Compare commits

...

5 Commits

Author SHA1 Message Date
cwez 58b35e5fa6 Adds migration configs. 2022-02-15 18:44:32 +00:00
cwez c6f921dd53 Adds event subscriber for post migration operations. 2022-02-15 18:44:20 +00:00
cwez 45b8ce5631 Adds custom migration source and process plugins. 2022-02-15 18:43:50 +00:00
cwez 88c410b4fc Adds module config form for admins. 2022-02-15 18:43:00 +00:00
cwez cdd10c02f5 Adds basic module files. 2022-02-15 18:39:51 +00:00
22 changed files with 1511 additions and 0 deletions
+14
View File
@@ -0,0 +1,14 @@
type: module
name: Demo Migration Module
description: 'Migrate content from a CSV file. DO NOT INSTALL'
package: Migration
core: 8.x
core_version_requirement: ^8 || ^9
dependencies:
- migrate
- migrate_plus
- migrate_tools
- migrate_source_csv
# Hides the module from the Admin UI.
hidden: true
+23
View File
@@ -0,0 +1,23 @@
<?php
/**
* @file
* Install and update functions for the Demo Migrate module.
*/
/**
* Implements hook_requirements().
*/
function cdesai_migrate_requirements($phase) {
$requirements = [];
if ($phase == 'install') {
// Abort installation of this module since it is not ready.
$requirements['cdesai_migrate_not_ready'] = [
'description' => t('Demo Migration module is not ready for the installation. This module heavily redacted version of a module used in a project. Do not install.'),
'severity' => REQUIREMENT_ERROR,
];
}
return $requirements;
}
+6
View File
@@ -0,0 +1,6 @@
# Settings form
cdesai_migrate.settings:
title: 'Demo Migrations'
description: 'Configure Demo migration settings.'
route_name: cdesai_migrate.settings
parent: system.admin_config_development
+3
View File
@@ -0,0 +1,3 @@
administer demo migrate:
title: 'Administer Demo Migrate'
restrict access: true
+7
View File
@@ -0,0 +1,7 @@
cdesai_migrate.settings:
path: '/admin/config/development/cdesai/migrate'
defaults:
_form: '\Drupal\cdesai_migrate\Form\CDesaiMigrateSettings'
_title: 'Demo Migrations'
requirements:
_permission: 'administer demo migrate'
+5
View File
@@ -0,0 +1,5 @@
services:
post_migration_subscriber:
class: '\Drupal\cdesai_migrate\EventSubscriber\DemoPostMigrationSubscriber'
tags:
- { name: 'event_subscriber' }
+22
View File
@@ -0,0 +1,22 @@
{
"name": "drupal/cdesai_migrate",
"description": "Demo of custom text formatters based on Unleashed Technologies challenge.",
"type": "drupal-module",
"homepage": "https://wezbyte.com",
"keywords": ["Drupal", "Chintan", "Desai", "Migration"],
"minimum-stability": "dev",
"license": "GPL-2.0-or-later",
"authors": [
{
"name": "Chintan Desai",
"email": "chintan@wezbyte.com"
}
],
"require": {
"drupal/core": "^8.9 || ^9.3",
"drupal/migrate_file": "^2.0",
"drupal/migrate_plus": "^5.1",
"drupal/migrate_source_csv": "^3.4",
"drupal/migrate_tools": "^5.0"
}
}
@@ -0,0 +1,95 @@
# Migration configuration for importing basic pages.
id: cdesai_basic_pages
label: Import Basic Pages from CSV
migration_group: CDESAI
source:
# The custom CSV source plugin will fetch the source path from configs.
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: 'cdesai_migrate_csv'
source_config_key: cdesai_migrate_basic_pages_source
delimiter: ','
enclosure: '"'
header_offset: 0
ids:
- page
process:
langcode: en
moderation_state: draft
title: title
uid:
plugin: default_value
default_value: 1
field_content/target_id:
