| 1 |
<?php |
| 2 |
/** |
| 3 |
* WP-Admin specific functionality for the plugin |
| 4 |
* |
| 5 |
* @package Substack_Importer |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace SubstackImporter; |
| 9 |
|
| 10 |
use WXR_Generator\File_Writer; |
| 11 |
use WXR_Generator\Generator; |
| 12 |
use WXR_Parser; |
| 13 |
use WP_Import; |
| 14 |
use WP_Error; |
| 15 |
|
| 16 |
|
| 17 |
/** |
| 18 |
* The admin specific functionality for the Substack Importer Plugin |
| 19 |
* |
| 20 |
* |
| 21 |
*/ |
| 22 |
class Importer_Admin { |
| 23 |
|
| 24 |
const EXPORT_FILE_OPTION = 'substack-export-attachment'; |
| 25 |
|
| 26 |
const SUBSTACK_URL_OPTION = 'substack-newsletter-url'; |
| 27 |
|
| 28 |
const WXR_FILE_OPTION = 'substack-wxr-attachement'; |
| 29 |
|
| 30 |
const SUBSTACK_PROGRESS_OPTION = 'substack-import-progress'; |
| 31 |
|
| 32 |
public function run() { |
| 33 |
|
| 34 |
$action = isset( $_GET['action'] ) ? $_GET['action'] : 'start'; |
| 35 |
|
| 36 |
switch ( $action ) { |
| 37 |
|
| 38 |
case 'start': |
| 39 |
default: |
| 40 |
$this->render_page( |
| 41 |
'start-screen', |
| 42 |
array( |
| 43 |
'progress' => get_option( self::SUBSTACK_PROGRESS_OPTION, false ), |
| 44 |
) |
| 45 |
); |
| 46 |
|
| 47 |
break; |
| 48 |
|
| 49 |
case 'upload': |
| 50 |
$upload_result = $this->upload(); |
| 51 |
|
| 52 |
if ( ! is_wp_error( $upload_result ) ) { |
| 53 |
$url = admin_url( 'admin.php?import=substack&action=progress' ); |
| 54 |
return wp_safe_redirect( $url ); |
| 55 |
} |
| 56 |
|
| 57 |
require_once ABSPATH . 'wp-admin/admin-header.php'; |
| 58 |
$this->render_page( |
| 59 |
'start-screen', |
| 60 |
array( |
| 61 |
'error' => $upload_result->get_error_message(), |
| 62 |
'progress' => get_option( self::SUBSTACK_PROGRESS_OPTION, false ), |
| 63 |
) |
| 64 |
); |
| 65 |
|
| 66 |
break; |
| 67 |
|
| 68 |
case 'progress': |
| 69 |
$this->render_page( 'progress' ); |
| 70 |
break; |
| 71 |
|
| 72 |
case 'pre-import': |
| 73 |
// Convert Substack export to WXR |
| 74 |
$wxr_path = $this->convert_substack_to_wxr(); |
| 75 |
|
| 76 |
// Parse the WXR and render the author mapping step. |
| 77 |
$import_data = $this->parse_wxr( $wxr_path ); |
| 78 |
$this->pre_import_page( $import_data ); |
| 79 |
break; |
| 80 |
|
| 81 |
case 'import': |
| 82 |
// Use WordPress importer to import the WXR |
| 83 |
$this->import(); |
| 84 |
break; |
| 85 |
} |
| 86 |
} |
| 87 |
|
| 88 |
/** |
| 89 |
* Progresses through the posts and downloads additional data (author info, comments) |
| 90 |
* through the Substack API. |
| 91 |
* |
| 92 |
* This method is used as an Ajax Action. |
| 93 |
* |
| 94 |
*/ |
| 95 |
public function progress() { |
| 96 |
|
| 97 |
$url = get_option( self::SUBSTACK_URL_OPTION ); |
| 98 |
$file = get_attached_file( get_option( self::EXPORT_FILE_OPTION ) ); |
| 99 |
$writer = new File_Writer( 'php://output' ); |
| 100 |
$converter = new Converter( new Generator( $writer ), $file, get_option( self::SUBSTACK_URL_OPTION ) ); |
| 101 |
$progress = get_option( self::SUBSTACK_PROGRESS_OPTION ); |
| 102 |
|
| 103 |
$result = $converter->load_meta_data( $progress, 1 ); |
| 104 |
|
| 105 |
// If no url was set, we can consider all posts as processed. |
| 106 |
if ( ! $url ) { |
| 107 |
