plugins/markdown/markdown.php

   1 <?php
   2
   3 /**
   4  * Plugin Markdown.
   5  *
   6  * Shaare's descriptions are parsed with Markdown.
   7  */
   8
   9 use Shaarli\Config\ConfigManager;
  10 use Shaarli\Router;
  11
  12 /*
  13  * If this tag is used on a shaare, the description won't be processed by Parsedown.
  14  */
  15 define('NO_MD_TAG', 'nomarkdown');
  16
  17 /**
  18  * Parse linklist descriptions.
  19  *
  20  * @param array         $data linklist data.
  21  * @param ConfigManager $conf instance.
  22  *
  23  * @return mixed linklist data parsed in markdown (and converted to HTML).
  24  */
  25 function hook_markdown_render_linklist($data, $conf)
  26 {
  27     foreach ($data['links'] as &$value) {
  28         if (!empty($value['tags']) && noMarkdownTag($value['tags'])) {
  29             $value = stripNoMarkdownTag($value);
  30             continue;
  31         }
  32         $value['description_src'] = $value['description'];
  33         $value['description'] = process_markdown(
  34             $value['description'],
  35             $conf->get('security.markdown_escape', true),
  36             $conf->get('security.allowed_protocols')
  37         );
  38     }
  39     return $data;
  40 }
  41
  42 /**
  43  * Parse feed linklist descriptions.
  44  *
  45  * @param array $data linklist data.
  46  * @param ConfigManager $conf instance.
  47  *
  48  * @return mixed linklist data parsed in markdown (and converted to HTML).
  49  */
  50 function hook_markdown_render_feed($data, $conf)
  51 {
  52     foreach ($data['links'] as &$value) {
  53         if (!empty($value['tags']) && noMarkdownTag($value['tags'])) {
  54             $value = stripNoMarkdownTag($value);
  55             continue;
  56         }
  57         $value['description'] = reverse_feed_permalink($value['description']);
  58         $value['description'] = process_markdown(
  59             $value['description'],
  60             $conf->get('security.markdown_escape', true),
  61             $conf->get('security.allowed_protocols')
  62         );
  63     }
  64
  65     return $data;
  66 }
  67
  68 /**
  69  * Parse daily descriptions.
  70  *
  71  * @param array         $data daily data.
  72  * @param ConfigManager $conf instance.
  73  *
  74  * @return mixed daily data parsed in markdown (and converted to HTML).
  75  */
  76 function hook_markdown_render_daily($data, $conf)
  77 {
  78     //var_dump($data);die;
  79     // Manipulate columns data
  80     foreach ($data['linksToDisplay'] as &$value) {
  81         if (!empty($value['tags']) && noMarkdownTag($value['tags'])) {
  82             $value = stripNoMarkdownTag($value);
  83             continue;
  84         }
  85         $value['formatedDescription'] = process_markdown(
  86             $value['formatedDescription'],
  87             $conf->get('security.markdown_escape', true),
  88             $conf->get('security.allowed_protocols')
  89         );
  90     }
  91
  92     return $data;
  93 }
  94
  95 /**
  96  * Check if noMarkdown is set in tags.
  97  *
  98  * @param string $tags tag list
  99  *
 100  * @return bool true if markdown should be disabled on this link.
 101  */
 102 function noMarkdownTag($tags)
 103 {
 104     return preg_match('/(^|\s)'. NO_MD_TAG .'(\s|$)/', $tags);
 105 }
 106
 107 /**
 108  * Remove the no-markdown meta tag so it won't be displayed.
 109  *
 110  * @param array $link Link data.
 111  *
 112  * @return array Updated link without no markdown tag.
 113  */
 114 function stripNoMarkdownTag($link)
 115 {
 116     if (! empty($link['taglist'])) {
 117         $offset = array_search(NO_MD_TAG, $link['taglist']);
 118         if ($offset !== false) {
 119             unset($link['taglist'][$offset]);
 120         }
 121     }
 122
 123     if (!empty($link['tags'])) {
 124         str_replace(NO_MD_TAG, '', $link['tags']);
 125     }
 126
 127     return $link;
 128 }
 129
 130 /**
 131  * When link list is displayed, include markdown CSS.
 132  *
 133  * @param array $data includes data.
 134  *
 135  * @return mixed - includes data with markdown CSS file added.