# Skips a row if skip_migration property is set to TRUE.
-
plugin: skip_on_value
equals: true
source: skip_migration
value: 'TRUE'
method: row
-
plugin: migration_lookup
migration: cdesai_paragraph_parents
no_stub: true
source: page
-
plugin: extract
index:
- '0'
field_content/target_revision_id:
-
plugin: migration_lookup
migration: cdesai_paragraph_parents
no_stub: true
source: page
-
plugin: extract
index:
- 1
field_category:
# Do not process further if the category value is empty.
-
plugin: skip_on_value
equals: true
source: category
value: ''
method: process
# The plugin looks up for an exiting term based on the value_key property.
# If no matching term is found, it creates one on the fly.
-
plugin: entity_generate
source: category
entity_type: taxonomy_term
bundle_key: vid
bundle: category
value_key: name
ignore_case: true
field_legacy_url: legacy_url
type:
plugin: default_value
default_value: basic_page
destination:
plugin: 'entity:node'
# make sure these migration execute before running current migration
migration_dependencies:
required:
- cdesai_paragraphs
- cdesai_paragraph_parents
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,60 @@
# Migration configuration for attaching paragraphs to parent paragraphs (row).
id: cdesai_paragraph_parents
label: 'Parent (Row) Paragraphs'
migration_group: CDESAI
source:
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: cdesai_paragraph_migrate_source
process:
langcode: en
field_bundles/target_id:
# Skips a row if skip_migration property is set to TRUE.
-
plugin: skip_on_value
equals: true
source: skip_migration
value: 'TRUE'
method: row
# Skips a row if scrapped property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: scrapped
value: 'FALSE'
method: row
-
plugin: migration_lookup
migration: cdesai_paragraphs
no_stub: true
source: page
-
plugin: extract
index:
- '0'
field_bundles/target_revision_id:
-
plugin: migration_lookup
migration: cdesai_paragraphs
no_stub: true
source: page
-
plugin: extract
index:
- '1'
destination:
plugin: 'entity_reference_revisions:paragraph'
default_bundle: row
# make sure these migration execute before running current migration
migration_dependencies:
required:
- cdesai_paragraphs
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,45 @@
# Migration configuration for importing content paragraphs.
id: cdesai_paragraphs
label: Import Content Paragraphs from Text File Archive
migration_group: CDESAI
source:
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: cdesai_paragraph_migrate_source
process:
langcode: en
field_body/value:
# Skips a row if skip_migration property is set to TRUE.
-
plugin: skip_on_value
equals: true
source: skip_migration
value: 'TRUE'
method: row
# Skips a row if scrapped property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: scrapped
value: 'FALSE'
method: row
# Uses the custom process plugin to extract paragraph contents from a text file.
-
plugin: extract_file_contents
source: content_file
field_body/format:
plugin: default_value
default_value: rich_text
destination:
plugin: 'entity_reference_revisions:paragraph'
default_bundle: basic
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,145 @@
# Migration configuration for importing profile pages.
id: cdesai_profile_pages
label: Import Profile Pages from CSV
migration_group: CDESAI
source:
# The custom CSV source plugin will fetch the source path from configs.
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: 'cdesai_migrate_csv'
source_config_key: cdesai_migrate_profile_pages_source
delimiter: ','
enclosure: '"'
header_offset: 0
ids:
- page
process:
langcode: en
moderation_state: draft
title: title
field_legacy_url: legacy_url
field_year_birth: year_birth
field_year_death: year_death
uid:
plugin: default_value
default_value: 1
body/value:
# Skips a row if skip_migration property is set to TRUE.
-
plugin: skip_on_value
equals: true
source: skip_migration
value: 'TRUE'
method: row
# Skips a row if scrapped property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: scrapped
value: 'FALSE'
method: row
-
plugin: migration_lookup
migration: cdesai_paragraphs
no_stub: true
source: page
-
plugin: extract_cdesai_migrate_paragraph
field_body/format:
plugin: default_value
default_value: rich_text
field_position:
