1<?php 2/** 3 * Changelog handling functions 4 * 5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6 * @author Andreas Gohr <andi@splitbrain.org> 7 */ 8 9// Constants for known core changelog line types. 10// Use these in place of string literals for more readable code. 11define('DOKU_CHANGE_TYPE_CREATE', 'C'); 12define('DOKU_CHANGE_TYPE_EDIT', 'E'); 13define('DOKU_CHANGE_TYPE_MINOR_EDIT', 'e'); 14define('DOKU_CHANGE_TYPE_DELETE', 'D'); 15define('DOKU_CHANGE_TYPE_REVERT', 'R'); 16 17/** 18 * parses a changelog line into it's components 19 * 20 * @author Ben Coburn <btcoburn@silicodon.net> 21 * 22 * @param string $line changelog line 23 * @return array|bool parsed line or false 24 */ 25function parseChangelogLine($line) { 26 $line = rtrim($line, "\n"); 27 $tmp = explode("\t", $line); 28 if ($tmp!==false && count($tmp)>1) { 29 $info = array(); 30 $info['date'] = (int)$tmp[0]; // unix timestamp 31 $info['ip'] = $tmp[1]; // IPv4 address (127.0.0.1) 32 $info['type'] = $tmp[2]; // log line type 33 $info['id'] = $tmp[3]; // page id 34 $info['user'] = $tmp[4]; // user name 35 $info['sum'] = $tmp[5]; // edit summary (or action reason) 36 $info['extra'] = $tmp[6]; // extra data (varies by line type) 37 if(isset($tmp[7]) && $tmp[7] !== '') { //last item has line-end|| 38 $info['sizechange'] = (int) $tmp[7]; 39 } else { 40 $info['sizechange'] = null; 41 } 42 return $info; 43 } else { 44 return false; 45 } 46} 47 48/** 49 * Add's an entry to the changelog and saves the metadata for the page 50 * 51 * @param int $date Timestamp of the change 52 * @param String $id Name of the affected page 53 * @param String $type Type of the change see DOKU_CHANGE_TYPE_* 54 * @param String $summary Summary of the change 55 * @param mixed $extra In case of a revert the revision (timestmp) of the reverted page 56 * @param array $flags Additional flags in a key value array. 57 * Available flags: 58 * - ExternalEdit - mark as an external edit. 59 * @param null|int $sizechange Change of filesize 60 * 61 * @author Andreas Gohr <andi@splitbrain.org> 62 * @author Esther Brunner <wikidesign@gmail.com> 63 * @author Ben Coburn <btcoburn@silicodon.net> 64 */ 65function addLogEntry($date, $id, $type=DOKU_CHANGE_TYPE_EDIT, $summary='', $extra='', $flags=null, $sizechange = null){ 66 global $conf, $INFO; 67 /** @var Input $INPUT */ 68 global $INPUT; 69 70 // check for special flags as keys 71 if (!is_array($flags)) { $flags = array(); } 72 $flagExternalEdit = isset($flags['ExternalEdit']); 73 74 $id = cleanid($id); 75 $file = wikiFN($id); 76 $created = @filectime($file); 77 $minor = ($type===DOKU_CHANGE_TYPE_MINOR_EDIT); 78 $wasRemoved = ($type===DOKU_CHANGE_TYPE_DELETE); 79 80 if(!$date) $date = time(); //use current time if none supplied 81 $remote = (!$flagExternalEdit)?clientIP(true):'127.0.0.1'; 82 $user = (!$flagExternalEdit)?$INPUT->server->str('REMOTE_USER'):''; 83 if($sizechange === null) { 84 $sizechange = ''; 85 } else { 86 $sizechange = (int) $sizechange; 87 } 88 89 $strip = array("\t", "\n"); 90 $logline = array( 91 'date' => $date, 92 'ip' => $remote, 93 'type' => str_replace($strip, '', $type), 94 'id' => $id, 95 'user' => $user, 96 'sum' => utf8_substr(str_replace($strip, '', $summary), 0, 255), 97 'extra' => str_replace($strip, '', $extra), 98 'sizechange' => $sizechange 99 ); 100 101 $wasCreated = ($type===DOKU_CHANGE_TYPE_CREATE); 102 $wasReverted = ($type===DOKU_CHANGE_TYPE_REVERT); 103 // update metadata 104 if (!$wasRemoved) { 105 $oldmeta = p_read_metadata($id); 106 $meta = array(); 107 if ($wasCreated && empty($oldmeta['persistent']['date']['created'])){ // newly created 108 $meta['date']['created'] = $created; 109 if ($user){ 110 $meta['creator'] = $INFO['userinfo']['name']; 111 $meta['user'] = $user; 112 } 113 } elseif (($wasCreated || $wasReverted) && !empty($oldmeta['persistent']['date']['created'])) { // re-created / restored 114 $meta['date']['created'] = $oldmeta['persistent']['date']['created']; 115 $meta['date']['modified'] = $created; // use the files ctime here 116 $meta['creator'] = $oldmeta['persistent']['creator']; 117 if ($user) $meta['contributor'][$user] = $INFO['userinfo']['name']; 118 } elseif (!$minor) { // non-minor modification 119 $meta['date']['modified'] = $date; 120 if ($user) $meta['contributor'][$user] = $INFO['userinfo']['name']; 121 } 122 $meta['last_change'] = $logline; 123 p_set_metadata($id, $meta); 124 } 125 126 // add changelog lines 127 $logline = implode("\t", $logline)."