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

…ltinational site annotations tag.
Novactive · May 31, 2024 · bb3f0d4 · bb3f0d4
1 parent 17637d6
commit bb3f0d4
Show file tree

Hide file tree

Showing 11 changed files with 528 additions and 77 deletions.
diff --git a/components/SEOBundle/bundle/Controller/SitemapController.php b/components/SEOBundle/bundle/Controller/SitemapController.php
diff --git a/components/SEOBundle/bundle/Core/Helper/SiteMapHelper.php b/components/SEOBundle/bundle/Core/Helper/SiteMapHelper.php
@@ -0,0 +1,173 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Novactive\Bundle\eZSEOBundle\Core\Helper;
+
+use Exception;
+use Ibexa\Contracts\Core\SiteAccess\ConfigResolverInterface;
+use Ibexa\Core\MVC\Symfony\Locale\LocaleConverter;
+use Ibexa\Core\MVC\Symfony\Routing\Generator\RouteReferenceGenerator;
+use Ibexa\Core\MVC\Symfony\Routing\UrlAliasRouter;
+use Ibexa\Core\MVC\Symfony\SiteAccess\SiteAccessServiceInterface;
+use Ibexa\Migration\Log\LoggerAwareTrait;
+use Psr\Log\LoggerInterface;
+use Psr\Log\NullLogger;
+use Symfony\Component\Routing\Generator\UrlGeneratorInterface;
+use Symfony\Component\Routing\RouterInterface;
+use Throwable;
+
+class SiteMapHelper
+{
+    use LoggerAwareTrait;
+
+    /**
+     * @var ConfigResolverInterface
+     */
+    private $configResolver;
+
+    /**
+     * @var RouteReferenceGenerator
+     */
+    private $routeReferenceGenerator;
+
+    /**
+     * @var RouterInterface
+     */
+    private $router;
+
+    /**
+     * @var LocaleConverter
+     */
+    private $localeConverter;
+
+    /**
+     * @var SiteAccessServiceInterface
+     */
+    private $siteAccessService;
+
+    public function __construct(
+        ConfigResolverInterface $configResolver,
+        SiteAccessServiceInterface $siteAccessService,
+        RouteReferenceGenerator $routeReferenceGenerator,
+        RouterInterface $router,
+        LocaleConverter $localeConverter,
+        ?LoggerInterface $logger = null
+    ) {
+        $this->configResolver = $configResolver;
+        $this->siteAccessService = $siteAccessService;
+        $this->routeReferenceGenerator = $routeReferenceGenerator;
+        $this->router = $router;
+        $this->localeConverter = $localeConverter;
+        $this->logger = $logger ?? new NullLogger();
+    }
+
+    public function generateLocationUrl(
+        int $locationId,
+        string $siteAccess = null
+    ): ?string {
+        try {
+            $routeParams['locationId'] = $locationId;
+            $routeReference = $this->routeReferenceGenerator->generate(
+                UrlAliasRouter::URL_ALIAS_ROUTE_NAME,
+                $routeParams
+            );
+            if ($siteAccess) {
+                $routeReference->set('siteaccess', $siteAccess);
+            }
+            $url = $this->router->generate(
+                UrlAliasRouter::URL_ALIAS_ROUTE_NAME,
+                $routeReference->getParams(),
+                UrlGeneratorInterface::ABSOLUTE_URL
+            );
+        } catch (Throwable $exception) {
+            $this->logger->error('NovaeZSEO: ' . $exception->getMessage());
+            $url = null;
+        }
+
+        return $url;
+    }
+    public function generateRouteUrl(
+        string $routeName,
+        string $siteAccess = null,
+        array $parameters = []
+    ): ?string {
+        try {
+            $url = $this->router->generate(
+                $routeName,
+                [...['siteaccess' => $siteAccess], ...$parameters],
+                UrlGeneratorInterface::ABSOLUTE_URL
+            );
+        } catch (Throwable $exception) {
+            $this->logger->error('NovaeZSEO: ' . $exception->getMessage());
+            $url = null;
+        }
+
+        return $url;
+    }
+    public function getSiteAccessesLocationIdLanguages(): array
+    {
+
+        static $rootLocationLanguages = null;
+
+        if (is_array($rootLocationLanguages)) {
+            return $rootLocationLanguages;
+        }
+
+        $rootLocationLanguages = [];
+        $siteAccesses = $this->configResolver->getParameter('translation_siteaccesses');
+        foreach ($siteAccesses as $siteAccess) {
+            $rootLocationLanguages[$siteAccess] = [
+                'rootLocationId' =>  $this->getSiteAccessRootLocationId($siteAccess),
+                'mainLanguage' => $this->getSiteAccessMainLanguage($siteAccess),
+                'languages' => $this->getSiteAccessLanguages($siteAccess)
+            ];
+        }
+
+        return $rootLocationLanguages;
+    }
+
+    public function getCurrentSiteAccess(): ?string
+    {
+        return $this->siteAccessService->getCurrent()?->name;
+    }
+
+    public function getCurrentSiteAccessRootLocationId(): ?int
+    {
+        return $this->getSiteAccessRootLocationId($this->getCurrentSiteAccess());
+    }
+    public function getCurrentSiteAccessMainLanguage(): ?string
+    {
+        return $this->getSiteAccessMainLanguage($this->getCurrentSiteAccess());
+    }
+
+    public function getSiteAccessRootLocationId(string $siteAccess): ?int
+    {
+        return $this->configResolver->getParameter('content.tree_root.location_id', null, $siteAccess);
+    }
+
+    public function getSiteAccessLanguages(string $siteAccess): array
+    {
+        return (array) $this->configResolver->getParameter('languages', null, $siteAccess);
+    }
+
+    public function getSiteAccessMainLanguage(string $siteAccess): string
+    {
+        $languages = $this->configResolver->getParameter('languages', null, $siteAccess);
+        return array_shift($languages);
+    }
+
+    public function getHrefLang(string $languageCode): string
+    {
+        return str_replace(
+            '_',
+            '-',
+            ($this->localeConverter->convertToPOSIX($languageCode) ?? '')
+        );
+    }
+
+    public function logException(Exception $exception): void
+    {
+        $this->logger?->error('NovaeZSEO: ' . $exception->getMessage());
+    }
+}
diff --git a/components/SEOBundle/bundle/Core/MetaNameSchema.php b/components/SEOBundle/bundle/Core/MetaNameSchema.php
@@ -74,6 +74,11 @@ class MetaNameSchema extends NameSchemaService
      */
     private $configurationResolver;
 
