Add hard and soft breaks

psalm.xml

@@ -27,6 +27,7 @@
                  <referencedMethod name="Erusev\Parsedown\Parsing\Lines::last" />
                  <referencedMethod name="Erusev\Parsedown\Configurables\StrictMode::enabled" />
                  <referencedMethod name="Erusev\Parsedown\Configurables\SafeMode::enabled" />
+                 <referencedMethod name="Erusev\Parsedown\Html\Renderables\Container::__construct" />
             </errorLevel>
         </PossiblyUnusedMethod>
     </issueHandlers>

src/Components/Inlines/HardBreak.php

@@ -0,0 +1,91 @@
+<?php
+
+namespace Erusev\Parsedown\Components\Inlines;
+
+use Erusev\Parsedown\AST\StateRenderable;
+use Erusev\Parsedown\Components\Inline;
+use Erusev\Parsedown\Html\Renderables\Element;
+use Erusev\Parsedown\Html\Renderables\Text;
+use Erusev\Parsedown\Parsedown;
+use Erusev\Parsedown\Parsing\Excerpt;
+use Erusev\Parsedown\State;
+
+final class HardBreak implements Inline
+{
+    use WidthTrait;
+
+    /** @var int */
+    private $position;
+
+    /**
+     * @param int $width
+     * @param int $position
+     */
+    public function __construct($width, $position)
+    {
+        $this->width = $width;
+        $this->position = $position;
+    }
+
+    /**
+     * @param Excerpt $Excerpt
+     * @param State $State
+     * @return static|null
+     */
+    public static function build(Excerpt $Excerpt, State $State)
+    {
+        $context = $Excerpt->context();
+        $offset = $Excerpt->offset();
+
+        if ($offset < 1) {
+            return null;
+        }
+
+        if (\substr($context, $offset -1, 1) === '\\') {
+            $trimTrailingWhitespace = \rtrim(\substr($context, 0, $offset -1));
+            $contentLen = \strlen($trimTrailingWhitespace);
+
+            return new self($offset - $contentLen, $contentLen);
+        }
+
+        if ($offset < 2) {
+            return null;
+        }
+
+        if (\substr($context, $offset -2, 2) === '  ') {
+            $trimTrailingWhitespace = \rtrim(\substr($context, 0, $offset));
+            $contentLen = \strlen($trimTrailingWhitespace);
+
+            return new self($offset - $contentLen, $contentLen);
+        }
+
+        return null;
+    }
+
+    /**
+     * Return an integer to declare that the inline should be treated as if it
+     * started from that position in the excerpt given to static::build.
+     * Return null to use the excerpt offset value.
+     * @return int|null
+     * */
+    public function modifyStartPositionTo()
+    {
+        return $this->position;
+    }
+
+    /**
+     * @return Element
+     */
+    public function stateRenderable(Parsedown $_)
+    {
+        return Element::selfClosing('br', []);
+    }
+
+    /**
+     * @return Text
+     */
+    public function bestPlaintext()
+    {
+        return new Text("\n");
+    }
+}

src/Components/Inlines/PlainText.php

@@ -2,11 +2,8 @@
 
 namespace Erusev\Parsedown\Components\Inlines;
 
-use Erusev\Parsedown\AST\Handler;
 use Erusev\Parsedown\AST\StateRenderable;
 use Erusev\Parsedown\Components\Inline;
-use Erusev\Parsedown\Html\Renderables\Container;
-use Erusev\Parsedown\Html\Renderables\Element;
 use Erusev\Parsedown\Html\Renderables\Text;
 use Erusev\Parsedown\Parsedown;
 use Erusev\Parsedown\Parsing\Excerpt;
@@ -45,34 +42,11 @@ final class PlainText implements Inline
     }
 
