migration projets: body markdown, images HD, légendes, personnes prénom/nom

- ProjetNode: body Markdown->HTML (strip HTML brut), images + légendes lues depuis public://drive_shared_leshed/Page{UID} - ImageResizeImport: nouveau process plugin, redimensionne les images en web HD (max 1920px) et crée le fichier managé (idempotent) - field_images: caption via image_field_caption, alt depuis les légendes - PersonneGenerate: nom dans field_nom (titre auto-généré via auto_entitylabel), dédup sur field_nom + field_prenom - title: skip_on_empty (ligne ignorée si "Nom du projet" vide) - CER: relation curateur sur field_projets_curates (distinct de field_projet_s_lie_s) - deps: + league/commonmark, image_field_caption, auto_entitylabel ; - migrate_file
2026-07-26 16:26:18 +02:00
parent 829653a9e9
commit a2cb5f9dcf
28 changed files with 1600 additions and 142 deletions
@@ -10,6 +10,10 @@ source:
  plugin: projet_node
  data_fetcher_plugin: http
  data_parser_plugin: google_sheets
+  constants:
+    # Trailing slash: file_import keeps the original filename and copies the
+    # image into this directory.
+    images_destination: 'public://projets/'
  # The feed file for the spreadsheet. The Google Spreadsheet should be either “Public” or set to “Anyone with link can
  # view” in order for the feed to work.
  # Template: 'https://sheets.googleapis.com/v4/spreadsheets/<SHEET>/values/<TAB>?key=<KEY>'
@@ -68,7 +72,32 @@ process:
  type:
    plugin: default_value
    default_value: projet
-  title: title
+  title:
+    plugin: skip_on_empty
+    method: row
+    source: title
+    message: 'Projet ignoré : « Nom du projet » vide.'
+  body/value: body_html
+  body/format:
+    plugin: default_value
+    default_value: wysiwyg
+  field_images:
+    plugin: sub_process
+    source: images
+    process:
+      target_id:
+        plugin: image_resize_import
+        source: uri
+        destination: destination
+        max_width: 1920
+        max_height: 1920
+        reuse: true
+        skip_on_missing_source: true
+      alt: alt
+      image_field_caption/value: caption
+      image_field_caption/format:
+        plugin: default_value
+        default_value: wysiwyg
  field_dates/value:
    plugin: leshed_date
    source: date_start
@@ -0,0 +1,179 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Drupal\migrate_leshed\Plugin\migrate\process;
+
+use Drupal\Core\Entity\EntityTypeManagerInterface;
+use Drupal\Core\File\FileSystemInterface;
+use Drupal\Core\Image\ImageFactory;
+use Drupal\Core\Plugin\ContainerFactoryPluginInterface;
+use Drupal\migrate\Attribute\MigrateProcess;
+use Drupal\migrate\MigrateExecutableInterface;
+use Drupal\migrate\MigrateSkipRowException;
+use Drupal\migrate\ProcessPluginBase;
+use Drupal\migrate\Row;
+use Symfony\Component\DependencyInjection\ContainerInterface;
+
+/**
+ * Copies a local image, downscaling it to a web-friendly size, as a File.
+ *
+ * The source value is a readable file path/URI. The image is scaled down
+ * (aspect ratio preserved, never upscaled) to fit within max_width x
+ * max_height, saved into the destination directory and registered as a
+ * managed file. Returns the file id.
+ *
+ * Configuration:
+ * - destination: (required) destination directory URI, e.g. 'public://projets/1/'.
+ * - max_width: maximum width in pixels (default 1920).
+ * - max_height: maximum height in pixels (default 1920).
+ * - uid: owner uid of the created file (default 0).
+ * - reuse: reuse an existing managed file at the destination URI (default TRUE).
+ * - skip_on_missing_source: skip the value when the source is missing
+ *   (default TRUE) instead of failing.
+ *
+ * @code
+ * field_images/target_id:
+ *   plugin: image_resize_import
+ *   source: uri
+ *   destination: destination
+ *   max_width: 1920
+ *   max_height: 1920
+ * @endcode
+ */
+#[MigrateProcess(id: 'image_resize_import')]
+final class ImageResizeImport extends ProcessPluginBase implements ContainerFactoryPluginInterface {
+
+  /**
+   * Per-run cache of already resolved destination URIs to file ids.
+   *
+   * @var array<string, int>
+   */
+  private array $fidCache = [];
+
+  public function __construct(
+    array $configuration,
+    $plugin_id,
+    $plugin_definition,
+    private readonly FileSystemInterface $fileSystem,
+    private readonly ImageFactory $imageFactory,
+    private readonly EntityTypeManagerInterface $entityTypeManager,
+  ) {
+    $configuration += [
+      'max_width' => 1920,
+      'max_height' => 1920,
+      'uid' => 0,
+      'reuse' => TRUE,
+      'skip_on_missing_source' => TRUE,
+    ];
+    parent::__construct($configuration, $plugin_id, $plugin_definition);
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create(ContainerInterface $container, array $configuration, $plugin_id, $plugin_definition): self {
+    return new self(
+      $configuration,
+      $plugin_id,
+      $plugin_definition,
