BlockMarkup
6 days ago
CSS
6 days ago
DataFormatConsumer
6 days ago
DataFormatProducer
6 days ago
EntityReader
6 days ago
EntityWriter
6 days ago
Importer
6 days ago
URL
6 days ago
vendor-patched
6 days ago
LICENSE.md
6 days ago
README.md
6 days ago
class-dataliberationexception.php
6 days ago
class-dataliberationhtmlprocessor.php
6 days ago
class-importentity.php
6 days ago
composer.json
6 days ago
rector.php
6 days ago
class-dataliberationhtmlprocessor.php
266 lines
| 1 | <?php |
| 2 | |
| 3 | namespace WordPress\DataLiberation; |
| 4 | |
| 5 | use WP_HTML_Processor; |
| 6 | use WP_HTML_Tag_Processor; |
| 7 | |
| 8 | /** |
| 9 | * HTML Processor helpers used by the Data Liberation import pipeline. |
| 10 | * |
| 11 | * This class intentionally keeps the same no-burden progressive upgrade model |
| 12 | * as the public HTML API. When the native extension is loaded, inherited cursor |
| 13 | * operations such as `next_tag()`, `next_token()`, `get_attribute()`, and |
| 14 | * `get_modifiable_text()` use the native processor automatically. Consumers do |
| 15 | * not need to opt in or choose a separate class. |
| 16 | * |
| 17 | * The Data Liberation-only helpers below expose source-string offsets and |
| 18 | * substrings: |
| 19 | * |
| 20 | * $p = DataLiberationHTMLProcessor::create_fragment( '<h1>Title</h1><p>Body</p>' ); |
| 21 | * $p->next_tag( 'H1' ); |
| 22 | * $inner_html = $p->get_inner_html(); // "Title". |
| 23 | * |
| 24 | * $p->skip_to_closer(); |
| 25 | * $after_h1 = substr( $html, $p->get_string_index_after_current_token() ); |
| 26 | * |
| 27 | * Those offsets are an implementation detail of the PHP parser; native HTML |
| 28 | * bookmarks are opaque and do not expose byte spans. To keep native processing |
| 29 | * available for normal scanning, these helpers replay the current cursor |
| 30 | * position on a PHP-backed mirror and read the offsets from that mirror. |
| 31 | */ |
| 32 | class DataLiberationHTMLProcessor extends WP_HTML_Processor { |
| 33 | |
| 34 | /** |
| 35 | * Number of public tokens consumed from this processor. |
| 36 | * |
| 37 | * This is enough to replay the cursor onto a PHP-backed mirror for the |
| 38 | * Data Liberation helpers that need PHP bookmark spans. |
| 39 | * |
| 40 | * @var int |
| 41 | */ |
| 42 | private $tokens_parsed_for_php_replay = 0; |
| 43 | |
| 44 | /** |
| 45 | * Token counts captured for public bookmarks. |
| 46 | * |
| 47 | * Native bookmarks are intentionally opaque. This map lets a later `seek()` |
| 48 | * restore the replay counter when the bookmark was created through this |
| 49 | * class, so the PHP mirror can land on the same token. |
| 50 | * |
| 51 | * @var array<string,int> |
| 52 | */ |
| 53 | private $php_replay_bookmarks = array(); |
| 54 | |
| 55 | /** |
| 56 | * Whether the current cursor can be replayed onto the PHP parser. |
| 57 | * |
| 58 | * @var bool |
| 59 | */ |
| 60 | private $can_replay_php_cursor = true; |
| 61 | |
| 62 | /** |
| 63 | * Whether this instance exists only as a PHP replay mirror. |
| 64 | * |
| 65 | * @var bool |
| 66 | */ |
| 67 | private $is_php_replay = false; |
| 68 | |
| 69 | public function next_token(): bool { |
| 70 | $matched = parent::next_token(); |
| 71 | if ( $matched ) { |
| 72 | ++$this->tokens_parsed_for_php_replay; |
| 73 | } |
| 74 | |
