1<?php 2 3namespace dokuwiki\ChangeLog; 4 5/** 6 * ChangeLog Prototype; methods for handling changelog 7 */ 8abstract class ChangeLog 9{ 10 use ChangeLogTrait; 11 12 /** @var string */ 13 protected $id; 14 /** @var int */ 15 protected $currentRevision; 16 /** @var array */ 17 protected $cache; 18 19 /** 20 * Constructor 21 * 22 * @param string $id page id 23 * @param int $chunk_size maximum block size read from file 24 */ 25 public function __construct($id, $chunk_size = 8192) 26 { 27 global $cache_revinfo; 28 29 $this->cache =& $cache_revinfo; 30 if (!isset($this->cache[$id])) { 31 $this->cache[$id] = array(); 32 } 33 34 $this->id = $id; 35 $this->setChunkSize($chunk_size); 36 // set property currentRevision and cache prior to getRevisionInfo($currentRev) call 37 $this->getCurrentRevisionInfo(); 38 } 39 40 /** 41 * Returns path to current page/media 42 * 43 * @return string path to file 44 */ 45 abstract protected function getFilename(); 46 47 /** 48 * Check whether given revision is the current page 49 * 50 * @param int $rev timestamp of current page 51 * @return bool true if $rev is current revision, otherwise false 52 */ 53 public function isCurrentRevision($rev) 54 { 55 return $rev == $this->currentRevision(); 56 } 57 58 /** 59 * Checks if the revision is last revision 60 * 61 * @param int $rev revision timestamp 62 * @return bool true if $rev is last revision, otherwise false 63 */ 64 public function isLastRevision($rev = null) 65 { 66 return $rev === $this->lastRevision(); 67 } 68 69 /** 70 * Return the current revision identifer 71 * 72 * The "current" revision means current version of the page or media file. It is either 73 * identical with or newer than the "last" revision, that depends on whether the file 74 * has modified, created or deleted outside of DokuWiki. 75 * The value of identifier can be determined by timestamp as far as the file exists, 76 * otherwise it must be assigned larger than any other revisions to keep them sortable. 77 * 78 * @return int|false revision timestamp 79 */ 80 public function currentRevision() 81 { 82 if (!isset($this->currentRevision)) { 83 // set ChangeLog::currentRevision property 84 $this->getCurrentRevisionInfo(); 85 } 86 return $this->currentRevision; 87 } 88 89 /** 90 * Return the last revision identifer, date value of the last entry of the changelog 91 * 92 * @return int|false revision timestamp 93 */ 94 public function lastRevision() 95 { 96 $revs = $this->getRevisions(-1, 1); 97 return empty($revs) ? false : $revs[0]; 98 } 99 100 /** 101 * Save revision info to the cache pool 102 * 103 * @param array $info Revision info structure 104 * @return bool 105 */ 106 protected function cacheRevisionInfo($info) 107 { 108 if (!is_array($info)) return false; 109 //$this->cache[$this->id][$info['date']] ??= $info; // since php 7.4 110 $this->cache[$this->id][$info['date']] = $this->cache[$this->id][$info['date']] ?? $info; 111 return true; 112 } 113 114 /** 115 * Get the changelog information for a specific revision (timestamp) 116 * 117 * Adjacent changelog lines are optimistically parsed and cached to speed up 118 * consecutive calls to getRevisionInfo. For large changelog files, only the chunk 119 * containing the requested changelog line is read. 120 * 121 * @param int $rev revision timestamp 122 * @return bool|array false or array with entries: 123 * - date: unix timestamp 124 * - ip: IPv4 address (127.0.0.1) 125 * - type: log line type 126 * - id: page id 127 * - user: user name 128 * - sum: edit summary (or action reason) 129 * - extra: extra data (varies by line type) 130 * - sizechange: change of filesize 131 * 132 * @author Ben Coburn <btcoburn@silicodon.net> 133 * @author Kate Arzamastseva <pshns@ukr.net> 134 */ 135 public function getRevisionInfo($rev) 136 { 137 $rev = max(0, $rev); 138 if (!