matecat / xliff-parser

src/XliffReplacer/SdlXliffSAXTranslationReplacer.php

Spacing +19 added lines, -19 removed lines

@@ -19,15 +19,15 @@ 
 
             // 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);
 
             // current 'translate' attribute of the current trans-unit
-            $this->currentTransUnitTranslate = isset($attr[ 'translate' ]) ? $attr[ 'translate' ] : 'yes';
+            $this->currentTransUnitTranslate = isset($attr['translate']) ? $attr['translate'] : 'yes';
         }
 
         // check if we are entering into a <target>
         if ('target' == $name) {
-            if($this->currentTransUnitTranslate === 'no'){
+            if ($this->currentTransUnitTranslate === 'no') {
                 $this->inTarget = false;
             } else {
                 $this->inTarget = true;
@@ -64,10 +64,10 @@ 
                 } elseif ('sdl:seg' == $name) {
 
                     // write the confidence level for this segment ( Translated, Draft, etc. )
-                    if (isset($this->segments[ 'matecat|' . $this->currentTransUnitId ]) && $_sdlStatus_confWritten == false) {
+                    if (isset($this->segments['matecat|' . $this->currentTransUnitId]) && $_sdlStatus_confWritten == false) {
 
                         // append definition attribute
-                        $tag .= $this->prepareTargetStatuses($this->lastTransUnit[ $this->markerPos ]);
+                        $tag .= $this->prepareTargetStatuses($this->lastTransUnit[$this->markerPos]);
 
                         //prepare for an eventual next cycle
                         $this->markerPos++;
@@ -94,15 +94,15 @@ 
             // is outside current buffer (in the latter case, it's in next buffer to be read by the while loop);
             // this check is necessary because we may have truncated a tag in half with current read,
             // and the other half may be encountered in the next buffer it will be passed
-            if (isset($this->currentBuffer[ $idx - $this->offset ])) {
+            if (isset($this->currentBuffer[$idx - $this->offset])) {
                 // if this tag entire lenght fitted in the buffer, the last char must be the last
                 // symbol before the '>'; if it's an empty tag, it is assumed that it's a '/'
                 $tmp_offset = $idx - $this->offset;
-                $lastChar   = $this->currentBuffer[ $idx - $this->offset ];
+                $lastChar   = $this->currentBuffer[$idx - $this->offset];
             } else {
                 //if it's out, simple use the last character of the chunk
                 $tmp_offset = $this->len - 1;
-                $lastChar   = $this->currentBuffer[ $this->len - 1 ];
+                $lastChar   = $this->currentBuffer[$this->len - 1];
             }
 
             // trim last space
@@ -136,7 +136,7 @@ 
                 } else {
                     //these are NOT source/seg-source/value empty tags, THERE IS A CONTENT, write it in buffer
                     $this->bufferIsActive = true;
-                    $this->CDATABuffer    .= $tag;
+                    $this->CDATABuffer .= $tag;
                 }
             } else {
                 $this->postProcAndFlush($this->outputFP, $tag);
@@ -156,11 +156,11 @@ 
     {
         $endTags = "";
 
-        $segment     = Strings::removeDangerousChars($seg [ 'segment' ]);
-        $translation = Strings::removeDangerousChars($seg [ 'translation' ]);
+        $segment     = Strings::removeDangerousChars($seg ['segment']);
+        $translation = Strings::removeDangerousChars($seg ['translation']);
         $dataRefMap  = (isset($seg['data_ref_map']) && $seg['data_ref_map'] !== null) ? Strings::jsonToArray($seg['data_ref_map']) : [];
 
-        if (is_null($seg [ 'translation' ]) || $seg [ 'translation' ] == '') {
+        if (is_null($seg ['translation']) || $seg ['translation'] == '') {
             $translation = $segment;
         } else {
             if ($this->callback) {
@@ -173,20 +173,20 @@ 
         // for Trados the trailing spaces after </mrk> are meaningful
         // so we trim the translation from Matecat DB and add them after </mrk>
         $trailingSpaces = '';
-        for ($s=0; $s < Strings::getTheNumberOfTrailingSpaces($translation); $s++){
+        for ($s = 0; $s < Strings::getTheNumberOfTrailingSpaces($translation); $s++) {
             $trailingSpaces .= ' ';
         }
 