# Do not process further if the position value is empty.
-
plugin: skip_on_value
equals: true
source: position
value: ''
method: process
# The plugin looks up for an exiting term based on the value_key property.
# If no matching term is found, it creates one on the fly.
-
plugin: entity_generate
source: position
entity_type: taxonomy_term
bundle_key: vid
bundle: position
value_key: name
ignore_case: true
field_agency:
# Do not process further if the position value is empty.
-
plugin: skip_on_value
equals: true
source: agency
value: ''
method: process
# The plugin looks up for an exiting term based on the value_key property.
# If no matching term is found, it creates one on the fly.
-
plugin: entity_generate
source: agency
entity_type: taxonomy_term
bundle_key: vid
bundle: agency
value_key: name
ignore_case: true
field_years_in_office:
-
plugin: skip_on_empty
source: years_in_office
method: process
message: 'Row does not contain years_in_office.'
-
plugin: explode
delimiter: ','
-
plugin: callback
callable: trim
field_media/target_id:
# Skips a row if has_images property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: has_images
value: 'FALSE'
method: row
-
plugin: skip_on_empty
method: row
message: 'Cannot import image file with an empty url.'
-
plugin: migration_lookup
migration: cdesai_profile_pages_media_items
source: page
type:
plugin: default_value
default_value: profile
destination:
plugin: 'entity:node'
# make sure these migration execute before running current migration
migration_dependencies:
required:
- cdesai_profile_pages_media_items
- cdesai_basic_pages
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,73 @@
# Migration configuration for importing image files.
id: cdesai_profile_pages_image_files
label: Import Profile Page Files
migration_group: CDESAI_MEDIA
source:
# The custom CSV source plugin will fetch the source path from configs.
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: 'cdesai_migrate_csv'
source_config_key: cdesai_migrate_profile_pages_source
delimiter: ','
enclosure: '"'
header_offset: 0
constants:
DRUPAL_FILE_DIRECTORY: 'public://images/profile/'
ids:
- page
process:
uid:
plugin: default_value
default_value: 1
pseudo_destination_filename:
# Skips a row if has_images property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: has_images
value: 'FALSE'
method: row
-
plugin: skip_on_empty
method: row
message: 'Cannot import image file with an empty url.'
-
plugin: callback
callable: basename
source: image_url
pseudo_destination_path:
-
plugin: concat
source:
- constants/DRUPAL_FILE_DIRECTORY
- '@pseudo_destination_filename'
uri:
-
plugin: skip_on_empty
method: process
source: image_url
-
plugin: file_copy
source:
- image_url
- '@pseudo_destination_path'
file_exists: rename
move: false
destination:
plugin: 'entity:file'
# make sure these migration execute before running current migration
migration_dependencies: null
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,72 @@
# Migration configuration for creating media items.
id: cdesai_profile_pages_media_items
label: Import Profile Page Media
migration_group: CDESAI_MEDIA
source:
# The custom CSV source plugin will fetch the source path from configs.
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: 'cdesai_migrate_csv'
source_config_key: cdesai_migrate_profile_pages_source
delimiter: ','
enclosure: '"'
header_offset: 0
ids:
- page
process:
pseudo_destination_filename:
# Skips a row if has_images property is set to FALSE.
-
plugin: skip_on_value
equals: true
source: has_images
value: 'FALSE'
method: row
-
plugin: skip_on_empty
method: row
message: 'Cannot import image file with an empty url.'
