Markup like CommonMark

src/Components/Blocks/Comment.php

@@ -1,88 +0,0 @@
-<?php
-
-namespace Erusev\Parsedown\Components\Blocks;
-
-use Erusev\Parsedown\AST\Handler;
-use Erusev\Parsedown\AST\StateRenderable;
-use Erusev\Parsedown\Components\Block;
-use Erusev\Parsedown\Components\ContinuableBlock;
-use Erusev\Parsedown\Configurables\SafeMode;
-use Erusev\Parsedown\Html\Renderables\RawHtml;
-use Erusev\Parsedown\Html\Renderables\Text;
-use Erusev\Parsedown\Parsing\Context;
-use Erusev\Parsedown\State;
-
-final class Comment implements ContinuableBlock
-{
-    use BlockAcquisition;
-
-    /** @var string */
-    private $html;
-
-    /** @var bool */
-    private $isClosed;
-
-    /**
-     * @param string $html
-     * @param bool $isClosed
-     */
-    public function __construct($html, $isClosed)
-    {
-        $this->html = $html;
-        $this->isClosed = $isClosed;
-    }
-
-    /**
-     * @param Context $Context
-     * @param Block|null $Block
-     * @param State|null $State
-     * @return static|null
-     */
-    public static function build(
-        Context $Context,
-        Block $Block = null,
-        State $State = null
-    ) {
-        if (\strpos($Context->line()->text(), '<!--') === 0) {
-            return new self(
-                $Context->line()->rawLine(),
-                \strpos($Context->line()->text(), '-->') !== false
-            );
-        }
-
-        return null;
-    }
-
-    /**
-     * @param Context $Context
-     * @return self|null
-     */
-    public function advance(Context $Context)
-    {
-        if ($this->isClosed) {
-            return null;
-        }
-
-        return new self(
-            $this->html . "\n" . $Context->line()->rawLine(),
-            \strpos($Context->line()->text(), '-->') !== false
-        );
-    }
-
-    /**
-     * @return Handler<Text|RawHtml>
-     */
-    public function stateRenderable()
-    {
-        return new Handler(
-            /** @return Text|RawHtml */
-            function (State $State) {
-                if ($State->get(SafeMode::class)->isEnabled()) {
-                    return new Text($this->html);
-                } else {
-                    return new RawHtml($this->html);
-                }
-            }
-        );
-    }
-}

src/Components/Blocks/Markup.php

@@ -19,15 +19,33 @@ final class Markup implements ContinuableBlock
 
     const REGEX_HTML_ATTRIBUTE = '[a-zA-Z_:][\w:.-]*+(?:\s*+=\s*+(?:[^"\'=<>`\s]+|"[^"]*+"|\'[^\']*+\'))?+';
 
+    /** @var array{2: string, 3: string, 4: string, 5: string} */
+    private static $simpleContainsEndConditions = [
+        2 => '-->',
+        3 => '?>',
+        4 => '>',
+        5 => ']]>'
+    ];
+
     /** @var string */
     private $html;
 
+    /** @var 1|2|3|4|5|6|7 */
+    private $type;
+
+    /** @var bool */
+    private $closed;
+
     /**
      * @param string $html
+     * @param 1|2|3|4|5|6|7 $type
+     * @param bool $closed
      */
-    public function __construct($html)
+    private function __construct($html, $type, $closed = false)
     {
         $this->html = $html;
+        $this->type = $type;
+        $this->closed = $closed;
     }
 
