[add] show favicon of site

[add] fetch and cache favicon

inc/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",
+                )
+            )
+        );
+    }
+}

inc/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;
+    }
+}

inc/shaarli.css

@@ -69,9 +69,9 @@ h1 { font-size:20pt; font-weight:bold; font-style:italic; margin-bottom:20px; }
 
 /* Small tab on the left of each link with edit/delete buttons. */
 .button_edit, .button_delete { border-radius:0; box-shadow:none; border-style:none; border-width:0; padding:0; background:none; }
-.linkeditbuttons {  
+.linkeditbuttons {
-	position:absolute; 
+	position:absolute;
-	left:-1px;  
+	left:-1px;
 	padding:4px 2px 2px 2px;
 	background-color:#f0f0f0;
 
@@ -111,7 +111,7 @@ cursor:pointer;
     box-shadow: 0 1px 2px rgba(0, 0, 0, 0.5);
 	width:auto;
 	padding:0 10px 5px 10px;
-	margin: auto; 
+	margin: auto;
 }
 
 #pageheader a
@@ -282,7 +282,7 @@ font-size:9pt;
 .thumbnail { float:left;  margin-right: 10px;   }
 .linkcontainer { position: static; margin-left:130px; }
 */
-
+.favicon {width:16px;height:16px;margin-right:0.1em;}
 
 
 
@@ -305,7 +305,7 @@ font-size:9pt;
   background-color: transparent;
   background-color: rgba(0, 0, 0, 0.4);  /* FF3+, Saf3+, Opera 10.10+, Chrome, IE9 */
             filter: progid:DXImageTransform.Microsoft.gradient(startColorstr=#66000000,endColorstr=#66000000); /* IE6–IE9 */
-text-shadow:2px 2px 1px #000000;    
+text-shadow:2px 2px 1px #000000;
 }
 
 /* Minimal customisation for jQuery widgets */
@@ -337,7 +337,7 @@ box-shadow:2px 2px 20px 2px #333333;
 
 div.daily
 {
-    font-family: Georgia, 'DejaVu Serif', Norasi, serif; 
+    font-family: Georgia, 'DejaVu Serif', Norasi, serif;
     background-color: #E6D6BE;
     /* Background paper texture by BashCorpo:
        http://www.bashcorpo.dk/textures.php
@@ -356,7 +356,7 @@ div.daily
 #daily_col3 { float:left;position:relative; width:33%;}
 
 div.dailyAbout
-{ 
+{
     float:left;
     border: 1px solid black;
     font-size: 8pt;
@@ -364,21 +364,21 @@ div.dailyAbout
     left:10px;
     top: 15px;
     padding: 5px 5px 5px 5px;
-    text-align:center; 
+    text-align:center;
 }
 div.dailyAbout  a { color: #890500; }
 div.dailyTitle
- { 
+ {
      font-weight: bold;
      font-size: 44pt;
-     text-align:center; 
+     text-align:center;
      padding:10px 20px 0px 20px;
 }
 div.dailyDate
- { 
+ {
      font-size: 12pt;
-     font-weight:bold; 
+     font-weight:bold;
-     text-align:center; 
+     text-align:center;
      padding:0px 20px 30px 20px;
 }
 
@@ -392,19 +392,19 @@ div.dailyEntry
 div.dailyEntry  a { text-decoration:none; color: #890500; }
 div.dailyEntryTags { font-size:7.75pt;  }
 div.dailyEntryTitle { font-size:18pt; font-weight:bold;}
-div.dailyEntryThumbnail 
+div.dailyEntryThumbnail
-{ 
+{
-    width:100%; 
+    width:100%;
-    text-align:center; 
+    text-align:center;
-    background-color:rgb(128,128,128); 
+    background-color:rgb(128,128,128);
     background:url(../images/50pc_transparent.png);
     padding:4px 0px 2px 0px;
 }
 div.dailyEntryDescription
-{ 
+{
     margin-top: 10px;
-    margin-bottom: 30px; 
+    margin-bottom: 30px;
-    text-align:justify; 
+    text-align:justify;
     overflow:auto;
 }
 
@@ -414,7 +414,7 @@ div.dailyEntryDescription
 }
 
 /* For lazy images loading in picture wall.
-   using http://www.appelsiini.net/projects/lazyload 
+   using http://www.appelsiini.net/projects/lazyload
  */
 .lazyimage { display:none; }
 
@@ -463,7 +463,7 @@ div.dailyDate  {  font-size: 11pt;padding:0px; display:block; }
 div.dailyEntryTitle { font-size:16pt; font-weight:bold;}
 div.dailyEntryDescription { font-size:10pt; }
 
