gfm.php
420 lines
| 1 | <?php |
| 2 | /** |
| 3 | * GitHub-Flavoured Markdown. Inspired by Evan's plugin, but modified. |
| 4 | * |
| 5 | * @author Evan Solomon |
| 6 | * @author Matt Wiebe <wiebe@automattic.com> |
| 7 | * @link https://github.com/evansolomon/wp-github-flavored-markdown-comments |
| 8 | * |
| 9 | * Add a few extras from GitHub's Markdown implementation. Must be used in a WordPress environment. |
| 10 | */ |
| 11 | |
| 12 | class WPCom_GHF_Markdown_Parser extends MarkdownExtra_Parser { |
| 13 | |
| 14 | /** |
| 15 | * Hooray somewhat arbitrary numbers that are fearful of 1.0.x. |
| 16 | */ |
| 17 | const WPCOM_GHF_MARDOWN_VERSION = '0.9.0'; |
| 18 | |
| 19 | /** |
| 20 | * Use a [code] shortcode when encountering a fenced code block |
| 21 | * @var boolean |
| 22 | */ |
| 23 | public $use_code_shortcode = true; |
| 24 | |
| 25 | /** |
| 26 | * Preserve shortcodes, untouched by Markdown. |
| 27 | * This requires use within a WordPress installation. |
| 28 | * @var boolean |
| 29 | */ |
| 30 | public $preserve_shortcodes = true; |
| 31 | |
| 32 | /** |
| 33 | * Preserve the legacy $latex your-latex-code-here$ style |
| 34 | * LaTeX markup |
| 35 | */ |
| 36 | public $preserve_latex = true; |
| 37 | |
| 38 | /** |
| 39 | * Preserve single-line <code> blocks. |
| 40 | * @var boolean |
| 41 | */ |
| 42 | public $preserve_inline_code_blocks = true; |
| 43 | |
| 44 | /** |
| 45 | * Strip paragraphs from the output. This is the right default for WordPress, |
| 46 | * which generally wants to create its own paragraphs with `wpautop` |
| 47 | * @var boolean |
| 48 | */ |
| 49 | public $strip_paras = true; |
| 50 | |
| 51 | // Will run through sprintf - you can supply your own syntax if you want |
| 52 | public $shortcode_start = '[code lang=%s]'; |
| 53 | public $shortcode_end = '[/code]'; |
| 54 | |
| 55 | // Stores shortcodes we remove and then replace |
| 56 | protected $preserve_text_hash = array(); |
| 57 | |
| 58 | /** |
| 59 | * Set environment defaults based on presence of key functions/classes. |
| 60 | */ |
| 61 | public function __construct() { |
| 62 | $this->use_code_shortcode = class_exists( 'SyntaxHighlighter' ); |
| 63 | /** |
| 64 | * Allow processing shortcode contents. |
| 65 | * |
| 66 | * @module markdown |
| 67 | * |
| 68 | * @since 4.4.0 |
| 69 | * |
| 70 | * @param boolean $preserve_shortcodes Defaults to $this->preserve_shortcodes. |
| 71 | */ |
| 72 | $this->preserve_shortcodes = apply_filters( 'jetpack_markdown_preserve_shortcodes', $this->preserve_shortcodes ) && function_exists( 'get_shortcode_regex' ); |
| 73 | $this->preserve_latex = function_exists( 'latex_markup' ); |
| 74 | $this->strip_paras = function_exists( 'wpautop' ); |
| 75 | |
| 76 | parent::__construct(); |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * Overload to specify heading styles only if the hash has space(s) after it. This is actually in keeping with |
| 81 | * the documentation and eases the semantic overload of the hash character. |
| 82 | * #Will Not Produce a Heading 1 |
| 83 | * # This Will Produce a Heading 1 |
| 84 | * |
| 85 | * @param string $text Markdown text |
| 86 | * @return string HTML-transformed text |
| 87 | */ |
| 88 | public function transform( $text ) { |