| 75 | return $matched; |
| 76 | } |
| 77 | |
| 78 | public function set_bookmark( $name ): bool { |
| 79 | $set = parent::set_bookmark( $name ); |
| 80 | if ( $set ) { |
| 81 | $this->php_replay_bookmarks[ $name ] = $this->tokens_parsed_for_php_replay; |
| 82 | } |
| 83 | |
| 84 | return $set; |
| 85 | } |
| 86 | |
| 87 | public function release_bookmark( $name ): bool { |
| 88 | $released = parent::release_bookmark( $name ); |
| 89 | if ( $released ) { |
| 90 | unset( $this->php_replay_bookmarks[ $name ] ); |
| 91 | } |
| 92 | |
| 93 | return $released; |
| 94 | } |
| 95 | |
| 96 | public function seek( $bookmark_name ): bool { |
| 97 | $found = parent::seek( $bookmark_name ); |
| 98 | if ( ! $found ) { |
| 99 | return false; |
| 100 | } |
| 101 | |
| 102 | if ( isset( $this->php_replay_bookmarks[ $bookmark_name ] ) ) { |
| 103 | $this->tokens_parsed_for_php_replay = $this->php_replay_bookmarks[ $bookmark_name ]; |
| 104 | } else { |
| 105 | $this->can_replay_php_cursor = false; |
| 106 | } |
| 107 | |
| 108 | return true; |
| 109 | } |
| 110 | |
| 111 | public function get_inner_html() { |
| 112 | if ( $this->has_active_native_processor() ) { |
| 113 | $php_processor = $this->create_php_replay_at_current_token(); |
| 114 | |
| 115 | return $php_processor instanceof self ? $php_processor->get_inner_html() : false; |
| 116 | } |
| 117 | |
| 118 | if ( '#tag' !== $this->get_token_type() ) { |
| 119 | return false; |
| 120 | } |
| 121 | |
| 122 | if ( $this->is_tag_closer() ) { |
| 123 | return false; |
| 124 | } |
| 125 | |
| 126 | if ( false === $this->set_bookmark( 'tag-start' ) ) { |
| 127 | return false; |
| 128 | } |
| 129 | |
| 130 | $this->skip_to_closer(); |
| 131 | |
| 132 | if ( false === $this->set_bookmark( 'tag-end' ) ) { |
| 133 | $this->release_bookmark( 'tag-start' ); |
| 134 | |
| 135 | return false; |
| 136 | } |
| 137 | |
| 138 | $tag_start = $this->get_bookmark_span( 'tag-start' ); |
| 139 | $tag_end = $this->get_bookmark_span( 'tag-end' ); |
| 140 | $inner_html_start = $tag_start->start + $tag_start->length; |
| 141 | $inner_html_end = $tag_end->start - $inner_html_start; |
| 142 | |
| 143 | if ( ! $this->is_php_replay ) { |
| 144 | $this->seek( 'tag-start' ); |
| 145 | $this->release_bookmark( 'tag-start' ); |
| 146 | $this->release_bookmark( 'tag-end' ); |
| 147 | } |
| 148 | |
| 149 | return substr( |
| 150 | $this->html, |
| 151 | $inner_html_start, |
| 152 | $inner_html_end |
| 153 | ); |
| 154 | } |
| 155 | |
| 156 | public function skip_to_closer() { |
| 157 | $starting_depth = $this->get_current_depth(); |
| 158 | while ( $this->next_token() ) { |
| 159 | if ( |
| 160 | '#tag' === $this->get_token_type() && |
| 161 | $this->is_tag_closer() && |
| 162 | $this->get_current_depth() === $starting_depth - 1 |
| 163 | ) { |
| 164 | return true; |
| 165 | } |
| 166 | } |
| 167 | |
| 168 | return false; |
| 169 | } |
| 170 | |
| 171 | public function get_string_index_after_current_token() { |
| 172 | if ( $this->has_active_native_processor() ) { |
| 173 | $php_processor = $this->create_php_replay_at_current_token(); |
| 174 | |
| 175 | return $php_processor instanceof self ? $php_processor->get_string_index_after_current_token() : false; |
| 176 | } |
| 177 | |
| 178 | $name = 'current_token'; |
| 179 | $this->set_bookmark( $name ); |
