some bugfixes

Author: oscarotero

src/Adapters/Sassmeister.php

@@ -24,6 +24,8 @@ public static function check(Request $request)
      */
     public function getCode()
     {
+        $this->width = null;
+        $this->height = 480;
         $id = $this->request->getDirectoryPosition(1);
 
         return "<p class=\"sassmeister\" data-gist-id=\"{$id}\" data-height=\"480\" data-theme=\"tomorrow\">".

src/Providers/Html.php

@@ -30,7 +30,7 @@ public function run()
         self::extractFromMeta($html, $this->bag);
 
         if ($main = self::getMainElement($html)) {
-            self::extractImages($main, $this->bag, $this->request->getDomain());
+            $this->extractImages($main);
         }
 
         //Title
@@ -273,50 +273,42 @@ protected static function extractFromMeta(\DOMDocument $html, Bag $bag)
      * Extract <img> elements.
      *
      * @param \DOMElement $html
-     * @param Bag         $bag
-     * @param null|string $domain
      */
-    protected static function extractImages(\DOMElement $html, Bag $bag, $domain = null)
+    protected function extractImages(\DOMElement $html)
     {
+        $domain = $this->request->getDomain();
+
         foreach ($html->getElementsByTagName('img') as $img) {
             if ($img->hasAttribute('src')) {
-                $src = new Url($img->getAttribute('src'));
+                $src = $this->request->createUrl($img->getAttribute('src'));
 
-                //Is src relative?
-                if (!$src->getDomain()) {
-                    $bag->add('images', $src->getUrl());
+                //Avoid external images
+                if ($src->getContent() === null && $src->getDomain() !== $domain) {
                     continue;
                 }
 
-                //Avoid external images or in external links
-                if ($domain !== null) {
-                    if ($src->getDomain() !== $domain) {
-                        continue;
-                    }
-
-                    $parent = $img->parentNode;
+                $parent = $img->parentNode;
 
-                    while ($parent && isset($parent->tagName)) {
-                        if ($parent->tagName === 'a') {
-                            if ($parent->hasAttribute('href')) {
-                                $href = new Url($parent->getAttribute('href'));
+                while ($parent && isset($parent->tagName)) {
+                    if ($parent->tagName === 'a') {
+                        if ($parent->hasAttribute('href')) {
+                            $href = $this->request->createUrl($parent->getAttribute('href'));
 
-                                if ($href->getDomain() && $src->getDomain() !== $domain) {
-                                    continue 2;
-                                }
-                            }
-                            if ($parent->hasAttribute('rel') && (string) $parent->getAttribute('rel') === 'nofollow') {
+                            if ($href->getDomain() !== $domain) {
                                 continue 2;
                             }
-
-                            break;
+                        }
+                        if ($parent->hasAttribute('rel') && (string) $parent->getAttribute('rel') === 'nofollow') {
+                            continue 2;
                         }
 
-                        $parent = $parent->parentNode;
+                        break;
                     }
 
-                    $bag->add('images', $src->getUrl());
+                    $parent = $parent->parentNode;
                 }
+
+                $this->bag->add('images', $src->getUrl());
             }
         }
     }

src/Request.php

@@ -83,11 +83,17 @@ public function getUrl()
 
     /**
      * Creates and returns an Url clone.
+     * 
+     * @param string|null $url
      *
      * @return Url
      */
-    public function createUrl()
+    public function createUrl($url = null)
     {
+        if ($url !== null) {
+            return new Url($this->getAbsolute($url));
+        }
+
         return new Url($this->getUrl());
     }
 

src/Url.php

@@ -9,6 +9,8 @@ class Url
 {
     protected $info;
 
+    public static $validate = false;
+
     /**
      * Constructor. Sets the url.
      *
@@ -436,7 +438,9 @@ protected function buildUrl()
      */
     protected function parseUrl($url)
     {
-        if (!filter_var($url, FILTER_VALIDATE_URL)) {
+        // do not validate urls because some real urls fails.
+        // Example: http://jouey-.deviantart.com/art/market-153836478 fails.
+        if (self::$validate && substr($url, 0, 5) !== 'data:' && !filter_var($url, FILTER_VALIDATE_URL)) {
             throw new Exceptions\InvalidUrlException("The url '{$url}' is not valid");
         }
 
@@ -478,6 +482,8 @@ protected function parseUrl($url)
      */
     public function getAbsolute($url)
     {
+        $url = trim($url);
+
         if (empty($url)) {
             return '';
         }

tests/SassmeisterTest.php

@@ -8,8 +8,6 @@ public function testGist()
             'http://sassmeister.com/gist/1122e441f0434ec6eb2a',
             [
                 'title' => 'SassMeister | The Sass Playground!',
-                'imageWidth' => 240,
-                'imageHeight' => 240,
                 'type' => 'rich',
                 'providerName' => 'sassmeister',
             ]

tests/ScribdTest.php

@@ -8,7 +8,6 @@ public function testOne()
             'http://www.scribd.com/doc/110799637/Synthesis-of-Knowledge-Effects-of-Fire-and-Thinning-Treatments-on-Understory-Vegetation-in-Dry-U-S-Forests',
             [
                 'title' => 'Synthesis of Knowledge: Effects of Fire and Thinning Treatments on Understory Vegetation in Dry U.S. Forests',
-                'description' => "A review of current literature on studies that address effects of prescribed fire and mechanical thinning treatments on understory vegetation, and the effects of prescribed burning on rare, threatened and endangered species. The studies' outcomes are presented in two sets of tables: (1) functional group results, and (2) species-specific results.",
                 'imageWidth' => 111,
                 'imageHeight' => 142,
                 'type' => 'rich',