-} 
+}
 
 /* Highlight search results */
 .highlight { background-color: #FFFF33; }

index.php

@@ -22,9 +22,11 @@
 $GLOBALS['config']['OPEN_SHAARLI'] = false; // If true, anyone can add/edit/delete links without having to login
 $GLOBALS['config']['HIDE_TIMESTAMPS'] = false; // If true, the moment when links were saved are not shown to users that are not logged in.
 $GLOBALS['config']['ENABLE_THUMBNAILS'] = true; // Enable thumbnails in links.
+$GLOBALS['config']['ENABLE_FAVICON'] = true; // Enable favicon in links.
 $GLOBALS['config']['CACHEDIR'] = 'cache'; // Cache directory for thumbnails for SLOW services (like flickr)
 $GLOBALS['config']['PAGECACHE'] = 'pagecache'; // Page cache directory.
 $GLOBALS['config']['ENABLE_LOCALCACHE'] = true; // Enable Shaarli to store thumbnail in a local cache. Disable to reduce webspace usage.
+                                                // Care if favicon is active and local cache are false serve page can be long
 $GLOBALS['config']['PUBSUBHUB_URL'] = ''; // PubSubHubbub support. Put an empty string to disable, or put your hub url here to enable.
 $GLOBALS['config']['UPDATECHECK_FILENAME'] = $GLOBALS['config']['DATADIR'].'/lastupdatecheck.txt'; // For updates check of Shaarli.
 $GLOBALS['config']['UPDATECHECK_INTERVAL'] = 86400 ; // Updates check frequency for Shaarli. 86400 seconds=24 hours
@@ -344,7 +346,7 @@ function isLoggedIn()
 }
 
 // Force logout.
-function logout() { if (isset($_SESSION)) { unset($_SESSION['uid']); unset($_SESSION['ip']); unset($_SESSION['username']); unset($_SESSION['privateonly']); }  
+function logout() { if (isset($_SESSION)) { unset($_SESSION['uid']); unset($_SESSION['ip']); unset($_SESSION['username']); unset($_SESSION['privateonly']); }
 setcookie('shaarli_staySignedIn', FALSE, 0, WEB_PATH);
 }
 
@@ -913,7 +915,7 @@ function showRSS()
     else $linksToDisplay = $LINKSDB;
     $nblinksToDisplay = 50;  // Number of links to display.
     if (!empty($_GET['nb']))  // In URL, you can specificy the number of links. Example: nb=200 or nb=all for all links.
-    { 
+    {
         $nblinksToDisplay = $_GET['nb']=='all' ? count($linksToDisplay) : max($_GET['nb']+0,1) ;
     }
 
@@ -988,7 +990,7 @@ function showATOM()
     else $linksToDisplay = $LINKSDB;
     $nblinksToDisplay = 50;  // Number of links to display.
     if (!empty($_GET['nb']))  // In URL, you can specificy the number of links. Example: nb=200 or nb=all for all links.
-    { 
+    {
         $nblinksToDisplay = $_GET['nb']=='all' ? count($linksToDisplay) : max($_GET['nb']+0,1) ;
     }
 
@@ -1563,7 +1565,7 @@ function renderPage()
             $title = (empty($_GET['title']) ? '' : $_GET['title'] ); // Get title if it was provided in URL (by the bookmarklet).
             $description = (empty($_GET['description']) ? '' : $_GET['description']); // Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
             $tags = (empty($_GET['tags']) ? '' : $_GET['tags'] ); // Get tags if it was provided in URL
-            $private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0); // Get private if it was provided in URL 
+            $private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0); // Get private if it was provided in URL
             if (($url!='') && parse_url($url,PHP_URL_SCHEME)=='') $url = 'http://'.$url;
             // If this is an HTTP link, we try go get the page to extact the title (otherwise we will to straight to the edit form.)
             if (empty($title) && parse_url($url,PHP_URL_SCHEME)=='http')
@@ -1574,7 +1576,7 @@ function renderPage()
  					 {
                         // Look for charset in html header.
  						preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
- 
+
  						// If found, extract encoding.
  						if (!empty($meta[0]))
  						{
@@ -1584,7 +1586,7 @@ function renderPage()
 							$html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
  						}
  						else { $html_charset = 'utf-8'; }
- 
+
  						// Extract title
  						$title = html_extract_title($data);
  						if (!empty($title))
@@ -2052,6 +2054,42 @@ function lazyThumbnail($url,$href=false)
     return $html;
 }
 
