| 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 |
* @author Brandon Kraft <kraft@automattic.com> -- Strikedown support converted from GPL code at https://github.com/annaesvensson/yellow-markdown/blob/main/markdown.php |
| 8 |
* @link https://github.com/evansolomon/wp-github-flavored-markdown-comments |
| 9 |
* |
| 10 |
* Add a few extras from GitHub's Markdown implementation. Must be used in a WordPress environment. |
| 11 |
*/ |
| 12 |
|
| 13 |
if ( ! defined( 'ABSPATH' ) ) { |
| 14 |
exit( 0 ); |
| 15 |
} |
| 16 |
|
| 17 |
class WPCom_GHF_Markdown_Parser extends MarkdownExtra_Parser { |
| 18 |
|
| 19 |
/** |
| 20 |
* Hooray somewhat arbitrary numbers that are fearful of 1.0.x. |
| 21 |
*/ |
| 22 |
const WPCOM_GHF_MARDOWN_VERSION = '0.9.1'; |
| 23 |
|
| 24 |
/** |
| 25 |
* Use a [code] shortcode when encountering a fenced code block |
| 26 |
* @var boolean |
| 27 |
*/ |
| 28 |
public $use_code_shortcode = true; |
| 29 |
|
| 30 |
/** |
| 31 |
* Preserve shortcodes, untouched by Markdown. |
| 32 |
* This requires use within a WordPress installation. |
| 33 |
* @var boolean |
| 34 |
*/ |
| 35 |
public $preserve_shortcodes = true; |
| 36 |
|
| 37 |
/** |
| 38 |
* Preserve the legacy $latex your-latex-code-here$ style |
| 39 |
* LaTeX markup |
| 40 |
*/ |
| 41 |
public $preserve_latex = true; |
| 42 |
|
| 43 |
/** |
| 44 |
* Preserve single-line <code> blocks. |
| 45 |
* @var boolean |
| 46 |
*/ |
| 47 |
public $preserve_inline_code_blocks = true; |
| 48 |
|
| 49 |
/** |
| 50 |
* Strip paragraphs from the output. This is the right default for WordPress, |
| 51 |
* which generally wants to create its own paragraphs with `wpautop` |
| 52 |
* @var boolean |
| 53 |
*/ |
| 54 |
public $strip_paras = true; |
| 55 |
|
| 56 |
// Will run through sprintf - you can supply your own syntax if you want |
| 57 |
public $shortcode_start = '[code lang=%s]'; |
| 58 |
public $shortcode_end = '[/code]'; |
| 59 |
|
| 60 |
// Stores shortcodes we remove and then replace |
| 61 |
protected $preserve_text_hash = array(); |
| 62 |
|
| 63 |
/** |
| 64 |
* Set environment defaults based on presence of key functions/classes. |
| 65 |
*/ |
| 66 |
public function __construct() { |
| 67 |
$this->use_code_shortcode = class_exists( 'SyntaxHighlighter' ); |
| 68 |
/** |
| 69 |
* Allow processing shortcode contents. |
| 70 |
* |
| 71 |
* @module markdown |
| 72 |
* |
| 73 |
* @since 4.4.0 |
| 74 |
* |
| 75 |
* @param boolean $preserve_shortcodes Defaults to $this->preserve_shortcodes. |
| 76 |
*/ |
| 77 |
$this->preserve_shortcodes = apply_filters( 'jetpack_markdown_preserve_shortcodes', $this->preserve_shortcodes ) && function_exists( 'get_shortcode_regex' ); |
| 78 |
$this->preserve_latex = function_exists( 'latex_markup' ); |
| 79 |
$this->strip_paras = function_exists( 'wpautop' ); |
| 80 |
|
| 81 |
$this->span_gamut['do_strikethrough'] = 55; |
| 82 |
|
| 83 |
parent::__construct(); |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Overload to specify heading styles only if the hash has space(s) after it. This is actually in keeping with |
| 88 |
* the documentation and eases the semantic overload of the hash character. |
| 89 |
* #Will Not Produce a Heading 1 |
| 90 |
* # This Will Produce a Heading 1 |
| 91 |
* |
| 92 |
* @param string $text Markdown text |
| 93 |
* @return string HTML-transformed text |
| 94 |
*/ |
| 95 |
public function transform( $text ) { |
| 96 |
// Preserve anything inside a single-line <code> element |
