| 1 |
<?php |
| 2 |
|
| 3 |
namespace Leadpages; |
| 4 |
|
| 5 |
defined('ABSPATH') || die('No script kiddies please!'); // Avoid direct file request |
| 6 |
|
| 7 |
use Leadpages\providers\Utils; |
| 8 |
use Leadpages\providers\http\Client; |
| 9 |
use Leadpages\providers\http\exceptions\ServerException; |
| 10 |
use Leadpages\providers\http\exceptions\NotFoundException; |
| 11 |
use Leadpages\models\Page; |
| 12 |
use Leadpages\models\Options; |
| 13 |
use Leadpages\PopupScope; |
| 14 |
use Leadpages\serving\PageSource; |
| 15 |
use Leadpages\serving\ClassicPageSource; |
| 16 |
use Leadpages\serving\NovaPageSource; |
| 17 |
|
| 18 |
/** |
| 19 |
* A class for serving Leadpages assets within the WordPress environment. |
| 20 |
* |
| 21 |
* The reverse-proxy mechanics (fetch, render, cache) are shared; the backend-specific behavior for |
| 22 |
* a connected page comes from a PageSource selected by the row's platform (Classic or Nova). |
| 23 |
*/ |
| 24 |
class Proxy { |
| 25 |
|
| 26 |
use Utils; |
| 27 |
|
| 28 |
/** @var Client */ |
| 29 |
private $client; |
| 30 |
|
| 31 |
/** The serving-tag meta value injected into the current response (set per request in render). */ |
| 32 |
private static $serving_tag = 'wordpress-official'; |
| 33 |
|
| 34 |
/** The pop-up embed <script> injected before </body> for the current response, or '' for none. */ |
| 35 |
private static $popup_embed = ''; |
| 36 |
|
| 37 |
public function __construct() { |
| 38 |
$this->client = new Client(); |
| 39 |
} |
| 40 |
|
| 41 |
/** |
| 42 |
* Proxy requests to connected Leadpages landing pages. Ignores any request methods that are not |
| 43 |
* GET and any paths not associated with a connected landing page. The page is served through the |
| 44 |
* PageSource for its platform. |
| 45 |
* |
| 46 |
* "path" and "slug" are used synonymously here and are the same as a WP "permalink" |
| 47 |
* |
| 48 |
* Renders the page html and exits the current process so no other code can execute. |
| 49 |
* |
| 50 |
* @return void |
| 51 |
*/ |
| 52 |
public function serve_landing_page() { |
| 53 |
$request_method = sanitize_key($_SERVER['REQUEST_METHOD']); |
| 54 |
if ('get' !== $request_method) { |
| 55 |
return; |
| 56 |
} |
| 57 |
|
| 58 |
$start = microtime(true); |
| 59 |
|
| 60 |
$current_url = $this->get_current_url(); |
| 61 |
$slug = sanitize_title($this->parse_request($current_url)); |
| 62 |
|
| 63 |
$page = Page::get_by_slug($slug); |
| 64 |
if (! $page) { |
| 65 |
$this->debug("Ignoring request to $current_url"); |
| 66 |
return; |
| 67 |
} |
| 68 |
|
| 69 |
$source = $this->source_for($page); |
| 70 |
$cache_key = $source->cache_key($page); |
| 71 |
|
| 72 |
// The proxy exits before wp_enqueue_scripts runs, so the site-wide pop-up enqueue never fires |
| 73 |
// for proxied pages. Inject the same embed here, gated by the same scope (matched on this |
| 74 |
// request's slug) so a "specific pages" selection covers proxied pages too. |
| 75 |
self::$popup_embed = $this->build_popup_embed($slug); |
| 76 |
|
| 77 |
$cached_value = Cache::get($cache_key); |
| 78 |
if ($cached_value) { |
| 79 |
$this->debug('Serving page from cache'); |
| 80 |
$this->render_html($cached_value, $source); |
| 81 |
} else { |
| 82 |
// Do not serve unpublished, deleted, or split-test-variation pages. |
| 83 |
if (! $page->current_edition || ! $page->connected || $page->deleted_at || $page->split_test) { |
| 84 |
$this->debug("Ignoring request to $current_url"); |
| 85 |
return; |
| 86 |
} |
| 87 |
|
| 88 |
$wp_url = strtok($current_url, '?'); |
