| commit | author | age | ||
| 48e9c1 | 1 | <?php |
| T | 2 | |
| 3 | /** | |
| 4 | * Class for operations on Sieve scripts | |
| 5 | * | |
| 6 | * Copyright (C) 2008-2011, The Roundcube Dev Team | |
| 7 | * Copyright (C) 2011, Kolab Systems AG | |
| 8 | * | |
| 07c6c6 | 9 | * This program is free software: you can redistribute it and/or modify |
| TB | 10 | * it under the terms of the GNU General Public License as published by |
| 11 | * the Free Software Foundation, either version 3 of the License, or | |
| 12 | * (at your option) any later version. | |
| 48e9c1 | 13 | * |
| T | 14 | * This program is distributed in the hope that it will be useful, |
| 15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
| 17 | * GNU General Public License for more details. | |
| 18 | * | |
| 07c6c6 | 19 | * You should have received a copy of the GNU General Public License |
| TB | 20 | * along with this program. If not, see http://www.gnu.org/licenses/. |
| 48e9c1 | 21 | */ |
| T | 22 | |
| 23 | class rcube_sieve_script | |
| 24 | { | |
| 25 | public $content = array(); // script rules array | |
| 26 | ||
| e499a1 | 27 | private $vars = array(); // "global" variables |
| AM | 28 | private $prefix = ''; // script header (comments) |
| 29 | private $supported = array( // supported Sieve extensions: | |
| ed837a | 30 | 'body', // RFC5173 |
| 48e9c1 | 31 | 'copy', // RFC3894 |
| ed837a | 32 | 'date', // RFC5260 |
| e499a1 | 33 | 'duplicate', // RFC7352 |
| ed837a | 34 | 'enotify', // RFC5435 |
| AM | 35 | 'envelope', // RFC5228 |
| 36 | 'ereject', // RFC5429 | |
| 37 | 'fileinto', // RFC5228 | |
| 48e9c1 | 38 | 'imapflags', // draft-melnikov-sieve-imapflags-06 |
| T | 39 | 'imap4flags', // RFC5232 |
| e499a1 | 40 | 'include', // RFC6609 |
| ed837a | 41 | 'index', // RFC5260 |
| e499a1 | 42 | 'notify', // RFC5435 |
| ed837a | 43 | 'regex', // draft-ietf-sieve-regex-01 |
| AM | 44 | 'reject', // RFC5429 |
| 45 | 'relational', // RFC3431 | |
| 46 | 'subaddress', // RFC5233 | |
| 47 | 'vacation', // RFC5230 | |
| 48 | 'vacation-seconds', // RFC6131 | |
| 49 | 'variables', // RFC5229 | |
| 50 | // @TODO: spamtest+virustest, mailbox | |
| 48e9c1 | 51 | ); |
| T | 52 | |
| 53 | /** | |
| 54 | * Object constructor | |
| 55 | * | |
| 56 | * @param string Script's text content | |
| 57 | * @param array List of capabilities supported by server | |
| 58 | */ | |
| 59 | public function __construct($script, $capabilities=array()) | |
| 60 | { | |
| 61 | $capabilities = array_map('strtolower', (array) $capabilities); | |
| 62 | ||
| 63 | // disable features by server capabilities | |
| 64 | if (!empty($capabilities)) { | |
| 65 | foreach ($this->supported as $idx => $ext) { | |
| 66 | if (!in_array($ext, $capabilities)) { | |
| 67 | unset($this->supported[$idx]); | |
| 68 | } | |
| 69 | } | |
| 70 | } | |
| 71 | ||
| 72 | // Parse text content of the script | |
| 73 | $this->_parse_text($script); | |
| 74 | } | |
| 75 | ||
| 76 | /** | |
| 77 | * Adds rule to the script (at the end) | |
| 78 | * | |
| 79 | * @param string Rule name | |
| 80 | * @param array Rule content (as array) | |
| 81 | * | |
| 82 | * @return int The index of the new rule | |
| 83 | */ | |
| 84 | public function add_rule($content) | |
| 85 | { | |
| 86 | // TODO: check this->supported | |
| 87 | array_push($this->content, $content); | |
| 88 | return sizeof($this->content)-1; | |
| 89 | } | |
| 90 | ||
| 91 | public function delete_rule($index) | |
| 92 | { | |
| 93 | if(isset($this->content[$index])) { | |
| 94 | unset($this->content[$index]); | |
| 95 | return true; | |
| 96 | } | |
| 97 | return false; | |
| 98 | } | |
| 99 | ||
| 100 | public function size() | |
| 101 | { | |
| 102 | return sizeof($this->content); | |
| 103 | } | |
| 104 | ||
| 105 | public function update_rule($index, $content) | |
| 106 | { | |
| 107 | // TODO: check this->supported | |
| 108 | if ($this->content[$index]) { | |
| 109 | $this->content[$index] = $content; | |
| 110 | return $index; | |
| 111 | } | |
| 112 | return false; | |
| 113 | } | |
| 114 | ||
| 115 | /** | |
| 116 | * Sets "global" variable | |
| 117 | * | |
| 118 | * @param string $name Variable name | |
| 119 | * @param string $value Variable value | |
| 120 | * @param array $mods Variable modifiers | |
| 121 | */ | |
| 122 | public function set_var($name, $value, $mods = array()) | |
| 123 | { | |
| 124 | // Check if variable exists | |
| 125 | for ($i=0, $len=count($this->vars); $i<$len; $i++) { | |
| 126 | if ($this->vars[$i]['name'] == $name) { | |
| 127 | break; | |
| 128 | } | |
| 129 | } | |
| 130 | ||
| 131 | $var = array_merge($mods, array('name' => $name, 'value' => $value)); | |
| 132 | $this->vars[$i] = $var; | |
| 133 | } | |
| 134 | ||
| 135 | /** | |
| 136 | * Unsets "global" variable | |
| 137 | * | |
| 138 | * @param string $name Variable name | |
| 139 | */ | |
| 140 | public function unset_var($name) | |
| 141 | { | |
| 142 | // Check if variable exists | |
| 143 | foreach ($this->vars as $idx => $var) { | |
| 144 | if ($var['name'] == $name) { | |
| 145 | unset($this->vars[$idx]); | |
| 146 | break; | |
| 147 | } | |
| 148 | } | |
| 149 | } | |
| 150 | ||
| 151 | /** | |
| 152 | * Gets the value of "global" variable | |
| 153 | * | |
| 154 | * @param string $name Variable name | |
| 155 | * | |
| 156 | * @return string Variable value | |
| 157 | */ | |
| 158 | public function get_var($name) | |
| 159 | { | |
| 160 | // Check if variable exists | |
| 161 | for ($i=0, $len=count($this->vars); $i<$len; $i++) { | |
| 162 | if ($this->vars[$i]['name'] == $name) { | |
| 163 | return $this->vars[$i]['name']; | |
| 164 | } | |
| 165 | } | |
| 166 | } | |
| 167 | ||
| 168 | /** | |