-        if ($seg[ 'mrk_id' ] !== null && $seg[ 'mrk_id' ] != '') {
+        if ($seg['mrk_id'] !== null && $seg['mrk_id'] != '') {
             if ($this->targetLang === 'ja-JP') {
-                $seg[ 'mrk_succ_tags' ] = ltrim($seg[ 'mrk_succ_tags' ]);
+                $seg['mrk_succ_tags'] = ltrim($seg['mrk_succ_tags']);
             }
 
-            $translation = "<mrk mid=\"" . $seg[ 'mrk_id' ] . "\" mtype=\"seg\">" . $seg[ 'mrk_prev_tags' ] . rtrim($translation) . $seg[ 'mrk_succ_tags' ] . "</mrk>" . $trailingSpaces;
+            $translation = "<mrk mid=\"" . $seg['mrk_id'] . "\" mtype=\"seg\">" . $seg['mrk_prev_tags'] . rtrim($translation) . $seg['mrk_succ_tags'] . "</mrk>" . $trailingSpaces;
         }
 
         // we need to trim succ_tags here because we already added the trailing spaces after </mrk>
-        $transUnitTranslation .= $seg[ 'prev_tags' ] . $translation . $endTags . ltrim($seg[ 'succ_tags' ]);
+        $transUnitTranslation .= $seg['prev_tags'] . $translation . $endTags . ltrim($seg['succ_tags']);
 
         return $transUnitTranslation;
     }
@@ -206,7 +206,7 @@ 
             'REJECTED'   => 'RejectedTranslation',
         ];
 
-        return "conf=\"{$statusMap[ $segment[ 'status' ] ]}\" ";
+        return "conf=\"{$statusMap[$segment['status']]}\" ";
     }
 
     /**
@@ -218,7 +218,7 @@ 
      */
     protected function setTransUnitState($seg, $state_prop, $lastMrkState)
     {
-        return [ null, null ];
+        return [null, null];
     }
 
     /**

src/Utils/Emoji.php

Spacing +5 added lines, -5 removed lines

@@ -7402,7 +7402,7 @@
             '

src/XliffUtils/DataRefReplacer.php

Spacing +68 added lines, -68 removed lines

@@ -17,7 +17,7 @@ 
      *
      * @param array $map
      */
-    public function __construct( array $map)
+    public function __construct(array $map)
     {
         $this->map = $map;
     }
@@ -48,14 +48,14 @@ 
         $html = HtmlParser::parse($string);
 
         // 1. Replace <ph>|<sc>|<ec> tags
-        foreach ($html as $node){
+        foreach ($html as $node) {
             $string = $this->recursiveAddEquivTextToPhTag($node, $string);
         }
 
         // 2. Replace <pc> tags
         $toBeEscaped = Strings::isAnEscapedHTML($string);
 
-        if($this->stringContainsPcTags($string, $toBeEscaped)){
+        if ($this->stringContainsPcTags($string, $toBeEscaped)) {
 
             // replace self-closed <pc />
             $string = $this->replaceSelfClosedPcTags($string, $toBeEscaped);
@@ -84,10 +84,10 @@ 
             'dataRefEnd',
         ];
 
-        foreach ($dataRefTags as $tag){
-            preg_match('/ '.$tag.'=[\\\\"](.*?)[\\\\"]/', $string, $matches);
+        foreach ($dataRefTags as $tag) {
+            preg_match('/ ' . $tag . '=[\\\\"](.*?)[\\\\"]/', $string, $matches);
 
-            if(count($matches) > 0){
+            if (count($matches) > 0) {
                 return true;
             }
         }
@@ -102,7 +102,7 @@ 
     {
         $html = HtmlParser::parse($string);
 
-        foreach ($html as $node){
+        foreach ($html as $node) {
             $string = $this->recursiveCleanFromEquivText($node, $string);
         }
 
