+					"\n[*]"),
+				$message);
+
+	$message = str_replace(array('[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'),
+		array('[b]', '[/b]', '[i]', '[/i]'), $message);
+
+	// Handling Yahoo style of mails
+	$message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
+
+	return(trim($message));
+}
+?>