1<?php 2 3namespace splitbrain\PHPArchive; 4 5/** 6 * Class Zip 7 * 8 * Creates or extracts Zip archives 9 * 10 * for specs see http://www.pkware.com/appnote 11 * 12 * @author Andreas Gohr <andi@splitbrain.org> 13 * @package splitbrain\PHPArchive 14 * @license MIT 15 */ 16class Zip extends Archive 17{ 18 19 protected $file = ''; 20 protected $fh; 21 protected $memory = ''; 22 protected $closed = true; 23 protected $writeaccess = false; 24 protected $ctrl_dir; 25 protected $complevel = 9; 26 27 /** 28 * Set the compression level. 29 * 30 * Compression Type is ignored for ZIP 31 * 32 * You can call this function before adding each file to set differen compression levels 33 * for each file. 34 * 35 * @param int $level Compression level (0 to 9) 36 * @param int $type Type of compression to use ignored for ZIP 37 * @return mixed 38 */ 39 public function setCompression($level = 9, $type = Archive::COMPRESS_AUTO) 40 { 41 $this->complevel = $level; 42 } 43 44 /** 45 * Open an existing ZIP file for reading 46 * 47 * @param string $file 48 * @throws ArchiveIOException 49 */ 50 public function open($file) 51 { 52 $this->file = $file; 53 $this->fh = @fopen($this->file, 'rb'); 54 if (!$this->fh) { 55 throw new ArchiveIOException('Could not open file for reading: '.$this->file); 56 } 57 $this->closed = false; 58 } 59 60 /** 61 * Read the contents of a ZIP archive 62 * 63 * This function lists the files stored in the archive, and returns an indexed array of FileInfo objects 64 * 65 * The archive is closed afer reading the contents, for API compatibility with TAR files 66 * Reopen the file with open() again if you want to do additional operations 67 * 68 * @throws ArchiveIOException 69 * @return FileInfo[] 70 */ 71 public function contents() 72 { 73 if ($this->closed || !$this->file) { 74 throw new ArchiveIOException('Can not read from a closed archive'); 75 } 76 77 $result = array(); 78 79 $centd = $this->readCentralDir(); 80 81 @rewind($this->fh); 82 @fseek($this->fh, $centd['offset']); 83 84 for ($i = 0; $i < $centd['entries']; $i++) { 85 $result[] = $this->header2fileinfo($this->readCentralFileHeader()); 86 } 87 88 $this->close(); 89 return $result; 90 } 91 92 /** 93 * Extract an existing ZIP archive 94 * 95 * The $strip parameter allows you to strip a certain number of path components from the filenames 96 * found in the tar file, similar to the --strip-components feature of GNU tar. This is triggered when 97 * an integer is passed as $strip. 98 * Alternatively a fixed string prefix may be passed in $strip. If the filename matches this prefix, 99 * the prefix will be stripped. It is recommended to give prefixes with a trailing slash. 100 * 101 * By default this will extract all files found in the archive. You can restrict the output using the $include 102 * and $exclude parameter. Both expect a full regular expression (including delimiters and modifiers). If 103 * $include is set only files that match this expression will be extracted. Files that match the $exclude 104 * expression will never be extracted. Both parameters can be used in combination. Expressions are matched against 105 * stripped filenames as described above. 106 * 107 * @param string $outdir the target directory for extracting 108 * @param int|string $strip either the number of path components or a fixed prefix to strip 109 * @param string $exclude a regular expression of files to exclude 110 * @param string $include a regular expression of files to include 111 * @throws ArchiveIOException 112 * @return FileInfo[] 113 */ 114 function extract($outdir, $strip = '', $exclude = '', $include = '') 115 { 116 if ($this->closed || !