-
plugin: callback
callable: basename
source: image_url
name: '@pseudo_destination_filename'
field_media_image/target_id:
plugin: migration_lookup
migration: cdesai_profile_pages_image_files
source: page
field_media_image/alt: title
thumbnail/target_id:
plugin: migration_lookup
migration: cdesai_profile_pages_image_files
source: page
thumbnail/alt: title
uid:
plugin: default_value
default_value: 1
langcode: en
destination:
plugin: 'entity:media'
default_bundle: image
# make sure these migration execute before running current migration
migration_dependencies:
required:
- cdesai_profile_pages_image_files
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
@@ -0,0 +1,65 @@
# Migration configuration for generating redirects from legacy url to Drupal nodes.
id: cdesai_migrate_page_redirects
label: 'Legacy Page Redirects'
migration_group: CDESAI
source:
# Upload source files at /admin/config/development/cdesai/migrate.
plugin: cdesai_paragraph_migrate_source
process:
language:
plugin: default_value
source: language
default_value: und
uid:
plugin: default_value
default_value: 1
status_code:
plugin: default_value
default_value: 301
redirect_source/path:
# Skips a row if skip_migration property is set to TRUE.
-
plugin: skip_on_value
equals: true
source: skip_migration
value: 'TRUE'
method: row
# Skips a row if legacy_url property is EMPTY.
-
plugin: skip_on_value
equals: true
source: legacy_url
value: ''
method: row
-
plugin: substr
source: legacy_url
start: 1
redirect_redirect/title: ''
redirect_redirect/uri:
# Uses the custom process plugin to extract paragraph contents from a text file.
-
plugin: generate_redirect_uri
source: legacy_url
destination:
plugin: entity:redirect
# make sure these migration execute before running current migration
migration_dependencies:
required:
- cdesai_articles
- cdesai_basic_pages
# use forced module dependency so uninstall/reinstall works properly
dependencies:
enforced:
module:
- cdesai_migrate
+60
View File
@@ -0,0 +1,60 @@
<?php
namespace Drupal\cdesai_migrate;
/**
* Configuration form.
*/
interface DemoMigrateConstantsInterface {
const CONFIG_NAME = 'cdesai_migrate.settings';
const FILE_TYPES = [
'spreadsheet' => ['csv'],
'archive' => ['zip'],
'text' => ['txt'],
];
const SOURCES = [
'cdesai_migrate_basic_pages_source' => [
'label' => 'Basic Pages Source',
'directory' => 'private://migration/sources',
'description' => 'File must be a CSV file.',
'source_file_type' => 'spreadsheet',
'required_columns' => [
'page',
'skip_migration',
'scrapped',
'category',
'title',
'legacy_url',
],
],
'cdesai_migrate_profile_pages_source' => [
'label' => 'Profile Pages Source',
'directory' => 'private://migration/sources',
'description' => 'File must be a CSV file.',
'source_file_type' => 'spreadsheet',
'required_columns' => [
'page',
'skip_migration',
'scrapped',
'position',
'title',
'legacy_url',
'year_birth',
'year_death',
'years_in_office',
'agency',
'image_url',
],
],
'cdesai_migrate_paragraphs_source' => [
'label' => 'Paragraphs Source',
'directory' => 'private://migration/sources',
'description' => 'File must be a ZIP archive containing one or more .txt files.',
'source_file_type' => 'archive',
'archive_content_file_types' => 'text',
],
];
}
@@ -0,0 +1,40 @@
<?php
namespace Drupal\cdesai_migrate\EventSubscriber;
use Drupal\migrate\Event\MigrateEvents;
use Drupal\migrate\Event\MigratePostRowSaveEvent;
use Symfony\Component\EventDispatcher\EventSubscriberInterface;
/**
* Class PostMigrationSubscriber.
*
* Run our user flagging after the last node migration is run.
*
* @package Drupal\YOUR_MODULE
*/
class DemoPostMigrationSubscriber implements EventSubscriberInterface {
/**
* Get subscribed events.
*
* @inheritdoc
*/
public static function getSubscribedEvents(): array {
$events[MigrateEvents::POST_ROW_SAVE][] = ['onMigratePostImport'];
return $events;
}
/**
* Callback for MigrateEvents::POST_ROW_SAVE event.
*
* @param \Drupal\migrate\Event\MigratePostRowSaveEvent $event
* The import event object.
