| 1 |
<?php |
| 2 |
|
| 3 |
namespace WPDeveloper\BetterDocs\Core; |
| 4 |
|
| 5 |
if ( ! defined( 'ABSPATH' ) ) { |
| 6 |
exit; // Exit if accessed directly |
| 7 |
} |
| 8 |
|
| 9 |
use WPDeveloper\BetterDocs\Utils\Database; |
| 10 |
|
| 11 |
/** |
| 12 |
* Render a doc as Markdown. |
| 13 |
* |
| 14 |
* AI assistants consume plain Markdown far more reliably than themed HTML, so |
| 15 |
* every published doc is convertible to a clean Markdown document. Two output |
| 16 |
* profiles are supported: |
| 17 |
* |
| 18 |
* - `endpoint` — YAML front matter + `# Title` + body. What `<doc-url>.md` |
| 19 |
* serves, and what AI crawlers index. |
| 20 |
* - `copy` — `# Title`, blockquoted description, `Source:` line, body. |
| 21 |
* What the "Copy page" button puts on the clipboard; no YAML, |
| 22 |
* because a human is about to paste this into a chat box. |
| 23 |
* |
| 24 |
* The body is produced by running the real `the_content` filter (so blocks, |
| 25 |
* shortcodes and embeds resolve exactly as they do on the page) and then walking |
| 26 |
* the resulting DOM. A node-removal pass first strips chrome that is meaningless |
| 27 |
* outside a browser — code-snippet toolbars, heading anchor links, glossary |
| 28 |
* tooltips, scripts. |
| 29 |
* |
| 30 |
* @since 4.8.0 |
| 31 |
*/ |
| 32 |
class MarkdownRenderer { |
| 33 |
/** |
| 34 |
* Cached body Markdown (no front matter — that is assembled per context at |
| 35 |
* request time, so one blob serves both profiles). |
| 36 |
*/ |
| 37 |
const META_BODY = '_betterdocs_markdown'; |
| 38 |
|
| 39 |
/** Signature the cached body was generated from. */ |
| 40 |
const META_SIG = '_betterdocs_markdown_sig'; |
| 41 |
|
| 42 |
/** Cache namespace for Database::get_cache_version(). */ |
| 43 |
const CACHE_NS = 'markdown'; |
| 44 |
|
| 45 |
/** |
| 46 |
* @var Settings |
| 47 |
*/ |
| 48 |
protected $settings; |
| 49 |
|
| 50 |
/** |
| 51 |
* @var Database |
| 52 |
*/ |
| 53 |
protected $database; |
| 54 |
|
| 55 |
/** |
| 56 |
* Re-entrancy guard. `the_content` can be filtered from inside a page that is |
| 57 |
* already rendering `the_content` (a shortcode placed inside a doc), which |
| 58 |
* would recurse forever. |
| 59 |
* |
| 60 |
* @var bool |
| 61 |
*/ |
| 62 |
protected $rendering = false; |
| 63 |
|
| 64 |
public function __construct( Settings $settings, Database $database ) { |
| 65 |
$this->settings = $settings; |
| 66 |
$this->database = $database; |
| 67 |
|
| 68 |
// Bust one doc's cache when it is saved. `wp_after_insert_post` — NOT |
| 69 |
// `save_post_docs` — because terms are not written yet at `save_post`, and |
| 70 |
// the front matter carries the category and tags. |
| 71 |
add_action( 'wp_after_insert_post', [ $this, 'flush_post' ], 10, 2 ); |
| 72 |
|
| 73 |
// Term renames and settings changes affect every doc's front matter, so bump |
| 74 |
// the namespace version instead of walking the post table. |
| 75 |
add_action( 'edited_doc_category', [ $this, 'flush_all' ] ); |
| 76 |
add_action( 'edited_doc_tag', [ $this, 'flush_all' ] ); |
| 77 |
add_action( 'delete_term', [ $this, 'flush_all' ] ); |
| 78 |
add_action( 'update_option_betterdocs_settings', [ $this, 'flush_all' ] ); |
| 79 |
} |
| 80 |
|
| 81 |
/** |
| 82 |
* Whether Markdown may be produced for this doc at all. |
| 83 |
* |
| 84 |
* The single gate both the `.md` endpoint and the AI Actions UI consult, so a |
| 85 |
* doc that cannot be served also never renders a button pointing at it. |
| 86 |
* |
| 87 |
* @param \WP_Post|int|null $post |
| 88 |
* @return bool |
| 89 |
*/ |
| 90 |
public function can_read( $post ) { |
| 91 |
$post = get_post( $post ); |
| 92 |
|
| 93 |
if ( ! $post instanceof \WP_Post || 'docs' !== $post->post_type ) { |
| 94 |
return false; |
| 95 |
} |
| 96 |
|
| 97 |
// Drafts, pending and private docs are readable only by someone who could |
| 98 |
// read them in the admin. |
| 99 |
if ( 'publish' !== $post->post_status && ! current_user_can( 'read_post', $post->ID ) ) { |
| 100 |
