Adds custom migration source and process plugins.

src/Plugin/migrate/process/ExtractFileContents.php

@@ -0,0 +1,54 @@
+<?php
+
+namespace Drupal\cdesai_migrate\Plugin\migrate\process;
+
+use Drupal\migrate\MigrateSkipRowException;
+use Drupal\migrate\Row;
+use Drupal\migrate\ProcessPluginBase;
+use Drupal\migrate\MigrateExecutableInterface;
+use Drush\Drush;
+
+/**
+ * Provides a 'Extract File Contents' migrate process plugin.
+ *
+ * Extracts contents of a file.
+ *
+ * @MigrateProcessPlugin(
+ *  id = "extract_file_contents"
+ * )
+ */
+class ExtractFileContents extends ProcessPluginBase {
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\migrate\MigrateSkipRowException
+   */
+  public function transform($value, MigrateExecutableInterface $migrate_executable, Row $row, $destination_property) {
+    $error = FALSE;
+    $page = $row->get('page');
+
+    if (is_null($value) || trim($value) === '') {
+      $error = "Skipping because content filepath is empty for {$page}.";
+    }
+    elseif (!is_file($value)) {
+      $error = "Skipping because content file is missing for {$page}.";
+    }
+    else {
+      $contents = file_get_contents($value);
+      if ($contents !== FALSE) {
+        return $contents;
+      }
+
+      $error = "Skipping because content file is unreadable for {$page}.";
+    }
+
+    if ($error) {
+      Drush::output()->writeln($error);
+      throw new MigrateSkipRowException($error);
+    }
+
+    return '';
+  }
+
+}

src/Plugin/migrate/process/ExtractFromParagraph.php

@@ -0,0 +1,55 @@
+<?php
+
+namespace Drupal\cdesai_migrate\Plugin\migrate\process;
+
+use Drupal\migrate\ProcessPluginBase;
+use Drupal\migrate\MigrateException;
+use Drupal\migrate\MigrateExecutableInterface;
+use Drupal\migrate\Row;
+use Drupal\paragraphs\Entity\Paragraph;
+
+/**
+ * Provides a 'Extract Paragraph' migrate process plugin.
+ *
+ * Extracts content of field_body from a paragraph entity.
+ *
+ * @MigrateProcessPlugin(
+ *   id = "extract_cdesai_migrate_paragraph",
+ *   handle_multiples = TRUE
+ * )
+ */
+class ExtractFromParagraph extends ProcessPluginBase {
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\migrate\MigrateException
+   * @throws \Drupal\Core\TypedData\Exception\MissingDataException
+   */
+  public function transform($value, MigrateExecutableInterface $migrate_executable, Row $row, $destination_property) {
+    if (!is_array($value)) {
+      throw new MigrateException(sprintf("Input should be an array, instead it was of type '%s'", gettype($value)));
+    }
+
+    if (count($value) != 2) {
+      throw new MigrateException(sprintf("Input array should have exactly two (2) elements, instead it had '%s'", count($value)));
+    }
+
+    $paragraph = Paragraph::load($value[0]);
+
+    if ($paragraph->hasField('field_body')) {
+      $body = $paragraph->get('field_body');
+      $item = $body->get(0);
+
+      if ($item) {
+        return $item->getValue()['value'];
+      }
+    }
+    else {
+      throw new MigrateException("Paragraph does not have field_body.");
+    }
+
+    throw new MigrateException("Unable to find paragraph.");
+  }
+
+}

src/Plugin/migrate/process/GenerateRedirectUri.php

@@ -0,0 +1,62 @@
+<?php
+
+namespace Drupal\cdesai_migrate\Plugin\migrate\process;
+
+use Drupal\Component\Plugin\Exception\PluginException;
+use Drupal\migrate\MigrateSkipRowException;
+use Drupal\migrate\Row;
+use Drupal\migrate\ProcessPluginBase;
+use Drupal\migrate\MigrateExecutableInterface;
+use Drush\Drush;
+
+/**
+ * Provides a 'Generate Redirect Uri' migrate process plugin.
+ *
+ * Generates a redirect URI from the legacy url.
+ *
+ * @MigrateProcessPlugin(
+ *  id = "generate_redirect_uri"
+ * )
+ */
+class GenerateRedirectUri extends ProcessPluginBase {
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\migrate\MigrateSkipRowException
+   */
+  public function transform($value, MigrateExecutableInterface $migrate_executable, Row $row, $destination_property) {
+    $redirect = '';
+
+    if (is_null($value) || trim($value) === '') {
+      $error = "Skipping because legacy_url is null or empty.";
+      Drush::output()->writeln($error);
+      throw new MigrateSkipRowException($error);
+    }
+
+    /** @var \Drupal\node\NodeStorageInterface $nodeStorage */
+    try {
+      $nodeStorage = \Drupal::entityTypeManager()->getStorage('node');
+      $nodes = $nodeStorage->loadByProperties([
+        'field_legacy_url' => trim($value),
+      ]);
+
+      if (!empty($nodes)) {
+        $redirect = 'entity:node/' . reset($nodes)->id();
+      }
+    }
+    catch (PluginException $e) {
+      $redirect = '';
+    }
+
+    if (!$redirect) {
+      $error = sprintf('Skipped... Unable to map legacy url "%s" to a node.', $value);
+      Drush::output()->writeln($error);
+      throw new MigrateSkipRowException($error);
+    }
+
+    Drush::output()->writeln($redirect);
+    return $redirect;
+  }
+
+}

