Add Html Content Validation and actuality Checks

src/ConfluencePageContentDownloader.php

@@ -8,6 +8,7 @@
 use Artemeon\Confluence\Endpoint\Download;
 use Artemeon\Confluence\Endpoint\Dto\ConfluencePage;
 use Artemeon\Confluence\MacroReplacer\MacroReplacerInterface;
+use DOMDocument;
 use Exception;
 
 class ConfluencePageContentDownloader
@@ -25,6 +26,8 @@ public function __construct(Content $contentEndpoint, Download $downloadEndpoint
 
     public function downloadPageContent(ConfluencePage $page, bool $withAttachments = true): void
     {
+        $page = $this->repairPageContent($page);
+
         try {
             foreach ($this->macroReplacers as $macroReplacer) {
                 if ($macroReplacer instanceof MacroReplacerInterface) {
@@ -42,10 +45,29 @@ public function downloadPageContent(ConfluencePage $page, bool $withAttachments
             foreach ($attachments as $attachment) {
                 $this->downloadEndpoint->downloadAttachment($attachment);
             }
-
         } catch (Exception $e) {
             echo 'An error has occurred: ' . $e->getMessage();
         }
     }
 
+    private function repairPageContent(ConfluencePage $page): ConfluencePage
+    {
+        $previousLibxmlState = libxml_use_internal_errors(true);
+
+        $domDocument = new DOMDocument();
+        $domDocument->loadHTML($page->getContent());
+        if (!$domDocument->validate()) {
+            $pageContent = '';
+            foreach ($domDocument->getElementsByTagName('body')->item(0)->childNodes as $child) {
+                $pageContent .= $domDocument->saveHTML($child);
+            }
+
+            $page->setContent($pageContent);
+        }
+
+        libxml_clear_errors();
+        libxml_use_internal_errors($previousLibxmlState);
+
+        return $page;
+    }
 }

src/Endpoint/Content.php

@@ -94,7 +94,11 @@ public function findChildAttachments(string $pageId): array
     {
         $response = $this->client->get(
             'wiki/rest/api/content/' . $pageId . '/child/attachment',
-            array_merge([], $this->auth->getAuthenticationArray())
+            array_merge([
+                'query' => [
+                    'expand' => 'history,history.lastUpdated'
+                ]
+            ], $this->auth->getAuthenticationArray())
         );
 
         if ($response->getStatusCode() === 200) {

src/Endpoint/Download.php

@@ -50,12 +50,33 @@ public function downloadAttachment(ConfluenceAttachment $attachment): void
             return;
         }
 
-        // Verwende den relativen Pfad aus der API, um das Attachment herunterzuladen
-        $attachmentContent = $this->client->get(
-            '/wiki/' . $attachment->findDownloadPath(),
-            array_merge([], $this->auth->getAuthenticationArray())
-        )->getBody()->getContents();
+        if ($this->shouldAttachmentBeUpdated($attachment)) {
+            // Verwende den relativen Pfad aus der API, um das Attachment herunterzuladen
+            $attachmentContent = $this->client->get(
+                '/wiki/' . $attachment->findDownloadPath(),
+                array_merge([], $this->auth->getAuthenticationArray())
+            )->getBody()->getContents();
 
-        file_put_contents($this->downloadFolder . '/' . $attachment->getTitle(), $attachmentContent);
+            file_put_contents($this->getAttachmentFilePath($attachment), $attachmentContent);
+        }
+    }
+
+    private function getAttachmentFilePath(ConfluenceAttachment $attachment): string
+    {
+        return $this->downloadFolder . '/' . $attachment->getTitle();
+    }
+
+    private function shouldAttachmentBeUpdated(ConfluenceAttachment $attachment): bool
+    {
+        $filepath = $this->getAttachmentFilePath($attachment);
+
+        if (file_exists($filepath)) {
+            $filemtime = filemtime($filepath);
+            if (is_int($filemtime)) {
+                return $filemtime < $attachment->getLastUpdated()->getTimestamp();
+            }
+        }
+
+        return true;
     }
 }

src/Endpoint/Dto/ConfluenceAttachment.php

@@ -4,17 +4,22 @@
 
 namespace Artemeon\Confluence\Endpoint\Dto;
 
+use DateTime;
+
 class ConfluenceAttachment
 {
     private array $rawData;
 
     private string $title;
 
+    private ?DateTime $lastUpdated;
+
     public function __construct(array $rawData)
     {
         $this->rawData = $rawData;
 
         $this->title = $rawData['title'];
+        $this->lastUpdated = isset($rawData['history']['lastUpdated']['when']) ? new DateTime($rawData['history']['lastUpdated']['when']) : null;
     }
 
     public function findDownloadPath(): ?string
@@ -26,4 +31,9 @@ public function getTitle(): string
     {
         return $this->title;
     }
+
+    public function getLastUpdated(): ?DateTime
+    {
+        return $this->lastUpdated;
+    }
 }