Receive parsed keywords instead of fetched HTML for tag suggestions

Kovah · Nov 1, 2023 · 3dc4a3d · 3dc4a3d
1 parent 31331bc
commit 3dc4a3d
Show file tree

Hide file tree

Showing 8 changed files with 106 additions and 39 deletions.
diff --git a/app/Http/Controllers/FetchController.php b/app/Http/Controllers/FetchController.php
@@ -10,6 +10,7 @@
 use Illuminate\Http\Request;
 use Illuminate\Http\Response;
 use Illuminate\Support\Facades\Http;
+use Masterminds\HTML5;
 
 class FetchController extends Controller
 {
@@ -121,9 +122,9 @@ public static function checkForUpdates(): JsonResponse
      * implementation.
      *
      * @param Request $request
-     * @return Response
+     * @return JsonResponse
      */
-    public function htmlForUrl(Request $request): Response
+    public function htmlKeywordsFromUrl(Request $request)
     {
         $request->validate([
             'url' => ['url'],
@@ -138,9 +139,20 @@ public function htmlForUrl(Request $request): Response
         $response = $newRequest->get($url);
 
         if ($response->successful()) {
-            return response($response->body());
+            $html5 = new HTML5();
+            $dom = $html5->loadHTML($response->body());
+            $keywords = [];
+            /** @var \DOMElement $metaTag */
+            foreach ($dom->getElementsByTagName('meta') as $metaTag) {
+                if (strtolower($metaTag->getAttribute('name')) === 'keywords') {
+                    $keywords = explode(',', $metaTag->getAttributeNode('content')?->value);
+                    $keywords = array_map(fn($keyword) => trim(e($keyword)), $keywords);
+                    array_push($keywords, ...$keywords);
+                }
+            }
+            return response()->json(['keywords' => $keywords]);
         }
 
-        return response(null);
+        return response()->json(['keywords' => null]);
     }
 }
diff --git a/composer.json b/composer.json
@@ -16,6 +16,7 @@
     "league/flysystem-aws-s3-v3": "^3.0",
     "league/flysystem-ftp": "^3.0",
     "league/flysystem-sftp-v3": "^3.0",
+    "masterminds/html5": "^2.8",
     "predis/predis": "^v2.1",
     "rap2hpoutre/laravel-log-viewer": "^v2.2.0",
     "sentry/sentry-laravel": "^3.3.0",

diff --git a/composer.lock b/composer.lock
diff --git a/resources/assets/js/components/TagsSelect.js b/resources/assets/js/components/TagsSelect.js
@@ -76,9 +76,7 @@ export default class TagsSelect {
 
     this.$suggestionsContent.innerHTML = '';
 
-    tags.forEach(newTag => {
-      newTag = newTag.trim();
-
+    tags.slice(0, 20).forEach(newTag => {
       const $tag = document.createElement('span');
       $tag.classList.add('btn', 'btn-outline-secondary', 'btn-xs');
       $tag.innerText = newTag;

diff --git a/resources/assets/js/components/UrlField.js b/resources/assets/js/components/UrlField.js
@@ -76,7 +76,7 @@ export default class UrlField {
       return;
     }
 
-    fetch(window.appData.routes.fetch.htmlForUrl, {
+    fetch(window.appData.routes.fetch.keywordsForUrl, {
       method: 'POST',
       credentials: 'same-origin',
       headers: {'Content-Type': 'application/json'},
@@ -85,25 +85,11 @@ export default class UrlField {
         url: url
       })
     })
-      .then(response => response.text())
-      .then(body => {
-        if (body !== null) {
-          this.parseHtmlForKeywords(body);
+      .then(response => response.json())
+      .then(data => {
+        if (data.keywords !== null) {
+          this.tagSuggestions.displayNewSuggestions(data.keywords);
         }
       });
   }
-
-  parseHtmlForKeywords (body) {
-    const parser = new DOMParser();
-    const doc = parser.parseFromString(body, 'text/html');
-
-    if (doc.head.children.length > 0) {
-      const keywords = doc.head.children.namedItem('keywords')
-        || doc.head.children.namedItem('Keywords');
-
-      if (keywords !== null && keywords.content.length > 0) {
-        this.tagSuggestions.displayNewSuggestions(keywords.content.split(','));
-      }
-    }
-  }
 }
diff --git a/resources/views/partials/header.blade.php b/resources/views/partials/header.blade.php
@@ -22,7 +22,7 @@
             'searchLists' => route('fetch-lists'),
             'searchTags' => route('fetch-tags'),
             'existingLinks' => route('fetch-existing-links'),
-            'htmlForUrl' => route('fetch-html-for-url'),
+            'keywordsForUrl' => route('fetch-keywords-for-url'),
             'updateCheck' => route('fetch-update-check'),
             'generateApiToken' => route('generate-api-token'),
             'generateCronToken' => route('generate-cron-token'),

diff --git a/routes/web.php b/routes/web.php
@@ -129,8 +129,8 @@
         ->name('fetch-lists');
     Route::post('fetch/existing-links', [FetchController::class, 'searchExistingUrls'])
         ->name('fetch-existing-links');
-    Route::post('fetch/html-for-url', [FetchController::class, 'htmlForUrl'])
-        ->name('fetch-html-for-url');
+    Route::post('fetch/keywords-for-url', [FetchController::class, 'htmlKeywordsFromUrl'])
+        ->name('fetch-keywords-for-url');
     Route::get('fetch/update-check', [FetchController::class, 'checkForUpdates'])
         ->name('fetch-update-check');
 

diff --git a/tests/Controller/FetchControllerTest.php b/tests/Controller/FetchControllerTest.php
@@ -82,46 +82,49 @@ public function testExistingUrlSearchWithoutResult(): void
         $response->assertOk()->assertJson(['linkFound' => false]);
     }
 
-    public function testGetHtmlForUrl(): void
+    public function testGetHtmlKeywordsForUrl(): void
     {
         $testHtml = '<!DOCTYPE html><head>' .
             '<title>Example Title</title>' .
             '<meta name="description" content="This an example description">' .
+            '<meta name="keywords" content="html, css, javascript">' .
             '</head></html>';
 
         Http::fake([
             'example.com' => Http::response($testHtml, 200),
         ]);
 
-        $response = $this->post('fetch/html-for-url', [
+        $response = $this->post('fetch/keywords-for-url', [
             'url' => 'https://example.com',
         ]);
 
         $response->assertOk();
 
-        $responseHtml = $response->content();
-        $this->assertEquals($testHtml, $responseHtml);
+        $keywords = $response->json('keywords');
+        $this->assertEquals('html', $keywords[0]);
+        $this->assertEquals('css', $keywords[1]);
+        $this->assertEquals('javascript', $keywords[2]);
     }
 
-    public function testGetHtmlForInvalidUrl(): void
+    public function testGetKeywordsForInvalidUrl(): void
     {
-        $response = $this->post('fetch/html-for-url', [
+        $response = $this->post('fetch/keywords-for-url', [
             'url' => 'not a url',
         ]);
 
         $response->assertSessionHasErrors('url');
     }
 
-    public function testGetHtmlForUrlWithFailure(): void
+    public function testGetKeywordsForUrlWithFailure(): void
     {
         Http::fake([
             'example.com' => Http::response('', 500),
         ]);
 
-        $response = $this->post('fetch/html-for-url', [
+        $response = $this->post('fetch/keywords-for-url', [
             'url' => 'https://example.com',
         ]);
 
-        $response->assertOk()->assertSee('');
+        $response->assertOk()->assertJson(['keywords' => null]);
     }
 }