<?php
/**
 * Self-hosted Google Fonts.
 *
 * Off until an administrator turns on "Host Google Fonts on this site" (and has accepted
 * the terms). Then Google Fonts stylesheets are downloaded once (server to Google, never
 * visitor to Google), stored in uploads/goconsent-fonts/ and served from the site. Downloads
 * run in the background and in scans, through the WordPress HTTP API; rewrite() runs inside
 * the output-buffer callback and only reads the stored map.
 *
 * @package GoConsent
 */

namespace GoConsent\Frontend;

defined( 'ABSPATH' ) || exit;

/**
 * Downloads and rewrites Google Fonts references.
 */
class LocalFonts {

	const OPTION = 'goconsent_fonts';
	const HOOK   = 'goconsent_fonts_fetch';
	const QUEUE  = 'goconsent_fonts_queue';

	/** Browser user agent: Google serves woff2 to it. */
	const UA = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0 Safari/537.36';

	/**
	 * Whether the administrator turned on font hosting (and accepted the terms).
	 * Nothing contacts Google Fonts unless this is true.
	 *
	 * @return bool
	 */
	public static function enabled() {
		return (bool) ( new \GoConsent\Settings() )->get( 'local_fonts' ) && ( new \GoConsent\Terms() )->accepted();
	}

	/**
	 * Background download hook. Visitors never wait on a download.
	 */
	public static function register() {
		add_action( self::HOOK, array( __CLASS__, 'run_queue' ) );
	}

	/**
	 * Map of stylesheet hash => { url, src, time } or { failed, reason, done, total, src }.
	 *
	 * @return array
	 */
	private static function map() {
		$m = get_option( self::OPTION, array() );
		return is_array( $m ) ? $m : array();
	}

	/**
	 * Normalise a stylesheet URL and hash it. Cache-busting "ver" values are dropped:
	 * Google ignores them, and themes that change them per page would never match.
	 *
	 * @param string $href URL.
	 * @return array{0:string,1:string}
	 */
	private static function key( $href ) {
		$href = html_entity_decode( trim( $href ), ENT_QUOTES | ENT_HTML5 );
		if ( 0 === strpos( $href, '//' ) ) {
			$href = 'https:' . $href;
		}
		$href = (string) preg_replace( '#(?<=[?&])ver=[^&]*(?:&|$)#', '', $href );
		$href = rtrim( $href, '?&' );
		return array( md5( $href ), $href );
	}

	/**
	 * Local URL if this stylesheet is already on the site.
	 *
	 * @param string $href URL.
	 * @return string
	 */
	public static function cached( $href ) {
		list( $key ) = self::key( $href );
		$map         = self::map();
		return empty( $map[ $key ]['url'] ) ? '' : $map[ $key ]['url'];
	}

	/**
	 * Queue a stylesheet for the background download.
	 *
	 * @param string $href URL.
	 */
	public static function queue( $href ) {
		if ( ! self::enabled() ) {
			return;
		}
		list( $key, $href ) = self::key( $href );
		$q                  = get_option( self::QUEUE, array() );
		$q                  = is_array( $q ) ? $q : array();
		if ( isset( $q[ $key ] ) ) {
			return;
		}
		$map = self::map();
		if ( ! empty( $map[ $key ]['failed'] ) && $map[ $key ]['failed'] > time() - HOUR_IN_SECONDS && empty( $map[ $key ]['done'] ) ) {
			return; // Failed outright recently; the next scan retries.
		}
		$q[ $key ] = $href;
		update_option( self::QUEUE, array_slice( $q, -20, null, true ), false );
		if ( ! wp_next_scheduled( self::HOOK ) ) {
			wp_schedule_single_event( time() + 5, self::HOOK );
		}
	}