| 97 |
if ( $this->preserve_inline_code_blocks ) { |
| 98 |
$text = $this->single_line_code_preserve( $text ); |
| 99 |
} |
| 100 |
// Remove all shortcodes so their interiors are left intact |
| 101 |
if ( $this->preserve_shortcodes ) { |
| 102 |
$text = $this->shortcode_preserve( $text ); |
| 103 |
} |
| 104 |
// Remove legacy LaTeX so it's left intact |
| 105 |
if ( $this->preserve_latex ) { |
| 106 |
$text = $this->latex_preserve( $text ); |
| 107 |
} |
| 108 |
|
| 109 |
// Do not process characters inside URLs. |
| 110 |
$text = $this->urls_preserve( $text ); |
| 111 |
|
| 112 |
// escape line-beginning # chars that do not have a space after them. |
| 113 |
$text = preg_replace_callback( '|^#{1,6}( )?|um', array( $this, '_doEscapeForHashWithoutSpacing' ), $text ); |
| 114 |
|
| 115 |
/** |
| 116 |
* Allow third-party plugins to define custom patterns that won't be processed by Markdown. |
| 117 |
* |
| 118 |
* @module markdown |
| 119 |
* |
| 120 |
* @since 3.9.2 |
| 121 |
* |
| 122 |
* @param array $custom_patterns Array of custom patterns to be ignored by Markdown. |
| 123 |
*/ |
| 124 |
$custom_patterns = apply_filters( 'jetpack_markdown_preserve_pattern', array() ); |
| 125 |
if ( is_array( $custom_patterns ) && ! empty( $custom_patterns ) ) { |
| 126 |
foreach ( $custom_patterns as $pattern ) { |
| 127 |
$text = preg_replace_callback( $pattern, array( $this, '_doRemoveText'), $text ); |
| 128 |
} |
| 129 |
} |
| 130 |
|
| 131 |
// run through core Markdown |
| 132 |
$text = parent::transform( $text ); |
| 133 |
|
| 134 |
// Occasionally Markdown Extra chokes on a para structure, producing odd paragraphs. |
| 135 |
$text = str_replace( "<p><</p>\n\n<p>p>", '<p>', $text ); |
| 136 |
|
| 137 |
// put start-of-line # chars back in place |
| 138 |
$text = $this->restore_leading_hash( $text ); |
| 139 |
|
| 140 |
// Strip paras if set |
| 141 |
if ( $this->strip_paras ) { |
| 142 |
$text = $this->unp( $text ); |
| 143 |
} |
| 144 |
|
| 145 |
// Restore preserved things like shortcodes/LaTeX |
| 146 |
$text = $this->do_restore( $text ); |
| 147 |
|
| 148 |
return $text; |
| 149 |
} |
| 150 |
|
| 151 |
/** |
| 152 |
* Prevents blocks like <code>__this__</code> from turning into <code><strong>this</strong></code> |
| 153 |
* @param string $text Text that may need preserving |
| 154 |
* @return string Text that was preserved if needed |
| 155 |
*/ |
| 156 |
public function single_line_code_preserve( $text ) { |
| 157 |
return preg_replace_callback( '|<code\b[^>]*>(.*?)</code>|', array( $this, 'do_single_line_code_preserve' ), $text ); |
| 158 |
} |
| 159 |
|
| 160 |
/** |
| 161 |
* Regex callback for inline code presevation |
| 162 |
* @param array $matches Regex matches |
| 163 |
* @return string Hashed content for later restoration |
| 164 |
*/ |
| 165 |
public function do_single_line_code_preserve( $matches ) { |
| 166 |
return '<code>' . $this->hash_block( $matches[1] ) . '</code>'; |
| 167 |
} |
| 168 |
|
| 169 |
/** |
| 170 |
* Preserve code block contents by HTML encoding them. Useful before getting to KSES stripping. |
| 171 |
* @param string $text Markdown/HTML content |
| 172 |
* @return string Markdown/HTML content with escaped code blocks |
| 173 |
*/ |
| 174 |
public function codeblock_preserve( $text ) { |
| 175 |
return preg_replace_callback( "/^([`~]{3})([^`\n]+)?\n([^`~]+)(\\1)/m", array( $this, 'do_codeblock_preserve' ), $text ); |
| 176 |
} |
| 177 |
|