@@ -122,10 +122,10 @@ 
      *
      * @return string
      */
-    private function recursiveAddEquivTextToPhTag( $node, $string)
+    private function recursiveAddEquivTextToPhTag($node, $string)
     {
-        if($node->has_children){
-            foreach ($node->inner_html as $childNode){
+        if ($node->has_children) {
+            foreach ($node->inner_html as $childNode) {
                 $string = $this->recursiveAddEquivTextToPhTag($childNode, $string);
             }
         } else {
@@ -134,18 +134,18 @@ 
                     return $string;
                 }
 
-                $a = $node->node;  // complete match. Eg:  <ph id="source1" dataRef="source1"/>
-                $b = $node->attributes['dataRef'];   // map identifier. Eg: source1
+                $a = $node->node; // complete match. Eg:  <ph id="source1" dataRef="source1"/>
+                $b = $node->attributes['dataRef']; // map identifier. Eg: source1
 
 
                 // if isset a value in the map calculate base64 encoded value
                 // otherwise skip
-                if(!in_array($b, array_keys($this->map))){
+                if (!in_array($b, array_keys($this->map))) {
                     return $string;
                 }
 
                 // check if is null, in this case convert it to NULL string
-                if(is_null($this->map[$b])){
+                if (is_null($this->map[$b])) {
                     $this->map[$b] = 'NULL';
                 }
 
@@ -157,19 +157,19 @@ 
                 }
 
                 // if there is no id copy it from dataRef
-                $id = (!isset($node->attributes['id'])) ? ' id="'.$b.'" removeId="true"': '';
+                $id = (!isset($node->attributes['id'])) ? ' id="' . $b . '" removeId="true"' : '';
 
                 // introduce dataType for <ec>/<sc> tag handling
-                $dataType = ($this->isAEcOrScTag($node)) ? ' dataType="'.$node->tagname.'"' : '';
+                $dataType = ($this->isAEcOrScTag($node)) ? ' dataType="' . $node->tagname . '"' : '';
 
                 // replacement
-                $d = str_replace('/', $id.$dataType. ' equiv-text="base64:'.$base64EncodedValue.'"/', $a);
-                $a = str_replace(['<','>','&gt;', '&lt;'], '', $a);
-                $d = str_replace(['<','>','&gt;', '&lt;'], '', $d);
+                $d = str_replace('/', $id . $dataType . ' equiv-text="base64:' . $base64EncodedValue . '"/', $a);
+                $a = str_replace(['<', '>', '&gt;', '&lt;'], '', $a);
+                $d = str_replace(['<', '>', '&gt;', '&lt;'], '', $d);
 
                 // convert <ec>/<sc> into <ph>
-                if($this->isAEcOrScTag($node)){
-                    $d = 'ph'.substr($d, 2);
+                if ($this->isAEcOrScTag($node)) {
+                    $d = 'ph' . substr($d, 2);
                     $d = trim($d);
                 }
 
@@ -191,7 +191,7 @@ 
         $regex = ($toBeEscaped) ? '/&lt;pc (.*?)&gt;/iu' : '/<pc (.*?)>/iu';
         preg_match_all($regex, $string, $openingPcMatches);
 
-        return (isset($openingPcMatches[0]) && count($openingPcMatches[0])>0);
+        return (isset($openingPcMatches[0]) && count($openingPcMatches[0]) > 0);
     }
 
     /**
@@ -202,27 +202,27 @@ 
      */
     private function replaceSelfClosedPcTags($string, $toBeEscaped)
     {
-        if($toBeEscaped){
-            $string = str_replace(['&lt;','&gt;'],['<','>'],$string);
+        if ($toBeEscaped) {
+            $string = str_replace(['&lt;', '&gt;'], ['<', '>'], $string);
         }
 
         $regex = '/<pc[^>]+?\/>/iu';
         preg_match_all($regex, $string, $selfClosedPcMatches);
 
-        foreach ($selfClosedPcMatches[0] as $match){
+        foreach ($selfClosedPcMatches[0] as $match) {
 
             $html = HtmlParser::parse($match);
             $node = $html[0];
             $attributes = $node->attributes;
 
