Keep space between marks

src/XliffReplacer/AbstractXliffReplacer.php

@@ -54,6 +54,8 @@ abstract class AbstractXliffReplacer {
             'eq_word_count'  => 0,
     ];
 
+    protected $mrkTagsMap = [];
+
     /**
      * AbstractXliffReplacer constructor.
      *
@@ -296,13 +298,46 @@ protected function characterData( $parser, string $data ): void {
      * postprocess escaped data and write to disk
      *
      * @param resource $fp
-     * @param string   $data
-     * @param bool     $treatAsCDATA
+     * @param string $data
+     * @param bool $treatAsCDATA
+     * @param bool $parseMarks
      */
-    protected function postProcAndFlush( $fp, string $data, bool $treatAsCDATA = false ) {
+    protected function postProcAndFlush($fp, string $data, bool $treatAsCDATA = false, $parseMarks = false ) {
         //postprocess string
         $data = preg_replace( "/" . self::$INTERNAL_TAG_PLACEHOLDER . '(.*?)' . self::$INTERNAL_TAG_PLACEHOLDER . "/", '&$1;', $data );
         $data = str_replace( ' ', ' ', $data );
+
+        // extract <mrk> map only for <seg-source> tag
+        if($parseMarks){
+            // check if there are spaces between <mrk> tags
+            preg_match_all('/<mrk \b[^>]*>(.*?)<\/mrk>(\s+)/', $data, $spacesBetweenMrkCheck);
+
+            if(!empty($spacesBetweenMrkCheck[0])){
+
+                // $spacesBetweenMrkCheck[0] // holds the complete tags
+                // $spacesBetweenMrkCheck[1] // holds the text
+                // $spacesBetweenMrkCheck[2] // holds the spaces
+
+                foreach ($spacesBetweenMrkCheck[0] as $index => $mrk){
+
+                    if($this instanceof Xliff20){
+                        preg_match('/id="(\d+)"/', $mrk, $markMatch);
+                    } else {
+                        preg_match('/mid="(\d+)"/', $mrk, $markMatch);
+                    }
+
+                    if(isset($markMatch[1])){
+
+                        if(!isset($this->mrkTagsMap[$this->currentTransUnitId])){
+                            $this->mrkTagsMap[$this->currentTransUnitId] = [];
+                        }
+
+                        $this->mrkTagsMap[$this->currentTransUnitId][$markMatch[1]] = $spacesBetweenMrkCheck[2][$index];
+                    }
+                }
+            }
+        }
+
         if ( !$treatAsCDATA ) {
             //unix2dos
             $data = str_replace( "\r\n", "\r", $data );

src/XliffReplacer/Xliff12.php

@@ -142,7 +142,7 @@ protected function tagClose( $parser, string $name ) {
                 $this->CDATABuffer    = "";
 
                 //flush to the pointer
-                $this->postProcAndFlush( $this->outputFP, $tag );
+                $this->postProcAndFlush( $this->outputFP, $tag, false, $name === 'seg-source' );
 
             } elseif ( $name === $this->tuTagName ) {
 
@@ -179,7 +179,7 @@ protected function tagClose( $parser, string $name ) {
             $this->CDATABuffer    = "";
 
             //flush to the pointer
-            $this->postProcAndFlush( $this->outputFP, $tag );
+            $this->postProcAndFlush( $this->outputFP, $tag, false, $name === 'seg-source'  );
 
         } else {
             //ok, nothing to be done; reset flag for next coming tag
@@ -228,10 +228,23 @@ protected function prepareTranslation( array $seg, string $transUnitTranslation
         return $transUnitTranslation;
     }
 
+    /**
+     * @param array $seg
+     * @param string $translation
+     * @return string
+     */
     protected function rebuildMarks( array $seg, string $translation ): string {
 
         if ( $seg[ 'mrk_id' ] !== null && $seg[ 'mrk_id' ] != '' ) {
             $translation = "<mrk mid=\"" . $seg[ 'mrk_id' ] . "\" mtype=\"seg\">" . $seg[ 'mrk_prev_tags' ] . $translation . $seg[ 'mrk_succ_tags' ] . "</mrk>";
+
+            // check if there is a trailing space in the map of this trans unit
+            if(
+                isset($this->mrkTagsMap[$this->currentTransUnitId]) and
+                isset($this->mrkTagsMap[$this->currentTransUnitId][$seg[ 'mrk_id' ]])
+            ){
+                $translation .= $this->mrkTagsMap[$this->currentTransUnitId][$seg[ 'mrk_id' ]];
+            }
         }
 
         return $translation;

src/XliffReplacer/Xliff20.php

@@ -233,7 +233,7 @@ protected function tagClose( $parser, string $name ) {
                 $this->CDATABuffer = "";
 
