Merge branch '290-favicons' into dev

4 files changed, 386 insertions, 32 deletions

diff --git a/lib/Favicon/DataAccess.php b/lib/Favicon/DataAccess.php
new file mode 100644
index 000000000..2bfdf640e
--- /dev/null
+++ b/lib/Favicon/DataAccess.php
@@ -0,0 +1,40 @@
+<?php
+
+namespace Favicon;
+
+/**
+ * DataAccess is a wrapper used to read/write data locally or remotly
+ * Aside from SOLID principles, this wrapper is also useful to mock remote resources in unit tests
+ * Note: remote access warning are silenced because we don't care if a website is unreachable
+ **/
+class DataAccess {
+	public function retrieveUrl($url) {
+	    $this->set_context();
+	    return @file_get_contents($url);
+	}
+	
+	public function retrieveHeader($url) {
+	    $this->set_context();
+		return @get_headers($url, TRUE);
+	}
+	
+    public function saveCache($file, $data) {
+        file_put_contents($file, $data);
+    }
+    
+    public function readCache($file) {
+    	return file_get_contents($file);
+    }
+    
+    private function set_context() {
+        stream_context_set_default(
+            array(
+                'http' => array(
+                    'method' => 'GET',
+                    'timeout' => 10,
+                    'header' => "User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:20.0; Favicon; +https://github.com/ArthurHoaro/favicon) Gecko/20100101 Firefox/32.0\r\n",
+                )
+            )
+        );
+    }
+}
\ No newline at end of file
diff --git a/lib/Favicon/Favicon.php b/lib/Favicon/Favicon.php
new file mode 100644
index 000000000..7ea6ccf16
--- /dev/null
+++ b/lib/Favicon/Favicon.php
@@ -0,0 +1,293 @@
+<?php
+
+namespace Favicon;
+
+class Favicon
+{
+    protected $url = '';
+    protected $cacheDir;
+    protected $cacheTimeout;
+    protected $dataAccess;
+
+    public function __construct($args = array())
+    {
+        if (isset($args['url'])) {
+            $this->url = $args['url'];
+        }
+        
+        $this->cacheDir = __DIR__ . '/../../resources/cache';
+        $this->dataAccess = new DataAccess();
+    }
+
+    public function cache($args = array()) {
+        if (isset($args['dir'])) {
+            $this->cacheDir = $args['dir'];
+        }
+
+        if (!empty($args['timeout'])) {
+                $this->cacheTimeout = $args['timeout'];
+        } else {
+                $this->cacheTimeout = 0;
+        }
+    }
+
+    public static function baseUrl($url, $path = false)
+    {
+        $return = '';
+
+        if (!$url = parse_url($url)) {
+            return FALSE;
+        }
+
+        // Scheme
+        $scheme = isset($url['scheme']) ? strtolower($url['scheme']) : null;
+        if ($scheme != 'http' && $scheme != 'https') {
+
+            return FALSE;
+        }
+        $return .= "{$scheme}://";
+
+        // Username and password
+        if (isset($url['user'])) {
+            $return .= $url['user'];
+            if (isset($url['pass'])) {
+                $return .= ":{$url['pass']}";
+            }
+            $return .= '@';
+        }
+
+        // Hostname
+        if( !isset($url['host']) ) {
+            return FALSE;
+        }
+        
+        $return .= $url['host'];
+
+        // Port
+        if (isset($url['port'])) {
+            $return .= ":{$url['port']}";
+        }
+
+        // Path
+        if( $path && isset($url['path']) ) {
+            $return .= $url['path'];
+        }
+        $return .= '/';
+
+        return $return;    
+    }
+
+    public function info($url)
+    {
+        if(empty($url) || $url === false) {
+            return false;
+        }
+        
+        $max_loop = 5;
+        
+        // Discover real status by following redirects. 
+        $loop = TRUE;
+        while ($loop && $max_loop-- > 0) {
+            $headers = $this->dataAccess->retrieveHeader($url);
+            $exploded = explode(' ', $headers[0]);
+            
+            if( !isset($exploded[1]) ) { 
+                return false;
+            }
+            list(,$status) = $exploded;
+            
+            switch ($status) {
+                case '301':
+                case '302':
+                    $url = $headers['Location'];
+                    break;
+                default:
+                    $loop = FALSE;
+                    break;
+            }
+        }
+
+        return array('status' => $status, 'url' => $url);
+    }
+    
+    public function endRedirect($url) {
+        $out = $this->info($url);
+        return !empty($out['url']) ? $out['url'] : false;
+    }
+
+    /**
+     * Find remote (or cached) favicon
+     * @return favicon URL, false if nothing was found
+     **/
+    public function get($url = '')
+    {
+        // URLs passed to this method take precedence.
+        if (!empty($url)) {
+            $this->url = $url;
+        }
+
+        // Get the base URL without the path for clearer concatenations.
+        $original = rtrim($this->baseUrl($this->url, true), '/');
+        $url = rtrim($this->endRedirect($this->baseUrl($this->url, false)), '/');
+
+        if(($favicon = $this->checkCache($url)) || ($favicon = $this->getFavicon($url))) {