return false; |
| 101 |
} |
| 102 |
|
| 103 |
// Deliberately no `?password=` query-arg escape hatch (unlike REST\Docs): |
| 104 |
// the cookie is already present on a front-end request, and a password in a |
| 105 |
// GET would leak into referrers, server logs and the LLM prompt. |
| 106 |
if ( post_password_required( $post ) ) { |
| 107 |
return false; |
| 108 |
} |
| 109 |
|
| 110 |
/** |
| 111 |
* Gate Markdown output for a doc. |
| 112 |
* |
| 113 |
* BetterDocs Pro hooks its Content Restriction / Access Control check here so |
| 114 |
* a restricted doc cannot be exfiltrated through `.md`. |
| 115 |
* |
| 116 |
* @since 4.8.0 |
| 117 |
* |
| 118 |
* @param bool $can_read |
| 119 |
* @param \WP_Post $post |
| 120 |
*/ |
| 121 |
return (bool) apply_filters( 'betterdocs_markdown_can_read', true, $post ); |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Full Markdown document for a doc. |
| 126 |
* |
| 127 |
* @param \WP_Post|int|null $post |
| 128 |
* @param string $context `endpoint` | `copy` |
| 129 |
* @return string Empty string when the caller is not entitled to the content. |
| 130 |
*/ |
| 131 |
public function render( $post, $context = 'endpoint' ) { |
| 132 |
$post = get_post( $post ); |
| 133 |
|
| 134 |
if ( ! $this->can_read( $post ) ) { |
| 135 |
return ''; |
| 136 |
} |
| 137 |
|
| 138 |
$body = $this->body( $post ); |
| 139 |
|
| 140 |
return $this->front_matter( $post, $context, $body ) . $body . "\n"; |
| 141 |
} |
| 142 |
|
| 143 |
/** |
| 144 |
* Body Markdown only, cached. |
| 145 |
* |
| 146 |
* @param \WP_Post $post |
| 147 |
* @return string |
| 148 |
*/ |
| 149 |
public function body( $post ) { |
| 150 |
$signature = $this->signature( $post ); |
| 151 |
$cacheable = $this->cacheable( $post ); |
| 152 |
|
| 153 |
if ( $cacheable ) { |
| 154 |
$cached = get_post_meta( $post->ID, self::META_BODY, true ); |
| 155 |
if ( is_string( $cached ) && '' !== $cached |
| 156 |
&& $signature === get_post_meta( $post->ID, self::META_SIG, true ) ) { |
| 157 |
return $cached; |
| 158 |
} |
| 159 |
} |
| 160 |
|
| 161 |
$body = $this->html_to_markdown( $this->content_html( $post ) ); |
| 162 |
|
| 163 |
if ( $cacheable && '' !== $body ) { |
| 164 |
update_post_meta( $post->ID, self::META_BODY, $body ); |
| 165 |
update_post_meta( $post->ID, self::META_SIG, $signature ); |
| 166 |
} |
| 167 |
|
| 168 |
return $body; |
| 169 |
} |
| 170 |
|
| 171 |
/** |
| 172 |
* Only published, unprotected, non-preview docs are worth persisting. Anything |
| 173 |
* else is either transient or user-specific. |
| 174 |
* |
| 175 |
* The cache is shared by every reader, so it may only hold what an anonymous |
| 176 |
* visitor would get. A logged-in reader's render can differ — membership |
| 177 |
* plugins, "logged-in only" blocks and per-user shortcodes run inside |
| 178 |
* the_content — so it is neither stored (it would be served to guests) nor |
| 179 |
* read (a guest's render would be served to the member). |
| 180 |
* |
| 181 |
* @param \WP_Post $post |
| 182 |
* @return bool |
| 183 |
*/ |
| 184 |
protected function cacheable( $post ) { |
| 185 |
$cacheable = 'publish' === $post->post_status |
| 186 |
&& ! post_password_required( $post ) |
| 187 |
&& ! is_preview() |
| 188 |
&& ! is_user_logged_in(); |
| 189 |
|
| 190 |
/** |
| 191 |
* Whether a doc's rendered Markdown may be stored and reused for other |
| 192 |
* readers. Return false for content that varies by visitor. |
| 193 |
* |
| 194 |
* @since 4.8.0 |
| 195 |
* |
| 196 |
* @param bool $cacheable |
| 197 |
* @param \WP_Post $post |
| 198 |
*/ |
| 199 |
return (bool) apply_filters( 'betterdocs_markdown_cacheable', $cacheable, $post ); |
| 200 |
} |
| 201 |
|
| 202 |
/** |
| 203 |
* @param \WP_Post $post |
| 204 |
* @return string |
| 205 |
*/ |
| 206 |
protected function signature( $post ) { |
| 207 |
return md5( |
| 208 |
implode( |
| 209 |
'|', |
| 210 |
[ |
| 211 |
$post->post_content, |
| 212 |
$post->post_title, |
| 213 |
$post->post_modified_gmt, |
| 214 |
