Merge pull request #150 from GSA/feature/DIGITAL-89-linkit

DIGITAL-89 Add linkit module

Author: mattsqd

composer.json

@@ -40,7 +40,7 @@
         "drupal/google_tag": "^2.0",
         "drupal/inline_entity_form": "^3.0@RC",
         "drupal/link_class": "^2.1",
-        "drupal/linkit": "^6.1",
+        "drupal/linkit": "^7.0",
         "drupal/log_stdout": "^1.5",
         "drupal/maillog": "dev-1.x",
         "drupal/mailsystem": "^4.4",

composer.lock

@@ -70,3 +70,5 @@ b2b17cbb32ac57e73928af9f52acb537|Oscar Merida|feature/DIGITAL-96-ckeditor-emoji|
 c73ec25d5b68a3e0e1bdfc83c24312ba|Matt Poole|feature/DIGITAL-396-update-deps-8-0|Wed Feb  5 15:32:16 EST 2025|./composer.sh update
 d0ca998c567e8cc218ee57bb5c9b6679|cathybaptista|feature/DIGITAL-119-configure-images|Thu Feb  6 11:06:05 EST 2025|./composer.sh require drupal/media_library_edit
 6b5963bf3c6740285c129be6d6273aed|cathybaptista|feature/DIGITAL-119-configure-images|Thu Feb  6 11:06:14 EST 2025|./composer.sh require drupal/edit_media_modal
+caeb4cc7f30e1f80c7425f9c3fcf8a98|Oscar Merida|feature/DIGITAL-89-linkit|Wed Feb  5 16:40:40 EST 2025|./composer.sh require drupal/linkit:^7.0
+c063010712247ba9dfd709c97fccb648|Oscar Merida|feature/DIGITAL-89-linkit|Thu Feb  6 16:27:25 EST 2025|./composer.sh require drupal/linkit:^7.0

composer.log

@@ -15,7 +15,19 @@ matchers:
     uuid: 556010a3-e317-48b3-b4ed-854c10f4b950
     settings:
       metadata: '[node:content-type:name] #[node:nid] | [node:created:medium] by [node:author]'
-      bundles: {  }
+      bundles:
+        authors: authors
+        basic_page: basic_page
+        community: community
+        event: event
+        guides: guides
+        guide_landing: guide_landing
+        landing_page: landing_page
+        news: news
+        resources: resources
+        short_post: short_post
+        source: source
+        topics: topics
       group_by_bundle: false
       substitution_type: canonical
       limit: 100

config/sync/linkit.linkit_profile.default.yml

@@ -4,6 +4,9 @@
 
 namespace Drupal\convert_text;
 
+use Drupal\Core\Url;
+use Drupal\node\Entity\Node;
+use Drupal\taxonomy\Entity\Term;
 use League\CommonMark\CommonMarkConverter;
 use LitEmoji\LitEmoji;
 
@@ -15,9 +18,9 @@ class ConvertText {
   /**
    * Converts text for the given $field_type.
    *
-   * @var string $source_text
+   * @param string $source_text
    *   The original source value.
-   * @var string $field_type
+   * @param string $field_type
    *   Either plain or html.
    *
    * @return string
@@ -37,18 +40,158 @@ protected static function convert(string $source_text, string $field_type): stri
 
         $converter = new CommonMarkConverter();
         $html = $converter->convert($source_text)->getContent();
-        return LitEmoji::encodeUnicode($html);
+        $html = LitEmoji::encodeUnicode($html);
+        return self::addLinkItMarkup($html);
 