+      $container->get('file_system'),
+      $container->get('image.factory'),
+      $container->get('entity_type.manager'),
+    );
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function transform($value, MigrateExecutableInterface $migrate_executable, Row $row, $destination_property): ?int {
+    $source = is_string($value) ? trim($value) : '';
+    if ($source === '') {
+      return NULL;
+    }
+
+    // Stream wrappers support is_file() directly; realpath() is only needed as
+    // a fallback for schemes it cannot resolve.
+    if (!is_file($source) && !is_file($this->fileSystem->realpath($source) ?: $source)) {
+      // Skipping throws MigrateSkipRowException because, inside a sub_process,
+      // only that exception drops the whole image item cleanly (SubProcess
+      // catches it and `continue`s); MigrateSkipProcessException would instead
+      // leave a broken delta with alt/caption but no target_id.
+      return $this->skipImage($migrate_executable, "Source image '$source' is missing.");
+    }
+
+    $directory = $this->getPropertyValue($this->configuration['destination'], $row);
+    if (!is_string($directory) || $directory === '') {
+      return $this->skipImage($migrate_executable, "Missing 'destination' for image '$source'.");
+    }
+    $directory = rtrim($directory, '/') . '/';
+    $destination = $directory . $this->fileSystem->basename($source);
+
+    // Reuse an already imported file (idempotency), memoized for the run.
+    if ($this->configuration['reuse']) {
+      if (isset($this->fidCache[$destination])) {
+        return $this->fidCache[$destination];
+      }
+      $existing = $this->entityTypeManager->getStorage('file')
+        ->loadByProperties(['uri' => $destination]);
+      if ($existing) {
+        return $this->fidCache[$destination] = (int) reset($existing)->id();
+      }
+    }
+
+    // Any infrastructure failure only drops the current image (logged), so a
+    // single unreadable/oversized file never aborts the whole batch.
+    try {
+      if (!$this->fileSystem->prepareDirectory($directory, FileSystemInterface::CREATE_DIRECTORY | FileSystemInterface::MODIFY_PERMISSIONS)) {
+        return $this->skipImage($migrate_executable, "Cannot prepare directory '$directory'.");
+      }
+
+      $image = $this->imageFactory->get($source);
+      if (!$image->isValid()) {
+        return $this->skipImage($migrate_executable, "Source '$source' is not a valid image.");
+      }
+      // Downscale only: scale() never upscales when upscale is FALSE.
+      $image->scale((int) $this->configuration['max_width'], (int) $this->configuration['max_height']);
+      if (!$image->save($destination)) {
+        return $this->skipImage($migrate_executable, "Failed to write resized image to '$destination'.");
+      }
+
+      $file = $this->entityTypeManager->getStorage('file')->create([
+        'uri' => $destination,
+        'uid' => (int) $this->getPropertyValue($this->configuration['uid'], $row),
+        'status' => 1,
+      ]);
+      $file->save();
+    }
+    catch (\Throwable $e) {
+      return $this->skipImage($migrate_executable, "Import of image '$source' failed: " . $e->getMessage());
+    }
+
+    return $this->fidCache[$destination] = (int) $file->id();
+  }
+
+  /**
+   * Logs a message and skips the current image item.
+   *
+   * @throws \Drupal\migrate\MigrateSkipRowException
+   */
+  private function skipImage(MigrateExecutableInterface $migrate_executable, string $message): never {
+    $migrate_executable->saveMessage($message);
+    throw new MigrateSkipRowException($message, FALSE);
+  }
+
+  /**
+   * Resolves a configuration value, supporting destination (@) references.
+   */
+  private function getPropertyValue(mixed $key, Row $row): mixed {
+    if (is_string($key) && str_starts_with($key, '@')) {
+      return $row->get(substr($key, 1));
+    }
+    if (is_string($key) && $row->hasSourceProperty($key)) {
+      return $row->getSourceProperty($key);
+    }
+
+    return $key;
+  }
+
+}
@@ -18,8 +18,9 @@ use Symfony\Component\DependencyInjection\ContainerInterface;
 * Input: a string such as "Julie Faitot et Jonathan Loppin" or
 * "Sonja Beaudouin, Alexandre Delabrière". Names are split on "et" and
 * commas. For each name, the first word is used as 'field_prenom' and the
- * rest as the node title. An existing node with the same title and prenom
- * is reused, otherwise it is created. Returns an array of node IDs.
+ * rest as 'field_nom'. The node title is generated from those two fields by
+ * the auto_entitylabel module on save. An existing node with the same nom and
+ * prenom is reused, otherwise it is created. Returns an array of node IDs.
 *
 * When 'type_de_personne' is set in the configuration, a taxonomy term with
 * that name is found or created (without duplicates) in the
@@ -103,7 +104,7 @@ final class PersonneGenerate extends ProcessPluginBase implements ContainerFacto

