path: root/import-posts.php

<?php
/**
 * Script to import posts from RefineryCMS to WordPress for Heavymetal.no
 *
 * We expect the posts to be in a zip-archive containing files as exported
 * by the export-articles script in the old install. That is html files
 * with a prepended frontmatter block containing the relevant metadata.
 *
 * We don't care about any subdirectories in the archive, or indeed file
 * names, as we only iterate through the archive linearly by index.
 *
 * @package hmno.migrate
 */

if ( $argc < 2 ) {
	echo "Usage: {$argv[0]} <filename.zip>" . PHP_EOL;
	exit( -1 );
}

$filename = $argv[1];


// Load WordPress
require_once getcwd() . '/wp-load.php';

// Required for the taxonomy functions (categories/tags)
require_once getcwd() . '/wp-admin/includes/taxonomy.php';

// Required for kses functionality
require_once getcwd() . '/wp-includes/kses.php';

// Import media handling to get post thumbnails to work
require_once getcwd() . '/wp-includes/media.php';

set_post_thumbnail_size( 500, 300, true );

$archive = new ZipArchive();
if ( true === $archive->open( $filename, ZipArchive::RDONLY ) ) {
	$files = $archive->count();
	echo "Opened archive with {$files} files." . PHP_EOL;

	for ( $i = 0; $i < $files; $i++ ) {
		echo " => {$archive->getNameIndex( $i )}..." . PHP_EOL;
		$meta = parse_old_post( $archive->getFromIndex( $i ) );
		if ( isset( $meta['body'] ) ) {
			migrate_post( $meta );
		}
	}

	$archive->close();
}

function parse_old_post( $input ) : array {
	$parts = explode( '---', $input );

	// $parts[0] is empty
	$meta = parse_post_frontmatter( $parts[1] );
	if ( isset( $parts[2] ) ) {
		$meta['body'] = trim( $parts[2] );
	}

	return $meta;
}

function parse_post_frontmatter( $input ) : array {
	$lines = array_filter( explode( "\n", $input ) );
	return array_reduce(
		$lines,
		function( $acc, $line ) {
			$kv = explode( ': ', $line, 2 );
			if ( isset( $kv[0], $kv[1] ) ) {
				$acc[ $kv[0] ] = $kv[1];
			}
			return $acc;
		},
		array()
	);
}

function get_wp_category( $name, $parent = 0 ) : int {
	$cat = category_exists( $name );

	if ( ! $cat ) {
		$cat = wp_create_category( $name, $parent );
	}

	return $cat;
}

function migrate_post( $meta ) {
    global $wpdb;

	$categories = array_map(
		fn( $cat ) => get_wp_category( $cat ),
		explode( ',', $meta['categories'] )
	);

	$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' => $res[0],
			'post_title' => $meta['title'],
			'post_excerpt' => wp_filter_kses( $meta['custom_teaser'] ),
			'post_status' => 'publish',
			'comment_status' => 'closed',
			'ping_status' => 'closed',
			'post_modified' => $meta['updated_at'],
			'post_modified_gmt' => $meta['updated_at'],
			'post_parent' => '0',
			'post_category' => $categories,
			'post_tags' => explode( ',', $meta['tags'] ),
			'menu_order' => '0',
			'post_type' => 'post',
		)
    );

	if ( 0 === $post_id ) {
		die( "Error: Unable to insert post: {$meta['title']}." . PHP_EOL );
	}

	if ( $thumb ) {
		set_post_thumbnail( $post_id, $thumb );
	}
}

function migrate_images( array $meta ) : array {
	$post_date = new DateTimeImmutable( $meta['published_at'] );
	if ( ! $post_date ) {
		die( "Error: Could not parse published_at date: {$meta['published_at']}." );
	}

	$thumb = null;

	foreach ( array( 'custom_teaser', 'body' ) as $section ) {
		if ( isset( $meta[$section] ) ) {
			$body = preg_replace_callback(
				'/<img src="([^"]*)" title="([^"]*)" alt="([^"]*)"[^>]*( class="[^"]*")[^>]*>/',
				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];
					$class = $matches[4] ?? '';

					$uploads = wp_upload_dir( $post_date->format( 'Y/m' ) );
					$target_file = implode( '/', array( $uploads['path'], $filename ) );
					$target_url = implode( '/', array( $uploads['url'], $filename ) );

					import_photo( $remote_filename, $target_file );

					$mime = wp_get_image_mime( $target_file );
					if ( ! $mime ) {
						die( "Error: Unknown mime type for {$target_file}" . PHP_EOL );
					}

					$attachment_id = wp_insert_attachment(
						array(
							'post_mime_type' => $mime,
							'guid' => $target_url,
							'post_paren' => 0,
							'post_title' => $filename,
						),
						$target_file
					);

					if ( 0 === $attachment_id ) {
						die( "Error: Could not create attachment for {$target_file}" . PHP_EOL );
					}

					if ( null === $thumb ) {
						$thumb = $attachment_id;
						return '<!-- image moved to thumbnail -->';
					} else {
						return "<img src=\"{$target_url}\" title=\"{$title}\" alt=\"{$alt}\"{$class}>";
					}
				},
				$meta[$section],
			);
		}
	}

	return array( $body, $thumb );
}

function import_photo( $remote_path, $target_filename ) {
	$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' );
	}

	curl_setopt( $curl, CURLOPT_RETURNTRANSFER, true );
	$res = curl_exec( $curl );
	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 );
}