     /**
-     * @return Handler<Container>
+     * @return Text
      */
     public function stateRenderable(Parsedown $_)
     {
-        return new Handler(
-            /** @return Container */
-            function (State $_) {
-                $Renderables = [];
-                $text = $this->text;
-
-                $text = \preg_replace('/(?<![ \t])[ ]\n/', "$1\n", $text);
-
-                while (\preg_match('/(?:[ ]*+[\\\]|[ ]{2,}+)\n/', $text, $matches, \PREG_OFFSET_CAPTURE)) {
-                    $offset = \intval($matches[0][1]);
-                    $before = \substr($text, 0, $offset);
-                    $after = \substr($text, $offset + \strlen($matches[0][0]));
-                    $Renderables[] = new Text($before);
-                    $Renderables[] = Element::selfClosing('br', []);
-                    $Renderables[] = new Text("\n");
-
-                    $text = $after;
-                }
-
-                $Renderables[] = new Text($text);
-
-                return new Container($Renderables);
-            }
-        );
+        return new Text($this->text);
     }
 
     /**

src/Components/Inlines/SoftBreak.php

@@ -0,0 +1,76 @@
+<?php
+
+namespace Erusev\Parsedown\Components\Inlines;
+
+use Erusev\Parsedown\AST\StateRenderable;
+use Erusev\Parsedown\Components\Inline;
+use Erusev\Parsedown\Html\Renderables\Text;
+use Erusev\Parsedown\Parsedown;
+use Erusev\Parsedown\Parsing\Excerpt;
+use Erusev\Parsedown\State;
+
+final class SoftBreak implements Inline
+{
+    use WidthTrait;
+
+    /** @var int */
+    private $position;
+
+    /**
+     * @param int $width
+     * @param int $position
+     */
+    public function __construct($width, $position)
+    {
+        $this->width = $width;
+        $this->position = $position;
+    }
+
+    /**
+     * @param Excerpt $Excerpt
+     * @param State $State
+     * @return static|null
+     */
+    public static function build(Excerpt $Excerpt, State $State)
+    {
+        $context = $Excerpt->context();
+        $offset = $Excerpt->offset();
+
+        $trimTrailingWhitespaceBefore = \rtrim(\substr($context, 0, $offset));
+        $trimLeadingWhitespaceAfter = \ltrim(\substr($context, $offset + 1));
+        $contentLenBefore = \strlen($trimTrailingWhitespaceBefore);
+        $contentLenAfter = \strlen($trimLeadingWhitespaceAfter);
+
+        $originalLen = \strlen($context);
+        $afterWidth = $originalLen - $offset - $contentLenAfter;
+
+        return new self($offset + $afterWidth - $contentLenBefore, $contentLenBefore);
+    }
+
+    /**
+     * Return an integer to declare that the inline should be treated as if it
+     * started from that position in the excerpt given to static::build.
+     * Return null to use the excerpt offset value.
+     * @return int|null
+     * */
+    public function modifyStartPositionTo()
+    {
+        return $this->position;
+    }
+
+    /**
+     * @return Text
+     */
+    public function stateRenderable(Parsedown $_)
+    {
+        return new Text("\n");
+    }
+
+    /**
+     * @return Text
+     */
+    public function bestPlaintext()
+    {
+        return new Text("\n");
+    }
+}

src/Parsedown.php

@@ -22,10 +22,12 @@ use Erusev\Parsedown\Components\Inlines\Code;
 use Erusev\Parsedown\Components\Inlines\Email;
 use Erusev\Parsedown\Components\Inlines\Emphasis;
 use Erusev\Parsedown\Components\Inlines\EscapeSequence;
+use Erusev\Parsedown\Components\Inlines\HardBreak;
 use Erusev\Parsedown\Components\Inlines\Image;
 use Erusev\Parsedown\Components\Inlines\Link;
 use Erusev\Parsedown\Components\Inlines\Markup as InlineMarkup;
 use Erusev\Parsedown\Components\Inlines\PlainText;
+use Erusev\Parsedown\Components\Inlines\SoftBreak;
 use Erusev\Parsedown\Components\Inlines\SpecialCharacter;
 use Erusev\Parsedown\Components\Inlines\Strikethrough;
 use Erusev\Parsedown\Components\Inlines\Url;
@@ -41,48 +43,13 @@ use Erusev\Parsedown\Parsing\Lines;
 
 final class Parsedown
 {
-    # ~
-
     const version = '2.0.0-dev';
 