    $query = $storage->getQuery()
      ->condition('type', 'personne')
-      ->condition('title', $nom)
+      ->condition('field_nom', $nom)
      ->accessCheck(FALSE)
      ->range(0, 1);
    if ($prenom !== '') {
@@ -121,9 +122,11 @@ final class PersonneGenerate extends ProcessPluginBase implements ContainerFacto
      return $nid;
    }

+    // The node title is generated from field_prenom + field_nom by
+    // auto_entitylabel on save; it is intentionally not set here.
    $values = [
      'type' => 'personne',
-      'title' => $nom,
+      'field_nom' => $nom,
      'field_prenom' => $prenom === '' ? NULL : $prenom,
      'status' => 1,
      'uid' => $this->configuration['author_uid'] ?? 3,
@@ -4,11 +4,31 @@ declare(strict_types=1);

 namespace Drupal\migrate_leshed\Plugin\migrate\source;

+use Drupal\Core\File\FileSystemInterface;
 use Drupal\migrate\Attribute\MigrateSource;
-use Drupal\migrate_plus\Plugin\migrate\source\Url;
+use Drupal\migrate\Plugin\MigrationInterface;
 use Drupal\migrate\Row;
+use Drupal\migrate_plus\DataParserPluginManager;
+use Drupal\migrate_plus\Plugin\migrate\source\Url;
+use League\CommonMark\CommonMarkConverter;
+use Symfony\Component\DependencyInjection\ContainerInterface;

