Merge pull request #8 from damianz5/content_separation_for_translation

Fixed tags separation (ezembed) for Encoder
ezsystems · Aug 16, 2018 · c2010df · c2010df
2 parents ed98546 + a1b6a04
commit c2010df
Show file tree

Hide file tree

Showing 9 changed files with 374 additions and 19 deletions.
diff --git a/lib/Encoder.php b/lib/Encoder.php
@@ -125,7 +125,7 @@ public function __construct(ContentTypeService $contentTypeService, ConfigResolv
             'ez_platform_automated_translation'
         );
 
-        $this->nonTranslatableTags              = ['ezembed'] + $tags;
+        $this->nonTranslatableTags              = ['ezvalue', 'ezconfig', 'ezembed'] + $tags;
         $this->nonTranslatableCharactersHashMap = ["\n" => 'XXXEOLXXX'] + $chars;
         $this->nonValidAttributeTags            = ['title'] + $attributes;
     }
@@ -207,19 +207,14 @@ public function decode(string $xml): array
      *
      * @return string
      */
-    public function richTextEncode(RichTextValue $value): string
+    private function richTextEncode(RichTextValue $value): string
     {
         $xmlString = (string) $value;
         $xmlString = substr($xmlString, strpos($xmlString, '>') + 1);
-        $xmlString = str_replace(
-            array_keys($this->nonTranslatableCharactersHashMap),
-            array_values($this->nonTranslatableCharactersHashMap),
-            $xmlString
-        );
-
+        $xmlString = $this->encodeNonTranslatableCharacters($xmlString);
         foreach ($this->nonTranslatableTags as $tag) {
             $xmlString = preg_replace_callback(
-                '#<' . $tag . '(.[^>]*)>(.*)</' . $tag . '>#uim',
+                '#<' . $tag . '(.[^>]*)>(.*)</' . $tag . '>#Uuim',
                 function ($matches) use ($tag) {
                     $hash                        = sha1($matches[0]);
                     $this->placeHolderMap[$hash] = $matches[0];
@@ -231,7 +226,7 @@ function ($matches) use ($tag) {
         }
         foreach ($this->nonValidAttributeTags as $tag) {
             $xmlString = preg_replace_callback(
-                '#<' . $tag . '(.[^>]*)>#uim',
+                '#<' . $tag . '(.[^>]*)>#Uuim',
                 function ($matches) use ($tag) {
                     $hash                        = sha1($matches[0]);
                     $this->placeHolderMap[$hash] = $matches[0];
@@ -251,16 +246,12 @@ function ($matches) use ($tag) {
      *
      * @return string
      */
-    public function richTextDecode(string $value): string
+    private function richTextDecode(string $value): string
     {
-        $value = str_replace(
-            array_values($this->nonTranslatableCharactersHashMap),
-            array_keys($this->nonTranslatableCharactersHashMap),
-            $value
-        );
-        foreach ($this->nonTranslatableTags as $tag) {
+        $value = $this->decodeNonTranslatableCharacters($value);
+        foreach (array_reverse($this->nonTranslatableTags) as $tag) {
             $value = preg_replace_callback(
-                '#<' . $tag . '>(.*)</' . $tag . '>#uim',
+                '#<' . $tag . '>(.*)</' . $tag . '>#Uuim',
                 function ($matches) {
                     return $this->placeHolderMap[trim($matches[1])];
                 },
@@ -269,15 +260,44 @@ function ($matches) {
         }
         foreach ($this->nonValidAttributeTags as $tag) {
             $value = preg_replace_callback(
-                '#<fake' . $tag . '(.[^>]*)>#uim',
+                '#<fake' . $tag . '(.[^>]*)>#Uuim',
                 function ($matches) {
                     return $this->placeHolderMap[trim($matches[1])];
                 },
                 $value
             );
             $value = str_replace("</fake{$tag}>", "</{$tag}>", $value);
         }
+        $value = $this->decodeNonTranslatableCharacters($value);
 
