| 1 |
<?php |
| 2 |
/** |
| 3 |
* Auto mark as spam - email address blocklist. |
| 4 |
* |
| 5 |
* A site-wide list of email addresses and wildcard patterns. A submission that |
| 6 |
* carries one of them in an email field is classified as spam by the |
| 7 |
* submission handler whatever the captcha of that form decided: a captcha |
| 8 |
* judges how a form was filled in, not who filled it in, so a known spammer |
| 9 |
* who solves the challenge correctly still has to be caught. |
| 10 |
* |
| 11 |
* The list is stored as the text the administrator typed (one entry per line, |
| 12 |
* '#' comment lines kept) rather than as an array, so the settings textarea can |
| 13 |
* round-trip it unchanged and the order and the comments survive a save. |
| 14 |
* |
| 15 |
* Deliberately hook-free - plain functions only, like includes/data-deletion.php |
| 16 |
* - and cheap when nothing is configured: the scan returns before it looks at a |
| 17 |
* single submitted value if the list is empty. |
| 18 |
* |
| 19 |
* @package Contact Forms |
| 20 |
* @since 2.3.1 |
| 21 |
*/ |
| 22 |
|
| 23 |
if ( ! defined( 'ABSPATH' ) ) exit; |
| 24 |
|
| 25 |
/** |
| 26 |
* The stored spam settings, with every key guaranteed present. |
| 27 |
* |
| 28 |
* @since 2.3.1 |
| 29 |
* @return array The settings, with at least the 'email_blocklist' key. |
| 30 |
*/ |
| 31 |
function accua_forms_spam_settings() { |
| 32 |
$defaults = array( |
| 33 |
'email_blocklist' => '', |
| 34 |
); |
| 35 |
$data = get_option( 'accua_forms_spam_data', array() ); |
| 36 |
if ( ! is_array( $data ) ) { |
| 37 |
$data = array(); |
| 38 |
} |
| 39 |
$data = $data + $defaults; |
| 40 |
|
| 41 |
// The list is stored as text, but an option written by hand, by a |
| 42 |
// migration or by an older shape can hold an array. Casting that to |
| 43 |
// string raises "Array to string conversion" in every caller, so fold it |
| 44 |
// back into lines instead of losing it. |
| 45 |
if ( is_array( $data['email_blocklist'] ) ) { |
| 46 |
$entries = array(); |
| 47 |
foreach ( $data['email_blocklist'] as $entry ) { |
| 48 |
if ( is_scalar( $entry ) ) { |
| 49 |
$entries[] = (string) $entry; |
| 50 |
} |
| 51 |
} |
| 52 |
$data['email_blocklist'] = implode( "\n", $entries ); |
| 53 |
} elseif ( ! is_string( $data['email_blocklist'] ) ) { |
| 54 |
$data['email_blocklist'] = is_scalar( $data['email_blocklist'] ) ? (string) $data['email_blocklist'] : ''; |
| 55 |
} |
| 56 |
|
| 57 |
return $data; |
| 58 |
} |
| 59 |
|
| 60 |
/** |
| 61 |
* Clean up a single blocklist entry. |
| 62 |
* |
| 63 |
* Accepts what people actually paste - "Spam Bot <bot@example.net>", |
| 64 |
* "mailto:bot@example.net" - and lowercases everything, since addresses are |
| 65 |
* matched case-insensitively. |
| 66 |
* |
| 67 |
* Returns '' for anything unusable, which includes the entries that would |
| 68 |
* silently mark every submission carrying an email address as spam: an entry |
| 69 |
* made only of wildcards and '@' ('*', '*@*', '?') is dropped rather than |
| 70 |
* stored. Blocking everything stays expressible per domain ('*@example.com'), |
| 71 |
* never by accident. |
| 72 |
* |
| 73 |
* @since 2.3.1 |
| 74 |
* @param string $entry One entry as typed. |
| 75 |
* @return string The normalized entry, or '' if it cannot be used. |
| 76 |
*/ |
| 77 |
function accua_forms_normalize_blocklist_entry( $entry ) { |
| 78 |
$entry = trim( (string) $entry ); |
| 79 |
if ( '' === $entry ) { |
| 80 |
return ''; |
| 81 |
} |
| 82 |
|
| 83 |
// "Display Name <address>" - keep the address. The trim() after it catches |
| 84 |
// an unbalanced bracket, which the pattern above does not match. |
| 85 |
if ( preg_match( '/<([^>]*)>/', $entry, $matches ) ) { |
| 86 |
$entry = $matches[1]; |
| 87 |
} |
| 88 |
$entry = trim( $entry, " \t<>" ); |
| 89 |
$entry = preg_replace( '/^mailto:/i', '', $entry ); |
| 90 |
$entry = strtolower( trim( $entry ) ); |
| 91 |
|
| 92 |
// Characters no address and no pattern over one can contain. Whitespace, |
| 93 |
// commas and semicolons are already gone (they separate entries), so what |
| 94 |
// is left here is a typo that could never match anything. |
| 95 |
if ( '' === $entry || preg_match( '/[<>"\'\\\\()\[\]:;,\s]/', $entry ) ) { |
| 96 |
return ''; |
| 97 |
} |
| 98 |
|
| 99 |
// A run of consecutive '*' means exactly what one '*' means, and each one |
| 100 |
// becomes a '.*' in the match regex: 20 of them in a row turn a failed |
| 101 |
// match into minutes of backtracking. Collapsing is not a restriction, |
| 102 |
// the pattern keeps matching precisely what it did. |
| 103 |
$entry = preg_replace( '/\*{2,}/', '*', $entry ); |
| 104 |
|
| 105 |
// At least one character that is not a wildcard or the '@' itself. |
| 106 |
if ( ! preg_match( '/[^*?@]/', $entry ) ) { |
| 107 |
return ''; |
| 108 |
} |
| 109 |
|
| 110 |
// An entry with no '@' is shorthand for a whole domain, so it has to look |
| 111 |
// like one: a dot, or a wildcard. Without this check a stray word from a |
| 112 |
// pasted line would become an entry of its own and be read as a domain - |
| 113 |
// "Spam Bot <bot@example.net>" splits on its spaces, and "spam" would then |
| 114 |
// match every address at a host called spam. |
| 115 |
if ( false === strpos( $entry, '@' ) |
| 116 |
&& false === strpos( $entry, '.' ) |
| 117 |
&& ! preg_match( '/[*?]/', $entry ) ) { |
| 118 |
return ''; |
| 119 |
} |
| 120 |
|
| 121 |
return $entry; |
| 122 |
} |
| 123 |
|
| 124 |
/** |
| 125 |
* Normalize a raw blocklist textarea into the text that gets stored. |
| 126 |
* |
| 127 |
* Entries may be separated by newlines, commas, semicolons or spaces; lines |
| 128 |
* starting with '#' are comments and are kept verbatim (the list doubles as |
| 129 |
* the note of why an address is on it). Duplicates are dropped. |
| 130 |
* |
| 131 |
* @since 2.3.1 |
| 132 |
* @param string $raw The submitted textarea content. |
| 133 |
* @return string One entry (or comment) per line. |
| 134 |
*/ |
| 135 |
function accua_forms_filter_email_blocklist( $raw ) { |
| 136 |
$lines = array(); |
| 137 |
$seen = array(); |
| 138 |
|
| 139 |
foreach ( preg_split( '/[\r\n]+/', (string) $raw ) as $line ) { |
| 140 |
$line = trim( $line ); |
| 141 |
if ( '' === $line ) { |
| 142 |
continue; |
| 143 |
} |
| 144 |
if ( 0 === strpos( $line, '#' ) ) { |
| 145 |
$lines[] = $line; |
| 146 |
continue; |
| 147 |
} |
| 148 |
// "Display Name <address>" has to be unwrapped before the line is split |
| 149 |
// on its spaces, or the display name becomes entries of its own. |
| 150 |
if ( preg_match_all( '/<([^>]*)>/', $line, $matches ) ) { |
| 151 |
$line = implode( ' ', $matches[1] ); |
| 152 |
} |
| 153 |
foreach ( preg_split( '/[\s,;]+/', $line ) as $entry ) { |
| 154 |
$entry = accua_forms_normalize_blocklist_entry( $entry ); |
| 155 |
if ( '' === $entry || isset( $seen[ $entry ] ) ) { |
| 156 |
continue; |
| 157 |
} |
| 158 |
$seen[ $entry ] = true; |
| 159 |
$lines[] = $entry; |
| 160 |
} |
| 161 |
} |
| 162 |
|
| 163 |
return implode( "\n", $lines ); |
| 164 |
} |
| 165 |
|
| 166 |
/** |
| 167 |
* The blocklist as a list of patterns, without comments or blank lines. |
| 168 |
* |
| 169 |
* @since 2.3.1 |
| 170 |
* @return array List of patterns. |
| 171 |
*/ |
| 172 |
function accua_forms_get_email_blocklist() { |
| 173 |
$settings = accua_forms_spam_settings(); |
| 174 |
$patterns = array(); |
| 175 |
|
| 176 |
foreach ( preg_split( '/[\r\n]+/', (string) $settings['email_blocklist'] ) as $line ) { |
| 177 |
$line = trim( $line ); |
| 178 |
if ( '' === $line || 0 === strpos( $line, '#' ) ) { |
| 179 |
continue; |
| 180 |
} |
| 181 |
$patterns[] = $line; |
| 182 |
} |
| 183 |
|
| 184 |
/** |
| 185 |
* Filter the email blocklist patterns. |
| 186 |
* |
| 187 |
* Lets code add addresses the settings page does not hold - a list shared |
| 188 |
* across a network, an external service - without touching the option. |
| 189 |
* |
| 190 |
* @since 2.3.1 |
| 191 |
* @param array $patterns Patterns read from the settings page. |
| 192 |
*/ |
| 193 |
$patterns = apply_filters( 'accua_forms_email_blocklist', $patterns ); |
| 194 |
|
| 195 |
if ( ! is_array( $patterns ) ) { |
| 196 |
return array(); |
| 197 |
} |
| 198 |
|
| 199 |
$clean = array(); |
| 200 |
foreach ( $patterns as $pattern ) { |
| 201 |
if ( is_scalar( $pattern ) && '' !== trim( (string) $pattern ) ) { |
| 202 |
$clean[] = trim( (string) $pattern ); |
| 203 |
} |
| 204 |
} |
| 205 |
return $clean; |
| 206 |
} |
| 207 |
|
| 208 |
/** |
| 209 |
* Does an email address match one blocklist pattern? |
| 210 |
* |
| 211 |
* Matching is case-insensitive over the whole address, with '*' standing for |
| 212 |
* any run of characters and '?' for exactly one. A pattern with no '@' |
| 213 |
* ("spammydomain.net") and one that starts with it ("@spammydomain.net") both |
| 214 |
* read as "*@spammydomain.net" - the common case, spelled the way people write |
| 215 |
* it. Subdomains are not implied: "*@*.spammydomain.net" is how you catch them. |
| 216 |
* |
| 217 |
* @since 2.3.1 |
| 218 |
* @param string $email The address to test. |
| 219 |
* @param string $pattern One blocklist entry. |
| 220 |
* @return bool |
| 221 |
*/ |
| 222 |
function accua_forms_email_matches_blocklist_pattern( $email, $pattern ) { |
| 223 |
$email = strtolower( trim( (string) $email ) ); |
| 224 |
$pattern = strtolower( trim( (string) $pattern ) ); |
| 225 |
|
| 226 |
if ( '' === $email || '' === $pattern ) { |
| 227 |
return false; |
| 228 |
} |
| 229 |
|
| 230 |
// Patterns from the accua_forms_email_blocklist filter, and any stored by |
| 231 |
// a version before this one, never passed through the normalizer: collapse |
| 232 |
// the wildcard runs here too. Semantics are unchanged, there is simply |
| 233 |
// less to walk. |
| 234 |
$pattern = preg_replace( '/\*{2,}/', '*', $pattern ); |
| 235 |
|
| 236 |
if ( false === strpos( $pattern, '@' ) ) { |
| 237 |
$pattern = '*@' . $pattern; |
| 238 |
} elseif ( 0 === strpos( $pattern, '@' ) ) { |
| 239 |
$pattern = '*' . $pattern; |
| 240 |
} |
| 241 |
|
| 242 |
return accua_forms_glob_matches( $pattern, $email ); |
| 243 |
} |
| 244 |
|
| 245 |
/** |
| 246 |
* Glob matching without a regex. |
| 247 |
* |
| 248 |
* The obvious implementation turns the pattern into a regex ('*' to '.*', |
| 249 |
* '?' to '.') and calls preg_match(). That is what this did until 2.3.2, and |
| 250 |
* it fails quietly: every '*' becomes a backtracking point, so a pattern as |
| 251 |
* ordinary as '*info*@*.example.com' exhausts PCRE's backtrack limit against a |
| 252 |
* long address that does not match. preg_match() then returns false, not 0, |
| 253 |
* and the caller's '1 === preg_match(...)' reads that as "no match" - the |
| 254 |
* blocklist entry silently stops blocking, and only for some addresses. |
| 255 |
* |
| 256 |
* This is the standard linear glob walk instead: advance through both strings, |
| 257 |
* remember the last '*' and the position it was matched at, and on a mismatch |
| 258 |
* return to it having consumed one more character. No recursion, no |
| 259 |
* backtracking stack, no limit to exhaust. |
| 260 |
* |
| 261 |
* Comparison is byte-wise, so '?' matches one byte exactly as the regex '.' |
| 262 |
* did without the /u modifier. Both arguments are already lowercased. |
| 263 |
* |
| 264 |
* @since 2.3.2 |
| 265 |
* @param string $pattern Glob pattern ('*' any run, '?' exactly one). |
| 266 |
* @param string $subject The string to test. |
| 267 |
* @return bool |
| 268 |
*/ |
| 269 |
function accua_forms_glob_matches( $pattern, $subject ) { |
| 270 |
$plen = strlen( $pattern ); |
| 271 |
$slen = strlen( $subject ); |
| 272 |
$p = 0; |
| 273 |
$s = 0; |
| 274 |
$star = -1; |
| 275 |
$mark = 0; |
| 276 |
|
| 277 |
while ( $s < $slen ) { |
| 278 |
if ( $p < $plen && ( '?' === $pattern[ $p ] || $pattern[ $p ] === $subject[ $s ] ) ) { |
| 279 |
$p++; |
| 280 |
$s++; |
| 281 |
} elseif ( $p < $plen && '*' === $pattern[ $p ] ) { |
| 282 |
$star = $p; |
| 283 |
$mark = $s; |
| 284 |
$p++; |
| 285 |
} elseif ( $star >= 0 ) { |
| 286 |
$p = $star + 1; |
| 287 |
$mark++; |
| 288 |
$s = $mark; |
| 289 |
} else { |
| 290 |
return false; |
| 291 |
} |
| 292 |
} |
| 293 |
|
| 294 |
while ( $p < $plen && '*' === $pattern[ $p ] ) { |
| 295 |
$p++; |
| 296 |
} |
| 297 |
|
| 298 |
return $p === $plen; |
| 299 |
} |
| 300 |
|
| 301 |
/** |
| 302 |
* Is this email address on the blocklist? |
| 303 |
* |
| 304 |
* @since 2.3.1 |
| 305 |
* @param string $email The address to test. |
| 306 |
* @return bool |
| 307 |
*/ |
| 308 |
function accua_forms_email_is_blocklisted( $email ) { |
| 309 |
$email = trim( (string) $email ); |
| 310 |
$blocked = false; |
| 311 |
|
| 312 |
if ( '' !== $email ) { |
| 313 |
foreach ( accua_forms_get_email_blocklist() as $pattern ) { |
| 314 |
if ( accua_forms_email_matches_blocklist_pattern( $email, $pattern ) ) { |
| 315 |
$blocked = true; |
| 316 |
break; |
| 317 |
} |
| 318 |
} |
| 319 |
} |
| 320 |
|
| 321 |
/** |
| 322 |
* Filter whether an email address is treated as spam. |
| 323 |
* |
| 324 |
* @since 2.3.1 |
| 325 |
* @param bool $blocked Whether the blocklist matched. |
| 326 |
* @param string $email The address that was tested. |
| 327 |
*/ |
| 328 |
return (bool) apply_filters( 'accua_forms_email_is_blocklisted', $blocked, $email ); |
| 329 |
} |
| 330 |
|
| 331 |
/** |
| 332 |
* The field types the blocklist reads. |
| 333 |
* |
| 334 |
* @since 2.3.1 |
| 335 |
* @return array Field type ids. |
| 336 |
*/ |
| 337 |
function accua_forms_spam_email_field_types() { |
| 338 |
/** |
| 339 |
* Filter which field types are scanned against the email blocklist. |
| 340 |
* |
| 341 |
* @since 2.3.1 |
| 342 |
* @param array $types Field type ids. |
| 343 |
*/ |
| 344 |
$types = apply_filters( 'accua_forms_spam_email_field_types', array( 'email', 'autoreply_email' ) ); |
| 345 |
|
| 346 |
return is_array( $types ) ? $types : array(); |
| 347 |
} |
| 348 |
|
| 349 |
/** |
| 350 |
* Does this submission carry a blocklisted address in one of its email fields? |
| 351 |
* |
| 352 |
* @since 2.3.1 |
| 353 |
* @param array $submitted_data Submitted values keyed by field instance id. |
| 354 |
* @param array $form_data The saved form data (needs 'fields'). |
| 355 |
* @return bool |
| 356 |
*/ |
| 357 |
function accua_forms_submission_has_blocklisted_email( $submitted_data, $form_data ) { |
| 358 |
if ( ! is_array( $submitted_data ) || empty( $form_data['fields'] ) || ! is_array( $form_data['fields'] ) ) { |
| 359 |
return false; |
| 360 |
} |
| 361 |
|
| 362 |
// Nothing configured: never look at the submitted values at all. |
| 363 |
if ( ! accua_forms_get_email_blocklist() ) { |
| 364 |
return false; |
| 365 |
} |
| 366 |
|
| 367 |
$avail_fields = get_option( 'accua_forms_avail_fields', array() ); |
| 368 |
$email_types = accua_forms_spam_email_field_types(); |
| 369 |
|
| 370 |
foreach ( $submitted_data as $istance_id => $value ) { |
| 371 |
if ( empty( $form_data['fields'][ $istance_id ]['ref'] ) ) { |
| 372 |
continue; |
| 373 |
} |
| 374 |
$ref = $form_data['fields'][ $istance_id ]['ref']; |
| 375 |
if ( empty( $avail_fields[ $ref ]['type'] ) |
| 376 |
|| ! in_array( $avail_fields[ $ref ]['type'], $email_types, true ) ) { |
| 377 |
continue; |
| 378 |
} |
| 379 |
foreach ( (array) $value as $single ) { |
| 380 |
if ( is_scalar( $single ) && accua_forms_email_is_blocklisted( $single ) ) { |
| 381 |
return true; |
| 382 |
} |
| 383 |
} |
| 384 |
} |
| 385 |
|
| 386 |
return false; |
| 387 |
} |
| 388 |
|