*/
public function onMigratePostImport(MigratePostRowSaveEvent $event) {
if ($event->getMigration()->getBaseId() == 'cdesai_basic_pages') {
// CODE REDACTED
}
}
}
+333
View File
@@ -0,0 +1,333 @@
<?php
namespace Drupal\cdesai_migrate\Form;
use Drupal\cdesai_migrate\DemoMigrateConstantsInterface;
use Drupal\Core\Archiver\ArchiverManager;
use Drupal\Core\Config\ConfigFactoryInterface;
use Drupal\Core\Extension\ModuleHandlerInterface;
use Drupal\Core\File\Exception\FileException;
use Drupal\Core\File\FileSystemInterface;
use Drupal\Core\Form\ConfigFormBase;
use Drupal\Core\Form\FormStateInterface;
use Drupal\Core\StreamWrapper\StreamWrapperManager;
use Drupal\file\Entity\File;
use Drupal\file\FileInterface;
use Symfony\Component\DependencyInjection\ContainerInterface;
/**
* Configuration form.
*/
class CDesaiMigrateSettings extends ConfigFormBase {
const FORM_ID = 'cdesai_migrate_settings';
/**
* The module handler.
*
* @var \Drupal\Core\Extension\ModuleHandlerInterface
*/
protected ModuleHandlerInterface $moduleHandler;
/**
* The file system.
*
* @var \Drupal\Core\File\FileSystemInterface
*/
protected FileSystemInterface $fileSystem;
/**
* The archive manager.
*
* @var \Drupal\Core\Archiver\ArchiverManager
*/
protected ArchiverManager $archiveManager;
/**
* Constructs a ThemeSettingsForm object.
*
* @param \Drupal\Core\Config\ConfigFactoryInterface $config_factory
* The factory for configuration objects.
* @param \Drupal\Core\Extension\ModuleHandlerInterface $module_handler
* The module handler instance to use.
* @param \Drupal\Core\File\FileSystemInterface $file_system
* The file system.
* @param \Drupal\Core\Archiver\ArchiverManager $archive_manager
* The archive manager.
*/
public function __construct(ConfigFactoryInterface $config_factory, ModuleHandlerInterface $module_handler, FileSystemInterface $file_system, ArchiverManager $archive_manager) {
parent::__construct($config_factory);
$this->moduleHandler = $module_handler;
$this->fileSystem = $file_system;
$this->archiveManager = $archive_manager;
}
/**
* {@inheritdoc}
*/
public static function create(ContainerInterface $container) {
return new static(
$container->get('config.factory'),
$container->get('module_handler'),
$container->get('file_system'),
$container->get('plugin.manager.archiver')
);
}
/**
* {@inheritdoc}
*/
protected function getEditableConfigNames(): array {
return [
DemoMigrateConstantsInterface::CONFIG_NAME,
];
}
/**
* {@inheritdoc}
*/
public function getFormId(): string {
return self::FORM_ID;
}
/**
* {@inheritdoc}
*/
public function buildForm(array $form, FormStateInterface $form_state): array {
$configs = $this->config(DemoMigrateConstantsInterface::CONFIG_NAME);
foreach (DemoMigrateConstantsInterface::SOURCES as $sourceKey => $source) {
$currentPath = $configs->get($sourceKey . '_path');
$text = $this->t('Upload a file');
if ($currentPath && $this->validatePath($currentPath)) {
$text = $currentPath;
}
$form[$sourceKey . '_current_path'] = [
'#type' => 'markup',
'#markup' =>
"<strong>Current {$source['label']} file:</strong> " . $text,
];
$form[$sourceKey . '_upload'] = [
'#type' => 'file',
'#title' => $this->t('Upload %title file', ['%title' => $source['label']]),
'#description' => $source['description'],
'#upload_validators' => [
'file_validate_extensions' => DemoMigrateConstantsInterface::FILE_TYPES[$source['source_file_type']],
],
'#suffix' => '<hr/>',
];
}
return parent::buildForm($form, $form_state);
}
/**
* {@inheritdoc}
*/
public function validateForm(array &$form, FormStateInterface $form_state) {
parent::validateForm($form, $form_state);
if ($this->moduleHandler->moduleExists('file')) {
foreach (DemoMigrateConstantsInterface::SOURCES as $sourceKey => $source) {
// Validate file only if a new file was uploaded.
