Blogger import improvements and news_title and news_body field size limitations removed. (DB Update Required)

e107_core/sql/core_sql.php

@@ -300,9 +300,9 @@ CREATE TABLE menus (
 
 CREATE TABLE news (
   news_id int(10) unsigned NOT NULL auto_increment,
-  news_title varchar(200) NOT NULL default '',
+  news_title varchar(255) NOT NULL default '',
   news_sef varchar(200) NOT NULL default '',
-  news_body text NOT NULL,
+  news_body longtext NOT NULL,
   news_extended text NOT NULL,
   news_meta_keywords  varchar(255) NOT NULL default '',
   news_meta_description text NOT NULL,

e107_plugins/import/providers/blogger_import_class.php

@@ -26,10 +26,6 @@
 // require_once('import_classes.php');
 require_once('rss_import_class.php');
 
-//$import_class_names['blogger_import'] 		= 'Blogger';
-//$import_class_comment['blogger_import'] 	= 'Import up to 500 items from yourblog.blogspot.com';
-//$import_class_support['blogger_import'] 	= array('news');
-//$import_default_prefix['blogger_import'] 	= '';
 
 class blogger_import extends rss_import
 {	
@@ -40,9 +36,7 @@ class blogger_import extends rss_import
 	
 	public $cleanupHtml = false;
 	public $defaultClass = false;
-	
+
-	
-	
 	
 	/*
 	 If the first 500 posts of your blog feed are here:
@@ -56,7 +50,13 @@ class blogger_import extends rss_import
 	function init()
 	{
 		$mes = e107::getMessage();
-	
+
+		if(E107_DEBUG_LEVEL > 0)
+		{
+			$this->action = 'preview'; // changes default action to 'preview' method below. (for testing)
+		}
+
+
 		if(vartrue($_POST['bloggerUrl']))
 		{			
 			$this->feedUrl = rtrim($_POST['bloggerUrl'],"/")."/feeds/posts/default?max-results=999&alt=rss";	
@@ -65,20 +65,26 @@ class blogger_import extends rss_import
 		if(vartrue($_POST['bloggerCleanup']))
 		{
 			$this->cleanupHtml = true;
-		}	
+		}
-		
+
 		$mes->addDebug("Blogger Feed:".$this->feedUrl);
+
+		if(!empty($_POST['preview']))
+		{
+			$this->preview();
+			return;
+		}
 	}
 		
 	
 	function config()
 	{
 		$var[0]['caption']	= "Blogger URL";
-		$var[0]['html'] 	= "<input class='tbox' type='text' name='bloggerUrl' size='80' value='{$_POST['bloggerUrl']}' maxlength='250' />";
+		$var[0]['html'] 	= e107::getForm()->text('bloggerUrl', $_POST['bloggerUrl'],255, 'size=xxlarge'); //  "<input class='tbox' type='text' name='bloggerUrl' size='120' value='{$_POST['bloggerUrl']}' maxlength='250' />";
 		$var[0]['help']		= "eg. http://blogname.blogspot.com";
 		
 		$var[1]['caption']	= "Cleanup HTML in content";
-		$var[1]['html'] 	= "<input class='tbox' type='checkbox' name='bloggerCleanup' size='80' value='1' />";
+		$var[1]['html'] 	= e107::getForm()->checkbox('bloggerCleanup',1, $_POST['bloggerCleanup']); // "<input class='tbox' type='checkbox' name='bloggerCleanup' value='1' />";
 		$var[1]['help']		= "Tick to enable";
 		