src/Plugin/migrate/source/CDesaiCSV.php

@@ -0,0 +1,103 @@
+<?php
+
+namespace Drupal\cdesai_migrate\Plugin\migrate\source;
+
+use Drupal\cdesai_migrate\DemoMigrateConstantsInterface;
+use Drupal\Core\Url;
+use Drupal\migrate\Plugin\MigrationInterface;
+use Drupal\migrate_source_csv\Plugin\migrate\source\CSV;
+use League\Csv\Reader;
+
+/**
+ * Source for CSV files (overridden to make source path dynamic).
+ *
+ * @see \Drupal\migrate_source_csv\Plugin\migrate\source\CSV
+ *
+ * Available configuration options:
+ * @see \Drupal\migrate_source_csv\Plugin\migrate\source\CSV for all options.
+ *
+ *
+ * Example with minimal options:
+ * @code
+ * source:
+ *   plugin: cdesai_migrate_csv
+ *   source_config_key: cdesai_migrate_basic_pages_source
+ *   ids: [id]
+ *   delimiter: '|'
+ *   enclosure: "'"
+ *   escape: '`'
+ *   header_offset: 0
+ *   fields:
+ *     -
+ *       name: id
+ *       label: ID
+ *     -
+ *       name: country
+ *       label: Country
+ *
+ * @MigrateSource(
+ *   id = "cdesai_migrate_csv",
+ *   source_module = "cdesai_migrate"
+ * )
+ */
+class CDesaiCSV extends CSV {
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \InvalidArgumentException
+   * @throws \Drupal\migrate\MigrateException
+   */
+  public function __construct(array $configuration, $plugin_id, $plugin_definition, MigrationInterface $migration) {
+    $configuration['path'] = 'provided by the configs';
+
+    // Source config key is required.
+    if (empty($this->configuration['source_config_key'])) {
+      throw new \InvalidArgumentException('You must declare the "source_config_key" in your source settings.');
+    }
+    // Source config key must be one of the valid values.
+    elseif (!array_key_exists($this->configuration['source_config_key'], DemoMigrateConstantsInterface::SOURCES)) {
+      throw new \InvalidArgumentException('Invalid value for "source_config_key".');
+    }
+
+    parent::__construct($configuration, $plugin_id, $plugin_definition, $migration);
+  }
+
+  /**
+   * Return a string representing the source file path.
+   *
+   * @return string
+   *   The file path.
+   */
+  public function __toString() {
+    return $this->getSourceFileName();
+  }
+
+  /**
+   * Construct a new CSV reader.
+   *
+   * @return \League\Csv\Reader
+   *   The reader.
+   */
+  protected function createReader(): Reader {
+    $path = $this->getSourceFileName();
+    if (!file_exists($path)) {
+      throw new \RuntimeException(sprintf('File "%s" was not found.', $path));
+    }
+    $csv = fopen($path, 'r');
+    if (!$csv) {
+      throw new \RuntimeException(sprintf('File "%s" could not be opened.', $path));
+    }
+    return Reader::createFromStream($csv);
+  }
+
+  /**
+   * Get source path from configs.
+   */
+  private function getSourceFileName(): string {
+    // CODE REDACTED
+
+    return "REDACTED";
+  }
+
+}