     /**
@@ -41,15 +59,40 @@ final class Markup implements ContinuableBlock
         Block $Block = null,
         State $State = null
     ) {
-        if (\preg_match('/^<[\/]?+(\w*)(?:[ ]*+'.self::REGEX_HTML_ATTRIBUTE.')*+[ ]*+(\/)?>/', $Context->line()->text(), $matches)) {
+        $text = $Context->line()->text();
+        $rawLine = $Context->line()->rawLine();
+
+        if (\preg_match('/^<(?:script|pre|style)(?:\s++|>|$)/i', $text)) {
+            return new self($rawLine, 1, self::closes12345TypeMarkup(1, $text));
+        }
+
+        if (\substr($text, 0, 4) === '<!--') {
+            return new self($rawLine, 2, self::closes12345TypeMarkup(2, $text));
+        }
+
+        if (\substr($text, 0, 2) === '<?') {
+            return new self($rawLine, 3, self::closes12345TypeMarkup(3, $text));
+        }
+
+        if (\preg_match('/^<![A-Z]/', $text)) {
+            return new self($rawLine, 4, self::closes12345TypeMarkup(4, $text));
+        }
+
+        if (\substr($text, 0, 9) === '<![CDATA[') {
+            return new self($rawLine, 5, self::closes12345TypeMarkup(5, $text));
+        }
+
+        if (\preg_match('/^<[\/]?+(\w*)(?:[ ]*+'.self::REGEX_HTML_ATTRIBUTE.')*+[ ]*+(\/)?>/', $text, $matches)) {
             $element = \strtolower($matches[1]);
 
             if (\array_key_exists($element, Element::$TEXT_LEVEL_ELEMENTS)) {
                 return null;
             }
 
-            return new self($Context->line()->text());
+            return new self($rawLine, 6);
         }
+
+        return null;
     }
 
     /**
@@ -58,13 +101,43 @@ final class Markup implements ContinuableBlock
      */
     public function advance(Context $Context)
     {
-        if ($Context->previousEmptyLines() > 0) {
+        $closed = $this->closed;
+        $type = $this->type;
+
+        if ($closed) {
             return null;
         }
 
-        $html = $this->html . "\n" . $Context->line()->rawLine();
+        if (($type === 6 || $type === 7) && $Context->previousEmptyLines() > 0) {
+            return null;
+        }
 
-        return new self($html);
+        if ($type === 1 || $type === 2 || $type === 3 || $type === 4 || $type === 5) {
+            $closed = self::closes12345TypeMarkup($type, $Context->line()->text());
+        }
+
+        $html = $this->html . \str_repeat("\n", $Context->previousEmptyLines() + 1);
+        $html .= $Context->line()->rawLine();
+
+        return new self($html, $type, $closed);
+    }
+
+    /**
+     * @param 1|2|3|4|5 $type
+     * @param string $text
+     * @return bool
+     */
+    private static function closes12345TypeMarkup($type, $text)
+    {
+        if ($type === 1) {
+            if (\preg_match('/<\/(?:script|pre|style)>/i', $text)) {
+                return true;
+            }
+        } elseif (\stripos($text, self::$simpleContainsEndConditions[$type]) !== false) {
+            return true;
+        }
+
+        return false;
     }
 
     /**

src/Configurables/BlockTypes.php

@@ -4,7 +4,6 @@ namespace Erusev\Parsedown\Configurables;
 
 use Erusev\Parsedown\Components\Block;
 use Erusev\Parsedown\Components\Blocks\BlockQuote;
-use Erusev\Parsedown\Components\Blocks\Comment;
 use Erusev\Parsedown\Components\Blocks\FencedCode;
 use Erusev\Parsedown\Components\Blocks\Header;
 use Erusev\Parsedown\Components\Blocks\IndentedCode;
@@ -38,7 +37,7 @@ final class BlockTypes implements Configurable
         '8' => [TList::class],
         '9' => [TList::class],
         ':' => [Table::class],
-        '<' => [Comment::class, BlockMarkup::class],
+        '<' => [BlockMarkup::class],
         '=' => [SetextHeader::class],
         '>' => [BlockQuote::class],
         '[' => [Reference::class],

tests/ParsedownTest.php

@@ -2,7 +2,6 @@
 
 namespace Erusev\Parsedown\Tests;
 
-use Erusev\Parsedown\Components\Blocks\Comment;
 use Erusev\Parsedown\Components\Blocks\Markup as BlockMarkup;
 use Erusev\Parsedown\Components\Inlines\Markup as InlineMarkup;
 use Erusev\Parsedown\Configurables\BlockTypes;
@@ -139,7 +138,7 @@ color: red;
 EXPECTED_HTML;
 
         $parsedownWithNoMarkup = new Parsedown(new State([
-            BlockTypes::initial()->removing([BlockMarkup::class, Comment::class]),
+            BlockTypes::initial()->removing([BlockMarkup::class]),
             InlineTypes::initial()->removing([InlineMarkup::class]),
         ]));