$this->file) { 117 throw new ArchiveIOException('Can not read from a closed archive'); 118 } 119 120 $outdir = rtrim($outdir, '/'); 121 @mkdir($outdir, 0777, true); 122 123 $extracted = array(); 124 125 $cdir = $this->readCentralDir(); 126 $pos_entry = $cdir['offset']; // begin of the central file directory 127 128 for ($i = 0; $i < $cdir['entries']; $i++) { 129 // read file header 130 @fseek($this->fh, $pos_entry); 131 $header = $this->readCentralFileHeader(); 132 $header['index'] = $i; 133 $pos_entry = ftell($this->fh); // position of the next file in central file directory 134 fseek($this->fh, $header['offset']); // seek to beginning of file header 135 $header = $this->readFileHeader($header); 136 $fileinfo = $this->header2fileinfo($header); 137 138 // apply strip rules 139 $fileinfo->strip($strip); 140 141 // skip unwanted files 142 if (!strlen($fileinfo->getPath()) || !$fileinfo->match($include, $exclude)) { 143 continue; 144 } 145 146 $extracted[] = $fileinfo; 147 148 // create output directory 149 $output = $outdir.'/'.$fileinfo->getPath(); 150 $directory = ($header['folder']) ? $output : dirname($output); 151 @mkdir($directory, 0777, true); 152 153 // nothing more to do for directories 154 if ($fileinfo->getIsdir()) { 155 continue; 156 } 157 158 // compressed files are written to temporary .gz file first 159 if ($header['compression'] == 0) { 160 $extractto = $output; 161 } else { 162 $extractto = $output.'.gz'; 163 } 164 165 // open file for writing 166 $fp = fopen($extractto, "wb"); 167 if (!$fp) { 168 throw new ArchiveIOException('Could not open file for writing: '.$extractto); 169 } 170 171 // prepend compression header 172 if ($header['compression'] != 0) { 173 $binary_data = pack( 174 'va1a1Va1a1', 175 0x8b1f, 176 chr($header['compression']), 177 chr(0x00), 178 time(), 179 chr(0x00), 180 chr(3) 181 ); 182 fwrite($fp, $binary_data, 10); 183 } 184 185 // read the file and store it on disk 186 $size = $header['compressed_size']; 187 while ($size != 0) { 188 $read_size = ($size < 2048 ? $size : 2048); 189 $buffer = fread($this->fh, $read_size); 190 $binary_data = pack('a'.$read_size, $buffer); 191 fwrite($fp, $binary_data, $read_size); 192 $size -= $read_size; 193 } 194 195 // finalize compressed file 196 if ($header['compression'] != 0) { 197 $binary_data = pack('VV', $header['crc'], $header['size']); 198 fwrite($fp, $binary_data, 8); 199 } 200 201 // close file 202 fclose($fp); 203 204 // unpack compressed file 205 if ($header['compression'] != 0) { 206 $gzp = @gzopen($extractto, 'rb'); 207 if (!$gzp) { 208 @unlink($extractto); 209 throw new ArchiveIOException('Failed file extracting. gzip support missing?'); 210 } 211 $fp = @fopen($output, 'wb'); 212 if (!$fp) { 213 throw new ArchiveIOException('Could not open file for writing: '.$extractto); 214 } 215 216 $size = $header['size']; 217 while ($size != 0) { 218 $read_size = ($size < 2048 ? $size : 2048); 219 $buffer = gzread($gzp, $read_size); 220 $binary_data = pack('a'.$read_size, $buffer); 221 @fwrite($fp, $binary_data, $read_size); 222 $size -= $read_size; 223 } 224 fclose($fp); 225 gzclose($gzp); 226 unlink($extractto); // remove temporary gz file 227 } 228 229 touch($output, $fileinfo->getMtime()); 230 //FIXME what about permissions? 231 } 232 233 $this->close(); 234 return $extracted; 235 } 236 237 /** 238 * Create a new ZIP file 239 * 240 * If $file is empty, the zip file will be created in memory 241 * 242 * @param string $file 243 * @throws ArchiveIOException 244 */ 245 public function create($file = '') 246 { 247 $this->file = $file; 248 $this->memory = ''; 249 $this->fh = 0; 250 251 if ($this->file) { 252 $this->fh = @fopen($this->file, 'wb'); 253 254 if (!$this->fh) { 255 throw new ArchiveIOException('Could not open file for writing: '.