                 //flush to the pointer
-                $this->postProcAndFlush( $this->outputFP, $tag );
+                $this->postProcAndFlush( $this->outputFP, $tag, false, $name === 'source' );
 
             } elseif ( 'segment' === $name ) {
 
@@ -278,7 +278,7 @@ protected function tagClose( $parser, string $name ) {
             $this->CDATABuffer    = "";
 
             //flush to the pointer
-            $this->postProcAndFlush( $this->outputFP, $tag );
+            $this->postProcAndFlush( $this->outputFP, $tag, false, $name === 'source' );
 
         } else {
             //ok, nothing to be done; reset flag for next coming tag

tests/XliffReplacerTest.php

@@ -1210,6 +1210,13 @@ public function should_replace_12_units_with_empty_segments_with_the_correct_sta
         $this->assertEquals( $status, $output[ 'files' ][ 1 ][ 'trans-units' ][ 3 ][ 'target' ][ 'attr' ][ 'state' ] );
     }
 
+    /**
+     * @test
+     */
+    public function should_replace_12_with_mrk_space() {
+
+    }
+
     /**
      * @test
      */
@@ -1218,14 +1225,68 @@ public function should_replace_12_units_with_entities() {
         $data = $this->getData( [
             [
                 'sid' => '1',
-                'segment' => 'Hello'' ',
-                'internal_id' => '2973331',
+                'segment' => 'No more digging. If it’s top of mind, you’ll find it at the top of your inbox.',
+                'internal_id' => '2975931',
                 'mrk_id' => '0',
                 'prev_tags' => '',
                 'succ_tags' => '',
                 'mrk_prev_tags' => NULL,
                 'mrk_succ_tags' => NULL,
-                'translation' => 'Ciao'' ',
+                'translation' => 'This is the translation',
+                'status' => 'APPROVED',
+                'error' => '',
+                'eq_word_count' => '1.34',
+                'raw_word_count' => '2.00',
+                'source_page' => NULL,
+                'r2' => NULL,
+                'data_ref_map' => NULL,
+            ],
+            [
+                'sid' => '1',
+                'segment' => 'No more digging. If it’s top of mind, you’ll find it at the top of your inbox.',
+                'internal_id' => '2975931',
+                'mrk_id' => '1',
+                'prev_tags' => '',
+                'succ_tags' => '',
+                'mrk_prev_tags' => NULL,
+                'mrk_succ_tags' => NULL,
+                'translation' => 'This is the second part of the translation',
+                'status' => 'APPROVED',
+                'error' => '',
+                'eq_word_count' => '1.34',
+                'raw_word_count' => '2.00',
+                'source_page' => NULL,
+                'r2' => NULL,
+                'data_ref_map' => NULL,
+            ],
+            [
+                'sid' => '2',
+                'segment' => 'Your messages are all here with a new look and feel. See how your mailbox uses AI to save you time.',
+                'internal_id' => '2976344',
+                'mrk_id' => '0',
+                'prev_tags' => '',
+                'succ_tags' => '',
+                'mrk_prev_tags' => NULL,
+                'mrk_succ_tags' => NULL,
+                'translation' => 'This is the translation',
+                'status' => 'APPROVED',
+                'error' => '',
+                'eq_word_count' => '1.34',
+                'raw_word_count' => '2.00',
+                'source_page' => NULL,
+                'r2' => NULL,
+                'data_ref_map' => NULL,
+            ],
+            [
+                'sid' => '2',
+                'segment' => 'Your messages are all here with a new look and feel. See how your mailbox uses AI to save you time.',
+                'internal_id' => '2976344',
+                'mrk_id' => '1',
+                'prev_tags' => '',
+                'succ_tags' => '',
+                'mrk_prev_tags' => NULL,
+                'mrk_succ_tags' => NULL,
+                'translation' => 'This is the second part of the translation',
                 'status' => 'APPROVED',
                 'error' => '',
                 'eq_word_count' => '1.34',
@@ -1236,13 +1297,13 @@ public function should_replace_12_units_with_entities() {
             ],
         ] );
 
