generate sitemap.xml per siteAccess && inject the multilingual and mu…

…ltinational site annotations tag.
Novactive · May 31, 2024 · a4d7791 · a4d7791
1 parent ba9e83c
commit a4d7791
Show file tree

Hide file tree

Showing 6 changed files with 251 additions and 68 deletions.
diff --git a/components/SEOBundle/bundle/Controller/SitemapController.php b/components/SEOBundle/bundle/Controller/SitemapController.php
@@ -21,10 +21,12 @@
 use Ibexa\Contracts\Core\Repository\Values\Content\Search\SearchResult;
 use Ibexa\Contracts\Core\Variation\VariationHandler;
 use Ibexa\Core\Helper\FieldHelper;
+use Ibexa\Core\MVC\Symfony\Routing\UrlAliasRouter;
 use Novactive\Bundle\eZSEOBundle\Core\Helper\SiteMapHelper;
 use Novactive\Bundle\eZSEOBundle\Core\Sitemap\QueryFactory;
 use Symfony\Component\HttpFoundation\Response;
 use Symfony\Component\Routing\Annotation\Route;
+use Symfony\Component\Routing\Generator\UrlGeneratorInterface;
 use Symfony\Component\Routing\RouterInterface;
 use Throwable;
 
@@ -71,16 +73,41 @@ public function __construct(
      */
     public function indexAction(QueryFactory $queryFactory): Response
     {
+        $isMultisiteAccess = $this->getConfigResolver()
+            ->getParameter('multi_siteaccess_sitemap', 'nova_ezseo') ?? false;
         // Dom Doc
         $sitemap = new DOMDocument('1.0', 'UTF-8');
         $sitemap->formatOutput = true;
         // Create an index for multi site
-        $root = $sitemap->createElement('sitemapindex');
-        $root->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
-        $sitemap->appendChild($root);
-
-        $this->fillSitemapIndex($sitemap, $root, $queryFactory);
+        if ($isMultisiteAccess) {
+            $root = $sitemap->createElement('sitemapindex');
+            $root->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
+            $sitemap->appendChild($root);
 
+            $this->fillSitemapMultiSiteIndex($sitemap, $root, $queryFactory);
+        } else {
+            $searchService = $this->getRepository()->getSearchService();
+            $query = $queryFactory();
+            $query->limit = 0;
+            $resultCount = $searchService->findLocations($query)->totalCount;
+        // create an index if we are greater than th PACKET_MAX
+            if ($resultCount > static::PACKET_MAX) {
+                $root = $sitemap->createElement('sitemapindex');
+                $root->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
+                $sitemap->appendChild($root);
+
+                $this->fillSitemapIndex($sitemap, $resultCount, $root);
+            } else {
+                // if we are less or equal than the PACKET_SIZE, redo the search with no limit and list directly the urlmap
+                $query->limit = $resultCount;
+                $results = $searchService->findLocations($query);
+                $root = $sitemap->createElement('urlset');
+                $root->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
+                $root->setAttribute('xmlns:image', 'http://www.google.com/schemas/sitemap-image/1.1');
+                $this->fillSitemap($sitemap, $root, $results);
+                $sitemap->appendChild($root);
+            }
+        }
         $response = new Response($sitemap->saveXML(), 200, ['Content-type' => 'text/xml']);
         $response->setSharedMaxAge(86400);
 