\n"; 128 io_saveFile(metaFN($id,'.changes'),$logline,true); //page changelog 129 io_saveFile($conf['changelog'],$logline,true); //global changelog cache 130} 131 132/** 133 * Add's an entry to the media changelog 134 * 135 * @author Michael Hamann <michael@content-space.de> 136 * @author Andreas Gohr <andi@splitbrain.org> 137 * @author Esther Brunner <wikidesign@gmail.com> 138 * @author Ben Coburn <btcoburn@silicodon.net> 139 * 140 * @param int $date Timestamp of the change 141 * @param String $id Name of the affected page 142 * @param String $type Type of the change see DOKU_CHANGE_TYPE_* 143 * @param String $summary Summary of the change 144 * @param mixed $extra In case of a revert the revision (timestmp) of the reverted page 145 * @param array $flags Additional flags in a key value array. 146 * Available flags: 147 * - (none, so far) 148 * @param null|int $sizechange Change of filesize 149 */ 150function addMediaLogEntry($date, $id, $type=DOKU_CHANGE_TYPE_EDIT, $summary='', $extra='', $flags=null, $sizechange = null){ 151 global $conf; 152 /** @var Input $INPUT */ 153 global $INPUT; 154 155 $id = cleanid($id); 156 157 if(!$date) $date = time(); //use current time if none supplied 158 $remote = clientIP(true); 159 $user = $INPUT->server->str('REMOTE_USER'); 160 if($sizechange === null) { 161 $sizechange = ''; 162 } else { 163 $sizechange = (int) $sizechange; 164 } 165 166 $strip = array("\t", "\n"); 167 $logline = array( 168 'date' => $date, 169 'ip' => $remote, 170 'type' => str_replace($strip, '', $type), 171 'id' => $id, 172 'user' => $user, 173 'sum' => utf8_substr(str_replace($strip, '', $summary), 0, 255), 174 'extra' => str_replace($strip, '', $extra), 175 'sizechange' => $sizechange 176 ); 177 178 // add changelog lines 179 $logline = implode("\t", $logline)."\n"; 180 io_saveFile($conf['media_changelog'],$logline,true); //global media changelog cache 181 io_saveFile(mediaMetaFN($id,'.changes'),$logline,true); //media file's changelog 182} 183 184/** 185 * returns an array of recently changed files using the 186 * changelog 187 * 188 * The following constants can be used to control which changes are 189 * included. Add them together as needed. 190 * 191 * RECENTS_SKIP_DELETED - don't include deleted pages 192 * RECENTS_SKIP_MINORS - don't include minor changes 193 * RECENTS_SKIP_SUBSPACES - don't include subspaces 194 * RECENTS_MEDIA_CHANGES - return media changes instead of page changes 195 * RECENTS_MEDIA_PAGES_MIXED - return both media changes and page changes 196 * 197 * @param int $first number of first entry returned (for paginating 198 * @param int $num return $num entries 199 * @param string $ns restrict to given namespace 200 * @param int $flags see above 201 * @return array recently changed files 202 * 203 * @author Ben Coburn <btcoburn@silicodon.net> 204 * @author Kate Arzamastseva <pshns@ukr.net> 205 */ 206function getRecents($first,$num,$ns='',$flags=0){ 207 global $conf; 208 $recent = array(); 209 $count = 0; 210 211 if(!$num) 212 return $recent; 213 214 // read all recent changes. (kept short) 215 if ($flags & RECENTS_MEDIA_CHANGES) { 216 $lines = @file($conf['media_changelog']); 217 } else { 218 $lines = @file($conf['changelog']); 219 } 220 $lines_position = count($lines)-1; 221 $media_lines_position = 0; 222 $media_lines = array(); 223 224 if ($flags & RECENTS_MEDIA_PAGES_MIXED) { 225 $media_lines = @file($conf['media_changelog']); 226 $media_lines_position = count($media_lines)-1; 227 } 228 229 $seen = array(); // caches seen lines, _handleRecent() skips them 230 231 // handle lines 232 while ($lines_position >= 0 || (($flags & RECENTS_MEDIA_PAGES_MIXED) && $media_lines_position >=0)) { 233 if (empty($rec) && $lines_position >= 0) { 234 $rec = _handleRecent(@$lines[$lines_position], $ns, $flags, $seen); 235 if (!$rec) { 236 $lines_position --; 237 continue; 238 } 239 } 240 if (($flags & RECENTS_MEDIA_PAGES_MIXED) && empty($media_rec) && $media_lines_position >= 0) { 241 $media_rec = _handleRecent(@$media_lines[$media_lines_position], $ns, $flags | RECENTS_MEDIA_CHANGES, $seen); 242 if (!