$upload_key = $sourceKey . '_upload';
$file = _file_save_upload_from_form($form[$upload_key], $form_state, 0, FileSystemInterface::EXISTS_REPLACE);
if ($file) {
if ($source['source_file_type'] === 'archive') {
$contentFileExts = DemoMigrateConstantsInterface::FILE_TYPES[$source['archive_content_file_types']];
$this->validateArchiveFile($upload_key, $contentFileExts, $file, $form_state);
}
else {
$this->validateCsvFile($upload_key, $source['required_columns'], $file, $form_state);
}
}
}
}
}
/**
* {@inheritdoc}
*/
public function submitForm(array &$form, FormStateInterface $form_state) {
parent::submitForm($form, $form_state);
$configs = $this->config(DemoMigrateConstantsInterface::CONFIG_NAME);
// Exclude unnecessary elements before saving.
$form_state->cleanValues();
$values = $form_state->getValues();
foreach (DemoMigrateConstantsInterface::SOURCES as $sourceKey => $source) {
// Check for a newly uploaded file.
$upload_key = $sourceKey . '_upload';
$path_key = $sourceKey . '_path';
try {
if (isset($values[$upload_key]) && $values[$upload_key] instanceof File) {
$this->fileSystem->mkdir($source['directory'], NULL, TRUE);
$filename = $this->fileSystem->copy($values[$upload_key]->getFileUri(), $source['directory'], FileSystemInterface::EXISTS_RENAME);
$configs->set($path_key, $filename);
}
}
catch (FileException $e) {
// Ignore.
$this->logger('Demo Migration')->error($e->getMessage());
$this->messenger()->addError($e->getMessage());
}
}
$configs->save();
}
/**
* Helper function for the cdesai_migrate_settings form.
*
* Attempts to validate normal system paths, paths relative to the public
* files directory, or stream wrapper URIs. If the given path is any of the
* above, returns a valid path or URI that the theme system can display.
*
* @param string $path
* A path relative to the Drupal root or to the public files directory, or
* a stream wrapper URI.
*
* @return false|string
* A valid path that can be displayed through the theme system, or FALSE if
* the path could not be validated.
*/
protected function validatePath(string $path) {
// Absolute local file paths are invalid.
if ($this->fileSystem->realpath($path) == $path) {
return FALSE;
}
// A path relative to the Drupal root or a fully qualified URI is valid.
if (is_file($path)) {
return $path;
}
// Prepend 'private://' for relative file paths within private filesystem.
if (StreamWrapperManager::getScheme($path) === FALSE) {
$path = 'private://' . $path;
if (is_file($path)) {
return $path;
}
}
return FALSE;
}
/**
* Validate the archive file.
*
* @param string $key
* Form field key.
* @param array $allowed_file_extensions
* File extensions allowed in the archive.
* @param \Drupal\file\FileInterface $file
* The file being uploaded.
* @param \Drupal\Core\Form\FormStateInterface $form_state
* The current state of the form.
*/
private function validateArchiveFile(string $key, array $allowed_file_extensions, FileInterface $file, FormStateInterface $form_state) {
$filepath = $this->fileSystem->realpath($file->getFileUri());
/** @var \Drupal\Core\Archiver\ArchiverInterface $archiver */
$archiver = $this->archiveManager->getInstance([
'filepath' => $filepath,
]);
if ($archiver && !empty($archiver->listContents())) {
$otherExt = FALSE;
foreach ($archiver->listContents() as $fileInArchive) {
$ext = pathinfo($fileInArchive, PATHINFO_EXTENSION);
if (!in_array($ext, $allowed_file_extensions)) {
$otherExt = TRUE;
break;
}
}
if ($otherExt) {
$form_state->setErrorByName($key, $this->t('Archive contains one or more files with an invalid extension. Allowed extensions are: %ext', [
'%ext' => implode(', ', $allowed_file_extensions),
]));
}
else {
// Put the temporary file in form_values so we can save it
// on submit.