-    # ~
-
     /** @var State */
     private $State;
 
-    public function __construct(State $State = null)
-    {
-        $this->State = $State ?: new State;
-    }
-
-    /**
-     * @param string $text
-     * @return string
-     */
-    public function text($text)
-    {
-        $InitialState = $this->State;
-
-        $StateRenderables = $this->lines(Lines::fromTextLines($text, 0));
-
-        $Renderables = $this->State->applyTo($StateRenderables);
-
-        $this->State = $InitialState;
-
-        $html = self::render($Renderables);
-
-        # trim line breaks
-        $html = \trim($html, "\n");
-
-        return $html;
-    }
-
-    #
-    # Lines
-    #
-
     /** @var array<array-key, class-string<Block>[]> */
-    protected $BlockTypes = [
+    private $BlockTypes = [
         '#' => [Header::class],
         '*' => [Rule::class, TList::class],
         '+' => [TList::class],
@@ -108,16 +75,57 @@ final class Parsedown
         '~' => [FencedCode::class],
     ];
 
-    # ~
-
     /** @var class-string<Block>[] */
-    protected $unmarkedBlockTypes = [
+    private $unmarkedBlockTypes = [
         IndentedCode::class,
     ];
 
-    #
-    # Blocks
-    #
+    /** @var array<array-key, class-string<Inline>[]> */
+    private $InlineTypes = [
+        '!' => [Image::class],
+        '*' => [Emphasis::class],
+        '_' => [Emphasis::class],
+        '&' => [SpecialCharacter::class],
+        '[' => [Link::class],
+        ':' => [Url::class],
+        '<' => [UrlTag::class, Email::class, InlineMarkup::class],
+        '`' => [Code::class],
+        '~' => [Strikethrough::class],
+        '\\' => [EscapeSequence::class],
+        "\n" => [HardBreak::class, SoftBreak::class],
+    ];
+
+    /** @var string */
+    private $inlineMarkers;
+
+    public function __construct(State $State = null)
+    {
+        $this->State = $State ?: new State;
+
+        $this->inlineMarkers = \implode('', \array_keys($this->InlineTypes));
+    }
+
+    /**
+     * @param string $text
+     * @return string
+     */
+    public function text($text)
+    {
+        $InitialState = $this->State;
+
+        $StateRenderables = $this->lines(Lines::fromTextLines($text, 0));
+
+        $Renderables = $this->State->applyTo($StateRenderables);
+
+        $this->State = $InitialState;
+
+        $html = self::render($Renderables);
+
+        # trim line breaks
+        $html = \trim($html, "\n");
+
+        return $html;
+    }
 
     /**
      * @return StateRenderable[]
@@ -225,29 +233,6 @@ final class Parsedown
         return $Blocks;
     }
 
-    #
-    # Inline Elements
-    #
-
-    /** @var array<array-key, class-string<Inline>[]> */
-    protected $InlineTypes = [
-        '!' => [Image::class],
-        '&' => [SpecialCharacter::class],
-        '*' => [Emphasis::class],
-        ':' => [Url::class],
-        '<' => [UrlTag::class, Email::class, InlineMarkup::class],
-        '[' => [Link::class],
-        '_' => [Emphasis::class],
-        '`' => [Code::class],
-        '~' => [Strikethrough::class],
-        '\\' => [EscapeSequence::class],
-    ];
-
-    # ~
-
-    /** @var string */
-    protected $inlineMarkerList = '!*_&[:<`~\\';
-
     #
     # ~
     #
@@ -280,9 +265,9 @@ final class Parsedown
         # $excerpt is based on the first occurrence of a marker
 
         for (
-            $Excerpt = (new Excerpt($text, 0))->pushingOffsetTo($this->inlineMarkerList);
+            $Excerpt = (new Excerpt($text, 0))->pushingOffsetTo($this->inlineMarkers);
             $Excerpt->text() !== '';
-            $Excerpt = $Excerpt->pushingOffsetTo($this->inlineMarkerList)
+            $Excerpt = $Excerpt->pushingOffsetTo($this->inlineMarkers)
         ) {
             $marker = \substr($Excerpt->text(), 0, 1);