almost every needed thing for corpus is ready, remains that all files are not migrated because of filepath :(

sites/all/modules/figli/edlp_migrate/config/install/migrate_plus.migration.d6_edlp_allfiles.yml

@@ -6,13 +6,13 @@ migration_tags:
   - Drupal 6
 migration_group: d6_edlp
 source:
-  plugin: d6_file
+  plugin: d6_edlp_file
   constants:
     # The tool configuring this migration must set source_base_path. It
     # represents the fully qualified path relative to which URIs in the files
     # table are specified, and must end with a /. See source_full_path
     # configuration in this migration's process pipeline as an example.
-    source_base_path: '../drupal6/'
+    source_base_path: '../drupal6'
 
 process:
     # If you are using both this migration and d6_user_picture_file in a custom

sites/all/modules/figli/edlp_migrate/config/install/migrate_plus.migration.d6_edlp_corpus.yml

@@ -40,109 +40,132 @@ process:
       plugin: default_value
       default_value: 1
 
-  status: status
+  # status: status
   created: created
   changed: changed
   promote: promote
   sticky: sticky
-  # 'body/format':
-  #   plugin: migration_lookup
-  #   migration: d6_filter_format
-  #   source: format
+  revision_uid: revision_uid
+  revision_log: log
+  revision_timestamp: timestamp
+
+  'field_description/value': body
+  'field_description/format':
+    plugin: default_value
+    default_value: wysiwyg
 
   # ARTICLE (body)
   # node ref -> text
-  # 'body/value': here we have to get the body value of article
-  # 'body/summary': teaser
+  'body/value': article
+  'body/format':
+    plugin: default_value
+    default_value: wysiwyg
 
-  revision_uid: revision_uid
-  revision_log: log
-  revision_timestamp: timestamp
-  #
-  # custom fields
-  # workflow
 
-  # text -> entity ref terms
-  # field_artiste: field_locuteur_s
-  # field_locuteurs_s:
-  #   plugin: entity_generate
-  #   entity_type: taxonomy_term
-  #   bundle_key: vid
-  #   bundle: locuteurs
-  #   source: field_artiste
-    # value_key: field_artiste_value
+  # taxo term -> entity ref terms
+  # use this one for fetchCol in plugin
+  field_entrees:
+    plugin: migration_lookup
+    migration: d6_edlp_entrees
+    source: entrees
+
+  # use this one for fetchAll in plugin
+  # field_entrees:
+  #   -
+  #     plugin: iterator
+  #     source: entrees
+  #     process:
+  #       id: tid
+  #   -
+  #     plugin: flatten
+  #   -
+  #     plugin: migration_lookup
+  #     migration: d6_edlp_entrees
+  #     no_stub: true
+
+  # field_entrees:
+  #   plugin: iterator
+  #   source: entrees
+  #   process:
+  #     plugin: migration_lookup
+  #     migration: d6_edlp_entrees
+
+
+  # field_genre -> field_genres
+  field_genres:
+    plugin: migration_lookup
+    migration: d6_edlp_genres
+    source: genres
+
+  # field_genres:
+  #   -
+  #     plugin: iterator
+  #     source: field_genre
+  #     process:
+  #       id: value
+  #   -
+  #     plugin: flatten
+  #   -
+  #     plugin: migration_lookup
+  #     migration: d6_edlp_genres
 
 
+  # taxo term -> entity ref terms
+  field_langues:
+    plugin: migration_lookup
+    migration: d6_edlp_langues
+    source: langues
+
+  # field_artiste: field_locuteurs
   # text -> entity ref terms
-  # field_compositeur: field_collectionneur_s
-  # field_collectionneur_s:
+  # instead of entity_generate, i create the term in prepareRow from plugin
+  field_locuteurs: locuteurs
+  # field_locuteurs:
   #   plugin: entity_generate
   #   entity_type: taxonomy_term
   #   bundle_key: vid
-  #   bundle: collectionneurs
-  #   source: field_compositeur
-    # value_key: field_compositeur_value
+  #   bundle: locuteurs
+  #   source: locuteurs
+  #   # value_key: field_artiste_value
 
-  # taxo term -> entity ref terms
-  # field_entree_s:
-  #   plugin: iterator
-  #   source:
-  #   plugin: migration_lookup
-  #   migration: d6_edlp_entrees
-  #   source: name
 
-  # field_genre: field_genre_s
-  field_genre_s:
-    -
-      plugin: iterator
-      source: field_genre
-      process:
-        tid: value
-    -
-      plugin: flatten
-    # -
-    #   plugin: migration_lookup
-    #   migration: d6_edlp_genres
-
-    # plugin: migration
-    # migration: d6_edlp_genres
-    # source: field_genre
-
-    # plugin: iterator
-    # source: field_genre
-    # process:
-    #   target_id:
-    #     plugin: migration_lookup
-    #     migration: d6_edlp_genres
-    #     source: value
+  # field_compositeur: field_collectionneurs
+  # text -> entity ref terms
+  # instead of entity_generate, i create the term in prepareRow from plugin
+  field_collectionneurs: collectionneurs
 