$result['processed'] = $result['total']; |
| 108 |
} |
| 109 |
|
| 110 |
update_option( self::SUBSTACK_PROGRESS_OPTION, $result['processed'] ); |
| 111 |
|
| 112 |
$result['status'] = $result['processed'] === $result['total'] |
| 113 |
? 'done' : 'processing'; |
| 114 |
|
| 115 |
wp_send_json( $result ); |
| 116 |
|
| 117 |
exit(); |
| 118 |
} |
| 119 |
|
| 120 |
/** |
| 121 |
* Try to upload the Substack Export and ensure it is a valid export that can be used in the converter. |
| 122 |
* |
| 123 |
* @return bool|WP_Error |
| 124 |
*/ |
| 125 |
protected function upload() { |
| 126 |
check_admin_referer( 'import-upload' ); |
| 127 |
$file = wp_import_handle_upload(); |
| 128 |
|
| 129 |
// If the upload handler already failed, don't attempt further checks |
| 130 |
if ( ! empty( $file['error'] ) ) { |
| 131 |
return new WP_Error( 'upload_error', esc_html( $file['error'] ) ); |
| 132 |
} |
| 133 |
|
| 134 |
if ( ! file_exists( $file['file'] ) ) { |
| 135 |
$error = sprintf( __( 'The export file could not be found at <code>%s</code>. It is likely that this was caused by a permissions problem.', 'substack-importer' ), esc_html( $file['file'] ) ); |
| 136 |
return new WP_Error( 'upload_error', $error ); |
| 137 |
} |
| 138 |
|
| 139 |
if ( mime_content_type( $file['file'] ) !== 'application/zip' ) { |
| 140 |
$error = sprintf( __( 'Invalid file type uploaded. Expected a zip file, got a %s file.', 'substack-importer' ), mime_content_type( $file['file'] ) ); |
| 141 |
return new WP_Error( 'upload_error', $error ); |
| 142 |
} |
| 143 |
|
| 144 |
$writer = new File_Writer( 'php://output' ); |
| 145 |
$converter = new Converter( new Generator( $writer ), $file['file'] ); |
| 146 |
|
| 147 |
$posts = $converter->get_posts(); |
| 148 |
|
| 149 |
// Something went wrong getting posts from the zip-file |
| 150 |
if ( is_wp_error( $posts ) ) { |
| 151 |
return $posts; |
| 152 |
} |
| 153 |
|
| 154 |
// The zip-file was valid and contained a posts.csv but it was empty. |
| 155 |
if ( null === $posts->current() ) { |
| 156 |
return new WP_Error( __( 'No posts were found in the uploaded export.', 'substack-importer' ) ); |
| 157 |
} |
| 158 |
|
| 159 |
// Check the substack URL. If it is not empty, the url must be valid for the uploaded export file. |
| 160 |
$url = ! empty( $_POST['substack-url'] ) |
| 161 |
? $this->sanitize_substack_url( $_POST['substack-url'] ) |
| 162 |
: null; |
| 163 |
|
| 164 |
if ( $url && ! $this->validate_substack_url( $url, $converter ) ) { |
| 165 |
return new WP_Error( 'upload_error', __( 'The provided Substack Newsletter URL is invalid', 'substack-importer' ) ); |
| 166 |
} |
| 167 |
|
| 168 |
update_option( self::SUBSTACK_PROGRESS_OPTION, 0 ); |
| 169 |
update_option( self::SUBSTACK_URL_OPTION, $url ); |
| 170 |
update_option( self::EXPORT_FILE_OPTION, $file['id'] ); |
| 171 |
|
| 172 |
return true; |
| 173 |
} |
| 174 |
|
| 175 |
/** |
| 176 |
* Validate that the provided (sanitized) Substack leads to the correct Substack Newsletter. |
| 177 |
* |
| 178 |
* @param $url |
| 179 |
* |
| 180 |
* @return bool |
| 181 |
*/ |
| 182 |
protected function validate_substack_url( $url, Converter $converter ) { |
| 183 |
|
| 184 |
// We need to get one post ID and check the posts comments endpoint. If we get a 200 response, the provided |