$this->file); 256 } 257 } 258 $this->writeaccess = true; 259 $this->closed = false; 260 $this->ctrl_dir = array(); 261 } 262 263 /** 264 * Add a file to the current ZIP archive using an existing file in the filesystem 265 * 266 * @param string $file path to the original file 267 * @param string|FileInfo $fileinfo either the name to us in archive (string) or a FileInfo oject with all meta data, empty to take from original 268 * @throws ArchiveIOException 269 */ 270 271 /** 272 * Add a file to the current archive using an existing file in the filesystem 273 * 274 * @param string $file path to the original file 275 * @param string|FileInfo $fileinfo either the name to us in archive (string) or a FileInfo oject with all meta data, empty to take from original 276 * @throws ArchiveIOException 277 */ 278 public function addFile($file, $fileinfo = '') 279 { 280 if (is_string($fileinfo)) { 281 $fileinfo = FileInfo::fromPath($file, $fileinfo); 282 } 283 284 if ($this->closed) { 285 throw new ArchiveIOException('Archive has been closed, files can no longer be added'); 286 } 287 288 $data = @file_get_contents($file); 289 if ($data === false) { 290 throw new ArchiveIOException('Could not open file for reading: '.$file); 291 } 292 293 // FIXME could we stream writing compressed data? gzwrite on a fopen handle? 294 $this->addData($fileinfo, $data); 295 } 296 297 /** 298 * Add a file to the current TAR archive using the given $data as content 299 * 300 * @param string|FileInfo $fileinfo either the name to us in archive (string) or a FileInfo oject with all meta data 301 * @param string $data binary content of the file to add 302 * @throws ArchiveIOException 303 */ 304 public function addData($fileinfo, $data) 305 { 306 if (is_string($fileinfo)) { 307 $fileinfo = new FileInfo($fileinfo); 308 } 309 310 if ($this->closed) { 311 throw new ArchiveIOException('Archive has been closed, files can no longer be added'); 312 } 313 314 // prepare info and compress data 315 $size = strlen($data); 316 $crc = crc32($data); 317 if ($this->complevel) { 318 $data = gzcompress($data, $this->complevel); 319 $data = substr($data, 2, -4); // strip compression headers 320 } 321 $csize = strlen($data); 322 $offset = $this->dataOffset(); 323 $name = $fileinfo->getPath(); 324 $time = $fileinfo->getMtime(); 325 326 // write local file header 327 $this->writebytes($this->makeLocalFileHeader( 328 $time, 329 $crc, 330 $size, 331 $csize, 332 $name, 333 (bool) $this->complevel 334 )); 335 336 // we store no encryption header 337 338 // write data 339 $this->writebytes($data); 340 341 // we store no data descriptor 342 343 // add info to central file directory 344 $this->ctrl_dir[] = $this->makeCentralFileRecord( 345 $offset, 346 $time, 347 $crc, 348 $size, 349 $csize, 350 $name, 351 (bool) $this->complevel 352 ); 353 } 354 355 /** 356 * Add the closing footer to the archive if in write mode, close all file handles 357 * 358 * After a call to this function no more data can be added to the archive, for 359 * read access no reading is allowed anymore 360 */ 361 public function close() 362 { 363 if ($this->closed) { 364 return; 365 } // we did this already 366 367 if ($this->writeaccess) { 368 // write central directory 369 $offset = $this->dataOffset(); 370 $ctrldir = join('', $this->ctrl_dir); 371 $this->writebytes($ctrldir); 372 373 // write end of central directory record 374 $this->writebytes("\x50\x4b\x05\x06"); // end of central dir signature 375 $this->writebytes(pack('v', 0)); // number of this disk 376 $this->writebytes(pack('v', 0)); // number of the disk with the start of the central directory 377 $this->writebytes(pack('v', 378 count($this->ctrl_dir))); // total