| 169 | * Sets script header content | |
| 170 | * | |
| 171 | * @param string $text Header content | |
| 172 | */ | |
| 173 | public function set_prefix($text) | |
| 174 | { | |
| 175 | $this->prefix = $text; | |
| 176 | } | |
| 177 | ||
| 178 | /** | |
| 179 | * Returns script as text | |
| 180 | */ | |
| 181 | public function as_text() | |
| 182 | { | |
| 183 | $output = ''; | |
| 184 | $exts = array(); | |
| 185 | $idx = 0; | |
| 186 | ||
| 187 | if (!empty($this->vars)) { | |
| 188 | if (in_array('variables', (array)$this->supported)) { | |
| 189 | $has_vars = true; | |
| 190 | array_push($exts, 'variables'); | |
| 191 | } | |
| 192 | foreach ($this->vars as $var) { | |
| 193 | if (empty($has_vars)) { | |
| 194 | // 'variables' extension not supported, put vars in comments | |
| 195 | $output .= sprintf("# %s %s\n", $var['name'], $var['value']); | |
| 196 | } | |
| 197 | else { | |
| 198 | $output .= 'set '; | |
| 199 | foreach (array_diff(array_keys($var), array('name', 'value')) as $opt) { | |
| 200 | $output .= ":$opt "; | |
| 201 | } | |
| 202 | $output .= self::escape_string($var['name']) . ' ' . self::escape_string($var['value']) . ";\n"; | |
| 203 | } | |
| 204 | } | |
| 205 | } | |
| eb1ee0 | 206 | |
| AM | 207 | $imapflags = in_array('imap4flags', $this->supported) ? 'imap4flags' : 'imapflags'; |
| 208 | $notify = in_array('enotify', $this->supported) ? 'enotify' : 'notify'; | |
| 48e9c1 | 209 | |
| T | 210 | // rules |
| 211 | foreach ($this->content as $rule) { | |
| 212 | $script = ''; | |
| 213 | $tests = array(); | |
| 214 | $i = 0; | |
| 215 | ||
| 216 | // header | |
| 217 | if (!empty($rule['name']) && strlen($rule['name'])) { | |
| 218 | $script .= '# rule:[' . $rule['name'] . "]\n"; | |
| 219 | } | |
| 220 | ||
| 221 | // constraints expressions | |
| 222 | if (!empty($rule['tests'])) { | |
| 223 | foreach ($rule['tests'] as $test) { | |
| 224 | $tests[$i] = ''; | |
| 225 | switch ($test['test']) { | |
| 226 | case 'size': | |
| 227 | $tests[$i] .= ($test['not'] ? 'not ' : ''); | |
| 228 | $tests[$i] .= 'size :' . ($test['type']=='under' ? 'under ' : 'over ') . $test['arg']; | |
| 229 | break; | |
| 230 | ||
| 231 | case 'true': | |
| 232 | $tests[$i] .= ($test['not'] ? 'false' : 'true'); | |
| 233 | break; | |
| 234 | ||
| 235 | case 'exists': | |
| 236 | $tests[$i] .= ($test['not'] ? 'not ' : ''); | |
| 237 | $tests[$i] .= 'exists ' . self::escape_string($test['arg']); | |
| 238 | break; | |
| 239 | ||
| 240 | case 'header': | |
| 241 | $tests[$i] .= ($test['not'] ? 'not ' : ''); | |
| 242 | $tests[$i] .= 'header'; | |
| 243 | ||
| ed837a | 244 | $this->add_index($test, $tests[$i], $exts); |
| AM | 245 | $this->add_operator($test, $tests[$i], $exts); |
| 48e9c1 | 246 | |
| T | 247 | $tests[$i] .= ' ' . self::escape_string($test['arg1']); |
| 248 | $tests[$i] .= ' ' . self::escape_string($test['arg2']); | |
| 249 | break; | |
| 250 | ||
| 251 | case 'address': | |
| 252 | case 'envelope': | |
| 253 | if ($test['test'] == 'envelope') { | |
| 254 | array_push($exts, 'envelope'); | |
| 255 | } | |
| 256 | ||
| 257 | $tests[$i] .= ($test['not'] ? 'not ' : ''); | |
| 258 | $tests[$i] .= $test['test']; | |
| 259 | ||
| ed837a | 260 | if ($test['test'] != 'envelope') { |
| AM | 261 | $this->add_index($test, $tests[$i], $exts); |
| 262 | } | |
| 263 | ||
| b2c586 | 264 | // :all address-part is optional, skip it |
| AM | 265 | if (!empty($test['part']) && $test['part'] != 'all') { |
| 48e9c1 | 266 | $tests[$i] .= ' :' . $test['part']; |
| T | 267 | if ($test['part'] == 'user' || $test['part'] == 'detail') { |
| 268 | array_push($exts, 'subaddress'); | |
| 269 | } | |
| 270 | } | |
| 271 | ||
| ed837a | 272 | $this->add_operator($test, $tests[$i], $exts); |
| 48e9c1 | 273 | |
| T | 274 | $tests[$i] .= ' ' . self::escape_string($test['arg1']); |
| 275 | $tests[$i] .= ' ' . self::escape_string($test['arg2']); | |
| 276 | break; | |
| 277 | ||
| 278 | case 'body': | |
| 279 | array_push($exts, 'body'); | |
| 280 | ||
| 281 | $tests[$i] .= ($test['not'] ? 'not ' : '') . 'body'; | |
| 282 | ||
| 283 | if (!empty($test['part'])) { | |
| 284 | $tests[$i] .= ' :' . $test['part']; | |
| 285 | ||
| 286 | if (!empty($test['content']) && $test['part'] == 'content') { | |
| 287 | $tests[$i] .= ' ' . self::escape_string($test['content']); | |
| 288 | } | |
| 289 | } | |
| 290 | ||
| ed837a | 291 | $this->add_operator($test, $tests[$i], $exts); |
| 48e9c1 | 292 | |
| T | 293 | $tests[$i] .= ' ' . self::escape_string($test['arg']); |
| ed837a | 294 | break; |
| AM | 295 | |
| 296 | case 'date': | |
| 297 | case 'currentdate': | |
| 298 | array_push($exts, 'date'); | |
| 299 | ||
| 300 | $tests[$i] .= ($test['not'] ? 'not ' : '') . $test['test']; | |
| 301 | ||
| 302 | $this->add_index($test, $tests[$i], $exts); | |
| 303 | ||
| 304 | if (!empty($test['originalzone']) && $test['test'] == 'date') { | |
| 305 | $tests[$i] .= ' :originalzone'; | |
| 306 | } | |
| 307 | else if (!empty($test['zone'])) { | |
| 308 | $tests[$i] .= ' :zone ' . self::escape_string($test['zone']); | |
| 309 | } | |
| 310 | ||
| 311 | $this->add_operator($test, $tests[$i], $exts); | |
| 312 | ||
| 313 | if ($test['test'] == 'date') { | |
| 314 | $tests[$i] .= ' ' . self::escape_string($test['header']); | |
| 315 | } | |
| 316 | ||
| 317 | $tests[$i] .= ' ' . self::escape_string($test['part']); | |
| 318 | $tests[$i] .= ' ' . self::escape_string($test['arg']); | |
| 319 | ||
| 48e9c1 | 320 | break; |
| e499a1 | 321 | |
| AM | 322 | case 'duplicate': |
| 323 | array_push($exts, 'duplicate'); | |
| 324 | ||
| 325 | $tests[$i] .= ($test['not'] ? 'not ' : '') . $test['test']; | |
| 326 | ||
| 327 | $tokens = array('handle', 'uniqueid', 'header'); | |