| 178 |
/** |
| 179 |
* Regex callback for code block preservation. |
| 180 |
* @param array $matches Regex matches |
| 181 |
* @return string Codeblock with escaped interior |
| 182 |
*/ |
| 183 |
public function do_codeblock_preserve( $matches ) { |
| 184 |
$block = stripslashes( $matches[3] ); |
| 185 |
$block = esc_html( $block ); |
| 186 |
$block = str_replace( '\\', '\\\\', $block ); |
| 187 |
$open = $matches[1] . $matches[2] . "\n"; |
| 188 |
return $open . $block . $matches[4]; |
| 189 |
} |
| 190 |
|
| 191 |
/** |
| 192 |
* Restore previously preserved (i.e. escaped) code block contents. |
| 193 |
* @param string $text Markdown/HTML content with escaped code blocks |
| 194 |
* @return string Markdown/HTML content |
| 195 |
*/ |
| 196 |
public function codeblock_restore( $text ) { |
| 197 |
return preg_replace_callback( "/^([`~]{3})([^`\n]+)?\n([^`~]+)(\\1)/m", array( $this, 'do_codeblock_restore' ), $text ); |
| 198 |
} |
| 199 |
|
| 200 |
/** |
| 201 |
* Regex callback for code block restoration (unescaping). |
| 202 |
* @param array $matches Regex matches |
| 203 |
* @return string Codeblock with unescaped interior |
| 204 |
*/ |
| 205 |
public function do_codeblock_restore( $matches ) { |
| 206 |
$block = html_entity_decode( $matches[3], ENT_QUOTES ); |
| 207 |
$open = $matches[1] . $matches[2] . "\n"; |
| 208 |
return $open . $block . $matches[4]; |
| 209 |
} |
| 210 |
|
| 211 |
/** |
| 212 |
* Called to preserve legacy LaTeX like $latex some-latex-text $ |
| 213 |
* @param string $text Text in which to preserve LaTeX |
| 214 |
* @return string Text with LaTeX replaced by a hash that will be restored later |
| 215 |
*/ |
| 216 |
protected function latex_preserve( $text ) { |
| 217 |
// regex from latex_remove() |
| 218 |
$regex = '% |
| 219 |
\$latex(?:=\s*|\s+) |
| 220 |
((?: |
| 221 |
[^$]+ # Not a dollar |
| 222 |
| |
| 223 |
(?<=(?<!\\\\)\\\\)\$ # Dollar preceded by exactly one slash |
| 224 |
)+) |
| 225 |
(?<!\\\\)\$ # Dollar preceded by zero slashes |
| 226 |
%ix'; |
| 227 |
$text = preg_replace_callback( $regex, array( $this, '_doRemoveText'), $text ); |
| 228 |
return $text; |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Called to preserve WP shortcodes from being formatted by Markdown in any way. |
| 233 |
* @param string $text Text in which to preserve shortcodes |
| 234 |
* @return string Text with shortcodes replaced by a hash that will be restored later |
| 235 |
*/ |
| 236 |
protected function shortcode_preserve( $text ) { |
| 237 |
$text = preg_replace_callback( $this->get_shortcode_regex(), array( $this, '_doRemoveText' ), $text ); |
| 238 |
return $text; |
| 239 |
} |
| 240 |
|
| 241 |
/** |
| 242 |
* Avoid characters inside URLs from being formatted by Markdown in any way. |
| 243 |
* |
| 244 |
* @param string $text Text in which to preserve URLs. |
| 245 |
* |
| 246 |
* @return string Text with URLs replaced by a hash that will be restored later. |
| 247 |
*/ |
| 248 |
protected function urls_preserve( $text ) { |
| 249 |
$text = preg_replace_callback( |
| 250 |
'#(?<!<)(?:https?|ftp)://([^\s<>"\'\[\]()]+|\[(?1)*+\]|\((?1)*+\))+(?<![_*.?])#i', |
| 251 |
array( $this, '_doRemoveText' ), |
| 252 |
$text |
| 253 |
); |
| 254 |
return $text; |
| 255 |
} |
| 256 |
|
| 257 |
/** |
| 258 |
* Restores any text preserved by $this->hash_block() |
| 259 |
* @param string $text Text that may have hashed preservation placeholders |
| 260 |