+    /**
+     * @var FieldTypeRegistry
+     */
+    protected $fieldTypeRegistry;
+
     public function __construct(
         ContentTypeHandler $contentTypeHandler,
         FieldTypeRegistry $fieldTypeRegistry,
@@ -83,6 +88,7 @@ public function __construct(
         ConfigResolverInterface $configurationResolver,
         array $settings = []
     ) {
+        $this->fieldTypeRegistry = $fieldTypeRegistry;
         $settings['limit'] = $this->fieldContentMaxLength;
         $handler = new ContentTypeDomainMapper(
             $contentTypeHandler,

diff --git a/components/SEOBundle/bundle/Core/Sitemap/QueryFactory.php b/components/SEOBundle/bundle/Core/Sitemap/QueryFactory.php
@@ -6,6 +6,7 @@
 
 use Ibexa\Bundle\Core\DependencyInjection\Configuration\ConfigResolver;
 use Ibexa\Contracts\Core\Repository\Exceptions\NotFoundException;
+use Ibexa\Contracts\Core\Repository\Exceptions\UnauthorizedException;
 use Ibexa\Contracts\Core\Repository\Repository;
 use Ibexa\Contracts\Core\Repository\Values\Content\Location;
 use Ibexa\Contracts\Core\Repository\Values\Content\LocationQuery as Query;
@@ -52,16 +53,28 @@ private function getRootLocation(): Location
         );
     }
 
