Add last-ditch attempt to recover from invalid XHTML when things go south

Author: mpdude

Integration/XPathHelperFactory.php

@@ -8,25 +8,47 @@
 
 namespace Webfactory\Bundle\LegacyIntegrationBundle\Integration;
 
+use Psr\Log\LoggerInterface;
+use Psr\Log\NullLogger;
 use Webfactory\Dom\BaseParsingHelper;
+use Webfactory\Dom\Exception\ParsingException;
 
 class XPathHelperFactory
 {
-
     /** @var BaseParsingHelper */
     protected $parser;
 
     /** @var LegacyApplication */
     protected $legacyApplication;
 
-    public function __construct(BaseParsingHelper $parser, LegacyApplication $legacy)
+    /**
+     * @var LoggerInterface
+     */
+    private $logger;
+
+    public function __construct(BaseParsingHelper $parser, LegacyApplication $legacy, LoggerInterface $logger = null)
     {
         $this->parser = $parser;
         $this->legacyApplication = $legacy;
+        $this->logger = $logger ?: new NullLogger();
     }
 
     public function createHelper()
     {
-        return new XPathHelper($this->parser, $this->legacyApplication->getResponse()->getContent());
+        $content = $this->legacyApplication->getResponse()->getContent();
+
+        try {
+            return new XPathHelper($this->parser, $content);
+        } catch (ParsingException $exception) {
+            if (function_exists('tidy_repair_string')) {
+                $this->logger->notice('Failed parsing the legacy response as XHTML, trying to clean up with Tidy.', ['exception' => $exception, 'legacy_response' => $exception->getXmlInput()]);
+
+                return new XPathHelper($this->parser, tidy_repair_string($content, ['output-xhtml' => true], 'utf8'));
+            } else {
+                $this->logger->warning('Failed to process legacy response as XHTML.', ['exception' => $exception, 'legacy_response' => $exception->getXmlInput()]);
+
+                throw $exception;
+            }
+        }
     }
 }

Resources/config/services.xml

@@ -41,6 +41,8 @@
                 <service class="%webfactory_legacy_integration.parser_class%"/>
             </argument>
             <argument type="service" id="webfactory_legacy_integration.legacy_application"/>
+            <argument type="service" id="logger" on-invalid="null" />
+            <tag name="monolog.logger" channel="webfactory_legacy_integration" />
         </service>
 
         <service id="webfactory_legacy_integration.xpath_helper"

composer.json

@@ -17,15 +17,15 @@
     },
 
     "require": {
-        "symfony/framework-bundle": "~2.2",
-        "symfony/dependency-injection": "~2.6",
+        "doctrine/annotations": "~1.0",
+        "psr/log": "^1.0",
         "symfony/config": "~2.2",
-        "symfony/http-kernel": "~2.2",
+        "symfony/dependency-injection": "~2.6",
+        "symfony/framework-bundle": "~2.2",
         "symfony/http-foundation": "~2.2",
-        "doctrine/annotations": "~1.0",
-        "webfactory/dom": "~1.0, >= 1.0.15",
-        "twig/twig": "^1.23"
-
+        "symfony/http-kernel": "~2.2",
+        "twig/twig": "^1.23",
+        "webfactory/dom": "~1.0, >= 1.0.15"
     }
 
 }