-            if(isset($attributes['dataRefStart']) && array_key_exists($node->attributes['dataRefStart'], $this->map)){
-                $replacement = '<ph id="'.$attributes['id'].'" dataType="pcSelf" originalData="'.base64_encode($match).'" dataRef="'.$attributes['dataRefStart'].'" equiv-text="base64:'.base64_encode($this->map[$node->attributes['dataRefStart']]).'"/>';
+            if (isset($attributes['dataRefStart']) && array_key_exists($node->attributes['dataRefStart'], $this->map)) {
+                $replacement = '<ph id="' . $attributes['id'] . '" dataType="pcSelf" originalData="' . base64_encode($match) . '" dataRef="' . $attributes['dataRefStart'] . '" equiv-text="base64:' . base64_encode($this->map[$node->attributes['dataRefStart']]) . '"/>';
                 $string = str_replace($match, $replacement, $string);
             }
         }
 
-        if($toBeEscaped){
-            $string = str_replace(['<','>'],['&lt;','&gt;'],$string);
+        if ($toBeEscaped) {
+            $string = str_replace(['<', '>'], ['&lt;', '&gt;'], $string);
         }
 
         return $string;
@@ -241,7 +241,7 @@ 
         $dataRefEndMap = [];
 
         foreach ($html as $index => $node) {
-            if ( $node->tagname === 'pc' ) {
+            if ($node->tagname === 'pc') {
                 $this->extractDataRefMapRecursively($node, $dataRefEndMap);
             }
         }
@@ -255,26 +255,26 @@ 
      * @param object $node
      * @param $dataRefEndMap
      */
-    private function extractDataRefMapRecursively( $node, &$dataRefEndMap)
+    private function extractDataRefMapRecursively($node, &$dataRefEndMap)
     {
-        if($this->nodeContainsNestedPcTags($node)) {
-            foreach ( $node->inner_html as $nestedNode ) {
+        if ($this->nodeContainsNestedPcTags($node)) {
+            foreach ($node->inner_html as $nestedNode) {
                 $this->extractDataRefMapRecursively($nestedNode, $dataRefEndMap);
             }
         }
 
         // EXCLUDE self closed <pc/>
-        if($node->tagname === 'pc' && $node->self_closed === false){
-            if(isset($node->attributes['dataRefEnd'])){
+        if ($node->tagname === 'pc' && $node->self_closed === false) {
+            if (isset($node->attributes['dataRefEnd'])) {
                 $dataRefEnd = $node->attributes['dataRefEnd'];
-            } elseif(isset($node->attributes['dataRefStart'])) {
+            } elseif (isset($node->attributes['dataRefStart'])) {
                 $dataRefEnd = $node->attributes['dataRefStart'];
             } else {
                 $dataRefEnd = null;
             }
 
             $dataRefEndMap[] = [
-                    'id' => isset($node->attributes['id'] ) ? $node->attributes['id'] : null,
+                    'id' => isset($node->attributes['id']) ? $node->attributes['id'] : null,
                     'dataRefEnd' => $dataRefEnd,
             ];
         }
@@ -288,12 +288,12 @@ 
      */
     private function recursiveCleanFromEquivText($node, $string)
     {
-        if($node->has_children){
-            foreach ($node->inner_html as $childNode){
+        if ($node->has_children) {
+            foreach ($node->inner_html as $childNode) {
                 $string = $this->recursiveCleanFromEquivText($childNode, $string);
             }
         } else {
-            if(isset($node->attributes['dataRef']) && array_key_exists($node->attributes['dataRef'], $this->map)){
+            if (isset($node->attributes['dataRef']) && array_key_exists($node->attributes['dataRef'], $this->map)) {
                 $cleaned = preg_replace('/ equiv-text="(.*?)"/', '', $node->node);
                 $string = str_replace($node->node, $cleaned, $string);
             }
@@ -315,29 +315,29 @@ 
         $regex = ($toBeEscaped) ? '/&lt;pc (.*?)&gt;/iu' : '/<pc (.*?)>/iu';
         preg_match_all($regex, $string, $openingPcMatches);
 
-        foreach ($openingPcMatches[0] as $index => $match){
+        foreach ($openingPcMatches[0] as $index => $match) {
             $attr = HtmlParser::getAttributes($openingPcMatches[1][$index]);
 