defined( 'BETTERDOCS_VERSION' ) ? BETTERDOCS_VERSION : '', |
| 215 |
get_locale(), |
| 216 |
(string) $this->database->get_cache_version( self::CACHE_NS ) |
| 217 |
] |
| 218 |
) |
| 219 |
); |
| 220 |
} |
| 221 |
|
| 222 |
/** |
| 223 |
* Drop one doc's cached Markdown. |
| 224 |
* |
| 225 |
* @param int $post_id |
| 226 |
* @param \WP_Post $post |
| 227 |
*/ |
| 228 |
public function flush_post( $post_id, $post = null ) { |
| 229 |
if ( $post instanceof \WP_Post && 'docs' !== $post->post_type ) { |
| 230 |
return; |
| 231 |
} |
| 232 |
delete_post_meta( $post_id, self::META_BODY ); |
| 233 |
delete_post_meta( $post_id, self::META_SIG ); |
| 234 |
} |
| 235 |
|
| 236 |
/** |
| 237 |
* Invalidate every doc's cached Markdown by bumping the namespace version. |
| 238 |
*/ |
| 239 |
public function flush_all() { |
| 240 |
$this->database->bump_cache_version( self::CACHE_NS ); |
| 241 |
} |
| 242 |
|
| 243 |
/** |
| 244 |
* Purge every stored Markdown blob. Called when the endpoint is switched off so |
| 245 |
* a disabled feature stops occupying postmeta. |
| 246 |
* |
| 247 |
* @global \wpdb $wpdb |
| 248 |
*/ |
| 249 |
public function purge() { |
| 250 |
global $wpdb; |
| 251 |
$wpdb->delete( $wpdb->postmeta, [ 'meta_key' => self::META_BODY ] ); // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key |
| 252 |
$wpdb->delete( $wpdb->postmeta, [ 'meta_key' => self::META_SIG ] ); // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_key |
| 253 |
} |
| 254 |
|
| 255 |
/* --------------------------------------------------------------------------- |
| 256 |
* Front matter |
| 257 |
* ------------------------------------------------------------------------- */ |
| 258 |
|
| 259 |
/** |
| 260 |
* @param \WP_Post $post |
| 261 |
* @param string $context |
| 262 |
* @param string $body Already-rendered body, used to derive a description. |
| 263 |
* @return string |
| 264 |
*/ |
| 265 |
protected function front_matter( $post, $context, $body ) { |
| 266 |
// get_the_title() runs the `the_title` filters, so an ampersand comes back as |
| 267 |
// `&`. Markdown is not HTML — entities have to be decoded or they show up |
| 268 |
// literally in the model's context. |
| 269 |
$title = $this->decode( get_the_title( $post ) ); |
| 270 |
$permalink = get_permalink( $post ); |
| 271 |
$description = $this->decode( $this->description( $post, $body ) ); |
| 272 |
|
| 273 |
$data = [ |
| 274 |
'title' => $title, |
| 275 |
'description' => $description, |
| 276 |
'url' => $permalink, |
| 277 |
'updated' => get_post_modified_time( 'Y-m-d', true, $post ), |
| 278 |
'category' => $this->category_path( $post ), |
| 279 |
'tags' => wp_get_post_terms( $post->ID, 'doc_tag', [ 'fields' => 'names' ] ) |
| 280 |
]; |
| 281 |
|
| 282 |
if ( is_wp_error( $data['tags'] ) ) { |
| 283 |
$data['tags'] = []; |
| 284 |
} |
| 285 |
$data['tags'] = array_map( [ $this, 'decode' ], $data['tags'] ); |
| 286 |
|
| 287 |
/** |
| 288 |
* Filter the Markdown front-matter data before it is serialised. |
| 289 |
* |
| 290 |
* @since 4.8.0 |
| 291 |
* |
| 292 |
* @param array $data |
| 293 |
* @param \WP_Post $post |
| 294 |
* @param string $context `endpoint` | `copy` |
| 295 |
*/ |
| 296 |
$data = (array) apply_filters( 'betterdocs_markdown_front_matter', $data, $post, $context ); |
| 297 |
|
| 298 |
if ( 'copy' === $context ) { |
| 299 |
// Human-facing: no YAML, because this is about to be pasted into a chat. |
| 300 |
$out = '# ' . $data['title'] . "\n\n"; |
| 301 |
if ( ! empty( $data['description'] ) ) { |
| 302 |
$out .= '> ' . $data['description'] . "\n\n"; |
| 303 |
} |
| 304 |
if ( ! empty( $data['url'] ) ) { |
| 305 |
$out .= 'Source: ' . $data['url'] . "\n\n"; |
| 306 |
} |
| 307 |
|
| 308 |
return $out; |
| 309 |
} |
| 310 |
|
| 311 |
$out = "---\n"; |
| 312 |
foreach ( $data as $key => $value ) { |
| 313 |
if ( is_array( $value ) ) { |
| 314 |
if ( empty( $value ) ) { |
| 315 |
continue; |
| 316 |
} |
| 317 |
$out .= $key . ":\n"; |
| 318 |
foreach ( $value as $item ) { |
