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