+            $base = true;
+        }
+        elseif(($favicon = $this->checkCache($original)) || ($favicon = $this->getFavicon($original, false))) {
+            $base = false;    
+        }
+        else
+            return false;
+            
+        // Save cache if necessary
+        $cache = $this->cacheDir . '/' . md5($base ? $url : $original);
+        if ($this->cacheTimeout && !file_exists($cache) || (is_writable($cache) && time() - filemtime($cache) > $this->cacheTimeout)) {
+            $this->dataAccess->saveCache($cache, $favicon);
+        }
+        
+        return $favicon;
+    }
+    
+    private function getFavicon($url, $checkDefault = true) {
+        $favicon = false;
+        
+        if(empty($url)) {
+            return false;
+        }
+        
+        // Try /favicon.ico first.
+        if( $checkDefault ) {
+            $info = $this->info("{$url}/favicon.ico");
+            if ($info['status'] == '200') {
+                $favicon = $info['url'];
+            }
+        }
+
+        // See if it's specified in a link tag in domain url.
+        if (!$favicon) {
+            $favicon = $this->getInPage($url);
+        }
+        
+        // Make sure the favicon is an absolute URL.
+        if( $favicon && filter_var($favicon, FILTER_VALIDATE_URL) === false ) {
+            $favicon = $url . '/' . $favicon;
+        }
+
+        // Sometimes people lie, so check the status.
+        // And sometimes, it's not even an image. Sneaky bastards!
+        // If cacheDir isn't writable, that's not our problem
+        if ($favicon && is_writable($this->cacheDir) && !$this->checkImageMType($favicon)) {
+            $favicon = false;
+        }
+
+        return $favicon;
+    }
+    
+    private function getInPage($url) {
+        $html = $this->dataAccess->retrieveUrl("{$url}/");
+        preg_match('!<head.*?>.*</head>!ims', $html, $match);
+        
+        if(empty($match) || count($match) == 0) {
+            return false;
+        }
+        
+        $head = $match[0];
+        
+        $dom = new \DOMDocument();
+        // Use error supression, because the HTML might be too malformed.
+        if (@$dom->loadHTML($head)) {
+            $links = $dom->getElementsByTagName('link');
+            foreach ($links as $link) {
+                if ($link->hasAttribute('rel') && strtolower($link->getAttribute('rel')) == 'shortcut icon') {
+                    return $link->getAttribute('href');
+                } elseif ($link->hasAttribute('rel') && strtolower($link->getAttribute('rel')) == 'icon') {
+                    return $link->getAttribute('href');
+                } elseif ($link->hasAttribute('href') && strpos($link->getAttribute('href'), 'favicon') !== FALSE) {
+                    return $link->getAttribute('href');
+                }
+            }
+        }
+        return false;
+    }
+    
+    private function checkCache($url) {
+        if ($this->cacheTimeout) {
+            $cache = $this->cacheDir . '/' . md5($url);
+            if (file_exists($cache) && is_readable($cache) && (time() - filemtime($cache) < $this->cacheTimeout)) {
+                return $this->dataAccess->readCache($cache);
+            }
+        } 
+        return false;
+    }
+    
+    private function checkImageMType($url) {
+        $tmpFile = $this->cacheDir . '/tmp.ico';
+        
+        $fileContent = $this->dataAccess->retrieveUrl($url);
+        $this->dataAccess->saveCache($tmpFile, $fileContent);
+        
+        $finfo = finfo_open(FILEINFO_MIME_TYPE);
+        $isImage = strpos(finfo_file($finfo, $tmpFile), 'image') !== false;
+        finfo_close($finfo);
+        
+        unlink($tmpFile);
+        
+        return $isImage;
+    }
+    
+    /**
+     * @return mixed
+     */
+    public function getCacheDir()
+    {
+        return $this->cacheDir;
+    }
+
+    /**
+     * @param mixed $cacheDir
+     */
+    public function setCacheDir($cacheDir)
+    {
+        $this->cacheDir = $cacheDir;
+    }
+
+    /**
+     * @return mixed
+     */
+    public function getCacheTimeout()
+    {
+        return $this->cacheTimeout;
+    }
+
+    /**
+     * @param mixed $cacheTimeout
+     */
+    public function setCacheTimeout($cacheTimeout)
+    {
+        $this->cacheTimeout = $cacheTimeout;
+    }
+
+    /**
+     * @return string
+     */
+    public function getUrl()
+    {
+        return $this->url;
+    }
+
+    /**
+     * @param string $url
+     */
+    public function setUrl($url)
+    {
+        $this->url = $url;
+    }
+
+    /**
+     * @param DataAccess $dataAccess
+     */
+    public function setDataAccess($dataAccess)
+    {
+        $this->dataAccess = $dataAccess;
+    }
+}
diff --git a/p/f.php b/p/f.php
index 5db9ab213..d00b36eb0 100644
--- a/p/f.php
+++ b/p/f.php
@@ -1,36 +1,59 @@
 <?php
+
 require('../constants.php');
+
+include(LIB_PATH . '/Favicon/Favicon.php');
+include(LIB_PATH . '/Favicon/DataAccess.php');
+
+
 $favicons_dir = DATA_PATH . '/favicons/';
+$default_favicon = PUBLIC_PATH . '/themes/icons/default_favicon.png';
+
 