-    public function __invoke(): Query
-    {
+    public function __invoke(
+        int $rootLocationId = null,
+        array $languages = [],
+        bool $matchAlwaysAvailable = true
+    ): Query {
         $query = new Query();
 
         // 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 (true === $limitToRootLocation) {
+        if ((int) $rootLocationId) {
+            try {
+                $rootLocation = $this->repository->getLocationService()->loadLocation($rootLocationId);
+            } catch (NotFoundException|UnauthorizedException $e) {
+                $rootLocation = null;
+            }
+            if ($rootLocation) {
+                $criterions[] = new Criterion\Subtree($rootLocation->pathString);
+            }
+        } elseif (true === $limitToRootLocation) {
             $criterions[] = new Criterion\Subtree($this->getRootLocation()->pathString);
         }
 
@@ -88,7 +101,8 @@ public function __invoke(): Query
             )
         );
 
-        $criterions[] = new Criterion\LanguageCode($this->configResolver->getParameter('languages'), true);
+        $languages = empty($languages) ? $this->configResolver->getParameter('languages') : $languages;
+        $criterions[] = new Criterion\LanguageCode($languages, $matchAlwaysAvailable);
 
         $query->query = new Criterion\LogicalAnd($criterions);
         $query->sortClauses = [new SortClause\DatePublished(Query::SORT_DESC)];

diff --git a/components/SEOBundle/bundle/DependencyInjection/Configuration.php b/components/SEOBundle/bundle/DependencyInjection/Configuration.php
@@ -28,8 +28,10 @@ public function getConfigTreeBuilder(): TreeBuilder
             ->scalarNode('google_gatracker')->defaultValue('~')->end()
             ->scalarNode('google_anonymizeIp')->defaultValue('~')->end()
             ->scalarNode('bing_verification')->defaultValue('~')->end()
-            ->booleanNode('limit_to_rootlocation')->defaultValue('~')->end()
-            ->booleanNode('display_images_in_sitemap')->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')
             ->isRequired()

diff --git a/components/SEOBundle/bundle/DependencyInjection/NovaeZSEOExtension.php b/components/SEOBundle/bundle/DependencyInjection/NovaeZSEOExtension.php
@@ -38,6 +38,7 @@ public function prepend(ContainerBuilder $container): void
             'ez_field_templates.yml' => 'ibexa',
             'variations.yml' => 'ibexa',
             'admin_ui/ez_field_templates.yml' => 'ibexa',
+            'ibexa_locale_conversion.yml' => 'ibexa',
         ];
 
         foreach ($configs as $fileName => $extensionName) {
@@ -68,6 +69,8 @@ public function load(array $configs, ContainerBuilder $container): void
         $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
@@ -11,8 +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: ~
-    nova_ezseo.default.display_images_in_sitemap: ~
+    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: []