-  # taxo term -> entity ref terms
-  # field_langue: field_langue_s
 
-  # file
-  # field_mp3: field_fichier_son
-  field_fichier_son:
+  # # file
+  # # field_mp3 -> field_son
+  field_son:
     plugin: migration_lookup
     migration: d6_edlp_allfiles
-    source: fid
+    source: sons_fid
+  # field_son:
+  # #   plugin: migration_lookup
+  # #   migration: d6_edlp_allfiles
+  # #   source: fid
+  #   plugin: iterator
+  #   source: field_mp3
+  #   process:
+  #     target_id:
+  #       plugin: migration_lookup
+  #       migration: d6_edlp_allfiles
+  #       source: fid
 
-    # plugin: iterator
-    # source: field_mp3
-    # process:
-    #   target_id:
-    #     plugin: migration_lookup
-    #     migration: d6_edlp_allfiles
-    #     source: fid
 
-  field_notes: body
 
+  # workflow
+  status:
+    plugin: static_map
+    source: workflow
+    map:
+      1: 0
+      2: 0
+      3: 1
+      4: 1
 
 
-#  unmapped d6 fields.
-#  tnid
-#  translate
-#  moderate
-#  comment
 
 destination:
   plugin: entity:node

sites/all/modules/figli/edlp_migrate/src/Plugin/migrate/source/D6EdlpCorpus.php

@@ -95,6 +95,11 @@ class D6EdlpCorpus extends D6Node {
    * {@inheritdoc}
    */
   public function prepareRow(Row $row) {
+    drush_print(' ');
+    drush_print('- - - - - - - - - - - - - - - - - - - - - - - -');
+    drush_print($row->getSourceProperty('title'));
+
+
     // format = 0 can happen when the body field is hidden. Set the format to 1
     // to avoid migration map issues (since the body field isn't used anyway).
     if ($row->getSourceProperty('format') === '0') {
@@ -104,17 +109,119 @@ class D6EdlpCorpus extends D6Node {
     if ($this->moduleExists('content') && $this->getModuleSchemaVersion('content') >= 6001) {
       foreach ($this->getFieldValues($row) as $field => $values) {
         $row->setSourceProperty($field, $values);
-        // \Drupal::logger('edlp_migrate')->debug(print_r($values));
-        // print_r($field);
-        // print_r("\n");
-        // print_r($values);
       }
     }
+    $row_nid = $row->getSourceProperty('nid');
+
 
     // Make sure we always have a translation set.
     if ($row->getSourceProperty('tnid') == 0) {
-      $row->setSourceProperty('tnid', $row->getSourceProperty('nid'));
+      $row->setSourceProperty('tnid', $row_nid);
+    }
+    // get the article content
+    // drush_print_r($row->getSourceProperty('field_article'));
+    $article_nid = $row->getSourceProperty('field_article')[0]['nid'];
+    if($article_nid){
+      drush_print($article_nid);
+      $query = $this->select('content_type_episode', 'ctp');
+      // // $query->fields('cte', ['field_entree_value']);
+      $query->condition('ctp.nid', $article_nid);
+      $query->join('node_revisions', 'nr', 'nr.vid = ctp.vid');
+      $query->addField('nr', 'body');
+      $article = $query->execute()
+        ->fetchField();
+      // drush_print_r($article);
+      $row->setSourceProperty('article', $article);
+    }
+
+    // get the entree values
+    $query = $this->select('term_node', 'tn');
+    // $query->fields('cte', ['field_entree_value']);
+    $query->condition('tn.nid', $row_nid);
+    $query->join('term_data', 'td', 'td.tid = tn.tid');
+    $query->condition('td.vid', 2); // vid of voc entrees
+    $query->addField('tn', 'tid');
+    $query->distinct();
+    $results = $query->execute()->fetchAll();
+
+    $entrees = [];
+    foreach ($results as $key => $value) {
+      $entrees[] = array($value['tid']);
+    }
+    $row->setSourceProperty('entrees', $entrees);
+
+    // Genres
+    $genres = [];
+    foreach ($row->getSourceProperty('field_genre') as $key => $value) {
+      $genres[] = array($value['value']);
+    }
+    $row->setSourceProperty('genres', $genres);
+
+    $langues = [];
+    foreach ($row->getSourceProperty('field_langue') as $key => $value) {
+      $langues[] = array($value['value']);
     }
+    $row->setSourceProperty('langues', $langues);
+
+    // locuteurs
+    $locuteurs = [];
+    foreach ($row->getSourceProperty('field_artiste') as $key => $value) {
+      $term_name = $value['value'];
+      // check if term already exists in configured vocabulary
+      $term = array_shift(taxonomy_term_load_multiple_by_name($term_name, 'locuteurs'));
+      if(!$term){
+        // if not create the term and get the tid
+        $term = \Drupal\taxonomy\Entity\Term::create([
+              'vid' => 'locuteurs',
+              'name' => $term_name,
+            ]);
+        $term->save();
+      }
+      // record the tid in array
+      $locuteurs[] = $term->get('tid')->value;
+    }
+    $row->setSourceProperty('locuteurs', $locuteurs);
+
+    // collectionneurs
+    $collectionneurs = [];
+    foreach ($row->getSourceProperty('field_compositeur') as $key => $value) {
+      $term_name = $value['value'];
+      // check if term already exists in configured vocabulary
+      $term = array_shift(taxonomy_term_load_multiple_by_name($term_name, 'collectionneurs'));
+      if(!$term){
+        // if not create the term and get the tid
+        $term = \Drupal\taxonomy\Entity\Term::create([
+              'vid' => 'collectionneurs',
+              'name' => $term_name,
+            ]);
+        $term->save();
+      }
+      // record the tid in array
+      $collectionneurs[] = $term->get('tid')->value;
+    }
+    $row->setSourceProperty('collectionneurs', $collectionneurs);
+
+    // son
+    // drush_print('field_mp3 : ', 0, null, false);
+    // drush_print_r($row->getSourceProperty('field_mp3'));
+    $sons = [];
+    foreach ($row->getSourceProperty('field_mp3') as $key => $value) {
+      $sons[] = array($value['fid']);
+    }
+    $row->setSourceProperty('sons_fid', $sons);
+
+    // TODO: get the workflow status
+    // get the entree values
+    $query = $this->select('workflow_node', 'wn');
+    $query->fields('wn', ['sid']);
+    $query->condition('wn.nid', $row_nid);
+    $results = $query->execute()->fetchField();
+
+    drush_print('workflow query results : ', 0, null, false);
+    drush_print_r($results);
+
+    $row->setSourceProperty('workflow', $results);
+
 