@@ -94,16 +121,45 @@ public function indexAction(QueryFactory $queryFactory): Response
      */
     public function pageAction(QueryFactory $queryFactory, int $page = 1): Response
     {
-        $schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9" .
-            " http://www.w3.org/1999/xhtml http://www.w3.org/2002/08/xhtml/xhtml1-strict.xsd".
-            " http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd http://www.google.com/schemas/sitemap-image/1.1".
-            " http://www.google.com/schemas/sitemap-image/1.1/sitemap-image.xsd" .
-            " http://www.google.com/schemas/sitemap-video/1.1";
+        $isMultisiteAccess = $this->getConfigResolver()
+            ->getParameter('multi_siteaccess_sitemap', 'nova_ezseo') ?? false;
+
         $sitemap = new DOMDocument('1.0', 'UTF-8');
         $root = $sitemap->createElement('urlset');
         $sitemap->formatOutput = true;
         $root->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
         $root->setAttribute('xmlns:image', 'http://www.google.com/schemas/sitemap-image/1.1');
+
+        // Create an index for multi site
+        if ($isMultisiteAccess) {
+            $this->multisiteAccessPage($sitemap, $root, $queryFactory, $page);
+        } else {
+            $sitemap->appendChild($root);
+            $query = $queryFactory();
+            $query->limit = static::PACKET_MAX;
+            $query->offset = static::PACKET_MAX * ($page - 1);
+            $searchService = $this->getRepository()->getSearchService();
+            $results = $searchService->findLocations($query);
+            $this->fillSitemap($sitemap, $root, $results);
+        }
+
+        $response = new Response($sitemap->saveXML(), 200, ['Content-type' => 'text/xml']);
+        $response->setSharedMaxAge(86400);
+
+        return $response;
+    }
+
+    public function multisiteAccessPage(
+        DOMDocument $sitemap,
+        DOMElement $root,
+        QueryFactory $queryFactory,
+        int $page = 1
+    ): void {
+        $schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9" .
+            " http://www.w3.org/1999/xhtml http://www.w3.org/2002/08/xhtml/xhtml1-strict.xsd".
+            " http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd http://www.google.com/schemas/sitemap-image/1.1".
+            " http://www.google.com/schemas/sitemap-image/1.1/sitemap-image.xsd" .
+            " http://www.google.com/schemas/sitemap-video/1.1";
         $root->setAttribute('xmlns:xsi', 'http://www.w3.org/2001/XMLSchema-instance');
         $root->setAttribute('xmlns:xhtml', 'http://www.w3.org/1999/xhtml');
         $root->setAttribute('xmlns:video', 'http://www.google.com/schemas/sitemap-video/1.1');
@@ -124,75 +180,44 @@ public function pageAction(QueryFactory $queryFactory, int $page = 1): Response
             $searchService = $this->getRepository()->getSearchService();
 
             $results = $searchService->findLocations($query);
-            $this->fillSitemap($sitemap, $root, $results);
+            $this->fillMultiLanguagesSitemap($sitemap, $root, $results);
         }
-
-
-        $response = new Response($sitemap->saveXML(), 200, ['Content-type' => 'text/xml']);
-        $response->setSharedMaxAge(86400);
-
-        return $response;
     }
 
     /**
      * Fill a sitemap.
      */
     protected function fillSitemap(DOMDocument $sitemap, DOMElement $root, SearchResult $results): void
     {
-        $currentSiteAccess = $this->siteMapHelper->getCurrentSiteAccess();
-        $siteAccesses = $this->getConfigResolver()->getParameter('translation_siteaccesses');
         foreach ($results->searchHits as $searchHit) {
             /**
              * @var Location  $location
              */
             $location = $searchHit->valueObject;
+            try {
+                $url = $this->generateUrl(
+                    UrlAliasRouter::URL_ALIAS_ROUTE_NAME,
+                    ['locationId' => $location->id],
+                    UrlGeneratorInterface::ABSOLUTE_URL
+                );
+            } catch (\Exception $exception) {
+                $this->siteMapHelper->logException($exception);
+                continue;
+            }
 
-            $mainLanguageUrl = $this->siteMapHelper->generateLocationUrl($location->id, $currentSiteAccess);
-            if (null === $mainLanguageUrl || 0 != strpos($mainLanguageUrl, 'view/content/')) {
+            if (0 != strpos($url, 'view/content/')) {
                 continue;
             }
 
             $modified = $location->contentInfo->modificationDate->format('c');
-            $loc = $sitemap->createElement('loc', $mainLanguageUrl);
+            $loc = $sitemap->createElement('loc', $url);
             $lastmod = $sitemap->createElement('lastmod', $modified);
             $urlElt = $sitemap->createElement('url');
 
-            $urlElt->appendChild($loc);
             // Inject the image tags if config is enabled
             $this->injectImageTag($location, $sitemap, $urlElt);
 
