Flickr-to-WordPress: a plugin to use a WordPress powered API to find/replace Flickr references in another WordPress site

jch-flickr-fixer.php

<?php
/**
 * Plugin Name: Flickr Fixer
 * Plugin URI: https://chrishardie.com/
 * Description: Find/replace Flickr references
 * Version: 1.0
 * Author: Chris Hardie
 * Author URI: https://chrishardie.com/
 */


if ( ! defined( 'WP_CLI' ) || ! WP_CLI ) {
	return;
}

class JCH_Flickr_Fixer extends WP_CLI_Command {

	/**
	 * Fix Flickr References
	 *
	 * ## OPTIONS
	 *
	 * [--post_id=<id>]
	 * : Specify an individual post ID to work with
	 *
	 * [--dry_run=false]
	 * : Actually update post content in a database write operation
	 *
	 * [--debug]
	 * : Display debug output
	 *
	 * @subcommand fix-refs [--dry_run=false] [--post_id=<id>] [--debug]
	 *
	 * @param array $args Args.
	 * @param array $args_assoc Associative args.

	 */
	public function fix_refs( $args, $assoc_args ) {

		global $wpdb;

		// Get a specific post, or all the posts.
		if ( isset( $assoc_args['post_id'] ) && is_numeric( $assoc_args['post_id'] ) ) {
			WP_CLI::line( 'Getting selected post...' );

			$flickr_post_ids   = $wpdb->get_col( "SELECT ID FROM {$wpdb->posts} WHERE post_type = 'post' AND ID = " . (int) $assoc_args['post_id'] );

		} else {
			WP_CLI::line( 'Getting all posts containing Flickr references...' );

			$flickr_post_ids   = $wpdb->get_col( "SELECT ID FROM {$wpdb->posts} WHERE post_type = 'post' AND ( post_content LIKE '%%" . esc_sql( $wpdb->esc_like( 'flickr.com/photos/chrishardie' ) ) . "%%' OR post_content LIKE '%%" . esc_sql( $wpdb->esc_like( 'flickr.com/photos/11288301@N00' ) ) . "%%' )" );

		}

		$flickr_post_count = count( $flickr_post_ids );

		WP_CLI::line( 'Found ' . (int) $flickr_post_count . ' posts to process.' );

		if ( ! $flickr_post_count ) {
			WP_CLI::line( 'No work to do! Exiting.' );
			exit;
		}

		$progress = \WP_CLI\Utils\make_progress_bar( 'Processing Flickr references', $flickr_post_count );

		// Keep track of all replacements across all posts.
		$replacements_made = 0;

		// Work on 20 posts at a time.
		while ( $post_ids = array_splice( $flickr_post_ids, 0, 20 ) ) {

			$posts = $wpdb->get_results( "SELECT ID, post_content FROM {$wpdb->posts} WHERE post_type = 'post' AND ID IN (" . implode( ',', $post_ids ) . ')' );
			if ( empty( $posts ) ) {
				break;
			}

			foreach ( $posts as $the_post ) {

				// I couldn't get edit_post_link() to work. This is dumb.
				$admin_edit_url = admin_url( 'post.php?post=' ) . $the_post->ID . '&action=edit';

				WP_CLI::debug( "Working on Post ID: {$the_post->ID}: " . $admin_edit_url );

				// We don't end up using the $original_content variable but you could use it for an extra comparison check if needed.
				$original_content = $the_post->post_content;
				$new_content      = $original_content;

				// Keep track of the number of Flickr references found in this post.
				$ref_count = 0;

				// Despite all this fancy automation, there are still going to be references that need special/manual handling.
				// This array is for that. The key is the string you want to find, the value is the string to replace it with.
				$manual_find_replace = array(
					'11288301@N00' => 'chrishardie', // Older references to my Flickr user ID
					'http://www.flickr.com/photos/chrishardie/sets/' => 'https://photos.chrishardie.com/flickrset/', // References to Flickr sets
					'https://www.flickr.com/photos/chrishardie/sets/' => 'https://photos.chrishardie.com/flickrset/',
					'http://www.flickr.com/photos/chrishardie/tags/' => 'https://photos.chrishardie.com/tag/', // Flickr tags
					'https://www.flickr.com/photos/chrishardie/tags/' => 'https://photos.chrishardie.com/tag/',
					'href="http://www.flickr.com/photos/chrishardie"' => 'href="https://photos.chrishardie.com/"', // Plain old links to my profile
					'href="http://www.flickr.com/photos/chrishardie/"' => 'href="https://photos.chrishardie.com/"',
				);

				// Given those manual items, fix them quickly and move on.
				foreach ( $manual_find_replace as $key => $manual_ref ) {

					$new_content = str_replace( $key, $manual_ref, $new_content, $manual_replacements_made );

					if ( 0 < $manual_replacements_made ) {
						$ref_count++;
						$replacements_made += $manual_replacements_made;
					}
				}