 /**
+ * Source plugin for projects, enriched with body and images from the drive.
+ *
+ * On top of the Google Sheets columns, this source reads the local files
+ * synced from the shared Google Drive (see the `sync_leshed_drive` Makefile
+ * target) that live under public://drive_shared_leshed. Files are grouped in
+ * per-project folders named "Page{UID}" (optionally suffixed, e.g.
+ * "Page1_Complet"), where {UID} matches the sheet UID column:
+ * - "{UID}_texte_*.md": the body, written in Markdown, converted to HTML.
+ * - "{UID}_images|Images/": one folder of images, each "{name}.jpg" paired
+ *   with a "{name}_legende.md" caption used as the (required) alt text.
+ *
+ * These are exposed as two extra source properties consumed by the migration:
+ * - body_html: the converted HTML string (or NULL).
+ * - images: a list of ['uri' => string, 'alt' => string] items.
+ *
 * @MigrateSource(
 *   id = "projet_node"
 * )
@@ -16,18 +36,65 @@ use Drupal\migrate\Row;
 #[MigrateSource(id: 'projet_node')]
 final class ProjetNode extends Url {

+  /**
+   * Stream wrapper URI of the synced drive folder.
+   */
+  private const DRIVE_URI = 'public://drive_shared_leshed';
+
+  /**
+   * Image file extensions to import into field_images.
+   */
+  private const IMAGE_EXTENSIONS = ['png', 'gif', 'jpg', 'jpeg', 'webp'];
+
+  /**
+   * Default destination directory when no constant is configured.
+   */
+  private const DEFAULT_IMAGE_DESTINATION = 'public://projets/';
+
+  /**
+   * Lazily instantiated Markdown converter.
+   */
+  private ?CommonMarkConverter $markdownConverter = NULL;
+
+  /**
+   * Memoized map of UID => project folder URI, built once from the drive root.
+   *
+   * @var array<string, string>|null
+   */
+  private ?array $folderMap = NULL;
+
+  public function __construct(
+    array $configuration,
+    $plugin_id,
+    $plugin_definition,
+    MigrationInterface $migration,
+    DataParserPluginManager $parserPluginManager,
+    private readonly FileSystemInterface $fileSystem,
+  ) {
+    parent::__construct($configuration, $plugin_id, $plugin_definition, $migration, $parserPluginManager);
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public static function create($container, array $configuration, $plugin_id, $plugin_definition, ?MigrationInterface $migration = NULL): self {
+    return new self(
+      $configuration,
+      $plugin_id,
+      $plugin_definition,
+      $migration,
+      $container->get('plugin.manager.migrate_plus.data_parser'),
+      $container->get('file_system'),
+    );
+  }
+
  /**
   * {@inheritdoc}
   */
  public function fields(): array {
-    // theese are source fields
-    return [
-      'id' => $this->t('Project ID'),
-      'title' => $this->t('Project Title'),
-      'date_start' =>  $this->t('Project date start'),
-      'date_end' =>  $this->t('Project date end'),
-      'projet_type' =>  $this->t('Type de projet'),
-      'curator' =>  $this->t('Commissaria')
+    return parent::fields() + [
+      'body_html' => $this->t('Project body (HTML converted from Markdown)'),
+      'images' => $this->t('Project images with captions'),
    ];
  }

@@ -45,16 +112,211 @@ final class ProjetNode extends Url {
  /**
   * {@inheritdoc}
   */
-  public function prepareRow(Row $row) {
-    // $nid = $row->getSourceProperty('entity_id');
-    // $vid = $row->getSourceProperty('revision_id');
-    // $type = $row->getSourceProperty('type');
-    // $language = $row->getSourceProperty('language');
-    // $title = $row->getSourceProperty('title');
-    // // drush_print('-- '.$nid."\t".$title."\t".$language);
+  public function prepareRow(Row $row): bool {
+    // Let the parent (and prepare-row event subscribers) decide first: no need
+    // to hit the filesystem for a row that will be skipped.
+    if (!parent::prepareRow($row)) {
+      return FALSE;
+    }

+    // Always define both properties so the process pipeline never references
+    // an undefined source key, even when no drive folder exists yet.
+    $row->setSourceProperty('body_html', NULL);
+    $row->setSourceProperty('images', []);