-        $inputFile  = __DIR__ . '/../tests/files/with-entities.xliff';
-        $outputFile = __DIR__ . '/../tests/files/output/with-entities.xliff';
+        $inputFile  = __DIR__ . '/../tests/files/mrk-with-space.xliff';
+        $outputFile = __DIR__ . '/../tests/files/output/mrk-with-space.xliff';
 
         ( new XliffParser() )->replaceTranslation( $inputFile, $data[ 'data' ], $data[ 'transUnits' ], 'it-it', $outputFile, false );
         $output = ( new XliffParser() )->xliffToArray( file_get_contents( $outputFile ) );
 
-        $this->assertEquals( "<mrk mid=\"0\" mtype=\"seg\">Ciao''</mrk> ", $output[ 'files' ][ 1 ][ 'trans-units' ][ 1 ][ 'target' ][ 'raw-content' ] );
+        $this->assertEquals( "<mrk mid=\"0\" mtype=\"seg\">This is the translation</mrk> <mrk mid=\"1\" mtype=\"seg\">This is the second part of the translation</mrk>", $output[ 'files' ][ 1 ][ 'trans-units' ][ 1 ][ 'target' ][ 'raw-content' ] );
     }
 
     /**

tests/files/mrk-with-space.xliff

@@ -0,0 +1,39 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?><xliff xmlns="urn:oasis:names:tc:xliff:document:1.2" xmlns:okp="okapi-framework:xliff-extensions" xmlns:yahoo="http://localizr.yahooapis.com/v1/localizrBase.xsd" version="1.2">
+<file datatype="x-textunit_dto" original="unknown" source-language="en" target-language="fr-FR">
+<header>
+<count-group name="total">
+    <count count-type="total" unit="trans-unit">2</count>
+    <count count-type="total" unit="word">40</count>
+    <count count-type="total" unit="x-fuzzy-100">null</count>
+    <count count-type="total" unit="x-fuzzy-95">null</count>
+    <count count-type="total" unit="x-fuzzy-85">null</count>
+    <count count-type="total" unit="x-fuzzy-75">null</count>
+    <count count-type="total" unit="x-fuzzy-new">40</count>
+</count-group>
+</header>
+<body>
+  <trans-unit id="2975931" resname="AYrjpi" xml:space="preserve">
+    <source xml:lang="en">No more digging. If it’s top of mind, you’ll find it at the top of your inbox.</source>
+    <seg-source><mrk mid="0" mtype="seg">No more digging.</mrk> <mrk mid="1" mtype="seg">If it’s top of mind, you’ll find it at the top of your inbox.</mrk></seg-source>
+    <target state="translated" state-qualifier="x-validation" xml:lang="fr-FR"><mrk mid="0" mtype="seg">No more digging.</mrk> <mrk mid="1" mtype="seg">If it’s top of mind, you’ll find it at the top of your inbox.</mrk> </target>
+    <note annotates="target">Reported in https://ouryahoo.atlassian.net/browse/MDF-3156, missing a space before/after a period.</note><note>Description for onboarding ai features modal priority step</note><note annotates="source">No more digging. If it’s top of mind, you’ll find it at the top of your inbox.</note>
+
+
+    <context-group purpose="location">
+<context context-type="url"/>
+</context-group>
+  </trans-unit>
+  <trans-unit id="2976344" resname="bNtCPN" xml:space="preserve">
+    <source xml:lang="en">Your messages are all here with a new look and feel. See how your mailbox uses AI to save you time.</source>
+    <seg-source><mrk mid="0" mtype="seg">Your messages are all here with a new look and feel.</mrk> <mrk mid="1" mtype="seg">See how your mailbox uses AI to save you time.</mrk></seg-source>
+    <target state="translated" state-qualifier="x-validation" xml:lang="fr-FR"><mrk mid="0" mtype="seg">Your messages are all here with a new look and feel.</mrk> <mrk mid="1" mtype="seg">See how your mailbox uses AI to save you time.</mrk> </target>
+    <note annotates="target">Reported in https://ouryahoo.atlassian.net/browse/MDF-3156, missing a space before/after a period.</note><note>Onboarding AI features card body</note><note annotates="source">Your messages are all here with a new look and feel. See how your mailbox uses AI to save you time.</note>
+
+
+    <context-group purpose="location">
+<context context-type="url"/>
+</context-group>
+  </trans-unit>
+</body>
+</file>
+</xliff>