| 319 |
$out .= ' - ' . $this->yaml_scalar( $item ) . "\n"; |
| 320 |
} |
| 321 |
continue; |
| 322 |
} |
| 323 |
if ( '' === (string) $value ) { |
| 324 |
continue; |
| 325 |
} |
| 326 |
$out .= $key . ': ' . $this->yaml_scalar( $value ) . "\n"; |
| 327 |
} |
| 328 |
$out .= "---\n\n"; |
| 329 |
$out .= '# ' . $data['title'] . "\n\n"; |
| 330 |
|
| 331 |
return $out; |
| 332 |
} |
| 333 |
|
| 334 |
/** |
| 335 |
* The doc's own excerpt when it has one, otherwise the opening of the body. |
| 336 |
* |
| 337 |
* `get_the_excerpt()` is avoided on purpose — it fires the `the_excerpt` |
| 338 |
* filters, which other plugins use to append read-more markup. |
| 339 |
* |
| 340 |
* @param \WP_Post $post |
| 341 |
* @param string $body |
| 342 |
* @return string |
| 343 |
*/ |
| 344 |
protected function description( $post, $body ) { |
| 345 |
if ( ! empty( $post->post_excerpt ) ) { |
| 346 |
return trim( wp_strip_all_tags( $post->post_excerpt ) ); |
| 347 |
} |
| 348 |
|
| 349 |
return trim( wp_trim_words( wp_strip_all_tags( $body ), 40, '…' ) ); |
| 350 |
} |
| 351 |
|
| 352 |
/** |
| 353 |
* Full hierarchical category path, e.g. "Getting Started / Installation". |
| 354 |
* |
| 355 |
* @param \WP_Post $post |
| 356 |
* @return string |
| 357 |
*/ |
| 358 |
protected function category_path( $post ) { |
| 359 |
$terms = get_the_terms( $post->ID, 'doc_category' ); |
| 360 |
if ( is_wp_error( $terms ) || empty( $terms ) ) { |
| 361 |
return ''; |
| 362 |
} |
| 363 |
|
| 364 |
$term = $terms[0]; |
| 365 |
$names = [ $term->name ]; |
| 366 |
|
| 367 |
foreach ( get_ancestors( $term->term_id, 'doc_category', 'taxonomy' ) as $ancestor_id ) { |
| 368 |
$ancestor = get_term( $ancestor_id, 'doc_category' ); |
| 369 |
if ( $ancestor && ! is_wp_error( $ancestor ) ) { |
| 370 |
array_unshift( $names, $ancestor->name ); |
| 371 |
} |
| 372 |
} |
| 373 |
|
| 374 |
return $this->decode( implode( ' / ', $names ) ); |
| 375 |
} |
| 376 |
|
| 377 |
/** |
| 378 |
* Turn HTML entities back into the characters they stand for. |
| 379 |
* |
| 380 |
* @param string $text |
| 381 |
* @return string |
| 382 |
*/ |
| 383 |
protected function decode( $text ) { |
| 384 |
return html_entity_decode( (string) $text, ENT_QUOTES | ENT_HTML5, 'UTF-8' ); |
| 385 |
} |
| 386 |
|
| 387 |
/** |
| 388 |
* Quote a YAML scalar only when it would otherwise break the mapping. |
| 389 |
* |
| 390 |
* @param mixed $value |
| 391 |
* @return string |
| 392 |
*/ |
| 393 |
protected function yaml_scalar( $value ) { |
| 394 |
$value = (string) $value; |
| 395 |
|
| 396 |
if ( '' === $value ) { |
| 397 |
return '""'; |
| 398 |
} |
| 399 |
// Line breaks would end the scalar and start a new key; inside double |
| 400 |
// quotes a backslash starts an escape, so it is escaped first. |
| 401 |
if ( preg_match( '/[:#\-\[\]\{\}&\*!\|>\'"%@`\r\n\t]/', $value ) || preg_match( '/^\s|\s$/', $value ) ) { |
| 402 |
return '"' . str_replace( [ '\\', '"', "\r", "\n", "\t" ], [ '\\\\', '\"', '\r', '\n', '\t' ], $value ) . '"'; |
| 403 |
} |
| 404 |
|
| 405 |
return $value; |
| 406 |
} |
| 407 |
|
| 408 |
/* --------------------------------------------------------------------------- |
| 409 |
* the_content pipeline |
| 410 |
* ------------------------------------------------------------------------- */ |
| 411 |
|
| 412 |
/** |
| 413 |
* Rendered HTML for a doc, with the filters that only make sense in a browser |
| 414 |
* temporarily unhooked. |
| 415 |
* |
| 416 |
* @param \WP_Post $post |
| 417 |
* @return string |
| 418 |
*/ |
| 419 |
protected function content_html( $post ) { |
| 420 |
if ( $this->rendering ) { |
| 421 |
return ''; |
| 422 |
} |
| 423 |
$this->rendering = true; |
| 424 |
|
| 425 |
global $wp_query; |
| 426 |
|
| 427 |
$prev_post = isset( $GLOBALS['post'] ) ? $GLOBALS['post'] : null; |
| 428 |
$prev_inloop = isset( $wp_query->in_the_loop ) ? $wp_query->in_the_loop : false; |
| 429 |
|
| 430 |
// Dynamic blocks and Pro's glossary wrapper read the global $post, so the |
| 431 |