$rev) return false; 139 140 // check if it's already in the memory cache 141 if (isset($this->cache[$this->id]) && isset($this->cache[$this->id][$rev])) { 142 return $this->cache[$this->id][$rev]; 143 } 144 145 //read lines from changelog 146 list($fp, $lines) = $this->readloglines($rev); 147 if ($fp) { 148 fclose($fp); 149 } 150 if (empty($lines)) return false; 151 152 // parse and cache changelog lines 153 foreach ($lines as $value) { 154 $info = $this->parseLogLine($value); 155 $this->cacheRevisionInfo($info); 156 } 157 if (!isset($this->cache[$this->id][$rev])) { 158 return false; 159 } 160 return $this->cache[$this->id][$rev]; 161 } 162 163 /** 164 * Return a list of page revisions numbers 165 * 166 * Does not guarantee that the revision exists in the attic, 167 * only that a line with the date exists in the changelog. 168 * By default the current revision is skipped. 169 * 170 * The current revision is automatically skipped when the page exists. 171 * See $INFO['meta']['last_change'] for the current revision. 172 * A negative $first let read the current revision too. 173 * 174 * For efficiency, the log lines are parsed and cached for later 175 * calls to getRevisionInfo. Large changelog files are read 176 * backwards in chunks until the requested number of changelog 177 * lines are recieved. 178 * 179 * @param int $first skip the first n changelog lines 180 * @param int $num number of revisions to return 181 * @return array with the revision timestamps 182 * 183 * @author Ben Coburn <btcoburn@silicodon.net> 184 * @author Kate Arzamastseva <pshns@ukr.net> 185 */ 186 public function getRevisions($first, $num) 187 { 188 $revs = array(); 189 $lines = array(); 190 $count = 0; 191 192 $logfile = $this->getChangelogFilename(); 193 if (!file_exists($logfile)) return $revs; 194 195 $num = max($num, 0); 196 if ($num == 0) { 197 return $revs; 198 } 199 200 if ($first < 0) { 201 $first = 0; 202 } else { 203 $fileLastMod = $this->getFilename(); 204 if (file_exists($fileLastMod) && $this->isLastRevision(filemtime($fileLastMod))) { 205 // skip last revision if the page exists 206 $first = max($first + 1, 0); 207 } 208 } 209 210 if (filesize($logfile) < $this->chunk_size || $this->chunk_size == 0) { 211 // read whole file 212 $lines = file($logfile); 213 if ($lines === false) { 214 return $revs; 215 } 216 } else { 217 // read chunks backwards 218 $fp = fopen($logfile, 'rb'); // "file pointer" 219 if ($fp === false) { 220 return $revs; 221 } 222 fseek($fp, 0, SEEK_END); 223 $tail = ftell($fp); 224 225 // chunk backwards 226 $finger = max($tail - $this->chunk_size, 0); 227 while ($count < $num + $first) { 228 $nl = $this->getNewlinepointer($fp, $finger); 229 230 // was the chunk big enough? if not, take another bite 231 if ($nl > 0 && $tail <= $nl) { 232 $finger = max($finger - $this->chunk_size, 0); 233 continue; 234 } else { 235 $finger = $nl; 236 } 237 238 // read chunk 239 $chunk = ''; 240 $read_size = max($tail - $finger, 0); // found chunk size 241 $got = 0; 242 while ($got < $read_size && !feof($fp)) { 243 $tmp = @fread($fp, max(min($this->chunk_size, $read_size - $got), 0)); 244 if ($tmp === false) { 245 break; 246 } //error state 247 $got += strlen($tmp); 248 $chunk .= $tmp; 249 } 250 $tmp = explode("\n", $chunk); 251 array_pop($tmp); // remove trailing newline 252 253 // combine with previous chunk 254 $count += count($tmp); 255 $lines = array_merge($tmp, $lines); 256 257 // next chunk 258 if ($finger == 0) { 259 break; 260 } else { // already read all the lines 261 $tail = $finger; 262 $finger = max($tail - $this->chunk_size, 0); 263 } 264 } 265 fclose($fp); 266 } 267 268 // skip parsing extra lines 269 $num = max(min(count($lines) - $first, $num), 0); 270 if ($first > 0 && $num > 0) { 271 $lines = array_slice($lines, max(count($lines) - $first - $num, 0), $num); 272 } elseif ($first > 0 && $num == 0) { 273 $lines = array_slice($lines, 0, max(count($lines) - $first, 0)); 274 } elseif ($first == 0 && $num > 0) { 275 $lines = array_slice($lines, max(count($lines) - $num, 0)); 276 } 277 278 // handle lines in reverse order 279 for ($i = count($lines) - 1; $i >= 0; $i--) { 280 $info = $this->parseLogLine($lines[$i]); 281 if ($this->cacheRevisionInfo($info)) { 282 $revs[] = $info['date']; 283 } 284 } 285 286 return $revs; 287 } 288 289 /** 290 * Get the nth revision left or right handside for a specific page id and revision (timestamp) 291 * 292 * For large changelog files, only the chunk containing the 293 * reference revision $rev is read and sometimes a next chunck. 294 * 295 * Adjacent changelog lines are optimistically parsed and cached to speed up 296 * consecutive calls to getRevisionInfo. 297 * 298 * @param int $rev revision timestamp used as startdate 299 * (doesn't need to be exact revision number) 300 * @param int $direction give position of returned revision with respect to $rev; 301 positive=next, negative=prev 302 * @return bool|int 303 * timestamp of the requested revision 304 * otherwise false 305 */ 306 public function getRelativeRevision($rev, $direction) 307 { 308 $rev = max($rev, 0); 309 $direction = (int)$direction; 310 311 //no direction given or last rev, so no follow-up 312 if (!$direction || ($direction > 0 && $this->isCurrentRevision($rev))) { 313 return false; 314 } 315 316 //get lines from changelog 317 list($fp, $lines, $head, $tail, $eof) = $this->readloglines($rev); 318 if (empty($lines)) return false; 319 320 // look for revisions later/earlier than $rev, when founded count till the wanted revision is reached 321 // also parse and cache changelog lines for getRevisionInfo(). 322 $revcounter = 0; 323 $relativerev = false; 324 $checkotherchunck = true; //always runs once 325 while (!$relativerev && $checkotherchunck) { 326 $info = array(); 327 //parse in normal or reverse order 328 $count = count($lines); 329 if ($direction > 0) { 330 $start = 0; 331 $step = 1; 332 } else { 333 $start = $count - 1; 334 $step = -1; 335 } 336 for ($i = $start; $i >= 0 && $i < $count; $i = $i + $step) { 337 $info = $this->parseLogLine($lines[$i]); 338 if ($this->cacheRevisionInfo($info)) { 339 //look for revs older/earlier then reference $rev and select $direction-th one 340 if (($direction > 0 && $info['date'] > $rev) || ($direction < 0 && $info['date'] < $rev)) { 341 $revcounter++; 342 if ($revcounter == abs($direction)) { 343 $relativerev = $info['date']; 344 } 345 } 346 } 347 } 348 349 //true when $rev is found, but not the wanted follow-up. 350 $checkotherchunck = $fp 351 && ($info['date'] == $rev || ($revcounter > 0 && !$relativerev)) 352 && !(($tail == $eof && $direction > 0) || ($head == 0 && $direction < 0)); 353 354 if ($checkotherchunck) { 355 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, $direction); 356 357 if (empty($lines)) break; 358 } 359 } 360 if ($fp) { 361 fclose($fp); 362 } 363 364 return $relativerev; 365 } 366 367 /** 368 * Returns revisions around rev1 and rev2 369 * When available it returns $max entries for each revision 370 * 371 * @param int $rev1 oldest revision timestamp 372 * @param int $rev2 newest revision timestamp (0 looks up last revision) 373 * @param int $max maximum number of revisions returned 374 * @return array with two arrays with revisions surrounding rev1 respectively rev2 375 */ 376 public function getRevisionsAround($rev1, $rev2, $max = 50) 377 { 378 $max = intval(abs($max) / 2) * 2 + 1; 379 $rev1 = max($rev1, 0); 380 $rev2 = max($rev2, 0); 381 382 if ($rev2) { 383 if ($rev2 < $rev1) { 384 $rev = $rev2; 385 $rev2 = $rev1; 386 $rev1 = $rev; 387 } 388 } else { 389 //empty right side means a removed page. Look up last revision. 390 $rev2 = $this->currentRevision(); 391 } 392 //collect revisions around rev2 393 list($revs2, $allrevs, $fp, $lines, $head, $tail) = $this->retrieveRevisionsAround($rev2, $max); 394 395 if (empty($revs2)) return array(array(), array()); 396 397 //collect revisions around rev1 398 $index = array_search($rev1, $allrevs); 399 if ($index === false) { 400 //no overlapping revisions 401 list($revs1, , , , ,) = $this->retrieveRevisionsAround($rev1, $max); 402 if (empty($revs1)) $revs1 = array(); 403 } else { 404 //revisions overlaps, reuse revisions around rev2 405 $lastrev = array_pop($allrevs); //keep last entry that could be external edit 406 $revs1 = $allrevs; 407 while ($head > 0) { 408 for ($i = count($lines) - 1; $i >= 0; $i--) { 409 $info = $this->parseLogLine($lines[$i]); 410 if ($this->cacheRevisionInfo($info)) { 411 $revs1[] = $info['date']; 412 $index++; 413 414 if ($index > intval($max / 2)) break 2; 415 } 416 } 417 418 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, -1); 419 } 420 sort($revs1); 421 $revs1[] = $lastrev; //push back last entry 422 423 //return wanted selection 424 $revs1 = array_slice($revs1, max($index - intval($max / 2), 0), $max); 425 } 426 427 return array(array_reverse($revs1), array_reverse($revs2)); 428 } 429 430 /** 431 * Return an existing revision for a specific date which is 432 * the current one or younger or equal then the date 433 * 434 * @param number $date_at timestamp 435 * @return string revision ('' for current) 436 */ 437 public function getLastRevisionAt($date_at) 438 { 439 $fileLastMod = $this->getFilename(); 440 //requested date_at(timestamp) younger or equal then modified_time($this->id) => load current 441 if (file_exists($fileLastMod) && $date_at >= @filemtime($fileLastMod)) { 442 return ''; 443 } else { 444 if ($rev = $this->getRelativeRevision($date_at + 1, -1)) { //+1 to get also the requested date revision 445 return $rev; 446 } else { 447 return false; 448 } 449 } 450 } 451 452 /** 453 * Collect the $max revisions near to the timestamp $rev 454 * 455 * Ideally, half of retrieved timestamps are older than $rev, another half are newer. 456 * The returned array $requestedrevs may not contain the reference timestamp $rev 457 * when it does not match any revision value recorded in changelog. 458 * 459 * @param int $rev revision timestamp 460 * @param int $max maximum number of revisions to be returned 461 * @return bool|array 462 * return array with entries: 463 * - $requestedrevs: array of with $max revision timestamps 464 * - $revs: all parsed revision timestamps 465 * - $fp: filepointer only defined for chuck reading, needs closing. 466 * - $lines: non-parsed changelog lines before the parsed revisions 467 * - $head: position of first readed changelogline 468 * - $lasttail: position of end of last readed changelogline 469 * otherwise false 470 */ 471 protected function retrieveRevisionsAround($rev, $max) 472 { 473 $revs = array(); 474 $aftercount = $beforecount = 0; 475 476 //get lines from changelog 477 list($fp, $lines, $starthead, $starttail, $eof) = $this->readloglines($rev); 478 if (empty($lines)) return false; 479 480 //parse changelog lines in chunk, and read forward more chunks until $max/2 is reached 481 $head = $starthead; 482 $tail = $starttail; 483 while (count($lines) > 0) { 484 foreach ($lines as $line) { 485 $info = $this->parseLogLine($line); 486 if ($this->cacheRevisionInfo($info)) { 487 $revs[] = $info['date']; 488 if ($info['date'] >= $rev) { 489 //count revs after reference $rev 490 $aftercount++; 491 if ($aftercount == 1) $beforecount = count($revs); 492 } 493 //enough revs after reference $rev? 494 if ($aftercount > intval($max / 2)) break 2; 495 } 496 } 497 //retrieve next chunk 498 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, 1); 499 } 500 $lasttail = $tail; 501 502 // add a possible revision of external edit, create or deletion 503 if ($lasttail == $eof && $aftercount <= intval($max / 2) && 504 count($revs) && !$this->isCurrentRevision($revs[count($revs)-1]) 505 ) { 506 $revs[] = $this->currentRevision; 507 $aftercount++; 508 } 509 510 if ($aftercount == 0) { 511 //given timestamp $rev is newer than the most recent line in chunk 512 return false; //FIXME: or proceed to collect older revisions? 513 } 514 515 //read more chunks backward until $max/2 is reached and total number of revs is equal to $max 516 $lines = array(); 517 $i = 0; 518 if ($aftercount > 0) { 519 $head = $starthead; 520 $tail = $starttail; 521 while ($head > 0) { 522 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, -1); 523 524 for ($i = count($lines) - 1; $i >= 0; $i--) { 525 $info = $this->parseLogLine($lines[$i]); 526 if ($this->cacheRevisionInfo($info)) { 527 $revs[] = $info['date']; 528 $beforecount++; 529 //enough revs before reference $rev? 530 if ($beforecount > max(intval($max / 2), $max - $aftercount)) break 2; 531 } 532 } 533 } 534 } 535 //keep only non-parsed lines 536 $lines = array_slice($lines, 0, $i); 537 538 sort($revs); 539 540 //trunk desired selection 541 $requestedrevs = array_slice($revs, -$max, $max); 542 543 return array($requestedrevs, $revs, $fp, $lines, $head, $lasttail); 544 } 545 546 /** 547 * Get the current revision information, considering external edit, create or deletion 548 * 549 * When the file has not modified since its last revision, the infomation of the last 550 * change that had already recorded in the changelog is returned as current change info. 551 * Otherwise, the change infomation since the last revision caused outside DokuWiki 552 * should be returned, which is referred as "external revision". 553 * 554 * The change date of the file can be determined by timestamp as far as the file exists, 555 * however this is not possible when the file has already deleted outside of DokuWiki. 556 * In such case we assign current time() value for the external deletion. The change 557 * date is used as revision identifier. 558 * As a result, the value of current revision identifier may change each time because: 559 * 1) the file has again modified outside of DokuWiki, or 560 * 2) the value is essentially volatile for deleted but once existed files. 561 * 562 * @return bool|array false when page had never existed or array with entries: 563 * - date: revision identifier (timestamp or last revision +1) 564 * - ip: IPv4 address (127.0.0.1) 565 * - type: log line type 566 * - id: id of page or media 567 * - user: user name 568 * - sum: edit summary (or action reason) 569 * - extra: extra data (varies by line type) 570 * - sizechange: change of filesize 571 * - timestamp: unix timestamp or false (key set only for external edit occurred) 572 * 573 * @author Satoshi Sahara <sahara.satoshi@gmail.com> 574 */ 575 public function getCurrentRevisionInfo() 576 { 577 global $lang; 578 579 if (isset($this->currentRevision)) return $this->getRevisionInfo($this->currentRevision); 580 581 // get revision id from the item file timestamp and chagelog 582 $fileLastMod = $this->getFilename(); 583 $fileRev = @filemtime($fileLastMod); // false when the file not exist 584 $lastRev = $this->lastRevision(); // false when no changelog 585 586 if (!$fileRev && !$lastRev) { // has never existed 587 $this->currentRevision = false; 588 return false; 589 } elseif ($fileRev === $lastRev) { // not external edit 590 $this->currentRevision = $lastRev; 591 return $this->getRevisionInfo($lastRev); 592 } 593 594 if (!$fileRev && $lastRev) { // item file does not exist 595 // check consistency against changelog 596 $revInfo = $this->getRevisionInfo($lastRev); 597 if ($revInfo['type'] == DOKU_CHANGE_TYPE_DELETE) { 598 $this->currentRevision = $lastRev; 599 return $this->getRevisionInfo($lastRev); 600 } 601 602 // externally deleted 603 $revInfo = [ 604 'date' => time(), // assign current time 605 'ip' => '127.0.0.1', 606 'type' => DOKU_CHANGE_TYPE_DELETE, 607 'id' => $this->id, 608 'user' => '', 609 'sum' => $lang['deleted'].' - '.$lang['external_edit'].' ('.$lang['unknowndate'].')', 610 'extra' => '', 611 'sizechange' => -io_getSizeFile($this->getFilename($lastRev)), 612 'timestamp' => false, 613 ]; 614 615 } elseif ($fileRev) { // item file exist 616 // here, file timestamp is different with last revision in changelog 617 $isJustCreated = $lastRev === false || ( 618 $fileRev > $lastRev && 619 $this->getRevisionInfo($lastRev)['type'] == DOKU_CHANGE_TYPE_DELETE 620 ); 621 $filesize_new = filesize($this->getFilename()); 622 $filesize_old = $isJustCreated ? 0 : io_getSizeFile($this->getFilename($lastRev)); 623 $sizechange = $filesize_new - $filesize_old; 624 625 if ($isJustCreated) { 626 $timestamp = $fileRev; 627 $sum = $lang['created'].' - '.$lang['external_edit']; 628 } elseif ($fileRev > $lastRev) { 629 $timestamp = $fileRev; 630 $sum = $lang['external_edit']; 631 } else { 632 // $fileRev is older than $lastRev, that is erroneous/incorrect occurence. 633 // try to change file modification time to the detection time 634 $timestamp = touch($fileLastMod) ? filemtime($fileLastMod) : false; 635 $sum = $lang['external_edit'].' ('.$lang['unknowndate'].')'; 636 } 637 638 // externally created or edited 639 $revInfo = [ 640 'date' => $timestamp ?: time(), 641 'ip' => '127.0.0.1', 642 'type' => $isJustCreated ? DOKU_CHANGE_TYPE_CREATE : DOKU_CHANGE_TYPE_EDIT, 643 'id' => $this->id, 644 'user' => '', 645 'sum' => $sum, 646 'extra' => '', 647 'sizechange' => $sizechange, 648 'timestamp' => $timestamp, 649 ]; 650 } 651 652 // cache current revision information of external edition 653 $this->currentRevision = $revInfo['date']; 654 $this->cache[$this->id][$this->currentRevision] = $revInfo; 655 return $this->getRevisionInfo($this->currentRevision); 656 } 657} 658