| 89 | // Preserve anything inside a single-line <code> element |
| 90 | if ( $this->preserve_inline_code_blocks ) { |
| 91 | $text = $this->single_line_code_preserve( $text ); |
| 92 | } |
| 93 | // Remove all shortcodes so their interiors are left intact |
| 94 | if ( $this->preserve_shortcodes ) { |
| 95 | $text = $this->shortcode_preserve( $text ); |
| 96 | } |
| 97 | // Remove legacy LaTeX so it's left intact |
| 98 | if ( $this->preserve_latex ) { |
| 99 | $text = $this->latex_preserve( $text ); |
| 100 | } |
| 101 | |
| 102 | // Do not process characters inside URLs. |
| 103 | $text = $this->urls_preserve( $text ); |
| 104 | |
| 105 | // escape line-beginning # chars that do not have a space after them. |
| 106 | $text = preg_replace_callback( '|^#{1,6}( )?|um', array( $this, '_doEscapeForHashWithoutSpacing' ), $text ); |
| 107 | |
| 108 | /** |
| 109 | * Allow third-party plugins to define custom patterns that won't be processed by Markdown. |
| 110 | * |
| 111 | * @module markdown |
| 112 | * |
| 113 | * @since 3.9.2 |
| 114 | * |
| 115 | * @param array $custom_patterns Array of custom patterns to be ignored by Markdown. |
| 116 | */ |
| 117 | $custom_patterns = apply_filters( 'jetpack_markdown_preserve_pattern', array() ); |
| 118 | if ( is_array( $custom_patterns ) && ! empty( $custom_patterns ) ) { |
| 119 | foreach ( $custom_patterns as $pattern ) { |
| 120 | $text = preg_replace_callback( $pattern, array( $this, '_doRemoveText'), $text ); |
| 121 | } |
| 122 | } |
| 123 | |
| 124 | // run through core Markdown |
| 125 | $text = parent::transform( $text ); |
| 126 | |
| 127 | // Occasionally Markdown Extra chokes on a para structure, producing odd paragraphs. |
| 128 | $text = str_replace( "<p><</p>\n\n<p>p>", '<p>', $text ); |
| 129 | |
| 130 | // put start-of-line # chars back in place |
| 131 | $text = $this->restore_leading_hash( $text ); |
| 132 | |
| 133 | // Strip paras if set |
| 134 | if ( $this->strip_paras ) { |
| 135 | $text = $this->unp( $text ); |
| 136 | } |
| 137 | |
| 138 | // Restore preserved things like shortcodes/LaTeX |
| 139 | $text = $this->do_restore( $text ); |
| 140 | |
| 141 | return $text; |
| 142 | } |
| 143 | |
| 144 | /** |
| 145 | * Prevents blocks like <code>__this__</code> from turning into <code><strong>this</strong></code> |
| 146 | * @param string $text Text that may need preserving |
| 147 | * @return string Text that was preserved if needed |
| 148 | */ |
| 149 | public function single_line_code_preserve( $text ) { |
| 150 | return preg_replace_callback( '|<code\b[^>]*>(.*?)</code>|', array( $this, 'do_single_line_code_preserve' ), $text ); |
| 151 | } |
| 152 | |
| 153 | /** |
| 154 | * Regex callback for inline code presevation |
| 155 | * @param array $matches Regex matches |
| 156 | * @return string Hashed content for later restoration |
| 157 | */ |
| 158 | public function do_single_line_code_preserve( $matches ) { |
| 159 | return '<code>' . $this->hash_block( $matches[1] ) . '</code>'; |
| 160 | } |
| 161 | |
| 162 | /** |
| 163 | * Preserve code block contents by HTML encoding them. Useful before getting to KSES stripping. |
| 164 | * @param string $text Markdown/HTML content |
| 165 | * @return string Markdown/HTML content with escaped code blocks |
| 166 | */ |
