<?php
/**
 * Stats Tracking_Pixel
 *
 * @package automattic/jetpack-stats
 */

namespace Automattic\Jetpack\Stats;

use Jetpack_Options;
use WP_Post;

/**
 * Stats Tracking_Pixel class.
 *
 * Responsible for embedding the Stats tracking pixel.
 *
 * @since 0.1.0
 */
class Tracking_Pixel {

	/**
	 * Array name.
	 *
	 * @var string $array_name The 'stats' array name
	 */
	const STATS_ARRAY_TO_STRING_FILTER = 'stats_array';

	const TRACKED_UTM_PARAMETERS = array(
		'utm_id',
		'utm_source',
		'utm_medium',
		'utm_campaign',
		'utm_term',
		'utm_content',
		'utm_source_platform',
		'utm_creative_format',
		'utm_marketing_tactic',
	);

	/**
	 * Stats Build View Data.
	 *
	 * @access public
	 * @return array
	 */
	public static function build_view_data() {
		global $wp_the_query;

		$blog        = Jetpack_Options::get_option( 'id' );
		$tz          = get_option( 'gmt_offset' );
		$v           = 'ext';
		$blog_url    = wp_parse_url( site_url() );
		$srv         = $blog_url['host'];
		$is_not_post = false;
		if ( $wp_the_query->is_single || $wp_the_query->is_page || $wp_the_query->is_posts_page ) {
			// Store and reset the queried_object and queried_object_id
			// Otherwise, redirect_canonical() will redirect to home_url( '/' ) for show_on_front = page sites where home_url() is not all lowercase.
			// Repro:
			// 1. Set home_url = https://ExamPle.com/
			// 2. Set show_on_front = page
			// 3. Set page_on_front = something
			// 4. Visit https://example.com/ !
			$queried_object    = $wp_the_query->queried_object ?? null;
			$queried_object_id = $wp_the_query->queried_object_id ?? null;
			try {
				$post_obj = $wp_the_query->get_queried_object();
				$post     = $post_obj instanceof WP_Post ? $post_obj->ID : '0';
			} finally {
				$wp_the_query->queried_object    = $queried_object;
				$wp_the_query->queried_object_id = $queried_object_id;
			}
		} else {
			$post        = '0';
			$is_not_post = true;
		}
		$view_data = compact( 'v', 'blog', 'post', 'tz', 'srv' );
		// Batcache removes some of the UTM params from $_GET, we need to extract them from uri directly instead.
		// phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- We're sanitizing individual params in the loop.
		$url_query = wp_parse_url( wp_unslash( $_SERVER['REQUEST_URI'] ?? '' ), PHP_URL_QUERY );
		parse_str( (string) $url_query, $url_params );
		foreach ( self::TRACKED_UTM_PARAMETERS as $utm_parameter ) {
			if ( isset( $url_params[ $utm_parameter ] ) && is_scalar( $url_params[ $utm_parameter ] ) ) {
				$view_data[ $utm_parameter ] = substr( sanitize_textarea_field( wp_unslash( $url_params[ $utm_parameter ] ) ), 0, 255 );
			}
		}

		if ( $is_not_post ) {
			if ( $wp_the_query->is_home() ) {
				$view_data['arch_home'] = '1';
			} elseif ( $wp_the_query->is_search() ) {
				$search_term               = $wp_the_query->query['s'] ?? $wp_the_query->query_vars['s'] ?? '';
				$view_data['arch_search']  = sanitize_text_field( $search_term );
				$view_data['arch_filters'] = sanitize_text_field( self::build_search_filters( $wp_the_query ) );
				$view_data['arch_results'] = $wp_the_query->posts ? $wp_the_query->post_count : 0;
			} elseif ( $wp_the_query->is_archive() ) {
				if ( $wp_the_query->is_date ) {
					$query                  = $wp_the_query->query;
					$date_parts             = array_filter( array( $query['year'] ?? null, $query['monthnum'] ?? null, $query['day'] ?? null ) );
					$date                   = implode( '/', $date_parts );
					$view_data['arch_date'] = $date;
				}
				if ( $wp_the_query->is_category ) {
					$view_data['arch_cat'] = $wp_the_query->query['category_name'] ?? $wp_the_query->query_vars['category_name'] ?? '';
				}
				if ( $wp_the_query->is_tag ) {
					$view_data['arch_tag'] = $wp_the_query->query['tag'] ?? $wp_the_query->query_vars['tag'] ?? '';
				}
				if ( $wp_the_query->is_author ) {
					$view_data['arch_author'] = $wp_the_query->query['author_name'] ?? '';
				}
				if ( $wp_the_query->is_tax ) {
					$query = $wp_the_query->query;
					if ( is_array( $query ) && count( $query ) === 1 ) {
						$view_data[ 'arch_tax_' . array_keys( $query )[0] ] = array_values( $query )[0];
					}
				}
				$view_data['arch_results'] = $wp_the_query->posts ? $wp_the_query->post_count : 0;
			} elseif ( $wp_the_query->is_404() ) {
				$view_data['arch_err'] = sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ?? '' ) );
			} else {
				$view_data['arch_other'] = sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ?? '' ) );
			}
		}
		return $view_data;
	}