| 180 | $bookmark = $this->get_bookmark_span( '_' . $name ); |
| 181 | $this->release_bookmark( $name ); |
| 182 | |
| 183 | return $bookmark->start + $bookmark->length; |
| 184 | } |
| 185 | |
| 186 | /** |
| 187 | * Returns a PHP bookmark span from this object or its PHP delegate. |
| 188 | * |
| 189 | * @param string $name Bookmark name. |
| 190 | * @return WP_HTML_Span Bookmark span. |
| 191 | */ |
| 192 | private function get_bookmark_span( $name ) { |
| 193 | if ( isset( $this->bookmarks[ $name ] ) ) { |
| 194 | return $this->bookmarks[ $name ]; |
| 195 | } |
| 196 | if ( isset( $this->bookmarks[ '_' . $name ] ) ) { |
| 197 | return $this->bookmarks[ '_' . $name ]; |
| 198 | } |
| 199 | |
| 200 | if ( property_exists( $this, 'php_processor' ) ) { |
| 201 | $property = new \ReflectionProperty( 'WP_HTML_PHP_Tag_Processor', 'bookmarks' ); |
| 202 | $property->setAccessible( true ); |
| 203 | $bookmarks = $property->getValue( $this->php_processor ); |
| 204 | if ( isset( $bookmarks[ $name ] ) ) { |
| 205 | return $bookmarks[ $name ]; |
| 206 | } |
| 207 | if ( isset( $bookmarks[ '_' . $name ] ) ) { |
| 208 | return $bookmarks[ '_' . $name ]; |
| 209 | } |
| 210 | } |
| 211 | |
| 212 | return null; |
| 213 | } |
| 214 | |
| 215 | /** |
| 216 | * Checks whether this processor has a native delegate. |
| 217 | * |
| 218 | * The pure PHP HTML Processor loaded on older runtimes does not define the |
| 219 | * native-wrapper helper method, so callers must guard the method before |
| 220 | * invoking it. |
| 221 | * |
| 222 | * @return bool Whether a native processor is active. |
| 223 | */ |
| 224 | private function has_active_native_processor() { |
| 225 | return method_exists( $this, 'has_native_processor' ) && $this->has_native_processor(); |
| 226 | } |
| 227 | |
| 228 | /** |
| 229 | * Creates a PHP-backed processor at the same public cursor token. |
| 230 | * |
| 231 | * Native cursor operations are faster for scanning, but native bookmarks do |
| 232 | * not expose byte offsets. For offset-only helpers this method replays the |
| 233 | * already-consumed public token count on a mirror with native delegation |
| 234 | * disabled. Example: |
| 235 | * |
| 236 | * $native = DataLiberationHTMLProcessor::create_fragment( '<h1>Title</h1><p>Body</p>' ); |
| 237 | * $native->next_tag( 'H1' ); |
| 238 | * |
| 239 | * // The mirror runs the PHP parser through the same first token, so its |
| 240 | * // bookmark spans point to the original `<h1>` source bytes. |
| 241 | * $php = $native->create_php_replay_at_current_token(); |
| 242 | * |
| 243 | * @return self|null PHP-backed processor at the current token, or null. |
| 244 | */ |
| 245 | private function create_php_replay_at_current_token() { |
| 246 | if ( ! $this->can_replay_php_cursor ) { |
| 247 | return null; |
| 248 | } |
| 249 | |
| 250 | $processor = static::create_fragment( $this->html ); |
| 251 | if ( ! $processor instanceof self ) { |
| 252 | return null; |
| 253 | } |
| 254 | |
| 255 | $processor->set_native_processor( null ); |
| 256 | $processor->is_php_replay = true; |
| 257 | for ( $i = 0; $i < $this->tokens_parsed_for_php_replay; ++$i ) { |
| 258 | if ( ! $processor->next_token() ) { |
| 259 | return null; |
| 260 | } |
| 261 | } |
| 262 | |
| 263 | return $processor; |
| 264 | } |
| 265 | } |
| 266 |