$media_rec) { 243 $media_lines_position --; 244 continue; 245 } 246 } 247 if (($flags & RECENTS_MEDIA_PAGES_MIXED) && @$media_rec['date'] >= @$rec['date']) { 248 $media_lines_position--; 249 $x = $media_rec; 250 $x['media'] = true; 251 $media_rec = false; 252 } else { 253 $lines_position--; 254 $x = $rec; 255 if ($flags & RECENTS_MEDIA_CHANGES) $x['media'] = true; 256 $rec = false; 257 } 258 if(--$first >= 0) continue; // skip first entries 259 $recent[] = $x; 260 $count++; 261 // break when we have enough entries 262 if($count >= $num){ break; } 263 } 264 return $recent; 265} 266 267/** 268 * returns an array of files changed since a given time using the 269 * changelog 270 * 271 * The following constants can be used to control which changes are 272 * included. Add them together as needed. 273 * 274 * RECENTS_SKIP_DELETED - don't include deleted pages 275 * RECENTS_SKIP_MINORS - don't include minor changes 276 * RECENTS_SKIP_SUBSPACES - don't include subspaces 277 * RECENTS_MEDIA_CHANGES - return media changes instead of page changes 278 * 279 * @param int $from date of the oldest entry to return 280 * @param int $to date of the newest entry to return (for pagination, optional) 281 * @param string $ns restrict to given namespace (optional) 282 * @param int $flags see above (optional) 283 * @return array of files 284 * 285 * @author Michael Hamann <michael@content-space.de> 286 * @author Ben Coburn <btcoburn@silicodon.net> 287 */ 288function getRecentsSince($from,$to=null,$ns='',$flags=0){ 289 global $conf; 290 $recent = array(); 291 292 if($to && $to < $from) 293 return $recent; 294 295 // read all recent changes. (kept short) 296 if ($flags & RECENTS_MEDIA_CHANGES) { 297 $lines = @file($conf['media_changelog']); 298 } else { 299 $lines = @file($conf['changelog']); 300 } 301 if(!$lines) return $recent; 302 303 // we start searching at the end of the list 304 $lines = array_reverse($lines); 305 306 // handle lines 307 $seen = array(); // caches seen lines, _handleRecent() skips them 308 309 foreach($lines as $line){ 310 $rec = _handleRecent($line, $ns, $flags, $seen); 311 if($rec !== false) { 312 if ($rec['date'] >= $from) { 313 if (!$to || $rec['date'] <= $to) { 314 $recent[] = $rec; 315 } 316 } else { 317 break; 318 } 319 } 320 } 321 322 return array_reverse($recent); 323} 324 325/** 326 * Internal function used by getRecents 327 * 328 * don't call directly 329 * 330 * @see getRecents() 331 * @author Andreas Gohr <andi@splitbrain.org> 332 * @author Ben Coburn <btcoburn@silicodon.net> 333 * 334 * @param string $line changelog line 335 * @param string $ns restrict to given namespace 336 * @param int $flags flags to control which changes are included 337 * @param array $seen listing of seen pages 338 * @return array|bool false or array with info about a change 339 */ 340function _handleRecent($line,$ns,$flags,&$seen){ 341 if(empty($line)) return false; //skip empty lines 342 343 // split the line into parts 344 $recent = parseChangelogLine($line); 345 if ($recent===false) { return false; } 346 347 // skip seen ones 348 if(isset($seen[$recent['id']])) return false; 349 350 // skip minors 351 if($recent['type']===DOKU_CHANGE_TYPE_MINOR_EDIT && ($flags & RECENTS_SKIP_MINORS)) return false; 352 353 // remember in seen to skip additional sights 354 $seen[$recent['id']] = 1; 355 356 // check if it's a hidden page 357 if(isHiddenPage($recent['id'])) return false; 358 359 // filter namespace 360 if (($ns) && (strpos($recent['id'],$ns.':') !== 0)) return false; 361 362 // exclude subnamespaces 363 if (($flags & RECENTS_SKIP_SUBSPACES) && (getNS($recent['id']) != $ns)) return false; 364 365 // check ACL 366 if ($flags & RECENTS_MEDIA_CHANGES) { 367 $recent['perms'] = auth_quickaclcheck(getNS($recent['id']).':*'); 368 } else { 369 $recent['perms'] = auth_quickaclcheck($recent['id']); 370 } 371 if ($recent['perms'] < AUTH_READ) return false; 372 373 // check existance 374 if($flags & RECENTS_SKIP_DELETED){ 375 $fn = (($flags & RECENTS_MEDIA_CHANGES) ? mediaFN($recent['id']) : wikiFN($recent['id'])); 376 if(!file_exists($fn)) return false; 377 } 378 379 return $recent; 380} 381 382/** 383 * Class ChangeLog 384 * methods for handling of changelog of pages or media files 385 */ 386abstract class ChangeLog { 387 388 /** @var string */ 389 protected $id; 390 /** @var int */ 391 protected $chunk_size; 392 /** @var array */ 393 protected $cache; 394 395 /** 396 * Constructor 397 * 398 * @param string $id page id 399 * @param int $chunk_size maximum block size read from file 400 */ 401 public function __construct($id, $chunk_size = 8192) { 402 global $cache_revinfo; 403 404 $this->cache =& $cache_revinfo; 405 if(!isset($this->cache[$id])) { 406 $this->cache[$id] = array(); 407 } 408 409 $this->id = $id; 410 $this->setChunkSize($chunk_size); 411 412 } 413 414 /** 415 * Set chunk size for file reading 416 * Chunk size zero let read whole file at once 417 * 418 * @param int $chunk_size maximum block size read from file 419 */ 420 public function setChunkSize($chunk_size) { 421 if(!is_numeric($chunk_size)) $chunk_size = 0; 422 423 $this->chunk_size = (int) max($chunk_size, 0); 424 } 425 426 /** 427 * Returns path to changelog 428 * 429 * @return string path to file 430 */ 431 abstract protected function getChangelogFilename(); 432 433 /** 434 * Returns path to current page/media 435 * 436 * @return string path to file 437 */ 438 abstract protected function getFilename(); 439 440 /** 441 * Get the changelog information for a specific page id and revision (timestamp) 442 * 443 * Adjacent changelog lines are optimistically parsed and cached to speed up 444 * consecutive calls to getRevisionInfo. For large changelog files, only the chunk 445 * containing the requested changelog line is read. 446 * 447 * @param int $rev revision timestamp 448 * @return bool|array false or array with entries: 449 * - date: unix timestamp 450 * - ip: IPv4 address (127.0.0.1) 451 * - type: log line type 452 * - id: page id 453 * - user: user name 454 * - sum: edit summary (or action reason) 455 * - extra: extra data (varies by line type) 456 * 457 * @author Ben Coburn <btcoburn@silicodon.net> 458 * @author Kate Arzamastseva <pshns@ukr.net> 459 */ 460 public function getRevisionInfo($rev) { 461 $rev = max($rev, 0); 462 463 // check if it's already in the memory cache 464 if(isset($this->cache[$this->id]) && isset($this->cache[$this->id][$rev])) { 465 return $this->cache[$this->id][$rev]; 466 } 467 468 //read lines from changelog 469 list($fp, $lines) = $this->readloglines($rev); 470 if($fp) { 471 fclose($fp); 472 } 473 if(empty($lines)) return false; 474 475 // parse and cache changelog lines 476 foreach($lines as $value) { 477 $tmp = parseChangelogLine($value); 478 if($tmp !== false) { 479 $this->cache[$this->id][$tmp['date']] = $tmp; 480 } 481 } 482 if(!isset($this->cache[$this->id][$rev])) { 483 return false; 484 } 485 return $this->cache[$this->id][$rev]; 486 } 487 488 /** 489 * Return a list of page revisions numbers 490 * 491 * Does not guarantee that the revision exists in the attic, 492 * only that a line with the date exists in the changelog. 493 * By default the current revision is skipped. 494 * 495 * The current revision is automatically skipped when the page exists. 496 * See $INFO['meta']['last_change'] for the current revision. 497 * A negative $first let read the current revision too. 498 * 499 * For efficiency, the log lines are parsed and cached for later 500 * calls to getRevisionInfo. Large changelog files are read 501 * backwards in chunks until the requested number of changelog 502 * lines are recieved. 503 * 504 * @param int $first skip the first n changelog lines 505 * @param int $num number of revisions to return 506 * @return array with the revision timestamps 507 * 508 * @author Ben Coburn <btcoburn@silicodon.net> 509 * @author Kate Arzamastseva <pshns@ukr.net> 510 */ 511 public function getRevisions($first, $num) { 512 $revs = array(); 513 $lines = array(); 514 $count = 0; 515 516 $num = max($num, 0); 517 if($num == 0) { 518 return $revs; 519 } 520 521 if($first < 0) { 522 $first = 0; 523 } else if(file_exists($this->getFilename())) { 524 // skip current revision if the page exists 525 $first = max($first + 1, 0); 526 } 527 528 $file = $this->getChangelogFilename(); 529 530 if(!file_exists($file)) { 531 return $revs; 532 } 533 if(filesize($file) < $this->chunk_size || $this->chunk_size == 0) { 534 // read whole file 535 $lines = file($file); 536 if($lines === false) { 537 return $revs; 538 } 539 } else { 540 // read chunks backwards 541 $fp = fopen($file, 'rb'); // "file pointer" 542 if($fp === false) { 543 return $revs; 544 } 545 fseek($fp, 0, SEEK_END); 546 $tail = ftell($fp); 547 548 // chunk backwards 549 $finger = max($tail - $this->chunk_size, 0); 550 while($count < $num + $first) { 551 $nl = $this->getNewlinepointer($fp, $finger); 552 553 // was the chunk big enough? if not, take another bite 554 if($nl > 0 && $tail <= $nl) { 555 $finger = max($finger - $this->chunk_size, 0); 556 continue; 557 } else { 558 $finger = $nl; 559 } 560 561 // read chunk 562 $chunk = ''; 563 $read_size = max($tail - $finger, 0); // found chunk size 564 $got = 0; 565 while($got < $read_size && !feof($fp)) { 566 $tmp = @fread($fp, max(min($this->chunk_size, $read_size - $got), 0)); 567 if($tmp === false) { 568 break; 569 } //error state 570 $got += strlen($tmp); 571 $chunk .= $tmp; 572 } 573 $tmp = explode("\n", $chunk); 574 array_pop($tmp); // remove trailing newline 575 576 // combine with previous chunk 577 $count += count($tmp); 578 $lines = array_merge($tmp, $lines); 579 580 // next chunk 581 if($finger == 0) { 582 break; 583 } // already read all the lines 584 else { 585 $tail = $finger; 586 $finger = max($tail - $this->chunk_size, 0); 587 } 588 } 589 fclose($fp); 590 } 591 592 // skip parsing extra lines 593 $num = max(min(count($lines) - $first, $num), 0); 594 if ($first > 0 && $num > 0) { $lines = array_slice($lines, max(count($lines) - $first - $num, 0), $num); } 595 else if($first > 0 && $num == 0) { $lines = array_slice($lines, 0, max(count($lines) - $first, 0)); } 596 else if($first == 0 && $num > 0) { $lines = array_slice($lines, max(count($lines) - $num, 0)); } 597 598 // handle lines in reverse order 599 for($i = count($lines) - 1; $i >= 0; $i--) { 600 $tmp = parseChangelogLine($lines[$i]); 601 if($tmp !== false) { 602 $this->cache[$this->id][$tmp['date']] = $tmp; 603 $revs[] = $tmp['date']; 604 } 605 } 606 607 return $revs; 608 } 609 610 /** 611 * Get the nth revision left or right handside for a specific page id and revision (timestamp) 612 * 613 * For large changelog files, only the chunk containing the 614 * reference revision $rev is read and sometimes a next chunck. 615 * 616 * Adjacent changelog lines are optimistically parsed and cached to speed up 617 * consecutive calls to getRevisionInfo. 618 * 619 * @param int $rev revision timestamp used as startdate (doesn't need to be revisionnumber) 620 * @param int $direction give position of returned revision with respect to $rev; positive=next, negative=prev 621 * @return bool|int 622 * timestamp of the requested revision 623 * otherwise false 624 */ 625 public function getRelativeRevision($rev, $direction) { 626 $rev = max($rev, 0); 627 $direction = (int) $direction; 628 629 //no direction given or last rev, so no follow-up 630 if(!$direction || ($direction > 0 && $this->isCurrentRevision($rev))) { 631 return false; 632 } 633 634 //get lines from changelog 635 list($fp, $lines, $head, $tail, $eof) = $this->readloglines($rev); 636 if(empty($lines)) return false; 637 638 // look for revisions later/earlier then $rev, when founded count till the wanted revision is reached 639 // also parse and cache changelog lines for getRevisionInfo(). 640 $revcounter = 0; 641 $relativerev = false; 642 $checkotherchunck = true; //always runs once 643 while(!$relativerev && $checkotherchunck) { 644 $tmp = array(); 645 //parse in normal or reverse order 646 $count = count($lines); 647 if($direction > 0) { 648 $start = 0; 649 $step = 1; 650 } else { 651 $start = $count - 1; 652 $step = -1; 653 } 654 for($i = $start; $i >= 0 && $i < $count; $i = $i + $step) { 655 $tmp = parseChangelogLine($lines[$i]); 656 if($tmp !== false) { 657 $this->cache[$this->id][$tmp['date']] = $tmp; 658 //look for revs older/earlier then reference $rev and select $direction-th one 659 if(($direction > 0 && $tmp['date'] > $rev) || ($direction < 0 && $tmp['date'] < $rev)) { 660 $revcounter++; 661 if($revcounter == abs($direction)) { 662 $relativerev = $tmp['date']; 663 } 664 } 665 } 666 } 667 668 //true when $rev is found, but not the wanted follow-up. 669 $checkotherchunck = $fp 670 && ($tmp['date'] == $rev || ($revcounter > 0 && !$relativerev)) 671 && !(($tail == $eof && $direction > 0) || ($head == 0 && $direction < 0)); 672 673 if($checkotherchunck) { 674 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, $direction); 675 676 if(empty($lines)) break; 677 } 678 } 679 if($fp) { 680 fclose($fp); 681 } 682 683 return $relativerev; 684 } 685 686 /** 687 * Returns revisions around rev1 and rev2 688 * When available it returns $max entries for each revision 689 * 690 * @param int $rev1 oldest revision timestamp 691 * @param int $rev2 newest revision timestamp (0 looks up last revision) 692 * @param int $max maximum number of revisions returned 693 * @return array with two arrays with revisions surrounding rev1 respectively rev2 694 */ 695 public function getRevisionsAround($rev1, $rev2, $max = 50) { 696 $max = floor(abs($max) / 2)*2 + 1; 697 $rev1 = max($rev1, 0); 698 $rev2 = max($rev2, 0); 699 700 if($rev2) { 701 if($rev2 < $rev1) { 702 $rev = $rev2; 703 $rev2 = $rev1; 704 $rev1 = $rev; 705 } 706 } else { 707 //empty right side means a removed page. Look up last revision. 708 $revs = $this->getRevisions(-1, 1); 709 $rev2 = $revs[0]; 710 } 711 //collect revisions around rev2 712 list($revs2, $allrevs, $fp, $lines, $head, $tail) = $this->retrieveRevisionsAround($rev2, $max); 713 714 if(empty($revs2)) return array(array(), array()); 715 716 //collect revisions around rev1 717 $index = array_search($rev1, $allrevs); 718 if($index === false) { 719 //no overlapping revisions 720 list($revs1,,,,,) = $this->retrieveRevisionsAround($rev1, $max); 721 if(empty($revs1)) $revs1 = array(); 722 } else { 723 //revisions overlaps, reuse revisions around rev2 724 $revs1 = $allrevs; 725 while($head > 0) { 726 for($i = count($lines) - 1; $i >= 0; $i--) { 727 $tmp = parseChangelogLine($lines[$i]); 728 if($tmp !== false) { 729 $this->cache[$this->id][$tmp['date']] = $tmp; 730 $revs1[] = $tmp['date']; 731 $index++; 732 733 if($index > floor($max / 2)) break 2; 734 } 735 } 736 737 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, -1); 738 } 739 sort($revs1); 740 //return wanted selection 741 $revs1 = array_slice($revs1, max($index - floor($max/2), 0), $max); 742 } 743 744 return array(array_reverse($revs1), array_reverse($revs2)); 745 } 746 747 /** 748 * Returns lines from changelog. 749 * If file larger than $chuncksize, only chunck is read that could contain $rev. 750 * 751 * @param int $rev revision timestamp 752 * @return array|false 753 * if success returns array(fp, array(changeloglines), $head, $tail, $eof) 754 * where fp only defined for chuck reading, needs closing. 755 * otherwise false 756 */ 757 protected function readloglines($rev) { 758 $file = $this->getChangelogFilename(); 759 760 if(!file_exists($file)) { 761 return false; 762 } 763 764 $fp = null; 765 $head = 0; 766 $tail = 0; 767 $eof = 0; 768 769 if(filesize($file) < $this->chunk_size || $this->chunk_size == 0) { 770 // read whole file 771 $lines = file($file); 772 if($lines === false) { 773 return false; 774 } 775 } else { 776 // read by chunk 777 $fp = fopen($file, 'rb'); // "file pointer" 778 if($fp === false) { 779 return false; 780 } 781 $head = 0; 782 fseek($fp, 0, SEEK_END); 783 $eof = ftell($fp); 784 $tail = $eof; 785 786 // find chunk 787 while($tail - $head > $this->chunk_size) { 788 $finger = $head + floor(($tail - $head) / 2.0); 789 $finger = $this->getNewlinepointer($fp, $finger); 790 $tmp = fgets($fp); 791 if($finger == $head || $finger == $tail) { 792 break; 793 } 794 $tmp = parseChangelogLine($tmp); 795 $finger_rev = $tmp['date']; 796 797 if($finger_rev > $rev) { 798 $tail = $finger; 799 } else { 800 $head = $finger; 801 } 802 } 803 804 if($tail - $head < 1) { 805 // cound not find chunk, assume requested rev is missing 806 fclose($fp); 807 return false; 808 } 809 810 $lines = $this->readChunk($fp, $head, $tail); 811 } 812 return array( 813 $fp, 814 $lines, 815 $head, 816 $tail, 817 $eof 818 ); 819 } 820 821 /** 822 * Read chunk and return array with lines of given chunck. 823 * Has no check if $head and $tail are really at a new line 824 * 825 * @param resource $fp resource filepointer 826 * @param int $head start point chunck 827 * @param int $tail end point chunck 828 * @return array lines read from chunck 829 */ 830 protected function readChunk($fp, $head, $tail) { 831 $chunk = ''; 832 $chunk_size = max($tail - $head, 0); // found chunk size 833 $got = 0; 834 fseek($fp, $head); 835 while($got < $chunk_size && !feof($fp)) { 836 $tmp = @fread($fp, max(min($this->chunk_size, $chunk_size - $got), 0)); 837 if($tmp === false) { //error state 838 break; 839 } 840 $got += strlen($tmp); 841 $chunk .= $tmp; 842 } 843 $lines = explode("\n", $chunk); 844 array_pop($lines); // remove trailing newline 845 return $lines; 846 } 847 848 /** 849 * Set pointer to first new line after $finger and return its position 850 * 851 * @param resource $fp filepointer 852 * @param int $finger a pointer 853 * @return int pointer 854 */ 855 protected function getNewlinepointer($fp, $finger) { 856 fseek($fp, $finger); 857 $nl = $finger; 858 if($finger > 0) { 859 fgets($fp); // slip the finger forward to a new line 860 $nl = ftell($fp); 861 } 862 return $nl; 863 } 864 865 /** 866 * Check whether given revision is the current page 867 * 868 * @param int $rev timestamp of current page 869 * @return bool true if $rev is current revision, otherwise false 870 */ 871 public function isCurrentRevision($rev) { 872 return $rev == @filemtime($this->getFilename()); 873 } 874 875 /** 876 * Return an existing revision for a specific date which is 877 * the current one or younger or equal then the date 878 * 879 * @param number $date_at timestamp 880 * @return string revision ('' for current) 881 */ 882 function getLastRevisionAt($date_at){ 883 //requested date_at(timestamp) younger or equal then modified_time($this->id) => load current 884 if($date_at >= @filemtime($this->getFilename())) { 885 return ''; 886 } else if ($rev = $this->getRelativeRevision($date_at+1, -1)) { //+1 to get also the requested date revision 887 return $rev; 888 } else { 889 return false; 890 } 891 } 892 893 /** 894 * Returns the next lines of the changelog of the chunck before head or after tail 895 * 896 * @param resource $fp filepointer 897 * @param int $head position head of last chunk 898 * @param int $tail position tail of last chunk 899 * @param int $direction positive forward, negative backward 900 * @return array with entries: 901 * - $lines: changelog lines of readed chunk 902 * - $head: head of chunk 903 * - $tail: tail of chunk 904 */ 905 protected function readAdjacentChunk($fp, $head, $tail, $direction) { 906 if(!$fp) return array(array(), $head, $tail); 907 908 if($direction > 0) { 909 //read forward 910 $head = $tail; 911 $tail = $head + floor($this->chunk_size * (2 / 3)); 912 $tail = $this->getNewlinepointer($fp, $tail); 913 } else { 914 //read backward 915 $tail = $head; 916 $head = max($tail - $this->chunk_size, 0); 917 while(true) { 918 $nl = $this->getNewlinepointer($fp, $head); 919 // was the chunk big enough? if not, take another bite 920 if($nl > 0 && $tail <= $nl) { 921 $head = max($head - $this->chunk_size, 0); 922 } else { 923 $head = $nl; 924 break; 925 } 926 } 927 } 928 929 //load next chunck 930 $lines = $this->readChunk($fp, $head, $tail); 931 return array($lines, $head, $tail); 932 } 933 934 /** 935 * Collect the $max revisions near to the timestamp $rev 936 * 937 * @param int $rev revision timestamp 938 * @param int $max maximum number of revisions to be returned 939 * @return bool|array 940 * return array with entries: 941 * - $requestedrevs: array of with $max revision timestamps 942 * - $revs: all parsed revision timestamps 943 * - $fp: filepointer only defined for chuck reading, needs closing. 944 * - $lines: non-parsed changelog lines before the parsed revisions 945 * - $head: position of first readed changelogline 946 * - $lasttail: position of end of last readed changelogline 947 * otherwise false 948 */ 949 protected function retrieveRevisionsAround($rev, $max) { 950 //get lines from changelog 951 list($fp, $lines, $starthead, $starttail, /* $eof */) = $this->readloglines($rev); 952 if(empty($lines)) return false; 953 954 //parse chunk containing $rev, and read forward more chunks until $max/2 is reached 955 $head = $starthead; 956 $tail = $starttail; 957 $revs = array(); 958 $aftercount = $beforecount = 0; 959 while(count($lines) > 0) { 960 foreach($lines as $line) { 961 $tmp = parseChangelogLine($line); 962 if($tmp !== false) { 963 $this->cache[$this->id][$tmp['date']] = $tmp; 964 $revs[] = $tmp['date']; 965 if($tmp['date'] >= $rev) { 966 //count revs after reference $rev 967 $aftercount++; 968 if($aftercount == 1) $beforecount = count($revs); 969 } 970 //enough revs after reference $rev? 971 if($aftercount > floor($max / 2)) break 2; 972 } 973 } 974 //retrieve next chunk 975 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, 1); 976 } 977 if($aftercount == 0) return false; 978 979 $lasttail = $tail; 980 981 //read additional chuncks backward until $max/2 is reached and total number of revs is equal to $max 982 $lines = array(); 983 $i = 0; 984 if($aftercount > 0) { 985 $head = $starthead; 986 $tail = $starttail; 987 while($head > 0) { 988 list($lines, $head, $tail) = $this->readAdjacentChunk($fp, $head, $tail, -1); 989 990 for($i = count($lines) - 1; $i >= 0; $i--) { 991 $tmp = parseChangelogLine($lines[$i]); 992 if($tmp !== false) { 993 $this->cache[$this->id][$tmp['date']] = $tmp; 994 $revs[] = $tmp['date']; 995 $beforecount++; 996 //enough revs before reference $rev? 997 if($beforecount > max(floor($max / 2), $max - $aftercount)) break 2; 998 } 999 } 1000 } 1001 } 1002 sort($revs); 1003 1004 //keep only non-parsed lines 1005 $lines = array_slice($lines, 0, $i); 1006 //trunk desired selection 1007 $requestedrevs = array_slice($revs, -$max, $max); 1008 1009 return array($requestedrevs, $revs, $fp, $lines, $head, $lasttail); 1010 } 1011} 1012 1013/** 1014 * Class PageChangelog handles changelog of a wiki page 1015 */ 1016class PageChangelog extends ChangeLog { 1017 1018 /** 1019 * Returns path to changelog 1020 * 1021 * @return string path to file 1022 */ 1023 protected function getChangelogFilename() { 1024 return metaFN($this->id, '.changes'); 1025 } 1026 1027 /** 1028 * Returns path to current page/media 1029 * 1030 * @return string path to file 1031 */ 1032 protected function getFilename() { 1033 return wikiFN($this->id); 1034 } 1035} 1036 1037/** 1038 * Class MediaChangelog handles changelog of a media file 1039 */ 1040class MediaChangelog extends ChangeLog { 1041 1042 /** 1043 * Returns path to changelog 1044 * 1045 * @return string path to file 1046 */ 1047 protected function getChangelogFilename() { 1048 return mediaMetaFN($this->id, '.changes'); 1049 } 1050 1051 /** 1052 * Returns path to current page/media 1053 * 1054 * @return string path to file 1055 */ 1056 protected function getFilename() { 1057 return mediaFN($this->id); 1058 } 1059} 1060 1061/** 1062 * Get the changelog information for a specific page id 1063 * and revision (timestamp). Adjacent changelog lines 1064 * are optimistically parsed and cached to speed up 1065 * consecutive calls to getRevisionInfo. For large 1066 * changelog files, only the chunk containing the 1067 * requested changelog line is read. 1068 * 1069 * @deprecated 2013-11-20 1070 * 1071 * @author Ben Coburn <btcoburn@silicodon.net> 1072 * @author Kate Arzamastseva <pshns@ukr.net> 1073 * 1074 * @param string $id 1075 * @param int $rev 1076 * @param int $chunk_size 1077 * @param bool $media 1078 * @return array|bool 1079 */ 1080function getRevisionInfo($id, $rev, $chunk_size = 8192, $media = false) { 1081 dbg_deprecated('class PageChangeLog or class MediaChangelog'); 1082 if($media) { 1083 $changelog = new MediaChangeLog($id, $chunk_size); 1084 } else { 1085 $changelog = new PageChangeLog($id, $chunk_size); 1086 } 1087 return $changelog->getRevisionInfo($rev); 1088} 1089 1090/** 1091 * Return a list of page revisions numbers 1092 * Does not guarantee that the revision exists in the attic, 1093 * only that a line with the date exists in the changelog. 1094 * By default the current revision is skipped. 1095 * 1096 * The current revision is automatically skipped when the page exists. 1097 * See $INFO['meta']['last_change'] for the current revision. 1098 * 1099 * For efficiency, the log lines are parsed and cached for later 1100 * calls to getRevisionInfo. Large changelog files are read 1101 * backwards in chunks until the requested number of changelog 1102 * lines are recieved. 1103 * 1104 * @deprecated 2013-11-20 1105 * 1106 * @author Ben Coburn <btcoburn@silicodon.net> 1107 * @author Kate Arzamastseva <pshns@ukr.net> 1108 * 1109 * @param string $id the page of interest 1110 * @param int $first skip the first n changelog lines 1111 * @param int $num number of revisions to return 1112 * @param int $chunk_size 1113 * @param bool $media 1114 * @return array 1115 */ 1116function getRevisions($id, $first, $num, $chunk_size = 8192, $media = false) { 1117 dbg_deprecated('class PageChangeLog or class MediaChangelog'); 1118 if($media) { 1119 $changelog = new MediaChangeLog($id, $chunk_size); 1120 } else { 1121 $changelog = new PageChangeLog($id, $chunk_size); 1122 } 1123 return $changelog->getRevisions($first, $num); 1124} 1125 1126