get favicon from icon atom feeds & search for icons not only in html head (closes #457)

2 files changed, 72 insertions, 55 deletions

diff --git a/include/functions.php b/include/functions.php
index 27c24f8ed..8f31241b8 100644
--- a/include/functions.php
+++ b/include/functions.php
@@ -332,9 +332,9 @@
 	 * @access public
 	 * @return mixed The favicon URL, or false if none was found.
 	 */
-	function get_favicon_url($url) {
+	function get_favicon_urls($url) {
 
-		$favicon_url = false;
+		$favicons = array();
 
 		if ($html = @fetch_file_contents($url)) {
 
@@ -350,68 +350,78 @@
 				break;
 			}
 
-			$entries = $xpath->query('/html/head/link[@rel="shortcut icon" or @rel="icon"]');
+			# Limiting the search to head will not find when in body
+			$entries = $xpath->query('//link[@rel="shortcut icon" or @rel="icon"]');
 			if (count($entries) > 0) {
 				foreach ($entries as $entry) {
-					$favicon_url = rewrite_relative_url($url, $entry->getAttribute("href"));
-					break;
+					array_push($favicons, rewrite_relative_url($url, $entry->getAttribute("href")));
 				}
 			}
 		}
 
-		if (!$favicon_url)
-			$favicon_url = rewrite_relative_url($url, "/favicon.ico");
-
-		return $favicon_url;
-	} // function get_favicon_url
+		array_push($favicons, rewrite_relative_url($url, "/favicon.ico"));
 
-	function check_feed_favicon($site_url, $feed, $link) {
-#		print "FAVICON [$site_url]: $favicon_url\n";
+		return $favicons;
+	} // function get_favicon_urls
 
+	function validate_favicon($url, $feed) {
+		// Limiting to "image" type misses those served with text/plain
+		$contents = fetch_file_contents($url); // , "image");
+		
+		if (!$contents)
+			return false;
+		
+		// Crude image type matching.
+		// Patterns gleaned from the file(1) source code.
+		if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
+			// 0       string  \000\000\001\000        MS Windows icon resource
+			//error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
+		}
+		elseif (preg_match('/^GIF8/', $contents)) {
+			// 0       string          GIF8            GIF image data
+			//error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
+		}
+		elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
+			// 0       string          \x89PNG\x0d\x0a\x1a\x0a         PNG image data
+			//error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
+		}
+		elseif (preg_match('/^\xff\xd8/', $contents)) {
+			// 0       beshort         0xffd8          JPEG image data
+			//error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
+		}
+		else {
+			//error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
+			return false;
+		}
+		
 		$icon_file = ICONS_DIR . "/$feed.ico";
+		$fp = @fopen($icon_file, "w");
+		if ($fp) {
+			fwrite($fp, $contents);
+			fclose($fp);
+			chmod($icon_file, 0644);
+		}
+		return true;		
+	}
+	
+	function check_feed_favicon($site_url, $feed, $link, $atom_icon) {
+#		print "FAVICON [$site_url]: $favicon_url\n";
 
-		if (!file_exists($icon_file)) {
-			$favicon_url = get_favicon_url($site_url);
-
-			if ($favicon_url) {
-				// Limiting to "image" type misses those served with text/plain
-				$contents = fetch_file_contents($favicon_url); // , "image");
-
-				if ($contents) {
-					// Crude image type matching.
-					// Patterns gleaned from the file(1) source code.
-					if (preg_match('/^\x00\x00\x01\x00/', $contents)) {
-						// 0       string  \000\000\001\000        MS Windows icon resource
-						//error_log("check_feed_favicon: favicon_url=$favicon_url isa MS Windows icon resource");
-					}
-					elseif (preg_match('/^GIF8/', $contents)) {
-						// 0       string          GIF8            GIF image data
-						//error_log("check_feed_favicon: favicon_url=$favicon_url isa GIF image");
-					}
-					elseif (preg_match('/^\x89PNG\x0d\x0a\x1a\x0a/', $contents)) {
-						// 0       string          \x89PNG\x0d\x0a\x1a\x0a         PNG image data
-						//error_log("check_feed_favicon: favicon_url=$favicon_url isa PNG image");
-					}
-					elseif (preg_match('/^\xff\xd8/', $contents)) {
-						// 0       beshort         0xffd8          JPEG image data
-						//error_log("check_feed_favicon: favicon_url=$favicon_url isa JPG image");
-					}
-					else {
-						//error_log("check_feed_favicon: favicon_url=$favicon_url isa UNKNOWN type");
-						$contents = "";
-					}
-				}
-
-				if ($contents) {
-					$fp = @fopen($icon_file, "w");
-
-					if ($fp) {
-						fwrite($fp, $contents);
-						fclose($fp);
-						chmod($icon_file, 0644);
-					}
-				}
-			}
+		if(!empty($atom_icon) && validate_favicon($atom_icon, $feed))
+			return;
+		
+		$favicon_urls = array($atom_icon);
+		
+		$favicon_urls = array_unique(array_merge($favicon_urls, get_favicon_urls($site_url)));
+		for ($i = 1; $i < count($favicon_urls); $i++) {
+			if (validate_favicon($favicon_urls[$i], $feed))
+				return;
+		}
+		
+		$favicon_urls = array_unique(array_merge($favicon_urls, get_favicon_urls(rewrite_relative_url($link, "/"))));
+		for (; $i < count($favicon_urls); $i++) {
+			if (validate_favicon($favicon_urls[$i], $feed))
+				return;
 		}
 	}
 
diff --git a/include/rssfuncs.php b/include/rssfuncs.php
index 48a246d88..cb985197c 100644
--- a/include/rssfuncs.php
+++ b/include/rssfuncs.php
@@ -441,7 +441,14 @@
 				_debug("update_rss_feed: checking favicon...");
 			}
 
-			check_feed_favicon($site_url, $feed, $link);
+			if (!file_exists(ICONS_DIR . "/$feed.ico")) {
+				if ($use_simplepie) {
+					$atom_icon = $rss->get_favicon();
+				} else {
+					$atom_icon = $rss->channel["icon"];
+				}
+					check_feed_favicon($site_url, $feed, $link, $atom_icon);
+			}
 
 			if (!$registered_title || $registered_title == "[Unknown]") {