matecat / xliff-parser

src/XliffReplacer/AbstractXliffReplacer.php

Spacing +117 added lines, -117 removed lines

@@ -7,29 +7,29 @@ 
 
 abstract class AbstractXliffReplacer {
     protected $originalFP;
-    protected $outputFP;                  // output stream pointer
+    protected $outputFP; // output stream pointer
 
-    protected string $tuTagName;                 // <trans-unit> (forXliff v 1.*) or <unit> (forXliff v 2.*)
-    protected bool   $inTU                  = false;  // flag to check whether we are in a <trans-unit>
-    protected bool   $inTarget              = false;  // flag to check whether we are in a <target>, to ignore everything
-    protected bool   $inAltTrans            = false;  // flag to check whether we are in an <alt-trans> (xliff 1.2) or <mtc:matches> (xliff 2.0)
+    protected string $tuTagName; // <trans-unit> (forXliff v 1.*) or <unit> (forXliff v 2.*)
+    protected bool   $inTU                  = false; // flag to check whether we are in a <trans-unit>
+    protected bool   $inTarget              = false; // flag to check whether we are in a <target>, to ignore everything
+    protected bool   $inAltTrans            = false; // flag to check whether we are in an <alt-trans> (xliff 1.2) or <mtc:matches> (xliff 2.0)
     protected string $alternativeMatchesTag = ""; // polymorphic tag name for xliff 1.2 and 2.0
-    protected bool   $isEmpty               = false;  // flag to check whether we are in an empty tag (<tag/>)
-    protected bool   $targetWasWritten      = false;  // flag to check is <target> was written in the current unit
-    protected string $CDATABuffer           = "";       // buffer for special tag
-    protected string $namespace             = "";       // Custom namespace
-    protected bool   $bufferIsActive        = false;    // flag for buffeting
+    protected bool   $isEmpty               = false; // flag to check whether we are in an empty tag (<tag/>)
+    protected bool   $targetWasWritten      = false; // flag to check is <target> was written in the current unit
+    protected string $CDATABuffer           = ""; // buffer for special tag
+    protected string $namespace             = ""; // Custom namespace
+    protected bool   $bufferIsActive        = false; // flag for buffeting
 
-    protected int $offset = 0;         // offset for SAX pointer
+    protected int $offset = 0; // offset for SAX pointer
 
-    protected string  $currentBuffer;             // the current piece of text it's been parsed
-    protected int     $len;                       // length of the currentBuffer
-    protected array   $segments;                  // array of translations
+    protected string  $currentBuffer; // the current piece of text it's been parsed
+    protected int     $len; // length of the currentBuffer
+    protected array   $segments; // array of translations
     protected array   $lastTransUnit                  = [];
     protected int     $segmentInUnitPosition          = 0;
-    protected ?string $currentTransUnitId             = null;        // id of current <trans-unit>
+    protected ?string $currentTransUnitId             = null; // id of current <trans-unit>
     protected ?string $currentTransUnitIsTranslatable = null; // 'translate' attribute of current <trans-unit>
-    protected bool    $hasWrittenCounts               = false;  // check if <unit> already wrote segment counts (forXliff v 2.*)
+    protected bool    $hasWrittenCounts               = false; // check if <unit> already wrote segment counts (forXliff v 2.*)
     protected string  $targetLang;
     protected bool    $sourceInTarget                 = false;
 
@@ -79,8 +79,8 @@ 
             XliffReplacerCallbackInterface $callback = null
     ) {
         self::$INTERNAL_TAG_PLACEHOLDER = $this->getInternalTagPlaceholder();
-        $this->createOutputFileIfDoesNotExist( $outputFilePath );
-        $this->setFileDescriptors( $originalXliffPath, $outputFilePath );
+        $this->createOutputFileIfDoesNotExist($outputFilePath);
+        $this->setFileDescriptors($originalXliffPath, $outputFilePath);
         $this->xliffVersion   = $xliffVersion;
         $this->segments       = $segments;
         $this->targetLang     = $trgLang;
@@ -91,17 +91,17 @@ 
     }
 
     public function replaceTranslation() {
-        fwrite( $this->outputFP, '<?xml version="1.0" encoding="UTF-8"?>' );
+        fwrite($this->outputFP, '<?xml version="1.0" encoding="UTF-8"?>');
 