-            try {
-                $languagesCodes = [];
-                $contentInfo = $location->contentInfo;
-                $contentLanguages = $this->repository->getContentService()
-                    ->loadVersionInfo($contentInfo)->getLanguages();
-                foreach ($contentLanguages as $language) {
-                    $languagesCodes[] = $language->languageCode;
-                }
-                foreach ($siteAccesses as $siteAccess) {
-                    $siteAccessMainLanguage = $this->siteMapHelper->getSiteAccessMainLanguage($siteAccess);
-                    if (!in_array($this->siteMapHelper->getSiteAccessMainLanguage($siteAccess), $languagesCodes)) {
-                        continue;
-                    }
-
-                    $url = $this->siteMapHelper->generateLocationUrl($location->id, $siteAccess);
-                    $hreflang = $this->siteMapHelper->getHrefLang($siteAccessMainLanguage);
-                    if (null === $mainLanguageUrl || 0 != strpos($url, 'view/content/')) {
-                        continue;
-                    }
-
-                    $xhtml = $sitemap->createElement('xhtml:link');
-                    $xhtml->setAttribute('rel', 'alternate');
-                    $xhtml->setAttribute('hreflang', $hreflang);
-                    $xhtml->setAttribute('href', $url);
-                    $urlElt->appendChild($xhtml);
-                }
-            } catch (Throwable $e) {
-                $this->siteMapHelper->logException($e);
-                continue;
-            }
-
-
+            $urlElt->appendChild($loc);
             $urlElt->appendChild($lastmod);
             $root->appendChild($urlElt);
         }
@@ -201,7 +226,33 @@ protected function fillSitemap(DOMDocument $sitemap, DOMElement $root, SearchRes
     /**
      * Fill the sitemap index.
      */
-    protected function fillSitemapIndex(
+    protected function fillSitemapIndex(DOMDocument $sitemap, int $numberOfResults, DOMElement $root): void
+    {
+        $numberOfPage = (int) ceil($numberOfResults / static::PACKET_MAX);
+        for ($sitemapNumber = 1; $sitemapNumber <= $numberOfPage; ++$sitemapNumber) {
+            $sitemapElt = $sitemap->createElement('sitemap');
+
+            try {
+                $locUrl = $this->generateUrl(
+                    '_novaseo_sitemap_page',
+                    ['page' => $sitemapNumber],
+                    UrlGeneratorInterface::ABSOLUTE_URL
+                );
+            } catch (\Exception $exception) {
+                $this->siteMapHelper->logException($exception);
+                continue;
+            }
+
+            $loc = $sitemap->createElement('loc', $locUrl);
+            $date = new DateTime();
+            $modificationDate = $date->format('c');
+            $mod = $sitemap->createElement('lastmod', $modificationDate);
+            $sitemapElt->appendChild($loc);
+            $sitemapElt->appendChild($mod);
+            $root->appendChild($sitemapElt);
+        }
+    }
+    protected function fillSitemapMultiSiteIndex(
         DOMDocument $sitemap,
         DOMElement $root,
         QueryFactory $queryFactory
@@ -243,8 +294,79 @@ protected function fillSitemapIndex(
         });
     }
 
-    public function injectImageTag($location, DOMDocument $sitemap, DOMElement $root): void
-    {
+    /**
+     * Fill a sitemap.
+     */
+    protected function fillMultiLanguagesSitemap(
+        DOMDocument $sitemap,
+        DOMElement $root,
+        SearchResult $results
+    ): void {
+        $currentSiteAccess = $this->siteMapHelper->getCurrentSiteAccess();
+        foreach ($results->searchHits as $searchHit) {
+            /**
+             * @var Location  $location
+             */
+            $location = $searchHit->valueObject;
+
+            $mainLanguageUrl = $this->siteMapHelper->generateLocationUrl($location->id, $currentSiteAccess);
+            if (null === $mainLanguageUrl || 0 != strpos($mainLanguageUrl, 'view/content/')) {
+                continue;
+            }
+
+            $modified = $location->contentInfo->modificationDate->format('c');
+            $loc = $sitemap->createElement('loc', $mainLanguageUrl);
+            $lastmod = $sitemap->createElement('lastmod', $modified);
+            $urlElt = $sitemap->createElement('url');
+
+            $urlElt->appendChild($loc);
+            // Inject the image tags if config is enabled
+            $this->injectImageTag($location, $sitemap, $urlElt);
+            // Inject the alternate lang tags if config is enabled
+            $this->injectAlternateLangTag($location, $sitemap, $urlElt);
+
+            $urlElt->appendChild($lastmod);
+            $root->appendChild($urlElt);
+        }
+    }
+
+    public function injectAlternateLangTag($location, DOMDocument $sitemap, DOMElement $root): void {
+        $isMultiLanguages = $this->getConfigResolver()->getParameter('multi_languages_sitemap', 'nova_ezseo');
+        if ($isMultiLanguages) {
+            try {
+                $siteAccesses = $this->getConfigResolver()->getParameter('translation_siteaccesses');
+                $languagesCodes = [];
+                $contentInfo = $location->contentInfo;
+                $contentLanguages = $this->repository->getContentService()
+                    ->loadVersionInfo($contentInfo)->getLanguages();
+                foreach ($contentLanguages as $language) {
+                    $languagesCodes[] = $language->languageCode;
+                }
+                foreach ($siteAccesses as $siteAccess) {
+                    $siteAccessMainLanguage = $this->siteMapHelper->getSiteAccessMainLanguage($siteAccess);
+                    if (!in_array($this->siteMapHelper->getSiteAccessMainLanguage($siteAccess), $languagesCodes)) {
+                        continue;
+                    }
+
+                    $url = $this->siteMapHelper->generateLocationUrl($location->id, $siteAccess);
+                    $hreflang = $this->siteMapHelper->getHrefLang($siteAccessMainLanguage);
+                    if (null === $url || 0 != strpos($url, 'view/content/')) {
+                        continue;
+                    }
+
+                    $xhtml = $sitemap->createElement('xhtml:link');
+                    $xhtml->setAttribute('rel', 'alternate');
+                    $xhtml->setAttribute('hreflang', $hreflang);
+                    $xhtml->setAttribute('href', $url);
+                    $root->appendChild($xhtml);
+                }
+            } catch (Throwable $e) {
+                $this->siteMapHelper->logException($e);
+            }
+        }
+    }
+
+    public function injectImageTag($location, DOMDocument $sitemap, DOMElement $root): void {
         $displayImage = $this->getConfigResolver()->getParameter('display_images_in_sitemap', 'nova_ezseo');
 