	/**
	 * Cron: work through the queue for up to 20 seconds, then reschedule.
	 */
	public static function run_queue() {
		if ( ! self::enabled() ) {
			delete_option( self::QUEUE );
			return;
		}
		$q = get_option( self::QUEUE, array() );
		if ( ! is_array( $q ) || ! $q ) {
			return;
		}
		$until = microtime( true ) + 20;
		foreach ( $q as $key => $href ) {
			if ( microtime( true ) > $until ) {
				break;
			}
			$r = self::localize( $href, (int) ( $until - microtime( true ) ) );
			$m = self::map();
			if ( $r || empty( $m[ $key ]['done'] ) ) {
				unset( $q[ $key ] ); // Done, or failed before any file arrived (retried by the next scan).
			}
		}
		update_option( self::QUEUE, $q, false );
		if ( $q ) {
			wp_schedule_single_event( time() + 30, self::HOOK );
		}
	}

	/**
	 * Font files for this stylesheet still being copied, or why the last try failed.
	 *
	 * @return array[] [ src, reason, done, total ] per stylesheet not yet on the site.
	 */
	public static function problems() {
		$out = array();
		foreach ( self::map() as $row ) {
			if ( empty( $row['url'] ) && ! empty( $row['failed'] ) ) {
				$out[] = array(
					'src'    => isset( $row['src'] ) ? $row['src'] : '',
					'reason' => isset( $row['reason'] ) ? $row['reason'] : '',
					'done'   => isset( $row['done'] ) ? (int) $row['done'] : 0,
					'total'  => isset( $row['total'] ) ? (int) $row['total'] : 0,
				);
			}
		}
		return $out;
	}

	/**
	 * Font family names in a Google Fonts URL.
	 *
	 * @param string $href URL.
	 * @return string
	 */
	public static function families( $href ) {
		$q = (string) wp_parse_url( html_entity_decode( $href, ENT_QUOTES | ENT_HTML5 ), PHP_URL_QUERY );
		preg_match_all( '#(?:^|&)family=([^&]+)#', $q, $m );
		$names = array();
		foreach ( $m[1] as $f ) {
			foreach ( explode( '|', rawurldecode( str_replace( '+', ' ', $f ) ) ) as $one ) {
				$names[] = trim( strtok( $one, ':' ) );
			}
		}
		return implode( ', ', array_unique( array_filter( $names ) ) );
	}

	/**
	 * Download a stylesheet and its font files. Already downloaded files are kept,
	 * so a run that runs out of time resumes where it stopped.
	 *
	 * @param string $href   Google Fonts stylesheet URL.
	 * @param int    $budget Seconds available.
	 * @return string Local CSS URL, or '' when not (yet) complete.
	 */
	public static function localize( $href, $budget = 20 ) {
		if ( ! self::enabled() ) {
			return '';
		}
		list( $key, $href ) = self::key( $href );
		$map                = self::map();
		if ( ! empty( $map[ $key ]['url'] ) ) {
			return $map[ $key ]['url'];
		}
		if ( get_transient( 'goconsent_font_lock_' . $key ) ) {
			return '';
		}
		set_transient( 'goconsent_font_lock_' . $key, 1, MINUTE_IN_SECONDS );
		$until = microtime( true ) + max( 5, (int) $budget );

		$uploads = wp_upload_dir();
		$dir     = trailingslashit( $uploads['basedir'] ) . 'goconsent-fonts/' . $key;
		$url     = trailingslashit( $uploads['baseurl'] ) . 'goconsent-fonts/' . $key;
		if ( ! wp_mkdir_p( $dir ) ) {
			return self::fail( $key, $href, __( 'The uploads folder isn’t writable, so the fonts can’t be stored on the site.', 'goconsent' ) );
		}

