Added support for <style> tag processing and fixed index file not cor…

…rectly identified. (#11)

* Added support for <style> tag processing.

* Fixed index file not correctly identified.

* Fixed formatter link title not being using description.

* Fixed cleanup of meta charset.

* Fixed URL encoding.

minisite.theme.inc

@@ -90,6 +90,8 @@ function template_preprocess_minisite_link(array &$variables) {
   // Fix in https://www.drupal.org/node/2646744.
   $variables['#cache']['contexts'][] = 'url.site';
 
+  $link_text = $file_entity->getFilename();
+
   // Asset path is provided.
   if (isset($variables['asset_path'])) {
     // Asset path is a file.
@@ -100,6 +102,11 @@ function template_preprocess_minisite_link(array &$variables) {
     else {
       $url = Url::fromUserInput($variables['asset_path'])->toString();
     }
+
+    // Use the description as the link text if available.
+    if (!empty($variables['description'])) {
+      $link_text = $variables['description'];
+    }
   }
   // Falling back to the archive file.
   else {
@@ -122,14 +129,8 @@ function template_preprocess_minisite_link(array &$variables) {
     $variables['attributes']->addClass($classes);
   }
 
-  // Use the description as the link text if available.
-  if (empty($variables['description'])) {
-    $link_text = $file_entity->getFilename();
-  }
-  else {
-    $link_text = $variables['description'];
-    $options['attributes']['title'] = $file_entity->getFilename();
-  }
+  // Set link title.
+  $options['attributes']['title'] = $link_text;
 
   $variables['link'] = Link::fromTextAndUrl($link_text, Url::fromUserInput($url, $options));
 }

src/Asset.php

@@ -378,7 +378,14 @@ public function getLanguage() {
    * {@inheritdoc}
    */
   public function isIndex() {
-    return UrlValidator::urlIsIndex($this->urlBag->getUri(), self::INDEX_FILE);
+    if (!UrlValidator::urlIsIndex($this->urlBag->getUri(), self::INDEX_FILE)) {
+      return FALSE;
+    }
+
+    $path = $this->urlBag->getPathInArchive();
+    $in_root = strpos($path, DIRECTORY_SEPARATOR) == FALSE;
+
+    return $in_root;
   }
 
   /**

src/PageProcessor.php

@@ -66,6 +66,10 @@ public function process() {
     foreach ($this->document->getElementsByTagName('script') as $item) {
       $this->processTagScript($item);
     }
+
+    foreach ($this->document->getElementsByTagName('style') as $item) {
+      $this->processTagStyle($item);
+    }
   }
 
   /**
@@ -198,6 +202,39 @@ protected function processTagScript(\DOMNode $item) {
     $item->setAttribute('src', $url);
   }
 
+  /**
+   * Process <style> tag.
+   *
+   * @param \DOMNode $item
+   *   Document node object to process.
+   */
+  protected function processTagStyle(\DOMNode $item) {
+    $content = $item->textContent;
+
+    // Replace imported styles.
+    preg_match_all('/@import url\(([^)]+)\)/i', $content, $matches, PREG_SET_ORDER);
+
+    if (empty($matches)) {
+      return;
+    }
+
+    foreach ($matches as $match) {
+      if (count($match) != 2) {
+        continue;
+      }
+
+      $url = $match[1];
+
+      $url = self::urlExtractPath($url);
+      $url = UrlValidator::relativeToRoot($url, $this->urlBag->getAssetDir() . '/' . $this->urlBag->getRootDir());
+
+      $str = str_replace($match[1], $url, $match[0]);
+      $content = str_replace($match[0], $str, $content);
+    }
+
+    $item->textContent = $content;
+  }
+
   /**
    * Process <img> tag.
    *
@@ -231,6 +268,8 @@ protected function loadDocument($content) {
 
     $document = new \DOMDocument();
 
+    $content = $this->cleanupContent($content);
+
     $loaded = $document->loadHTML($content);
     if (!$loaded || empty($document) || empty($document->textContent)) {
       throw new PageProcessorException(sprintf('Unable to parse document: %s', libxml_get_last_error()));
@@ -239,6 +278,16 @@ protected function loadDocument($content) {
     return $document;
   }
 
+  /**
+   * Cleanup content of the page before loading it int internal document.
+   */
+  protected function cleanupContent($content) {
+    $content = preg_replace('/\<meta\s+http-equiv\s*=\s*\"content-type\"\s+content\s*=\s*\".*charset=ISO-8859-1\"\s*(\/?)\>/i', '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">', $content);
+    $content = mb_convert_encoding($content, 'HTML-ENTITIES', "UTF-8");
+
+    return $content;
+  }
+
   /**
    * Extract path from the URL.
    *

src/UrlValidator.php

@@ -92,6 +92,8 @@ public static function relativeToRoot($url, $parent) {
     $url = str_replace('../', '', $url);
     $url = rtrim($parent, '/') . '/' . ltrim($url, '/');
     $url = \Drupal::service('stream_wrapper_manager')->isValidUri($url) ? file_url_transform_relative(file_create_url($url)) : $url;
+    // Decode URL encoded in file_create_url().
+    $url = rawurldecode($url);
     $url = '/' . ltrim($url, '/');
 
     return $url;

tests/src/Functional/AssetTest.php

@@ -155,4 +155,36 @@ public function dataProviderAssetFromValues() {
     ];
   }
 
+  /**
+   * Test Asset::isIndex().
+   *
+   * @dataProvider dataProviderIsIndex
+   * @covers \Drupal\minisite\Asset::isIndex
+   */
+  public function testIsIndex($path, $is_index) {
+    $instance = new Asset(
+      'node',
+      $this->contentType,
+      1,
+      2,
+      Language::LANGCODE_DEFAULT,
+      'field_minisite_test',
+      $path
+    );
+
+    $this->assertEqual($instance->isIndex(), $is_index);
+  }
+
+  /**
+   * Data provider for testIsIndex.
+   */
+  public function dataProviderIsIndex() {
+    return [
+      ['public://minisite/static/24c22dd1-2cf1-47ae-ac8a-23a7ff8b86c5/rootpath/index.html', TRUE],
+      ['public://minisite/static/24c22dd1-2cf1-47ae-ac8a-23a7ff8b86c5/rootpath/page.html', FALSE],
+      ['public://minisite/static/24c22dd1-2cf1-47ae-ac8a-23a7ff8b86c5/rootpath/subpath/index.html', FALSE],
+      ['public://minisite/static/24c22dd1-2cf1-47ae-ac8a-23a7ff8b86c5/rootpath/subpath/page.html', FALSE],
+    ];
+  }
+
 }