| 1 |
<?php |
| 2 |
/** |
| 3 |
* Page-builder content extraction. |
| 4 |
* |
| 5 |
* SEO analysis reads `post_content`, which is only the real text on a classic |
| 6 |
* post. Page builders keep the words somewhere else, and every server-side |
| 7 |
* scoring path — bulk analysis, the post-list SEO Overview column, the MCP |
| 8 |
* abilities, cron reports — saw an empty page as a result: |
| 9 |
* |
| 10 |
* - Oxygen / Breakdance leave `post_content` completely EMPTY and store the |
| 11 |
* node tree in postmeta. Nothing to render, nothing to strip: the analyzer |
| 12 |
* reported "No content" on pages with well over a thousand visible words. |
| 13 |
* - Elementor does the same via `_elementor_data`. |
| 14 |
* - Divi 5 and Gutenberg do store block markup in `post_content`, but Divi |
| 15 |
* keeps module text inside the block's JSON attributes — inside an HTML |
| 16 |
* comment, which tag stripping removes wholesale. |
| 17 |
* - Divi 4 and other shortcode builders keep text in shortcode attributes. |
| 18 |
* |
| 19 |
* Extraction reads the builder's own stored data rather than invoking its |
| 20 |
* render engine. Rendering an Oxygen page outside a front-end request is slow, |
| 21 |
* stateful and can fatal in an admin context, whereas the stored tree is just |
| 22 |
* JSON — cheap, side-effect free and safe to touch during a bulk run. |
| 23 |
* |
| 24 |
* @package ThinkRank\SEO |
| 25 |
* @since 1.23.0 |
| 26 |
*/ |
| 27 |
|
| 28 |
declare(strict_types=1); |
| 29 |
|
| 30 |
namespace ThinkRank\SEO; |
| 31 |
|
| 32 |
if (!defined('ABSPATH')) { |
| 33 |
exit; |
| 34 |
} |
| 35 |
|
| 36 |
/** |
| 37 |
* Resolves the analyzable content of a post, whatever built it. |
| 38 |
*/ |
| 39 |
class Builder_Content { |
| 40 |
|
| 41 |
/** |
| 42 |
* Post meta keys that hold builder data, in priority order. |
| 43 |
* |
| 44 |
* Several generations of the same builder are listed on purpose: Oxygen 6 |
| 45 |
* is Breakdance under the hood (`_breakdance_data`), while earlier Oxygen |
| 46 |
* releases used `_oxygen_data` or the shortcode-based |
| 47 |
* `ct_builder_shortcodes`. A site can only have one of them. |
| 48 |
* |
| 49 |
* @var string[] |
| 50 |
*/ |
| 51 |
private const BUILDER_META_KEYS = [ |
| 52 |
'_breakdance_data', // Oxygen 6+ / Breakdance |
| 53 |
'_oxygen_data', // Oxygen (earlier releases) |
| 54 |
'ct_builder_shortcodes', // Oxygen classic |
| 55 |
'_elementor_data', // Elementor |
| 56 |
]; |
| 57 |
|
| 58 |
/** |
| 59 |
* JSON keys whose values are user-visible text. |
| 60 |
* |
| 61 |
* Builder trees mix content with configuration, so a blind string sweep |
| 62 |
* would count CSS classes and option slugs as words. Matching on the key |
| 63 |
* keeps the word count honest. |
| 64 |
* |
| 65 |
* @var string[] |
| 66 |
*/ |
| 67 |
private const CONTENT_KEYS = [ |
| 68 |
'text', 'title', 'subtitle', 'heading', 'subheading', 'content', |
| 69 |
'description', 'caption', 'excerpt', 'label', 'value', 'html', |
| 70 |
'editor', 'quote', 'answer', 'question', 'body', 'button_text', |
| 71 |
]; |
| 72 |
|
| 73 |
/** |
| 74 |
* Resolve the content worth analyzing for a post. |
| 75 |
* |
| 76 |
* @param \WP_Post $post Post being analyzed. |
| 77 |
* @return string HTML/text to analyze. |
| 78 |
*/ |
| 79 |
public static function resolve(\WP_Post $post): string { |
| 80 |
return self::resolve_markup((string) $post->post_content, $post); |
| 81 |
} |
| 82 |
|
| 83 |
/** |
| 84 |
* Resolve an arbitrary chunk of editor markup for the given post. |
| 85 |
* |
| 86 |
* The editor sends its live content to the scorer so an author sees their |
| 87 |
* unsaved edits reflected. On a builder page that live string is the raw |
| 88 |
* builder markup — the block editor hands over Divi's |
| 89 |
* `<!-- wp:divi/... -->` comments verbatim, because it cannot render |
| 90 |
* blocks it has no client-side registration for. Analyzed as-is it reads |
| 91 |
* as zero words, which is how a Divi page could show a correct saved score |
| 92 |
* while the live Content Analysis panel next to it still said |
| 93 |
