| 1 |
<?php |
| 2 |
|
| 3 |
|
| 4 |
if (!defined('ABSPATH')) { |
| 5 |
exit; |
| 6 |
} |
| 7 |
|
| 8 |
/** Stores a message and its importance. */ |
| 9 |
class ABJ_404_Solution_UserRequest { |
| 10 |
|
| 11 |
/** @var self|null */ |
| 12 |
private static $instance = null; |
| 13 |
|
| 14 |
/** @var string|null */ |
| 15 |
private $requestURIWithoutCommentsPage = null; |
| 16 |
|
| 17 |
/** @var string */ |
| 18 |
private $requestURI = null; |
| 19 |
|
| 20 |
/** @var array<string, int|string>|null */ |
| 21 |
private $urlParts = null; |
| 22 |
|
| 23 |
/** @var string */ |
| 24 |
private $queryString = null; |
| 25 |
|
| 26 |
/** @var string */ |
| 27 |
private $commentPagePart = null; |
| 28 |
|
| 29 |
/** |
| 30 |
* Replace the singleton instance. Used by integration tests that need |
| 31 |
* a pre-populated UserRequest (or a programmable test double) so the |
| 32 |
* pipeline reads stable path / slug / query-string values without |
| 33 |
* relying on `$_SERVER['REQUEST_URI']` being set during the test run. |
| 34 |
* |
| 35 |
* @param self|null $instance |
| 36 |
* @return void |
| 37 |
*/ |
| 38 |
public static function setInstance($instance) { |
| 39 |
self::$instance = $instance; |
| 40 |
} |
| 41 |
|
| 42 |
/** @return self|null */ |
| 43 |
public static function getInstance() { |
| 44 |
if (self::$instance == null) { |
| 45 |
if (!self::initialize()) { |
| 46 |
$abj404logging = abj_service('logging'); |
| 47 |
$abj404logging->errorMessage('Issue initializing ' . __CLASS__, |
| 48 |
new Exception('Issue initializing ' . __CLASS__)); |
| 49 |
} |
| 50 |
} |
| 51 |
|
| 52 |
return self::$instance; |
| 53 |
} |
| 54 |
|
| 55 |
/** @return bool */ |
| 56 |
public static function initialize(): bool { |
| 57 |
global $wp_rewrite; |
| 58 |
|
| 59 |
$abj404logging = abj_service('logging'); |
| 60 |
$f = abj_service('functions'); |
| 61 |
$abj404logic = abj_service('plugin_logic'); |
| 62 |
$sanitizer = abj_service('sanitizer'); |
| 63 |
|
| 64 |
$urlToParse = $sanitizer->normalizeUrlString($_SERVER['REQUEST_URI'] ?? ''); |
| 65 |
|
| 66 |
// if the user somehow requested an invalid URL that's too long then fix it. |
| 67 |
if ($f->strlen($urlToParse) > ABJ404_MAX_URL_LENGTH) { |
| 68 |
$matches = null; |
| 69 |
$f->regexMatch("image (.+);base64,", $urlToParse, $matches); |
| 70 |
if ($matches != null && $f->strlen($matches[0]) > 0) { |
| 71 |
$instrPattern = $matches[0]; |
| 72 |
$truncateHere = $f->strpos($urlToParse, $instrPattern); |
| 73 |
$truncatedRequest = $f->substr($urlToParse, 0, ($truncateHere !== false ? $truncateHere : null)); |
| 74 |
$urlToParse = $truncatedRequest; |
| 75 |
} |
| 76 |
|
| 77 |
if ($f->strlen($urlToParse) > ABJ404_MAX_URL_LENGTH) { |
| 78 |
// just truncate it to something reasonable. |
| 79 |
$urlToParse = $f->substr($urlToParse, 0, ABJ404_MAX_URL_LENGTH); |
| 80 |
} |
| 81 |
} |
| 82 |
|
| 83 |
// hanlde the case where '///?gf_page=upload' is returned as the request URI. |
| 84 |
$containsHost = $f->strpos($urlToParse, "://"); |
| 85 |
|
| 86 |
if (($containsHost === false) || ($containsHost >= 7) || (!is_array(parse_url(esc_url($urlToParse))))) { |
| 87 |
// we have something like //login.php and it needs to be http://host.com/login.php |
| 88 |
while ($f->strpos($urlToParse, "//") !== false) { |