	/**
	 * Collect the tracking data for a search page.
	 *
	 * @access private
	 * @param  \WP_Query $query The WP_Query object to parse all the filters from.
	 * @return string The search filters in a URL query string format.
	 */
	private static function build_search_filters( $query ) {
		$data = array(
			'posts_per_page' => $query->get( 'posts_per_page' ),
			'paged'          => ( $query->get( 'paged' ) ) ? absint( $query->get( 'paged' ) ) : 1,
			'orderby'        => $query->get( 'orderby' ),
			'order'          => $query->get( 'order' ),
		);

		if ( $query->get( 'author_name' ) ) {
			$data['author_name'] = $query->get( 'author_name' );
		}
		$filters = http_build_query( $data );

		$the_tax_query = $query->tax_query;
		$terms         = array();
		if ( ! empty( $the_tax_query->queried_terms ) && is_array( $the_tax_query->queried_terms ) ) {
			foreach ( $the_tax_query->queries as $tax_query ) {
				if ( ! is_array( $tax_query ) || ! isset( $tax_query['taxonomy'] ) ) {
					continue;
				}
				$taxonomy = $tax_query['taxonomy'];
				if ( ! isset( $terms[ $taxonomy ] ) || ! is_array( $terms[ $taxonomy ] ) ) {
					$terms[ $taxonomy ] = array();
				}
				$terms[ $taxonomy ] = array_merge( $terms[ $taxonomy ], $tax_query['terms'] );
			}
		}
		if ( ! empty( $terms ) ) {
			$filters .= '&terms=' . wp_json_encode( $terms, JSON_UNESCAPED_SLASHES | JSON_HEX_TAG | JSON_HEX_AMP );
		}
		return $filters;
	}

	/**
	 * Build the Stats tracking details.
	 *
	 * @since 0.6.0
	 *
	 * @access private
	 * @param array $data Array of options about the site and page for the inline (non-AMP) tracker.
	 * @return string
	 */
	private static function build_stats_details( $data ) {
		$data_stats_array = self::stats_array_to_string( $data );

		$pushes = sprintf(
			'_stq.push([ "view", %1$s ]);
_stq.push([ "clickTrackerInit", "%2$s", "%3$s" ]);',
			$data_stats_array,
			$data['blog'],
			$data['post']
		);

		// OFF (default): byte-for-byte identical to the historical output.
		if ( ! Options::get_option( 'honor_cookie_consent' ) ) {
			return "_stq = window._stq || [];\n" . $pushes;
		}

		// Fail closed when the WP Consent API plugin is active (an unavailable client-side API
		// means "wait", not "fire"); fail open otherwise to preserve historical tracking.
		return self::build_consent_gate( $pushes, ! function_exists( 'wp_has_consent' ) );
	}

	/**
	 * Wrap the tracking pushes in a WP Consent API gate.
	 *
	 * The check runs in the browser because cached HTML is shared across visitors, deferred to
	 * DOMContentLoaded (and re-run on the `wp_consent_type_defined` readiness event) so a
	 * late-loading consent plugin is still honored. The check is idempotent.
	 *
	 * `_jpStatsFire.done` is set before the pushes, not after, so the gate is at-most-once even
	 * if a push throws. Retrying can't recover: the stats sender assigns the beacon `src` before
	 * any of its fallible DOM work, so a later exception means the view was already counted and
	 * a replay would double-count it.
	 *
	 * @access private
	 * @param string $pushes    The `_stq.push(...)` statements to gate.
	 * @param bool   $fail_open Whether to fire when the client-side WP Consent API is unavailable.
	 * @return string
	 */
	private static function build_consent_gate( $pushes, $fail_open ) {
		$fail_open_literal = $fail_open ? 'true' : 'false';

		return sprintf(
			'_stq = window._stq || [];
function _jpStatsFire() {
	if ( _jpStatsFire.done ) { return; }
	_jpStatsFire.done = true;
	%1$s
}
function _jpStatsCheck() {
	if ( typeof window.wp_has_consent === "function" ) {
		var consented;
		try {
			consented = window.wp_has_consent( "statistics" );
		} catch ( e ) {
			consented = %2$s;
		}
		if ( consented ) { _jpStatsFire(); }
		return;
	}
	if ( %2$s ) { _jpStatsFire(); }
}
document.addEventListener( "wp_listen_for_consent_change", function ( event ) {
	if ( event && event.detail && event.detail.statistics === "allow" ) { _jpStatsFire(); }
} );
document.addEventListener( "wp_consent_type_defined", _jpStatsCheck );
window.addEventListener( "wp_consent_type_defined", _jpStatsCheck );
if ( document.readyState === "loading" ) {
	document.addEventListener( "DOMContentLoaded", _jpStatsCheck, { once: true } );
} else {
	_jpStatsCheck();
}',
			$pushes,
			$fail_open_literal
		);
	}