number of entries in the central directory on this disk 379 $this->writebytes(pack('v', count($this->ctrl_dir))); // total number of entries in the central directory 380 $this->writebytes(pack('V', strlen($ctrldir))); // size of the central directory 381 $this->writebytes(pack('V', 382 $offset)); // offset of start of central directory with respect to the starting disk number 383 $this->writebytes(pack('v', 0)); // .ZIP file comment length 384 385 $this->ctrl_dir = array(); 386 } 387 388 // close file handles 389 if ($this->file) { 390 fclose($this->fh); 391 $this->file = ''; 392 $this->fh = 0; 393 } 394 395 $this->writeaccess = false; 396 $this->closed = true; 397 } 398 399 /** 400 * Returns the created in-memory archive data 401 * 402 * This implicitly calls close() on the Archive 403 */ 404 public function getArchive() 405 { 406 $this->close(); 407 408 return $this->memory; 409 } 410 411 /** 412 * Save the created in-memory archive data 413 * 414 * Note: It's more memory effective to specify the filename in the create() function and 415 * let the library work on the new file directly. 416 * 417 * @param $file 418 * @throws ArchiveIOException 419 */ 420 public function save($file) 421 { 422 if (!file_put_contents($file, $this->getArchive())) { 423 throw new ArchiveIOException('Could not write to file: '.$file); 424 } 425 } 426 427 /** 428 * Read the central directory 429 * 430 * This key-value list contains general information about the ZIP file 431 * 432 * @return array 433 */ 434 protected function readCentralDir() 435 { 436 $size = filesize($this->file); 437 if ($size < 277) { 438 $maximum_size = $size; 439 } else { 440 $maximum_size = 277; 441 } 442 443 @fseek($this->fh, $size - $maximum_size); 444 $pos = ftell($this->fh); 445 $bytes = 0x00000000; 446 447 while ($pos < $size) { 448 $byte = @fread($this->fh, 1); 449 $bytes = (($bytes << 8) & 0xFFFFFFFF) | ord($byte); 450 if ($bytes == 0x504b0506) { 451 break; 452 } 453 $pos++; 454 } 455 456 $data = unpack( 457 'vdisk/vdisk_start/vdisk_entries/ventries/Vsize/Voffset/vcomment_size', 458 fread($this->fh, 18) 459 ); 460 461 if ($data['comment_size'] != 0) { 462 $centd['comment'] = fread($this->fh, $data['comment_size']); 463 } else { 464 $centd['comment'] = ''; 465 } 466 $centd['entries'] = $data['entries']; 467 $centd['disk_entries'] = $data['disk_entries']; 468 $centd['offset'] = $data['offset']; 469 $centd['disk_start'] = $data['disk_start']; 470 $centd['size'] = $data['size']; 471 $centd['disk'] = $data['disk']; 472 return $centd; 473 } 474 475 /** 476 * Read the next central file header 477 * 478 * Assumes the current file pointer is pointing at the right position 479 * 480 * @return array 481 */ 482 protected function readCentralFileHeader() 483 { 484 $binary_data = fread($this->fh, 46); 485 $header = unpack( 486 'vchkid/vid/vversion/vversion_extracted/vflag/vcompression/vmtime/vmdate/Vcrc/Vcompressed_size/Vsize/vfilename_len/vextra_len/vcomment_len/vdisk/vinternal/Vexternal/Voffset', 487 $binary_data 488 ); 489 490 if ($header['filename_len'] != 0) { 491 $header['filename'] = fread($this->fh, $header['filename_len']); 492 } else { 493 $header['filename'] = ''; 494 } 495 496 if ($header['extra_len'] != 0) { 497 $header['extra'] = fread($this->fh, $header['extra_len']); 498 } else { 499 $header['extra'] = ''; 500 } 501 502 if ($header['comment_len'] != 0) { 503 $header['comment'] = fread($this->fh, $header['comment_len']); 504 } else { 505 $header['comment'] = ''; 506 } 507 508 $header['mtime'] = $this->makeUnixTime($header['mdate'], $header['mtime']); 509 $header['stored_filename'] = $header['filename']; 510 $header['status'] = 'ok'; 511 if (substr($header['filename'], -1) == '/') { 512 $header['external'] = 0x41FF0010; 513 } 514 $header['folder'] = ($header['external'] == 0x41FF0010 || $header['external'] == 16) ? 