         return $value;
     }
+
+    /**
+     * @param string $value
+     *
+     * @return string
+     */
+    private function encodeNonTranslatableCharacters(string $value): string
+    {
+        return str_replace(
+            array_keys($this->nonTranslatableCharactersHashMap),
+            array_values($this->nonTranslatableCharactersHashMap),
+            $value
+        );
+    }
+
+    /**
+     * @param string $value
+     *
+     * @return string
+     */
+    private function decodeNonTranslatableCharacters(string $value): string
+    {
+        return str_replace(
+            array_values($this->nonTranslatableCharactersHashMap),
+            array_keys($this->nonTranslatableCharactersHashMap),
+            $value
+        );
+    }
 }
diff --git a/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_expectedEncodeResult.xml b/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_expectedEncodeResult.xml
@@ -0,0 +1,2 @@
+<?xml version="1.0"?>
+<response><field_1_richtext type="eZ\Publish\Core\FieldType\RichText\Value"><fakecdata>XXXEOLXXX<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml" xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">XXXEOLXXX    <para>Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.</para>XXXEOLXXX</section>XXXEOLXXX</fakecdata></field_1_richtext><field_2_richtext type="eZ\Publish\Core\FieldType\RichText\Value"><fakecdata>XXXEOLXXX<section xmlns="http://ez.no/namespaces/ezpublish5/xhtml5/edit">XXXEOLXXX    <p>Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish. TheXXXEOLXXX    basic staples are native foods such as corn, beans and chili peppers.</p>XXXEOLXXX    <h2>Main Features</h2>XXXEOLXXX    <div data-ezelement="ezembed" data-href="ezcontent://192" data-ezview="embed" class="ez-embed-type-image" data-ezalign="right"><span data-ezelement="ezconfig"><span data-ezelement="ezvalue" data-ezvalue-key="size">large</span> </span></div>XXXEOLXXX    <p>The ability to cook well, called "saz&#xF3;n" (lit. seasoning) is considered to be a gift generally gained fromXXXEOLXXX    experience and a sense of commitment to the diners.</p>XXXEOLXXX    <div data-ezelement="ezembed" data-href="ezcontent://191" data-ezview="embed" class="ez-embed-type-image" data-ezalign="left"><span data-ezelement="ezconfig"><span data-ezelement="ezvalue" data-ezvalue-key="size">large</span> </span></div>XXXEOLXXX    <p>Mexican regional home cooking is completely different from the food served in most Mexican restaurants outsideXXXEOLXXX    Mexico, which is usually some variety of Tex-Mex.</p>XXXEOLXXX</section>XXXEOLXXX</fakecdata></field_2_richtext></response>
diff --git a/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_field1_richtext.xml b/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_field1_richtext.xml
@@ -0,0 +1,5 @@
+<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink"
+         xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml"
+         xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">
+    <para>Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.</para>
+</section>
diff --git a/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_field2_richtext.xml b/tests/fixtures/testEncodeTwoRichTextWithTwoEzembed_field2_richtext.xml
@@ -0,0 +1,15 @@
+<section xmlns="http://ez.no/namespaces/ezpublish5/xhtml5/edit">
+    <p>Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish. The
+    basic staples are native foods such as corn, beans and chili peppers.</p>
+    <h2>Main Features</h2>
+    <div data-ezelement="ezembed" data-href="ezcontent://192" data-ezview="embed" class="ez-embed-type-image"
+       data-ezalign="right"><span data-ezelement="ezconfig"><span data-ezelement="ezvalue"
+                                                                  data-ezvalue-key="size">large</span> </span></div>
+    <p>The ability to cook well, called "sazón" (lit. seasoning) is considered to be a gift generally gained from
+    experience and a sense of commitment to the diners.</p>
+    <div data-ezelement="ezembed" data-href="ezcontent://191" data-ezview="embed" class="ez-embed-type-image"
+       data-ezalign="left"><span data-ezelement="ezconfig"><span data-ezelement="ezvalue"
+                                                                 data-ezvalue-key="size">large</span> </span></div>
+    <p>Mexican regional home cooking is completely different from the food served in most Mexican restaurants outside
+    Mexico, which is usually some variety of Tex-Mex.</p>
+</section>
diff --git a/tests/fixtures/testEncodeTwoRichText_expectedEncodeResult.xml b/tests/fixtures/testEncodeTwoRichText_expectedEncodeResult.xml
@@ -0,0 +1,2 @@
+<?xml version="1.0"?>
+<response><field_1_richtext type="eZ\Publish\Core\FieldType\RichText\Value"><fakecdata>XXXEOLXXX<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml" xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">XXXEOLXXX    <para>XXXEOLXXX        Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.XXXEOLXXX    </para>XXXEOLXXX</section>XXXEOLXXX</fakecdata></field_1_richtext><field_2_richtext type="eZ\Publish\Core\FieldType\RichText\Value"><fakecdata>XXXEOLXXX<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml" xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">XXXEOLXXX    <para>XXXEOLXXX        Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.XXXEOLXXX    </para>XXXEOLXXX    <para>File 2</para>XXXEOLXXX</section>XXXEOLXXX</fakecdata></field_2_richtext></response>
diff --git a/tests/fixtures/testEncodeTwoRichText_field1_richtext.xml b/tests/fixtures/testEncodeTwoRichText_field1_richtext.xml
@@ -0,0 +1,7 @@
+<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink"
+         xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml"
+         xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">
+    <para>
+        Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.
+    </para>
+</section>
diff --git a/tests/fixtures/testEncodeTwoRichText_field2_richtext.xml b/tests/fixtures/testEncodeTwoRichText_field2_richtext.xml
@@ -0,0 +1,8 @@
+<section xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink"
+         xmlns:ezxhtml="http://ez.no/xmlns/ezpublish/docbook/xhtml"
+         xmlns:ezcustom="http://ez.no/xmlns/ezpublish/docbook/custom" version="5.0-variant ezpublish-1.0">
+    <para>
+        Mexican cuisine is primarily a fusion of indigenous Mesoamerican cooking with European, especially Spanish.
+    </para>
+    <para>File 2</para>
+</section>
diff --git a/tests/fixtures/testEncodeTwoTextline_expectedEncodeResult.xml b/tests/fixtures/testEncodeTwoTextline_expectedEncodeResult.xml
@@ -0,0 +1,2 @@
+<?xml version="1.0"?>
+<response><field_1_textline type="eZ\Publish\Core\FieldType\TextLine\Value">Some text 1</field_1_textline><field_2_textline type="eZ\Publish\Core\FieldType\TextLine\Value">Some text 2</field_2_textline></response>