| 1 |
<?php |
| 2 |
/** |
| 3 |
* Status and information regarding the site visitor. |
| 4 |
* |
| 5 |
* @package automattic/jetpack-status |
| 6 |
*/ |
| 7 |
|
| 8 |
namespace Automattic\Jetpack\Status; |
| 9 |
|
| 10 |
use Automattic\Jetpack\IP\Utils as IP_Utils; |
| 11 |
|
| 12 |
/** |
| 13 |
* Visitor class. |
| 14 |
*/ |
| 15 |
class Visitor { |
| 16 |
|
| 17 |
/** |
| 18 |
* Gets current user IP address. |
| 19 |
* |
| 20 |
* Only a value that parses as an IP address is returned. With `$check_all_headers`, the |
| 21 |
* forwarded headers are tried in order, a comma-separated list yields its first valid entry, |
| 22 |
* and a header holding no valid address is skipped. |
| 23 |
* |
| 24 |
* A site with a trusted header configured does not use that sweep at all. Brute force |
| 25 |
* protection stores which header carries the visitor address, how far back to count in its |
| 26 |
* list, and whether that list runs in reverse, all worked out for this site's own proxy |
| 27 |
* setup. Once that answer exists it is the answer, including when the request did not carry |
| 28 |
* the header — `IP\Utils::get_ip()` falls back to `REMOTE_ADDR` itself in that case. Reading |
| 29 |
* the sweep afterwards would let a request that simply omits the trusted header pick its own |
| 30 |
* address out of a header it fully controls, and would hand Jetpack two answers for one |
| 31 |
* request, since brute force protection resolves the same visitor through `IP\Utils`. |
| 32 |
* |
| 33 |
* That is a guarantee about which source is consulted, not about the address itself: it |
| 34 |
* still assumes the proxy named in the stored configuration rewrites the header. A request |
| 35 |
* reaching the origin directly can present that header itself. |
| 36 |
* |
| 37 |
* The address is normalized by `IP\Utils::clean_ip()`: it is lowercased, anything following an |
| 38 |
* " unless " separator is dropped, and a port suffix, IPv6 brackets, or an `::ffff:` IPv4 |
| 39 |
* mapping are reduced to the bare address. Code comparing this value against a stored or |
| 40 |
* configured address should normalize that address the same way. |
| 41 |
* |
| 42 |
* @param bool $check_all_headers Check all headers? Default is `false`. |
| 43 |
* |
| 44 |
* @return string Current user IP address, or an empty string if no valid address could be determined. |
| 45 |
*/ |
| 46 |
public function get_ip( $check_all_headers = false ) { |
| 47 |
if ( $check_all_headers ) { |
| 48 |
$trusted_header_data = get_site_option( 'trusted_ip_header' ); |
| 49 |
if ( isset( $trusted_header_data->trusted_header ) ) { |
| 50 |
$trusted_ip = IP_Utils::get_ip(); |
| 51 |
return false !== $trusted_ip ? $trusted_ip : ''; |
| 52 |
} |
| 53 |
|
| 54 |
foreach ( array( |
| 55 |
'HTTP_CF_CONNECTING_IP', |
| 56 |
'HTTP_CLIENT_IP', |
| 57 |
'HTTP_X_FORWARDED_FOR', |
| 58 |
'HTTP_X_FORWARDED', |
| 59 |
'HTTP_X_CLUSTER_CLIENT_IP', |
| 60 |
'HTTP_FORWARDED_FOR', |
| 61 |
'HTTP_FORWARDED', |
| 62 |
'HTTP_VIA', |
| 63 |
) as $key ) { |
| 64 |
if ( empty( $_SERVER[ $key ] ) ) { |
| 65 |
continue; |
| 66 |
} |
| 67 |
// Proxies append to the list, so the leftmost entry is the client. |
| 68 |
foreach ( explode( ',', (string) wp_unslash( $_SERVER[ $key ] ) ) as $candidate ) { // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- Each entry is validated by clean_ip() below. |
| 69 |
$ip = IP_Utils::clean_ip( $candidate ); |
| 70 |
if ( false !== $ip ) { |
| 71 |
return $ip; |
| 72 |
} |
| 73 |
} |
| 74 |
} |
| 75 |
} |
| 76 |
|
| 77 |
$ip = empty( $_SERVER['REMOTE_ADDR'] ) ? false : IP_Utils::clean_ip( wp_unslash( $_SERVER['REMOTE_ADDR'] ) ); // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- clean_ip() validates it. |
| 78 |
return false !== $ip ? $ip : ''; |
| 79 |
} |
| 80 |
|
| 81 |
/** |
| 82 |
* Simple gate check for a11n feature testing purposes using AT_PROXIED_REQUEST constant. |
| 83 |
* IMPORTANT: Only use it for internal feature test purposes, not authorization. |
| 84 |
* |
| 85 |
* The goal of this function is to help us gate features by using a similar function name |
| 86 |
* we find on simple sites: is_automattician(). |
| 87 |
* |
| 88 |
* @return bool True if the current request is PROXIED, false otherwise. |
| 89 |
*/ |
| 90 |
public function is_automattician_feature_flags_only() { |
| 91 |
return ( defined( 'AT_PROXIED_REQUEST' ) && AT_PROXIED_REQUEST ); |
| 92 |
} |
| 93 |
|
| 94 |
/** |
| 95 |
* Whether the current request should be attributed to an Automattician in analytics. |
| 96 |
* |
| 97 |
* True for an identified Automattician on WordPress.com Simple, and for A8C-proxied |
| 98 |
* requests on both Simple and WoA. Use it to tag Tracks events as internal traffic so |
| 99 |
* it can be filtered out of product reporting — this matters most for newly launched |
| 100 |
* features, where a small amount of internal testing is a large share of the totals |
| 101 |
* and there is no way to separate it after the fact. |
| 102 |
* |
| 103 |
* IMPORTANT: Reporting signal only, never authorization. A proxied request says |
| 104 |
* something about where the request came from, not who the user is. |
| 105 |
* |
| 106 |
* @since 6.4.0 |
| 107 |
* |
| 108 |
* @return bool True if the request looks like Automattician traffic, false otherwise. |
| 109 |
*/ |
| 110 |
public function is_tracking_automattician() { |
| 111 |
// Identified Automattician on WordPress.com Simple. |
| 112 |
if ( function_exists( 'is_automattician' ) && \is_automattician() ) { |
| 113 |
return true; |
| 114 |
} |
| 115 |
|
| 116 |
// Proxied A8C request on WordPress.com Simple. |
| 117 |
if ( function_exists( 'wpcom_is_proxied_request' ) && \wpcom_is_proxied_request() ) { |
| 118 |
return true; |
| 119 |
} |
| 120 |
|
| 121 |
// Proxied A8C request on WoA. |
| 122 |
return $this->is_automattician_feature_flags_only(); |
| 123 |
} |
| 124 |
} |
| 125 |
|