             // CASE 1 - Missing `dataRefStart`
-            if( isset($attr['dataRefEnd']) && !isset($attr['dataRefStart'])  ){
+            if (isset($attr['dataRefEnd']) && !isset($attr['dataRefStart'])) {
                 $attr['dataRefStart'] = $attr['dataRefEnd'];
             }
 
             // CASE 2 - Missing `dataRefEnd`
-            if( isset($attr['dataRefStart']) && !isset($attr['dataRefEnd'])  ){
+            if (isset($attr['dataRefStart']) && !isset($attr['dataRefEnd'])) {
                 $attr['dataRefEnd'] = $attr['dataRefStart'];
             }
 
-            if(isset($attr['dataRefStart'])){
+            if (isset($attr['dataRefStart'])) {
                 $startOriginalData = $match; // opening <pc>
                 $startValue = $this->map[$attr['dataRefStart']] ? $this->map[$attr['dataRefStart']] : 'NULL'; //handling null values in original data map
                 $base64EncodedStartValue = base64_encode($startValue);
                 $base64StartOriginalData = base64_encode($startOriginalData);
 
                 // conversion for opening <pc> tag
-                $openingPcConverted  = '<ph '. ((isset($attr['id'])) ? 'id="'.$attr['id'].'_1"' : '') .' dataType="pcStart" originalData="'.$base64StartOriginalData.'" dataRef="'
-                        .$attr['dataRefStart'].'" equiv-text="base64:'
-                        .$base64EncodedStartValue.'"/>';
+                $openingPcConverted = '<ph ' . ((isset($attr['id'])) ? 'id="' . $attr['id'] . '_1"' : '') . ' dataType="pcStart" originalData="' . $base64StartOriginalData . '" dataRef="'
+                        .$attr['dataRefStart'] . '" equiv-text="base64:'
+                        .$base64EncodedStartValue . '"/>';
 
                 $string = str_replace($startOriginalData, $openingPcConverted, $string);
             }
@@ -362,20 +362,20 @@ 
         preg_match_all($regex, $string, $closingPcMatches, PREG_OFFSET_CAPTURE);
         $delta = 0;
 
-        foreach ( $closingPcMatches[ 0 ] as $index => $match ) {
+        foreach ($closingPcMatches[0] as $index => $match) {
             $offset = $match[1];
             $length = strlen($match[0]);
             $attr = $dataRefEndMap[$index];
 
-            if(!empty($attr) && isset($attr['dataRefEnd'])){
+            if (!empty($attr) && isset($attr['dataRefEnd'])) {
                 $endOriginalData = $match[0]; // </pc>
                 $endValue = $this->map[$attr['dataRefEnd']] ? $this->map[$attr['dataRefEnd']] : 'NULL';
                 $base64EncodedEndValue = base64_encode($endValue);
                 $base64EndOriginalData = base64_encode($endOriginalData);
 
                 // conversion for closing <pc> tag
-                $closingPcConverted = '<ph '. ((isset($attr['id'])) ? 'id="'.$attr['id'].'_2"': '') .' dataType="pcEnd" originalData="'.$base64EndOriginalData.'" dataRef="'
-                        .$attr['dataRefEnd'].'" equiv-text="base64:' .$base64EncodedEndValue.'"/>';
+                $closingPcConverted = '<ph ' . ((isset($attr['id'])) ? 'id="' . $attr['id'] . '_2"' : '') . ' dataType="pcEnd" originalData="' . $base64EndOriginalData . '" dataRef="'
+                        .$attr['dataRefEnd'] . '" equiv-text="base64:' . $base64EncodedEndValue . '"/>';
 
                 $realOffset = ($delta === 0) ? $offset : ($offset + $delta);
 
