From e5992003e88545ffe5df0acc57c850733facdd5e Mon Sep 17 00:00:00 2001
From: Friendika <info@friendika.com>
Date: Fri, 28 Oct 2011 02:50:00 -0700
Subject: [PATCH] better handling of remote urls - especially FB opengraph

---
 include/bb2diaspora.php |  6 ++--
 include/diaspora.php    | 10 ++++--
 mod/parse_url.php       | 72 +++++++++++++++++++++++++++++++++++++++--
 3 files changed, 82 insertions(+), 6 deletions(-)

diff --git a/include/bb2diaspora.php b/include/bb2diaspora.php
index efbf347587..aec2b74bc6 100644
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@@ -22,7 +22,7 @@ function diaspora2bb($s) {
 }
 
 
-function scale_diaspora_images($s) {
+function scale_diaspora_images($s,$include_link = true) {
 
 	$matches = null;
 	$c = preg_match_all('/\[img\](.*?)\[\/img\]/ism',$s,$matches,PREG_SET_ORDER);
@@ -38,7 +38,9 @@ function scale_diaspora_images($s) {
 						$new_width = $ph->getWidth();
 						$new_height = $ph->getHeight();
 						$s = str_replace($mtch[0],'[img=' . $new_width . 'x' . $new_height. ']' . $mtch[1] . '[/img]'
-							. "\n" . '[url=' . $mtch[1] . ']' . t('view full size') . '[/url]' . "\n",$s);
+							. "\n" . ((! $include_link) 
+								? '[url=' . $mtch[1] . ']' . t('view full size') . '[/url]' . "\n"
+								: ''),$s);
 					}
 				}
 			}
diff --git a/include/diaspora.php b/include/diaspora.php
index ad5e2adb2a..712ddd7b08 100644
--- a/include/diaspora.php
+++ b/include/diaspora.php
@@ -635,12 +635,18 @@ function diaspora_reshare($importer,$xml) {
 	$x = str_replace(array('<activity_streams-photo>','</activity_streams-photo>'),array('<asphoto>','</asphoto>'),$x);
 	$source_xml = parse_xml_string($x,false);
 
-	if(strlen($source_xml->post->asphoto->objectId) && ($source_xml->post->asphoto->objectId != 0) && ($source_xml->post->asphoto->image_url))
+	if(strlen($source_xml->post->asphoto->objectId) && ($source_xml->post->asphoto->objectId != 0) && ($source_xml->post->asphoto->image_url)) {
 		$body = '[url=' . notags(unxmlify($source_xml->post->asphoto->image_url)) . '][img]' . notags(unxmlify($source_xml->post->asphoto->objectId)) . '[/img][/url]' . "\n";
-	elseif($source_xml->post->asphoto->image_url)
+		$body = scale_diaspora_images($body,false);
+	}
+	elseif($source_xml->post->asphoto->image_url) {
 		$body = '[img]' . notags(unxmlify($source_xml->post->asphoto->image_url)) . '[/img]' . "\n";
+		$body = scale_diaspora_images($body);
+	}
 	elseif($source_xml->post->status_message) {
 		$body = diaspora2bb($source_xml->post->status_message->raw_message);
+		$body = scale_diaspora_images($body);
+
 	}
 	else {
 		logger('diaspora_reshare: no reshare content found: ' . print_r($source_xml,true));
diff --git a/mod/parse_url.php b/mod/parse_url.php
index 7a520ba92f..2df9de4763 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -46,6 +46,7 @@ function parse_url_content(&$a) {
 		killme();
 	}
 
+
 	if($url && $title && $text) {
 
 		$text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
@@ -67,13 +68,26 @@ function parse_url_content(&$a) {
 		killme();
 	}
 
-	logger('parse_url: data: ' . $s, LOGGER_DATA);
+//	logger('parse_url: data: ' . $s, LOGGER_DATA);
 
 	if(! $s) {
 		echo sprintf($template,$url,$url,'') . $str_tags;
 		killme();
 	}
 
+	$matches = '';
+	$c = preg_match('/\<head(.*?)\>(.*?)\<\/head\>/ism',$s,$matches);
+	if($c) {
+//		logger('parse_url: header: ' . $matches[2], LOGGER_DATA);
+		try {
+			$domhead = HTML5_Parser::parse($matches[2]);
+		} catch (DOMException $e) {
+			logger('scrape_dfrn: parse error: ' . $e);
+		}
+		if($domhead)
+			logger('parsed header');
+	}
+
 	if(! $title) {
 		if(strpos($s,'<title>')) {
 			$title = substr($s,strpos($s,'<title>')+7,64);
@@ -84,10 +98,11 @@ function parse_url_content(&$a) {
 
 	$config = HTMLPurifier_Config::createDefault();
 	$config->set('Cache.DefinitionImpl', null);
-
 	$purifier = new HTMLPurifier($config);
 	$s = $purifier->purify($s);
 
+//	logger('purify_output: ' . $s);
+
 	try {
 		$dom = HTML5_Parser::parse($s);
 	} catch (DOMException $e) {
@@ -157,10 +172,63 @@ function parse_url_content(&$a) {
 		}
 	}
 
+	if(! $text) {
+		logger('parsing meta');
+		$items = $domhead->getElementsByTagName('meta');
+		if($items) {
+			foreach($items as $item) {
+				$property = $item->getAttribute('property');
+				if($property && (stristr($property,':description'))) {
+
+					$text = $item->getAttribute('content');
+					if(stristr($text,'<script')) {
+						$text = '';
+						continue;
+					}
+					$text = strip_tags($text);
+
+
+					$text = substr($text,0,250) . '...' ;
+				}
+				if($property && (stristr($property,':image'))) {
+
+					$image = $item->getAttribute('content');
+					if(stristr($text,'<script')) {
+						$image = '';
+						continue;
+					}
+					$image = strip_tags($image);
+					
+					$i = fetch_url($image);
+					if($i) {
+						require_once('include/Photo.php');
+						$ph = new Photo($i);
+						if($ph->is_valid()) {
+							if($ph->getWidth() > 300 || $ph->getHeight() > 300) {
+								$ph->scaleImage(300);
+								$new_width = $ph->getWidth();
+								$new_height = $ph->getHeight();
+								$image = '<br /><br /><img height="' . $new_height . '" width="' . $new_width . '" src="' .$image . '" alt="photo" />';
+							}
+							else
+								$image = '<br /><br /><img src="' . $image . '" alt="photo" />';
+						}
+						else
+							$image = '';
+					
+					}
+				}
+			}
+		}
+	}
+
 	if(strlen($text)) {
 		$text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 	}
 
+	if($image) {
+		$text = $image . '<br />' . $text;
+	}
 	$title = str_replace(array("\r","\n"),array('',''),$title);
 
 	$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;