 136  */
 137 function hook_markdown_render_includes($data)
 138 {
 139     if ($data['_PAGE_'] == Router::$PAGE_LINKLIST
 140         || $data['_PAGE_'] == Router::$PAGE_DAILY
 141         || $data['_PAGE_'] == Router::$PAGE_EDITLINK
 142     ) {
 143         $data['css_files'][] = PluginManager::$PLUGINS_PATH . '/markdown/markdown.css';
 144     }
 145
 146     return $data;
 147 }
 148
 149 /**
 150  * Hook render_editlink.
 151  * Adds an help link to markdown syntax.
 152  *
 153  * @param array $data data passed to plugin
 154  *
 155  * @return array altered $data.
 156  */
 157 function hook_markdown_render_editlink($data)
 158 {
 159     // Load help HTML into a string
 160     $txt = file_get_contents(PluginManager::$PLUGINS_PATH .'/markdown/help.html');
 161     $translations = [
 162         t('Description will be rendered with'),
 163         t('Markdown syntax documentation'),
 164         t('Markdown syntax'),
 165     ];
 166     $data['edit_link_plugin'][] = vsprintf($txt, $translations);
 167     // Add no markdown 'meta-tag' in tag list if it was never used, for autocompletion.
 168     if (! in_array(NO_MD_TAG, $data['tags'])) {
 169         $data['tags'][NO_MD_TAG] = 0;
 170     }
 171
 172     return $data;
 173 }
 174
 175
 176 /**
 177  * Remove HTML links auto generated by Shaarli core system.
 178  * Keeps HREF attributes.
 179  *
 180  * @param string $description input description text.
 181  *
 182  * @return string $description without HTML links.
 183  */
 184 function reverse_text2clickable($description)
 185 {
 186     $descriptionLines = explode(PHP_EOL, $description);
 187     $descriptionOut = '';
 188     $codeBlockOn = false;
 189     $lineCount = 0;
 190
 191     foreach ($descriptionLines as $descriptionLine) {
 192         // Detect line of code: starting with 4 spaces,
 193         // except lists which can start with +/*/- or `2.` after spaces.
 194         $codeLineOn = preg_match('/^    +(?=[^\+\*\-])(?=(?!\d\.).)/', $descriptionLine) > 0;
 195         // Detect and toggle block of code
 196         if (!$codeBlockOn) {
 197             $codeBlockOn = preg_match('/^```/', $descriptionLine) > 0;
 198         } elseif (preg_match('/^```/', $descriptionLine) > 0) {
 199             $codeBlockOn = false;
 200         }
 201
 202         $hashtagTitle = ' title="Hashtag [^"]+"';
 203         // Reverse `inline code` hashtags.
 204         $descriptionLine = preg_replace(
 205             '!(`[^`\n]*)<a href="[^ ]*"'. $hashtagTitle .'>([^<]+)</a>([^`\n]*`)!m',
 206             '$1$2$3',
 207             $descriptionLine
 208         );
 209
 210         // Reverse all links in code blocks, only non hashtag elsewhere.
 211         $hashtagFilter = (!$codeBlockOn && !$codeLineOn) ? '(?!'. $hashtagTitle .')': '(?:'. $hashtagTitle .')?';
 212         $descriptionLine = preg_replace(
 213             '#<a href="[^ ]*"'. $hashtagFilter .'>([^<]+)</a>#m',
 214             '$1',
 215             $descriptionLine
 216         );
 217
 218         // Make hashtag links markdown ready, otherwise the links will be ignored with escape set to true
 219         if (!$codeBlockOn && !$codeLineOn) {
 220             $descriptionLine = preg_replace(
 221                 '#<a href="([^ ]*)"'. $hashtagTitle .'>([^<]+)</a>#m',
 222                 '[$2]($1)',
 223                 $descriptionLine
 224             );
 225         }
 226
 227         $descriptionOut .= $descriptionLine;
 228         if ($lineCount++ < count($descriptionLines) - 1) {
 229             $descriptionOut .= PHP_EOL;
 230         }
 231     }
 232     return $descriptionOut;
 233 }
 234
 235 /**
 236  * Remove <br> tag to let markdown handle it.
 237  *
 238  * @param string $description input description text.
 239  *
 240  * @return string $description without <br> tags.
