application/formatter/BookmarkMarkdownFormatter.php

   1 <?php
   2
   3 namespace Shaarli\Formatter;
   4
   5 use Shaarli\Config\ConfigManager;
   6
   7 /**
   8  * Class BookmarkMarkdownFormatter
   9  *
  10  * Format bookmark description into Markdown format.
  11  *
  12  * @package Shaarli\Formatter
  13  */
  14 class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter
  15 {
  16     /**
  17      * When this tag is present in a bookmark, its description should not be processed with Markdown
  18      */
  19     const NO_MD_TAG = 'nomarkdown';
  20
  21     /** @var \Parsedown instance */
  22     protected $parsedown;
  23
  24     /** @var bool used to escape HTML in Markdown or not.
  25      *            It MUST be set to true for shared instance as HTML content can
  26      *            introduce XSS vulnerabilities.
  27      */
  28     protected $escape;
  29
  30     /**
  31      * @var array List of allowed protocols for links inside bookmark's description.
  32      */
  33     protected $allowedProtocols;
  34
  35     /**
  36      * LinkMarkdownFormatter constructor.
  37      *
  38      * @param ConfigManager $conf instance
  39      */
  40     public function __construct(ConfigManager $conf)
  41     {
  42         parent::__construct($conf);
  43         $this->parsedown = new \Parsedown();
  44         $this->escape = $conf->get('security.markdown_escape', true);
  45         $this->allowedProtocols = $conf->get('security.allowed_protocols', []);
  46     }
  47
  48     /**
  49      * @inheritdoc
  50      */
  51     public function formatDescription($bookmark)
  52     {
  53         if (in_array(self::NO_MD_TAG, $bookmark->getTags())) {
  54             return parent::formatDescription($bookmark);
  55         }
  56
  57         $processedDescription = $bookmark->getDescription();
  58         $processedDescription = $this->filterProtocols($processedDescription);
  59         $processedDescription = $this->formatHashTags($processedDescription);
  60         $processedDescription = $this->parsedown
  61             ->setMarkupEscaped($this->escape)
  62             ->setBreaksEnabled(true)
  63             ->text($processedDescription);
  64         $processedDescription = $this->sanitizeHtml($processedDescription);
  65
  66         if (!empty($processedDescription)) {
  67             $processedDescription = '<div class="markdown">'. $processedDescription . '</div>';
  68         }
  69
  70         return $processedDescription;
  71     }
  72
  73     /**
  74      * Remove the NO markdown tag if it is present
  75      *
  76      * @inheritdoc
  77      */
  78     protected function formatTagList($bookmark)
  79     {
  80         $out = parent::formatTagList($bookmark);
  81         if (($pos = array_search(self::NO_MD_TAG, $out)) !== false) {
  82             unset($out[$pos]);
  83             return array_values($out);
  84         }
  85         return $out;
  86     }
  87
  88     /**
  89      * Replace not whitelisted protocols with http:// in given description.
  90      * Also adds `index_url` to relative links if it's specified
  91      *
  92      * @param string $description      input description text.
  93      *
  94      * @return string $description without malicious link.
  95      */
  96     protected function filterProtocols($description)
  97     {
  98         $allowedProtocols = $this->allowedProtocols;
  99         $indexUrl = ! empty($this->contextData['index_url']) ? $this->contextData['index_url'] : '';
 100
 101         return preg_replace_callback(
 102             '#]\((.*?)\)#is',
 103             function ($match) use ($allowedProtocols, $indexUrl) {
 104                 $link = startsWith($match[1], '?') || startsWith($match[1], '/') ? $indexUrl : '';
 105                 $link .= whitelist_protocols($match[1], $allowedProtocols);
 106                 return ']('. $link.')';
 107             },
 108             $description
 109         );
 110     }
 111
 112     /**
 113      * Replace hashtag in Markdown links format
 114      * E.g. `#hashtag` becomes `[#hashtag](?addtag=hashtag)`
 115      * It includes the index URL if specified.
 116      *
 117      * @param string $description
 118      *
 119      * @return string
 120      */
 121     protected function formatHashTags($description)
 122     {
 123         $indexUrl = ! empty($this->contextData['index_url']) ? $this->contextData['index_url'] : '';
 124
 125         /*
 126          * To support unicode: http://stackoverflow.com/a/35498078/1484919
 127          * \p{Pc} - to match underscore
 128          * \p{N} - numeric character in any script
 129          * \p{L} - letter from any language
 130          * \p{Mn} - any non marking space (accents, umlauts, etc)
 131          */
 132         $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
 133         $replacement = '$1[#$2]('. $indexUrl .'?addtag=$2)';
 134
 135         $descriptionLines = explode(PHP_EOL, $description);
 136         $descriptionOut = '';
 137         $codeBlockOn = false;
 138         $lineCount = 0;
 139
 140         foreach ($descriptionLines as $descriptionLine) {
 141             // Detect line of code: starting with 4 spaces,
 142             // except lists which can start with +/*/- or `2.` after spaces.
 143             $codeLineOn = preg_match('/^    +(?=[^\+\*\-])(?=(?!\d\.).)/', $descriptionLine) > 0;
 144             // Detect and toggle block of code
 145             if (!$codeBlockOn) {
 146                 $codeBlockOn = preg_match('/^```/', $descriptionLine) > 0;
 147             } elseif (preg_match('/^```/', $descriptionLine) > 0) {
 148                 $codeBlockOn = false;
 149             }
 150
 151             if (!$codeBlockOn && !$codeLineOn) {
 152                 $descriptionLine = preg_replace($regex, $replacement, $descriptionLine);
 153             }
 154
 155             $descriptionOut .= $descriptionLine;
 156             if ($lineCount++ < count($descriptionLines) - 1) {
 157                 $descriptionOut .= PHP_EOL;
 158             }
 159         }
 160
 161         return $descriptionOut;
 162     }
 163
 164     /**
 165      * Remove dangerous HTML tags (tags, iframe, etc.).
 166      * Doesn't affect <code> content (already escaped by Parsedown).
 167      *
 168      * @param string $description input description text.
 169      *
 170      * @return string given string escaped.
 171      */
 172     protected function sanitizeHtml($description)
 173     {
 174         $escapeTags = array(
 175             'script',
 176             'style',
 177             'link',
 178             'iframe',
 179             'frameset',
 180             'frame',
 181         );
 182         foreach ($escapeTags as $tag) {
 183             $description = preg_replace_callback(
 184                 '#<\s*'. $tag .'[^>]*>(.*</\s*'. $tag .'[^>]*>)?#is',
 185                 function ($match) {
 186                     return escape($match[0]);
 187                 },
 188                 $description
 189             );
 190         }
 191         $description = preg_replace(
 192             '#(<[^>]+\s)on[a-z]*="?[^ "]*"?#is',
 193             '$1',
 194             $description
 195         );
 196         return $description;
 197     }
 198 }