 /* Télécharge le favicon d'un site et le place sur le serveur */
-function download_favicon ($website, $dest) {
-	$ok = false;
-	$url = 'http://g.etfv.co/' . $website;
-
-	$c = curl_init ($url);
-	curl_setopt ($c, CURLOPT_HEADER, false);
-	curl_setopt ($c, CURLOPT_RETURNTRANSFER, true);
-	curl_setopt ($c, CURLOPT_BINARYTRANSFER, true);
-	$imgRaw = curl_exec ($c);
-
-	if (curl_getinfo ($c, CURLINFO_HTTP_CODE) == 200) {
-		$file = fopen ($dest, 'w');
+function download_favicon($website, $dest) {
+	global $favicons_dir;
+
+	$favicon_getter = new \Favicon\Favicon();
+	$favicon_getter->setCacheDir($favicons_dir);
+	$favicon_url = $favicon_getter->get($website);
+
+	if ($favicon_url === false) {
+		return false;
+	}
+
+	$c = curl_init($favicon_url);
+	curl_setopt($c, CURLOPT_HEADER, false);
+	curl_setopt($c, CURLOPT_RETURNTRANSFER, true);
+	curl_setopt($c, CURLOPT_BINARYTRANSFER, true);
+	$img_raw = curl_exec($c);
+	$status_code = curl_getinfo($c, CURLINFO_HTTP_CODE);
+	curl_close($c);
+
+	if ($status_code === 200) {
+		$file = fopen($dest, 'w');
 		if ($file !== false) {
-			fwrite ($file, $imgRaw);
-			fclose ($file);
-			$ok = true;
+			fwrite($file, $img_raw);
+			fclose($file);
+			return true;
 		}
 	}
-	curl_close ($c);
-	if (!$ok) {
-		header('Location: ' . $url);
-		return false;
-	}
-	return true;
+
+	return false;
 }
 
-$id = isset($_SERVER['QUERY_STRING']) ? $_SERVER['QUERY_STRING'] : '0';
 
+function show_default_favicon() {
+	global $default_favicon;
+
+	header('HTTP/1.1 404 Not Found');
+	header('Content-Type: image/png');
+	readfile($default_favicon);
+	die();
+}
+
+
+$id = isset($_SERVER['QUERY_STRING']) ? $_SERVER['QUERY_STRING'] : '0';
 if (!ctype_xdigit($id)) {
 	$id = '0';
 }
@@ -38,19 +61,17 @@ if (!ctype_xdigit($id)) {
 $txt = $favicons_dir . $id . '.txt';
 $ico = $favicons_dir . $id . '.ico';
 
-$icoMTime = @filemtime($ico);
-$txtMTime = @filemtime($txt);
+$ico_mtime = @filemtime($ico);
+$txt_mtime = @filemtime($txt);
 
-if (($icoMTime == false) || ($txtMTime > $icoMTime)) {
-	if ($txtMTime == false) {
-		header('HTTP/1.1 404 Not Found');
-		header('Content-Type: image/gif');
-		readfile(PUBLIC_PATH . '/themes/icons/grey.gif');	//TODO: Better 404 favicon
-		die();
+if (($ico_mtime == false) || ($txt_mtime > $ico_mtime)) {
+	if ($txt_mtime == false) {
+		show_default_favicon();
 	}
+
 	$url = file_get_contents($txt);
 	if (!download_favicon($url, $ico)) {
-		die();
+		show_default_favicon();
 	}
 }
 
@@ -59,6 +80,6 @@ require(LIB_PATH . '/http-conditional.php');
 header('Content-Type: image/x-icon');
 header('Content-Disposition: inline; filename="' . $id . '.ico"');
 
-if (!httpConditional($icoMTime, 2592000, 2)) {
+if (!httpConditional($ico_mtime, 2592000, 2)) {
 	readfile($ico);
 }
diff --git a/p/themes/icons/default_favicon.png b/p/themes/icons/default_favicon.png
new file mode 100644
index 000000000..1e660b740
--- /dev/null
+++ b/p/themes/icons/default_favicon.png