| 89 |
$urlToParse = $f->str_replace('//', '/', $urlToParse); |
| 90 |
} |
| 91 |
$urlToParse = ltrim($abj404logic->urlNormalization()->removeHomeDirectory($urlToParse), '/'); |
| 92 |
$urlToParse = get_site_url() . '/' . $urlToParse; |
| 93 |
} |
| 94 |
|
| 95 |
$urlParts = parse_url($urlToParse); |
| 96 |
if (!is_array($urlParts)) { |
| 97 |
$abj404logging->errorMessage('parse_url returned a non-array value. REQUEST_URI: "' . |
| 98 |
$sanitizer->normalizeUrlString($_SERVER['REQUEST_URI']) . '", parse_url result: "' . json_encode($urlParts) . '", ' . |
| 99 |
'urlToParse result: ' . $urlToParse); |
| 100 |
return false; |
| 101 |
} |
| 102 |
// make things work with foreign languages while avoiding XSS issues. |
| 103 |
foreach ($urlParts as $key => $value) { |
| 104 |
if ($key === 'query') { |
| 105 |
// For query strings, preserve reserved characters while removing invalid bytes. |
| 106 |
parse_str($value, $queryArray); |
| 107 |
$safeQueryArray = $sanitizer->sanitizeUrlComponent($queryArray); |
| 108 |
$urlParts[$key] = http_build_query(is_array($safeQueryArray) ? $safeQueryArray : $queryArray); |
| 109 |
} else { |
| 110 |
// Sanitize path/host/etc. without stripping reserved URL characters. |
| 111 |
$urlParts[$key] = $sanitizer->sanitizeUrlComponent($value); |
| 112 |
} |
| 113 |
} |
| 114 |
|
| 115 |
// remove a pointless trailing /amp |
| 116 |
$urlPath = isset($urlParts['path']) && is_string($urlParts['path']) ? $urlParts['path'] : ''; |
| 117 |
if ($urlPath !== '' && |
| 118 |
($f->endsWithCaseInsensitive($urlPath, '/amp') || |
| 119 |
$f->endsWithCaseInsensitive($urlPath, '/amp/') |
| 120 |
) |
| 121 |
&& $f->strlen($urlPath) >= 6) { |
| 122 |
$urlParts['path'] = $f->substr($urlPath, 0, $f->strlen($urlPath) - 4); |
| 123 |
} |
| 124 |
|
| 125 |
// remove any "/comment-page-???/" if there is one. |
| 126 |
/* tested with: |
| 127 |
* http://localhost:8888/404solution-site/2019/02/hello-world2/comment-page-2/#comment-26 |
| 128 |
* http://localhost:8888/404solution-site/2019/02/hello-world2/comment-page-2/ |
| 129 |
* http://localhost:8888/404solution-site/2019/02/hello-world2/comment-page-2 |
| 130 |
* http://localhost:8888/404solution-site/2019/02/hello-world2/comment-page-2/?quer=true |
| 131 |
*/ |
| 132 |
// Fix for PHP 8.2: Handle URLs with no path component (e.g., http://example.com) |
| 133 |
$urlWithoutCommentPage = (isset($urlParts['path']) && is_string($urlParts['path'])) ? $urlParts['path'] : '/'; |
| 134 |
$commentPagePart = ''; |
| 135 |
$results = array(); |
| 136 |
if (isset($wp_rewrite) && isset($wp_rewrite->comments_pagination_base)) { |
| 137 |
$safeBase = preg_quote($wp_rewrite->comments_pagination_base); |
| 138 |
$commentregex = '(.*)\/(' . $safeBase . '-[0-9]{1,})(\/|\z)?(.*)'; |
| 139 |
$f->regexMatch($commentregex, $urlWithoutCommentPage, $results); |
| 140 |
|
| 141 |
if (!empty($results)) { |
| 142 |
$urlWithoutCommentPage = $results[1]; |
| 143 |
$commentPagePart = $results[2]; |
| 144 |
$commentPagePart = ($commentPagePart == '') ? '' : $commentPagePart . '/'; |
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
$queryString = ''; |
| 149 |
if (!array_key_exists('query', $urlParts) || @$urlParts['query'] == "") { |
| 150 |
$queryString = ''; |
| 151 |
} else { |
| 152 |
$queryString = $urlParts['query']; |
| 153 |
} |
| 154 |
|
| 155 |
/** @var array<string, int|string> $urlPartsSafe */ |
| 156 |
$urlPartsSafe = $urlParts; |
| 157 |
self::$instance = new ABJ_404_Solution_UserRequest(new ABJ_404_Solution_UserRequestParts( |
| 158 |
$urlToParse, $urlPartsSafe, $urlWithoutCommentPage, $commentPagePart, $queryString)); |
| 159 |
|
| 160 |
return true; |
| 161 |
} |
| 162 |
|
| 163 |
/** |
| 164 |
* @param ABJ_404_Solution_UserRequestParts $parts |
| 165 |
*/ |
| 166 |
private function __construct(ABJ_404_Solution_UserRequestParts $parts) { |
| 167 |
$this->requestURI = $parts->requestURI; |
| 168 |
$this->urlParts = $parts->urlParts; |
| 169 |
$this->requestURIWithoutCommentsPage = $parts->urlWithoutCommentPage; |
| 170 |
$this->commentPagePart = $parts->commentPagePart; |
| 171 |
|
| 172 |
$this->queryString = $parts->queryString; |
| 173 |
} |
| 174 |
|
| 175 |
/** @return string|null */ |
| 176 |
function getRequestURI() { |
| 177 |
return $this->requestURI; |
| 178 |
} |
| 179 |
|
| 180 |
/** @return string|null */ |
| 181 |
function getRequestURIWithoutCommentsPage() { |
| 182 |
return $this->requestURIWithoutCommentsPage; |
| 183 |
} |
| 184 |
|
| 185 |
/** http://s.com/404solution-site/hello-world/comment-page-2/#comment-26?query_info=true becomes |
| 186 |
* /404solution-site/hello-world/comment-page-2/ |
| 187 |
* @return string |
| 188 |
*/ |
| 189 |
function getPath() { |
| 190 |
if ($this->urlParts === null || !array_key_exists('path', $this->urlParts)) { |
| 191 |
// this happens for a request with no path. like http://example.com |
| 192 |
return ''; |
| 193 |
} |
| 194 |
|
| 195 |
return (string)($this->urlParts['path']); |
| 196 |
} |
| 197 |
|
| 198 |
/** @return string */ |
| 199 |
function getPathWithSortedQueryString(): string { |
| 200 |
$requestedURL = $this->getPath(); |
| 201 |
/** @var array<string, string> $urlPartsForSort */ |
| 202 |
$urlPartsForSort = $this->getUrlParts() ?? array(); |
| 203 |
$urlParts = abj_service('query_string_helper')->sortQueryString($urlPartsForSort); |
| 204 |
if ($urlParts != null && trim($urlParts) != '') { |
| 205 |
$requestedURL .= '?' . $urlParts; |
| 206 |
} |
| 207 |
|
| 208 |
// otherwise various queries break. |
| 209 |
$requestedURL = abj_service('url_encoder')->urlencodeEmojis($requestedURL); |
| 210 |
|
| 211 |
return $requestedURL; |
| 212 |
} |
| 213 |
|
| 214 |
/** http://s.com/404solution-site/hello-world/comment-page-2/#comment-26?query_info=true becomes |
| 215 |
* /hello-world/comment-page-2/ |
| 216 |
* @return string |
| 217 |
*/ |
| 218 |
function getOnlyTheSlug() { |
| 219 |
$abj404logic = abj_service('plugin_logic'); |
| 220 |
$path = $this->getRequestURIWithoutCommentsPage(); |
| 221 |
return $abj404logic->urlNormalization()->removeHomeDirectory($path); |
| 222 |
} |
| 223 |
|
| 224 |
/** @return array<string, int|string>|null */ |
| 225 |
function getUrlParts() { |
| 226 |
return $this->urlParts; |
| 227 |
} |
| 228 |
|
| 229 |
/** @return string|null */ |
| 230 |
function getQueryString() { |
| 231 |
return $this->queryString; |
| 232 |
} |
| 233 |
|
| 234 |
/** @return string|null */ |
| 235 |
function getCommentPagePart() { |
| 236 |
return $this->commentPagePart; |
| 237 |
} |
| 238 |
|
| 239 |
} |
| 240 |
|