1 : 0; 515 516 return $header; 517 } 518 519 /** 520 * Reads the local file header 521 * 522 * This header precedes each individual file inside the zip file. Assumes the current file pointer is pointing at 523 * the right position already. Enhances the given central header with the data found at the local header. 524 * 525 * @param array $header the central file header read previously (see above) 526 * @return array 527 */ 528 protected function readFileHeader($header) 529 { 530 $binary_data = fread($this->fh, 30); 531 $data = unpack( 532 'vchk/vid/vversion/vflag/vcompression/vmtime/vmdate/Vcrc/Vcompressed_size/Vsize/vfilename_len/vextra_len', 533 $binary_data 534 ); 535 536 $header['filename'] = fread($this->fh, $data['filename_len']); 537 if ($data['extra_len'] != 0) { 538 $header['extra'] = fread($this->fh, $data['extra_len']); 539 } else { 540 $header['extra'] = ''; 541 } 542 543 $header['compression'] = $data['compression']; 544 foreach (array( 545 'size', 546 'compressed_size', 547 'crc' 548 ) as $hd) { // On ODT files, these headers are 0. Keep the previous value. 549 if ($data[$hd] != 0) { 550 $header[$hd] = $data[$hd]; 551 } 552 } 553 $header['flag'] = $data['flag']; 554 $header['mtime'] = $this->makeUnixTime($data['mdate'], $data['mtime']); 555 556 $header['stored_filename'] = $header['filename']; 557 $header['status'] = "ok"; 558 $header['folder'] = ($header['external'] == 0x41FF0010 || $header['external'] == 16) ? 1 : 0; 559 return $header; 560 } 561 562 /** 563 * Create fileinfo object from header data 564 * 565 * @param $header 566 * @return FileInfo 567 */ 568 protected function header2fileinfo($header) 569 { 570 $fileinfo = new FileInfo(); 571 $fileinfo->setPath($header['filename']); 572 $fileinfo->setSize($header['size']); 573 $fileinfo->setCompressedSize($header['compressed_size']); 574 $fileinfo->setMtime($header['mtime']); 575 $fileinfo->setComment($header['comment']); 576 $fileinfo->setIsdir($header['external'] == 0x41FF0010 || $header['external'] == 16); 577 return $fileinfo; 578 } 579 580 /** 581 * Write to the open filepointer or memory 582 * 583 * @param string $data 584 * @throws ArchiveIOException 585 * @return int number of bytes written 586 */ 587 protected function writebytes($data) 588 { 589 if (!$this->file) { 590 $this->memory .= $data; 591 $written = strlen($data); 592 } else { 593 $written = @fwrite($this->fh, $data); 594 } 595 if ($written === false) { 596 throw new ArchiveIOException('Failed to write to archive stream'); 597 } 598 return $written; 599 } 600 601 /** 602 * Current data pointer position 603 * 604 * @fixme might need a -1 605 * @return int 606 */ 607 protected function dataOffset() 608 { 609 if ($this->file) { 610 return ftell($this->fh); 611 } else { 612 return strlen($this->memory); 613 } 614 } 615 616 /** 617 * Create a DOS timestamp from a UNIX timestamp 618 * 619 * DOS timestamps start at 1980-01-01, earlier UNIX stamps will be set to this date 620 * 621 * @param $time 622 * @return int 623 */ 624 protected function makeDosTime($time) 625 { 626 $timearray = getdate($time); 627 if ($timearray['year'] < 1980) { 628 $timearray['year'] = 1980; 629 $timearray['mon'] = 1; 630 $timearray['mday'] = 1; 631 $timearray['hours'] = 0; 632 $timearray['minutes'] = 0; 633 $timearray['seconds'] = 0; 634 } 635 return (($timearray['year'] - 1980) << 25) | 636 ($timearray['mon'] << 21) | 637 ($timearray['mday'] << 16) | 638 ($timearray['hours'] << 11) | 639 ($timearray['minutes'] << 5) | 640 ($timearray['seconds'] >> 1); 641 } 642 643 /** 644 * Create