* "No content". |
| 94 |
* |
| 95 |
* Running the live string through the same chain as stored content keeps |
| 96 |
* both paths honest, and falling through to the post's builder storage |
| 97 |
* covers builders (Oxygen) whose editor content is empty to begin with. |
| 98 |
* |
| 99 |
* @since 1.23.0 |
| 100 |
* |
| 101 |
* @param string $raw Markup to analyze. |
| 102 |
* @param \WP_Post $post Post the markup belongs to. |
| 103 |
* @return string Content to analyze. |
| 104 |
*/ |
| 105 |
public static function resolve_markup(string $raw, \WP_Post $post): string { |
| 106 |
$content = self::render_post_content($raw); |
| 107 |
|
| 108 |
// Block markup that renders to nothing usually means the builder that |
| 109 |
// owns those blocks did not register them in this context — Divi 5 |
| 110 |
// loads its module library lazily per-request, so in CLI, REST, admin |
| 111 |
// and block-editor requests do_blocks() yields an empty string while |
| 112 |
// the words sit right there in the block attributes. Read them |
| 113 |
// directly. |
| 114 |
if (self::is_blank($content)) { |
| 115 |
$from_blocks = self::from_block_attributes($raw); |
| 116 |
if (!self::is_blank($from_blocks)) { |
| 117 |
$content = $from_blocks; |
| 118 |
} |
| 119 |
} |
| 120 |
|
| 121 |
// Only reach for builder storage when the markup yielded nothing — a |
| 122 |
// classic post must never pay for this. |
| 123 |
if (self::is_blank($content)) { |
| 124 |
$builder = self::from_builder_meta((int) $post->ID); |
| 125 |
if (!self::is_blank($builder)) { |
| 126 |
$content = $builder; |
| 127 |
} |
| 128 |
} |
| 129 |
|
| 130 |
// A resolution that collapsed to nothing is worse than the raw markup. |
| 131 |
if (self::is_blank($content) && !self::is_blank($raw)) { |
| 132 |
$content = $raw; |
| 133 |
} |
| 134 |
|
| 135 |
/** |
| 136 |
* Filter the content ThinkRank analyzes for a post. |
| 137 |
* |
| 138 |
* Use this to teach ThinkRank about a builder it does not know, or to |
| 139 |
* override extraction for one it does. |
| 140 |
* |
| 141 |
* @since 1.23.0 |
| 142 |
* |
| 143 |
* @param string $content Resolved content. |
| 144 |
* @param \WP_Post $post Post being analyzed. |
| 145 |
* @param string $raw Markup this resolution started from. |
| 146 |
*/ |
| 147 |
return (string) apply_filters('thinkrank_analyzable_content', $content, $post, $raw); |
| 148 |
} |
| 149 |
|
| 150 |
/** |
| 151 |
* Render blocks and shortcodes found in post_content. |
| 152 |
* |
| 153 |
* Best-effort: a third-party block that fatals must not take the whole |
| 154 |
* score down with it. |
| 155 |
* |
| 156 |
* @param string $raw Raw post content. |
| 157 |
* @return string Rendered content. |
| 158 |
*/ |
| 159 |
private static function render_post_content(string $raw): string { |
| 160 |
if ('' === trim($raw)) { |
| 161 |
return ''; |
| 162 |
} |
| 163 |
|
| 164 |
$content = $raw; |
| 165 |
|
| 166 |
try { |
| 167 |
if (function_exists('has_blocks') && function_exists('do_blocks') && has_blocks($raw)) { |
| 168 |
$content = do_blocks($raw); |
| 169 |
} |
| 170 |
|
| 171 |
// Block output can itself contain shortcodes, so this runs either way. |
| 172 |
if (function_exists('do_shortcode') && strpos($content, '[') !== false) { |
| 173 |
$content = do_shortcode($content); |
| 174 |
} |
| 175 |
} catch (\Throwable $e) { |
| 176 |
return $raw; |
| 177 |
} |
| 178 |
|
| 179 |
return self::is_blank($content) ? $raw : $content; |
| 180 |
} |
| 181 |
|
| 182 |
/** |
| 183 |
* Extract text from the attributes of parsed blocks. |
| 184 |
* |
| 185 |
* @param string $raw Raw post content containing block markup. |
| 186 |
* @return string Collected text, or '' when nothing was found. |
| 187 |
*/ |
| 188 |
private static function from_block_attributes(string $raw): string { |
| 189 |
if (!function_exists('parse_blocks') || !function_exists('has_blocks') || !has_blocks($raw)) { |
| 190 |
return ''; |
| 191 |
} |
| 192 |
|
| 193 |
try { |
| 194 |
$blocks = parse_blocks($raw); |
| 195 |
} catch (\Throwable $e) { |
| 196 |
return ''; |
| 197 |
} |
| 198 |
|
| 199 |