diff --git a/components/SEOBundle/bundle/Resources/config/ibexa_locale_conversion.yml b/components/SEOBundle/bundle/Resources/config/ibexa_locale_conversion.yml
@@ -0,0 +1,15 @@
+#list from https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
+locale_conversion:
+  bre-BT: br_BT
+  rum-RO: ro_RO
+  mlt-MT: mt_MT
+  gle-IE: ga_IE
+  est-EE: et_EE
+  slv-SI: sl_SI
+  sla-MK: sl_MK
+  lav-LV: lv_LV
+  lit-LT: lt_LT
+  geo-GE: ka_GE
+  per-IR: fa_IR
+  bul-BG: bg_BG
+  bel-BY: be_BY
diff --git a/components/SEOBundle/bundle/Resources/config/services.yml b/components/SEOBundle/bundle/Resources/config/services.yml
@@ -51,9 +51,9 @@ services:
     Novactive\Bundle\eZSEOBundle\Core\MetaNameSchema:
         lazy: true
         arguments:
-            $contentTypeHandler: "@Ibexa\\Contracts\\Core\\Persistence\\Content\\Type\\Handler"
-            $languageHandler: "@Ibexa\\Core\\Persistence\\Cache\\ContentLanguageHandler"
-            $translationHelper: "@Ibexa\\Core\\Helper\\TranslationHelper"
+            $contentTypeHandler: '@Ibexa\Contracts\Core\Persistence\Content\Type\Handler'
+            $languageHandler: '@Ibexa\Core\Persistence\Cache\ContentLanguageHandler'
+            $translationHelper: '@Ibexa\Core\Helper\TranslationHelper'
         calls:
             - [setRichTextConverter, ["@Ibexa\\FieldTypeRichText\\RichText\\Converter\\Html5"]]
             # Note: injecting lower layer Variation Handler (AliasGenerator) as a workaround for missing Public API objects context
@@ -73,6 +73,8 @@ services:
         $ioService: '@ezseo_importurls.ibexa.core.io_service'
         $cacheDirectory: '%kernel.cache_dir%'
 
+    Novactive\Bundle\eZSEOBundle\Core\Helper\SiteMapHelper:
+
     Novactive\Bundle\eZSEOBundle\Core\SiteAccessAwareEntityManagerFactory:
         arguments:
             $repositoryConfigurationProvider: "@Ibexa\\Bundle\\Core\\ApiLoader\\RepositoryConfigurationProvider"

diff --git a/components/SEOBundle/documentation/USAGE.md b/components/SEOBundle/documentation/USAGE.md
@@ -141,6 +141,48 @@ nova_ezseo:
                 contentTypeIdentifiers: ['footer','something']
 ```
 
+
+Set `multi_siteaccess_sitemap` to true to automatically generate the index page sitemap.xml per site access.
+
+```yml
+nova_ezseo:
+    system:
+        default:
+            multi_siteaccess_sitemap: true
+```
+
+```html
+<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
+    <sitemap>
+        <loc>https://www.example.com/fr/sitemap-1.xml</loc>
+        <lastmod>2024-05-31T01:04:52+00:00</lastmod>
+    </sitemap>
+    <sitemap>
+        <loc>https://www.example.com/en/sitemap-1.xml</loc>
+        <lastmod>2024-05-31T01:04:52+00:00</lastmod>
+    </sitemap>
+</sitemapindex>
+```
+
+Set `multi_languages_sitemap` to inject the multilingual and multinational site annotations tag
+Notice: this doesn't work with `multi_siteaccess_sitemap: false`.
+```html
+<url xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
+    <loc>https://www.example.com/pageFr</loc>
+    <xhtml:link xmlns:xhtml="http://www.w3.org/1999/xhtml" rel="alternate" hreflang="en-GB" href="https://www.example.com/en/pageEn" />
+    <xhtml:link xmlns:xhtml="http://www.w3.org/1999/xhtml" rel="alternate" hreflang="fr-FR" href="https://www.example.com/pageFr" />
+    <lastmod>2024-05-03T08:48:52+00:00</lastmod>
+</url>
+```
+
+```yml
+nova_ezseo:
+    system:
+        default:
+            multi_siteaccess_sitemap: true
+            multi_languages_sitemap: true
+```
+
 Set "display_images_in_sitemap" to true to inject the image tags.
 Notice: this doesn't work with `limit_to_rootlocation: true`.
 

diff --git a/composer.json b/composer.json
@@ -61,6 +61,7 @@
         "knplabs/knp-menu": "^3.1",
         "behat/behat": "^3.8",
         "friends-of-behat/mink-extension": "^2.5",
-        "ext-pdo": "*"
+        "ext-pdo": "*",
+        "ext-dom": "*"
     }
 }