$form_state->setValue($key, $file);
}
}
else {
$form_state->setErrorByName($key, $this->t('Not a valid archive file.'));
}
}
/**
* Validate the CSV file.
*
* @param string $key
* Form field key.
* @param array $required_columns
* An array column keys that should be in the CSV.
* @param \Drupal\file\FileInterface $file
* The file being uploaded.
* @param \Drupal\Core\Form\FormStateInterface $form_state
* The current state of the form.
*/
private function validateCsvFile(string $key, array $required_columns, FileInterface $file, FormStateInterface $form_state) {
$readFile = FALSE;
if (($handle = fopen($file->getFileUri(), 'r')) !== FALSE) {
$row = fgetcsv($handle, 1000);
if ($row !== FALSE) {
$readFile = TRUE;
// Sanitize keys in the header.
$header = array_map([__CLASS__, 'sanitizeKey'], $row);
// Check for required keys.
if ($required_columns) {
$missingRequiredKeys = array_diff($required_columns, $header);
if (!empty($missingRequiredKeys)) {
$form_state->setErrorByName($key, $this->t('CSV file is missing required columns: %columns', [
'%columns' => implode(', ', $missingRequiredKeys),
]));
}
else {
// Put the temporary file in form_values so we can save it
// on submit.
$form_state->setValue($key, $file);
}
}
}
fclose($handle);
}
if (!$readFile) {
$form_state->setErrorByName($key, $this->t('Unable to read from the file.'));
}
}
/**
* Remove whitespaces form a string and converts it to lowercase.
*
* @param string $key
* A string to sanitize.
*
* @return string
* A sanitized string.
*/
public static function sanitizeKey(string $key): string {
$key = trim($key);
$key = str_replace(' ', '_', $key);
return strtolower($key);
}
}
@@ -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($error);
}
return '';
}
}
@@ -0,0 +1,55 @@
<?php
namespace Drupal\cdesai_migrate\Plugin\migrate\process;
use Drupal\migrate\ProcessPluginBase;
use Drupal\migrate\MigrateException;
use Drupal\migrate\MigrateExecutableInterface;
use Drupal\migrate\Row;
use Drupal\paragraphs\Entity\Paragraph;
/**
* Provides a 'Extract Paragraph' migrate process plugin.
*
* Extracts content of field_body from a paragraph entity.
*
* @MigrateProcessPlugin(
* id = "extract_cdesai_migrate_paragraph",
* handle_multiples = TRUE
* )
*/
class ExtractFromParagraph extends ProcessPluginBase {
/**
* {@inheritdoc}
*
* @throws \Drupal\migrate\MigrateException
* @throws \Drupal\Core\TypedData\Exception\MissingDataException
*/
public function transform($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 $item->getValue()['value'];
}
}
else {
throw new MigrateException("Paragraph does not have field_body.");
}
throw new MigrateException("Unable to find paragraph.");
}
}
@@ -0,0 +1,62 @@
<?php
namespace Drupal\cdesai_migrate\Plugin\migrate\process;
use Drupal\Component\Plugin\Exception\PluginException;
use Drupal\migrate\MigrateSkipRowException;
use Drupal\migrate\Row;
use Drupal\migrate\ProcessPluginBase;
use Drupal\migrate\MigrateExecutableInterface;
use Drush\Drush;
/**
* Provides a 'Generate Redirect Uri' migrate process plugin.
*
* Generates a redirect URI from the legacy url.
*
* @MigrateProcessPlugin(
* id = "generate_redirect_uri"
* )
*/
class GenerateRedirectUri extends ProcessPluginBase {
/**
* {@inheritdoc}
*
* @throws \Drupal\migrate\MigrateSkipRowException
*/
public function transform($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;
}
}
+103
View File
@@ -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";
}
}
@@ -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;
}
}