fix: use mb_encode_numericentity for #424 (#439)

* fix: use mb_encode_numericentity for #424 * test: add copyright-trademark test to Element * fix: use mb_encode_numericentity for #424 * test: add copyright-trademark test to Element
PhpGt · Jul 20, 2023 · 4679248 · 4679248
1 parent 5e759d5
commit 4679248
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 4 deletions.
diff --git a/phpmd.xml b/phpmd.xml
@@ -45,7 +45,7 @@
 
 	<rule ref="rulesets/naming.xml/ShortVariable">
 		<properties>
-			<property name="exceptions" value="ch,mh,sh,id,tr,td,i" />
+			<property name="exceptions" value="ch,mh,sh,id,tr,td,i,h1,h2,h3,h4,h5,h6" />
 		</properties>
 	</rule>
 </ruleset>
diff --git a/src/Element.php b/src/Element.php
@@ -162,10 +162,15 @@ protected function __prop_set_innerHTML(string $innerHTML):void {
 			$child->parentNode->removeChild($child);
 		}
 
-		$innerHTML = mb_convert_encoding(
+		if($innerHTML === "") {
+			return;
+		}
+
+		$conversionMap = [0x80, 0x10FFFF, 0, 0x1FFFFF];
+		$innerHTML = mb_encode_numericentity(
 			$innerHTML,
-			"HTML-ENTITIES",
-			"utf-8"
+			$conversionMap,
+			"UTF-8"
 		);
 
 		$tempDocument = new HTMLDocument();

diff --git a/test/phpunit/ElementTest.php b/test/phpunit/ElementTest.php
@@ -639,4 +639,15 @@ public function testRemoveAttributeNS():void {
 		self::assertFalse(
 			$sut->hasAttributeNS($ns, "foo"));
 	}
+
+	public function testConstruct_copyrightTrademark():void {
+		$copyright = "©";
+		$trademark = "™";
+		$text = "Copyright $copyright PHP.Gt, DOM$trademark";
+		$sut = new HTMLDocument("<!doctype html><h1>PHP.Gt/Dom Unit Test</h1>");
+		$h1 = $sut->querySelector("h1");
+		$h1->innerHTML = $text;
+		self::assertStringContainsString($copyright, $h1->innerHTML);
+		self::assertStringContainsString($trademark, $h1->innerHTML);
+	}
 }
diff --git a/test/phpunit/HTMLDocumentTest.php b/test/phpunit/HTMLDocumentTest.php
@@ -781,4 +781,18 @@ public function testEscapedCharacters_entireDom():void {
 			self::assertStringNotContainsString($needle, $domString);
 		}
 	}
+
+	public function testConstruct_copyrightTrademark():void {
+		$copyright = "©";
+		$trademark = "™";
+		$text = "Copyright $copyright PHP.Gt, DOM$trademark";
+		$sut = new HTMLDocument("<!doctype html><h1>$text</h1>");
+		$h1 = $sut->querySelector("h1");
+		self::assertStringContainsString($copyright, $h1->innerHTML);
+		self::assertStringContainsString($trademark, $h1->innerHTML);
+
+		$sut->documentElement->innerHTML = $text;
+		self::assertStringContainsString($copyright, $sut->documentElement->innerHTML);
+		self::assertStringContainsString($trademark, $sut->documentElement->innerHTML);
+	}
 }