XmlStreamReader
3 weeks ago
partner-discount-sdk
3 weeks ago
api.php
3 weeks ago
arraytoxml.php
3 weeks ago
chunk.php
3 weeks ago
config.php
2 years ago
download.php
3 weeks ago
error.php
3 weeks ago
handler.php
3 weeks ago
helper.php
3 weeks ago
input.php
3 weeks ago
nested.php
3 weeks ago
rapidaddon.php
3 weeks ago
render.php
3 weeks ago
session.php
9 months ago
upload.php
3 weeks ago
zip.php
10 years ago
chunk.php
441 lines
| 1 | <?php |
| 2 | |
| 3 | if ( ! defined( 'ABSPATH' ) ) exit; |
| 4 | include __DIR__ . "/XmlStreamReader/autoload.php"; |
| 5 | |
| 6 | use Prewk\XmlStringStreamer; |
| 7 | use Prewk\XmlStringStreamer\Parser; |
| 8 | use Prewk\XmlStringStreamer\Stream; |
| 9 | |
| 10 | /** |
| 11 | * Chunk |
| 12 | * |
| 13 | * Reads a large file in as chunks for easier parsing. |
| 14 | * |
| 15 | * |
| 16 | * @package default |
| 17 | * @author Max Tsiplyakov |
| 18 | */ |
| 19 | class PMXI_Chunk { |
| 20 | /** |
| 21 | * options |
| 22 | * |
| 23 | * @var array Contains all major options |
| 24 | * @access public |
| 25 | */ |
| 26 | public $options = array( |
| 27 | 'path' => './', // string The path to check for $file in |
| 28 | 'element' => '', // string The XML element to return |
| 29 | 'type' => 'upload', |
| 30 | 'encoding' => 'UTF-8', |
| 31 | 'pointer' => 1, |
| 32 | 'chunkSize' => 1024, |
| 33 | 'filter' => true, |
| 34 | 'get_cloud' => false |
| 35 | ); |
| 36 | |
| 37 | /** |
| 38 | * file |
| 39 | * |
| 40 | * @var string The filename being read |
| 41 | * @access public |
| 42 | */ |
| 43 | public $file = ''; |
| 44 | /** |
| 45 | * pointer |
| 46 | * |
| 47 | * @var integer The current position the file is being read from |
| 48 | * @access public |
| 49 | */ |
| 50 | public $reader; |
| 51 | public $cloud = array(); |
| 52 | public $loop = 1; |
| 53 | public $is_404 = false; |
| 54 | public $parser_type = false; |
| 55 | |
| 56 | /** |
| 57 | * handle |
| 58 | * |
| 59 | * @var resource The fopen() resource |
| 60 | * @access private |
| 61 | */ |
| 62 | private $handle = null; |
| 63 | /** |
| 64 | * reading |
| 65 | * |
| 66 | * @var boolean Whether the script is currently reading the file |
| 67 | * @access private |
| 68 | */ |
| 69 | |
| 70 | /** |
| 71 | * __construct |
| 72 | * |
| 73 | * Builds the Chunk object |
| 74 | * |
| 75 | * @param string $file The filename to work with |
| 76 | * @param array $options The options with which to parse the file |
| 77 | * |
| 78 | * @author Dom Hastings |
| 79 | * @access public |
| 80 | */ |
| 81 | public function __construct( $file, $options = array(), $parser_type = false ) { |
| 82 | try { |
| 83 | |
| 84 | // merge the options together |
| 85 | $this->options = array_merge( $this->options, ( is_array( $options ) ? $options : array() ) ); |
| 86 | |
| 87 | $this->options['chunkSize'] *= PMXI_Plugin::getInstance()->getOption( 'chunk_size' ); |
| 88 | |
| 89 | // set the filename |
| 90 | $this->file = $file; |
| 91 | |
| 92 | $this->parser_type = empty( $parser_type ) ? 'xmlreader' : $parser_type; |
| 93 | |
| 94 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 95 | $sleep = apply_filters( 'wp_all_import_shard_delay', 0 ); |
| 96 | usleep( $sleep ); |
| 97 | |
| 98 | $is_html = false; |
| 99 | $f = @fopen( $file, "rb" ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fopen |
| 100 | if ( is_resource( $file ) ) { |
| 101 | while ( ! @feof( $f ) ) { |
| 102 | $chunk = @fread( $f, 1024 ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fread |
| 103 | if ( strpos( $chunk, "<!DOCTYPE" ) === 0 ) { |
| 104 | $is_html = true; |
| 105 | } |
| 106 | break; |
| 107 | } |
| 108 | @fclose( $f ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose |
| 109 | } |
| 110 | |
| 111 | if ( $is_html ) { |
| 112 | $path = $this->get_file_path(); |
| 113 | |
| 114 | $this->is_404 = true; |
| 115 | |
| 116 | $this->reader = new XMLReader(); |
| 117 | @$this->reader->open( $path ); |
| 118 | @$this->reader->setParserProperty( XMLReader::VALIDATE, false ); |
| 119 | |
| 120 | return; |
| 121 | } |
| 122 | |
| 123 | $input = new PMXI_Input(); |
| 124 | $import_id = $input->get( 'id', 0 ); |
| 125 | if ( empty( $import_id ) ) { |
| 126 | $import_id = $input->get( 'import_id', 0 ); |
| 127 | } |
| 128 | |
| 129 | if ( PMXI_Plugin::getInstance()->getOption( 'force_stream_reader' ) ) { |
| 130 | $this->parser_type = 'xmlstreamer'; |
| 131 | } else { |
| 132 | if ( ! empty( $import_id ) ) { |
| 133 | $this->parser_type = empty( $parser_type ) ? 'xmlreader' : $parser_type; |
| 134 | $import = new PMXI_Import_Record(); |
| 135 | $import->getById( $import_id ); |
| 136 | if ( ! $import->isEmpty() ) { |
| 137 | $this->parser_type = empty( $import->options['xml_reader_engine'] ) ? 'xmlreader' : 'xmlstreamer'; |
| 138 | } |
| 139 | } else { |
| 140 | $this->parser_type = empty( $parser_type ) ? get_option( 'wpai_parser_type', 'xmlreader' ) : $parser_type; |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | if ( empty( $this->options['element'] ) or $this->options['get_cloud'] ) { |
| 145 | $path = $this->get_file_path(); |
| 146 | |
| 147 | if ( $this->parser_type == 'xmlreader' ) { |
| 148 | $reader = new XMLReader(); |
| 149 | $reader->open( $path ); |
| 150 | $reader->setParserProperty( XMLReader::VALIDATE, false ); |
| 151 | while ( @$reader->read() ) { |
| 152 | switch ( $reader->nodeType ) { |
| 153 | case ( XMLREADER::ELEMENT ): |
| 154 | $localName = str_replace( "_colon_", ":", $reader->localName ); |
| 155 | if ( array_key_exists( str_replace( ":", "_", $localName ), $this->cloud ) ) { |
| 156 | $this->cloud[ str_replace( ":", "_", $localName ) ] ++; |
| 157 | } else { |
| 158 | $this->cloud[ str_replace( ":", "_", $localName ) ] = 1; |
| 159 | } |
| 160 | break; |
| 161 | default: |
| 162 | |
| 163 | break; |
| 164 | } |
| 165 | } |
| 166 | unset( $reader ); |
| 167 | } else { |
| 168 | $CHUNK_SIZE = 1024; |
| 169 | $streamProvider = new Prewk\XmlStringStreamer\Stream\File( $path, $CHUNK_SIZE ); |
| 170 | $parseroptions = array( |
| 171 | "extractContainer" => false, // Required option |
| 172 | ); |
| 173 | // Works like an XmlReader, and walks the XML tree node by node. Captures by node depth setting. |
| 174 | $parser = new Parser\StringWalker( $parseroptions ); |
| 175 | // Create the streamer |
| 176 | $streamer = new XmlStringStreamer( $parser, $streamProvider ); |
| 177 | while ( $node = $streamer->getNode() ) { |
| 178 | // $simpleXmlNode = simplexml_load_string($node); |
| 179 | // echo (string)$simpleXmlNode->firstName; |
| 180 | } |
| 181 | |
| 182 | $this->cloud = $parser->cloud; |
| 183 | |
| 184 | } |
| 185 | |
| 186 | if ( ! empty( $this->cloud ) and empty( $this->options['element'] ) ) { |
| 187 | |
| 188 | arsort( $this->cloud ); |
| 189 | |
| 190 | $main_elements = array( |
| 191 | 'node', |
| 192 | 'product', |
| 193 | 'job', |
| 194 | 'deal', |
| 195 | 'entry', |
| 196 | 'item', |
| 197 | 'property', |
| 198 | 'listing', |
| 199 | 'hotel', |
| 200 | 'record', |
| 201 | 'article', |
| 202 | 'post', |
| 203 | 'book', |
| 204 | 'item_0' |
| 205 | ); |
| 206 | |
| 207 | foreach ( $this->cloud as $element_name => $value ) { |
| 208 | if ( in_array( strtolower( $element_name ), $main_elements ) ) { |
| 209 | $this->options['element'] = $element_name; |
| 210 | break; |
| 211 | } |
| 212 | } |
| 213 | |
| 214 | if ( empty( $this->options['element'] ) ) { |
| 215 | foreach ( $this->cloud as $el => $count ) { |
| 216 | $this->options['element'] = $el; |
| 217 | break; |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 222 | $this->options['element'] = apply_filters( 'wp_all_import_root_element', $this->options['element'], $import_id, $this->cloud ); |
| 223 | } |
| 224 | } |
| 225 | |
| 226 | $path = $this->get_file_path(); |
| 227 | |
| 228 | if ( $this->parser_type == 'xmlreader' ) { |
| 229 | $this->reader = new XMLReader(); |
| 230 | @$this->reader->open( $path ); |
| 231 | @$this->reader->setParserProperty( XMLReader::VALIDATE, false ); |
| 232 | } else { |
| 233 | $parseroptions = array( |
| 234 | "uniqueNode" => $this->options['element'] |
| 235 | ); |
| 236 | $CHUNK_SIZE = 1024; |
| 237 | $streamProvider = new Prewk\XmlStringStreamer\Stream\File( $path, $CHUNK_SIZE ); |
| 238 | $parser = new Parser\UniqueNode( $parseroptions ); |
| 239 | $this->reader = new XmlStringStreamer( $parser, $streamProvider ); |
| 240 | } |
| 241 | } catch ( Throwable $e ) { |
| 242 | |
| 243 | if (defined('WP_DEBUG') && WP_DEBUG) { |
| 244 | error_log('Error in PMXI_Chunk constructor: ' . $e->getMessage()); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log |
| 245 | } |
| 246 | |
| 247 | $this->options = array(); |
| 248 | $this->reader = null; |
| 249 | $this->file = ''; |
| 250 | $this->cloud = array(); |
| 251 | $this->loop = 1; |
| 252 | $this->is_404 = false; |
| 253 | $this->parser_type = false; |
| 254 | } |
| 255 | } |
| 256 | |
| 257 | function get_file_path() { |
| 258 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 259 | $is_enabled_stream_filter = apply_filters( 'wp_all_import_is_enabled_stream_filter', true ); |
| 260 | if ( function_exists( 'stream_filter_register' ) and $this->options['filter'] and $is_enabled_stream_filter and $this->parser_type == 'xmlreader' ) { |
| 261 | stream_filter_register( 'preprocessxml', 'preprocessXml_filter' ); |
| 262 | if ( defined( 'HHVM_VERSION' ) ) { |
| 263 | $path = $this->file; |
| 264 | } else { |
| 265 | $path = 'php://filter/read=preprocessxml/resource=' . $this->file; |
| 266 | } |
| 267 | } else { |
| 268 | $path = $this->file; |
| 269 | } |
| 270 | |
| 271 | return $path; |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * __destruct |
| 276 | * |
| 277 | * Cleans up |
| 278 | * |
| 279 | * @return void |
| 280 | * @author Dom Hastings |
| 281 | * @access public |
| 282 | */ |
| 283 | public function __destruct() { |
| 284 | // close the file resource |
| 285 | unset( $this->reader ); |
| 286 | } |
| 287 | |
| 288 | /** |
| 289 | * read |
| 290 | * |
| 291 | * Reads the first available occurence of the XML element $this->options['element'] |
| 292 | * |
| 293 | * @return string The XML string from $this->file |
| 294 | * @author Dom Hastings |
| 295 | * @access public |
| 296 | */ |
| 297 | public function read( $debug = false ) { |
| 298 | |
| 299 | // Constructor may have set reader to null on failure (e.g. empty path). |
| 300 | if ( ! $this->reader ) { |
| 301 | return false; |
| 302 | } |
| 303 | |
| 304 | // trim it |
| 305 | $element = trim( $this->options['element'] ); |
| 306 | |
| 307 | $xml = ''; |
| 308 | |
| 309 | if ( $this->parser_type == 'xmlreader' ) { |
| 310 | try { |
| 311 | while ( @$this->reader->read() ) { |
| 312 | switch ( $this->reader->nodeType ) { |
| 313 | case ( XMLREADER::ELEMENT ): |
| 314 | |
| 315 | $localName = str_replace( "_colon_", ":", $this->reader->localName ); |
| 316 | |
| 317 | if ( strtolower( str_replace( ":", "_", $localName ) ) == strtolower( $element ) ) { |
| 318 | |
| 319 | if ( $this->loop < $this->options['pointer'] ) { |
| 320 | $this->loop ++; |
| 321 | continue( 2 ); |
| 322 | } |
| 323 | |
| 324 | $xml = @$this->reader->readOuterXML(); |
| 325 | |
| 326 | break( 2 ); |
| 327 | } |
| 328 | break; |
| 329 | default: |
| 330 | // code ... |
| 331 | break; |
| 332 | } |
| 333 | } |
| 334 | } catch ( XmlImportException $e ) { |
| 335 | $xml = false; |
| 336 | } |
| 337 | } else { |
| 338 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 339 | $is_preprocess_enabled = apply_filters( 'is_xml_preprocess_enabled', true ); |
| 340 | |
| 341 | while ( $xml = $this->reader->getNode() ) { |
| 342 | |
| 343 | if ( $this->loop < $this->options['pointer'] ) { |
| 344 | $this->loop ++; |
| 345 | continue; |
| 346 | } |
| 347 | |
| 348 | if ( $is_preprocess_enabled ) { |
| 349 | // the & symbol is not valid in XML, so replace it with temporary word _ampersand_ |
| 350 | $xml = str_replace( "&", "_ampersand_", $xml ); |
| 351 | $xml = preg_replace( '/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', str_replace( ":", "_colon_", $xml ) ); |
| 352 | } |
| 353 | |
| 354 | break; |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | return ( ! empty( $xml ) ) ? self::removeColonsFromRSS( preg_replace( '%xmlns\s*=\s*([\'"]).*\1%sU', '', $xml ) ) : false; |
| 359 | |
| 360 | } |
| 361 | |
| 362 | public static function removeColonsFromRSS( $feed ) { |
| 363 | |
| 364 | $feed = str_replace( "_colon_", ":", $feed ); |
| 365 | |
| 366 | // pull out colons from start tags |
| 367 | // (<\w+):(\w+>) |
| 368 | $pattern = '/(<\w+):([\w+|\.|-]+[ |>]{1})/i'; |
| 369 | $replacement = '$1_$2'; |
| 370 | $feed = preg_replace( $pattern, $replacement, $feed ); |
| 371 | // pull out colons from end tags |
| 372 | // (<\/\w+):(\w+>) |
| 373 | $pattern = '/(<\/\w+):([\w+|\.|-]+>)/i'; |
| 374 | $replacement = '$1_$2'; |
| 375 | $feed = preg_replace( $pattern, $replacement, $feed ); |
| 376 | |
| 377 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 378 | $is_replace_colons = apply_filters( 'wp_all_import_replace_colons_in_attribute_names', true ); |
| 379 | if ( $is_replace_colons ) { |
| 380 | // pull out colons from attributes |
| 381 | $pattern = '/(\s+\w+):(\w+[=]{1})/i'; |
| 382 | $replacement = '$1_$2'; |
| 383 | $feed = preg_replace( $pattern, $replacement, $feed ); |
| 384 | } |
| 385 | // pull colons from single element |
| 386 | // (<\w+):(\w+\/>) |
| 387 | $pattern = '/(<\w+):([\w+|\.|-]+\/>)/i'; |
| 388 | $replacement = '$1_$2'; |
| 389 | $feed = preg_replace( $pattern, $replacement, $feed ); |
| 390 | |
| 391 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 392 | $is_preprocess_enabled = apply_filters( 'is_xml_preprocess_enabled', true ); |
| 393 | if ( $is_preprocess_enabled ) { |
| 394 | // replace temporary word _ampersand_ back to & symbol |
| 395 | $feed = str_replace( "_ampersand_", "&", $feed ); |
| 396 | } |
| 397 | |
| 398 | // replace all standalone & symbols ( which is not in htmlentities e.q. and not wrapped in CDATA section ) to & |
| 399 | PMXI_Import_Record::preprocessXml( $feed ); |
| 400 | |
| 401 | return $feed; |
| 402 | |
| 403 | } |
| 404 | |
| 405 | } |
| 406 | |
| 407 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedClassFound |
| 408 | class preprocessXml_filter extends php_user_filter { |
| 409 | #[\ReturnTypeWillChange] |
| 410 | public function filter( $in, $out, &$consumed, $closing ) { |
| 411 | while ( $bucket = stream_bucket_make_writeable( $in ) ) { |
| 412 | // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound |
| 413 | $is_preprocess_enabled = apply_filters( 'is_xml_preprocess_enabled', true ); |
| 414 | if ( $is_preprocess_enabled ) { |
| 415 | // the & symbol is not valid in XML, so replace it with temporary word _ampersand_ |
| 416 | $bucket->data = str_replace( "&", "_ampersand_", $bucket->data ); |
| 417 | $cleanXML = preg_replace( '/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $this->replace_colons( $bucket->data ) ); |
| 418 | if ( $cleanXML == null && preg_last_error() == PREG_BAD_UTF8_ERROR ) { |
| 419 | $cleanXML = preg_replace( '/[^\x09\x0a\x0d\x20-\xFF]+/', ' ', $this->replace_colons( $bucket->data ) ); |
| 420 | } |
| 421 | if ( $cleanXML == null && preg_last_error() == PREG_BAD_UTF8_ERROR ) { |
| 422 | if ( function_exists( 'mb_ereg_replace' ) ) { |
| 423 | mb_regex_encoding( 'UTF-8' ); |
| 424 | $cleanXML = mb_ereg_replace( '/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $this->replace_colons( $bucket->data ) ); |
| 425 | } |
| 426 | } |
| 427 | $bucket->data = empty( $cleanXML ) ? $this->replace_colons( $bucket->data ) : $cleanXML; |
| 428 | } |
| 429 | $consumed += $bucket->datalen; |
| 430 | stream_bucket_append( $out, $bucket ); |
| 431 | } |
| 432 | |
| 433 | return PSFS_PASS_ON; |
| 434 | } |
| 435 | |
| 436 | function replace_colons( $data ) { |
| 437 | return str_replace( ":", "_colon_", $data ); |
| 438 | } |
| 439 | |
| 440 | } |
| 441 |