Remove extra body tags from processed contents

StenopePHP · Jan 31, 2023 · e641ad3 · e641ad3
1 parent f936b12
commit e641ad3
Show file tree

Hide file tree

Showing 5 changed files with 15 additions and 9 deletions.
diff --git a/src/Processor/HtmlIdProcessor.php b/src/Processor/HtmlIdProcessor.php
@@ -62,21 +62,21 @@ public function __invoke(array &$data, Content $content): void
 
     private function setIdFromContent(\DOMElement $element): void
     {
-        if (!$id = $element->getAttribute('id')) {
+        if (!$element->getAttribute('id')) {
             $element->setAttribute('id', $this->slugify($element->textContent));
         }
     }
 
     private function setIdFromHashedContent(\DOMElement $element): void
     {
-        if (!$id = $element->getAttribute('id')) {
+        if (!$element->getAttribute('id')) {
             $element->setAttribute('id', $this->hash($element->textContent));
         }
     }
 
     private function setIdForImage(\DOMElement $element): void
     {
-        if (!$id = $element->getAttribute('id')) {
+        if (!$element->getAttribute('id')) {
             $name = $element->getAttribute('alt') ?: basename($element->getAttribute('src'));
             $element->setAttribute('id', $this->slugify($name));
         }

diff --git a/src/Service/NaiveHtmlCrawlerManager.php b/src/Service/NaiveHtmlCrawlerManager.php
@@ -40,16 +40,16 @@ public function save(Content $content, array &$data, string $property): void
         $key = "{$content->getType()}:{$content->getSlug()}";
 
         if (isset($this->crawlers[$key][$property])) {
-            $data[$property] = $this->crawlers[$key][$property]->html();
+            $data[$property] = $this->crawlers[$key][$property]->filterXPath('//body')->first()->html();
             unset($this->crawlers[$key][$property]);
         }
     }
 
     public function saveAll(Content $content, array &$data): void
     {
-        foreach ($this->crawlers as $key => $crawlers) {
+        foreach ($this->crawlers as $crawlers) {
             foreach ($crawlers as $property => $crawler) {
-                $data[$property] = $crawler->html();
+                $data[$property] = $crawler->filterXPath('//body')->first()->html();
             }
         }
 

diff --git a/src/Service/Parsedown.php b/src/Service/Parsedown.php
@@ -63,13 +63,15 @@ protected function blockAdmonition($line, $block = null)
 
             return $block;
         }
+
+        return null;
     }
 
     protected function blockAdmonitionContinue($line, $block = null)
     {
         // A blank newline has occurred, or text without indent:
         if (isset($block['interrupted']) || $line['indent'] < 4) {
-            return;
+            return null;
         }
 
         $previous = $block['$admonitionContentRef'] ?? "\n";

diff --git a/src/Service/SharedHtmlCrawlerManager.php b/src/Service/SharedHtmlCrawlerManager.php
@@ -54,7 +54,7 @@ public function saveAll(Content $content, array &$data): void
         }
 
         foreach ($this->crawlers[$key] as $property => $crawler) {
-            $data[$property] = $crawler->html();
+            $data[$property] = $crawler->filterXPath('//body')->first()->html();
         }
 
         unset($this->crawlers[$key]);

diff --git a/tests/Unit/Processor/ResolveContentLinksProcessorTest.php b/tests/Unit/Processor/ResolveContentLinksProcessorTest.php
@@ -70,7 +70,11 @@ public function testResolveLinks(): void
                 <a href="/other-contents-route-path/another-contents#some-anchor">Another content with anchor</a>
             </body>
             HTML,
-            $data['content']
+            <<<HTML
+            <body>
+            {$data['content']}
+            </body>
+            HTML,
         );
     }
 }