a-toy-markdown-parser.php

<?php
declare(strict_types=1);

class Node
{
    protected string $text;
    protected array $params;
    public function __construct(string $text, ...$params)
    {
        $this->text = $text;
        $this->params = $params;
    }
}

class Heading extends Node
{
}

class Bold extends Node
{
}

class Italic extends Node
{
}

class Text extends Node
{
}

class Parser
{
    protected string $text;
    protected int $size;

    public function __construct(string $text)
    {
        $this->text = $text;
        $this->size = mb_strlen($this->text);
    }

    public function parse(): array
    {
        $nodes = [];
        $line = 1;
        $linePos = 0;
        $buffer = '';

        $addTextNode = function () use (&$buffer, &$nodes) {
            if ($buffer === '') {
                return;
            }
            $nodes[] = new Text($buffer);
            $buffer = '';
        };

        for ($i = 0; $i < $this->size; $i++, $linePos++) {
            $char = mb_substr($this->text, $i, 1);

            switch ($char) {
                case '#':
                    if ($linePos !== 0) {
                        $buffer .= $char;
                        break;
                    }
                    $headingSize = $this->readSameToken('#', $i);

                    $addTextNode();
                    $nodes[] = new Heading(
                        ltrim($this->readToLineBreak($i)),
                        strlen($headingSize)
                    );
                    break;
                case '*':
                    $asterSize = strlen($this->readSameToken('*', $i));

                    $class = null;
                    if ($asterSize === 1) {
                        $text = $this->readToToken('*', $i);
                        $class = Bold::class;
                    } elseif ($asterSize === 2) {
                        $text = $this->readToToken('**', $i);
                        $class = Italic::class;
                    }

                    $addTextNode();
                    $nodes[] = new $class($text);

                    $i++;
                    break;
                case "\n":
                    $line++;
                    $linePos = -1;
                    break;
                default:
                    $buffer .= $char;
                    break;
            }
        }

        $addTextNode();
        return $nodes;
    }

    protected function readToLineBreak(&$i): string
    {
        return $this->readToToken("\n", $i);
    }

    protected function readToToken(string $token, &$i): string
    {
        $text = '';
        $tokenSize = mb_strlen($token);
        for (; $i < $this->size; $i++) {
            $char = mb_substr($this->text, $i, $tokenSize);
            if ($char !== $token) {
                $text .= mb_substr($this->text, $i, 1);
            } else {
                // Back
                $i += $tokenSize - 1;
                $i--;
                break;
            }
        }
        return $text;
    }


    protected function readSameToken(string $token, &$i): string
    {
        $text = '';
        $tokenSize = mb_strlen($token);
        for (; $i < $this->size; $i++) {
            $char = mb_substr($this->text, $i, $tokenSize);
            if ($char !== $token) {
                $i += $tokenSize - 1;
                break;
            }
            $text .= mb_substr($this->text, $i, 1);
        }
        return $text;
    }
}

/**
array(9) {
  [0]=>
  object(Heading)#3 (2) {
    ["text":protected]=>
    string(4) "aaaa"
    ["params":protected]=>
    array(1) {
      [0]=>
      int(1)
    }
  }
  [1]=>
  object(Heading)#4 (2) {
    ["text":protected]=>
    string(4) "aaaa"
    ["params":protected]=>
    array(1) {
      [0]=>
      int(2)
    }
  }
  [2]=>
  object(Text)#5 (2) {
    ["text":protected]=>
    string(8) "hogehoge"
    ["params":protected]=>
    array(0) {
    }
  }
  [3]=>
  object(Italic)#6 (2) {
    ["text":protected]=>
    string(8) "biyobiyo"
    ["params":protected]=>
    array(0) {
    }
  }
  [4]=>
  object(Text)#7 (2) {
    ["text":protected]=>
    string(4) "nyan"
    ["params":protected]=>
    array(0) {
    }
  }
  [5]=>
  object(Bold)#8 (2) {
    ["text":protected]=>
    string(4) "neko"
    ["params":protected]=>
    array(0) {
    }
  }
  [6]=>
  object(Text)#9 (2) {
    ["text":protected]=>
    string(4) "guee"
    ["params":protected]=>
    array(0) {
    }
  }
  [7]=>
  object(Heading)#10 (2) {
    ["text":protected]=>
    string(4) "aaaa"
    ["params":protected]=>
    array(1) {
      [0]=>
      int(3)
    }
  }
  [8]=>
  object(Text)#11 (2) {
    ["text":protected]=>
    string(5) "nya-n"
    ["params":protected]=>
    array(0) {
    }
  }
}
 */

var_dump(
    (new Parser(
        $markdown = <<< _
        # aaaa
        ## aaaa
        hogehoge**biyobiyo**nyan*neko*guee
        
        ### aaaa
        nya-n
        _
    ))->parse()
);