Fix only first child being saved, improve UTF8 support

AmaraLiving · Mar 13, 2017 · bfff619 · bfff619
1 parent 12267b8
commit bfff619
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 10 deletions.
diff --git a/src/Amara/Varcon/HtmlCrawler.php b/src/Amara/Varcon/HtmlCrawler.php
@@ -48,15 +48,26 @@ public function crawlAndModify($content, callable $callable)
         $this->stripDoctypeHtmlBodyAndHeadElements($document);
 
         $xpath = new DOMXPath($document);
+        $nodes = $xpath->query(implode('|', $this->xpathExpressions));
 
-        $textNodes = $xpath->query(implode('|', $this->xpathExpressions));
+        $isContentModified = false;
 
-        /** @var DOMNode $textNode */
-        foreach ($textNodes as $textNode) {
-            $textNode->nodeValue = $callable($textNode->nodeValue);
+        /** @var DOMNode $node */
+        foreach ($nodes as $node) {
+            $modifiedNodeValue = $callable($node->nodeValue);
+
+            if ($node->nodeValue != $modifiedNodeValue) {
+                $isContentModified = true;
+
+                $node->nodeValue = $modifiedNodeValue;
+            }
+        }
+
+        if (false === $isContentModified) {
+            return $content;
         }
 
-        return $document->saveHTML($document->documentElement);
+        return trim($document->saveHTML());
     }
 
     /**

diff --git a/tests/Amara/Varcon/HtmlCrawlerTest.php b/tests/Amara/Varcon/HtmlCrawlerTest.php
@@ -46,6 +46,10 @@ public function provideCrawlAndModifyWithChangedXpathExpressions()
                 $html = '<span unsupported="Text">Text text text</span>',
                 $changedHtml = '<span unsupported="Changed">Text text text</span>',
             ],
+            [
+                $html = '<span>Unchanged</span>',
+                $changedHtml = '<span>Unchanged</span>',
+            ],
         ];
     }
 }
diff --git a/tests/Amara/Varcon/HtmlTranslatorTest.php b/tests/Amara/Varcon/HtmlTranslatorTest.php
@@ -13,7 +13,7 @@ public function testTranslatePreservesWhitespace()
         $htmlTranslator = new HtmlTranslator();
 
         $this->assertSame(
-            '<p>Colour <strong>pyjama</strong> паралелепипед</p>', // Tests UTF8 characters as well
+            '<p>Colour <strong>pyjama</strong> &#1087;&#1072;&#1088;&#1072;&#1083;&#1077;&#1083;&#1077;&#1087;&#1080;&#1087;&#1077;&#1076;</p>', // Tests UTF8 characters as well
             $htmlTranslator->translate(
                 '<p>Color <strong>pajama</strong> паралелепипед</p>',
                 'A',
@@ -58,13 +58,12 @@ public function provideTranslate()
     {
         return [
             [
-                $html = '<p>Text & text text</p>',
-                $translatedHtml = '<p>Translated &amp; translated translated</p>',
+                $html = '<p>Text & text text</p><p>More text</p><p>More text</p>',
+                $translatedHtml = '<p>Translated &amp; translated translated</p><p>More translated</p><p>More translated</p>',
             ],
             [
-                // Some day, this will stay as &bull; ..some ..day
                 $html = '<p>&bull; Text &amp; <strong>text</strong>: text</p>',
-                $translatedHtml = '<p>• Translated &amp; <strong>translated</strong>: translated</p>',
+                $translatedHtml = '<p>&bull; Translated &amp; <strong>translated</strong>: translated</p>',
             ],
             [
                 $html = '<img src="#" alt="Text text text">',