         //create Sax parser
         $xmlParser = $this->initSaxParser();
 
-        while ( $this->currentBuffer = fread( $this->originalFP, 4096 ) ) {
+        while ($this->currentBuffer = fread($this->originalFP, 4096)) {
             /*
                preprocess file
              */
             // obfuscate entities because sax automatically does html_entity_decode
-            $temporary_check_buffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer );
+            $temporary_check_buffer = preg_replace("/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer);
 
             //avoid cutting entities in half:
             //the last fread could have truncated an entity (say, '&lt;' in '&l'), thus invalidating the escaping
@@ -109,27 +109,27 @@ 
             // 9 is the max length of an entity. So, suppose that the & is at the end of buffer,
             // add 9 Bytes and substitute the entities, if the & is present, and it is not at the end
             //it can't be an entity, exit the loop
-            while ( true ) {
-                $_ampPos = strpos( $temporary_check_buffer, '&' );
+            while (true) {
+                $_ampPos = strpos($temporary_check_buffer, '&');
 
                 //check for real entity or escape it to safely exit from the loop!!!
-                if ( $_ampPos === false || strlen( substr( $temporary_check_buffer, $_ampPos ) ) > 9 ) {
+                if ($_ampPos === false || strlen(substr($temporary_check_buffer, $_ampPos)) > 9) {
                     break;
                 }
 
                 //if an entity is still present, fetch some more and repeat the escaping
-                $this->currentBuffer    .= fread( $this->originalFP, 9 );
-                $temporary_check_buffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer );
+                $this->currentBuffer .= fread($this->originalFP, 9);
+                $temporary_check_buffer = preg_replace("/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer);
             }
 
             //free stuff outside the loop
-            unset( $temporary_check_buffer );
+            unset($temporary_check_buffer);
 
-            $this->currentBuffer = preg_replace( "/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer );
-            $this->currentBuffer = str_replace( "&", self::$INTERNAL_TAG_PLACEHOLDER . 'amp' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer );
+            $this->currentBuffer = preg_replace("/&(.*?);/", self::$INTERNAL_TAG_PLACEHOLDER . '$1' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer);
+            $this->currentBuffer = str_replace("&", self::$INTERNAL_TAG_PLACEHOLDER . 'amp' . self::$INTERNAL_TAG_PLACEHOLDER, $this->currentBuffer);
 
             //get length of chunk
-            $this->len = strlen( $this->currentBuffer );
+            $this->len = strlen($this->currentBuffer);
 
             /*
             * Get the accumulated this->offset in the document:
@@ -139,12 +139,12 @@ 
             $this->offset += $this->len;
 
             //parse chunk of text
-            $this->runParser( $xmlParser );
+            $this->runParser($xmlParser);
 
         }
 
         // close Sax parser
-        $this->closeSaxParser( $xmlParser );
+        $this->closeSaxParser($xmlParser);
 
     }
 
@@ -153,15 +153,15 @@ 
      *
      * @return void
      */
-    protected function runParser( $xmlParser ) {
+    protected function runParser($xmlParser) {
         //parse chunk of text
-        if ( !xml_parse( $xmlParser, $this->currentBuffer, feof( $this->originalFP ) ) ) {
+        if (!xml_parse($xmlParser, $this->currentBuffer, feof($this->originalFP))) {
             //if unable, raise an exception
-            throw new RuntimeException( sprintf(
+            throw new RuntimeException(sprintf(
                     "XML error: %s at line %d",
-                    xml_error_string( xml_get_error_code( $xmlParser ) ),
-                    xml_get_current_line_number( $xmlParser )
-            ) );
+                    xml_error_string(xml_get_error_code($xmlParser)),
+                    xml_get_current_line_number($xmlParser)
+            ));
         }
     }
 