| 167 | public function codeblock_preserve( $text ) { |
| 168 | return preg_replace_callback( "/^([`~]{3})([^`\n]+)?\n([^`~]+)(\\1)/m", array( $this, 'do_codeblock_preserve' ), $text ); |
| 169 | } |
| 170 | |
| 171 | /** |
| 172 | * Regex callback for code block preservation. |
| 173 | * @param array $matches Regex matches |
| 174 | * @return string Codeblock with escaped interior |
| 175 | */ |
| 176 | public function do_codeblock_preserve( $matches ) { |
| 177 | $block = stripslashes( $matches[3] ); |
| 178 | $block = esc_html( $block ); |
| 179 | $block = str_replace( '\\', '\\\\', $block ); |
| 180 | $open = $matches[1] . $matches[2] . "\n"; |
| 181 | return $open . $block . $matches[4]; |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Restore previously preserved (i.e. escaped) code block contents. |
| 186 | * @param string $text Markdown/HTML content with escaped code blocks |
| 187 | * @return string Markdown/HTML content |
| 188 | */ |
| 189 | public function codeblock_restore( $text ) { |
| 190 | return preg_replace_callback( "/^([`~]{3})([^`\n]+)?\n([^`~]+)(\\1)/m", array( $this, 'do_codeblock_restore' ), $text ); |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Regex callback for code block restoration (unescaping). |
| 195 | * @param array $matches Regex matches |
| 196 | * @return string Codeblock with unescaped interior |
| 197 | */ |
| 198 | public function do_codeblock_restore( $matches ) { |
| 199 | $block = html_entity_decode( $matches[3], ENT_QUOTES ); |
| 200 | $open = $matches[1] . $matches[2] . "\n"; |
| 201 | return $open . $block . $matches[4]; |
| 202 | } |
| 203 | |
| 204 | /** |
| 205 | * Called to preserve legacy LaTeX like $latex some-latex-text $ |
| 206 | * @param string $text Text in which to preserve LaTeX |
| 207 | * @return string Text with LaTeX replaced by a hash that will be restored later |
| 208 | */ |
| 209 | protected function latex_preserve( $text ) { |
| 210 | // regex from latex_remove() |
| 211 | $regex = '% |
| 212 | \$latex(?:=\s*|\s+) |
| 213 | ((?: |
| 214 | [^$]+ # Not a dollar |
| 215 | | |
| 216 | (?<=(?<!\\\\)\\\\)\$ # Dollar preceded by exactly one slash |
| 217 | )+) |
| 218 | (?<!\\\\)\$ # Dollar preceded by zero slashes |
| 219 | %ix'; |
| 220 | $text = preg_replace_callback( $regex, array( $this, '_doRemoveText'), $text ); |
| 221 | return $text; |
| 222 | } |
| 223 | |
| 224 | /** |
| 225 | * Called to preserve WP shortcodes from being formatted by Markdown in any way. |
| 226 | * @param string $text Text in which to preserve shortcodes |
| 227 | * @return string Text with shortcodes replaced by a hash that will be restored later |
| 228 | */ |
| 229 | protected function shortcode_preserve( $text ) { |
| 230 | $text = preg_replace_callback( $this->get_shortcode_regex(), array( $this, '_doRemoveText' ), $text ); |
| 231 | return $text; |
| 232 | } |
| 233 | |
| 234 | /** |
| 235 | * Avoid characters inside URLs from being formatted by Markdown in any way. |
| 236 | * |
| 237 | * @param string $text Text in which to preserve URLs. |
| 238 | * |
| 239 | * @return string Text with URLs replaced by a hash that will be restored later. |
| 240 | */ |
| 241 | protected function urls_preserve( $text ) { |
| 242 | $text = preg_replace_callback( |
