Fix image import.

It appears the url contains a base64 encoded array of arrays with
information about the orinal image. (I originally thought it was a
secret token of some sort.) This makes it easy to fetch the original
unprocessed image from the old server.

1 files changed, 40 insertions, 14 deletions

diff --git a/import-posts.php b/import-posts.php
index 6697289..113d356 100644
--- a/import-posts.php
+++ b/import-posts.php
@@ -38,7 +38,9 @@ if ( true === $archive->open( $filename, ZipArchive::RDONLY ) ) {
 	for ( $i = 0; $i < $files; $i++ ) {
 		echo " => {$archive->getNameIndex( $i )}..." . PHP_EOL;
 		$meta = parse_old_post( $archive->getFromIndex( $i ) );
-		migrate_post( $meta );
+		if ( isset( $meta['body'] ) ) {
+			migrate_post( $meta );
+		}
 	}
 
 	$archive->close();
@@ -49,7 +51,9 @@ function parse_old_post( $input ) : array {
 
 	// $parts[0] is empty
 	$meta = parse_post_frontmatter( $parts[1] );
-	$meta['body'] = trim( $parts[2] );
+	if ( isset( $parts[2] ) ) {
+		$meta['body'] = trim( $parts[2] );
+	}
 
 	return $meta;
 }
@@ -87,14 +91,16 @@ function migrate_post( $meta ) {
 		explode( ',', $meta['categories'] )
 	);
 
-	$images = migrate_images( $meta );
+	$res = migrate_images( $meta );
+	$body = $res[0];
+	$thumb = $res[1];
 
 	$post_id = wp_insert_post(
 		array(
 			'post_author' => $meta['user_id'],
 			'post_date' => $meta['published_at'],
 			'post_date_gmt' => $meta['published_at'],
-			'post_content' => $meta['body'],
+			'post_content' => $res[0],
 			'post_title' => $meta['title'],
 			'post_excerpt' => wp_filter_kses( $meta['custom_teaser'] ),
 			'post_status' => 'publish',
@@ -114,7 +120,9 @@ function migrate_post( $meta ) {
 		die( "Error: Unable to insert post: {$meta['title']}." . PHP_EOL );
 	}
 
-    return array( $post_id, $thumb );
+	if ( $thumb ) {
+		set_post_thumbnail( $post_id, $thumb );
+	}
 }
 
 function migrate_images( array $meta ) : array {
@@ -126,17 +134,17 @@ function migrate_images( array $meta ) : array {
 	$thumb = null;
 
 	if ( isset( $meta['body'] ) ) {
-		$meta['body'] = preg_replace_callback(
+		$body = preg_replace_callback(
 			'/<img src="([^"]*)" title="([^"]*)" alt="([^"]*)"[^>]*>/',
-			function( array $matches ) {
-				$remote_filename = $matches[1];
+			function( array $matches ) use ( $post_date, &$thumb ) {
+				$remote_filename = get_real_image_name( $matches[1] );
 				$filename = basename( $remote_filename );
 				$title = $matches[2];
 				$alt = $matches[3];
 
 				$uploads = wp_upload_dir( $post_date->format( 'Y/m' ) );
 				$target_file = implode( '/', array( $uploads['path'], $filename ) );
-				$target_url = implide( '/', array( $uploads['url'], $filename ) );
+				$target_url = implode( '/', array( $uploads['url'], $filename ) );
 
 				import_photo( $remote_filename, $target_file );
 
@@ -159,9 +167,9 @@ function migrate_images( array $meta ) : array {
 					die( "Error: Could not create attachment for {$target_file}" . PHP_EOL );
 				}
 
-				if ( ! $thumb ) {
+				if ( null === $thumb ) {
 					$thumb = $attachment_id;
-					return '';
+					return '<!-- image moved to thumbnail -->';
 				} else {
 					return "<img src=\"{$target_url}\" title=\"{$title}\" alt=\"{$alt}\">";
 				}
@@ -170,11 +178,12 @@ function migrate_images( array $meta ) : array {
 		);
 	}
 
-	return $thumb;
+	return array( $body, $thumb );
 }
 
 function import_photo( $remote_path, $target_filename ) {
-	$curl = curl_init( implode( '/', array( 'https://heavymetal.no', $remote_path ) ) );
+	$remote_url = implode( '/', array( 'https://heavymetal.no/system/refinery/images', $remote_path ) );
+	$curl = curl_init( $remote_url );
 
 	if ( false === $curl ) {
 		die( 'Error: Unable to init curl' );
@@ -182,5 +191,22 @@ function import_photo( $remote_path, $target_filename ) {
 
 	curl_setopt( $curl, CURLOPT_RETURNTRANSFER, true );
 	$res = curl_exec( $curl );
-	file_put_contents( $target_filename, $res );
+	if ( $res ) {
+		$status = curl_getinfo( $curl, CURLINFO_HTTP_CODE );
+		if ( $status > 299 ) {
+			die( "Error: Could not fetch '{$remote_url}', status = {$status}." . PHP_EOL );
+		}
+		file_put_contents( $target_filename, $res );
+	}
+}
+
+function get_real_image_name( string $orig ) : string {
+	$encoded = basename( dirname( $orig ) );
+	$decoded = base64_decode( $encoded );
+
+	if ( preg_match( '/\["f","([^"]+)"\]/', $decoded, $matches ) ) {
+		return $matches[1];
+	}
+
+	die( "Filed to decode filename: {$encoded}: {$decoded}" . PHP_EOL );
 }