     $row = parent::prepareRow($row);
     return $row;

sites/all/modules/figli/edlp_migrate/src/Plugin/migrate/source/D6EdlpFile.php

@@ -0,0 +1,111 @@
+<?php
+
+namespace Drupal\edlp_migrate\Plugin\migrate\source;
+
+use Drupal\migrate\Row;
+use Drupal\migrate_drupal\Plugin\migrate\source\DrupalSqlBase;
+
+/**
+ * Drupal 6 file source from database.
+ *
+ * @MigrateSource(
+ *   id = "d6_edlp_file"
+ * )
+ */
+class D6EdlpFile extends DrupalSqlBase {
+
+  /**
+   * The file directory path.
+   *
+   * @var string
+   */
+  protected $filePath;
+
+  /**
+   * The temporary file path.
+   *
+   * @var string
+   */
+  protected $tempFilePath;
+
+  /**
+   * Flag for private or public file storage.
+   *
+   * @var bool
+   */
+  protected $isPublic;
+
+  /**
+   * {@inheritdoc}
+   */
+  public function query() {
+    return $this->select('files', 'f')
+      ->fields('f')
+      ->orderBy('timestamp')
+      // If two or more files have the same timestamp, they'll end up in a
+      // non-deterministic order. Ordering by fid (or any other unique field)
+      // will prevent this.
+      ->orderBy('f.fid');
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  protected function initializeIterator() {
+    $site_path = isset($this->configuration['site_path']) ? $this->configuration['site_path'] : 'sites/default';
+    // $filepath = $this->variableGet('file_directory_path', $site_path . '/files') . '/';
+    // $filepath = str_replace('//', '/', $filepath);
+    // drush_print('filepath : '.$filepath);
+    $this->filePath = $this->variableGet('file_directory_path', $site_path . '/files') . '/';
+
+    $this->tempFilePath = $this->variableGet('file_directory_temp', '/tmp') . '/';
+
+    // FILE_DOWNLOADS_PUBLIC == 1 and FILE_DOWNLOADS_PRIVATE == 2.
+    $this->isPublic = $this->variableGet('file_downloads', 1) == 1;
+    return parent::initializeIterator();
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function prepareRow(Row $row) {
+    drush_print("D6EdlpFile");
+    // drush_print_r($row);
+
+    $filepath = $row->getSourceProperty('filepath');
+    $filepath = str_replace('//', '/', $filepath);
+    $filepath = preg_replace('/^\//', '', $filepath);
+    drush_print('filePath : '.$filepath);
+    $row->setSourceProperty('filepath', $filepath);
+
+    $row->setSourceProperty('file_directory_path', $this->filePath);
+    $row->setSourceProperty('temp_directory_path', $this->tempFilePath);
+    $row->setSourceProperty('is_public', $this->isPublic);
+    return parent::prepareRow($row);
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function fields() {
+    return [
+      'fid' => $this->t('File ID'),
+      'uid' => $this->t('The {users}.uid who added the file. If set to 0, this file was added by an anonymous user.'),
+      'filename' => $this->t('File name'),
+      'filepath' => $this->t('File path'),
+      'filemime' => $this->t('File MIME Type'),
+      'status' => $this->t('The published status of a file.'),
+      'timestamp' => $this->t('The time that the file was added.'),
+      'file_directory_path' => $this->t('The Drupal files path.'),
+      'is_public' => $this->t('TRUE if the files directory is public otherwise FALSE.'),
+    ];
+  }
+  /**
+   * {@inheritdoc}
+   */
+  public function getIds() {
+    $ids['fid']['type'] = 'integer';
+    return $ids;
+  }
+
+}