Improve consistency between PHP and Libxml versions, add support for …

…UTF8 characters
AmaraLiving · Mar 13, 2017 · 35a7b00 · 35a7b00
1 parent 9ab2b79
commit 35a7b00
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 8 deletions.
diff --git a/composer.json b/composer.json
@@ -13,8 +13,7 @@
     }
   ],
   "require": {
-    "php": ">=5.5.0,<8.0",
-    "lib-libxml": ">=2.7.8"
+    "php": ">=5.5.0,<8.0"
   },
   "require-dev": {
     "phpunit/phpunit": "~4.8"

diff --git a/src/Amara/Varcon/HtmlCrawler.php b/src/Amara/Varcon/HtmlCrawler.php
@@ -39,10 +39,15 @@ public function setXpathExpressions(array $xpathExpressions)
      */
     public function crawlAndModify($content, callable $callable)
     {
-        $dom = new DOMDocument();
-        $dom->loadHTML($content, LIBXML_HTML_NOIMPLIED|LIBXML_HTML_NODEFDTD);
+        $document = new DOMDocument();
+        $document->loadHTML(mb_convert_encoding(
+            sprintf('<div>%s</div>', $content),
+            'HTML-ENTITIES',
+            'UTF-8'
+        ));
+        $this->stripDoctypeHtmlBodyAndHeadElements($document);
 
-        $xpath = new DOMXPath($dom);
+        $xpath = new DOMXPath($document);
 
         $textNodes = $xpath->query(implode('|', $this->xpathExpressions));
 
@@ -51,6 +56,26 @@ public function crawlAndModify($content, callable $callable)
             $textNode->nodeValue = $callable($textNode->nodeValue);
         }
 
-        return $dom->saveHTML($dom->documentElement);
+        return $document->saveHTML($document->documentElement);
+    }
+
+    /**
+     * This method is a short hack to avoid incompatibilities between different PHP and Libxml setups. It has the same
+     * effect as passing the LIBXML_HTML_NOIMPLIED and LIBXML_HTML_NODEFDTD flags to loadHtml's options
+     *
+     * @param DOMDocument $document
+     */
+    private function stripDoctypeHtmlBodyAndHeadElements(DOMDocument $document)
+    {
+        $container = $document->getElementsByTagName('div')->item(0);
+        $container = $container->parentNode->removeChild($container);
+
+        while ($document->firstChild) {
+            $document->removeChild($document->firstChild);
+        }
+
+        while ($container->firstChild) {
+            $document->appendChild($container->firstChild);
+        }
     }
 }
diff --git a/tests/Amara/Varcon/HtmlTranslatorTest.php b/tests/Amara/Varcon/HtmlTranslatorTest.php
@@ -13,9 +13,9 @@ public function testTranslatePreservesWhitespace()
         $htmlTranslator = new HtmlTranslator();
 
         $this->assertSame(
-            '<p>Colour <strong>pyjama</strong> test</p>',
+            '<p>Colour <strong>pyjama</strong> паралелепипед</p>', // Tests UTF8 characters as well
             $htmlTranslator->translate(
-                '<p>Color <strong>pajama</strong> test</p>',
+                '<p>Color <strong>pajama</strong> паралелепипед</p>',
                 'A',
                 'B'
             )