@@ -394,12 +394,12 @@ 
      */
     private function nodeContainsNestedPcTags($node)
     {
-        if(!$node->has_children){
+        if (!$node->has_children) {
             return false;
         }
 
         foreach ($node->inner_html as $nestedNode) {
-            if($nestedNode->tagname === 'pc' && (isset($node->attributes['dataRefEnd']) || isset($node->attributes['dataRefStart']))){
+            if ($nestedNode->tagname === 'pc' && (isset($node->attributes['dataRefEnd']) || isset($node->attributes['dataRefStart']))) {
                 return true;
             }
         }
@@ -423,7 +423,7 @@ 
         $string = str_replace(' equiv-text=""', '', $string);
         $html = HtmlParser::parse($string);
 
-        foreach ($html as $node){
+        foreach ($html as $node) {
             $string = $this->recursiveRemoveOriginalData($node, $string);
         }
 
@@ -438,43 +438,43 @@ 
      */
     private function recursiveRemoveOriginalData($node, $string)
     {
-        if($node->has_children){
-            foreach ($node->inner_html as $childNode){
+        if ($node->has_children) {
+            foreach ($node->inner_html as $childNode) {
                 $string = $this->recursiveRemoveOriginalData($childNode, $string);
             }
         } else {
 
-            if(!isset($node->attributes['dataRef'])){
+            if (!isset($node->attributes['dataRef'])) {
                 return $string;
             }
 
-            $a = $node->node;                  // complete match. Eg:  <ph id="source1" dataRef="source1"/>
+            $a = $node->node; // complete match. Eg:  <ph id="source1" dataRef="source1"/>
             $b = $node->attributes['dataRef']; // map identifier. Eg: source1
-            $c = $node->terminator;            // terminator: Eg: >
+            $c = $node->terminator; // terminator: Eg: >
 
             // if isset a value in the map calculate base64 encoded value
             // or it is an empty string
             // otherwise skip
-            if(!in_array($b, array_keys($this->map))  ){
+            if (!in_array($b, array_keys($this->map))) {
                 return $string;
             }
 
             // check if is null, in this case convert it to NULL string
-            if(is_null($this->map[$b])){
+            if (is_null($this->map[$b])) {
                 $this->map[$b] = 'NULL';
             }
 
             // remove id?
-            $removeId = (isset($node->attributes['removeId']) && $node->attributes['removeId'] === "true") ? ' id="'.$b.'" removeId="true"' : '';
+            $removeId = (isset($node->attributes['removeId']) && $node->attributes['removeId'] === "true") ? ' id="' . $b . '" removeId="true"' : '';
 
             // grab dataType attribute for <ec>/<sc> tag handling
-            $dataType = ($this->wasAEcOrScTag($node)) ? ' dataType="'.$node->attributes['dataType'].'"' : '';
+            $dataType = ($this->wasAEcOrScTag($node)) ? ' dataType="' . $node->attributes['dataType'] . '"' : '';
 
-            $d = str_replace($removeId.$dataType.' equiv-text="base64:'.base64_encode($this->map[$b]).'"/'.$c, '/'.$c, $a);
+            $d = str_replace($removeId . $dataType . ' equiv-text="base64:' . base64_encode($this->map[$b]) . '"/' . $c, '/' . $c, $a);
 
             // replace original <ec>/<sc> tag
-            if($this->wasAEcOrScTag($node)){
-                $d = $node->attributes['dataType'].substr($d, 3);
+            if ($this->wasAEcOrScTag($node)) {
+                $d = $node->attributes['dataType'] . substr($d, 3);
                 $d = trim($d);
             }
 
@@ -487,7 +487,7 @@ 
             $string = str_replace($a, $d, $string);
 
             // restoring <pc/> self-closed here
-            if(Strings::contains('dataType="pcSelf"', $d)){
+            if (Strings::contains('dataType="pcSelf"', $d)) {
                 preg_match('/\s?originalData="(.*?)"\s?/', $d, $originalDataMatches);
 
                 if (isset($originalDataMatches[1])) {
@@ -546,6 +546,6 @@ 
      */
     private function wasAEcOrScTag($node)
     {
-        return (isset($node->attributes['dataType']) && ( $node->attributes['dataType'] === 'ec' || $node->attributes['dataType'] === 'sc'));
+        return (isset($node->attributes['dataType']) && ($node->attributes['dataType'] === 'ec' || $node->attributes['dataType'] === 'sc'));
     }
 }