Add setup script and filter to import content from remote REST API

env/import-content.php

@@ -0,0 +1,94 @@
+#!/usr/bin/php
+<?php
+
+namespace WordPress_org\Parent_2021\ImportTestContent;
+
+/**
+ * CLI script for generating local test content, fetched from the live learn.wordpress.org site.
+ *
+ * This needs to be run in a wp-env, for example:
+ *
+ * yarn run wp-env run cli "php bin/import-test-content.php"
+ */
+
+// This script should only be called in a CLI environment.
+if ( 'cli' != php_sapi_name() ) {
+	die();
+}
+
+
+$opts = getopt( '', array( 'url:' ) );
+
+require dirname( dirname( __FILE__ ) ) . '/wp-load.php';
+
+if ( 'local' !== wp_get_environment_type() ) {
+	die( 'Not safe to run on ' . esc_html( get_site_url() ) );
+}
+
+if ( empty( $opts[ 'url' ] ) || $opts[ 'url' ] !== esc_url_raw( $opts[ 'url' ], ['https'] ) ) {
+	die( 'Invalid url parameter ' . esc_html( $opts[ 'url' ] ) );
+}
+
+/**
+ * Sanitize postmeta from the rest API for the format required by wp_insert_post.
+ *
+ * @return array An array suitable for meta_input.
+ */
+function sanitize_meta_input( $meta ) {
+	$meta = array( $meta );
+	foreach ( $meta as $k => $v ) {
+		if ( is_array( $v ) ) {
+			$meta[ $k ] = implode( ',', $v );
+		}
+	}
+
+	return $meta;
+}
+
+/**
+ * Import posts from a remote REST API to the local test site.
+ *
+ * @param string $rest_url The remote REST API endpoint URL.
+ */
+function import_rest_to_posts( $rest_url ) {
+	$response = wp_remote_get( $rest_url );
+	$status_code = wp_remote_retrieve_response_code( $response );
+
+	if ( is_wp_error( $response ) ) {
+		die( esc_html( $response->get_error_message() ) );
+	} elseif ( 200 !== wp_remote_retrieve_response_code( $response ) ) {
+		die( esc_html( "HTTP Error $status_code \n" ) );
+	}
+
+	$body = wp_remote_retrieve_body( $response );
+	$data = json_decode( $body );
+
+	foreach ( $data as $post ) {
+		echo esc_html( "Got {$post->type} {$post->id} {$post->slug}\n" );
+
+		// Surely there's a neater way to do this.
+		$newpost = array(
+			'import_id' => $post->id,
+			'post_date' => gmdate( 'Y-m-d H:i:s', strtotime( $post->date ) ),
+			'post_name' => $post->slug,
+			'post_title' => $post->title,
+			'post_status' => $post->status,
+			'post_type' => $post->type,
+			'post_title' => $post->title->rendered,
+			'post_content' => ( $post->content_raw ?? $post->content->rendered ),
+			'post_parent' => $post->parent,
+			'comment_status' => $post->comment_status,
+			'meta_input' => sanitize_meta_input( $post->meta ),
+		);
+
+		$new_post_id = wp_insert_post( $newpost, true );
+
+		if ( is_wp_error( $new_post_id ) ) {
+			die( esc_html( $new_post_id->get_error_message() ) );
+		}
+
+		echo esc_html( "Inserted $post->type $post->id as $new_post_id\n" );
+	}
+}
+
+import_rest_to_posts( $opts['url'] );

env/setup.sh

@@ -14,3 +14,6 @@ wp import "${root}/env/data.xml" --authors=create
 
 wp option update show_on_front 'page'
 wp option update page_on_front 8891
+
+npm run wp-env run cli "php env/import-content.php --url 'https://wordpress.org/main-test/wp-json/wp/v2/posts?context=wporg_export&per_page=50'"
+npm run wp-env run cli "php env/import-content.php --url 'https://wordpress.org/main-test/wp-json/wp/v2/pages?context=wporg_export&per_page=50'"

source/wp-content/themes/wporg-main-2022/functions.php

@@ -21,3 +21,14 @@ function enqueue_assets() {
 		filemtime( __DIR__ . '/style.css' )
 	);
 }
+
+/**
+ * Make posts and pages available for export from the staging site, so the import script can
+ * fetch them to a local dev environment.
+ */
+add_filter( 'wporg_export_context_post_types', function( $types ) {
+    return array_merge( $types, [
+        'post',
+        'page',
+        ]);
+} );