				// Now for the fun stuff.
				// Find <a> references with or without <img> references inside
				preg_match_all( '/(<a[^>]+?href="https?:\/\/www.flickr.com\/photos\/chrishardie\/\d+.+?[<img.+?\/>]?<\/a>)/', $new_content, $refs_complex, PREG_SET_ORDER );

				foreach ( $refs_complex as $ref ) {

					$ref_count++;

					// It would be too hard to use regexp parsing given all the different ways the attributes of a tag can be ordered.
					// Instead, use DOMDocument to programmatically parse the tag attributes and update them.
					$dom = new DOMDocument();

					// Get the HTML of the pattern we matched above.
					$dom->loadHTML( $ref[0] );

					// Get the A and IMG tags into an array
					$a_tags   = $dom->getElementsByTagName( 'a' );
					$img_tags = $dom->getElementsByTagName( 'img' );

					// We're only expecting one link and one image per match.
					if ( 1 < count( $a_tags ) || 1 < count( $img_tags ) ) {
						WP_CLI::warning( 'More than one a tag or img tag: ' . $admin_edit_url );
					} else {
						// Extract the Flickr link in particular we want to replace
						$flickr_link = preg_replace( '/(https?:\/\/www\.flickr\.com\/photos\/chrishardie\/\d+\/)(in\/\S+)?/', '$1', $a_tags[0]->getAttribute( 'href' ) );

						// For our API lookup, add a trailing slash if there isn't one
						$flickr_link = rtrim( $flickr_link, '/' ) . '/';

						// Get width and height for the img tag, if there is one
						// This allows us to retrieve an img src value appropriate for replacing, instead of full size
						if ( ! empty( $img_tags[0] ) ) {
							$width  = $img_tags[0]->getAttribute( 'width' );
							$height = $img_tags[0]->getAttribute( 'height' );
						} else {
							$width  = null;
							$height = null;
						}

						// Look it up on our WordPress photo site
						$photo_data = $this->get_photo_post_by_flickr_id( $flickr_link, $width, $height );

						// See if we got something back
						if ( ! $photo_data ) {
							WP_CLI::warning( 'Could not find a valid photo site post for ' . $flickr_link . ' in ' . $admin_edit_url );
							continue;
						} else {
							WP_CLI::debug( 'Found valid photo site post for Flickr URL.' );
							WP_CLI::debug( $flickr_link . ' --> ' . $photo_data->permalink );

							// Replace it
							$a_tags[0]->setAttribute( 'href', $photo_data->permalink );
							$dom->saveHTML( $a_tags[0] );
							$replacements_made++;

							// If we're working with an img tag, update it too
							if ( ! empty( $img_tags[0] ) ) {

								if ( empty( $photo_data->thumbnail_url ) ) {
									WP_CLI::warning( 'No valid replacement image src url in ' . $admin_edit_url );
								} else {
									$old_img_src = $img_tags[0]->getAttribute( 'src' );
									WP_CLI::debug( $old_img_src . ' --> ' . $photo_data->thumbnail_url );
									$img_tags[0]->setAttribute( 'src', $photo_data->thumbnail_url );
									$dom->saveHTML( $img_tags[0] );
									$replacements_made++;
								}
							}

							# Strip out the DOCTYPE, html & body tags that DOMDocument adds to the HTML doc
							$final_html = preg_replace( '~<(?:!DOCTYPE|/?(?:html|body))[^>]*>\s*~i', '', $dom->saveHTML() );

							// Update the content variable with the right HTML
							$new_content = str_replace( $ref[0], $final_html, $new_content );

						}
					}
				}

				// Find remaining references - usually a link on a line by itself for oembed use, or inside an <a> tag
				preg_match_all( '/(https?:\/\/www\.flickr\.com\/photos\/chrishardie\/\d+\/)(in\S+)?/', $new_content, $refs_solo, PREG_SET_ORDER );

				foreach ( $refs_solo as $ref ) {
					$ref_count++;

					$found_flickr_ref = $ref[1];

					// The first array slot contains the matched Flickr photo URL
					$photo_data = self::get_photo_post_by_flickr_id( $found_flickr_ref );

					if ( ! $photo_data ) {
						WP_CLI::warning( 'Could not find a valid photo site post for ' . $found_flickr_ref . ' in ' . $admin_edit_url );
						continue;
					} else {
						WP_CLI::debug( 'Found valid photo site post for Flickr URL.' );
						WP_CLI::debug( $found_flickr_ref . ' --> ' . $photo_data->permalink );

						// Update the content
						$new_content = str_replace(
							$ref[0], // The original matched pattern in full
							$photo_data->permalink,
							$new_content,
							$solo_replacements_made
						);

						$replacements_made += $solo_replacements_made;

