Merge branch '3485600-allow-reporting-pages' into '1.0.x'

Issue #3485600: Allow reporting pages that are missing to migrate

See merge request !4

src/Drush/Commands/WordpressMigrateSqlUrlCommands.php

+<?php
+
+namespace Drupal\wordpress_migrate_sql\Drush\Commands;
+
+use Consolidation\OutputFormatters\StructuredData\RowsOfFields;
+use Drupal\Core\Database\Database;
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\Core\File\FileUrlGeneratorInterface;
+use Drupal\Core\StringTranslation\StringTranslationTrait;
+use Drupal\redirect\RedirectRepository;
+use Drupal\wordpress_migrate_sql\MigrationStatus;
+use Drupal\wordpress_migrate_sql\WordpressPost;
+use Drush\Attributes as CLI;
+use Drush\Commands\DrushCommands;
+use FontLib\Table\Type\name;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+use Symfony\Component\Mime\MimeTypeGuesserInterface;
+final class WordpressMigrateSqlUrlCommands extends DrushCommands {
+
+  use StringTranslationTrait;
+
+  public function __construct(protected RedirectRepository $redirectRepository, protected FileUrlGeneratorInterface $fileUrlGenerator) {}
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container): self {
+    return new static(
+      $container->get('redirect.repository'),
+      $container->get('file_url_generator'),
+    );
+  }
+
+  /**
+   * Command description here.
+   */
+  #[CLI\Command(name: 'wordpress_migrate_sql:missing-urls', aliases: ['wms:mu'])]
+  #[CLI\Argument(name: 'database_name', description: 'Database name.')]
+  #[CLI\Argument(name: 'result_file', description: 'Result file.')]
+  #[Cli\Option(name: 'post-types', description: 'Post types.')]
+  #[Cli\Option(name: 'wordpress-base-url', description: 'Wordpress base url.')]
+  #[CLI\Usage(name: 'wordpress_migrate_sql:missing-urls', description: 'Get list of missing urls')]
+  public function analyzeAttachments(
+    $database_name,
+    string $result_file = 'public://wordpress-migrate-sql-missing-urls.csv',
+    array $options = [
+      'post-types' => 'post',
+      'wordpress-base-url' => '',
+    ]) {
+    $database = Database::getConnection('default', $database_name);
+
+    $migration_status = new MigrationStatus($this->redirectRepository, $database, explode(',', $options['post-types']));
+    $missing_urls = $migration_status->getMissingUrls();
+
+    if (!empty($missing_urls)) {
+      $base_url = !empty($options['wordpress-base-url']) ? $options['wordpress-base-url'] . '/' : '';
+      $missing_urls_csv = implode(',', ['Post ID', 'Post type', 'Wordpress URL']);
+
+      $missing_urls_csv = $missing_urls_csv . PHP_EOL . implode(PHP_EOL, array_map(function (WordpressPost $post) use ($base_url) {
+        return implode(',', [
+          $post->getId(),
+          $post->getType(),
+          $base_url . $post->getUrl(),
+        ]);
+      }, $missing_urls));
+      file_put_contents($result_file, $missing_urls_csv);
+      $this->logger()->success(sprintf('Missing %s urls. Full report: %s', count($missing_urls), $this->fileUrlGenerator->generateAbsoluteString($result_file)));
+    }
+  }
+
+}

src/MigrationStatus.php

+<?php
+
+namespace Drupal\wordpress_migrate_sql;
+
+use Drupal\Core\Database\Connection;
+use Drupal\Core\Database\Database;
+use Drupal\Core\Url;
+use Drupal\redirect\Exception\RedirectLoopException;
+use Drupal\redirect\RedirectRepository;
+use Drupal\redirect\Entity\Redirect;
+
+class MigrationStatus {
+
+  protected RedirectRepository $redirectRepository;
+  protected Connection $database;
+
+  protected array $postTypes;
+
+  public function __construct(RedirectRepository $redirectRepository, Connection $database, array $post_types = ['post']) {
+    $this->redirectRepository = $redirectRepository;
+    $this->database = $database;
+    $this->postTypes = $post_types;
+  }
+
+  public function getMissingUrls() {
+    return array_filter($this->getAllUrls(), [$this, 'urlNotAvailable']);
+  }
+  public function getUrlsWithRedirectLoop() {
+    return array_filter($this->getAllUrls(), [$this, 'urlHsaRedirectLoop']);
+  }
+
+  protected function urlNotAvailable(WordpressPost $post) {
+    $drupal_url = Url::fromUserInput('/' . $post->getUrl());
+    return !$drupal_url->isRouted() && !$this->redirectRepository->findMatchingRedirect($post->getUrl()) instanceof Redirect;
+  }
+
+  public function urlHsaRedirectLoop(WordpressPost $post) {
+    $url = $post->getUrl();
+    foreach ([$url, '/' . $url] as $url_candidate) {
+      try {
+        $this->redirectRepository->findMatchingRedirect($url_candidate);
+      }
+      catch (RedirectLoopException $e) {
+        return TRUE;
+      }
+    }
+    return FALSE;
+  }
+
+  public function getAllUrls() {
+    $posts_query_result = $this->database->select('posts')
+      ->fields('posts', ['id', 'post_parent', 'post_name', 'post_type'])
+      ->condition('post_status', 'publish')
+      ->condition('post_type', $this->postTypes, 'IN')
+      ->orderBy('id')
+      ->execute();
+
+    $post_urls = [];
+
+    while ($post = $posts_query_result->fetch()) {
+      $post_urls[$post->id] = new WordpressPost($post->id, $this->getPostUrl($post), $post->post_type);
+    }
+    return $post_urls;
+  }
+
+  protected function getPostUrl($post) {
+    if ($post->post_parent == 0) {
+      return $post->post_name;
+    }
+    else {
+      $parent = $this->findPostParent($post->post_parent);
+      return !empty($parent) ? $this->getPostUrl($parent) . '/' . $post->post_name : NULL;
+    }
+  }
+
+  protected function findPostParent($post_id) {
+    $post_parent_name_query_result = $this->database->select('posts')->fields('posts', ['id', 'post_parent','post_name'])->condition('id', $post_id)->execute();
+    return $post_parent_name_query_result->fetch();
+  }
+
+}

src/WordpressPost.php

+<?php
+
+namespace Drupal\wordpress_migrate_sql;
+
+class WordpressPost
+{
+
+  public function __construct(protected int $id, protected string $url, protected string $type)
+  {
+  }
+
+
+  public function getId(): int {
+    return $this->id;
+  }
+
+  public function getUrl(): string {
+    return $this->url;
+  }
+
+  public function getType(): string {
+    return $this->type;
+  }
+
+}