// loop has to be standing before `the_content` runs. |
| 432 |
$GLOBALS['post'] = $post; |
| 433 |
if ( isset( $wp_query ) ) { |
| 434 |
$wp_query->in_the_loop = true; |
| 435 |
} |
| 436 |
setup_postdata( $post ); |
| 437 |
|
| 438 |
// Presentation-only filters. `do_blocks`, `wpautop`, `do_shortcode` and |
| 439 |
// `WP_Embed::autoembed` all stay — removing autoembed makes do_shortcode |
| 440 |
// delete the bare URL outright instead of leaving it as text. |
| 441 |
$removed = [ |
| 442 |
[ 'wptexturize', 10 ], // smart quotes corrupt CLI and code samples |
| 443 |
[ 'capital_P_dangit', 11 ], // rewrites "Wordpress" inside code samples |
| 444 |
[ 'wp_filter_content_tags', 12 ], // srcset/sizes/loading/decoding noise |
| 445 |
[ 'convert_smilies', 20 ] // ":)" becomes <img class="wp-smiley"> |
| 446 |
]; |
| 447 |
|
| 448 |
foreach ( $removed as $filter ) { |
| 449 |
remove_filter( 'the_content', $filter[0], $filter[1] ); |
| 450 |
} |
| 451 |
|
| 452 |
try { |
| 453 |
$html = apply_filters( 'the_content', $post->post_content ); |
| 454 |
} finally { |
| 455 |
foreach ( $removed as $filter ) { |
| 456 |
add_filter( 'the_content', $filter[0], $filter[1] ); |
| 457 |
} |
| 458 |
|
| 459 |
if ( isset( $wp_query ) ) { |
| 460 |
$wp_query->in_the_loop = $prev_inloop; |
| 461 |
} |
| 462 |
$GLOBALS['post'] = $prev_post; |
| 463 |
wp_reset_postdata(); |
| 464 |
|
| 465 |
$this->rendering = false; |
| 466 |
} |
| 467 |
|
| 468 |
return (string) $html; |
| 469 |
} |
| 470 |
|
| 471 |
/* --------------------------------------------------------------------------- |
| 472 |
* HTML -> Markdown |
| 473 |
* ------------------------------------------------------------------------- */ |
| 474 |
|
| 475 |
/** |
| 476 |
* Convert an HTML fragment to Markdown by walking the DOM. |
| 477 |
* |
| 478 |
* @param string $html |
| 479 |
* @return string |
| 480 |
*/ |
| 481 |
public function html_to_markdown( $html ) { |
| 482 |
if ( '' === trim( (string) $html ) ) { |
| 483 |
return ''; |
| 484 |
} |
| 485 |
|
| 486 |
// ext-dom is not a declared requirement (see composer.json), and |
| 487 |
// Core\SampleDocBuilder already guards for it. Degrade to plain text rather |
| 488 |
// than fataling on a host without php-xml. |
| 489 |
if ( ! class_exists( '\DOMDocument' ) ) { |
| 490 |
return trim( wp_strip_all_tags( $html ) ); |
| 491 |
} |
| 492 |
|
| 493 |
$dom = new \DOMDocument(); |
| 494 |
libxml_use_internal_errors( true ); |
| 495 |
$dom->loadHTML( |
| 496 |
'<?xml encoding="utf-8"?><body>' . $html . '</body>', |
| 497 |
LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD |
| 498 |
); |
| 499 |
libxml_clear_errors(); |
| 500 |
|
| 501 |
$this->prune( $dom ); |
| 502 |
|
| 503 |
$body = $dom->getElementsByTagName( 'body' )->item( 0 ); |
| 504 |
$md = $body ? $this->children_md( $body ) : wp_strip_all_tags( $html ); |
| 505 |
|
| 506 |
$md = preg_replace( "/[ \t]+\n/", "\n", $md ); // trailing spaces |
| 507 |
$md = preg_replace( "/\n{3,}/", "\n\n", $md ); // collapse blank runs |
| 508 |
|
| 509 |
return trim( $md ); |
| 510 |
} |
| 511 |
|
| 512 |
/** |
| 513 |
* Strip browser-only chrome before the walk, and unwrap glossary tooltips so |
| 514 |
* the defined term survives as plain text. |
| 515 |
* |
| 516 |
* @param \DOMDocument $dom |
| 517 |
*/ |
| 518 |
protected function prune( $dom ) { |
| 519 |
$xpath = new \DOMXPath( $dom ); |
| 520 |
|
| 521 |
/** |
| 522 |
* XPath queries for nodes removed before HTML is converted to Markdown. |
| 523 |
* |
| 524 |
* @since 4.8.0 |
| 525 |
* |
| 526 |
* @param string[] $queries |
| 527 |
*/ |
| 528 |
$queries = (array) apply_filters( |
| 529 |
'betterdocs_markdown_remove_selectors', |
| 530 |
[ |
| 531 |
'//script', |
| 532 |
'//style', |
| 533 |
'//noscript', |
| 534 |
'//svg', |
| 535 |
'//form', |
| 536 |
'//button', |
| 537 |
'//*[@aria-hidden="true"]', |
| 538 |
'//*[' . $this->has_class( 'screen-reader-text' ) . ']', |
| 539 |
'//*[' . $this->has_class( 'betterdocs-sr-only' ) . ']', |
| 540 |
