Handles migrating images to local database/file-system

trisha-import-images.php

<?php
/*
Plugin Name: Fix Image Paths
Plugin URI: http://webdevstudios.com
Description: Plugin used to fix images that did not import properly in the MigrationService Plugin.
Author: WebDevStudios
Author URI: http://webdevstudios.com
Version: 1.0.0
License: GPLv2
*/
///wp-admin?run-trisha-import=true&posts-offset=0


add_action( 'admin_menu', 'add_image_path_fix_menu' );
function add_image_path_fix_menu() {

	add_management_page(
		'Fix Image Paths',
		'Fix Image Paths',
		'manage_options',
		'fix-image-paths',
		'display_fix_image_path_page'
	);

}

function display_fix_image_path_page() {?>
	<form name="form1" method="post" action="">
        <p><label for="accepted"><input type="checkbox" value="Accept" id="accepted" name="accepted" /> I have reviewed the settings above and want to run the tool.</label></p>
        <p class="submit">
        <input type="submit" name="Submit" class="button-primary" value="<?php esc_attr_e('Fix the Image Paths') ?>" />
        </p>
    </form>
<?
}

class TrishasImageImporter {

	public function __construct( $args, $test_run ) {
		$this->test_run = $test_run;
		$this->process_all_posts( $args );
	}

	public function process_all_posts( $args ) {
		$posts = get_posts( $args );
		if ( $posts ) {
			foreach ( $posts as $post ) {
				$this->post_id = $post->ID;
				$post->post_content = $this->formatHTML( $post->post_content, true );

				if ( $this->test_run ) {
					wp_die( '<xmp>'. print_r( $post->post_content, true ) .'</xmp>' );
				}
				wp_update_post( $post );
			}
		}
	}

	public function maybe_import_this( $content ) {
		$pattern = '([^\"]*\.(jpg|png|bmp|JPG))';
		$found_match = preg_match( $pattern, $content, $matches );
		if ( ! $found_match ) {
			// No images to import here
			return false;
		}

		if ( false != stripos( $matches[0], site_url() ) ) {
			// Image is already local to this site.. do not import
			return false;
		}

		return $matches[0];
	}

	public function maybe_sideload( $image, $attr ) {
		$imagesrc = $image->getAttribute( $attr );

		if ( isset( $this->images_imported[ $imagesrc ] ) ) {
			$image->setAttribute( $attr, $this->images_imported[ $imagesrc ][ 'url' ] );
			return false;
		}

		$imagesrc = $this->maybe_import_this( $imagesrc );
		if ( ! $imagesrc ) {
			return false;
		}

		echo  '<xmp>'. print_r( $imagesrc, true ) .'</xmp>';

		$attachment = $this->handle_sideload( $image_src );

		//wp_die( '<xmp>'. print_r( $attachment, true ) .'</xmp>' );

		echo 'Error Messages: ';
		if ( is_wp_error( $attachment ) ) {
			// log error, show error
			echo $attachment->get_error_message( $this->post_id ) . '<br />';
			return false;
		}
		$new_url = $attachment[ 'url' ];
		$image->setAttribute( $attr, $new_url );
		return $image;
	}

	/**
	 * @param $html_content
	 * @param string|bool $replaceUrl If string, replace URLs, if set to true then it will import
	 *
	 * @return mixed
	 */
	public function formatHTML( $html_content, $replaceUrl ) {

		if ( empty( $html_content ) ) {
			return $html_content;
		}

		// check for emojicons
		$html_content = str_replace( "\xF0\x9F\x98\x83", ":)", $html_content );

		$dom = new DOMDocument;

		libxml_use_internal_errors( true ); //hide errors on the load due to the HTML 5 glitch
		$dom->loadHTML( $html_content );
		libxml_use_internal_errors( false ); //show errors moving forward

		// Check for images
		$images = $dom->getElementsByTagName( 'img' );
		foreach ( $images as $image ) {
			$image = $this->maybe_sideload( $image, 'src' );
		}

		// check for links to images/files that are full size.
		$imagelinks = $dom->getElementsByTagName( 'a' );
		foreach ( $imagelinks as $imagelink ) {
			$imagelink = $this->maybe_sideload( $imagelink, 'href' );
		}


		// //now check for internal links that do not have the full bing blogs url
		// $links = $dom->getElementsByTagName( 'a' );
		// foreach ( $links as $link ) {
		// 	if ( substr( $link->getAttribute( 'href' ), 0, 3 ) == "/b/" ) {
		// 		$link->setAttribute( 'href', 'http://' . $this->this_domain . $link->getAttribute( 'href' ) );
		// 	}
		// }

		$updated_html_content = preg_replace( '/^<!DOCTYPE.+?>/', '', str_replace( array(
			'<html>',
			'</html>',
			'<body>',
			'</body>'
		), array(
			'',
			'',
			'',
			''
		), $dom->saveHTML() ) );

		//$updated_html_content = $dom->saveHTML();
		return $updated_html_content;
	}

	public function handle_sideload( $file, $generate_metadata = false ) {

		if ( isset( $this->images_imported[ $file ] ) ) {
			return $this->images_imported[ $file ];
		}

		// Download file to temp location
		$tmp = download_url( $file );

		// Set variables for storage
		// fix file filename for query strings
		preg_match( '/[^\?]+\.(jpe?g|jpe|JPG|gif|png|bmp)\b/i', $file, $matches );

		$file_array = array(
			'name' => basename( $matches[1] ),
			'tmp_name' => $tmp
		);

		// If error storing temporarily
		if ( is_wp_error( $tmp ) ) {
			$file_array[ 'tmp_name' ] = '';

			return $tmp;
		}

		// do the validation and storage stuff
		$id = media_handle_sideload( $file_array, $this->post_id );

		// If error storing permanently, unlink
		if ( is_wp_error( $id ) ) {
			if ( !empty( $file_array[ 'tmp_name' ] ) ) {
				@unlink( $file_array[ 'tmp_name' ] );
			}

			return $id;
		}

		$fullsizepath = get_attached_file( $id );

		if ( false === $fullsizepath || ! file_exists( $fullsizepath ) ) {
			return new WP_Error( '', 'Image not saved correctly for resizing' );
		}

		if ( $generate_metadata ) {
			$metadata = wp_generate_attachment_metadata( $id, $fullsizepath );

			if ( is_wp_error( $metadata ) ) {
				return $metadata;
			}
			elseif ( empty( $metadata ) ) {
				return new WP_Error( '', 'Image resizing incomplete' );
			}

			// If this fails, then it just means that nothing was changed (old value == new value)
			wp_update_attachment_metadata( $id, $metadata );
		}

		$url = wp_get_attachment_url( $id );

		return array( 'ID' => $id, 'url' => $url );

	}

}

add_action( 'admin_init', 'trisha_script_init' );
function trisha_script_init() {
	if ( isset( $_GET['run-trisha-import'] ) ) {
		$per_page = 5;
		$args = array(
			'posts_per_page' => $per_page,
			'offset' => isset( $_GET['posts-offset'] ) ? ( absint( $_GET['posts-offset'] ) * $per_page ) : 0,
		);
		$test_run = 'test-run' == $_GET['run-trisha-import'];
		new TrishasImageImporter( $args, $test_run );
	}
}