		$css = is_readable( $dir . '/source.css' ) ? (string) file_get_contents( $dir . '/source.css' ) : ''; // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- local file.
		if ( false === strpos( $css, '@font-face' ) ) {
			$got = self::get_css( $href );
			if ( '' === $got['css'] ) {
				return self::fail( $key, $href, $got['reason'] );
			}
			$css = $got['css'];
			file_put_contents( $dir . '/source.css', $css ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_file_put_contents -- writing to uploads.
		}

		preg_match_all( '#url\((["\']?)(https://fonts\.gstatic\.com/[^)"\']+)\1\)#i', $css, $m );
		$files = array();
		foreach ( array_unique( $m[2] ) as $remote ) {
			$files[ $remote ] = substr( md5( $remote ), 0, 12 ) . '-' . sanitize_file_name( basename( (string) wp_parse_url( $remote, PHP_URL_PATH ) ) );
		}
		$missing = array();
		foreach ( $files as $remote => $name ) {
			if ( ! file_exists( $dir . '/' . $name ) ) {
				$missing[ $remote ] = $name;
			}
		}

		$errors = array();
		foreach ( $missing as $remote => $name ) {
			if ( microtime( true ) > $until ) {
				break;
			}
			$body = self::fetch( $remote );
			if ( '' !== $body ) {
				file_put_contents( $dir . '/' . $name, $body ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_file_put_contents -- writing to uploads.
			} else {
				$errors[] = $remote;
			}
		}

		$done = 0;
		foreach ( $files as $name ) {
			$done += file_exists( $dir . '/' . $name ) ? 1 : 0;
		}
		if ( $done < count( $files ) ) {
			$reason = $errors
				? __( 'Some font files didn’t download. GoConsent keeps the ones it has and tries the rest again.', 'goconsent' )
				: __( 'Still copying: this font has many files. GoConsent continues in the background.', 'goconsent' );
			$out    = self::fail( $key, $href, $reason, $done, count( $files ) );
			if ( ! $errors ) {
				self::queue( $href );
			}
			return $out;
		}

		$local = str_replace(
			array_keys( $files ),
			array_map(
				static function ( $n ) use ( $url ) {
					return $url . '/' . $n;
				},
				array_values( $files )
			),
			$css
		);
		if ( false !== stripos( $local, 'fonts.gstatic.com' ) ) {
			return self::fail( $key, $href, __( 'The font stylesheet uses a format GoConsent can’t copy.', 'goconsent' ) );
		}
		file_put_contents( $dir . '/fonts.css', $local ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_file_put_contents -- writing to uploads.

		$map         = self::map();
		$map[ $key ] = array(
			'url'  => $url . '/fonts.css',
			'src'  => $href,
			'time' => time(),
		);
		update_option( self::OPTION, $map, false );
		delete_transient( 'goconsent_font_lock_' . $key );
		return $map[ $key ]['url'];
	}

	/**
	 * Fetch the Google Fonts stylesheet through the WordPress HTTP API.
	 *
	 * @param string $href URL.
	 * @return array{css:string,reason:string}
	 */
	private static function get_css( $href ) {
		$res  = wp_safe_remote_get(
			$href,
			array(
				'timeout'    => 10,
				'user-agent' => self::UA,
			)
		);
		$code = is_wp_error( $res ) ? 0 : (int) wp_remote_retrieve_response_code( $res );
		$css  = is_wp_error( $res ) ? '' : (string) wp_remote_retrieve_body( $res );
		if ( 200 === $code && false !== strpos( $css, '@font-face' ) ) {
			return array(
				'css'    => $css,
				'reason' => '',
			);
		}
		if ( is_wp_error( $res ) ) {
			return array(
				'css'    => '',
				/* translators: %s: error message. */
				'reason' => sprintf( __( 'The server couldn’t reach Google Fonts (%s). The host or another plugin may block outgoing connections.', 'goconsent' ), $res->get_error_message() ),
			);
		}
		return array(
			'css'    => '',
			/* translators: %d: HTTP status. */
			'reason' => sprintf( __( 'Google Fonts answered with an error (HTTP %d). The font address on the site may be wrong.', 'goconsent' ), $code ),
		);
	}