| 185 |
// substack url matches the export file. |
| 186 |
$post = $converter->get_posts()->current(); |
| 187 |
|
| 188 |
$id = (int) $post['post_id']; |
| 189 |
$api_endpoint = sprintf( '%s/api/v1/post/%d/comments?limit=1', $url, $id ); |
| 190 |
|
| 191 |
$response = wp_remote_get( $api_endpoint ); |
| 192 |
|
| 193 |
return ! is_wp_error( $response ) && 200 === $response['response']['code']; |
| 194 |
} |
| 195 |
|
| 196 |
/** |
| 197 |
* Clean up the Substack url provided by the user to only include scheme + host. |
| 198 |
* |
| 199 |
* Returns false if the url is invalid and can not be parsed. |
| 200 |
* |
| 201 |
* @param string $url URL of Substack newsletter as provided by the user. |
| 202 |
* |
| 203 |
* @return string|bool |
| 204 |
*/ |
| 205 |
protected function sanitize_substack_url( $url ) { |
| 206 |
|
| 207 |
// If scheme is missing, add it |
| 208 |
if ( ! preg_match( '|^.*//|', $url ) ) { |
| 209 |
$url = '//' . $url; |
| 210 |
} |
| 211 |
|
| 212 |
$url_parts = wp_parse_url( $url ); |
| 213 |
|
| 214 |
if ( false === $url_parts ) { |
| 215 |
return false; |
| 216 |
} |
| 217 |
|
| 218 |
return 'https://' . $url_parts['host']; |
| 219 |
} |
| 220 |
|
| 221 |
/** |
| 222 |
* Convert the Substack export to a WXR and render a pre-import. |
| 223 |
* |
| 224 |
* @return string The path of the WXR. |
| 225 |
* |
| 226 |
* @throws \Exception |
| 227 |
*/ |
| 228 |
protected function convert_substack_to_wxr() { |
| 229 |
|
| 230 |
$file = get_attached_file( get_option( self::EXPORT_FILE_OPTION ) ); |
| 231 |
|
| 232 |
// Temporarily store the WXR before sideloading it. |
| 233 |
$tmp_wxr = wp_tempnam( 'substack-wxr.xml' ); |
| 234 |
|
| 235 |
$writer = new File_Writer( $tmp_wxr ); |
| 236 |
|
| 237 |
$converter = new Converter( new Generator( $writer ), $file ); |
| 238 |
|
| 239 |
// Convert the export file to a WXR. |
| 240 |
$converter->convert(); |
| 241 |
$writer->close(); |
| 242 |
|
| 243 |
return $this->store_wxr( $tmp_wxr ); |
| 244 |
} |
| 245 |
|
| 246 |
protected function pre_import_page( $import_data ) { |
| 247 |
$wp_importer = new WP_Import(); |
| 248 |
$wp_importer->get_authors_from_import( $import_data ); |
| 249 |
|
| 250 |
// The wordpress-importer renders a form. The following filter overwrites |
| 251 |
// the action url of that form. This ensures the substack-importer will handle the |
| 252 |
// form submission. |
| 253 |
add_filter( |
| 254 |
'admin_url', |
| 255 |
function( $url ) { |
| 256 |
|
| 257 |
if ( false === strpos( $url, 'import=wordpress' ) ) { //phpcs:ignore WordPress.WP.CapitalPDangit.Misspelled |
| 258 |
return $url; |
| 259 |
} |
| 260 |
|
| 261 |
return wp_nonce_url( add_query_arg( array( 'action' => 'import' ) ), 'import-substack' ); |
| 262 |
} |
| 263 |
); |
| 264 |
|
| 265 |
$this->render_page( |
| 266 |
'pre-import-screen', |
| 267 |
array( |
| 268 |
'wp_importer' => $wp_importer, |
| 269 |
) |
| 270 |
); |
| 271 |
} |
| 272 |
|
| 273 |
/** |
| 274 |
* Import the WXR. |
| 275 |
*/ |
| 276 |
protected function import() { |
| 277 |
|
| 278 |
// To allow podcast uploads, we need to allow the mimetype. |
| 279 |
$this->allow_mpga_mime(); |
| 280 |
|
| 281 |
$wp_importer = new WP_Import(); |
| 282 |
|
| 283 |