         if (true === $displayImage) {

diff --git a/components/SEOBundle/bundle/Core/Sitemap/QueryFactory.php b/components/SEOBundle/bundle/Core/Sitemap/QueryFactory.php
@@ -63,22 +63,19 @@ public function __invoke(
         // always here, we want visible Contents
         $criterions = [new Criterion\Visibility(Criterion\Visibility::VISIBLE)];
 
+        // do we want to limit per Root Location, but default we don't
+        $limitToRootLocation = $this->configResolver->getParameter('limit_to_rootlocation', 'nova_ezseo');
         if ((int) $rootLocationId) {
             try {
                 $rootLocation = $this->repository->getLocationService()->loadLocation($rootLocationId);
             } catch (NotFoundException|UnauthorizedException $e) {
                 $rootLocation = null;
             }
-        } else {
-            $rootLocation = null;
-        }
-
-        if (!$rootLocation) {
-            $rootLocation = $this->getRootLocation();
-        }
-
-        if ($rootLocation) {
-            $criterions[] = new Criterion\Subtree($rootLocation->pathString);
+            if ($rootLocation){
+                $criterions[] = new Criterion\Subtree($rootLocation->pathString);
+            }
+        } elseif (true === $limitToRootLocation) {
+            $criterions[] = new Criterion\Subtree($this->getRootLocation()->pathString);
         }
 
         // Inclusions

diff --git a/components/SEOBundle/bundle/DependencyInjection/Configuration.php b/components/SEOBundle/bundle/DependencyInjection/Configuration.php
@@ -28,6 +28,9 @@ public function getConfigTreeBuilder(): TreeBuilder
             ->scalarNode('google_gatracker')->defaultValue('~')->end()
             ->scalarNode('google_anonymizeIp')->defaultValue('~')->end()
             ->scalarNode('bing_verification')->defaultValue('~')->end()
+            ->booleanNode('limit_to_rootlocation')->defaultFalse()->end()
+            ->booleanNode('multi_siteaccess_sitemap')->defaultFalse()->end()
+            ->booleanNode('multi_languages_sitemap')->defaultFalse()->end()
             ->booleanNode('display_images_in_sitemap')->defaultFalse()->end()
             ->scalarNode('fieldtype_metas_identifier')->defaultValue('metas')->end()
             ->arrayNode('fieldtype_metas')

diff --git a/components/SEOBundle/bundle/DependencyInjection/NovaeZSEOExtension.php b/components/SEOBundle/bundle/DependencyInjection/NovaeZSEOExtension.php
@@ -67,7 +67,10 @@ public function load(array $configs, ContainerBuilder $container): void
         $processor->mapSetting('google_gatracker', $config);
         $processor->mapSetting('google_anonymizeIp', $config);
         $processor->mapSetting('bing_verification', $config);
+        $processor->mapSetting('limit_to_rootlocation', $config);
         $processor->mapSetting('display_images_in_sitemap', $config);
+        $processor->mapSetting('multi_siteaccess_sitemap', $config);
+        $processor->mapSetting('multi_languages_sitemap', $config);
         $processor->mapSetting('robots', $config);
         $processor->mapConfigArray('fieldtype_metas', $config, ContextualizerInterface::MERGE_FROM_SECOND_LEVEL);
         $processor->mapConfigArray('default_metas', $config);

diff --git a/components/SEOBundle/bundle/Resources/config/default_settings.yml b/components/SEOBundle/bundle/Resources/config/default_settings.yml
@@ -11,7 +11,10 @@ parameters:
     nova_ezseo.default.google_anonymizeIp: ~
     nova_ezseo.default.bing_verification: ~
     nova_ezseo.default.custom_fallback_service: ~
+    nova_ezseo.default.limit_to_rootlocation: false
     nova_ezseo.default.display_images_in_sitemap: false
+    nova_ezseo.default.multi_siteaccess_sitemap: false
+    nova_ezseo.default.multi_languages_sitemap: false
     nova_ezseo.default.sitemap_excludes:
         locations: []
         subtrees: []