<?php
/**
 * Asset_Combiner — concatenates enqueued local CSS / JS into a single
 * combined file per type. Hooked from LegacyMinifier when the
 * `combine_css` / `combine_js` toggles are on.
 *
 * Algorithm (CSS):
 *   1. wp_enqueue_scripts @ 999 — walk WP_Styles->queue, partition into
 *      local + external. External (full http(s):// to other origins,
 *      data: URIs, protocol-relative pointing elsewhere) stay enqueued
 *      as-is; local handles get pulled out of the queue.
 *   2. Build cache key = md5(JSON({handle => [src, mtime]})). When the
 *      combined file already exists for that key, skip generation.
 *   3. Otherwise: read each source body, resolve recursive @import
 *      statements (depth-limited), rewrite url(...) paths to absolute,
 *      concat with a small `/* xspeed: HANDLE *​/` header per chunk for
 *      debug-traceability, write to XSPEED_CACHE_DIR/min/combined/.
 *   4. Register the combined file as a single new handle
 *      `xspeed-combined-css` and re-add it to the queue. The original
 *      handles stay registered (so other plugins that look them up
 *      still find their metadata) but are pulled from the queue —
 *      they won't print <link> tags.
 *
 * JS path is the same, minus @import (no JS analogue) and url()
 * rewriting (JS strings are too varied to safely rewrite). External
 * + async + deferred scripts (deferred via WP_Scripts->add_data
 * 'strategy' OR the script_loader_tag filter from Minify_Filters)
 * stay un-combined.
 *
 * Cache lives in {$min_dir}/combined/ — separate from the per-file
 * minify cache so purge can target them independently if needed.
 *
 * @package XSpeed
 */

declare(strict_types=1);

namespace XSpeed;

defined( 'ABSPATH' ) || exit;

final class Asset_Combiner {

	public const MAX_IMPORT_DEPTH = 3;

	/**
	 * Path to the combine cache dir. Created on first write.
	 */
	public static function cache_dir(): string {
		return trailingslashit( XSPEED_CACHE_DIR ) . 'min/combined';
	}

	/**
	 * URL prefix matching cache_dir(). Built from content_url, not by
	 * string-replacing filesystem paths (see class-minifier.php for the
	 * same rationale).
	 *
	 * The scheme is forced to match the page's — `content_url()` derives
	 * its scheme from `is_ssl()`, which returns false behind a TLS-
	 * terminating reverse proxy / load balancer (common on managed hosts),
	 * so it can hand back an `http://` URL on an `https` page. The browser
	 * then blocks the combined stylesheet as mixed content and the whole
	 * page renders unstyled. Re-scheme the URL to the site's actual scheme
	 * so the <link> always matches the page. (FBS-83633)
	 */
	public static function cache_url(): string {
		$url = trailingslashit( content_url( 'cache/xspeed' ) ) . 'min/combined';
		// Match the site's registered scheme (home_url), NOT is_ssl() —
		// which set_url_scheme() would consult with no explicit scheme, and
		// which is the very signal that misreports behind a proxy.
		$scheme = wp_parse_url( home_url(), PHP_URL_SCHEME ) ?: 'https';
		return set_url_scheme( $url, $scheme );
	}

	/**
	 * Combine local enqueued styles into one file.
	 */
	public static function combine_styles(): void {
		global $wp_styles;
		if ( ! $wp_styles instanceof \WP_Styles || empty( $wp_styles->queue ) ) {
			return;
		}

		// Group combinable handles by media type. Historically every sheet
		// whose media wasn't all/screen was dropped from combining — but on
		// page-builder sites (Elementor + Essential Addons + BetterDocs) a large
		// share of the stylesheets carry responsive/print media, so dropping
		// them starved the `all` bucket below the 2-handle floor and the whole
		// combine step silently no-op'd (the page shipped 60 separate <link>s
		// even with combine_css ON). Instead we bucket PER media type and emit
		// one combined file per group with the correct `media` attribute, so
		// nothing is dropped and the combinable majority always merges. (FBS-83653)
		$buckets = self::collect_local_handles( $wp_styles );
		foreach ( $buckets as $media => $bucket ) {
			if ( count( $bucket ) < 2 ) {
				continue; // nothing to gain from combining a single file in this group.
			}
			self::combine_media_group( $wp_styles, $media, $bucket );
		}
	}