| 243 | '#(?<!<)(?:https?|ftp)://([^\s<>"\'\[\]()]+|\[(?1)*+\]|\((?1)*+\))+(?<![_*.?])#i', |
| 244 | array( $this, '_doRemoveText' ), |
| 245 | $text |
| 246 | ); |
| 247 | return $text; |
| 248 | } |
| 249 | |
| 250 | /** |
| 251 | * Restores any text preserved by $this->hash_block() |
| 252 | * @param string $text Text that may have hashed preservation placeholders |
| 253 | * @return string Text with hashed preseravtion placeholders replaced by original text |
| 254 | */ |
| 255 | protected function do_restore( $text ) { |
| 256 | // Reverse hashes to ensure nested blocks are restored. |
| 257 | $hashes = array_reverse( $this->preserve_text_hash, true ); |
| 258 | foreach( $hashes as $hash => $value ) { |
| 259 | $placeholder = $this->hash_maker( $hash ); |
| 260 | $text = str_replace( $placeholder, $value, $text ); |
| 261 | } |
| 262 | // reset the hash |
| 263 | $this->preserve_text_hash = array(); |
| 264 | return $text; |
| 265 | } |
| 266 | |
| 267 | /** |
| 268 | * Regex callback for text preservation |
| 269 | * @param array $m Regex $matches array |
| 270 | * @return string A placeholder that will later be replaced by the original text |
| 271 | */ |
| 272 | protected function _doRemoveText( $m ) { |
| 273 | return $this->hash_block( $m[0] ); |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Call this to store a text block for later restoration. |
| 278 | * @param string $text Text to preserve for later |
| 279 | * @return string Placeholder that will be swapped out later for the original text |
| 280 | */ |
| 281 | protected function hash_block( $text ) { |
| 282 | $hash = md5( $text ); |
| 283 | $this->preserve_text_hash[ $hash ] = $text; |
| 284 | $placeholder = $this->hash_maker( $hash ); |
| 285 | return $placeholder; |
| 286 | } |
| 287 | |
| 288 | /** |
| 289 | * Less glamorous than the Keymaker |
| 290 | * @param string $hash An md5 hash |
| 291 | * @return string A placeholder hash |
| 292 | */ |
| 293 | protected function hash_maker( $hash ) { |
| 294 | return 'MARKDOWN_HASH' . $hash . 'MARKDOWN_HASH'; |
| 295 | } |
| 296 | |
| 297 | /** |
| 298 | * Remove bare <p> elements. <p>s with attributes will be preserved. |
| 299 | * @param string $text HTML content |
| 300 | * @return string <p>-less content |
| 301 | */ |
| 302 | public function unp( $text ) { |
| 303 | return preg_replace( "#<p>(.*?)</p>(\n|$)#ums", '$1$2', $text ); |
| 304 | } |
| 305 | |
| 306 | /** |
| 307 | * A regex of all shortcodes currently registered by the current |
| 308 | * WordPress installation |
| 309 | * @uses get_shortcode_regex() |
| 310 | * @return string A regex for grabbing shortcodes. |
| 311 | */ |
| 312 | protected function get_shortcode_regex() { |
| 313 | $pattern = get_shortcode_regex(); |
| 314 | |
| 315 | // don't match markdown link anchors that could be mistaken for shortcodes. |
| 316 | $pattern .= '(?!\()'; |
| 317 | |
| 318 | return "/$pattern/s"; |
| 319 | } |
| 320 | |
| 321 | /** |
| 322 | * Since we escape unspaced #Headings, put things back later. |
| 323 | * @param string $text text with a leading escaped hash |
| 324 | * @return string text with leading hashes unescaped |
| 325 | */ |
| 326 | protected function restore_leading_hash( $text ) { |
| 327 | return preg_replace( "/^(<p>)?(#|\\\\#)/um", "$1#", $text ); |
| 328 | } |