+function returnFavicon($url){
+  if(!$GLOBALS['config']['ENABLE_FAVICON']){
+    return;
+  }
+  $faviconHash = md5($url);
+  $path = substr($faviconHash, 0,2).'/'.substr($faviconHash, 2,2);
+  $faviconPath = $GLOBALS['config']['CACHEDIR'].'/'.$path.'/'.$faviconHash.'.ico';
+  if($GLOBALS['config']['ENABLE_LOCALCACHE'] === true && file_exists($faviconPath)){
+    $content = file_get_contents($faviconPath);
+    return '<img class="favicon" alt="favicon" src="data:image/ico;base64,'.base64_encode($content).'"/>';
+  }
+  if(file_exists($GLOBALS['config']['CACHEDIR'].'/'.$path.'/'.$faviconHash)){
+    return;
+  }
+  require_once 'inc/DataAccess.php';
+  require_once 'inc/Favicon.php';
+  $favicon = new \Favicon\Favicon();
+  $urlOfFavicon = $favicon->get($url);
+  if(!$urlOfFavicon){
+    if($GLOBALS['config']['ENABLE_LOCALCACHE'] === true){
+      mkdir($GLOBALS['config']['CACHEDIR'].'/'.$path,0777,true);
+      touch($GLOBALS['config']['CACHEDIR'].'/'.$path.'/'.$faviconHash);
+      return;
+    } else {
+      return;
+    }
+  }
+  if($GLOBALS['config']['ENABLE_LOCALCACHE'] === true && !is_dir($GLOBALS['config']['CACHEDIR'].'/'.$path.'/')){
+    mkdir($GLOBALS['config']['CACHEDIR'].'/'.$path,0777,true);
+  }
+  $content = file_get_contents($urlOfFavicon);
+  if($GLOBALS['config']['ENABLE_LOCALCACHE'] === true){
+    file_put_contents($faviconPath, $content);
+  }
+  return '<img  class="favicon" alt="favicon" src="data:image/ico;base64,'.base64_encode($content).'"/>';
+}
 
 // -----------------------------------------------------------------------------------------------
 // Installation

tpl/linklist.html

@@ -40,7 +40,7 @@
                         <input type="hidden" name="token" value="{$token}"><input type="hidden" name="delete_link"><input type="image" alt="Delete" src="images/delete_icon.png#" title="Delete" class="button_delete" onClick="return confirmDeleteLink();"></form>
                     </div>
                 {/if}
-                <span class="linktitle"><a href="{$redirector}{$value.url|htmlspecialchars}">{$value.title|htmlspecialchars}</a></span>
+                <span class="linktitle">{function="returnFavicon($value.url)"}<a href="{$redirector}{$value.url|htmlspecialchars}">{$value.title|htmlspecialchars}</a></span>
                 <br>
                 {if="$value.description"}<div class="linkdescription"{if condition="$search_type=='permalink'"} style="max-height:none !important;"{/if}>{$value.description}</div>{/if}
                 {if="!$GLOBALS['config']['HIDE_TIMESTAMPS'] || isLoggedIn()"}
@@ -48,8 +48,8 @@
                 {else}
                     <span class="linkdate" title="Short link here"><a href="?{$value.linkdate|smallHash}">permalink</a> - </span>
                 {/if}
-                <div style="position:relative;display:inline;"><a href="http://qrfree.kaywa.com/?l=1&s=8&d={$scripturl|urlencode}%3F{$value.linkdate|smallHash}" 
+                <div style="position:relative;display:inline;"><a href="http://qrfree.kaywa.com/?l=1&s=8&d={$scripturl|urlencode}%3F{$value.linkdate|smallHash}"
-                    onclick="showQrCode(this); return false;" class="qrcode" data-permalink="{$scripturl}?{$value.linkdate|smallHash}"><img src="images/qrcode.png#" width="13" height="13" title="QR-Code"></a></div> - 
+                    onclick="showQrCode(this); return false;" class="qrcode" data-permalink="{$scripturl}?{$value.linkdate|smallHash}"><img src="images/qrcode.png#" width="13" height="13" title="QR-Code"></a></div> -
                 <span class="linkurl" title="Short link">{$value.url|htmlspecialchars}</span><br>
                 {if="$value.tags"}
                     <div class="linktaglist">
@@ -65,12 +65,12 @@
 
 </div>
 
-    {include="page.footer"} 
+    {include="page.footer"}
 
 <script language="JavaScript">
 // Remove any displayed QR-Code
 function remove_qrcode()
-{ 
+{
     var elem = document.getElementById("permalinkQrcode");
     if (elem) elem.parentNode.removeChild(elem);
     return false;
@@ -78,7 +78,7 @@
 
 // Show the QR-Code of a permalink (when the QR-Code icon is clicked).
 function showQrCode(caller,loading=false)
-{ 
+{
     // Dynamic javascript lib loading: We only load qr.js if the QR code icon is clicked:
     if (typeof(qr)=='undefined') // Load qr.js only if not present.
     {
@@ -94,18 +94,18 @@
 
     // Remove previous qrcode if present.
     remove_qrcode();
-    
+
     // Build the div which contains the QR-Code:
     var element = document.createElement('div');
     element.id="permalinkQrcode";
 	// Make QR-Code div commit sepuku when clicked:
     if ( element.attachEvent ){ element.attachEvent('onclick', 'this.parentNode.removeChild(this);' ); } // Damn IE
     else { element.setAttribute('onclick', 'this.parentNode.removeChild(this);' ); }
-    
+
     // Build the QR-Code:
     var image = qr.image({size: 8,value: caller.dataset.permalink});
     if (image)
-    { 
+    {
         element.appendChild(image);
         element.innerHTML+= "<br>Click to close";
         caller.parentNode.appendChild(element);