From 31653aef64c94ade956b552dad65bfe15f0ad491 Mon Sep 17 00:00:00 2001 From: Michael Date: Sun, 19 Feb 2012 20:36:59 +0100 Subject: Brand new html2bbcode. Some other changes in diaspora to bbcode handling --- include/html2bbcode.php | 390 +++++++++++++++++++++++++++++++++--------------- 1 file changed, 271 insertions(+), 119 deletions(-) (limited to 'include/html2bbcode.php') 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 @@ (.*?)\<\/pre\>/is', - '/\/is', - '/\<\/p\>/is', - '/\(.*?)\<\/b\>/is', - '/\(.*?)\<\/i\>/is', - '/\(.*?)\<\/u\>/is', - '/\(.*?)\<\/ul\>/is', - '/\(.*?)\<\/li\>/is', - '/\/is', - '/\/is', - '/\/is', - '/\/is', - '/\/is', - '/\(.*?)\<\/div\>/is', - '/\/is', - '/\(.*?)\<\/strong\>/is', - '/\(.*?)\<\/a\>/is', - '/\(.*?)\<\/code\>/is', - '/\(.*?)\<\/span\>/is', - '/\(.*?)\<\/span\>/is', - '/\(.*?)\<\/blockquote\>/is', - '/\(.*?)\<\/video\>/is', - '/\(.*?)\<\/audio\>/is', - '/\(.*?)\<\/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, $node) { - // First, check for 
 tag
-    if(strpos($string, '
') === false)
-    {
-        return str_replace("\n","", $string);
-    }
-
-    // If there is a 
, 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 
. If it does, set $preFound to true
-        if(strpos($line, "
") !== false)
-        {
-            $preFound=true;
-        }
-        elseif(strpos($line, "
") !== 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(
+					"
  • ",
+					"
    • "),
+				array(
+					"
  • ",
+					"
    • "),
+				$message);
+
+	// remove namespaces
+	$message = preg_replace('=<(\w+):(.+?)>=', '', $message);
+	$message = preg_replace('==', '', $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("<", ">", "
    ", " ", ""), $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(" ", " ", $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("<"), $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));
+}
+?>
-- 
cgit v1.2.3


From 064c843119576c1345a61a531dfaa1076a54f3f0 Mon Sep 17 00:00:00 2001
From: Michael 
Date: Sun, 19 Feb 2012 21:07:45 +0100
Subject: Handling nested bbcodes. Some small design changes in html2bbcode.
 Change in poller.php is just for testing purposes.

---
 include/html2bbcode.php | 35 +++++++++++++++++++++--------------
 1 file changed, 21 insertions(+), 14 deletions(-)

(limited to 'include/html2bbcode.php')

diff --git a/include/html2bbcode.php b/include/html2bbcode.php
index 1b4b5e99a..4297ff2f9 100755
--- a/include/html2bbcode.php
+++ b/include/html2bbcode.php
@@ -187,9 +187,13 @@ function html2bbcode($message)
 
 	node2bbcode($doc, 'hr', array(), "[hr]", "");
 
-	node2bbcode($doc, 'table', array(), "", "");
-	node2bbcode($doc, 'tr', array(), "\n", "");
-	node2bbcode($doc, 'td', array(), "\t", "");
+	//node2bbcode($doc, 'table', array(), "", "");
+	//node2bbcode($doc, 'tr', array(), "\n", "");
+	//node2bbcode($doc, 'td', array(), "\t", "");
+	node2bbcode($doc, 'table', array(), "[table]", "[/table]");
+	node2bbcode($doc, 'th', array(), "[th]", "[/th]");
+	node2bbcode($doc, 'tr', array(), "[tr]", "[/tr]");
+	node2bbcode($doc, 'td', array(), "[td]", "[/td]");
 
 	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");
@@ -246,27 +250,30 @@ function html2bbcode($message)
 		$message = str_replace("\n\n\n", "\n\n", $message);
 	} while ($oldmessage != $message);
 
-	$message = str_replace(array(
+	do {
+		$oldmessage = $message;
+		$message = str_replace(array(
 					"[/size]\n\n",
-					"\n\n[hr]\n",
-					"\n[hr]\n\n",
-					"\n\n[list",
-					"[/list]\n\n",
+					"\n[hr]",
+					"[hr]\n",
+					"\n[list",
+					"[/list]\n",
 					"\n[/list]",
 					"[list]\n",
 					"[list=1]\n",
-					"\n\n[*]"),
+					"\n[*]"),
 				array(
 					"[/size]\n",
-					"\n[hr]\n",
-					"\n[hr]\n",
-					"\n[list",
-					"[/list]\n",
+					"[hr]",
+					"[hr]",
+					"[list",
+					"[/list]",
 					"[/list]",
 					"[list]",
 					"[list=1]",
-					"\n[*]"),
+					"[*]"),
 				$message);
+	} while ($message != $oldmessage);
 
 	$message = str_replace(array('[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'),
 		array('[b]', '[/b]', '[i]', '[/i]'), $message);
-- 
cgit v1.2.3