@@ -170,18 +170,18 @@ 
      *
      * @return string
      */
-    protected function getLastCharacter( $parser ): string {
+    protected function getLastCharacter($parser): string {
 
         //this logic helps detecting empty tags
         //get current position of SAX pointer in all the stream of data is has read so far:
         //it points at the end of current tag
-        $idx = xml_get_current_byte_index( $parser );
+        $idx = xml_get_current_byte_index($parser);
 
         //check whether the bounds of current tag are entirely in current buffer or the end of the current tag
         //is outside current buffer (in the latter case, it's in next buffer to be read by the while loop);
         //this check is necessary because we may have truncated a tag in half with current read,
         //and the other half may be encountered in the next buffer it will be passed
-        return $this->currentBuffer[ $idx - $this->offset ] ?? $this->currentBuffer[ $this->len - 1 ];
+        return $this->currentBuffer[$idx - $this->offset] ?? $this->currentBuffer[$this->len - 1];
 
     }
 
@@ -192,19 +192,19 @@ 
         return "§" .
                 substr(
                         str_replace(
-                                [ '+', '/' ],
+                                ['+', '/'],
                                 '',
-                                base64_encode( openssl_random_pseudo_bytes( 10, $_crypto_strong ) )
+                                base64_encode(openssl_random_pseudo_bytes(10, $_crypto_strong))
                         ),
                         0,
                         4
                 );
     }
 
-    private function createOutputFileIfDoesNotExist( string $outputFilePath ) {
+    private function createOutputFileIfDoesNotExist(string $outputFilePath) {
         // create output file
-        if ( !file_exists( $outputFilePath ) ) {
-            touch( $outputFilePath );
+        if (!file_exists($outputFilePath)) {
+            touch($outputFilePath);
         }
     }
 
@@ -212,13 +212,13 @@ 
      * @param string $originalXliffPath
      * @param string $outputFilePath
      */
-    private function setFileDescriptors( string $originalXliffPath, string $outputFilePath ) {
-        $this->outputFP = fopen( $outputFilePath, 'w+' );
+    private function setFileDescriptors(string $originalXliffPath, string $outputFilePath) {
+        $this->outputFP = fopen($outputFilePath, 'w+');
 
         $streamArgs = null;
 
-        if ( !( $this->originalFP = fopen( $originalXliffPath, "r", false, stream_context_create( $streamArgs ) ) ) ) {
-            throw new RuntimeException( "could not open XML input" );
+        if (!($this->originalFP = fopen($originalXliffPath, "r", false, stream_context_create($streamArgs)))) {
+            throw new RuntimeException("could not open XML input");
         }
     }
 
@@ -228,12 +228,12 @@ 
     public function __destruct() {
         //this stream can be closed outside the class
         //to permit multiple concurrent downloads, so suppress warnings
-        if ( is_resource( $this->originalFP ) ) {
-            fclose( $this->originalFP );
+        if (is_resource($this->originalFP)) {
+            fclose($this->originalFP);
         }
 
-        if ( is_resource( $this->outputFP ) ) {
-            fclose( $this->outputFP );
+        if (is_resource($this->outputFP)) {
+            fclose($this->outputFP);
         }
 