	/**
	 * Download one font file through the WordPress HTTP API.
	 *
	 * @param string $url fonts.gstatic.com URL.
	 * @return string File contents, or '' on failure.
	 */
	private static function fetch( $url ) {
		$res = wp_safe_remote_get(
			$url,
			array(
				'timeout'    => 15,
				'user-agent' => self::UA,
			)
		);
		return ( ! is_wp_error( $res ) && 200 === (int) wp_remote_retrieve_response_code( $res ) ) ? (string) wp_remote_retrieve_body( $res ) : '';
	}

	/**
	 * Record a failure or partial progress.
	 *
	 * @param string $key    Stylesheet hash.
	 * @param string $href   Stylesheet URL.
	 * @param string $reason Plain-language reason.
	 * @param int    $done   Files copied.
	 * @param int    $total  Files in total.
	 * @return string Always ''.
	 */
	private static function fail( $key, $href = '', $reason = '', $done = 0, $total = 0 ) {
		$map         = self::map();
		$map[ $key ] = array(
			'failed' => time(),
			'src'    => $href,
			'reason' => $reason,
			'done'   => $done,
			'total'  => $total,
		);
		update_option( self::OPTION, $map, false );
		delete_transient( 'goconsent_font_lock_' . $key );
		return '';
	}

	/**
	 * Rewrite Google Fonts <link> and @import references; drop preconnect hints to Google.
	 * Stylesheets not yet on the site are queued and copied in the background.
	 *
	 * @param string $html HTML.
	 * @return string
	 */
	public static function rewrite( $html ) {
		if ( ! self::enabled() ) {
			return $html;
		}
		if ( false === stripos( $html, 'fonts.googleapis.com' ) && false === stripos( $html, 'fonts.gstatic.com' ) ) {
			return $html;
		}
		// Visitors never wait on a download: unknown stylesheets are queued for the background job.
		$local = static function ( $href ) {
			$cached = self::cached( $href );
			if ( '' === $cached ) {
				self::queue( $href );
			}
			return $cached;
		};

		$out = preg_replace_callback(
			'#<link\b[^>]*>#i',
			static function ( $m ) use ( $local ) {
				$tag = $m[0];
				if ( preg_match( '#\brel\s*=\s*["\']?(preconnect|dns-prefetch)#i', $tag ) && preg_match( '#fonts\.(googleapis|gstatic)\.com#i', $tag ) ) {
					return '';
				}
				if ( ! preg_match( '#\bhref\s*=\s*(["\'])((?:https?:)?//fonts\.googleapis\.com/css[^"\']*)\1#i', $tag, $h ) ) {
					return $tag;
				}
				$new = $local( $h[2] );
				return $new ? str_replace( $h[0], 'href=' . $h[1] . esc_url( $new ) . $h[1], $tag ) : $tag;
			},
			$html
		);
		if ( null === $out ) {
			return $html;
		}
		$out2 = preg_replace_callback(
			'#@import\s+(?:url\()?\s*(["\']?)((?:https?:)?//fonts\.googleapis\.com/css[^"\')\s]*)\1\s*\)?\s*;#i',
			static function ( $m ) use ( $local ) {
				$new = $local( $m[2] );
				return $new ? '@import url("' . esc_url( $new ) . '");' : $m[0];
			},
			$out
		);
		return null === $out2 ? $out : $out2;
	}

	/**
	 * Delete downloaded fonts (uninstall).
	 */
	public static function purge() {
		$uploads = wp_upload_dir();
		$base    = trailingslashit( $uploads['basedir'] ) . 'goconsent-fonts';
		if ( ! is_dir( $base ) ) {
			return;
		}
		$it = new \RecursiveIteratorIterator( new \RecursiveDirectoryIterator( $base, \FilesystemIterator::SKIP_DOTS ), \RecursiveIteratorIterator::CHILD_FIRST );
		foreach ( $it as $f ) {
			$f->isDir() ? rmdir( $f->getPathname() ) : wp_delete_file( $f->getPathname() ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_rmdir
		}
		rmdir( $base ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_rmdir
	}
}