| 328 | foreach ($tokens as $token) | |
| 329 | if ($test[$token] !== null && $test[$token] !== '') { | |
| 330 | $tests[$i] .= " :$token " . self::escape_string($test[$token]); | |
| 331 | } | |
| 332 | ||
| 333 | if (!empty($test['seconds'])) { | |
| 334 | $tests[$i] .= ' :seconds ' . intval($test['seconds']); | |
| 335 | } | |
| 336 | ||
| 337 | if (!empty($test['last'])) { | |
| 338 | $tests[$i] .= ' :last'; | |
| 339 | } | |
| 340 | ||
| 341 | break; | |
| 48e9c1 | 342 | } |
| e499a1 | 343 | |
| 48e9c1 | 344 | $i++; |
| T | 345 | } |
| 346 | } | |
| 347 | ||
| 348 | // disabled rule: if false #.... | |
| 349 | if (!empty($tests)) { | |
| 350 | $script .= 'if ' . ($rule['disabled'] ? 'false # ' : ''); | |
| 351 | ||
| 352 | if (count($tests) > 1) { | |
| 353 | $tests_str = implode(', ', $tests); | |
| 354 | } | |
| 355 | else { | |
| 356 | $tests_str = $tests[0]; | |
| 357 | } | |
| 358 | ||
| 359 | if ($rule['join'] || count($tests) > 1) { | |
| 360 | $script .= sprintf('%s (%s)', $rule['join'] ? 'allof' : 'anyof', $tests_str); | |
| 361 | } | |
| 362 | else { | |
| 363 | $script .= $tests_str; | |
| 364 | } | |
| 365 | $script .= "\n{\n"; | |
| 366 | } | |
| 367 | ||
| 368 | // action(s) | |
| 369 | if (!empty($rule['actions'])) { | |
| 370 | foreach ($rule['actions'] as $action) { | |
| 371 | $action_script = ''; | |
| 372 | ||
| 373 | switch ($action['type']) { | |
| 374 | ||
| 375 | case 'fileinto': | |
| 376 | array_push($exts, 'fileinto'); | |
| 377 | $action_script .= 'fileinto '; | |
| 378 | if ($action['copy']) { | |
| 379 | $action_script .= ':copy '; | |
| 380 | array_push($exts, 'copy'); | |
| 381 | } | |
| 382 | $action_script .= self::escape_string($action['target']); | |
| 383 | break; | |
| 384 | ||
| 385 | case 'redirect': | |
| 386 | $action_script .= 'redirect '; | |
| 387 | if ($action['copy']) { | |
| 388 | $action_script .= ':copy '; | |
| 389 | array_push($exts, 'copy'); | |
| 390 | } | |
| 391 | $action_script .= self::escape_string($action['target']); | |
| 392 | break; | |
| 393 | ||
| 394 | case 'reject': | |
| 395 | case 'ereject': | |
| 396 | array_push($exts, $action['type']); | |
| 397 | $action_script .= $action['type'].' ' | |
| 398 | . self::escape_string($action['target']); | |
| 399 | break; | |
| 400 | ||
| 401 | case 'addflag': | |
| 402 | case 'setflag': | |
| 403 | case 'removeflag': | |
| eb1ee0 | 404 | array_push($exts, $imapflags); |
| 48e9c1 | 405 | $action_script .= $action['type'].' ' |
| T | 406 | . self::escape_string($action['target']); |
| 407 | break; | |
| 408 | ||
| 409 | case 'keep': | |
| 410 | case 'discard': | |
| 411 | case 'stop': | |
| 412 | $action_script .= $action['type']; | |
| 413 | break; | |
| 414 | ||
| 415 | case 'include': | |
| 416 | array_push($exts, 'include'); | |
| 417 | $action_script .= 'include '; | |
| 418 | foreach (array_diff(array_keys($action), array('target', 'type')) as $opt) { | |
| 419 | $action_script .= ":$opt "; | |
| 420 | } | |
| 421 | $action_script .= self::escape_string($action['target']); | |
| 422 | break; | |
| 423 | ||
| 424 | case 'set': | |
| 425 | array_push($exts, 'variables'); | |
| 426 | $action_script .= 'set '; | |
| 427 | foreach (array_diff(array_keys($action), array('name', 'value', 'type')) as $opt) { | |
| 428 | $action_script .= ":$opt "; | |
| 429 | } | |
| 430 | $action_script .= self::escape_string($action['name']) . ' ' . self::escape_string($action['value']); | |
| 431 | break; | |
| 432 | ||
| 2ac00a | 433 | case 'notify': |
| eb1ee0 | 434 | array_push($exts, $notify); |
| 2ac00a | 435 | $action_script .= 'notify'; |
| eb1ee0 | 436 | |
| f22de5 | 437 | $method = $action['method']; |
| AM | 438 | unset($action['method']); |
| 439 | $action['options'] = (array) $action['options']; | |
| 440 | ||
| 441 | // Here we support draft-martin-sieve-notify-01 used by Cyrus | |
| cd97f0 | 442 | if ($notify == 'notify') { |
| AM | 443 | switch ($action['importance']) { |
| 444 | case 1: $action_script .= " :high"; break; | |
| f22de5 | 445 | //case 2: $action_script .= " :normal"; break; |
| cd97f0 | 446 | case 3: $action_script .= " :low"; break; |
| AM | 447 | } |
| f22de5 | 448 | |
| AM | 449 | // Old-draft way: :method "mailto" :options "email@address" |
| 450 | if (!empty($method)) { | |
| 451 | $parts = explode(':', $method, 2); | |
| 452 | $action['method'] = $parts[0]; | |
| 453 | array_unshift($action['options'], $parts[1]); | |
| 454 | } | |
| 455 | ||
| cd97f0 | 456 | unset($action['importance']); |
| f22de5 | 457 | unset($action['from']); |
| AM | 458 | unset($method); |
| cd97f0 | 459 | } |
| AM | 460 | |
| f22de5 | 461 | foreach (array('id', 'importance', 'method', 'options', 'from', 'message') as $n_tag) { |
| 2ac00a | 462 | if (!empty($action[$n_tag])) { |
| PS | 463 | $action_script .= " :$n_tag " . self::escape_string($action[$n_tag]); |
| 464 | } | |
| 465 | } | |
| cd97f0 | 466 | |
| eb1ee0 | 467 | if (!empty($method)) { |
| f22de5 | 468 | $action_script .= ' ' . self::escape_string($method); |
| eb1ee0 | 469 | } |
| AM | 470 | |
| 2ac00a | 471 | break; |
| PS | 472 | |
| 48e9c1 | 473 | case 'vacation': |
| T | 474 | array_push($exts, 'vacation'); |
| 475 | $action_script .= 'vacation'; | |
| b6fa7d | 476 | if (isset($action['seconds'])) { |
| AM | 477 | array_push($exts, 'vacation-seconds'); |
| 478 | $action_script .= " :seconds " . intval($action['seconds']); | |
| 479 | } | |
| 480 | else if (!empty($action['days'])) { | |
| 481 | $action_script .= " :days " . intval($action['days']); | |
| 482 | } | |
| 48e9c1 | 483 | if (!empty($action['addresses'])) |
| T | 484 | $action_script .= " :addresses " . self::escape_string($action['addresses']); |