| 89 |
$this->debug("Proxying $current_url"); |
| 90 |
$response = $this->fetch_page_html($source, $page, $wp_url); |
| 91 |
if (! $response) { |
| 92 |
$this->debug('Something went wrong, aborting proxy'); |
| 93 |
return; |
| 94 |
} |
| 95 |
|
| 96 |
$this->render_html($response, $source); |
| 97 |
|
| 98 |
// Never cache a response that carries Set-Cookie: the cached copy would replay that |
| 99 |
// same cookie value to every visitor served from cache (a shared-identity vector). |
| 100 |
// Per-visitor experiment/personalization responses are already no-store (ttl null); |
| 101 |
// this also covers a cacheable page that happens to set a first-party cookie. |
| 102 |
$ttl = $source->cache_ttl($page, $response); |
| 103 |
if (null !== $ttl && empty(wp_remote_retrieve_cookies($response))) { |
| 104 |
Cache::set($cache_key, $response, $ttl); |
| 105 |
} |
| 106 |
} |
| 107 |
|
| 108 |
$end = microtime(true); |
| 109 |
$time_taken = ( $end - $start ) * 1000; |
| 110 |
$this->debug("Successful served page in $time_taken ms"); |
| 111 |
|
| 112 |
$this->lp_exit(0); |
| 113 |
} |
| 114 |
|
| 115 |
/** |
| 116 |
* Select the serving strategy for a page based on its platform. |
| 117 |
* |
| 118 |
* @param object $page |
| 119 |
* @return PageSource |
| 120 |
*/ |
| 121 |
private function source_for( $page ) { |
| 122 |
if (isset($page->platform) && 'nova' === $page->platform) { |
| 123 |
return new NovaPageSource(); |
| 124 |
} |
| 125 |
return new ClassicPageSource(); |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Strip the base url and params out of the request url and differentiate the |
| 130 |
* result against the users specified permalink structure. The result will be |
| 131 |
* the slug we can expect a landing page to be published under. |
| 132 |
* |
| 133 |
* @param string $url |
| 134 |
* @return string |
| 135 |
*/ |
| 136 |
private function parse_request( $url ) { |
| 137 |
$path_and_params = substr($url, strlen(home_url())); |
| 138 |
$path = explode('?', $path_and_params); |
| 139 |
$tokens = explode('/', $path[0]); |
| 140 |
|
| 141 |
$permalink_structure = $this->clean_permalink_for_leadpage(); |
| 142 |
$tokens = array_diff($tokens, $permalink_structure); |
| 143 |
|
| 144 |
foreach ($tokens as $index => $token) { |
| 145 |
if (empty($token)) { |
| 146 |
unset($tokens[ $index ]); |
| 147 |
} else { |
| 148 |
$tokens[ $index ] = sanitize_title($token); |
| 149 |
} |
| 150 |
} |
| 151 |
$tokens = array_values($tokens); |
| 152 |
$slug = implode('/', $tokens); |
| 153 |
|
| 154 |
return $slug; |
| 155 |
} |
| 156 |
|
| 157 |
|
| 158 |
/** |
| 159 |
* Get the WordPress permalink structure with any %parameters% removed |
| 160 |
* |
| 161 |
* @return string[] |
| 162 |
*/ |
| 163 |
private function clean_permalink_for_leadpage() { |
| 164 |
$permalink_structure = explode('/', Options::get(Options::$permalink_structure)); |
| 165 |
foreach ($permalink_structure as $key => $value) { |
| 166 |
if (empty($value) || strpos($value, '%') !== false) { |
| 167 |
unset($permalink_structure[ $key ]); |
| 168 |
} |
| 169 |
} |
| 170 |
return $permalink_structure; |
| 171 |
} |
| 172 |
|
| 173 |
/** |
| 174 |
* Fetch the upstream page HTML through the given source. Retries once on server errors. Returns |
| 175 |
* the response to render, or null to decline serving (letting WordPress handle the request). |
| 176 |
* |
| 177 |
* @param PageSource $source |
| 178 |
* @param object $page |
| 179 |
* @param string $wp_url the public WordPress URL being served |
