Function that fixes duplicate filename bug found in attachment_url_to_postid

attachment_url_to_postid_fix.php

function attachment_url_to_postid_case_sensitive( $url ) {
	global $wpdb;

	$dir = wp_get_upload_dir();
	$path = $url;

	$site_url = parse_url( $dir['url'] );
	$image_path = parse_url( $path );

	//force the protocols to match if needed
	if ( isset( $image_path['scheme'] ) && ( $image_path['scheme'] !== $site_url['scheme'] ) ) {
		$path = str_replace( $image_path['scheme'], $site_url['scheme'], $path );
	}

	if ( 0 === strpos( $path, $dir['baseurl'] . '/' ) ) {
		$path = substr( $path, strlen( $dir['baseurl'] . '/' ) );
	}

	$path_array = explode( '.', $path );

	if( count( $path_array ) ) {
		// Pop the last element off the array to grab the filename
		$path_duplicated_filename = array_pop( $path_array );

		// Piece everything back together, putting back any existing '.' that were in the filename
		$path_duplicated_base     = implode( '.', $path_array ) . '-';
		$path_duplicated          = $path_duplicated_base . '%' . $path_duplicated_filename;
	}

	// Perform the search, looking for the exact value OR a file in the format 'Filename-'
	$sql = $wpdb->prepare(
		"SELECT post_id, meta_value FROM $wpdb->postmeta WHERE meta_key = '_wp_attached_file' AND ( meta_value = %s OR meta_value LIKE %s )",
		$path,
		$path_duplicated
	);

	$results      = $wpdb->get_results( $sql );

	// Check to see if the path matches of the meta values returned (case-sensitive). If so, we've found the ID
	$meta_values             = wp_list_pluck( $results, 'meta_value', 'post_id' );
	$post_id                 = array_search( $path, $meta_values, true );
	$similarity_percent      = 0;
	$similarity_percent_temp = 0;

	// If there is no exact match and multiple results, parse through and match for the correct URL.
	if( count( $results ) > 1 &&
		empty( $post_id ) ) {

		// If an image has been uploaded as untitled.jpg and then another one is uploaded named Untitled.jpg, we need
		// to match capitalization for the file. That file is most likely in the format Untitled-1.jpg in the meta_value
		// field, so we'll need to check for that being case-sensitive
		foreach( $results as $result ) {

			// Separate just the name of the file.
			$path_temp_array = explode( '.', $result->meta_value );
			$path_temp = implode( '.', $path_temp_array );

			// Checks to see if the base is a part of the string (case-sensitive). If so, we found our match.
			if( false !== strpos( $path_temp, $path_duplicated_base ) ) {

				// Calculate similarity % to get the most similar result if we have other filenames that are very similar.
				similar_text( $path_temp, $path_duplicated_base, $similarity_percent_temp );

				if( $similarity_percent_temp > $similarity_percent ) {
					$similarity_percent = $similarity_percent_temp;
					$post_id = $result->post_id;
				}
			}
		}
	}

	if( empty( $post_id ) ) {
		$post_id = 0;
	}

	/**
	 * Filters an attachment id found by URL.
	 *
	 * @since 4.2.0
	 *
	 * @param int|null $post_id The post_id (if any) found by the function.
	 * @param string   $url     The URL being looked up.
	 */
	return (int) apply_filters( 'attachment_url_to_postid', $post_id, $url );
}