	/**
	 * Combine one media group's handles into a single stylesheet and wire it
	 * onto the group's carrier handle.
	 *
	 * @param string                     $media  The media attribute for this group ('all', 'print', …).
	 * @param array<string,array<mixed>> $bucket handle => info map.
	 */
	private static function combine_media_group( \WP_Styles $wp_styles, string $media, array $bucket ): void {
		$key = self::cache_key( $bucket );
		$dir = self::cache_dir();
		$out_file = $dir . '/combined-' . $key . '.css';
		$out_url  = self::cache_url() . '/combined-' . $key . '.css';

		if ( ! file_exists( $out_file ) ) {
			self::ensure_dir( $dir );
			$contents = '';
			foreach ( $bucket as $handle => $info ) {
				$body      = self::read_local_file( $info['path'] );
				if ( '' === $body ) {
					continue;
				}
				$body      = self::resolve_imports( $body, $info['url'], 0 );
				$body      = self::rewrite_url_paths( $body, $info['url'] );
				$contents .= "/* xspeed: $handle */\n" . $body . "\n";
			}
			// Atomic write: file_put_contents with LOCK_EX so concurrent
			// renders don't race.
			// phpcs:ignore WordPress.WP.AlternativeFunctions.file_put_contents_file_put_contents -- WP_Filesystem requires admin context, unavailable on frontend.
			file_put_contents( $out_file, $contents, LOCK_EX );
		}

		// Point the FIRST combined handle at the combined file and blank the
		// rest. This is deliberate — we do NOT enqueue a fresh
		// `xspeed-combined-css` handle, because WordPress would print it at
		// the tail of the queue, AFTER any non-combinable stylesheets
		// (media-query sheets like woocommerce-smallscreen, wc-blocks-*,
		// external fonts) that originally sat between/after the combined
		// handles. That reorders the cascade and breaks layout — e.g. the
		// WooCommerce/Astra grid + sidebar widths get overridden by rules
		// that should have lower priority. By reusing the first combined
		// handle's own queue slot for the combined <link>, the merged CSS
		// prints exactly where the earliest source stylesheet used to be,
		// preserving cascade order. (FBS-83114/83116)
		//
		// The remaining combined handles keep their registration + queue
		// membership (src blanked) so their wp_add_inline_style() data still
		// prints — WordPress only emits inline data for handles still in the
		// print queue, and some themes (Astra) attach that dynamic CSS on a
		// hook LATER than this priority-999 pass, so we can't harvest it now.
		// Dropping it is what made "combine CSS break the site".
		// The carrier is the FIRST bucket handle that WordPress hasn't already
		// printed. A block theme (Twenty Twenty-Five, etc.) prints some of its
		// per-block style handles BEFORE this priority-999 pass, marking them
		// `done`; pointing a done handle at the combined file emits no <link>
		// at all — the merged CSS silently vanishes and the whole site renders
		// unstyled. Skipping done handles guarantees the carrier still prints.
		// If every bucket handle is already done, register a dedicated combined
		// handle so the CSS is never lost (cascade tail is far better than no
		// styles). (FBS-83633)
		$done       = (array) $wp_styles->done;
		$carrier_set = false;
		foreach ( $bucket as $handle => $info ) {
			$reg = $wp_styles->registered[ $handle ] ?? null;
			if ( ! $reg instanceof \_WP_Dependency ) {
				continue;
			}
			if ( ! $carrier_set && ! in_array( $handle, $done, true ) ) {
				// Carry the combined file on this (not-yet-printed) handle's slot.
				$reg->src    = $out_url;
				$reg->ver    = $key;
				$reg->args   = $media;
				$carrier_set = true;
			} else {
				// Inline-only carrier: no <link>, keep inline CSS printable.
				$reg->src = false;
				$reg->ver = null;
			}
		}

