Merge pull request #7837 from wallabag/use-domcrawler-component

Use DomCrawler component

src/ExpressionLanguage/AuthenticatorProvider.php

@@ -3,6 +3,7 @@
 namespace Wallabag\ExpressionLanguage;
 
 use GuzzleHttp\ClientInterface;
+use Symfony\Component\DomCrawler\Crawler;
 use Symfony\Component\ExpressionLanguage\ExpressionFunction;
 use Symfony\Component\ExpressionLanguage\ExpressionFunctionProviderInterface;
 
@@ -69,27 +70,19 @@ class AuthenticatorProvider implements ExpressionFunctionProviderInterface
                 throw new \Exception('Not supported');
             },
             function (array $arguments, $xpathQuery, $html) {
-                $useInternalErrors = libxml_use_internal_errors(true);
+                try {
+                    $crawler = new Crawler((string) $html);
 
-                $doc = new \DOMDocument();
+                    $crawler = $crawler->filterXPath($xpathQuery);
-                $doc->loadHTML((string) $html, \LIBXML_NOCDATA | \LIBXML_NOWARNING | \LIBXML_NOERROR);
+                } catch (\Throwable $e) {
-
-                $xpath = new \DOMXPath($doc);
-                $domNodeList = $xpath->query($xpathQuery);
-
-                if (0 === $domNodeList->length) {
                     return '';
                 }
 
-                $domNode = $domNodeList->item(0);
+                if (0 === $crawler->count()) {
-
-                libxml_use_internal_errors($useInternalErrors);
-
-                if (null === $domNode || null === $domNode->attributes) {
                     return '';
                 }
 
-                return $domNode->attributes->getNamedItem('value')->nodeValue;
+                return (string) $crawler->first()->attr('value');
             }
         );
     }

src/Import/HtmlImport.php

@@ -2,6 +2,7 @@
 
 namespace Wallabag\Import;
 
+use Symfony\Component\DomCrawler\Crawler;
 use Wallabag\Entity\Entry;
 use Wallabag\Event\EntrySavedEvent;
 
@@ -29,27 +30,23 @@ abstract class HtmlImport extends AbstractImport
             return false;
         }
 
-        $html = new \DOMDocument();
+        $crawler = new Crawler(file_get_contents($this->filepath));
 
-        libxml_use_internal_errors(true);
+        $hrefs = $crawler->filterXPath('//a');
-        $html->loadHTMLFile($this->filepath);
-        $hrefs = $html->getElementsByTagName('a');
-        libxml_use_internal_errors(false);
 
-        if (0 === $hrefs->length) {
+        if (0 === $hrefs->count()) {
             $this->logger->error('Wallabag HTML: no entries in imported file');
 
             return false;
         }
 
-        $entries = [];
+        $entries = $hrefs->each(function (Crawler $node) {
-        foreach ($hrefs as $href) {
+            return [
-            $entry = [];
+                'url' => $node->attr('href'),
-            $entry['url'] = $href->getAttribute('href');
+                'tags' => $node->attr('tags'),
-            $entry['tags'] = $href->getAttribute('tags');
+                'created_at' => $node->attr('add_date'),
-            $entry['created_at'] = $href->getAttribute('add_date');
+            ];
-            $entries[] = $entry;
+        });
-        }
 
         if ($this->producer) {
             $this->parseEntriesForProducer($entries);

src/Import/PocketHtmlImport.php

@@ -2,6 +2,8 @@
 
 namespace Wallabag\Import;
 
+use Symfony\Component\DomCrawler\Crawler;
+
 class PocketHtmlImport extends HtmlImport
 {
     protected $filepath;
@@ -44,27 +46,23 @@ class PocketHtmlImport extends HtmlImport
             return false;
         }
 
-        $html = new \DOMDocument();
+        $crawler = new Crawler(file_get_contents($this->filepath));
 
-        libxml_use_internal_errors(true);
+        $hrefs = $crawler->filterXPath('//a');
-        $html->loadHTMLFile($this->filepath);
-        $hrefs = $html->getElementsByTagName('a');
-        libxml_use_internal_errors(false);
 
-        if (0 === $hrefs->length) {
+        if (0 === $hrefs->count()) {
             $this->logger->error('Pocket HTML: no entries in imported file');
 
             return false;
         }
 
-        $entries = [];
+        $entries = $hrefs->each(function (Crawler $node) {
-        foreach ($hrefs as $href) {
+            return [
-            $entry = [];
+                'url' => $node->attr('href'),
-            $entry['url'] = $href->getAttribute('href');
+                'tags' => $node->attr('tags'),
-            $entry['tags'] = $href->getAttribute('tags');
+                'created_at' => $node->attr('time_added'),
-            $entry['created_at'] = $href->getAttribute('time_added');
+            ];
-            $entries[] = $entry;
+        });
-        }
 
         if ($this->producer) {
             $this->parseEntriesForProducer($entries);

src/SiteConfig/Authenticator/LoginFormAuthenticator.php

@@ -4,6 +4,7 @@ namespace Wallabag\SiteConfig\Authenticator;
 
 use GuzzleHttp\ClientInterface;
 use GuzzleHttp\Cookie\CookieJar;
+use Symfony\Component\DomCrawler\Crawler;
 use Symfony\Component\ExpressionLanguage\ExpressionLanguage;
 use Wallabag\ExpressionLanguage\AuthenticatorProvider;
 use Wallabag\SiteConfig\SiteConfig;
@@ -54,22 +55,16 @@ class LoginFormAuthenticator implements Authenticator
 
     public function isLoginRequired($html)
     {
-        $useInternalErrors = libxml_use_internal_errors(true);
-
         // need to check for the login dom element ($options['not_logged_in_xpath']) in the HTML
-        $doc = new \DOMDocument();
+        try {
-        $doc->loadHTML($html);
+            $crawler = new Crawler((string) $html);
 
-        $xpath = new \DOMXPath($doc);
+            $loggedIn = $crawler->evaluate((string) $this->siteConfig->getNotLoggedInXpath());
-        $loggedIn = $xpath->evaluate((string) $this->siteConfig->getNotLoggedInXpath());
+        } catch (\Throwable $e) {
-
-        if (false === $loggedIn) {
             return false;
         }
 
-        libxml_use_internal_errors($useInternalErrors);
+        return \count($loggedIn) > 0;
-
-        return $loggedIn->length > 0;
     }
 
     /**