From 614030f9947a54868797d6a77acbeb6d6a096b29 Mon Sep 17 00:00:00 2001
From: Harald Eilertsen <haraldei@anduin.net>
Date: Sun, 31 Jul 2022 15:59:59 +0200
Subject: Initial commit

Importing the posts works, but not images.
---
 import-posts.php | 186 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 186 insertions(+)
 create mode 100644 import-posts.php

(limited to 'import-posts.php')

diff --git a/import-posts.php b/import-posts.php
new file mode 100644
index 0000000..6697289
--- /dev/null
+++ b/import-posts.php
@@ -0,0 +1,186 @@
+<?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 = $argv[1];
+
+
+// Load WordPress
+require_once getcwd() . '/wp-load.php';
+
+// Required for the taxonomy functions (categories/tags)
+require_once getcwd() . '/wp-admin/includes/taxonomy.php';
+
+// Required for kses functionality
+require_once getcwd() . '/wp-includes/kses.php';
+
+
+$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 ) );
+		migrate_post( $meta );
+	}
+
+	$archive->close();
+}
+
+function parse_old_post( $input ) : array {
+	$parts = explode( '---', $input );
+
+	// $parts[0] is empty
+	$meta = parse_post_frontmatter( $parts[1] );
+	$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'] )
+	);
+
+	$images = migrate_images( $meta );
+
+	$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_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 );
+	}
+
+    return array( $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;
+
+	if ( isset( $meta['body'] ) ) {
+		$meta['body'] = preg_replace_callback(
+			'/<img src="([^"]*)" title="([^"]*)" alt="([^"]*)"[^>]*>/',
+			function( array $matches ) {
+				$remote_filename = $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 ) );
+
+				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 ( ! $thumb ) {
+					$thumb = $attachment_id;
+					return '';
+				} else {
+					return "<img src=\"{$target_url}\" title=\"{$title}\" alt=\"{$alt}\">";
+				}
+			},
+			$meta['body'],
+		);
+	}
+
+	return $thumb;
+}
+
+function import_photo( $remote_path, $target_filename ) {
+	$curl = curl_init( implode( '/', array( 'https://heavymetal.no', $remote_path ) ) );
+
+	if ( false === $curl ) {
+		die( 'Error: Unable to init curl' );
+	}
+
+	curl_setopt( $curl, CURLOPT_RETURNTRANSFER, true );
+	$res = curl_exec( $curl );
+	file_put_contents( $target_filename, $res );
+}
-- 
cgit v1.2.3