| 180 |
* @param bool $retry whether to retry the request on server errors |
| 181 |
* @return array|null response object or null |
| 182 |
*/ |
| 183 |
private function fetch_page_html( $source, $page, $wp_url, $retry = true ) { |
| 184 |
$url = $source->build_url($page, $wp_url); |
| 185 |
$options = $source->build_request_options(); |
| 186 |
|
| 187 |
try { |
| 188 |
$response = $this->client->get($url, $options); |
| 189 |
} catch (NotFoundException $e) { |
| 190 |
// Classic serves the Leadpages 404 page; Nova declines so it never echoes an error page. |
| 191 |
$response = $source->serves_not_found() ? $e->response : null; |
| 192 |
} catch (ServerException $e) { |
| 193 |
$status_code = wp_remote_retrieve_response_code($e->response); |
| 194 |
if ($status_code >= 500 && $retry) { |
| 195 |
$response = $this->fetch_page_html($source, $page, $wp_url, false); |
| 196 |
} else { |
| 197 |
$response = null; |
| 198 |
} |
| 199 |
} catch (\Exception $e) { |
| 200 |
$response = null; |
| 201 |
} |
| 202 |
|
| 203 |
return $response; |
| 204 |
} |
| 205 |
|
| 206 |
/** |
| 207 |
* Render HTML from an HTTP response object to the page with the same status code as the response |
| 208 |
* and mirror the source's cookies back to the visitor. |
| 209 |
* |
| 210 |
* @param array $response |
| 211 |
* @param PageSource $source |
| 212 |
*/ |
| 213 |
public function render_html( $response, $source ) { |
| 214 |
if (ob_get_length() > 0) { |
| 215 |
ob_clean(); |
| 216 |
} |
| 217 |
|
| 218 |
$html = $response['body']; |
| 219 |
$status = wp_remote_retrieve_response_code($response); |
| 220 |
|
| 221 |
status_header($status); |
| 222 |
foreach ($source->cookies_to_mirror($response) as $cookie) { |
| 223 |
// Mirror on the WordPress host with a site-wide path so per-visitor experiment / |
| 224 |
// personalization cookies are sent on every page (not just the proxied slug's |
| 225 |
// directory, which is what a bare setcookie() would default to). Mark Secure on |
| 226 |
// HTTPS and SameSite=Lax. HttpOnly is intentionally NOT forced: Nova's client-side |
| 227 |
// scripts read experiment cookies (e.g. _hp_exp_*) from document.cookie. |
| 228 |
setcookie( |
| 229 |
$cookie->name, |
| 230 |
$cookie->value, |
| 231 |
[ |
| 232 |
'expires' => $cookie->expires ?? 0, |
| 233 |
'path' => '/', |
| 234 |
'secure' => is_ssl(), |
| 235 |
'samesite' => 'Lax', |
| 236 |
] |
| 237 |
); |
| 238 |
} |
| 239 |
|
| 240 |
self::$serving_tag = $source->serving_tag(); |
| 241 |
|
| 242 |
ob_start([ get_called_class(), 'preprocess_html' ]); |
| 243 |
// phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 244 |
echo $html; |
| 245 |
ob_end_flush(); |
| 246 |
} |
| 247 |
|
| 248 |
/** |
| 249 |
* Build the pop-up embed <script> tag for a served slug, or '' when no pop-up is selected or the |
| 250 |
* slug is outside the configured scope. The URL is the same public embed the site-wide enqueue |
| 251 |
* uses; PopupScope is the single source of truth so proxied pages honor the same selection. |
| 252 |
* |
| 253 |
* @param string $slug the slug of the page being served |
| 254 |
* @return string |
| 255 |
*/ |
| 256 |
private function build_popup_embed( $slug ) { |
| 257 |
$popup_id = PopupScope::selected_popup_id(); |
| 258 |
if (empty($popup_id) || ! PopupScope::allows_slug($slug)) { |
| 259 |
return ''; |
| 260 |
} |
| 261 |
// The proxy injects into already-rendered HTML and exits before wp_enqueue_scripts runs, so |
| 262 |
// the embed cannot go through the enqueue pipeline and is emitted as a literal script tag. |