'//*[' . $this->has_class( 'betterdocs-ai-actions' ) . ']', |
| 541 |
'//*[' . $this->has_class( 'betterdocs-code-snippet-header' ) . ']', |
| 542 |
'//*[' . $this->has_class( 'betterdocs-code-snippet-line-numbers' ) . ']', |
| 543 |
// Heading permalink anchors injected by FrontEnd\TemplateTags. |
| 544 |
'//a[' . $this->has_class( 'batterdocs-anchor' ) . ']' |
| 545 |
] |
| 546 |
); |
| 547 |
|
| 548 |
foreach ( $queries as $query ) { |
| 549 |
$nodes = $xpath->query( $query ); |
| 550 |
if ( ! $nodes ) { |
| 551 |
continue; |
| 552 |
} |
| 553 |
foreach ( iterator_to_array( $nodes ) as $node ) { |
| 554 |
if ( $node->parentNode ) { |
| 555 |
$node->parentNode->removeChild( $node ); |
| 556 |
} |
| 557 |
} |
| 558 |
} |
| 559 |
|
| 560 |
// Glossary tooltips are BetterDocs Pro's only `the_content` injector. Keep |
| 561 |
// the term, drop the tooltip markup wrapped around it. |
| 562 |
$tooltips = $xpath->query( '//*[' . $this->has_class( 'glossary-tooltip-container' ) . ']' ); |
| 563 |
if ( $tooltips ) { |
| 564 |
foreach ( iterator_to_array( $tooltips ) as $node ) { |
| 565 |
$this->unwrap( $node ); |
| 566 |
} |
| 567 |
} |
| 568 |
} |
| 569 |
|
| 570 |
/** |
| 571 |
* XPath predicate matching one class name, whitespace-safe. |
| 572 |
* |
| 573 |
* @param string $class |
| 574 |
* @return string |
| 575 |
*/ |
| 576 |
protected function has_class( $class ) { |
| 577 |
return 'contains(concat(" ", normalize-space(@class), " "), " ' . $class . ' ")'; |
| 578 |
} |
| 579 |
|
| 580 |
/** |
| 581 |
* Replace an element with its own child nodes. |
| 582 |
* |
| 583 |
* @param \DOMNode $node |
| 584 |
*/ |
| 585 |
protected function unwrap( $node ) { |
| 586 |
if ( ! $node->parentNode ) { |
| 587 |
return; |
| 588 |
} |
| 589 |
while ( $node->firstChild ) { |
| 590 |
$node->parentNode->insertBefore( $node->firstChild, $node ); |
| 591 |
} |
| 592 |
$node->parentNode->removeChild( $node ); |
| 593 |
} |
| 594 |
|
| 595 |
/** |
| 596 |
* @param \DOMNode $node |
| 597 |
* @param bool $in_code Inside <pre>/<code>, where Markdown is not escaped. |
| 598 |
* @return string |
| 599 |
*/ |
| 600 |
protected function children_md( $node, $in_code = false ) { |
| 601 |
$out = ''; |
| 602 |
foreach ( $node->childNodes as $child ) { |
| 603 |
$out .= $this->node_to_md( $child, $in_code ); |
| 604 |
} |
| 605 |
|
| 606 |
return $out; |
| 607 |
} |
| 608 |
|
| 609 |
/** |
| 610 |
* @param \DOMNode $node |
| 611 |
* @param bool $in_code |
| 612 |
* @return string |
| 613 |
*/ |
| 614 |
protected function node_to_md( $node, $in_code = false ) { |
| 615 |
if ( XML_TEXT_NODE === $node->nodeType ) { |
| 616 |
$text = preg_replace( '/\s+/', ' ', $node->nodeValue ); |
| 617 |
|
| 618 |
return $in_code ? $text : $this->escape( $text ); |
| 619 |
} |
| 620 |
if ( XML_ELEMENT_NODE !== $node->nodeType ) { |
| 621 |
return ''; |
| 622 |
} |
| 623 |
|
| 624 |
$tag = strtolower( $node->nodeName ); |
| 625 |
|
| 626 |
// BetterDocs code snippets carry their own toolbar and line-number column; |
| 627 |
// the toolbar is already pruned, so emit just the code with its language. |
| 628 |
if ( 'div' === $tag && $this->node_has_class( $node, 'betterdocs-code-snippet-wrapper' ) ) { |
| 629 |
return $this->fence( $node->textContent, $node->getAttribute( 'data-language' ) ); |
| 630 |
} |
| 631 |
|
| 632 |
$inner = $this->children_md( $node, $in_code ); |
| 633 |
|
| 634 |
switch ( $tag ) { |
| 635 |
case 'h1': |
| 636 |
case 'h2': |
| 637 |
case 'h3': |
| 638 |
case 'h4': |
| 639 |
case 'h5': |
| 640 |
case 'h6': |
| 641 |
return "\n\n" . str_repeat( '#', (int) substr( $tag, 1 ) ) . ' ' . trim( $inner ) . "\n\n"; |
| 642 |
|
| 643 |
case 'p': |
| 644 |
return "\n\n" . trim( $inner ) . "\n\n"; |
| 645 |
|
| 646 |
case 'br': |
| 647 |
return " \n"; |
| 648 |
|
| 649 |
case 'hr': |
| 650 |
return "\n\n---\n\n"; |
| 651 |
|
| 652 |
case 'strong': |
| 653 |
case 'b': |
| 654 |
return '' === trim( $inner ) ? '' : '**' . trim( $inner ) . '**'; |
| 655 |
|
| 656 |
case 'em': |
| 657 |
case 'i': |
