Fix single/double quote in strings (removed obsolete code)

matecat · Sep 21, 2023 · 453adbb · 453adbb
1 parent eafd676
commit 453adbb
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 9 deletions.
diff --git a/src/Filters/Html/HtmlParser.php b/src/Filters/Html/HtmlParser.php
@@ -123,10 +123,12 @@ public function transform( $segment ) {
             } elseif ( $state == static::STATE_HTML ) {
                 switch ( $char ) {
                     case '<':
-                        // ignore '<' if inside a quote
-                        if ( $in_quote_char ) {
-                            break;
-                        }
+                        // is not possible to have angle brackets inside a tag, this case can not happen
+                        // this code would ignore '>' if inside a quote, useless
+                        // for more info see https://www.w3.org/TR/xml/#charsets
+//                        if ( $in_quote_char ) {
+//                            break;
+//                        }
 
                         // if we found a second less than symbol the first one IS NOT a tag,
                         // treat the html_buffer as plain text and attach to the output
@@ -135,10 +137,12 @@ public function transform( $segment ) {
                         break;
 
                     case '>':
-                        // ignore '>' if inside a quote
-                        if ( $in_quote_char ) {
-                            break;
-                        }
+                        // is not possible to have angle brackets inside a tag, this case can not happen
+                        // this code would ignore '>' if inside a quote, useless
+                        // for more info see https://www.w3.org/TR/xml/#charsets
+//                        if ( $in_quote_char ) {
+//                            break;
+//                        }
 
                         if ( in_array( substr( $html_buffer, 0, 8 ), [ '<script ', '<style', '<script', '<style ' ] ) ) {
                             $html_buffer .= $char;

diff --git a/tests/MateCatSubFilteringTest.php b/tests/MateCatSubFilteringTest.php
@@ -12,7 +12,6 @@
 use Matecat\SubFiltering\MateCatFilter;
 use Matecat\SubFiltering\Tests\Mocks\Features\AirbnbFeature;
 use Matecat\SubFiltering\Tests\Mocks\FeatureSet;
-use Matecat\SubFiltering\Utils\CatUtils;
 use PHPUnit\Framework\TestCase;
 
 class MateCatSubFilteringTest extends TestCase {
@@ -44,6 +43,16 @@ public function testSimpleString() {
         $this->assertEquals( $segmentL1, $filter->fromLayer2ToLayer1( $tmpLayer2 ) );
     }
 
+    public function testHTMLStringWithApostrophe()
+    {
+        $filter = $this->getFilterInstance();
+
+        $segment   = "&lt;Value&gt; &lt;![CDATA[Visitez Singapour et détendez-vous sur l'île de Langkawi]]&gt; &lt;/Value&gt;";
+        $segmentL1 = $filter->fromLayer0ToLayer1( $segment );
+
+        $this->assertEquals( $segment, $filter->fromLayer1ToLayer0( $segmentL1 ) );
+    }
+
     /**
      * @throws Exception
      */