+    $uid = trim((string) $row->getSourceProperty('id'));
+    $folder = $uid !== '' ? $this->findProjectFolder($uid) : NULL;
+    if ($folder !== NULL) {
+      $row->setSourceProperty('body_html', $this->readBody($folder, $uid));
+      $row->setSourceProperty('images', $this->readImages($folder, $uid, (string) $row->getSourceProperty('title')));
+    }

-    return parent::prepareRow($row);
+    return TRUE;
  }
+
+  /**
+   * Returns the public:// URI of the project folder, or NULL when missing.
+   */
+  private function findProjectFolder(string $uid): ?string {
+    return $this->getFolderMap()[$uid] ?? NULL;
+  }
+
+  /**
+   * Builds (once) the UID => project folder URI map from the drive root.
+   *
+   * @return array<string, string>
+   *   Map keyed by the numeric part of each "Page{UID}[_suffix]" folder.
+   */
+  private function getFolderMap(): array {
+    if ($this->folderMap !== NULL) {
+      return $this->folderMap;
+    }
+
+    $this->folderMap = [];
+    $base = $this->fileSystem->realpath(self::DRIVE_URI);
+    if ($base === FALSE || !is_dir($base)) {
+      return $this->folderMap;
+    }
+    // Match "Page{N}" exactly or with a suffix ("_Complet", ...). Sorting keeps
+    // the resolution deterministic when e.g. "Page5" and "Page5_Complet"
+    // coexist (the plain, longer-suffixed folder wins consistently).
+    $entries = scandir($base) ?: [];
+    sort($entries, SORT_NATURAL | SORT_FLAG_CASE);
+    foreach ($entries as $entry) {
+      if ($entry === '.' || $entry === '..' || !is_dir($base . '/' . $entry)) {
+        continue;
+      }
+      if (preg_match('/^Page(\d+)(_.*)?$/', $entry, $matches)) {
+        $this->folderMap[$matches[1]] = self::DRIVE_URI . '/' . $entry;
+      }
+    }
+
+    return $this->folderMap;
+  }
+
+  /**
+   * Reads the Markdown body file and converts it to HTML.
+   */
+  private function readBody(string $folderUri, string $uid): ?string {
+    $dir = $this->fileSystem->realpath($folderUri);
+    if ($dir === FALSE) {
+      return NULL;
+    }
+    // Prefer "{UID}_texte*.md", fall back to any Markdown file at the root.
+    $preferred = '/^' . preg_quote($uid, '/') . '_texte.*\.md$/i';
+    $entries = scandir($dir) ?: [];
+    sort($entries, SORT_NATURAL | SORT_FLAG_CASE);
+    $markdown = NULL;
+    foreach ($entries as $entry) {
+      if (!is_file($dir . '/' . $entry) || !preg_match('/\.md$/i', $entry)) {
+        continue;
+      }
+      if (preg_match($preferred, $entry)) {
+        $markdown = $dir . '/' . $entry;
+        break;
+      }
+      $markdown ??= $dir . '/' . $entry;
+    }
+    if ($markdown === NULL) {
+      return NULL;
+    }
+
+    $contents = file_get_contents($markdown);
+    if ($contents === FALSE || trim($contents) === '') {
+      return NULL;
+    }
+
+    return (string) $this->getMarkdownConverter()->convert($contents);
+  }
+
+  /**
+   * Builds the list of images (URI + alt) for the project.
+   *
+   * @return array<int, array{uri: string, alt: string, caption: string, destination: string}>
+   *   The images to import.
+   */
+  private function readImages(string $folderUri, string $uid, string $title): array {
+    $dir = $this->fileSystem->realpath($folderUri);
+    if ($dir === FALSE) {
+      return [];
+    }
+    // The destination is passed through each item because a sub_process row
+    // cannot read the parent migration's `constants`. It is namespaced per
+    // project (UID) so identically-named files from different projects never