| 658 |
return '' === trim( $inner ) ? '' : '*' . trim( $inner ) . '*'; |
| 659 |
|
| 660 |
case 'del': |
| 661 |
case 's': |
| 662 |
case 'strike': |
| 663 |
return '' === trim( $inner ) ? '' : '~~' . trim( $inner ) . '~~'; |
| 664 |
|
| 665 |
case 'mark': |
| 666 |
return trim( $inner ); |
| 667 |
|
| 668 |
case 'kbd': |
| 669 |
case 'samp': |
| 670 |
return '`' . trim( $node->textContent ) . '`'; |
| 671 |
|
| 672 |
case 'sup': |
| 673 |
return '^' . trim( $inner ); |
| 674 |
|
| 675 |
case 'sub': |
| 676 |
return '~' . trim( $inner ); |
| 677 |
|
| 678 |
case 'a': |
| 679 |
$href = $node->getAttribute( 'href' ); |
| 680 |
$text = trim( $inner ); |
| 681 |
if ( '' === $href ) { |
| 682 |
return $text; |
| 683 |
} |
| 684 |
|
| 685 |
return '[' . ( '' === $text ? $href : $text ) . '](' . $href . ')'; |
| 686 |
|
| 687 |
case 'img': |
| 688 |
$src = $node->getAttribute( 'src' ); |
| 689 |
|
| 690 |
return '' === $src ? '' : ''; |
| 691 |
|
| 692 |
case 'figure': |
| 693 |
return "\n\n" . trim( $inner ) . "\n\n"; |
| 694 |
|
| 695 |
case 'figcaption': |
| 696 |
return '' === trim( $inner ) ? '' : "\n" . trim( $inner ) . "\n"; |
| 697 |
|
| 698 |
case 'iframe': |
| 699 |
$src = $node->getAttribute( 'src' ); |
| 700 |
if ( '' === $src ) { |
| 701 |
return ''; |
| 702 |
} |
| 703 |
$label = $node->getAttribute( 'title' ); |
| 704 |
|
| 705 |
return "\n\n[" . ( '' === $label ? $src : $this->escape( $label ) ) . '](' . $src . ")\n\n"; |
| 706 |
|
| 707 |
case 'code': |
| 708 |
// Inline code only; fenced blocks are handled by <pre>. |
| 709 |
return '`' . trim( $node->textContent ) . '`'; |
| 710 |
|
| 711 |
case 'pre': |
| 712 |
return $this->fence( $node->textContent, $this->language_of( $node ) ); |
| 713 |
|
| 714 |
case 'blockquote': |
| 715 |
$quote = trim( $this->children_md( $node, $in_code ) ); |
| 716 |
|
| 717 |
return "\n\n" . preg_replace( '/^/m', '> ', $quote ) . "\n\n"; |
| 718 |
|
| 719 |
case 'ul': |
| 720 |
return "\n\n" . $this->list_md( $node, false ) . "\n\n"; |
| 721 |
|
| 722 |
case 'ol': |
| 723 |
return "\n\n" . $this->list_md( $node, true ) . "\n\n"; |
| 724 |
|
| 725 |
case 'table': |
| 726 |
return $this->table_md( $node ); |
| 727 |
|
| 728 |
case 'dl': |
| 729 |
return "\n\n" . trim( $inner ) . "\n\n"; |
| 730 |
|
| 731 |
case 'dt': |
| 732 |
return "\n" . '**' . trim( $inner ) . '**' . "\n"; |
| 733 |
|
| 734 |
case 'dd': |
| 735 |
return ': ' . trim( $inner ) . "\n"; |
| 736 |
|
| 737 |
default: |
| 738 |
return $inner; // unwrap unknown containers (div/span/section/…) |
| 739 |
} |
| 740 |
} |
| 741 |
|
| 742 |
/** |
| 743 |
* Escape the Markdown metacharacters that would otherwise reformat prose. |
| 744 |
* |
| 745 |
* Intra-word underscores are deliberately left alone: CommonMark does not treat |
| 746 |
* them as emphasis, so escaping them only makes identifiers like `my_var` uglier |
| 747 |
* for the model reading them. |
| 748 |
* |
| 749 |
* @param string $text |
| 750 |
* @return string |
| 751 |
*/ |
| 752 |
protected function escape( $text ) { |
| 753 |
return preg_replace( '/([\\\\`\*\[\]])/', '\\\\$1', (string) $text ); |
| 754 |
} |
| 755 |
|
| 756 |
/** |
| 757 |
* @param \DOMNode $node |
| 758 |
* @param string $class |
| 759 |
* @return bool |
| 760 |
*/ |
| 761 |
protected function node_has_class( $node, $class ) { |
| 762 |
if ( ! $node instanceof \DOMElement ) { |
| 763 |
return false; |
| 764 |
} |
| 765 |
|
| 766 |
return in_array( $class, preg_split( '/\s+/', trim( $node->getAttribute( 'class' ) ) ), true ); |
| 767 |
} |
| 768 |
|
| 769 |
/** |
| 770 |
* Language hint for a fenced block, read from the usual `language-*` class on |
| 771 |
* the <pre> or its child <code>. |
| 772 |
* |
| 773 |
* @param \DOMNode $node |
| 774 |
* @return string |
| 775 |
*/ |
| 776 |
protected function language_of( $node ) { |
| 777 |
$candidates = [ $node ]; |
| 778 |
|
| 779 |
foreach ( $node->childNodes as $child ) { |
| 780 |
if ( XML_ELEMENT_NODE === $child->nodeType && 'code' === strtolower( $child->nodeName ) ) { |