 241  */
 242 function reverse_nl2br($description)
 243 {
 244     return preg_replace('!<br */?>!im', '', $description);
 245 }
 246
 247 /**
 248  * Remove HTML spaces '&nbsp;' auto generated by Shaarli core system.
 249  *
 250  * @param string $description input description text.
 251  *
 252  * @return string $description without HTML links.
 253  */
 254 function reverse_space2nbsp($description)
 255 {
 256     return preg_replace('/(^| )&nbsp;/m', '$1 ', $description);
 257 }
 258
 259 function reverse_feed_permalink($description)
 260 {
 261     return preg_replace('@&#8212; <a href="([^"]+)" title="[^"]+">(\w+)</a>$@im', '&#8212; [$2]($1)', $description);
 262 }
 263
 264 /**
 265  * Replace not whitelisted protocols with http:// in given description.
 266  *
 267  * @param string $description      input description text.
 268  * @param array  $allowedProtocols list of allowed protocols.
 269  *
 270  * @return string $description without malicious link.
 271  */
 272 function filter_protocols($description, $allowedProtocols)
 273 {
 274     return preg_replace_callback(
 275         '#]\((.*?)\)#is',
 276         function ($match) use ($allowedProtocols) {
 277             return ']('. whitelist_protocols($match[1], $allowedProtocols) .')';
 278         },
 279         $description
 280     );
 281 }
 282
 283 /**
 284  * Remove dangerous HTML tags (tags, iframe, etc.).
 285  * Doesn't affect <code> content (already escaped by Parsedown).
 286  *
 287  * @param string $description input description text.
 288  *
 289  * @return string given string escaped.
 290  */
 291 function sanitize_html($description)
 292 {
 293     $escapeTags = array(
 294         'script',
 295         'style',
 296         'link',
 297         'iframe',
 298         'frameset',
 299         'frame',
 300     );
 301     foreach ($escapeTags as $tag) {
 302         $description = preg_replace_callback(
 303             '#<\s*'. $tag .'[^>]*>(.*</\s*'. $tag .'[^>]*>)?#is',
 304             function ($match) {
 305                 return escape($match[0]);
 306             },
 307             $description
 308         );
 309     }
 310     $description = preg_replace(
 311         '#(<[^>]+\s)on[a-z]*="?[^ "]*"?#is',
 312         '$1',
 313         $description
 314     );
 315     return $description;
 316 }
 317
 318 /**
 319  * Render shaare contents through Markdown parser.
 320  *   1. Remove HTML generated by Shaarli core.
 321  *   2. Reverse the escape function.
 322  *   3. Generate markdown descriptions.
 323  *   4. Sanitize sensible HTML tags for security.
 324  *   5. Wrap description in 'markdown' CSS class.
 325  *
 326  * @param string $description input description text.
 327  * @param bool   $escape      escape HTML entities
 328  *
 329  * @return string HTML processed $description.
 330  */
 331 function process_markdown($description, $escape = true, $allowedProtocols = [])
 332 {
 333     $parsedown = new Parsedown();
 334
 335     $processedDescription = $description;
 336     $processedDescription = reverse_nl2br($processedDescription);
 337     $processedDescription = reverse_space2nbsp($processedDescription);
 338     $processedDescription = reverse_text2clickable($processedDescription);
 339     $processedDescription = filter_protocols($processedDescription, $allowedProtocols);
 340     $processedDescription = unescape($processedDescription);
 341     $processedDescription = $parsedown
 342         ->setMarkupEscaped($escape)
 343         ->setBreaksEnabled(true)
 344         ->text($processedDescription);
 345     $processedDescription = sanitize_html($processedDescription);
 346
 347     if (!empty($processedDescription)) {
 348         $processedDescription = '<div class="markdown">'. $processedDescription . '</div>';
 349     }
 350
 351     return $processedDescription;
 352 }
 353
 354 /**
 355  * This function is never called, but contains translation calls for GNU gettext extraction.
 356  */
 357 function markdown_dummy_translation()
 358 {
 359     // meta
 360     t('Render shaare description with Markdown syntax.<br><strong>Warning</strong>:
 361 If your shaared descriptions contained HTML tags before enabling the markdown plugin,
 362 enabling it might break your page.
 363 See the <a href="https://github.com/shaarli/Shaarli/tree/master/plugins/markdown#html-rendering">README</a>.');
 364 }