$attrs = []; |
| 200 |
$collect = static function (array $items) use (&$collect, &$attrs): void { |
| 201 |
foreach ($items as $block) { |
| 202 |
if (!empty($block['attrs']) && is_array($block['attrs'])) { |
| 203 |
$attrs[] = $block['attrs']; |
| 204 |
} |
| 205 |
if (!empty($block['innerBlocks']) && is_array($block['innerBlocks'])) { |
| 206 |
$collect($block['innerBlocks']); |
| 207 |
} |
| 208 |
} |
| 209 |
}; |
| 210 |
$collect($blocks); |
| 211 |
|
| 212 |
return empty($attrs) ? '' : self::text_from_tree($attrs); |
| 213 |
} |
| 214 |
|
| 215 |
/** |
| 216 |
* Pull text out of whichever builder stored this post. |
| 217 |
* |
| 218 |
* @param int $post_id Post ID. |
| 219 |
* @return string Extracted text, or '' when no builder data was found. |
| 220 |
*/ |
| 221 |
private static function from_builder_meta(int $post_id): string { |
| 222 |
foreach (self::BUILDER_META_KEYS as $key) { |
| 223 |
$stored = get_post_meta($post_id, $key, true); |
| 224 |
|
| 225 |
if (is_string($stored) && '' !== trim($stored)) { |
| 226 |
$decoded = json_decode($stored, true); |
| 227 |
|
| 228 |
// JSON node tree (Breakdance/Oxygen 6, Elementor). |
| 229 |
if (is_array($decoded)) { |
| 230 |
$text = self::text_from_tree($decoded); |
| 231 |
if (!self::is_blank($text)) { |
| 232 |
return $text; |
| 233 |
} |
| 234 |
continue; |
| 235 |
} |
| 236 |
|
| 237 |
// Shortcode tree (Oxygen classic). |
| 238 |
if (strpos($stored, '[') !== false && function_exists('do_shortcode')) { |
| 239 |
try { |
| 240 |
$rendered = do_shortcode($stored); |
| 241 |
} catch (\Throwable $e) { |
| 242 |
$rendered = $stored; |
| 243 |
} |
| 244 |
if (!self::is_blank($rendered)) { |
| 245 |
return $rendered; |
| 246 |
} |
| 247 |
} |
| 248 |
|
| 249 |
continue; |
| 250 |
} |
| 251 |
|
| 252 |
// Some builders store an already-decoded array. |
| 253 |
if (is_array($stored)) { |
| 254 |
$text = self::text_from_tree($stored); |
| 255 |
if (!self::is_blank($text)) { |
| 256 |
return $text; |
| 257 |
} |
| 258 |
} |
| 259 |
} |
| 260 |
|
| 261 |
return ''; |
| 262 |
} |
| 263 |
|
| 264 |
/** |
| 265 |
* Walk a builder node tree and collect the user-visible text. |
| 266 |
* |
| 267 |
* Values are joined with block-level markup so downstream heading, link and |
| 268 |
* image detection keeps working on the result. |
| 269 |
* |
| 270 |
* @param array $tree Decoded builder tree. |
| 271 |
* @return string Collected HTML. |
| 272 |
*/ |
| 273 |
private static function text_from_tree(array $tree): string { |
| 274 |
$collected = []; |
| 275 |
|
| 276 |
$walk = static function ($node, $key = null) use (&$walk, &$collected): void { |
| 277 |
if (is_array($node)) { |
| 278 |
foreach ($node as $child_key => $child) { |
| 279 |
$walk($child, is_string($child_key) ? $child_key : $key); |
| 280 |
} |
| 281 |
return; |
| 282 |
} |
| 283 |
|
| 284 |
if (!is_string($node) || '' === trim($node)) { |
| 285 |
return; |
| 286 |
} |
| 287 |
|
| 288 |
$is_content_key = is_string($key) |
| 289 |
&& in_array(strtolower($key), self::CONTENT_KEYS, true); |
| 290 |
|
| 291 |
// Markup is content wherever it appears; bare strings only count |
| 292 |
// when their key says they are content, so slugs and class names |
| 293 |
// stay out of the word count. |
| 294 |
if ($is_content_key || strpos($node, '<') !== false) { |
| 295 |
$collected[] = $node; |
| 296 |
} |
| 297 |
}; |
| 298 |
|
| 299 |
$walk($tree); |
| 300 |
|
| 301 |
if (empty($collected)) { |
| 302 |
return ''; |
| 303 |
} |
| 304 |
|
| 305 |
// De-duplicate: builder trees often repeat a value across responsive |
| 306 |
// breakpoints, which would otherwise multiply the word count. |
| 307 |
$collected = array_unique($collected); |
| 308 |
|
| 309 |
return implode("\n", $collected); |
| 310 |
} |
| 311 |
|
| 312 |
/** |
| 313 |
* Whether a value carries no readable text. |
| 314 |
* |
| 315 |
* @param string $value Candidate content. |
| 316 |
* @return bool |
| 317 |
*/ |
| 318 |
private static function is_blank(string $value): bool { |
| 319 |
return '' === trim(wp_strip_all_tags($value)); |
| 320 |
} |
| 321 |
} |
| 322 |
|