     }
@@ -244,11 +244,11 @@ 
      * @return resource
      */
     protected function initSaxParser() {
-        $xmlSaxParser = xml_parser_create( 'UTF-8' );
-        xml_set_object( $xmlSaxParser, $this );
-        xml_parser_set_option( $xmlSaxParser, XML_OPTION_CASE_FOLDING, false );
-        xml_set_element_handler( $xmlSaxParser, 'tagOpen', 'tagClose' );
-        xml_set_character_data_handler( $xmlSaxParser, 'characterData' );
+        $xmlSaxParser = xml_parser_create('UTF-8');
+        xml_set_object($xmlSaxParser, $this);
+        xml_parser_set_option($xmlSaxParser, XML_OPTION_CASE_FOLDING, false);
+        xml_set_element_handler($xmlSaxParser, 'tagOpen', 'tagClose');
+        xml_set_character_data_handler($xmlSaxParser, 'characterData');
 
         return $xmlSaxParser;
     }
@@ -256,8 +256,8 @@ 
     /**
      * @param resource $xmlSaxParser
      */
-    protected function closeSaxParser( $xmlSaxParser ) {
-        xml_parser_free( $xmlSaxParser );
+    protected function closeSaxParser($xmlSaxParser) {
+        xml_parser_free($xmlSaxParser);
     }
 
     /**
@@ -267,7 +267,7 @@ 
      *
      * @return mixed
      */
-    abstract protected function tagOpen( $parser, string $name, array $attr );
+    abstract protected function tagOpen($parser, string $name, array $attr);
 
     /**
      * @param resource $parser
@@ -275,7 +275,7 @@ 
      *
      * @return mixed
      */
-    abstract protected function tagClose( $parser, string $name );
+    abstract protected function tagClose($parser, string $name);
 
     /**
      * @param resource $parser
@@ -283,11 +283,11 @@ 
      *
      * @return mixed
      */
-    protected function characterData( $parser, string $data ): void {
+    protected function characterData($parser, string $data): void {
         // don't write <target> data
-        if ( !$this->inTarget && !$this->bufferIsActive ) {
-            $this->postProcAndFlush( $this->outputFP, $data );
-        } elseif ( $this->bufferIsActive ) {
+        if (!$this->inTarget && !$this->bufferIsActive) {
+            $this->postProcAndFlush($this->outputFP, $data);
+        } elseif ($this->bufferIsActive) {
             $this->CDATABuffer .= $data;
         }
     }
@@ -299,19 +299,19 @@ 
      * @param string   $data
      * @param bool     $treatAsCDATA
      */
-    protected function postProcAndFlush( $fp, string $data, bool $treatAsCDATA = false ) {
+    protected function postProcAndFlush($fp, string $data, bool $treatAsCDATA = false) {
         //postprocess string
-        $data = preg_replace( "/" . self::$INTERNAL_TAG_PLACEHOLDER . '(.*?)' . self::$INTERNAL_TAG_PLACEHOLDER . "/", '&$1;', $data );
-        $data = str_replace( '&nbsp;', ' ', $data );
-        if ( !$treatAsCDATA ) {
+        $data = preg_replace("/" . self::$INTERNAL_TAG_PLACEHOLDER . '(.*?)' . self::$INTERNAL_TAG_PLACEHOLDER . "/", '&$1;', $data);
+        $data = str_replace('&nbsp;', ' ', $data);
+        if (!$treatAsCDATA) {
             //unix2dos
-            $data = str_replace( "\r\n", "\r", $data );
-            $data = str_replace( "\n", "\r", $data );
-            $data = str_replace( "\r", "\r\n", $data );
+            $data = str_replace("\r\n", "\r", $data);
+            $data = str_replace("\n", "\r", $data);
+            $data = str_replace("\r", "\r\n", $data);
         }
 
         //flush to disk
-        fwrite( $fp, $data );
+        fwrite($fp, $data);
     }
 
     /**
@@ -320,19 +320,19 @@ 
      *
      * @return void
      */
-    protected function handleOpenUnit( string $name, array $attr ) {
+    protected function handleOpenUnit(string $name, array $attr) {
 
         // check if we are entering into a <trans-unit> (xliff v1.*) or <unit> (xliff v2.*)
-        if ( $this->tuTagName === $name ) {
+        if ($this->tuTagName === $name) {
             $this->inTU = true;
 