		// Fallback: every bucket handle was already printed, so no carrier
		// could emit the combined <link>. Register + enqueue a dedicated
		// handle so the merged CSS still loads (appended at the tail — not
		// cascade-ideal, but infinitely better than a fully unstyled page).
		if ( ! $carrier_set ) {
			$combined_handle = 'xspeed-combined-css-' . $media;
			wp_register_style( $combined_handle, $out_url, array(), $key, $media );
			wp_enqueue_style( $combined_handle );
		}
	}

	/**
	 * Combine local enqueued scripts into one file.
	 */
	public static function combine_scripts(): void {
		global $wp_scripts;
		if ( ! $wp_scripts instanceof \WP_Scripts || empty( $wp_scripts->queue ) ) {
			return;
		}

		$bucket = self::collect_local_script_handles( $wp_scripts );
		if ( count( $bucket ) < 2 ) {
			return;
		}

		$key = self::cache_key( $bucket );
		$dir = self::cache_dir();
		$out_file = $dir . '/combined-' . $key . '.js';
		$out_url  = self::cache_url() . '/combined-' . $key . '.js';

		if ( ! file_exists( $out_file ) ) {
			self::ensure_dir( $dir );
			$contents = '';
			foreach ( $bucket as $handle => $info ) {
				$body = self::read_local_file( $info['path'] );
				if ( '' === $body ) {
					continue;
				}
				$contents .= "/* xspeed: $handle */\n" . $body . "\n;\n";
			}
			// phpcs:ignore WordPress.WP.AlternativeFunctions.file_put_contents_file_put_contents -- WP_Filesystem unavailable on frontend.
			file_put_contents( $out_file, $contents, LOCK_EX );
		}

		foreach ( $bucket as $handle => $info ) {
			$wp_scripts->dequeue( $handle );
		}
		$combined_handle = 'xspeed-combined-js';
		wp_register_script( $combined_handle, $out_url, array(), $key, true );
		wp_enqueue_script( $combined_handle );
	}

	/**
	 * Walk WP_Styles->queue, return the handles whose src is a local file we
	 * can safely combine, grouped BY media type so each media gets its own
	 * combined file. Shape:
	 *   [ media => [ handle => [ 'url' => …, 'path' => …, 'mtime' => int, 'src' => … ] ] ].
	 * '' and 'screen' media fold into the 'all' group.
	 */
	private static function collect_local_handles( \WP_Styles $wp_styles ): array {
		$groups = array();
		foreach ( $wp_styles->queue as $handle ) {
			if ( ! isset( $wp_styles->registered[ $handle ] ) ) {
				continue;
			}
			$reg = $wp_styles->registered[ $handle ];
			$src = (string) ( $reg->src ?? '' );
			if ( '' === $src ) {
				continue;
			}
			// Leave WordPress core block styles alone. Block themes (Twenty
			// Twenty-*, and any FSE theme) load per-block CSS conditionally and
			// print/track these handles through their own separated-styles
			// pipeline, often BEFORE this pass. Pulling them into a combined
			// file fights that pipeline and leaves the page unstyled. These are
			// already tiny + conditionally loaded, so there's little to gain.
			// Matches `wp-block-*` handles and any src under wp-includes/blocks/
			// or the block-library dist dir. (FBS-83633)
			if (
				0 === strpos( $handle, 'wp-block-' )
				|| false !== strpos( $src, '/wp-includes/blocks/' )
				|| false !== strpos( $src, '/block-library/' )
			) {
				continue;
			}
			$abs = self::to_absolute_url( $src );
			$info = self::local_info( $abs );
			if ( null === $info ) {
				continue; // external or unresolvable — leave in queue.
			}
			// Bucket by media type. '' and 'screen' fold into 'all' (both mean
			// "the on-screen document"); every other media value (print,
			// max-width queries, …) gets its own group so we can emit one
			// combined file per media with the right attribute — instead of
			// dropping non-'all' sheets and starving the combinable bucket on
			// builder sites. (FBS-83653)
			$media = (string) ( $reg->args ?? 'all' );
			if ( '' === $media || 'screen' === $media ) {
				$media = 'all';
			}
			$groups[ $media ][ $handle ] = $info + array( 'src' => $src );
		}
		return $groups;
	}