					}
				}


				// See if we found any references that we touched, so we know if an actual post content update is needed.
				if ( 0 < $ref_count ) {
					WP_CLI::debug( $ref_count . ' Flickr references found, so updating content...' );

					$post_to_update = array(
						'ID'           => $the_post->ID,
						'post_content' => $new_content,
					);

					// Only do an actual database update if the dry_run=false flag is passed.
					if ( ! empty( $assoc_args['dry_run'] ) && 'false' === $assoc_args['dry_run'] ) {
						$update_result = wp_update_post( $post_to_update );
						if ( is_wp_error( $update_result ) ) {
							WP_CLI::error( 'There was a problem updating the post content.' );
						} else {
							WP_CLI::debug( 'Post successfully updated.' );
							clean_post_cache( $the_post->ID );
							WP_CLI::line( 'Updated post: ' . get_the_permalink( $the_post->ID ) );
						}
					} else {
						WP_CLI::debug( 'Dry run only, so not actually making DB changes.' );
					}
				} else {
					WP_CLI::warning( 'No references found in processing, probably something missing in ' . $admin_edit_url );
				}

				$progress->tick();

			}

			// Do some memory cleanup so we don't lose control.
			self::stop_the_insanity();

		}

		$progress->finish();

		WP_CLI::success( $replacements_made . ' replacement(s) made across all posts' );

	}

	/**
	 * Clear all of the caches for memory management
	 */
	public static function stop_the_insanity() {
		/**
		 * @var \WP_Object_Cache $wp_object_cache
		 * @var \wpdb $wpdb
		 */
		global $wpdb, $wp_object_cache;

		$wpdb->queries = array(); // or define( 'WP_IMPORTING', true );

		if ( is_object( $wp_object_cache ) ) {
			$wp_object_cache->group_ops      = array();
			$wp_object_cache->stats          = array();
			$wp_object_cache->memcache_debug = array();
			$wp_object_cache->cache          = array();

			if ( method_exists( $wp_object_cache, '__remoteset' ) ) {
				$wp_object_cache->__remoteset(); // important
			}
		}
	}

	/**
	 * Look up the photo site post info via API using the Flickr ID and possibly width/height
	 */
	public function get_photo_post_by_flickr_id( $flickr_url = null, $width = null, $height = null ) {

		$cache_group  = 'flickr_fixer';
		$cache_expire = 7 * DAY_IN_SECONDS;

		if ( empty( $flickr_url ) ) {
			return false;
		}

		// Always use the SSL version, since that's what was provided in the Flickr site data export/import process.
		$flickr_url = preg_replace( '/^http:/i', 'https:', $flickr_url );

		$photo_site_api_url = 'https://my-wp-photo-website.com/wp-json/myphotos/v1/find-by-flickr-url/';

		$cache_key = $flickr_url;

		// This is the main key we use to look up the presence of a Flickr photo on the new WordPress site
		$photo_site_api_url = add_query_arg(
			array(
				'flickr-url' => $flickr_url,
			),
			$photo_site_api_url
		);

		// If we received a width, use it
		if ( is_numeric( $width ) ) {
			$photo_site_api_url = add_query_arg(
				array(
					'width' => $width,
				),
				$photo_site_api_url
			);

			$cache_key .= '--' . $width;

		}

		// If we received a height, use it
		if ( is_numeric( $height ) ) {
			$photo_site_api_url = add_query_arg(
				array(
					'height' => $height,
				),
				$photo_site_api_url
			);

			$cache_key .= '--' . $height;
		}

		// It makes sense to cache the API responses from our WordPress-hosted photo site, they'll rarely change.
		WP_CLI::debug( 'Checking cache for ' . $cache_key );

		$cache_data = wp_cache_get( $cache_key, $cache_group, false, $found );

		if ( ( true === $found ) && ! empty( $cache_data ) ) {
			WP_CLI::debug( 'Cache hit' );
			return $cache_data;
		}

		// If no cache hit, do an actual lookup via API.
		WP_CLI::debug( 'Fetching ' . $photo_site_api_url );

		$photo_site_request = wp_remote_get( $photo_site_api_url );

		// Make sure we got a valid JSON response
		if ( is_wp_error( $photo_site_request ) ) {
			return false;
		}

		$photo_site_response = wp_remote_retrieve_body( $photo_site_request );

		$photo_data = json_decode( $photo_site_response, false );

		// If we got a valid result from the API, cache and return it
		if ( ! empty( $photo_data ) && 'found' === $photo_data->result ) {

			$cache_set_result = wp_cache_set( $cache_key, $photo_data, $cache_group, $cache_expire );

			if ( false === $cache_set_result ) {
				WP_CLI::error( 'Failed to set the cache value for key ' . $cache_key );
			}

			return $photo_data;
		} else {
			return false;
		}
	}

}

// Register this command.
WP_CLI::add_command( 'flickr-fixer', 'JCH_Flickr_Fixer' );