 		return $var;
@@ -86,16 +92,55 @@ class blogger_import extends rss_import
 
 	function process($type,$source)
 	{
-				
+
+		$tp = e107::getParser();
+
+		$allowedTags        = array('html', 'body', 'a','img','table','tr', 'td', 'th', 'tbody', 'thead', 'colgroup', 'b',
+		'i', 'pre','code', 'strong', 'u', 'em','ul', 'ol', 'li','img','h1','h2','h3','h4','h5','h6','p','iframe',
+		'div','pre','section','article', 'blockquote','hgroup','aside','figure', 'video', 'span', 'br',
+		'small', 'caption', 'noscript'
+		);
+
+		$allowedAttributes  = array(
+			'default'   => array('id'),
+			'img'       => array('id', 'src', 'style', 'class', 'alt', 'title', 'width', 'height'),
+			'a'         => array('id', 'href', 'class', 'title', 'target'),
+			'script'	=> array('type', 'src', 'language'),
+			'iframe'	=> array('id', 'src', 'frameborder', 'class', 'width', 'height', 'style')
+		);
+
+		$tp->setAllowedTags($allowedTags);
+		$tp->setAllowedAttributes($allowedAttributes);
+
 		switch ($type) 
 		{
 			case 'description':
 				$body = $source[$type][0];
-				if($this->cleanupHtml == TRUE)
+
+				if($this->cleanupHtml == true)
 				{
-					$body = preg_replace("/font-family: [\w]*;/i","", $body);
+				//	$body = preg_replace("/font-family: [\w]*;/i","", $body);
-					$body = preg_replace('/class="[\w]*" /i',"", $body);
+			//		$body = preg_replace('/class="[\w]*" /i',"", $body);
 					$body = str_replace("<br>","<br />",$body);
+
+					$body = $tp->cleanHtml($body,false);
+
+					$patterns = array();
+					$patterns[0] = '/<div[^>]*>/';
+					$patterns[1] = '/<\/div>/';
+
+					$replacements = array();
+					$replacements[2] = '';
+					$replacements[1] = '';
+
+					$body = preg_replace($patterns, $replacements, $body);
+
+					$srch = array('<span>', '</span>', '<br /><br /><br /><br />');
+					$repl = array("", "", "<br /><br />");
+
+					$body = str_replace($srch, $repl, $body);
+
+
 					return $body;
 				}
 				else 
@@ -103,6 +148,16 @@ class blogger_import extends rss_import
 					return $body;
 				}		
 			break;
+
+			case 'sef':
+
+				if(!empty($source['link'][0]))
+				{
+					return str_replace(".html","", basename($source['link'][0]));
+				}
+
+				return "";
+			break;
 			
 			default:
 				return $source[$type][0];
@@ -112,6 +167,28 @@ class blogger_import extends rss_import
 		
 	}
 
+
+	function preview()
+	{
+		$file = $this->feedUrl;
+		$array = e107::getXml()->loadXMLfile($file,'advanced');
+
+	//	print_a($array); // raw xml data;
+
+
+		foreach($array['channel']['item'] as $src)
+		{
+		//	$data = $this->process('description', $src);
+		//	$this->process('sef',$src);
+			$data = $this->copyNewsData($target,$src);
+			print_a($data);
+
+		}
+
+
+	}
+
+
 	//TODO Comment Import: 
 	//http://blogname.blogspot.com/feeds/comments/default?alt=rss
 	

e107_plugins/import/providers/rss_import_class.php

@@ -149,6 +149,7 @@ class rss_import extends base_import_class
 				
 		$body 			= $this->saveImages($body,'news');
 		$keywords 		= $this->process('category',$source);
+		$sef            = $this->process('sef',$source);
 							
 		if(!vartrue($source['title'][0]))
 		{
@@ -165,7 +166,7 @@ class rss_import extends base_import_class
 		}
 		
 		$target['news_title']					= $title;
-		//	$target['news_sef']					= $source['post_name'];
+		$target['news_sef']					    = $sef;
 		$target['news_body']					= "[html]".$body."[/html]";
 		//	$target['news_extended']			= '';
 		$target['news_meta_keywords']			= implode(",",$keywords);
@@ -223,6 +224,10 @@ class rss_import extends base_import_class
 					return $keywords;		
 				}
 			break;
+
+			case 'sef':
+				return '';
+			break;
 			
 			default:
 				return varset($source[$type][0]);