       default:
         throw new \Exception("Invalid \$field_type of $field_type given");
+    }
+  }
+
+  /**
+   * Update local link tags with linkit data attributes.
+   */
+  protected static function addLinkItMarkup(string $source_text): string {
+
+    // Consider these domains local.
+    $base_domains = [\Drupal::request()->getHost(), 'digital.gov', 'www.digital.gov'];
+
+    $dom = new \DOMDocument();
+    $dom->loadHTML($source_text);
+
+    foreach ($dom->getElementsByTagName('a') as $link) {
+      $href = $link->getAttribute('href');
+      if (!$href || str_starts_with($href, 'mailto:') || str_starts_with($href, '#')) {
+        continue;
+      }
+
+      $anchor = '';
+      if (preg_match('/\#(.*)$/', $href, $matches)) {
+        $anchor = $matches[0];
+      }
+
+      // Add a trailing slash for links with just the domain w/o trailing slash.
+      if (preg_match('/^https?:\/\/(' . implode('|', $base_domains) . ')$/', $href, $matches)) {
+        $href = $matches[0] . '/';
+      }
+      // Now, strip the local domains from links that include them.
+      $href = preg_replace(
+        '/^https?:\/\/(' . implode('|', $base_domains) . ')/',
+        '',
+        $href, count: $replaced);
+      if ($replaced > 0) {
+        // HREF here includes any anchor.
+        $link->setAttribute('href', $href);
+      }
+
+      $host = parse_url($href, PHP_URL_HOST) ?? '';
+      if ($host === '') {
+        $alias = parse_url($href, PHP_URL_PATH);
 
+        $sysPath = \Drupal::service('path_alias.manager')->getPathByAlias($alias);
+
+        // If a link already has all the linkit attributes, leave it be.
+        if (
+          $link->hasAttribute('data-entity-type')
+          && $link->hasAttribute('data-entity-uuid')
+          && $link->hasAttribute('data-entity-substitution')
+        ) {
+          continue;
+        }
+
+        $url = Url::fromUserInput($sysPath);
+
+        if (!$url->isRouted()) {
+          // What do we have here?
+          $client = \Drupal::httpClient();
+
+          $host = \Drupal::request()->getSchemeAndHttpHost();
+          $response = $client->get($host . $sysPath, [
+            // Don't throw exceptions on error codes.
+            'http_errors' => FALSE,
+            'allow_redirects' => [
+              'max' => 10,
+              'track_redirects' => TRUE,
+            ],
+          ]);
+
+          if ($response->getStatusCode() > 400) {
+            continue;
+          }
+          if ($response->getStatusCode() === 200) {
+            if ($history = $response->getHeader('X-Guzzle-Redirect-History')) {
+              $finalURI = $history[array_key_last($history)];
+              $redirHost = parse_url($finalURI, PHP_URL_HOST);
+              if (!str_ends_with($host, $redirHost)) {
+                // We were redirected off site. Let's fix the link and move on.
+                $link->setAttribute('href', $finalURI . $anchor);
+                continue;
+              }
+              $redirPath = parse_url($finalURI, PHP_URL_PATH) ?? '/';
+              $url = Url::fromUserInput($redirPath);
+              if (!$url->isRouted()) {
+                // Redirected to an unrouted path, not much we can do here.
+                continue;
+              }
+              // Replace the alias in the href with the internal path.
+              $sysPath = \Drupal::request()->getBasePath() . '/' . $url->getInternalPath();
+            }
+          }
+          else {
+            // Leave 40x links alone.
+            continue;
+          }
+        }
+
+        switch ($url->getRouteName()) {
+          case 'entity.node.canonical':
+            $params = $url->getRouteParameters();
+            $node = Node::load($params['node']);
+            if (!$node) {
+              continue 2;
+            }
+            $entityType = 'node';
+            $uuid = $node->uuid();
+            break;
+
+          case 'entity.taxonomy_term.canonical':
+            $entityType = 'term';
+            $params = $url->getRouteParameters();
+            $term = Term::load($params['taxonomy_term']);
+            if (!$term) {
+              continue 2;
+            }
+            $uuid = $term->uuid();
+            break;
+
+          case '<front>':
+            // Ensure the link to the homepage doesn't have a domain name.
+            // Don't need to add any other attributes or anchor.
+            $link->setAttribute('href', $href);
+            continue 2;
+
+          default:
+            continue 2;
+        }
+
+        // Linkit saves the internal path and renders the alias.
+        $link->setAttribute('href', $sysPath . $anchor);
+        $link->setAttribute('data-entity-type', $entityType);
+        $link->setAttribute('data-entity-uuid', $uuid);
+        $link->setAttribute('data-entity-substitution', 'canonical');
+      }
     }
+
+    $body = $dom->getElementsByTagName('body')->item(0);
+    $html = $dom->saveHTML($body);
+    // No good way to keep white space AND omit the body tag automatically.
+    return preg_replace(['/^<body>/', '/<\/body>$/'], '', $html);
   }
 