| 485 | if (!empty($action['subject'])) | |
| 486 | $action_script .= " :subject " . self::escape_string($action['subject']); | |
| 487 | if (!empty($action['handle'])) | |
| 488 | $action_script .= " :handle " . self::escape_string($action['handle']); | |
| 489 | if (!empty($action['from'])) | |
| 490 | $action_script .= " :from " . self::escape_string($action['from']); | |
| 491 | if (!empty($action['mime'])) | |
| 492 | $action_script .= " :mime"; | |
| 493 | $action_script .= " " . self::escape_string($action['reason']); | |
| 494 | break; | |
| 495 | } | |
| 496 | ||
| 497 | if ($action_script) { | |
| 498 | $script .= !empty($tests) ? "\t" : ''; | |
| 499 | $script .= $action_script . ";\n"; | |
| 500 | } | |
| 501 | } | |
| 502 | } | |
| 503 | ||
| 504 | if ($script) { | |
| 505 | $output .= $script . (!empty($tests) ? "}\n" : ''); | |
| 506 | $idx++; | |
| 507 | } | |
| 508 | } | |
| 509 | ||
| 510 | // requires | |
| b6fa7d | 511 | if (!empty($exts)) { |
| AM | 512 | $exts = array_unique($exts); |
| 513 | ||
| 514 | if (in_array('vacation-seconds', $exts) && ($key = array_search('vacation', $exts)) !== false) { | |
| 515 | unset($exts[$key]); | |
| 516 | } | |
| 517 | ||
| ed837a | 518 | sort($exts); // for convenience use always the same order |
| AM | 519 | |
| b6fa7d | 520 | $output = 'require ["' . implode('","', $exts) . "\"];\n" . $output; |
| AM | 521 | } |
| 48e9c1 | 522 | |
| T | 523 | if (!empty($this->prefix)) { |
| 524 | $output = $this->prefix . "\n\n" . $output; | |
| 525 | } | |
| 526 | ||
| 527 | return $output; | |
| 528 | } | |
| 529 | ||
| 530 | /** | |
| 531 | * Returns script object | |
| 532 | * | |
| 533 | */ | |
| 534 | public function as_array() | |
| 535 | { | |
| 536 | return $this->content; | |
| 537 | } | |
| 538 | ||
| 539 | /** | |
| 540 | * Returns array of supported extensions | |
| 541 | * | |
| 542 | */ | |
| 543 | public function get_extensions() | |
| 544 | { | |
| 545 | return array_values($this->supported); | |
| 546 | } | |
| 547 | ||
| 548 | /** | |
| 549 | * Converts text script to rules array | |
| 550 | * | |
| 551 | * @param string Text script | |
| 552 | */ | |
| 553 | private function _parse_text($script) | |
| 554 | { | |
| 80cb36 | 555 | $prefix = ''; |
| AM | 556 | $options = array(); |
| 557 | $position = 0; | |
| 558 | $length = strlen($script); | |
| 48e9c1 | 559 | |
| 80cb36 | 560 | while ($position < $length) { |
| AM | 561 | // skip whitespace chars |
| 562 | $position = self::ltrim_position($script, $position); | |
| 563 | $rulename = ''; | |
| 48e9c1 | 564 | |
| T | 565 | // Comments |
| 80cb36 | 566 | while ($script[$position] === '#') { |
| AM | 567 | $endl = strpos($script, "\n", $position) ?: $length; |
| 568 | $line = substr($script, $position, $endl - $position); | |
| 48e9c1 | 569 | |
| T | 570 | // Roundcube format |
| 571 | if (preg_match('/^# rule:\[(.*)\]/', $line, $matches)) { | |
| 572 | $rulename = $matches[1]; | |
| 573 | } | |
| 574 | // KEP:14 variables | |
| 575 | else if (preg_match('/^# (EDITOR|EDITOR_VERSION) (.+)$/', $line, $matches)) { | |
| 576 | $this->set_var($matches[1], $matches[2]); | |
| 577 | } | |
| 578 | // Horde-Ingo format | |
| 579 | else if (!empty($options['format']) && $options['format'] == 'INGO' | |
| 580 | && preg_match('/^# (.*)/', $line, $matches) | |
| 581 | ) { | |
| 582 | $rulename = $matches[1]; | |
| 583 | } | |
| 584 | else if (empty($options['prefix'])) { | |
| 585 | $prefix .= $line . "\n"; | |
| 586 | } | |
| 587 | ||
| 80cb36 | 588 | $position = $endl + 1; |
| 48e9c1 | 589 | } |
| T | 590 | |
| 591 | // handle script header | |
| 592 | if (empty($options['prefix'])) { | |
| 593 | $options['prefix'] = true; | |
| 594 | if ($prefix && strpos($prefix, 'horde.org/ingo')) { | |
| 595 | $options['format'] = 'INGO'; | |
| 596 | } | |
| 597 | } | |
| 598 | ||
| 599 | // Control structures/blocks | |
| 80cb36 | 600 | if (preg_match('/^(if|else|elsif)/i', substr($script, $position, 5))) { |
| AM | 601 | $rule = $this->_tokenize_rule($script, $position); |
| 48e9c1 | 602 | if (strlen($rulename) && !empty($rule)) { |
| T | 603 | $rule['name'] = $rulename; |
| 604 | } | |
| 605 | } | |
| 606 | // Simple commands | |
| 607 | else { | |
| 80cb36 | 608 | $rule = $this->_parse_actions($script, $position, ';'); |
| 48e9c1 | 609 | if (!empty($rule[0]) && is_array($rule)) { |
| T | 610 | // set "global" variables |
| 611 | if ($rule[0]['type'] == 'set') { | |
| 612 | unset($rule[0]['type']); | |
| 613 | $this->vars[] = $rule[0]; | |
| 022f51 | 614 | unset($rule); |
| 48e9c1 | 615 | } |
| T | 616 | else { |
| 617 | $rule = array('actions' => $rule); | |
| 618 | } | |
| 619 | } | |
| 620 | } | |
| 621 | ||
| 622 | if (!empty($rule)) { | |
| 623 | $this->content[] = $rule; | |
| 624 | } | |
| 625 | } | |
| 626 | ||
| 627 | if (!empty($prefix)) { | |
| 628 | $this->prefix = trim($prefix); | |
| 629 | } | |
| 630 | } | |
| 631 | ||
| 632 | /** | |
| 633 | * Convert text script fragment to rule object | |
| 634 | * | |
| 80cb36 | 635 | * @param string $content The whole script content |
| AM | 636 | * @param int &$position Start position in the script |
| 48e9c1 | 637 | * |
| T | 638 | * @return array Rule data |
| 639 | */ | |
| 80cb36 | 640 | private function _tokenize_rule($content, &$position) |
| 48e9c1 | 641 | { |
| 80cb36 | 642 | $cond = strtolower(self::tokenize($content, 1, $position)); |
| 48e9c1 | 643 | if ($cond != 'if' && $cond != 'elsif' && $cond != 'else') { |
| T | 644 | return null; |
| 645 | } | |
| 646 | ||
| 647 | $disabled = false; | |
| 648 | $join = false; | |
| 6af79f | 649 | $join_not = false; |
| 80cb36 | 650 | $length = strlen($content); |
| 48e9c1 | 651 | |