	private static function collect_local_script_handles( \WP_Scripts $wp_scripts ): array {
		$out = array();
		foreach ( $wp_scripts->queue as $handle ) {
			if ( ! isset( $wp_scripts->registered[ $handle ] ) ) {
				continue;
			}
			$reg = $wp_scripts->registered[ $handle ];
			$src = (string) ( $reg->src ?? '' );
			if ( '' === $src ) {
				continue;
			}
			// Skip scripts that carry inline-after data (they expect
			// to run at their original spot).
			if ( ! empty( $reg->extra['after'] ) || ! empty( $reg->extra['before'] ) || ! empty( $reg->extra['data'] ) ) {
				continue;
			}
			// Skip async / defer-via-strategy.
			$strategy = $reg->extra['strategy'] ?? '';
			if ( 'async' === $strategy || 'defer' === $strategy ) {
				continue;
			}
			$abs = self::to_absolute_url( $src );
			$info = self::local_info( $abs );
			if ( null === $info ) {
				continue;
			}
			$out[ $handle ] = $info + array( 'src' => $src );
		}
		return $out;
	}

	/**
	 * Convert a possibly-relative `src` into an absolute URL.
	 */
	private static function to_absolute_url( string $src ): string {
		if ( '' === $src ) {
			return '';
		}
		if ( 0 === strpos( $src, '//' ) ) {
			return ( is_ssl() ? 'https:' : 'http:' ) . $src;
		}
		if ( 0 === strpos( $src, '/' ) ) {
			$home = home_url();
			$home = (string) preg_replace( '#/$#', '', $home );
			return $home . $src;
		}
		return $src;
	}

	/**
	 * Resolve an absolute URL to a local filesystem path + mtime, or
	 * return null if the URL isn't on this site / outside web root.
	 *
	 * @return array{url:string,path:string,mtime:int}|null
	 */
	public static function local_info( string $url ): ?array {
		if ( '' === $url ) {
			return null;
		}
		$home = home_url();
		if ( 0 !== strpos( $url, $home ) ) {
			return null;
		}
		// Strip query / fragment for filesystem lookup; keep them in
		// the URL we hash against.
		$clean = strtok( $url, '?' );
		if ( ! is_string( $clean ) ) {
			return null;
		}
		$path = ABSPATH . ltrim( str_replace( $home, '', $clean ), '/' );
		if ( ! file_exists( $path ) || ! is_readable( $path ) ) {
			return null;
		}
		return array(
			'url'   => $url,
			'path'  => $path,
			'mtime' => (int) filemtime( $path ),
		);
	}

	private static function cache_key( array $bucket ): string {
		$signature = array();
		foreach ( $bucket as $handle => $info ) {
			$signature[ $handle ] = array( $info['src'] ?? '', $info['mtime'] ?? 0 );
		}
		return md5( wp_json_encode( $signature ) );
	}

	private static function read_local_file( string $path ): string {
		// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- WP_Filesystem unavailable on frontend; we already validated existence + readability.
		$body = file_get_contents( $path );
		return is_string( $body ) ? $body : '';
	}

	/**
	 * Recursively inline `@import url(...)` (and `@import "...";`)
	 * statements. Cycles detected via depth limit; cross-origin imports
	 * are left alone.
	 */
	public static function resolve_imports( string $css, string $base_url, int $depth ): string {
		if ( $depth > self::MAX_IMPORT_DEPTH ) {
			return $css;
		}
		return (string) preg_replace_callback(
			'#@import\s+(?:url\s*\(\s*)?["\']?([^"\')]+)["\']?\s*\)?\s*([^;]*);#i',
			static function ( $m ) use ( $base_url, $depth ) {
				$target = trim( (string) $m[1] );
				$media  = trim( (string) $m[2] );
				$abs    = self::resolve_relative( $target, $base_url );
				$info   = self::local_info( $abs );
				if ( null === $info ) {
					return $m[0]; // external or unresolvable; leave as-is.
				}
				$body = self::read_local_file( $info['path'] );
				if ( '' === $body ) {
					return $m[0];
				}
				$body = self::rewrite_url_paths( $body, $info['url'] );
				$body = self::resolve_imports( $body, $info['url'], $depth + 1 );
				if ( '' !== $media ) {
					return '@media ' . $media . " {\n" . $body . "\n}\n";
				}
				return $body;
			},
			$css
		);
	}

	/**
	 * Rewrite every `url(...)` whose argument is a relative path so it
	 * becomes absolute (resolved against the source file's URL). The
	 * combined file lives at a different location, so relative paths
	 * would otherwise break.
	 *
	 * Skips: absolute URLs (http://, https://, //), data: URIs,
	 * `#fragment-only`, blob:, javascript: (which shouldn't appear in
	 * CSS but won't crash).
	 */
	public static function rewrite_url_paths( string $css, string $base_url ): string {
		return (string) preg_replace_callback(
			'#url\(\s*(["\']?)([^"\')]+)\1\s*\)#i',
			static function ( $m ) use ( $base_url ) {
				$quote = $m[1];
				$raw   = trim( (string) $m[2] );
				if ( '' === $raw ) {
					return $m[0];
				}
				if (
					0 === strpos( $raw, 'data:' )
					|| 0 === strpos( $raw, 'blob:' )
					|| 0 === strpos( $raw, '#' )
					|| 0 === strpos( $raw, 'http://' )
					|| 0 === strpos( $raw, 'https://' )
					|| 0 === strpos( $raw, '//' )
				) {
					return $m[0];
				}
				$abs = self::resolve_relative( $raw, $base_url );
				return 'url(' . $quote . $abs . $quote . ')';
			},
			$css
		);
	}

	/**
	 * Resolve a relative URL (no scheme, no leading /) against a base
	 * URL. Public so tests can exercise it directly.
	 */
	public static function resolve_relative( string $target, string $base_url ): string {
		// Order matters — '//' is a prefix of '/' so the protocol-relative
		// check must happen BEFORE the leading-slash anchor.
		if ( 0 === strpos( $target, '//' ) ) {
			return ( is_ssl() ? 'https:' : 'http:' ) . $target;
		}
		if ( 0 === strpos( $target, '/' ) ) {
			$parts = wp_parse_url( $base_url );
			if ( ! is_array( $parts ) ) {
				return $target;
			}
			$origin = ( $parts['scheme'] ?? 'http' ) . '://' . ( $parts['host'] ?? '' );
			if ( isset( $parts['port'] ) ) {
				$origin .= ':' . $parts['port'];
			}
			return $origin . $target;
		}
		if ( 0 === strpos( $target, 'http://' ) || 0 === strpos( $target, 'https://' ) ) {
			return $target;
		}
		// Relative. Strip filename from base, resolve.
		$base_path = (string) wp_parse_url( $base_url, PHP_URL_PATH );
		$base_dir  = rtrim( str_replace( basename( $base_path ), '', $base_path ), '/' );
		$parts     = wp_parse_url( $base_url );
		$origin    = ( $parts['scheme'] ?? 'http' ) . '://' . ( $parts['host'] ?? '' );
		if ( isset( $parts['port'] ) ) {
			$origin .= ':' . $parts['port'];
		}
		// Collapse ../
		$joined  = $base_dir . '/' . $target;
		$segments = array();
		foreach ( explode( '/', $joined ) as $seg ) {
			if ( '' === $seg || '.' === $seg ) {
				continue;
			}
			if ( '..' === $seg ) {
				array_pop( $segments );
				continue;
			}
			$segments[] = $seg;
		}
		return $origin . '/' . implode( '/', $segments );
	}

	private static function ensure_dir( string $dir ): void {
		if ( ! is_dir( $dir ) ) {
			wp_mkdir_p( $dir );
		}
		$silence = $dir . '/index.php';
		if ( ! file_exists( $silence ) ) {
			// phpcs:ignore WordPress.WP.AlternativeFunctions.file_put_contents_file_put_contents -- bootstrap-time helper, WP_Filesystem unavailable.
			file_put_contents( $silence, "<?php\n// Silence is golden.\n" );
		}
	}
}