Remove extra body tags from processed contents

src/Behaviour/HtmlCrawlerManagerInterface.php

@@ -20,7 +20,7 @@ interface HtmlCrawlerManagerInterface
     /**
      * Get HTML Crawler for the given property (creates it if needed)
      */
-    public function get(Content $content, array &$data, string $property): ?Crawler;
+    public function get(Content $content, array $data, string $property): ?Crawler;
 
     /**
      * Dump the current state of the HTML Crawler into data for the given property.

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));
         }

src/Service/NaiveHtmlCrawlerManager.php

@@ -19,7 +19,7 @@ class NaiveHtmlCrawlerManager implements HtmlCrawlerManagerInterface
      */
     private array $crawlers = [];
 
-    public function get(Content $content, array &$data, string $property): ?Crawler
+    public function get(Content $content, array $data, string $property): ?Crawler
     {
         $key = "{$content->getType()}:{$content->getSlug()}";
         $crawler = $this->createCrawler($data[$property]);
@@ -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();
             }
         }
 

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";

src/Service/SharedHtmlCrawlerManager.php

@@ -19,7 +19,7 @@ class SharedHtmlCrawlerManager implements HtmlCrawlerManagerInterface
      */
     private array $crawlers = [];
 
-    public function get(Content $content, array &$data, string $property): ?Crawler
+    public function get(Content $content, array $data, string $property): ?Crawler
     {
         $key = "{$content->getType()}:{$content->getSlug()}";
 
@@ -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]);

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,
         );
     }
 }

tests/Unit/Service/HtmlCrawlerManagerTest.php

@@ -0,0 +1,117 @@
+<?php
+
+/*
+ * This file is part of the "StenopePHP/Stenope" bundle.
+ *
+ * @author Thomas Jarrand <[email protected]>
+ */
+
+namespace Stenope\Bundle\Tests\Unit\Service;
+
+use PHPUnit\Framework\TestCase;
+use Stenope\Bundle\Behaviour\HtmlCrawlerManagerInterface;
+use Stenope\Bundle\Content;
+use Stenope\Bundle\Service\NaiveHtmlCrawlerManager;
+use Stenope\Bundle\Service\SharedHtmlCrawlerManager;
+
+class HtmlCrawlerManagerTest extends TestCase
+{
+    /**
+     * @dataProvider provideNoExtraBodyData
+     */
+    public function testSaveNoExtraBody(
+        HtmlCrawlerManagerInterface $manager,
+        string $html,
+        string $expected
+    ): void {
+        if ($manager instanceof SharedHtmlCrawlerManager) {
+            $this->markTestSkipped('SharedHtmlCrawlerManager does nothing on save()');
+        }
+
+        $content = new Content('slug', 'type', $html, 'html');
+        $data = ['content' => $html];
+
+        $manager->get($content, $data, 'content');
+        $manager->save($content, $data, 'content');
+
+        self::assertXmlStringEqualsXmlString(<<<HTML
+            <html>$expected</html>
+            HTML,
+            <<<HTML
+            <html>{$data['content']}</html>
+            HTML,
+        );
+    }
+
+    /**
+     * @dataProvider provideNoExtraBodyData
+     */
+    public function testSaveAllNoExtraBody(
+        HtmlCrawlerManagerInterface $manager,
+        string $html,
+        string $expected
+    ): void {
+        $content = new Content('slug', 'type', $html, 'html');
+        $data = ['content' => $html];
+
+        $manager->get($content, $data, 'content');
+        $manager->saveAll($content, $data);
+
+        self::assertXmlStringEqualsXmlString(<<<HTML
+            <html>$expected</html>
+            HTML,
+            <<<HTML
+            <html>{$data['content']}</html>
+            HTML,
+        );
+    }
+
+    public function provideNoExtraBodyData(): iterable
+    {
+        $html = <<<HTML
+            <html>
+                <head>
+                    <title>My title</title>
+                </head>
+                <body>
+                    <h1>My title</h1>
+                    <p>My content</p>
+                </body>
+            </html>
+        HTML;
+
+        $expected = <<<HTML
+            <h1>My title</h1>
+            <p>My content</p>
+        HTML;
+
+        yield 'with full HTML and naive manager' => [
+            new NaiveHtmlCrawlerManager(),
+            $html,
+            $expected,
+        ];
+
+        yield 'with full HTML and shared manager' => [
+            new SharedHtmlCrawlerManager(),
+            $html,
+            $expected,
+        ];
+
+        $html = <<<HTML
+            <h1>My title</h1>
+            <p>My content</p>
+        HTML;
+
+        yield 'with partial HTML and naive manager' => [
+            new NaiveHtmlCrawlerManager(),
+            $html,
+            $expected,
+        ];
+
+        yield 'with partial HTML and shared manager' => [
+            new SharedHtmlCrawlerManager(),
+            $html,
+            $expected,
+        ];
+    }
+}