feat: #15 confluence HTML to markdown

Author: shaunthegeek

app/Confluence.php

@@ -2,13 +2,18 @@
 
 namespace App;
 
+use League\HTMLToMarkdown\HtmlConverter;
+
 class Confluence
 {
     private \DOMDocument $document;
+    private HtmlConverter $htmlConverter;
 
-    public function __construct(\DOMDocument $document = null)
+    public function __construct(\DOMDocument $document = null, HtmlConverter $htmlConverter = null)
     {
         $this->document = $document ?? new \DOMDocument();
+        $this->htmlConverter = $htmlConverter ?? new HtmlConverter();
+        $this->htmlConverter->getConfig()->setOption('strip_tags', true);
     }
 
     public function parsePageHtml(string $filename, string $spaceName): array
@@ -24,4 +29,13 @@ public function parsePageHtml(string $filename, string $spaceName): array
             'content' => $content,
         ];
     }
+
+    public function htmlFile2Markdown(string $filename)
+    {
+        libxml_use_internal_errors(true);
+        $this->document->loadHTMLFile($filename);
+
+        $html = $this->document->saveHTML($this->document->getElementById('main-content'));
+        return $this->htmlConverter->convert($html);
+    }
 }

composer.json

@@ -21,7 +21,8 @@
         "ext-json": "*",
         "ext-libxml": "*",
         "laravel-fans/confluence": "^0.1.1",
-        "laravel-zero/framework": "^8.8"
+        "laravel-zero/framework": "^8.8",
+        "league/html-to-markdown": "^5.0"
     },
     "require-dev": {
         "fakerphp/faker": "^1.14",

composer.lock

@@ -16,4 +16,11 @@ public function testParsePageHtml()
             'content' => '你好',
         ], $result);
     }
+
+    public function testHtmlFile2Markdown()
+    {
+        $confluence = new Confluence();
+        $markdown = $confluence->htmlFile2Markdown($this->dataDir . 'confluence/space1/text-demo_65601.html');
+        $this->assertEquals("你好\n==", $markdown);
+    }
 }