+    // collide when file_import reuses existing files (reuse: true).
+    $base = $this->configuration['constants']['images_destination']
+      ?? self::DEFAULT_IMAGE_DESTINATION;
+    $destination = rtrim($base, '/') . '/' . $uid . '/';
+    // Locate the images subfolder ("{UID}_images", case-insensitive).
+    $imagesDirUri = NULL;
+    $pattern = '/^' . preg_quote($uid, '/') . '_images$/i';
+    foreach (scandir($dir) ?: [] as $entry) {
+      if (preg_match($pattern, $entry) && is_dir($dir . '/' . $entry)) {
+        $imagesDirUri = $folderUri . '/' . $entry;
+        break;
+      }
+    }
+    if ($imagesDirUri === NULL) {
+      return [];
+    }
+
+    $imagesDir = $this->fileSystem->realpath($imagesDirUri);
+    if ($imagesDir === FALSE) {
+      return [];
+    }
+    $images = [];
+    $files = scandir($imagesDir) ?: [];
+    sort($files, SORT_NATURAL | SORT_FLAG_CASE);
+    foreach ($files as $file) {
+      $extension = strtolower(pathinfo($file, PATHINFO_EXTENSION));
+      if (!in_array($extension, self::IMAGE_EXTENSIONS, TRUE)) {
+        continue;
+      }
+      $name = pathinfo($file, PATHINFO_FILENAME);
+      $legende = $this->readCaptionFile($imagesDir . '/' . $name . '_legende.md');
+      // Plain text (required alt), falling back to the project title. Capped to
+      // the image field's alt column length (varchar 512).
+      $alt = $legende !== '' ? $this->toPlainText($legende) : $title;
+      $alt = mb_substr($alt, 0, 512);
+      // Rich caption (image_field_caption): Markdown converted to HTML.
+      $caption = $legende !== '' ? (string) $this->getMarkdownConverter()->convert($legende) : '';
+      $images[] = [
+        'uri' => $imagesDirUri . '/' . $file,
+        'alt' => $alt,
+        'caption' => $caption,
+        'destination' => $destination,
+      ];
+    }
+
+    return $images;
+  }
+
+  /**
+   * Returns the raw contents of a caption (legende) file, or '' when missing.
+   */
+  private function readCaptionFile(string $captionPath): string {
+    if (!is_file($captionPath)) {
+      return '';
+    }
+    $contents = file_get_contents($captionPath);
+    if ($contents === FALSE) {
+      return '';
+    }
+    // Drive exports are occasionally Windows-1252 rather than UTF-8.
+    if (!mb_check_encoding($contents, 'UTF-8')) {
+      $contents = mb_convert_encoding($contents, 'UTF-8', 'Windows-1252');
+    }
+
+    return trim($contents);
+  }
+
+  /**
+   * Reduces Markdown/HTML text to a single trimmed plain-text line (for alt).
+   */
+  private function toPlainText(string $text): string {
+    $text = html_entity_decode(strip_tags($text), ENT_QUOTES | ENT_HTML5, 'UTF-8');
+    // Drop Markdown emphasis/heading markers left after tag stripping.
+    $text = preg_replace('/[*_#`>]+/u', '', $text) ?? $text;
+
+    return trim((string) preg_replace('/\s+/u', ' ', $text));
+  }
+
+  /**
+   * Returns the shared Markdown converter.
+   *
+   * Raw HTML embedded in the source Markdown (alignment <div>, <br>, ...) is
+   * stripped: only the Markdown itself is converted to HTML, keeping the body
+   * clean and safe for the wysiwyg format.
+   */
+  private function getMarkdownConverter(): CommonMarkConverter {
+    if ($this->markdownConverter === NULL) {
+      $this->markdownConverter = new CommonMarkConverter([
+        'html_input' => 'strip',
+        'allow_unsafe_links' => FALSE,
+      ]);
+    }
+
+    return $this->markdownConverter;
+  }
+
 }