   /**
    * Gets text ready to be stored in plain text fields.
    *
-   * @var string $source_text
+   * @param string $source_text
    *   The original source value.
    *
    * @return string
@@ -61,7 +204,7 @@ public static function plainText(string $source_text): string {
   /**
    * Gets text ready to be stored in html text fields.
    *
-   * @var string $source_text
+   * @param string $source_text
    *   The original source value.
    *
    * @return string

web/modules/custom/convert_text/src/ConvertText.php

@@ -32,6 +32,26 @@ public function buildForm(array $form, FormStateInterface $form_state): array {
         '#default_value' => $converted_text,
         '#disabled' => TRUE,
       ];
+
+      switch ($form_state->getValue('dest')) {
+        case 'html':
+          $form['html_filter'] = [
+            '#markup' => '<h5>HTML Filter</h5>' . check_markup($converted_text, 'html'),
+          ];
+          $form['multiline_filter'] = [
+            '#markup' => '<h5>Multiline HTML Filter</h5>' . check_markup($converted_text, 'multiline_inline_html'),
+          ];
+          $form['single_inline_filter'] = [
+            '#markup' => '<h5>Single Inline Filter</h5>' . check_markup($converted_text, 'single_inline_html'),
+          ];
+          break;
+
+        case 'plain_text':
+          $form['Plain text'] = [
+            '#markup' => '<h5>Plain text</h5>' . check_markup($converted_text, 'plain_text'),
+          ];
+          break;
+      }
     }
 
     $form['source_text'] = [

web/modules/custom/convert_text/src/Form/ConvertTextForm.php

@@ -39,7 +39,7 @@ default:
       pathauto: 0
   body:
     -
-      value: '<p><a href="/node/19" data-entity-type="node" data-entity-uuid="7086f5d5-7c6f-41ef-8a60-70cd267a9860" data-entity-substitution="canonical">RESOURCES</a>&nbsp;</p>'
+      value: '<p><a href="/node/14" data-entity-type="node" data-entity-uuid="fefddcf3-e162-4622-9176-a3f4c2dadd8e" data-entity-substitution="canonical">RESOURCES</a>&nbsp;</p>'
       format: html
       summary: ''
   field_deck:

web/modules/custom/default_content_config/content/node/ad7aa7a5-33cf-400e-93cf-11820e467111.yml

@@ -36,7 +36,7 @@ default:
       pathauto: 0
   body:
     -
-      value: '<p><strong>The page you''re looking for cannot be found.</strong> Here are a few things that you can try:</p><ul><li>Go back to the <a href="/">homepage</a>.</li><li><a href="http://find.digitalgov.gov/search?affiliate=digitalgov">Search</a> for the page.</li><li>View our <a href="/news">latest news</a>.</li><li>Explore topics in our <a href="/resources">guides and resources</a>.</li><li>Send us an email and <a href="mailto:[email protected]">let us know this page is missing</a>.</li></ul>'
+      value: '<p><strong>The page you''re looking for cannot be found.</strong> Here are a few things that you can try:</p><ul><li>Go back to the <a href="/">homepage</a>.</li><li><a href="http://find.digitalgov.gov/search?affiliate=digitalgov">Search</a> for the page.</li><li>View our <a href="/node/21" data-entity-type="node" data-entity-uuid="986de625-5015-47c0-9ccc-273f985b69b8" data-entity-substitution="canonical">latest news</a>.</li><li>Explore topics in our <a href="/node/14" data-entity-type="node" data-entity-uuid="fefddcf3-e162-4622-9176-a3f4c2dadd8e" data-entity-substitution="canonical">guides and resources</a>.</li><li>Send us an email and <a href="mailto:[email protected]">let us know this page is missing</a>.</li></ul>'
       format: html
       summary: ''
   field_deck: