matecat /
xliff-parser
| Total Complexity | 49 |
| Total Lines | 474 |
| Duplicated Lines | 0 % |
| Changes | 5 | ||
| Bugs | 0 | Features | 0 |
Complex classes like AbstractXliffReplacer often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use AbstractXliffReplacer, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 8 | abstract class AbstractXliffReplacer { |
||
| 9 | protected $originalFP; |
||
| 10 | protected $outputFP; // output stream pointer |
||
| 11 | |||
| 12 | protected string $tuTagName; // <trans-unit> (forXliff v 1.*) or <unit> (forXliff v 2.*) |
||
| 13 | protected bool $inTU = false; // flag to check whether we are in a <trans-unit> |
||
| 14 | protected bool $inTarget = false; // flag to check whether we are in a <target>, to ignore everything |
||
| 15 | protected bool $isEmpty = false; // flag to check whether we are in an empty tag (<tag/>) |
||
| 16 | protected bool $targetWasWritten = false; // flag to check is <target> was written in the current unit |
||
| 17 | protected string $CDATABuffer = ""; // buffer for special tag |
||
| 18 | protected bool $bufferIsActive = false; // buffer for special tag |
||
| 19 | |||
| 20 | protected int $offset = 0; // offset for SAX pointer |
||
| 21 | |||
| 22 | protected string $currentBuffer; // the current piece of text it's been parsed |
||
| 23 | protected int $len; // length of the currentBuffer |
||
| 24 | protected array $segments; // array of translations |
||
| 25 | protected array $lastTransUnit = []; |
||
| 26 | protected int $segmentInUnitPosition = 0; |
||
| 27 | protected ?string $currentTransUnitId = null; // id of current <trans-unit> |
||
| 28 | protected ?string $currentTransUnitIsTranslatable = null; // 'translate' attribute of current <trans-unit> |
||
| 29 | protected bool $unitContainsMda = false; // check if <unit> already contains a <mda:metadata> (forXliff v 2.*) |
||
| 30 | protected bool $hasWrittenCounts = false; // check if <unit> already wrote segment counts (forXliff v 2.*) |
||
| 31 | protected string $targetLang; |
||
| 32 | protected bool $sourceInTarget = false; |
||
| 33 | |||
| 34 | protected array $nodesToBuffer; |
||
| 35 | |||
| 36 | protected array $transUnits; |
||
| 37 | |||
| 38 | /** @var int */ |
||
| 39 | protected int $xliffVersion; |
||
| 40 | |||
| 41 | /** |
||
| 42 | * @var XliffReplacerCallbackInterface|null |
||
| 43 | */ |
||
| 44 | protected ?XliffReplacerCallbackInterface $callback; |
||
| 45 | |||
| 46 | protected ?LoggerInterface $logger; |
||
| 47 | |||
| 48 | protected static $INTERNAL_TAG_PLACEHOLDER; |
||
| 49 | |||
| 50 | protected $counts = [ |
||
| 51 | 'raw_word_count' => 0, |
||
| 52 | 'eq_word_count' => 0, |
||
| 53 | ]; |
||
| 54 | |||
| 55 | /** |
||
| 56 | * AbstractXliffReplacer constructor. |
||
| 57 | * |
||
| 58 | * @param string $originalXliffPath |
||
| 59 | * @param int $xliffVersion |
||
| 60 | * @param array $segments |
||
| 61 | * @param array $transUnits |
||
| 62 | * @param string $trgLang |
||
| 63 | * @param string $outputFilePath |
||
| 64 | * @param bool $setSourceInTarget |
||
| 65 | * @param LoggerInterface|null $logger |
||
