Added HTML processing to the Commonmark dispatcher.

Author: Martin Brecht-Precht

example/commonmark-01.php

@@ -9,5 +9,5 @@
 
 $handler = new CommonmarkHandler();
 $dispatcher = new JsonDispatcher($handler);
-$dispatcher->processFile(__DIR__ . '/example-data.json');
+$dispatcher->processFile(__DIR__ . '/example-data-simple.json');
 fwrite(STDOUT, $handler->getResult());

example/commonmark-02.php

@@ -9,5 +9,5 @@
 
 $handler = new CommonmarkHandler();
 $dispatcher = new CommonmarkDispatcher($handler);
-$dispatcher->processFile(__DIR__ . '/result.md');
+$dispatcher->processFile(__DIR__ . '/example-data.md');
 fwrite(STDOUT, $handler->getResult());

example/example-data-simple.json

@@ -15,39 +15,31 @@
             ]
         },
         {
-            "type": "Paragraph",
+            "type": "Heading",
+            "level": 3,
             "contents": [
                 {
                     "type": "Text",
-                    "text": "this "
-                },
-                {
-                    "type": "Emphasis",
-                    "level": 1,
-                    "contents": [
-                        {
-                            "type": "Text",
-                            "text": "is"
-                        }
-                    ]
-                },
+                    "text": "HTML Block"
+                }
+            ]
+        },
+        {
+            "type": "Paragraph",
+            "contents": [
                 {
                     "type": "Text",
-                    "text": " a not so simple "
-                },
-                {
-                    "type": "Emphasis",
-                    "level": 2,
-                    "contents": [
-                        {
-                            "type": "Text",
-                            "text": "paragraph"
-                        }
-                    ]
-                },
+                    "text": "kjsdfkjhasd <p>Test<\/p> ksajdhfgkasjhfgd"
+                }
+            ]
+        },
+        {
+            "type": "Heading",
+            "level": 3,
+            "contents": [
                 {
                     "type": "Text",
-                    "text": " with different kinds of "
+                    "text": "blocks in a quote block"
                 }
             ]
         }

example/example-data.md

@@ -0,0 +1,13 @@
+# some markdown
+
+### HTML Block
+
+kjsdfkjhasd <span style="color:#ff0000;">Test</span> ksajdhfgkasjhfgd
+
+<p style="color:#ff00ff;">Block of HTML</p>
+
+sdklfjhalksdhflakjshdf
+
+klasjhdflkasghdlkfhjg
+
+### blocks in a quote block

src/Dispatcher/CommonmarkDispatcher.php

@@ -6,6 +6,7 @@
 use League\CommonMark\Environment;
 use Markdom\Dispatcher\CommonmarkUtil\DocumentProcessor;
 use Markdom\Dispatcher\Exception\DispatcherException;
+use Markdom\Dispatcher\HtmlProcessor\HtmlProcessorInterface;
 use Markdom\DispatcherInterface\DispatcherInterface;
 use Markdom\HandlerInterface\HandlerInterface;
 
@@ -22,6 +23,11 @@ class CommonmarkDispatcher implements DispatcherInterface
 	 */
 	private $markdomHandler;
 
+	/**
+	 * @var HtmlProcessorInterface
+	 */
+	private $htmlProcessor = null;
+
 	/**
 	 * Parser constructor.
 	 *
@@ -32,6 +38,24 @@ public function __construct(HandlerInterface $commonmarkHandler)
 		$this->markdomHandler = $commonmarkHandler;
 	}
 
+	/**
+	 * @return HtmlProcessorInterface
+	 */
+	public function getHtmlProcessor()
+	{
+		return $this->htmlProcessor;
+	}
+
+	/**
+	 * @param HtmlProcessorInterface $htmlProcessor
+	 * @return $this
+	 */
+	public function setHtmlProcessor($htmlProcessor)
+	{
+		$this->htmlProcessor = $htmlProcessor;
+		return $this;
+	}
+
 	/**
 	 * @param string $sourceFile
 	 * @return $this
@@ -55,7 +79,7 @@ public function processFile($sourceFile)
 	public function process($source)
 	{
 		$commonMarkEnvironment = Environment::createCommonMarkEnvironment();
-		$commonMarkEnvironment->addDocumentProcessor(new DocumentProcessor($this->markdomHandler));
+		$commonMarkEnvironment->addDocumentProcessor(new DocumentProcessor($this->markdomHandler, $this->htmlProcessor));
 		$docParser = new DocParser($commonMarkEnvironment);
 		$docParser->parse($source);
 		return $this;

src/Dispatcher/CommonmarkUtil/DocumentProcessor.php

@@ -4,6 +4,8 @@
 
 use League\CommonMark\Block\Element\Document;
 use League\CommonMark\DocumentProcessorInterface;
+use Markdom\Dispatcher\HtmlProcessor\HtmlProcessorInterface;
+use Markdom\Dispatcher\HtmlProcessor\HtmlTextProcessor;
 use Markdom\HandlerInterface\HandlerInterface;
 use Markenwerk\StackUtil\Stack;
 
@@ -42,6 +44,11 @@ final class DocumentProcessor implements DocumentProcessorInterface
 	 */
 	private $markdomHandler;
 
+	/**
+	 * @var HtmlProcessorInterface
+	 */
+	private $htmlProcessor;
+
 	/**
 	 * @var Stack
 	 */
@@ -51,11 +58,16 @@ final class DocumentProcessor implements DocumentProcessorInterface
 	 * DocumentProcessor constructor.
 	 *
 	 * @param HandlerInterface $markdomHandler
+	 * @param HtmlProcessorInterface $htmlProcessor
 	 */
-	public function __construct(HandlerInterface $markdomHandler)
+	public function __construct(HandlerInterface $markdomHandler, HtmlProcessorInterface $htmlProcessor = null)
 	{
 		$this->markdomHandler = $markdomHandler;
 		$this->imageStack = new Stack();
+		if (is_null($htmlProcessor)) {
+			$htmlProcessor = new HtmlTextProcessor();
+		}
+		$this->htmlProcessor = $htmlProcessor;
 	}
 
 	/**
@@ -65,7 +77,7 @@ public function __construct(HandlerInterface $markdomHandler)
 	 */
 	public function processDocument(Document $document)
 	{
-		$markdomHandlerEventDispatcher = new MarkdomEventBridge($this->markdomHandler);
+		$markdomHandlerEventDispatcher = new MarkdomEventBridge($this->markdomHandler, $this->htmlProcessor);
 		$walker = $document->walker();
 		while ($event = $walker->next()) {
 			$node = $event->getNode();

src/Dispatcher/CommonmarkUtil/MarkdomEventBridge.php

@@ -10,6 +10,7 @@
 use League\CommonMark\Block\Element\ListItem;
 use League\CommonMark\Inline\Element\Code;
 use League\CommonMark\Inline\Element\Emphasis;
+use League\CommonMark\Inline\Element\HtmlInline;
 use League\CommonMark\Inline\Element\Image;
 use League\CommonMark\Inline\Element\Link;
 use League\CommonMark\Inline\Element\Newline;
@@ -20,6 +21,7 @@
 use Markdom\Common\ContentType;
 use Markdom\Common\EmphasisLevel;
 use Markdom\Dispatcher\Exception\DispatcherException;
+use Markdom\Dispatcher\HtmlProcessor\HtmlProcessorInterface;
 use Markdom\HandlerInterface\HandlerInterface;
 
 /**
@@ -35,6 +37,11 @@ final class MarkdomEventBridge
 	 */
 	private $markdomHandler;
 
+	/**
+	 * @var HtmlProcessorInterface
+	 */
+	private $htmlProcessor;
+
 	/**
 	 * @var Node
 	 */
@@ -44,10 +51,12 @@ final class MarkdomEventBridge
 	 * MarkdomHandlerEventDispatcher constructor.
 	 *
 	 * @param HandlerInterface $commonmarkHandler
+	 * @param HtmlProcessorInterface $htmlProcessor
 	 */
-	public function __construct(HandlerInterface $commonmarkHandler)
+	public function __construct(HandlerInterface $commonmarkHandler, HtmlProcessorInterface $htmlProcessor)
 	{
 		$this->markdomHandler = $commonmarkHandler;
+		$this->htmlProcessor = $htmlProcessor;
 	}
 