$wp_importer->fetch_attachments = ( ! empty( $_POST['fetch_attachments'] ) && $wp_importer->allow_fetch_attachments() ); |
| 284 |
$file = get_attached_file( get_option( self::WXR_FILE_OPTION ) ); |
| 285 |
|
| 286 |
set_time_limit( 0 ); |
| 287 |
|
| 288 |
$wp_importer->import( $file ); |
| 289 |
|
| 290 |
delete_option( self::SUBSTACK_PROGRESS_OPTION ); |
| 291 |
delete_option( self::SUBSTACK_URL_OPTION ); |
| 292 |
delete_option( self::EXPORT_FILE_OPTION ); |
| 293 |
} |
| 294 |
|
| 295 |
protected function allow_mpga_mime() { |
| 296 |
add_filter( |
| 297 |
'upload_mimes', |
| 298 |
function( $mimes ) { |
| 299 |
$mimes['mpga'] = 'audio/mpeg'; |
| 300 |
return $mimes; |
| 301 |
} |
| 302 |
); |
| 303 |
} |
| 304 |
|
| 305 |
/** |
| 306 |
* Parse WXR file |
| 307 |
* |
| 308 |
* @param $wxr_path |
| 309 |
* |
| 310 |
* @return array|\WP_Error |
| 311 |
*/ |
| 312 |
protected function parse_wxr( $wxr_path ) { |
| 313 |
$parser = new WXR_Parser(); |
| 314 |
return $parser->parse( $wxr_path ); |
| 315 |
} |
| 316 |
|
| 317 |
/** |
| 318 |
* Sideload the WXR and store the ID as an option. |
| 319 |
* |
| 320 |
* @param string $wxr_path The path of the temporary WXR file that has to be sideloaded. |
| 321 |
* |
| 322 |
* @return string The path of the sideloaded WXR |
| 323 |
*/ |
| 324 |
protected function store_wxr( $wxr_path ) { |
| 325 |
|
| 326 |
$filedata = array( |
| 327 |
'error' => null, |
| 328 |
'tmp_name' => $wxr_path, |
| 329 |
'name' => 'substackw-wxr.xml', |
| 330 |
'type' => 'text/plain', |
| 331 |
); |
| 332 |
|
| 333 |
$overrides = array( |
| 334 |
'test_form' => false, |
| 335 |
'test_type' => false, |
| 336 |
); |
| 337 |
$sideload = wp_handle_sideload( $filedata, $overrides ); |
| 338 |
|
| 339 |
// Construct the object array.s |
| 340 |
$object = array( |
| 341 |
'post_title' => wp_basename( $sideload['file'] ), |
| 342 |
'post_content' => $sideload['url'], |
| 343 |
'post_mime_type' => mime_content_type( $sideload['file'] ), |
| 344 |
'guid' => $sideload['url'], |
| 345 |
'context' => 'import', |
| 346 |
'post_status' => 'private', |
| 347 |
); |
| 348 |
|
| 349 |
// Save the data. |
| 350 |
$id = wp_insert_attachment( $object, $sideload['file'] ); |
| 351 |
|
| 352 |
/* |
| 353 |
* Schedule a cleanup for one day from now in case of failed |
| 354 |
* import or missing wp_import_cleanup() call. |
| 355 |
*/ |
| 356 |
wp_schedule_single_event( time() + DAY_IN_SECONDS, 'importer_scheduled_cleanup', array( $id ) ); |
| 357 |
|
| 358 |
update_option( self::WXR_FILE_OPTION, $id ); |
| 359 |
|
| 360 |
return $sideload['file']; |
| 361 |
} |
| 362 |
|
| 363 |
/** |
| 364 |
* Render a partial template. |
| 365 |
* |
| 366 |
* @param string $partial The name of the partial. |
| 367 |
* @param array $vars Variables to load into the partial |
| 368 |
*/ |
| 369 |
protected function render_page( $partial, $vars = array() ) { |
| 370 |
|
| 371 |
extract( $vars, EXTR_SKIP ); //phpcs:ignore WordPress.PHP.DontExtract.extract_extract --internal usage only |
| 372 |
|
| 373 |
ob_start(); |
| 374 |
include __DIR__ . '/../partials/' . $partial . '.php'; |
| 375 |
$content = ob_get_clean(); |
| 376 |
|
| 377 |
include __DIR__ . '/../partials/container.php'; |
| 378 |
} |
| 379 |
|
| 380 |
} |
| 381 |
|