| 781 |
$candidates[] = $child; |
| 782 |
} |
| 783 |
} |
| 784 |
|
| 785 |
foreach ( $candidates as $candidate ) { |
| 786 |
if ( ! $candidate instanceof \DOMElement ) { |
| 787 |
continue; |
| 788 |
} |
| 789 |
if ( preg_match( '/(?:language|lang|brush:)[-\s]([a-z0-9#+_-]+)/i', $candidate->getAttribute( 'class' ), $m ) ) { |
| 790 |
return strtolower( $m[1] ); |
| 791 |
} |
| 792 |
} |
| 793 |
|
| 794 |
return ''; |
| 795 |
} |
| 796 |
|
| 797 |
/** |
| 798 |
* @param string $code |
| 799 |
* @param string $language |
| 800 |
* @return string |
| 801 |
*/ |
| 802 |
protected function fence( $code, $language = '' ) { |
| 803 |
$code = rtrim( ltrim( (string) $code, "\r\n" ) ); |
| 804 |
|
| 805 |
if ( '' === trim( $code ) ) { |
| 806 |
return ''; |
| 807 |
} |
| 808 |
|
| 809 |
// Use a longer fence when the snippet itself contains a triple backtick. |
| 810 |
$fence = preg_match( '/^\s*```/m', $code ) ? '````' : '```'; |
| 811 |
|
| 812 |
return "\n\n" . $fence . $language . "\n" . $code . "\n" . $fence . "\n\n"; |
| 813 |
} |
| 814 |
|
| 815 |
/** |
| 816 |
* @param \DOMNode $node The <ul>/<ol> element. |
| 817 |
* @param bool $ordered |
| 818 |
* @return string |
| 819 |
*/ |
| 820 |
protected function list_md( $node, $ordered ) { |
| 821 |
$lines = []; |
| 822 |
$index = 1; |
| 823 |
|
| 824 |
foreach ( $node->childNodes as $li ) { |
| 825 |
if ( XML_ELEMENT_NODE !== $li->nodeType || 'li' !== strtolower( $li->nodeName ) ) { |
| 826 |
continue; |
| 827 |
} |
| 828 |
|
| 829 |
$marker = $ordered ? ( $index++ . '. ' ) : '- '; |
| 830 |
$content = trim( $this->children_md( $li ) ); |
| 831 |
|
| 832 |
// Collapse blank lines inside the item so a nested list attaches directly |
| 833 |
// under its parent marker, then indent continuation lines to the marker |
| 834 |
// width (2 for "- ", 3 for "1. "). |
| 835 |
$content = preg_replace( "/\n{2,}/", "\n", $content ); |
| 836 |
$content = str_replace( "\n", "\n" . str_repeat( ' ', strlen( $marker ) ), $content ); |
| 837 |
|
| 838 |
$lines[] = $marker . $content; |
| 839 |
} |
| 840 |
|
| 841 |
return implode( "\n", $lines ); |
| 842 |
} |
| 843 |
|
| 844 |
/** |
| 845 |
* Convert a <table> to a GFM pipe table. Falls back to unwrapped text when the |
| 846 |
* table has no rows we can line up. |
| 847 |
* |
| 848 |
* @param \DOMNode $node |
| 849 |
* @return string |
| 850 |
*/ |
| 851 |
protected function table_md( $node ) { |
| 852 |
$dom = $node->ownerDocument; |
| 853 |
$rows = []; |
| 854 |
|
| 855 |
$tr_nodes = ( new \DOMXPath( $dom ) )->query( './/tr', $node ); |
| 856 |
if ( ! $tr_nodes || 0 === $tr_nodes->length ) { |
| 857 |
return "\n\n" . trim( $this->children_md( $node ) ) . "\n\n"; |
| 858 |
} |
| 859 |
|
| 860 |
foreach ( $tr_nodes as $tr ) { |
| 861 |
$cells = []; |
| 862 |
foreach ( $tr->childNodes as $cell ) { |
| 863 |
if ( XML_ELEMENT_NODE !== $cell->nodeType ) { |
| 864 |
continue; |
| 865 |
} |
| 866 |
$name = strtolower( $cell->nodeName ); |
| 867 |
if ( 'td' !== $name && 'th' !== $name ) { |
| 868 |
continue; |
| 869 |
} |
| 870 |
// A pipe table cell is a single line; a literal pipe must be escaped. |
| 871 |
$text = trim( preg_replace( "/\s*\n\s*/", ' ', $this->children_md( $cell ) ) ); |
| 872 |
$cells[] = str_replace( '|', '\|', $text ); |
| 873 |
} |
| 874 |
if ( $cells ) { |
| 875 |
$rows[] = $cells; |
| 876 |
} |
| 877 |
} |
| 878 |
|
| 879 |
if ( ! $rows ) { |
| 880 |
return ''; |
| 881 |
} |
| 882 |
|
| 883 |
$columns = max( array_map( 'count', $rows ) ); |
| 884 |
$header = array_shift( $rows ); |
| 885 |
$header = array_pad( $header, $columns, '' ); |
| 886 |
|
| 887 |
$out = '| ' . implode( ' | ', $header ) . " |\n"; |
| 888 |
$out .= '| ' . implode( ' | ', array_fill( 0, $columns, '---' ) ) . " |\n"; |
| 889 |
|
| 890 |
foreach ( $rows as $row ) { |
| 891 |
$out .= '| ' . implode( ' | ', array_pad( $row, $columns, '' ) ) . " |\n"; |
| 892 |
} |
| 893 |
|
| 894 |
return "\n\n" . rtrim( $out ) . "\n\n"; |
| 895 |
} |
| 896 |
} |
| 897 |
|
