Use XPath instead of getElementsByTagName

Bug: T336917
wikimedia · May 23, 2023 · 3987b63 · 3987b63
1 parent eb9d5a2
commit 3987b63
Showing 1 changed file with 24 additions and 12 deletions.
diff --git a/src/Model/Svg/SvgFile.php b/src/Model/Svg/SvgFile.php
@@ -15,6 +15,7 @@
 use DOMDocument;
 use DOMElement;
 use DOMNode;
+use DOMNodeList;
 use DOMXpath;
 use Krinkle\Intuition\Intuition;
 use Psr\Log\LoggerAwareTrait;
@@ -158,15 +159,15 @@ protected function makeTranslationReady(): void
         }
 
         // Check that there is something to translate.
-        $texts = $this->document->getElementsByTagName('text');
+        $texts = $this->getElementsByTagName('text');
         $textLength = $texts->length;
         if (0 === $textLength) {
             // Nothing to translate. Given special handling in TranslateController.
             $this->logFileProblem('File {file} has nothing to translate');
             return;
         }
 
-        $styles = $this->document->getElementsByTagName('style');
+        $styles = $this->getElementsByTagName('style');
         $styleLength = $styles->length;
         for ($i = 0; $i < $styleLength; $i++) {
             $style = $styles->item($i);
@@ -187,16 +188,16 @@ protected function makeTranslationReady(): void
         }
 
         // tref tags are not supported.
-        $trefs = $this->document->getElementsByTagName('tref');
+        $trefs = $this->getElementsByTagName('tref');
         if (0 !== $trefs->length) {
             throw new SvgStructureException('structure-error-contains-tref', $trefs->item(0));
         }
 
         // Strip empty tspans, texts, fill $idsInUse
         $idsInUse = [ 0 ];
         $translatableNodes = [];
-        $tspans = $this->document->getElementsByTagName('tspan');
-        $texts = $this->document->getElementsByTagName('text');
+        $tspans = $this->getElementsByTagName('tspan');
+        $texts = $this->getElementsByTagName('text');
         foreach ($tspans as $tspan) {
             if ($tspan->childNodes->length > 1
                 || ( 1 == $tspan->childNodes->length && XML_TEXT_NODE !== $tspan->childNodes->item(0)->nodeType )
@@ -256,8 +257,8 @@ protected function makeTranslationReady(): void
 
         // Reset $translatableNodes
         $translatableNodes = [];
-        $tspans = $this->document->getElementsByTagName('tspan');
-        $texts = $this->document->getElementsByTagName('text');
+        $tspans = $this->getElementsByTagName('tspan');
+        $texts = $this->getElementsByTagName('text');
         foreach ($tspans as $tspan) {
             array_push($translatableNodes, $tspan);
         }
@@ -274,7 +275,7 @@ protected function makeTranslationReady(): void
             }
         }
 
-        $texts = $this->document->getElementsByTagName('text');
+        $texts = $this->getElementsByTagName('text');
         $textLength = $texts->length;
         for ($i = 0; $i < $textLength; $i++) {
             /** @var DOMElement $text */
@@ -322,9 +323,9 @@ protected function makeTranslationReady(): void
             }
         }
 
-        $switchLength = $this->document->getElementsByTagName('switch')->length;
+        $switchLength = $this->getElementsByTagName('switch')->length;
         for ($i = 0; $i < $switchLength; $i++) {
-            $switch = $this->document->getElementsByTagName('switch')->item($i);
+            $switch = $this->getElementsByTagName('switch')->item($i);
             $siblings = $switch->childNodes;
             $existingLangs = [];
             foreach ($siblings as $sibling) {
@@ -396,7 +397,7 @@ protected function makeTranslationReady(): void
      */
     protected function analyse(): void
     {
-        $switches = $this->document->getElementsByTagName('switch');
+        $switches = $this->getElementsByTagName('switch');
         $number = $switches->length;
         $translations = [];
         $this->filteredTextNodes = []; // Reset
@@ -470,6 +471,17 @@ protected function analyse(): void
         $this->savedLanguages = array_unique($this->savedLanguages);
     }
 
+    /**
+     * Workaround for O(N^2) performance of DOMDocument::getElementsByTagName() T336917
+     *
+     * @param string $name
+     * @return DOMNodeList
+     */
+    private function getElementsByTagName(string $name): DOMNodeList
+    {
+        return $this->xpath->query("//*[local-name()=\"$name\"]");
+    }
+
     /**
      * Returns a list of translations present in the loaded file, in the following format:
      *
@@ -561,7 +573,7 @@ public function switchToTranslationSet(array $translations): array
         $currentLanguages = $this->getSavedLanguages();
         $expanded = $started = [];
 
-        $switches = $this->document->getElementsByTagName('switch');
+        $switches = $this->getElementsByTagName('switch');
         $number = $switches->length;
         for ($i = 0; $i < $number; $i++) {
             $switch = $switches->item($i);