 	/**
@@ -203,7 +212,7 @@ private function transmitContainerBeginEvent(Node $node)
 					$comment = trim($comment);
 					$this->markdomHandler->onCommentBlock($comment);
 				} else {
-					// TODO: Handle HTML block
+					$this->htmlProcessor->handleHtmlBlock($node, $this->markdomHandler);
 				}
 				break;
 			case DocumentProcessor::BLOCK_NODE_IMAGE:
@@ -303,8 +312,6 @@ private function transmitContainerEndEvent(Node $node)
 				/** @var HtmlBlock $node */
 				if ($node->getType() == $node::TYPE_2_COMMENT) {
 					$this->dispatchBlockEndEvents($node, BlockType::TYPE_COMMENT);
-				} else {
-					// TODO: Handle HTML block
 				}
 				break;
 			case DocumentProcessor::BLOCK_NODE_IMAGE:
@@ -375,7 +382,8 @@ private function transmitInlineBeginEvent(Node $node)
 				$this->markdomHandler->onCodeContent($node->getContent());
 				break;
 			case DocumentProcessor::INLINE_NODE_HTML_INLINE:
-				// TODO: Handle inline HTML
+				/** @var HtmlInline $node */
+				$this->htmlProcessor->handleInlineHtml($node, $this->markdomHandler);
 				break;
 			case DocumentProcessor::INLINE_NODE_NEWLINE:
 				$this->dispatchContentBeginEvents(ContentType::TYPE_LINE_BREAK);
@@ -408,7 +416,6 @@ private function transmitInlineEndEvent()
 				$this->dispatchContentEndEvents($this->recentInlineNode, ContentType::TYPE_CODE);
 				break;
 			case DocumentProcessor::INLINE_NODE_HTML_INLINE:
-				// TODO: Handle inline HTML
 				break;
 			case DocumentProcessor::INLINE_NODE_NEWLINE:
 				$this->dispatchContentEndEvents($this->recentInlineNode, ContentType::TYPE_LINE_BREAK);

src/Dispatcher/HtmlProcessor/HtmlProcessorInterface.php

@@ -0,0 +1,31 @@
+<?php
+
+namespace Markdom\Dispatcher\HtmlProcessor;
+
+use League\CommonMark\Block\Element\HtmlBlock;
+use League\CommonMark\Inline\Element\HtmlInline;
+use Markdom\HandlerInterface\HandlerInterface;
+
+/**
+ * Interface HtmlProcessorInterface
+ *
+ * @package Markdom\Dispatcher\HtmlProcessor
+ */
+interface HtmlProcessorInterface
+{
+
+	/**
+	 * @param HtmlBlock $htmlBlock
+	 * @param HandlerInterface $markdomHandler
+	 * @return void
+	 */
+	public function handleHtmlBlock(HtmlBlock $htmlBlock, HandlerInterface $markdomHandler);
+
+	/**
+	 * @param HtmlInline $htmlInline
+	 * @param HandlerInterface $markdomHandler
+	 * @return void
+	 */
+	public function handleInlineHtml(HtmlInline $htmlInline, HandlerInterface $markdomHandler);
+
+}

src/Dispatcher/HtmlProcessor/HtmlTextProcessor.php

@@ -0,0 +1,48 @@
+<?php
+
+namespace Markdom\Dispatcher\HtmlProcessor;
+
+use League\CommonMark\Block\Element\HtmlBlock;
+use League\CommonMark\Inline\Element\HtmlInline;
+use Markdom\Common\BlockType;
+use Markdom\Common\ContentType;
+use Markdom\HandlerInterface\HandlerInterface;
+
+/**
+ * Class HtmlTextProcessor
+ *
+ * @package Markdom\Dispatcher\HtmlProcessor
+ */
+class HtmlTextProcessor implements HtmlProcessorInterface
+{
+
+	/**
+	 * @param HtmlBlock $htmlBlock
+	 * @param HandlerInterface $markdomHandler
+	 * @return void
+	 */
+	public function handleHtmlBlock(HtmlBlock $htmlBlock, HandlerInterface $markdomHandler)
+	{
+		$plaintext = strip_tags($htmlBlock->getStringContent());
+		$markdomHandler->onBlockBegin(BlockType::TYPE_PARAGRAPH);
+		$markdomHandler->onParagraphBlockBegin();
+		$markdomHandler->onContentsBegin();
+		$markdomHandler->onContentBegin(ContentType::TYPE_TEXT);
+		$markdomHandler->onTextContent($plaintext);
+		$markdomHandler->onContentEnd(ContentType::TYPE_TEXT);
+		$markdomHandler->onContentsEnd();
+		$markdomHandler->onParagraphBlockEnd();
+		$markdomHandler->onBlockEnd(BlockType::TYPE_PARAGRAPH);
+		// FIXME: Why the hack does the CommonmarkDispatcher not add a blank line after this paragraph?!
+	}
+
+	/**
+	 * @param HtmlInline $htmlInline
+	 * @param HandlerInterface $markdomHandler
+	 * @return void
+	 */
+	public function handleInlineHtml(HtmlInline $htmlInline, HandlerInterface $markdomHandler)
+	{
+	}
+
+}

src/Handler/CommonmarkUtil/HandlerDelimiter.php

@@ -7,7 +7,7 @@
  *
  * @package Markdom\Handler\CommonmarkUtil
  */
-class HandlerDelimiter
+final class HandlerDelimiter
 {
 
 	/**