src/Plugin/migrate/source/CDesaiParagraphMigrateSource.php

@@ -0,0 +1,169 @@
+<?php
+
+namespace Drupal\cdesai_migrate\Plugin\migrate\source;
+
+use Drupal\cdesai_migrate\DemoMigrateConstantsInterface;
+use Drupal\Core\Url;
+use Drupal\migrate\MigrateException;
+use Drupal\migrate\Plugin\migrate\source\SourcePluginBase;
+use Drupal\migrate\Plugin\MigrationInterface;
+
+/**
+ * Source plugin to import paragraphs from content files.
+ *
+ * @MigrateSource(
+ *   id = "cdesai_paragraph_migrate_source"
+ * )
+ */
+class CDesaiParagraphMigrateSource extends SourcePluginBase {
+
+  /**
+   * The paragraph contents archive file containing one or more text files.
+   *
+   * @var string
+   */
+  protected string $paragraphsSourceFile;
+
+  /**
+   * The CSV source files for pages.
+   *
+   * @var string[]
+   */
+  protected array $pageSourceFiles;
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\migrate\MigrateException
+   */
+  public function __construct(array $configuration, $plugin_id, $plugin_definition, MigrationInterface $migration) {
+    parent::__construct($configuration, $plugin_id, $plugin_definition, $migration);
+    $configs = \Drupal::config(DemoMigrateConstantsInterface::CONFIG_NAME);
+    $this->paragraphsSourceFile = (string) $configs->get('cdesai_migrate_paragraphs_source_path');
+    $this->pageSourceFiles = [
+      (string) $configs->get('cdesai_migrate_articles_source_path'),
+      (string) $configs->get('cdesai_migrate_basic_pages_source_path'),
+    ];
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function getIds(): array {
+    return [
+      'page' => [
+        'type' => 'string',
+      ],
+    ];
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function fields(): array {
+    return [
+      'page' => $this->t("Page"),
+      'skip_migration' => $this->t('Skip Migration'),
+      'scrapped' => $this->t('Scrapped Content'),
+      'content_file' => $this->t('Paragraph Content File'),
+    ];
+  }
+
+  /**
+   * {@inheritdoc}
+   */
+  public function __toString() {
+    $this->initializeIterator();
+
+    return "Content Archive: \n\t{$this->paragraphsSourceFile} \n"
+      . "Page Metadata Files:\n\t" . implode("\n\t", $this->pageSourceFiles);
+  }
+
+  /**
+   * {@inheritdoc}
+   *
+   * @throws \Drupal\migrate\MigrateException
+   */
+  protected function initializeIterator() {
+    $this->requiredFilesExist();
+    $pages = $this->extractPageData();
+
+    // CODE REDACTED
+
+    // Migrate needs an Iterator class, not just an array.
+    return new \ArrayIterator($pages);
+  }
+
+  /**
+   * Make sure all required source files exists.
+   *
+   * @throws \Drupal\migrate\MigrateException
+   */
+  private function requiredFilesExist() {
+    $fileExists = TRUE;
+
+    // CODE REDACTED
+
+    if (!$fileExists) {
+      $fileUploadUri = Url::fromRoute('cdesai_migrate.settings')->getInternalPath();
+      throw new MigrateException('Upload required source files at ' . $fileUploadUri);
+    }
+  }
+
+  /**
+   * Retrieves paragraph content files from the archive.
+   *
+   * @throws \Drupal\migrate\MigrateException
+   */
+  private function getParagraphContentFiles(string $archive_file): array {
+    $contentFiles = [];
+    $extractedDir = $this->extractArchiveFile($archive_file);
+    /** @var \Drupal\Core\File\FileSystem $fileSystem */
+    $fileSystem = \Drupal::service('file_system');
+    $files = $fileSystem->scanDirectory($extractedDir, '/.*txt/');
+    if (empty($files)) {
+      throw new MigrateException('There was an issue extracting files form ' . $archive_file);
+    }
+
+    foreach ($files as $file) {
+      $key = substr($file->filename, 0, strlen($file->filename) - 4);
+      $contentFiles[$key] = $file->uri;
+    }
+
+    return $contentFiles;
+  }
+
+  /**
+   * Extract the archive in a temp directory.
+   *
+   * @param string $filepath
+   *   The file path.
+   *
+   * @return string
+   *   The temp directory path containing the extracted archive file.
+   */
+  private function extractArchiveFile(string $filepath): string {
+    /** @var \Drupal\Core\Archiver\ArchiverInterface $archiver */
+    $archiveManager = \Drupal::service('plugin.manager.archiver');
+    $tempPath = "";
+
+    // CODE REDACTED
+
+    return 'temporary://' . $tempPath;
+  }
+
+  /**
+   * Extract page data from CSV source files.
+   *
+   * @return array
+   *   An associative array containing page data.
+   */
+  private function extractPageData(): array {
+    $pages = [];
+
+    // CODE REDACTED
+
+    return $pages;
+  }
+
+}