| 263 |
// phpcs:ignore WordPress.WP.EnqueuedResources.NonEnqueuedScript -- injected into proxied HTML. |
| 264 |
return '<script async src="' . esc_url(PopupScope::embed_src($popup_id)) . '"></script>'; |
| 265 |
} |
| 266 |
|
| 267 |
/** |
| 268 |
* Wrap the exit construct for testing purposes |
| 269 |
*/ |
| 270 |
public function lp_exit() { |
| 271 |
exit(0); |
| 272 |
} |
| 273 |
|
| 274 |
private static function get_current_url() { |
| 275 |
$host = isset($_SERVER['HTTP_HOST']) ? sanitize_text_field(wp_unslash($_SERVER['HTTP_HOST'])) : ''; |
| 276 |
$uri = isset($_SERVER['REQUEST_URI']) ? sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])) : ''; |
| 277 |
return esc_url_raw(( is_ssl() ? 'https://' : 'http://' ) . $host . $uri); |
| 278 |
} |
| 279 |
|
| 280 |
/** |
| 281 |
* Process HTML content for output buffering. |
| 282 |
* |
| 283 |
* @param string $content html |
| 284 |
* @return string |
| 285 |
*/ |
| 286 |
public static function preprocess_html( $content ) { |
| 287 |
$html = self::modify_url_tag($content); |
| 288 |
$html = self::modify_serving_tags($html); |
| 289 |
$html = self::modify_popup_embed($html); |
| 290 |
return $html; |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Output buffering callback to inject the Nova pop-up embed <script> before the closing </body> |
| 295 |
* tag. No-op when no pop-up applies to this response (none selected or slug out of scope) or when |
| 296 |
* the HTML has no </body>. |
| 297 |
* |
| 298 |
* @param string $content html |
| 299 |
* @return string |
| 300 |
*/ |
| 301 |
public static function modify_popup_embed( $content ) { |
| 302 |
if ('' === self::$popup_embed) { |
| 303 |
return $content; |
| 304 |
} |
| 305 |
return str_replace('</body>', self::$popup_embed . '</body>', $content); |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* Output buffering callback to add a "leadpages-serving-tags" meta tag for analytics. This |
| 310 |
* helps us differentiate WordPress traffic (and Classic vs Nova) in our system. |
| 311 |
* |
| 312 |
* @param string $content html |
| 313 |
* @return string |
| 314 |
*/ |
| 315 |
public static function modify_serving_tags( $content ) { |
| 316 |
$search = '</head>'; |
| 317 |
$replace = '<meta name="leadpages-serving-tags" content="' . self::$serving_tag . '"></head>'; |
| 318 |
return str_replace($search, $replace, $content); |
| 319 |
} |
| 320 |
|
| 321 |
/** |
| 322 |
* Output buffering callback for "og:url" meta tag. Replace the tag content with |
| 323 |
* the url of the WordPress page. |
| 324 |
* |
| 325 |
* Open Graph meta tags are snippets of code that control how URLs are displayed when shared |
| 326 |
* on social media. A link to this page shared on social media should point to the users WP |
| 327 |
* site and not the page within Leadpages. |
| 328 |
* |
| 329 |
* @param string $content html |
| 330 |
* @return string |
| 331 |
*/ |
| 332 |
public static function modify_url_tag( $content ) { |
| 333 |
global $wp; |
| 334 |
if (empty($wp)) { |
| 335 |
// we can't build the correct WP url in this case, so return the original page |
| 336 |
return $content; |
| 337 |
} |
| 338 |
|
| 339 |
$url = self::get_current_url(); |
| 340 |
$regex = '/(<meta property="og:url" content=")[^"]+(">)/'; |
| 341 |
$html = preg_replace($regex, '${1}' . $url . '${2}', $content); |
| 342 |
if (null === $html) { |
| 343 |
// An error occured so we return the original content |
| 344 |
return $content; |
| 345 |
} |
| 346 |
return $html; |
| 347 |
} |
| 348 |
} |
| 349 |
|