             // get id
             // trim to first 100 characters because this is the limit on Matecat's DB
-            $this->currentTransUnitId = substr( $attr[ 'id' ], 0, 100 );
+            $this->currentTransUnitId = substr($attr['id'], 0, 100);
 
             // `translate` attribute can be only yes or no
             // current 'translate' attribute of the current trans-unit
-            $this->currentTransUnitIsTranslatable = empty( $attr[ 'translate' ] ) ? 'yes' : $attr[ 'translate' ];
+            $this->currentTransUnitIsTranslatable = empty($attr['translate']) ? 'yes' : $attr['translate'];
 
             $this->setLastTransUnitSegments();
 
@@ -346,15 +346,15 @@ 
      *
      * @return string
      */
-    protected function handleOpenXliffTag( string $name, array $attr, string $tag ): string {
+    protected function handleOpenXliffTag(string $name, array $attr, string $tag): string {
 
         // Add MateCat specific namespace.
         // Add trgLang
-        if ( $name === 'xliff' ) {
-            if ( !array_key_exists( 'xmlns:' . $this->namespace, $attr ) ) {
+        if ($name === 'xliff') {
+            if (!array_key_exists('xmlns:' . $this->namespace, $attr)) {
                 $tag .= ' xmlns:' . $this->namespace . '="https://www.matecat.com" ';
             }
-            $tag = preg_replace( '/trgLang="(.*?)"/', 'trgLang="' . $this->targetLang . '"', $tag );
+            $tag = preg_replace('/trgLang="(.*?)"/', 'trgLang="' . $this->targetLang . '"', $tag);
         }
 
         return $tag;
@@ -366,11 +366,11 @@ 
      *
      * @return void
      */
-    protected function checkSetInTarget( string $name ) {
+    protected function checkSetInTarget(string $name) {
 
         // check if we are entering into a <target>
-        if ( 'target' == $name && !$this->inAltTrans ) {
-            if ( $this->currentTransUnitIsTranslatable === 'no' ) {
+        if ('target' == $name && !$this->inAltTrans) {
+            if ($this->currentTransUnitIsTranslatable === 'no') {
                 $this->inTarget = false;
             } else {
                 $this->inTarget = true;
@@ -379,12 +379,12 @@ 
 
     }
 
-    protected function trySetAltTrans( string $name ) {
+    protected function trySetAltTrans(string $name) {
         $this->inAltTrans = $this->inAltTrans || $this->alternativeMatchesTag == $name;
     }
 
-    protected function tryUnsetAltTrans( string $name ) {
-        if ( $this->alternativeMatchesTag == $name ) {
+    protected function tryUnsetAltTrans(string $name) {
+        if ($this->alternativeMatchesTag == $name) {
             $this->inAltTrans = false;
         }
     }
@@ -394,8 +394,8 @@ 
      *
      * @return void
      */
-    protected function setInBuffer( string $name ) {
-        if ( in_array( $name, $this->nodesToBuffer ) ) {
+    protected function setInBuffer(string $name) {
+        if (in_array($name, $this->nodesToBuffer)) {
             $this->bufferIsActive = true;
         }
     }
@@ -403,24 +403,24 @@ 
     /**
      * @param array $seg
      */
-    protected function updateSegmentCounts( array $seg = [] ) {
+    protected function updateSegmentCounts(array $seg = []) {
 
-        $raw_word_count = $seg[ 'raw_word_count' ];
-        $eq_word_count  = ( floor( $seg[ 'eq_word_count' ] * 100 ) / 100 );
+        $raw_word_count = $seg['raw_word_count'];
+        $eq_word_count  = (floor($seg['eq_word_count'] * 100) / 100);
 
-        $this->counts[ 'segments_count_array' ][ $seg[ 'sid' ] ] = [
+        $this->counts['segments_count_array'][$seg['sid']] = [
                 'raw_word_count' => $raw_word_count,
                 'eq_word_count'  => $eq_word_count,
         ];
 