| 66 | * @param XliffReplacerCallbackInterface|null $callback |
||
| 67 | */ |
||
| 68 | public function __construct( |
||
| 69 | string $originalXliffPath, |
||
| 70 | int $xliffVersion, |
||
| 71 | array $segments, |
||
| 72 | array $transUnits, |
||
| 73 | string $trgLang, |
||
| 74 | string $outputFilePath, |
||
| 75 | bool $setSourceInTarget, |
||
| 76 | LoggerInterface $logger = null, |
||
| 77 | XliffReplacerCallbackInterface $callback = null |
||
| 78 | ) { |
||
| 79 | self::$INTERNAL_TAG_PLACEHOLDER = $this->getInternalTagPlaceholder(); |
||
| 80 | $this->createOutputFileIfDoesNotExist( $outputFilePath ); |
||
| 81 | $this->setFileDescriptors( $originalXliffPath, $outputFilePath ); |
||
| 82 | $this->xliffVersion = $xliffVersion; |
||
| 83 | $this->tuTagName = ( $this->xliffVersion === 2 ) ? 'unit' : 'trans-unit'; |
||
| 84 | $this->segments = $segments; |
||
| 85 | $this->targetLang = $trgLang; |
||
| 86 | $this->sourceInTarget = $setSourceInTarget; |
||
| 87 | $this->transUnits = $transUnits; |
||
| 88 | $this->logger = $logger; |
||
| 89 | $this->callback = $callback; |
||
| 90 | } |
||
| 91 | |||
| 92 | public function replaceTranslation() { |
||
| 93 | fwrite( $this->outputFP, '<?xml version="1.0" encoding="UTF-8"?>' ); |
||
| 94 | |||
| 95 | //create Sax parser |
||
| 96 | $xmlParser = $this->initSaxParser(); |
||
| 97 | |||
| 98 | while ( $this->currentBuffer = fread( $this->originalFP, 4096 ) ) { |
||
| 99 | /* |
||
| 100 | preprocess file |
||
| 101 | */ |
||
| 102 | // obfuscate entities because sax automatically does html_entity_decode |
||
| 103 | $temporary_check_buffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer ); |
||
| 104 | |||
| 105 | //avoid cutting entities in half: |
||
| 106 | //the last fread could have truncated an entity (say, '<' in '&l'), thus invalidating the escaping |
||
| 107 | //***** and if there is an & that it is not an entity, this is an infinite loop !!!!! |
||
| 108 | // 9 is the max length of an entity. So, suppose that the & is at the end of buffer, |
||
| 109 | // add 9 Bytes and substitute the entities, if the & is present, and it is not at the end |
||
| 110 | //it can't be an entity, exit the loop |
||
| 111 | while ( true ) { |
||
| 112 | $_ampPos = strpos( $temporary_check_buffer, '&' ); |
||
| 113 | |||
| 114 | //check for real entity or escape it to safely exit from the loop!!! |
||
| 115 | if ( $_ampPos === false || strlen( substr( $temporary_check_buffer, $_ampPos ) ) > 9 ) { |
||
| 116 | break; |
||
| 117 | } |
||
| 118 | |||
| 119 | //if an entity is still present, fetch some more and repeat the escaping |
||
| 120 | $this->currentBuffer .= fread( $this->originalFP, 9 ); |
||
| 121 | $temporary_check_buffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer ); |
||
| 122 | } |
||
| 123 | |||
| 124 | //free stuff outside the loop |
||
| 125 | unset( $temporary_check_buffer ); |
||
| 126 | |||
| 127 | $this->currentBuffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer ); |
||
| 128 | $this->currentBuffer = str_replace( "&", self::$INTERNAL_TAG_PLACEHOLDER . 'amp' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer ); |
||
| 129 | |||
| 130 | //get length of chunk |
||
| 131 | $this->len = strlen( $this->currentBuffer ); |
||
| 132 | |||
| 133 | //parse chunk of text |
||
| 134 | if ( !xml_parse( $xmlParser, $this->currentBuffer, feof( $this->originalFP ) ) ) { |
||
| 135 | //if unable, raise an exception |
||
| 136 | throw new RuntimeException( sprintf( |
||
| 137 | "XML error: %s at line %d", |
||
| 138 | xml_error_string( xml_get_error_code( $xmlParser ) ), |
||
| 139 | xml_get_current_line_number( $xmlParser ) |
||
| 140 | ) ); |
||
| 141 | } |
||
| 142 | /* |
||
| 143 | * Get the accumulated this->offset in the document: |
||
| 144 | * as long as SAX pointer advances, we keep track of total bytes it has seen so far; |
||
| 145 | * this way, we can translate its global pointer in an address local to the current buffer of text to retrieve the last char of tag |
||
| 146 | */ |
||
| 147 | $this->offset += $this->len; |
||
| 148 | } |
||
| 149 | |||
| 150 | // close Sax parser |
||
| 151 | $this->closeSaxParser( $xmlParser ); |
||
| 152 | |||
| 153 | } |
||
| 154 | |||
| 155 | /** |
||
| 156 | * @param resource $parser |
||
| 157 | * |
||
| 158 | * @return string |
||
| 159 | */ |
||
| 160 | protected function getLastCharacter( $parser ): string { |
||
| 161 | |||
| 162 | //this logic helps detecting empty tags |
||
| 163 | //get current position of SAX pointer in all the stream of data is has read so far: |
||
| 164 | //it points at the end of current tag |
||
| 165 | $idx = xml_get_current_byte_index( $parser ); |
||
| 166 | |||
| 167 | //check whether the bounds of current tag are entirely in current buffer or the end of the current tag |
||
| 168 | //is outside current buffer (in the latter case, it's in next buffer to be read by the while loop); |
||
| 169 | //this check is necessary because we may have truncated a tag in half with current read, |
||
| 170 | //and the other half may be encountered in the next buffer it will be passed |
||
| 171 | if ( isset( $this->currentBuffer[ $idx - $this->offset ] ) ) { |
||
| 172 | //if this tag entire lenght fitted in the buffer, the last char must be the last |
||
| 173 | //symbol before the '>'; if it's an empty tag, it is assumed that it's a '/' |
||
| 174 | $lastChar = $this->currentBuffer[ $idx - $this->offset ]; |
||
| 175 | } else { |
||
| 176 | //if it's out, simple use the last character of the chunk |
||
| 177 | $lastChar = $this->currentBuffer[ $this->len - 1 ]; |
||
| 178 | } |
||
| 179 | |||
| 180 | return $lastChar; |
||
| 181 | |||
| 182 | } |
||
| 183 | |||
| 184 | /** |
||
| 185 | * @return string |
||
| 186 | */ |
||
| 187 | private function getInternalTagPlaceholder(): string { |
||
| 188 | return "§" . |
||
| 189 | substr( |
||
| 190 | str_replace( |
||
| 191 | [ '+', '/' ], |
||
| 192 | '', |
||
| 193 | base64_encode( openssl_random_pseudo_bytes( 10, $_crypto_strong ) ) |
||
| 194 | ), |
||
| 195 | 0, |
||
| 196 | 4 |
||
| 197 | ); |
||
| 198 | } |
||
| 199 | |||
| 200 | private function createOutputFileIfDoesNotExist( string $outputFilePath ) { |
||
| 201 | // create output file |
||
| 202 | if ( !file_exists( $outputFilePath ) ) { |
||
| 203 | touch( $outputFilePath ); |
||
| 204 | } |
||
| 205 | } |
||
| 206 | |||
| 207 | /** |
||
| 208 | * @param string $originalXliffPath |
||
| 209 | * @param string $outputFilePath |
||
| 210 | */ |
||
| 211 | private function setFileDescriptors( string $originalXliffPath, string $outputFilePath ) { |
||
| 212 | $this->outputFP = fopen( $outputFilePath, 'w+' ); |
||
| 213 | |||
| 214 | $streamArgs = null; |
||
| 215 | |||
| 216 | if ( !( $this->originalFP = fopen( $originalXliffPath, "r", false, stream_context_create( $streamArgs ) ) ) ) { |
||
| 217 | throw new RuntimeException( "could not open XML input" ); |
||
| 218 | } |
||
| 219 | } |
||
| 220 | |||
| 221 | /** |
||
| 222 | * AbstractXliffReplacer destructor. |
||
| 223 | */ |
||
| 224 | public function __destruct() { |
||
| 225 | //this stream can be closed outside the class |
||
| 226 | //to permit multiple concurrent downloads, so suppress warnings |
||
| 227 | @fclose( $this->originalFP ); |
||
|
|
|||
| 228 | fclose( $this->outputFP ); |
||
| 229 | } |
||
| 230 | |||
| 231 | /** |
||
| 232 | * Init Sax parser |
||
| 233 | * |
||
| 234 | * @return resource |
||
| 235 | */ |
||
| 236 | protected function initSaxParser() { |
||
| 237 | $xmlSaxParser = xml_parser_create( 'UTF-8' ); |
||
| 238 | xml_set_object( $xmlSaxParser, $this ); |
||
| 239 | xml_parser_set_option( $xmlSaxParser, XML_OPTION_CASE_FOLDING, false ); |
||
| 240 | xml_set_element_handler( $xmlSaxParser, 'tagOpen', 'tagClose' ); |
||
| 241 | xml_set_character_data_handler( $xmlSaxParser, 'characterData' ); |
||
| 242 | |||
| 243 | return $xmlSaxParser; |
||
| 244 | } |
||
| 245 | |||
| 246 | /** |
||
| 247 | * @param resource $xmlSaxParser |
||
| 248 | */ |
||
| 249 | protected function closeSaxParser( $xmlSaxParser ) { |
||
| 250 | xml_parser_free( $xmlSaxParser ); |
||
| 251 | } |
||
| 252 | |||
| 253 | /** |
||
| 254 | * @param resource $parser |
||
| 255 | * @param string $name |
||
| 256 | * @param array $attr |
||
| 257 | * |
||
| 258 | * @return mixed |
||
| 259 | */ |
||
| 260 | abstract protected function tagOpen( $parser, string $name, array $attr ); |
||
| 261 | |||
| 262 | /** |
||
| 263 | * @param resource $parser |
||
| 264 | * @param string $name |
||
| 265 | * |
||
| 266 | * @return mixed |
||
| 267 | */ |
||
| 268 | abstract protected function tagClose( $parser, string $name ); |
||
| 269 | |||
| 270 | /** |
||
| 271 | * @param resource $parser |
||
| 272 | * @param string $data |
||
| 273 | * |
||
| 274 | * @return mixed |
||
| 275 | */ |
||
| 276 | protected function characterData( $parser, string $data ): void { |
||
| 282 | } |
||
| 283 | } |
||
| 284 | |||
| 285 | /** |
||
| 286 | * postprocess escaped data and write to disk |
||
| 287 | * |
||
| 288 | * @param resource $fp |
||
| 289 | * @param string $data |
||
| 290 | * @param ?bool $treatAsCDATA |
||
| 291 | */ |
||
| 292 | protected function postProcAndFlush( $fp, string $data, ?bool $treatAsCDATA = false ) { |
||
| 293 | //postprocess string |
||
| 294 | $data = preg_replace( "/" . self::$INTERNAL_TAG_PLACEHOLDER . '(.*?)' . self::$INTERNAL_TAG_PLACEHOLDER . "/", '&$1;', $data ); |
||
| 295 | $data = str_replace( ' ', ' ', $data ); |
||
| 296 | if ( !$treatAsCDATA ) { |
||
| 297 | //unix2dos |
||
| 298 | $data = str_replace( "\r\n", "\r", $data ); |
||
| 299 | $data = str_replace( "\n", "\r", $data ); |
||
| 300 | $data = str_replace( "\r", "\r\n", $data ); |
||
| 301 | } |
||
| 302 | |||
| 303 | //flush to disk |
||
| 304 | fwrite( $fp, $data ); |
||
| 305 | } |
||
| 306 | |||
| 307 | /** |
||
| 308 | * @param string $name |
||
| 309 | * @param array $attr |
||
| 310 | * |
||
| 311 | * @return void |
||
| 312 | */ |
||
| 313 | protected function handleOpenUnit( string $name, array $attr ) { |
||
| 314 | |||
| 315 | // check if we are entering into a <trans-unit> (xliff v1.*) or <unit> (xliff v2.*) |
||
| 316 | if ( $this->tuTagName === $name ) { |
||
| 317 | $this->inTU = true; |
||
| 318 | |||
| 319 | // get id |
||
| 320 | // trim to first 100 characters because this is the limit on Matecat's DB |
||
| 321 | $this->currentTransUnitId = substr( $attr[ 'id' ], 0, 100 ); |
||
| 322 | |||
| 323 | // `translate` attribute can be only yes or no |
||
| 324 | // current 'translate' attribute of the current trans-unit |
||
| 325 | $this->currentTransUnitIsTranslatable = empty( $attr[ 'translate' ] ) ? 'yes' : $attr[ 'translate' ]; |
||
| 326 | |||
| 327 | $this->setLastTransUnitSegments(); |
||
| 328 | |||
| 329 | } |
||
| 330 | } |
||
| 331 | |||
| 332 | /** |
||
| 333 | * @param string $name |
||
| 334 | * @param array $attr |
||
| 335 | * @param string $tag |
||
| 336 | * |
||
| 337 | * @return string |
||
| 338 | */ |
||
| 339 | protected function handleOpenXliffTag( string $name, array $attr, string $tag ): string { |
||
| 340 | |||
| 341 | // Add MateCat specific namespace. |
||
| 342 | // Add trgLang |
||
| 343 | if ( $name === 'xliff' ) { |
||
| 344 | if ( !array_key_exists( 'xmlns:mtc', $attr ) ) { |
||
| 345 | $tag .= ' xmlns:mtc="https://www.matecat.com" '; |
||
| 346 | } |
||
| 347 | $tag = preg_replace( '/trgLang="(.*?)"/', 'trgLang="' . $this->targetLang . '"', $tag ); |
||
| 348 | } |
||
| 349 | |||
| 350 | return $tag; |
||
| 351 | |||
| 352 | } |
||
| 353 | |||
| 354 | /** |
||
| 355 | * @param string $name |
||
| 356 | * |
||
| 357 | * @return void |
||
| 358 | */ |
||
| 359 | protected function checkSetInTarget( string $name ) { |
||
| 360 | // check if we are entering into a <target> |
||
| 361 | if ( 'target' === $name ) { |
||
| 362 | if ( $this->currentTransUnitIsTranslatable === 'no' ) { |
||
| 363 | $this->inTarget = false; |
||
| 364 | } else { |
||
| 365 | $this->inTarget = true; |
||
| 366 | } |
||
| 367 | } |
||
| 368 | } |
||
| 369 | |||
| 370 | /** |
||
| 371 | * @param string $name |
||
| 372 | * |
||
| 373 | * @return void |
||
| 374 | */ |
||
| 375 | protected function setInBuffer( string $name ) { |
||
| 376 | if ( in_array( $name, $this->nodesToBuffer ) ) { |
||
| 377 | $this->bufferIsActive = true; |
||
| 378 | } |
||
| 379 | } |
||
| 380 | |||
| 381 | /** |
||
| 382 | * @param array $seg |
||
| 383 | */ |
||
| 384 | protected function updateSegmentCounts( array $seg = [] ) { |
||
| 396 | } |
||
| 397 | |||
| 398 | protected function resetCounts() { |
||
| 399 | $this->counts[ 'segments_count_array' ] = []; |
||
| 400 | $this->counts[ 'raw_word_count' ] = 0; |
||
| 401 | $this->counts[ 'eq_word_count' ] = 0; |
||
| 402 | } |
||
| 403 | |||
| 404 | /** |
||
| 405 | * @param resource $parser |
||
| 406 | * @param string $tag |
||
| 407 | * |
||
| 408 | * @return void |
||
| 409 | */ |
||
| 410 | protected function checkForSelfClosedTagAndFlush( $parser, string $tag ) { |
||
| 411 | |||
| 412 | $lastChar = $this->getLastCharacter( $parser ); |
||
| 413 | |||
| 414 | //trim last space |
||
| 415 | $tag = rtrim( $tag ); |
||
| 416 | |||
| 417 | //detect empty tag |
||
| 418 | $this->isEmpty = $lastChar == '/'; |
||
| 419 | if ( $this->isEmpty ) { |
||
| 420 | $tag .= $lastChar; |
||
| 421 | } |
||
| 422 | |||
| 423 | //add tag ending |
||
| 424 | $tag .= ">"; |
||
| 425 | |||
| 426 | //set a Buffer for the segSource Source tag |
||
| 427 | if ( $this->bufferIsActive ) { // we are opening a critical CDATA section |
||
| 428 | //these are NOT source/seg-source/value empty tags, THERE IS A CONTENT, write it in buffer |
||
| 429 | $this->CDATABuffer .= $tag; |
||
| 430 | } else { |
||
| 431 | $this->postProcAndFlush( $this->outputFP, $tag ); |
||
| 432 | } |
||
| 433 | |||
| 434 | } |
||
| 435 | |||
| 436 | /** |
||
| 437 | * A trans-unit can contain a list of segments because of mrk tags |
||
| 438 | * Copy the segment's list for this trans-unit in a different structure |
||
| 439 | * |
||
| 440 | * @return void |
||
| 441 | */ |
||
| 442 | protected function setLastTransUnitSegments() { |
||
| 443 | |||
| 444 | /* |
||
| 445 | * At the end of every cycle the segment grouping information is lost: unset( 'matecat|' . $this->currentId ) |
||
| 446 | * |
||
| 447 | * We need to take the info about the last segment parsed |
||
| 448 | * ( normally more than 1 db row because of mrk tags ) |
||
| 449 | * |
||
| 450 | * So, copy the current segment data group into another structure to keep the last segment |
||
| 451 | * for the next tagOpen ( possible sdl:seg-defs ) |
||
| 452 | * |
||
| 453 | */ |
||
| 454 | $this->lastTransUnit = []; |
||
| 455 | |||
| 456 | if ( !isset( $this->transUnits[ $this->currentTransUnitId ] ) ) { |
||
| 457 | return; |
||
| 458 | } |
||
| 459 | |||
| 460 | $listOfSegmentsIds = $this->transUnits[ $this->currentTransUnitId ]; |
||
| 461 | $last_value = null; |
||
| 462 | $segmentsCount = count( $listOfSegmentsIds ); |
||
| 463 | for ( $i = 0; $i < $segmentsCount; $i++ ) { |
||
| 464 | $id = $listOfSegmentsIds[ $i ]; |
||
| 465 | if ( isset( $this->segments[ $id ] ) && ( $i == 0 || $last_value + 1 == $listOfSegmentsIds[ $i ] ) ) { |
||
| 466 | $last_value = $listOfSegmentsIds[ $i ]; |
||
| 467 | $this->lastTransUnit[] = $this->segments[ $id ]; |
||
| 468 | } |
||
| 469 | } |
||
| 470 | |||
| 471 | } |
||
| 472 | |||
| 473 | /** |
||
| 474 | * @return array |
||
| 475 | */ |
||
| 476 | protected function getCurrentSegment(): array { |
||
| 482 | } |
||
| 483 | |||
| 484 | } |
Loading history...
If you suppress an error, we recommend checking for the error condition explicitly: