Permit Some Backwards Compatibility

oleibman · oleibman · commit cf1fd2a11241 · 2024-05-09T17:35:13.000-07:00
It's debatable, but allow unescaped ampersand and unknown charset.
diff --git a/phpunit.xml.dist b/phpunit.xml.dist
@@ -7,10 +7,10 @@
     <exclude>
       <directory suffix=".php">./src/PhpWord/Shared/PCLZip</directory>
     </exclude>
-    <report>
+    <!--report>
       <clover outputFile="./build/logs/clover.xml"/>
       <html outputDirectory="./build/coverage"/>
-    </report>
+    </report-->
   </coverage>
   <php>
     <ini name="error_reporting" value="E_ALL"/>
diff --git a/src/PhpWord/Shared/Html.php b/src/PhpWord/Shared/Html.php
@@ -112,14 +112,14 @@ public static function addHtml($element, $html, $fullHTML = false, $preserveWhit
         $dom->preserveWhiteSpace = $preserveWhiteSpace;
 
         try {
-            $result = $dom->loadHTML($html);
+            $result = $dom->loadHTML($html, LIBXML_NOWARNING | LIBXML_NOERROR);
             $exceptionMessage = 'DOM loadHTML failed';
         } catch (Exception $e) {
             $result = false;
             $exceptionMessage = $e->getMessage();
         }
         if ($result === false) {
-            throw new Exception($exceptionMessage);
+            throw new Exception($exceptionMessage); // @codeCoverageIgnore
         }
         self::removeAnnoyingWhitespaceTextNodes($dom);
         static::$xpath = new DOMXPath($dom);
diff --git a/tests/PhpWordTests/Reader/Html/CharsetTest.php b/tests/PhpWordTests/Reader/Html/CharsetTest.php
@@ -57,7 +57,7 @@ public static function providerCharset(): array
             ['charset.UTF-16.bebom.html', 'À1'],
             ['charset.UTF-16.lebom.html', 'À1'],
             ['charset.gb18030.html', '电视机'],
-            ['charset.unknown.html', 'exception'],
+            'loadhtml gives its best shot' => ['charset.unknown.html', "Ã\u{80}1"],
         ];
     }
 }
diff --git a/tests/PhpWordTests/Shared/HtmlTest.php b/tests/PhpWordTests/Shared/HtmlTest.php
@@ -90,14 +90,14 @@ public function testAddHtml(): void
         $expectd .= "'Single Quoted Text'";
         $content .= '"Double Quoted Text"';
         $expectd .= '"Double Quoted Text"';
-        $content .= '&amp; Ampersand';
+        $content .= '& Ampersand';
         $expectd .= '& Ampersand';
-        $content .= '&lt;&gt;&ldquo;&lsquo;&rsquo;&laquo;&raquo;&lsaquo;&rsaquo;';
-        $expectd .= '<>“‘’«»‹›';
+        $content .= '&lt;&gt;&ldquo;&rdquo;&lsquo;&rsquo;&laquo;&raquo;&lsaquo;&rsaquo;';
+        $expectd .= '<>“”‘’«»‹›';
         $content .= '&amp;&bull;&deg;&hellip;&trade;&copy;&reg;&mdash;';
         $expectd .= '&•°…™©®—';
         $content .= '&ndash;&nbsp;&emsp;&ensp;&sup2;&sup3;&frac14;&frac12;&frac34;';
-        $expectd .= "–\u{a0}  ²³¼½¾";
+        $expectd .= "–\u{a0}\u{2003}\u{2002}²³¼½¾";
         Html::addHtml($section, $content);
         $elements = $section->getElements();
         foreach ($elements as $element) {

Original file line number	Diff line number	Diff line change
`@@ -57,7 +57,7 @@ public static function providerCharset(): array`
`57`	`57`	`['charset.UTF-16.bebom.html', 'À1'],`
`58`	`58`	`['charset.UTF-16.lebom.html', 'À1'],`
`59`	`59`	`['charset.gb18030.html', '电视机'],`
`60`		`- ['charset.unknown.html', 'exception'],`
	`60`	`+ 'loadhtml gives its best shot' => ['charset.unknown.html', "Ã\u{80}1"],`
`61`	`61`	`];`
`62`	`62`	`}`
`63`	`63`	`}`