| 329 | |
| 330 | /** |
| 331 | * Overload to support ```-fenced code blocks for pre-Markdown Extra 1.2.8 |
| 332 | * https://help.github.com/articles/github-flavored-markdown#fenced-code-blocks |
| 333 | */ |
| 334 | public function doFencedCodeBlocks( $text ) { |
| 335 | // If we're at least at 1.2.8, native fenced code blocks are in. |
| 336 | // Below is just copied from it in case we somehow got loaded on |
| 337 | // top of someone else's Markdown Extra |
| 338 | if ( version_compare( MARKDOWNEXTRA_VERSION, '1.2.8', '>=' ) ) |
| 339 | return parent::doFencedCodeBlocks( $text ); |
| 340 | |
| 341 | # |
| 342 | # Adding the fenced code block syntax to regular Markdown: |
| 343 | # |
| 344 | # ~~~ |
| 345 | # Code block |
| 346 | # ~~~ |
| 347 | # |
| 348 | $less_than_tab = $this->tab_width; |
| 349 | |
| 350 | $text = preg_replace_callback('{ |
| 351 | (?:\n|\A) |
| 352 | # 1: Opening marker |
| 353 | ( |
| 354 | (?:~{3,}|`{3,}) # 3 or more tildes/backticks. |
| 355 | ) |
| 356 | [ ]* |
| 357 | (?: |
| 358 | \.?([-_:a-zA-Z0-9]+) # 2: standalone class name |
| 359 | | |
| 360 | '.$this->id_class_attr_catch_re.' # 3: Extra attributes |
| 361 | )? |
| 362 | [ ]* \n # Whitespace and newline following marker. |
| 363 | |
| 364 | # 4: Content |
| 365 | ( |
| 366 | (?> |
| 367 | (?!\1 [ ]* \n) # Not a closing marker. |
| 368 | .*\n+ |
| 369 | )+ |
| 370 | ) |
| 371 | |
| 372 | # Closing marker. |
| 373 | \1 [ ]* (?= \n ) |
| 374 | }xm', |
| 375 | array($this, '_doFencedCodeBlocks_callback'), $text); |
| 376 | |
| 377 | return $text; |
| 378 | } |
| 379 | |
| 380 | /** |
| 381 | * Callback for pre-processing start of line hashes to slyly escape headings that don't |
| 382 | * have a leading space |
| 383 | * @param array $m preg_match matches |
| 384 | * @return string possibly escaped start of line hash |
| 385 | */ |
| 386 | public function _doEscapeForHashWithoutSpacing( $m ) { |
| 387 | if ( ! isset( $m[1] ) ) |
| 388 | $m[0] = '\\' . $m[0]; |
| 389 | return $m[0]; |
| 390 | } |
| 391 | |
| 392 | /** |
| 393 | * Overload to support Viper's [code] shortcode. Because awesome. |
| 394 | */ |
| 395 | public function _doFencedCodeBlocks_callback( $matches ) { |
| 396 | // in case we have some escaped leading hashes right at the start of the block |
| 397 | $matches[4] = $this->restore_leading_hash( $matches[4] ); |
| 398 | // just MarkdownExtra_Parser if we're not going ultra-deluxe |
| 399 | if ( ! $this->use_code_shortcode ) { |
| 400 | return parent::_doFencedCodeBlocks_callback( $matches ); |
| 401 | } |
| 402 | |
| 403 | // default to a "text" class if one wasn't passed. Helps with encoding issues later. |
| 404 | if ( empty( $matches[2] ) ) { |
| 405 | $matches[2] = 'text'; |
| 406 | } |
| 407 | |
| 408 | $classname =& $matches[2]; |
| 409 | $codeblock = preg_replace_callback('/^\n+/', array( $this, '_doFencedCodeBlocks_newlines' ), $matches[4] ); |
| 410 | |
| 411 | if ( $classname[0] == '.' ) |
| 412 | $classname = substr( $classname, 1 ); |
| 413 | |
| 414 | $codeblock = esc_html( $codeblock ); |
| 415 | $codeblock = sprintf( $this->shortcode_start, $classname ) . "\n{$codeblock}" . $this->shortcode_end; |
| 416 | return "\n\n" . $this->hashBlock( $codeblock ). "\n\n"; |
| 417 | } |
| 418 | |
| 419 | } |
| 420 |