	/**
	 * Add fetchpriority="low" to the Stats script attributes.
	 *
	 * Reduces network contention with resources in the critical rendering path (e.g., the LCP
	 * element image). This benefits Safari and Firefox, which don't automatically assign low
	 * priority to async/defer scripts (unlike Chrome).
	 *
	 * @since 0.19.5
	 *
	 * @param array $attributes Script tag attributes.
	 * @return array Modified attributes.
	 */
	public static function add_low_fetchpriority( $attributes ) {
		// WordPress derives the tag id from the enqueue handle as "{handle}-js", so the
		// 'jetpack-stats' script (registered in enqueue_stats_script()) prints as
		// 'jetpack-stats-js'. Keep this in sync if the handle is ever renamed.
		if ( isset( $attributes['id'] ) && 'jetpack-stats-js' === $attributes['id'] ) {
			$attributes['fetchpriority'] = 'low';
		}
		return $attributes;
	}

	/**
	 * Remove the dns-prefetch resource hint for stats.wp.com.
	 *
	 * WordPress automatically adds dns-prefetch hints for enqueued script hosts via
	 * wp_dependencies_unique_hosts(). Since we're deprioritizing the stats script,
	 * the dns-prefetch is counterproductive — it front-loads DNS resolution for a
	 * resource we're intentionally delaying.
	 *
	 * @since 0.19.5
	 *
	 * @param array  $urls          Array of resource hint URLs.
	 * @param string $relation_type The relation type (dns-prefetch, preconnect, etc.).
	 * @return array Filtered URLs.
	 */
	public static function remove_stats_dns_prefetch( $urls, $relation_type ) {
		if ( 'dns-prefetch' !== $relation_type ) {
			return $urls;
		}

		return array_filter(
			$urls,
			static function ( $url ) {
				// Resource hints can be arrays that carry the URL under an 'href' key.
				if ( is_array( $url ) ) {
					$candidate = ( isset( $url['href'] ) && is_string( $url['href'] ) ) ? $url['href'] : '';
				} elseif ( is_string( $url ) ) {
					$candidate = $url;
				} else {
					return true; // Unknown entry shape; leave it untouched.
				}

				// dns-prefetch entries arrive in several shapes: WordPress core emits bare
				// hosts ('stats.wp.com') via wp_dependencies_unique_hosts(), while other
				// filters may add scheme-relative ('//stats.wp.com') or full URLs. Normalize
				// each to a host so we drop stats.wp.com exactly without removing look-alike
				// hosts such as 'mystats.wp.com' or 'stats.wp.com.evil.tld'.
				if ( str_starts_with( $candidate, '//' ) ) {
					$host = wp_parse_url( 'https:' . $candidate, PHP_URL_HOST );
				} elseif ( str_contains( $candidate, '://' ) ) {
					$host = wp_parse_url( $candidate, PHP_URL_HOST );
				} else {
					$host = $candidate; // Bare host form, e.g. 'stats.wp.com'.
				}

				return ! is_string( $host ) || 'stats.wp.com' !== strtolower( $host );
			}
		);
	}

	/**
	 * Enqueue the Stats pixel.
	 * Do not use this function directly, it is hooked into `wp_enqueue_scripts`.
	 *
	 * @access public
	 * @return void
	 */
	public static function enqueue_stats_script() {
		if ( self::is_amp_request() ) {
			return;
		}

		wp_enqueue_script(
			'jetpack-stats',
			'https://stats.wp.com/e-' . gmdate( 'YW' ) . '.js',
			array(),
			null, // phpcs:ignore WordPress.WP.EnqueuedResourceParameters.MissingVersion -- The version is set in the URL.
			array(
				'in_footer' => true,
				'strategy'  => 'defer',
			)
		);
		add_filter( 'wp_script_attributes', array( static::class, 'add_low_fetchpriority' ) );
		add_filter( 'wp_resource_hints', array( static::class, 'remove_stats_dns_prefetch' ), 100, 2 );

		$data = self::build_view_data();