-        $this->counts[ 'raw_word_count' ] += $raw_word_count;
-        $this->counts[ 'eq_word_count' ]  += $eq_word_count;
+        $this->counts['raw_word_count'] += $raw_word_count;
+        $this->counts['eq_word_count']  += $eq_word_count;
     }
 
     protected function resetCounts() {
-        $this->counts[ 'segments_count_array' ] = [];
-        $this->counts[ 'raw_word_count' ]       = 0;
-        $this->counts[ 'eq_word_count' ]        = 0;
+        $this->counts['segments_count_array'] = [];
+        $this->counts['raw_word_count']       = 0;
+        $this->counts['eq_word_count']        = 0;
     }
 
     /**
@@ -429,16 +429,16 @@ 
      *
      * @return void
      */
-    protected function checkForSelfClosedTagAndFlush( $parser, string $tag ) {
+    protected function checkForSelfClosedTagAndFlush($parser, string $tag) {
 
-        $lastChar = $this->getLastCharacter( $parser );
+        $lastChar = $this->getLastCharacter($parser);
 
         //trim last space
-        $tag = rtrim( $tag );
+        $tag = rtrim($tag);
 
         //detect empty tag
         $this->isEmpty = $lastChar == '/';
-        if ( $this->isEmpty ) {
+        if ($this->isEmpty) {
             $tag .= $lastChar;
         }
 
@@ -446,11 +446,11 @@ 
         $tag .= ">";
 
         //set a Buffer for the segSource Source tag
-        if ( $this->bufferIsActive ) { // we are opening a critical CDATA section
+        if ($this->bufferIsActive) { // we are opening a critical CDATA section
             //these are NOT source/seg-source/value empty tags, THERE IS A CONTENT, write it in buffer
             $this->CDATABuffer .= $tag;
         } else {
-            $this->postProcAndFlush( $this->outputFP, $tag );
+            $this->postProcAndFlush($this->outputFP, $tag);
         }
 
     }
@@ -475,18 +475,18 @@ 
          */
         $this->lastTransUnit = [];
 
-        if ( !isset( $this->transUnits[ $this->currentTransUnitId ] ) ) {
+        if (!isset($this->transUnits[$this->currentTransUnitId])) {
             return;
         }
 
-        $listOfSegmentsIds = $this->transUnits[ $this->currentTransUnitId ];
+        $listOfSegmentsIds = $this->transUnits[$this->currentTransUnitId];
         $last_value        = null;
-        $segmentsCount     = count( $listOfSegmentsIds );
-        for ( $i = 0; $i < $segmentsCount; $i++ ) {
-            $id = $listOfSegmentsIds[ $i ];
-            if ( isset( $this->segments[ $id ] ) && ( $i == 0 || $last_value + 1 == $listOfSegmentsIds[ $i ] ) ) {
-                $last_value            = $listOfSegmentsIds[ $i ];
-                $this->lastTransUnit[] = $this->segments[ $id ];
+        $segmentsCount     = count($listOfSegmentsIds);
+        for ($i = 0; $i < $segmentsCount; $i++) {
+            $id = $listOfSegmentsIds[$i];
+            if (isset($this->segments[$id]) && ($i == 0 || $last_value + 1 == $listOfSegmentsIds[$i])) {
+                $last_value            = $listOfSegmentsIds[$i];
+                $this->lastTransUnit[] = $this->segments[$id];
             }
         }
     }
@@ -497,10 +497,10 @@ 
     protected function getCurrentSegment(): array {
         if (
             $this->currentTransUnitIsTranslatable !== 'no' &&
-            isset( $this->transUnits[ $this->currentTransUnitId ] ) &&
-            isset($this->segments[ $this->segmentInUnitPosition ])
+            isset($this->transUnits[$this->currentTransUnitId]) &&
+            isset($this->segments[$this->segmentInUnitPosition])
         ) {
-            return $this->segments[ $this->segmentInUnitPosition ];
+            return $this->segments[$this->segmentInUnitPosition];
         }
 
         return [];