* @return string Text with hashed preseravtion placeholders replaced by original text |
| 261 |
*/ |
| 262 |
protected function do_restore( $text ) { |
| 263 |
// Reverse hashes to ensure nested blocks are restored. |
| 264 |
$hashes = array_reverse( $this->preserve_text_hash, true ); |
| 265 |
foreach( $hashes as $hash => $value ) { |
| 266 |
$placeholder = $this->hash_maker( $hash ); |
| 267 |
$text = str_replace( $placeholder, $value, $text ); |
| 268 |
} |
| 269 |
// reset the hash |
| 270 |
$this->preserve_text_hash = array(); |
| 271 |
return $text; |
| 272 |
} |
| 273 |
|
| 274 |
/** |
| 275 |
* Regex callback for text preservation |
| 276 |
* @param array $m Regex $matches array |
| 277 |
* @return string A placeholder that will later be replaced by the original text |
| 278 |
*/ |
| 279 |
protected function _doRemoveText( $m ) { |
| 280 |
return $this->hash_block( $m[0] ); |
| 281 |
} |
| 282 |
|
| 283 |
/** |
| 284 |
* Call this to store a text block for later restoration. |
| 285 |
* @param string $text Text to preserve for later |
| 286 |
* @return string Placeholder that will be swapped out later for the original text |
| 287 |
*/ |
| 288 |
protected function hash_block( $text ) { |
| 289 |
$hash = md5( $text ); |
| 290 |
$this->preserve_text_hash[ $hash ] = $text; |
| 291 |
$placeholder = $this->hash_maker( $hash ); |
| 292 |
return $placeholder; |
| 293 |
} |
| 294 |
|
| 295 |
/** |
| 296 |
* Less glamorous than the Keymaker |
| 297 |
* @param string $hash An md5 hash |
| 298 |
* @return string A placeholder hash |
| 299 |
*/ |
| 300 |
protected function hash_maker( $hash ) { |
| 301 |
return 'MARKDOWN_HASH' . $hash . 'MARKDOWN_HASH'; |
| 302 |
} |
| 303 |
|
| 304 |
/** |
| 305 |
* Remove bare <p> elements. <p>s with attributes will be preserved. |
| 306 |
* @param string $text HTML content |
| 307 |
* @return string <p>-less content |
| 308 |
*/ |
| 309 |
public function unp( $text ) { |
| 310 |
return preg_replace( "#<p>(.*?)</p>(\n|$)#ums", '$1$2', $text ); |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* A regex of all shortcodes currently registered by the current |
| 315 |
* WordPress installation |
| 316 |
* @uses get_shortcode_regex() |
| 317 |
* @return string A regex for grabbing shortcodes. |
| 318 |
*/ |
| 319 |
protected function get_shortcode_regex() { |
| 320 |
$pattern = get_shortcode_regex(); |
| 321 |
|
| 322 |
// don't match markdown link anchors that could be mistaken for shortcodes. |
| 323 |
$pattern .= '(?!\()'; |
| 324 |
|
| 325 |
return "/$pattern/s"; |
| 326 |
} |
| 327 |
|
| 328 |
/** |
| 329 |
* Since we escape unspaced #Headings, put things back later. |
| 330 |
* @param string $text text with a leading escaped hash |
| 331 |
* @return string text with leading hashes unescaped |
| 332 |
*/ |
| 333 |
protected function restore_leading_hash( $text ) { |
| 334 |
return preg_replace( "/^(<p>)?(#|\\\\#)/um", "$1#", $text ); |
| 335 |
} |
| 336 |
|
| 337 |
/** |
| 338 |
* Overload to support ```-fenced code blocks for pre-Markdown Extra 1.2.8 |
| 339 |
* https://help.github.com/articles/github-flavored-markdown#fenced-code-blocks |
| 340 |
*/ |
| 341 |
public function doFencedCodeBlocks( $text ) { |
| 342 |
// If we're at least at 1.2.8, native fenced code blocks are in. |
| 343 |
// Below is just copied from it in case we somehow got loaded on |
| 344 |
// top of someone else's Markdown Extra |
| 345 |
if ( version_compare( MARKDOWNEXTRA_VERSION, '1.2.8', '>=' ) ) |
| 346 |
return parent::doFencedCodeBlocks( $text ); |
| 347 |
|
| 348 |
# |
| 349 |
# Adding the fenced code block syntax to regular Markdown: |
| 350 |
# |
| 351 |
# ~~~ |
| 352 |
# Code block |
| 353 |
# ~~~ |
| 354 |
# |
| 355 |
$less_than_tab = $this->tab_width; |
| 356 |
|
| 357 |
$text = preg_replace_callback('{ |
| 358 |
(?:\n|\A) |
| 359 |
# 1: Opening marker |
| 360 |
( |
| 361 |
(?:~{3,}|`{3,}) # 3 or more tildes/backticks. |
| 362 |
) |
| 363 |
[ ]* |
| 364 |
(?: |
| 365 |
\.?([-_:a-zA-Z0-9]+) # 2: standalone class name |
| 366 |
| |
| 367 |
'.$this->id_class_attr_catch_re.' # 3: Extra attributes |
| 368 |
)? |
| 369 |
[ ]* \n # Whitespace and newline following marker. |
| 370 |
|
| 371 |
# 4: Content |
| 372 |
( |
| 373 |
(?> |
| 374 |
(?!\1 [ ]* \n) # Not a closing marker. |
| 375 |
.*\n+ |
| 376 |
)+ |
| 377 |
) |
| 378 |
|
| 379 |
# Closing marker. |
| 380 |
\1 [ ]* (?= \n ) |
| 381 |
}xm', |
| 382 |
array($this, '_doFencedCodeBlocks_callback'), $text); |
| 383 |
|
| 384 |
return $text; |
| 385 |
} |
| 386 |
|
| 387 |
/** |
| 388 |
* Callback for pre-processing start of line hashes to slyly escape headings that don't |
| 389 |
* have a leading space |
| 390 |
* @param array $m preg_match matches |
| 391 |
* @return string possibly escaped start of line hash |
| 392 |
*/ |
| 393 |
public function _doEscapeForHashWithoutSpacing( $m ) { |
| 394 |
if ( ! isset( $m[1] ) ) |
| 395 |
$m[0] = '\\' . $m[0]; |
| 396 |
return $m[0]; |
| 397 |
} |
| 398 |
|
| 399 |
/** |
| 400 |
* Overload to support Viper's [code] shortcode. Because awesome. |
| 401 |
*/ |
| 402 |
public function _doFencedCodeBlocks_callback( $matches ) { |
| 403 |
// in case we have some escaped leading hashes right at the start of the block |
| 404 |
$matches[4] = $this->restore_leading_hash( $matches[4] ); |
| 405 |
// just MarkdownExtra_Parser if we're not going ultra-deluxe |
| 406 |
if ( ! $this->use_code_shortcode ) { |
| 407 |
return parent::_doFencedCodeBlocks_callback( $matches ); |
| 408 |
} |
| 409 |
|
| 410 |
// default to a "text" class if one wasn't passed. Helps with encoding issues later. |
| 411 |
if ( empty( $matches[2] ) ) { |
| 412 |
$matches[2] = 'text'; |
| 413 |
} |
| 414 |
|
| 415 |
$classname =& $matches[2]; |
| 416 |
$codeblock = preg_replace_callback('/^\n+/', array( $this, '_doFencedCodeBlocks_newlines' ), $matches[4] ); |
| 417 |
|
| 418 |
if ( $classname[0] == '.' ) |
| 419 |
$classname = substr( $classname, 1 ); |
| 420 |
|
| 421 |
$codeblock = esc_html( $codeblock ); |
| 422 |
$codeblock = sprintf( $this->shortcode_start, $classname ) . "\n{$codeblock}" . $this->shortcode_end; |
| 423 |
return "\n\n" . $this->hashBlock( $codeblock ). "\n\n"; |
| 424 |
} |
| 425 |
|
| 426 |
/** |
| 427 |
* Add strikethrough support. |
| 428 |
* |
| 429 |
* GPL code modified from https://github.com/annaesvensson/yellow-markdown/blob/main/markdown.php |
| 430 |
*/ |
| 431 |
public function do_strikethrough($text) { |
| 432 |
$parts = preg_split("/(?<![~])(~~)(?![~])/", $text, -1, PREG_SPLIT_DELIM_CAPTURE); |
| 433 |
if (count($parts)>3) { |
| 434 |
$text = ""; |
| 435 |
$open = false; |
| 436 |
foreach ($parts as $part) { |
| 437 |
if ($part=="~~") { |
| 438 |
$text .= $open ? "</del>" : "<del>"; |
| 439 |
$open = !$open; |
| 440 |
} else { |
| 441 |
$text .= $part; |
| 442 |
} |
| 443 |
} |
| 444 |
if ($open) $text .= "</del>"; |
| 445 |
} |
| 446 |
return $text; |
| 447 |
} |
| 448 |
|
| 449 |
} |
| 450 |
|