		/**
		 * Filter the parameters added to the JavaScript stats tracking code.
		 *
		 * @module stats
		 *
		 * @since-jetpack 10.9
		 *
		 * @param array $data Array of options about the site and page you're on.
		 */
		$data = (array) apply_filters( 'jetpack_stats_footer_js_data', $data );

		$triggers = self::build_stats_details( $data );
		wp_add_inline_script(
			'jetpack-stats',
			$triggers,
			'before'
		);
	}

	/**
	 * Gets the tracking pixel URL for AMP output.
	 *
	 * @access private
	 * @param array $data Array of data for the AMP pixel tracker.
	 * @return string Returns the URL for the Stats tracker in an AMP scenario.
	 */
	private static function get_amp_pixel_url( $data ) {
		/**
		 * Filter the parameters added to the AMP pixel tracking code.
		 *
		 * @module stats
		 *
		 * @since-jetpack 10.9
		 *
		 * @param array $data Array of options about the site and page you're on.
		 */
		$data = (array) apply_filters( 'jetpack_stats_footer_amp_data', $data );

		$data['host'] = isset( $_SERVER['HTTP_HOST'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_HOST'] ) ) : ''; // input var ok.
		$data['rand'] = 'RANDOM'; // AMP placeholder.
		$data['ref']  = 'DOCUMENT_REFERRER'; // AMP placeholder.
		$data         = array_map( 'rawurlencode', $data );
		return add_query_arg( $data, 'https://pixel.wp.com/g.gif' );
	}

	/**
	 * Build an AMP pixel.
	 * Do not use this function directly, it is hooked into `wp_footer`.
	 *
	 * @access public
	 * @return void
	 */
	public static function add_amp_pixel() {
		$data = self::build_view_data();
		if ( ! self::is_amp_request() ) {
			return;
		}

		printf( '<amp-pixel src="%s"></amp-pixel>', esc_url( self::get_amp_pixel_url( $data ) ) );
	}

	/**
	 * Stats Footer.
	 *
	 * @deprecated 0.6.0
	 *
	 * @access public
	 * @return void
	 */
	public static function add_to_footer() {
		_deprecated_function( __METHOD__, '0.6.0' );
	}

	/**
	 * Gets the footer to add for the Stats tracker.
	 *
	 * @deprecated 0.6.0
	 *
	 * @access public
	 * @param array $data Array of data for the JS stats tracker.
	 * @return void
	 */
	public static function get_footer_to_add( $data ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
		_deprecated_function( __METHOD__, '0.6.0' );
	}

	/**
	 * Render the stats footer. Kept for backward compatibility on legacy AMF views.
	 *
	 * @deprecated 0.6.0
	 *
	 * @access public
	 * @param array $data Array of data for the JS stats tracker.
	 */
	public static function render_footer( $data ) { // phpcs:ignore VariableAnalysis.CodeAnalysis.VariableAnalysis.UnusedVariable
		_deprecated_function( __METHOD__, '0.6.0' );
	}

	/**
	 * Render the stats footer for AMP output. Kept for backward compatibility.
	 *
	 * @access public
	 * @param array $data Array of data for the AMP pixel tracker.
	 */
	public static function render_amp_footer( $data ) {
		printf( '<amp-pixel src="%s"></amp-pixel>', esc_url( self::get_amp_pixel_url( $data ) ) );
	}

	/**
	 * Creates the "array" string used as part of the JS tracker.
	 *
	 * @access private
	 * @param array $kvs Array of options about the site and page you're on.
	 * @return string
	 */
	private static function stats_array_to_string( $kvs ) {
		/**
		 * Filters the options added to the JavaScript Stats tracking code.
		 *
		 * @since-jetpack 1.1.0
		 *
		 * @param array $kvs Array of options about the site and page you're on.
		 */
		$kvs = (array) apply_filters( self::STATS_ARRAY_TO_STRING_FILTER, $kvs );
		$kvs = array_map( 'strval', $kvs );

		// Encode into JSON object for direct use in JS.
		return wp_json_encode( $kvs, JSON_UNESCAPED_SLASHES | JSON_HEX_TAG | JSON_HEX_AMP );
	}

	/**
	 * Does the page return AMP content.
	 *
	 * @return bool $is_amp_request Are we on AMP view.
	 */
	private static function is_amp_request() {
		$is_amp_request = ( function_exists( 'amp_is_request' ) && amp_is_request() );
		$is_amp_request = $is_amp_request || ( function_exists( 'ampforwp_is_amp_endpoint' ) && ampforwp_is_amp_endpoint() );

		/**
		 * Returns true if the current request should return valid AMP content.
		 *
		 * @since 6.2.0
		 *
		 * @param boolean $is_amp_request Is this request supposed to return valid AMP content?
		 */
		return apply_filters( 'jetpack_is_amp_request', $is_amp_request );
	}
}