| T | 652 | // disabled rule (false + comment): if false # ..... |
| 80cb36 | 653 | if (preg_match('/^\s*false\s+#\s*/i', substr($content, $position, 20), $m)) { |
| AM | 654 | $position += strlen($m[0]); |
| 48e9c1 | 655 | $disabled = true; |
| T | 656 | } |
| 657 | ||
| 80cb36 | 658 | while ($position < $length) { |
| AM | 659 | $tokens = self::tokenize($content, true, $position); |
| 48e9c1 | 660 | $separator = array_pop($tokens); |
| T | 661 | |
| 662 | if (!empty($tokens)) { | |
| 663 | $token = array_shift($tokens); | |
| 664 | } | |
| 665 | else { | |
| 666 | $token = $separator; | |
| 667 | } | |
| 668 | ||
| 669 | $token = strtolower($token); | |
| 670 | ||
| 671 | if ($token == 'not') { | |
| 672 | $not = true; | |
| 673 | $token = strtolower(array_shift($tokens)); | |
| 674 | } | |
| 675 | else { | |
| 676 | $not = false; | |
| 677 | } | |
| 678 | ||
| 6af79f | 679 | // we support "not allof" as a negation of allof sub-tests |
| AM | 680 | if ($join_not) { |
| 681 | $not = !$not; | |
| 682 | } | |
| 683 | ||
| 48e9c1 | 684 | switch ($token) { |
| T | 685 | case 'allof': |
| 6af79f | 686 | $join = true; |
| AM | 687 | $join_not = $not; |
| 48e9c1 | 688 | break; |
| 6af79f | 689 | |
| 48e9c1 | 690 | case 'anyof': |
| T | 691 | break; |
| 692 | ||
| 693 | case 'size': | |
| 6af79f | 694 | $test = array('test' => 'size', 'not' => $not); |
| ed837a | 695 | |
| AM | 696 | $test['arg'] = array_pop($tokens); |
| 697 | ||
| 48e9c1 | 698 | for ($i=0, $len=count($tokens); $i<$len; $i++) { |
| T | 699 | if (!is_array($tokens[$i]) |
| 700 | && preg_match('/^:(under|over)$/i', $tokens[$i]) | |
| 701 | ) { | |
| ed837a | 702 | $test['type'] = strtolower(substr($tokens[$i], 1)); |
| 48e9c1 | 703 | } |
| T | 704 | } |
| 705 | ||
| ed837a | 706 | $tests[] = $test; |
| 48e9c1 | 707 | break; |
| T | 708 | |
| 709 | case 'header': | |
| 710 | case 'address': | |
| 711 | case 'envelope': | |
| ed837a | 712 | $test = array('test' => $token, 'not' => $not); |
| AM | 713 | |
| 714 | $test['arg2'] = array_pop($tokens); | |
| 715 | $test['arg1'] = array_pop($tokens); | |
| 716 | ||
| 717 | $test += $this->test_tokens($tokens); | |
| 718 | ||
| 719 | if ($token != 'header' && !empty($tokens)) { | |
| 720 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| 721 | if (!is_array($tokens[$i]) && preg_match('/^:(localpart|domain|all|user|detail)$/i', $tokens[$i])) { | |
| 722 | $test['part'] = strtolower(substr($tokens[$i], 1)); | |
| 723 | } | |
| 48e9c1 | 724 | } |
| T | 725 | } |
| 726 | ||
| ed837a | 727 | $tests[] = $test; |
| 48e9c1 | 728 | break; |
| T | 729 | |
| 730 | case 'body': | |
| ed837a | 731 | $test = array('test' => 'body', 'not' => $not); |
| 48e9c1 | 732 | |
| ed837a | 733 | $test['arg'] = array_pop($tokens); |
| AM | 734 | |
| 735 | $test += $this->test_tokens($tokens); | |
| 736 | ||
| 737 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| 738 | if (!is_array($tokens[$i]) && preg_match('/^:(raw|content|text)$/i', $tokens[$i])) { | |
| 739 | $test['part'] = strtolower(substr($tokens[$i], 1)); | |
| 740 | ||
| 741 | if ($test['part'] == 'content') { | |
| 742 | $test['content'] = $tokens[++$i]; | |
| 48e9c1 | 743 | } |
| T | 744 | } |
| 745 | } | |
| 746 | ||
| ed837a | 747 | $tests[] = $test; |
| AM | 748 | break; |
| 749 | ||
| 750 | case 'date': | |
| 751 | case 'currentdate': | |
| 752 | $test = array('test' => $token, 'not' => $not); | |
| 753 | ||
| 754 | $test['arg'] = array_pop($tokens); | |
| 755 | $test['part'] = array_pop($tokens); | |
| 756 | ||
| 757 | if ($token == 'date') { | |
| 758 | $test['header'] = array_pop($tokens); | |
| 759 | } | |
| 760 | ||
| 761 | $test += $this->test_tokens($tokens); | |
| 762 | ||
| 763 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| 764 | if (!is_array($tokens[$i]) && preg_match('/^:zone$/i', $tokens[$i])) { | |
| 765 | $test['zone'] = $tokens[++$i]; | |
| 766 | } | |
| 767 | else if (!is_array($tokens[$i]) && preg_match('/^:originalzone$/i', $tokens[$i])) { | |
| 768 | $test['originalzone'] = true; | |
| 769 | } | |
| 770 | } | |
| 771 | ||
| 772 | $tests[] = $test; | |
| 48e9c1 | 773 | break; |
| T | 774 | |
| e499a1 | 775 | case 'duplicate': |
| AM | 776 | $test = array('test' => $token, 'not' => $not); |
| 777 | ||
| 778 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| 779 | if (!is_array($tokens[$i])) { | |
| 780 | if (preg_match('/^:(handle|header|uniqueid|seconds)$/i', $tokens[$i], $m)) { | |
| 781 | $test[strtolower($m[1])] = $tokens[++$i]; | |
| 782 | } | |
| 783 | else if (preg_match('/^:last$/i', $tokens[$i])) { | |
| 784 | $test['last'] = true; | |
| 785 | } | |
| 786 | } | |
| 787 | } | |
| 788 | ||
| 789 | $tests[] = $test; | |
| 790 | break; | |
| 791 | ||
| 48e9c1 | 792 | case 'exists': |
| 6af79f | 793 | $tests[] = array('test' => 'exists', 'not' => $not, |
| 48e9c1 | 794 | 'arg' => array_pop($tokens)); |
| T | 795 | break; |
| 796 | ||
| 797 | case 'true': | |
| 6af79f | 798 | $tests[] = array('test' => 'true', 'not' => $not); |
| 48e9c1 | 799 | break; |
| T | 800 | |
| 801 | case 'false': | |
| 6af79f | 802 | $tests[] = array('test' => 'true', 'not' => !$not); |
| 48e9c1 | 803 | break; |
| T | 804 | } |
| 805 | ||
| 806 | // goto actions... | |
| 807 | if ($separator == '{') { | |
| 808 | break; | |
| 809 | } | |
| 810 | } | |
| 811 | ||
| 812 | // ...and actions block | |
| 80cb36 | 813 | $actions = $this->_parse_actions($content, $position); |
| 48e9c1 | 814 | |
| T | 815 | if ($tests && $actions) { |
| 816 | $result = array( | |
| 817 | 'type' => $cond, | |
| 818 | 'tests' => $tests, | |
| 819 | 'actions' => $actions, | |
| 820 | 'join' => $join, | |
| 821 | 'disabled' => $disabled, | |
| 822 | ); | |
| 823 | } | |
| 824 | ||
| 825 | return $result; | |
| 826 | } | |
| 827 | ||
| 828 | /** | |
| 829 | * Parse body of actions section | |
| 830 | * | |
| 80cb36 | 831 | * @param string $content The whole script content |
| AM | 832 | * @param int &$position Start position in the script |
| 833 | * @param string $end End of text separator | |
| 48e9c1 | 834 | * |
| T | 835 | * @return array Array of parsed action type/target pairs |
| 836 | */ | |
| 80cb36 | 837 | private function _parse_actions($content, &$position, $end = '}') |
| 48e9c1 | 838 | { |
| T | 839 | $result = null; |
| 80cb36 | 840 | $length = strlen($content); |
| 48e9c1 | 841 | |
| 80cb36 | 842 | while ($position < $length) { |
| AM | 843 | $tokens = self::tokenize($content, true, $position); |
| 48e9c1 | 844 | $separator = array_pop($tokens); |
| ed837a | 845 | $token = !empty($tokens) ? array_shift($tokens) : $separator; |
| 48e9c1 | 846 | |
| T | 847 | switch ($token) { |
| 848 | case 'discard': | |
| 849 | case 'keep': | |
| 850 | case 'stop': | |
| 851 | $result[] = array('type' => $token); | |
| 852 | break; | |
| 853 | ||
| 854 | case 'fileinto': | |
| 855 | case 'redirect': | |
| ed837a | 856 | $action = array('type' => $token, 'target' => array_pop($tokens)); |
| AM | 857 | $args = array('copy'); |
| 858 | $action += $this->action_arguments($tokens, $args); | |
| 48e9c1 | 859 | |
| ed837a | 860 | $result[] = $action; |
| AM | 861 | break; |
| 48e9c1 | 862 | |
| ed837a | 863 | case 'vacation': |
| AM | 864 | $action = array('type' => 'vacation', 'reason' => array_pop($tokens)); |
| 865 | $args = array('mime'); | |
| 866 | $vargs = array('seconds', 'days', 'addresses', 'subject', 'handle', 'from'); | |
| 867 | $action += $this->action_arguments($tokens, $args, $vargs); | |
| 868 | ||
| 869 | $result[] = $action; | |
| 48e9c1 | 870 | break; |
| T | 871 | |
| 872 | case 'reject': | |
| 873 | case 'ereject': | |
| 874 | case 'setflag': | |
| 875 | case 'addflag': | |
| 876 | case 'removeflag': | |
| ed837a | 877 | $result[] = array('type' => $token, 'target' => array_pop($tokens)); |
| 48e9c1 | 878 | break; |
| T | 879 | |
| 880 | case 'include': | |
| ed837a | 881 | $action = array('type' => 'include', 'target' => array_pop($tokens)); |
| AM | 882 | $args = array('once', 'optional', 'global', 'personal'); |
| 883 | $action += $this->action_arguments($tokens, $args); | |
| 48e9c1 | 884 | |
| ed837a | 885 | $result[] = $action; |
| 48e9c1 | 886 | break; |
| T | 887 | |
| 888 | case 'set': | |
| ed837a | 889 | $action = array('type' => 'set', 'value' => array_pop($tokens), 'name' => array_pop($tokens)); |
| AM | 890 | $args = array('lower', 'upper', 'lowerfirst', 'upperfirst', 'quotewildcard', 'length'); |
| 891 | $action += $this->action_arguments($tokens, $args); | |
| 48e9c1 | 892 | |
| ed837a | 893 | $result[] = $action; |
| 48e9c1 | 894 | break; |
| T | 895 | |
| 896 | case 'require': | |
| 897 | // skip, will be build according to used commands | |
| ed837a | 898 | // $result[] = array('type' => 'require', 'target' => array_pop($tokens)); |
| 48e9c1 | 899 | break; |
| T | 900 | |
| 2ac00a | 901 | case 'notify': |
| ed837a | 902 | $action = array('type' => 'notify'); |
| AM | 903 | $priorities = array('high' => 1, 'normal' => 2, 'low' => 3); |
| f22de5 | 904 | $vargs = array('from', 'id', 'importance', 'options', 'message', 'method'); |
| ed837a | 905 | $args = array_keys($priorities); |
| AM | 906 | $action += $this->action_arguments($tokens, $args, $vargs); |
| 2ac00a | 907 | |
| f22de5 | 908 | // Here we'll convert draft-martin-sieve-notify-01 into RFC 5435 |
| ed837a | 909 | if (!isset($action['importance'])) { |
| AM | 910 | foreach ($priorities as $key => $val) { |
| 911 | if (isset($action[$key])) { | |
| 912 | $action['importance'] = $val; | |
| 913 | unset($action[$key]); | |
| cd97f0 | 914 | } |
| 2ac00a | 915 | } |
| PS | 916 | } |
| eb1ee0 | 917 | |
| f22de5 | 918 | $action['options'] = (array) $action['options']; |
| ed837a | 919 | |
| f22de5 | 920 | // Old-draft way: :method "mailto" :options "email@address" |
| AM | 921 | if (!empty($action['method']) && !empty($action['options'])) { |
| 922 | $action['method'] .= ':' . array_shift($action['options']); | |
| 923 | } | |
| 924 | // unnamed parameter is a :method in enotify extension | |
| 925 | else if (!isset($action['method'])) { | |
| 926 | $action['method'] = array_pop($tokens); | |
| 2ac00a | 927 | } |
| PS | 928 | |
| ed837a | 929 | $result[] = $action; |
| 2ac00a | 930 | break; |
| 48e9c1 | 931 | } |
| T | 932 | |
| 933 | if ($separator == $end) | |
| 934 | break; | |
| 935 | } | |
| 936 | ||
| 937 | return $result; | |
| 938 | } | |
| 939 | ||
| 940 | /** | |
| ed837a | 941 | * Add comparator to the test |
| 48e9c1 | 942 | */ |
| T | 943 | private function add_comparator($test, &$out, &$exts) |
| 944 | { | |
| 945 | if (empty($test['comparator'])) { | |
| 946 | return; | |
| 947 | } | |
| 948 | ||
| 949 | if ($test['comparator'] == 'i;ascii-numeric') { | |
| 950 | array_push($exts, 'relational'); | |
| 951 | array_push($exts, 'comparator-i;ascii-numeric'); | |
| 952 | } | |
| 953 | else if (!in_array($test['comparator'], array('i;octet', 'i;ascii-casemap'))) { | |
| 954 | array_push($exts, 'comparator-' . $test['comparator']); | |
| 955 | } | |
| 956 | ||
| 957 | // skip default comparator | |
| 958 | if ($test['comparator'] != 'i;ascii-casemap') { | |
| 959 | $out .= ' :comparator ' . self::escape_string($test['comparator']); | |
| 960 | } | |
| 961 | } | |
| 962 | ||
| 963 | /** | |
| ed837a | 964 | * Add index argument to the test |
| AM | 965 | */ |
| 966 | private function add_index($test, &$out, &$exts) | |
| 967 | { | |
| 968 | if (!empty($test['index'])) { | |
| 969 | array_push($exts, 'index'); | |
| 970 | $out .= ' :index ' . intval($test['index']) . ($test['last'] ? ' :last' : ''); | |
| 971 | } | |
| 972 | } | |
| 973 | ||
| 974 | /** | |
| 975 | * Add operators to the test | |
| 976 | */ | |
| 977 | private function add_operator($test, &$out, &$exts) | |
| 978 | { | |
| 979 | if (empty($test['type'])) { | |
| 980 | return; | |
| 981 | } | |
| 982 | ||
| 889c76 | 983 | // relational operator |
| ed837a | 984 | if (preg_match('/^(value|count)-([gteqnl]{2})/', $test['type'], $m)) { |
| AM | 985 | array_push($exts, 'relational'); |
| 986 | ||
| 439fd7 | 987 | $out .= ' :' . $m[1] . ' "' . $m[2] . '"'; |
| ed837a | 988 | } |
| AM | 989 | else { |
| 990 | if ($test['type'] == 'regex') { | |
| 991 | array_push($exts, 'regex'); | |
| 992 | } | |
| 993 | ||
| 994 | $out .= ' :' . $test['type']; | |
| 995 | } | |
| 889c76 | 996 | |
| AM | 997 | $this->add_comparator($test, $out, $exts); |
| ed837a | 998 | } |
| AM | 999 | |
| 1000 | /** | |
| 1001 | * Extract test tokens | |
| 1002 | */ | |
| 1003 | private function test_tokens(&$tokens) | |
| 1004 | { | |
| 1005 | $test = array(); | |
| 1006 | $result = array(); | |
| 1007 | ||
| 1008 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| e499a1 | 1009 | if (!is_array($tokens[$i]) && preg_match('/^:comparator$/i', $tokens[$i])) { |
| AM | 1010 | $test['comparator'] = $tokens[++$i]; |
| 302aff | 1011 | } |
| e499a1 | 1012 | else if (!is_array($tokens[$i]) && preg_match('/^:(count|value)$/i', $tokens[$i])) { |
| AM | 1013 | $test['type'] = strtolower(substr($tokens[$i], 1)) . '-' . $tokens[++$i]; |
| 1014 | } | |
| 1015 | else if (!is_array($tokens[$i]) && preg_match('/^:(is|contains|matches|regex)$/i', $tokens[$i])) { | |
| 1016 | $test['type'] = strtolower(substr($tokens[$i], 1)); | |
| 1017 | } | |
| 1018 | else if (!is_array($tokens[$i]) && preg_match('/^:index$/i', $tokens[$i])) { | |
| 1019 | $test['index'] = intval($tokens[++$i]); | |
| 1020 | if ($tokens[$i+1] && preg_match('/^:last$/i', $tokens[$i+1])) { | |
| 1021 | $test['last'] = true; | |
| 1022 | $i++; | |
| 1023 | } | |
| 1024 | } | |
| 1025 | else { | |
| 1026 | $result[] = $tokens[$i]; | |
| 1027 | } | |
| ed837a | 1028 | } |
| AM | 1029 | |
| 1030 | $tokens = $result; | |
| 1031 | ||
| 1032 | return $test; | |
| 1033 | } | |
| 1034 | ||
| 1035 | /** | |
| 1036 | * Extract action arguments | |
| 1037 | */ | |
| 1038 | private function action_arguments(&$tokens, $bool_args, $val_args = array()) | |
| 1039 | { | |
| 1040 | $action = array(); | |
| 1041 | $result = array(); | |
| 1042 | ||
| 1043 | for ($i=0, $len=count($tokens); $i<$len; $i++) { | |
| 1044 | $tok = $tokens[$i]; | |
| 1045 | if (!is_array($tok) && $tok[0] == ':') { | |
| 1046 | $tok = strtolower(substr($tok, 1)); | |
| 1047 | if (in_array($tok, $bool_args)) { | |
| 1048 | $action[$tok] = true; | |
| 1049 | } | |
| 1050 | else if (in_array($tok, $val_args)) { | |
| 1051 | $action[$tok] = $tokens[++$i]; | |
| 1052 | } | |
| 1053 | else { | |
| 1054 | $result[] = $tok; | |
| 1055 | } | |
| 1056 | } | |
| 1057 | else { | |
| 1058 | $result[] = $tok; | |
| 1059 | } | |
| 1060 | } | |
| 1061 | ||
| 1062 | $tokens = $result; | |
| 1063 | ||
| 1064 | return $action; | |
| 1065 | } | |
| 1066 | ||
| 1067 | /** | |
| 48e9c1 | 1068 | * Escape special chars into quoted string value or multi-line string |
| T | 1069 | * or list of strings |
| 1070 | * | |
| 1071 | * @param string $str Text or array (list) of strings | |
| 1072 | * | |
| 1073 | * @return string Result text | |
| 1074 | */ | |
| 1075 | static function escape_string($str) | |
| 1076 | { | |
| 1077 | if (is_array($str) && count($str) > 1) { | |
| 1078 | foreach($str as $idx => $val) | |
| 1079 | $str[$idx] = self::escape_string($val); | |
| 1080 | ||
| 1081 | return '[' . implode(',', $str) . ']'; | |
| 1082 | } | |
| 1083 | else if (is_array($str)) { | |
| 1084 | $str = array_pop($str); | |
| 1085 | } | |
| 1086 | ||
| 1087 | // multi-line string | |
| 1088 | if (preg_match('/[\r\n\0]/', $str) || strlen($str) > 1024) { | |
| 1089 | return sprintf("text:\n%s\n.\n", self::escape_multiline_string($str)); | |
| 1090 | } | |
| 1091 | // quoted-string | |
| 1092 | else { | |
| 1093 | return '"' . addcslashes($str, '\\"') . '"'; | |
| 1094 | } | |
| 1095 | } | |
| 1096 | ||
| 1097 | /** | |
| 1098 | * Escape special chars in multi-line string value | |
| 1099 | * | |
| 1100 | * @param string $str Text | |
| 1101 | * | |
| 1102 | * @return string Text | |
| 1103 | */ | |
| 1104 | static function escape_multiline_string($str) | |
| 1105 | { | |
| 1106 | $str = preg_split('/(\r?\n)/', $str, -1, PREG_SPLIT_DELIM_CAPTURE); | |
| 1107 | ||
| 1108 | foreach ($str as $idx => $line) { | |
| 1109 | // dot-stuffing | |
| 1110 | if (isset($line[0]) && $line[0] == '.') { | |
| 1111 | $str[$idx] = '.' . $line; | |
| 1112 | } | |
| 1113 | } | |
| 1114 | ||
| 1115 | return implode($str); | |
| 1116 | } | |
| 1117 | ||
| 1118 | /** | |
| 1119 | * Splits script into string tokens | |
| 1120 | * | |
| 80cb36 | 1121 | * @param string $str The script |
| AM | 1122 | * @param mixed $num Number of tokens to return, 0 for all |
| 1123 | * or True for all tokens until separator is found. | |
| 1124 | * Separator will be returned as last token. | |
| 1125 | * @param int &$position Parsing start position | |
| 48e9c1 | 1126 | * |
| T | 1127 | * @return mixed Tokens array or string if $num=1 |
| 1128 | */ | |
| 80cb36 | 1129 | static function tokenize($str, $num = 0, &$position = 0) |
| 48e9c1 | 1130 | { |
| T | 1131 | $result = array(); |
| 80cb36 | 1132 | $length = strlen($str); |
| 48e9c1 | 1133 | |
| T | 1134 | // remove spaces from the beginning of the string |
| 80cb36 | 1135 | while ($position < $length && (!$num || $num === true || count($result) < $num)) { |
| AM | 1136 | // skip whitespace chars |
| 1137 | $position = self::ltrim_position($str, $position); | |
| 1138 | ||
| 1139 | switch ($str[$position]) { | |
| 48e9c1 | 1140 | |
| T | 1141 | // Quoted string |
| 1142 | case '"': | |
| 80cb36 | 1143 | for ($pos = $position + 1; $pos < $length; $pos++) { |
| 48e9c1 | 1144 | if ($str[$pos] == '"') { |
| T | 1145 | break; |
| 1146 | } | |
| 1147 | if ($str[$pos] == "\\") { | |
| 1148 | if ($str[$pos + 1] == '"' || $str[$pos + 1] == "\\") { | |
| 1149 | $pos++; | |
| 1150 | } | |
| 1151 | } | |
| 1152 | } | |
| 1153 | if ($str[$pos] != '"') { | |
| 1154 | // error | |
| 1155 | } | |
| 80cb36 | 1156 | |
| 48e9c1 | 1157 | // we need to strip slashes for a quoted string |
| 80cb36 | 1158 | $result[] = stripslashes(substr($str, $position + 1, $pos - $position - 1)); |
| AM | 1159 | $position = $pos + 1; |
| 48e9c1 | 1160 | break; |
| T | 1161 | |
| 1162 | // Parenthesized list | |
| 1163 | case '[': | |
| 80cb36 | 1164 | $position++; |
| AM | 1165 | $result[] = self::tokenize($str, 0, $position); |
| 48e9c1 | 1166 | break; |
| T | 1167 | case ']': |
| 80cb36 | 1168 | $position++; |
| 48e9c1 | 1169 | return $result; |
| T | 1170 | break; |
| 1171 | ||
| 1172 | // list/test separator | |
| 1173 | case ',': | |
| 1174 | // command separator | |
| 1175 | case ';': | |
| 1176 | // block/tests-list | |
| 1177 | case '(': | |
| 1178 | case ')': | |
| 1179 | case '{': | |
| 1180 | case '}': | |
| 80cb36 | 1181 | $sep = $str[$position]; |
| AM | 1182 | $position++; |
| 48e9c1 | 1183 | if ($num === true) { |
| T | 1184 | $result[] = $sep; |
| 1185 | break 2; | |
| 1186 | } | |
| 1187 | break; | |
| 1188 | ||
| 1189 | // bracket-comment | |
| 1190 | case '/': | |
| 80cb36 | 1191 | if ($str[$position + 1] == '*') { |
| AM | 1192 | if ($end_pos = strpos($str, '*/', $position + 2)) { |
| 1193 | $position = $end_pos + 2; | |
| 48e9c1 | 1194 | } |
| T | 1195 | else { |
| 1196 | // error | |
| 80cb36 | 1197 | $position = $length; |
| 48e9c1 | 1198 | } |
| T | 1199 | } |
| 1200 | break; | |
| 1201 | ||
| 1202 | // hash-comment | |
| 1203 | case '#': | |
| 80cb36 | 1204 | if ($lf_pos = strpos($str, "\n", $position)) { |
| AM | 1205 | $position = $lf_pos + 1; |
| 48e9c1 | 1206 | break; |
| T | 1207 | } |
| 1208 | else { | |
| 80cb36 | 1209 | $position = $length; |
| 48e9c1 | 1210 | } |
| T | 1211 | |
| 1212 | // String atom | |
| 1213 | default: | |
| 1214 | // empty or one character | |
| 80cb36 | 1215 | if ($position == $length) { |
| 48e9c1 | 1216 | break 2; |
| T | 1217 | } |
| e499a1 | 1218 | if ($length - $position < 2) { |
| AM | 1219 | $result[] = substr($str, $position); |
| 1220 | $position = $length; | |
| 1221 | break; | |
| 1222 | } | |
| 48e9c1 | 1223 | |
| T | 1224 | // tag/identifier/number |
| e499a1 | 1225 | if (preg_match('/[a-zA-Z0-9:_]+/', $str, $m, PREG_OFFSET_CAPTURE, $position) |
| AM | 1226 | && $m[0][1] == $position |
| 1227 | ) { | |
| 1228 | $atom = $m[0][0]; | |
| 1229 | $position += strlen($atom); | |
| 48e9c1 | 1230 | |
| 80cb36 | 1231 | if ($atom != 'text:') { |
| AM | 1232 | $result[] = $atom; |
| 48e9c1 | 1233 | } |
| T | 1234 | // multiline string |
| 1235 | else { | |
| 80cb36 | 1236 | // skip whitespace chars (except \r\n) |
| AM | 1237 | $position = self::ltrim_position($str, $position, false); |
| 1238 | ||
| 48e9c1 | 1239 | // possible hash-comment after "text:" |
| 80cb36 | 1240 | if ($str[$position] === '#') { |
| AM | 1241 | $endl = strpos($str, "\n", $position); |
| 1242 | $position = $endl ?: $length; | |
| 48e9c1 | 1243 | } |
| T | 1244 | |
| 80cb36 | 1245 | // skip \n or \r\n |
| AM | 1246 | if ($str[$position] == "\n") { |
| 1247 | $position++; | |
| 1248 | } | |
| 1249 | else if ($str[$position] == "\r" && $str[$position] == "\n") { | |
| 1250 | $position += 2; | |
| 1251 | } | |
| 1252 | ||
| 1253 | $text = ''; | |
| 1254 | ||
| 1255 | // get text until alone dot in a line | |
| 1256 | while ($position < $length) { | |
| 1257 | $pos = strpos($str, "\n.", $position); | |
| 1258 | if ($pos === false) { | |
| 1259 | break; | |
| 1260 | } | |
| 1261 | ||
| 1262 | $text .= substr($str, $position, $pos - $position); | |
| 1263 | $position = $pos + 2; | |
| 1264 | ||
| 1265 | if ($str[$pos] == "\n" | |
| 1266 | || ($str[$pos] == "\r" && $str[$pos + 1] == "\n") | |
| 1267 | ) { | |
| 1268 | break; | |
| 1269 | } | |
| 1270 | } | |
| 1271 | ||
| 1272 | // remove dot-stuffing | |
| 1273 | $text = str_replace("\n..", "\n.", $text); | |
| 1274 | ||
| 48e9c1 | 1275 | $result[] = $text; |
| 80cb36 | 1276 | $position++; |
| 48e9c1 | 1277 | } |
| T | 1278 | } |
| 1279 | // fallback, skip one character as infinite loop prevention | |
| 1280 | else { | |
| 80cb36 | 1281 | $position++; |
| 48e9c1 | 1282 | } |
| T | 1283 | |
| 1284 | break; | |
| 1285 | } | |
| 1286 | } | |
| 1287 | ||
| 1288 | return $num === 1 ? (isset($result[0]) ? $result[0] : null) : $result; | |
| 1289 | } | |
| 1290 | ||
| 80cb36 | 1291 | /** |
| AM | 1292 | * Skip whitespace characters in a string from specified position. |
| 1293 | */ | |
| 1294 | static function ltrim_position($content, $position, $br = true) | |
| 1295 | { | |
| 1296 | $blanks = array("\t", "\0", "\x0B", " "); | |
| 1297 | ||
| 1298 | if ($br) { | |
| 1299 | $blanks[] = "\r"; | |
| 1300 | $blanks[] = "\n"; | |
| 1301 | } | |
| 1302 | ||
| 1303 | while (isset($content[$position]) && isset($content[$position + 1]) | |
| 1304 | && in_array($content[$position], $blanks, true) | |
| 1305 | ) { | |
| 1306 | $position++; | |
| 1307 | } | |
| 1308 | ||
| 1309 | return $position; | |
| 1310 | } | |
| 48e9c1 | 1311 | } |