a UNIX timestamp from a DOS timestamp 645 * 646 * @param $mdate 647 * @param $mtime 648 * @return int 649 */ 650 protected function makeUnixTime($mdate = null, $mtime = null) 651 { 652 if ($mdate && $mtime) { 653 $year = (($mdate & 0xFE00) >> 9) + 1980; 654 $month = ($mdate & 0x01E0) >> 5; 655 $day = $mdate & 0x001F; 656 657 $hour = ($mtime & 0xF800) >> 11; 658 $minute = ($mtime & 0x07E0) >> 5; 659 $seconde = ($mtime & 0x001F) << 1; 660 661 $mtime = mktime($hour, $minute, $seconde, $month, $day, $year); 662 } else { 663 $mtime = time(); 664 } 665 666 return $mtime; 667 } 668 669 /** 670 * Returns a local file header for the given data 671 * 672 * @param int $offset location of the local header 673 * @param int $ts unix timestamp 674 * @param int $crc CRC32 checksum of the uncompressed data 675 * @param int $len length of the uncompressed data 676 * @param int $clen length of the compressed data 677 * @param string $name file name 678 * @param boolean|null $comp if compression is used, if null it's determined from $len != $clen 679 * @return string 680 */ 681 protected function makeCentralFileRecord($offset, $ts, $crc, $len, $clen, $name, $comp = null) 682 { 683 if(is_null($comp)) $comp = $len != $clen; 684 $comp = $comp ? 8 : 0; 685 $dtime = dechex($this->makeDosTime($ts)); 686 687 $header = "\x50\x4b\x01\x02"; // central file header signature 688 $header .= pack('v', 14); // version made by - VFAT 689 $header .= pack('v', 20); // version needed to extract - 2.0 690 $header .= pack('v', 0); // general purpose flag - no flags set 691 $header .= pack('v', $comp); // compression method - deflate|none 692 $header .= pack( 693 'H*', 694 $dtime[6] . $dtime[7] . 695 $dtime[4] . $dtime[5] . 696 $dtime[2] . $dtime[3] . 697 $dtime[0] . $dtime[1] 698 ); // last mod file time and date 699 $header .= pack('V', $crc); // crc-32 700 $header .= pack('V', $clen); // compressed size 701 $header .= pack('V', $len); // uncompressed size 702 $header .= pack('v', strlen($name)); // file name length 703 $header .= pack('v', 0); // extra field length 704 $header .= pack('v', 0); // file comment length 705 $header .= pack('v', 0); // disk number start 706 $header .= pack('v', 0); // internal file attributes 707 $header .= pack('V', 0); // external file attributes @todo was 0x32!? 708 $header .= pack('V', $offset); // relative offset of local header 709 $header .= $name; // file name 710 711 return $header; 712 } 713 714 /** 715 * Returns a local file header for the given data 716 * 717 * @param int $ts unix timestamp 718 * @param int $crc CRC32 checksum of the uncompressed data 719 * @param int $len length of the uncompressed data 720 * @param int $clen length of the compressed data 721 * @param string $name file name 722 * @param boolean|null $comp if compression is used, if null it's determined from $len != $clen 723 * @return string 724 */ 725 protected function makeLocalFileHeader($ts, $crc, $len, $clen, $name, $comp = null) 726 { 727 if(is_null($comp)) $comp = $len != $clen; 728 $comp = $comp ? 8 : 0; 729 $dtime = dechex($this->makeDosTime($ts)); 730 731 $header = "\x50\x4b\x03\x04"; // local file header signature 732 $header .= pack('v', 20); // version needed to extract - 2.0 733 $header .= pack('v', 0); // general purpose flag - no flags set 734 $header .= pack('v', $comp); // compression method - deflate|none 735 $header .= pack( 736 'H*', 737 $dtime[6] . $dtime[7] . 738 $dtime[4] . $dtime[5] . 739 $dtime[2] . $dtime[3] . 740 $dtime[0] . $dtime[1] 741 ); // last mod file time and date 742 $header .= pack('V', $crc); // crc-32 743 $header .= pack('V', $clen); // compressed size 744 $header .= pack('V', $len); // uncompressed size 745 $header .= pack('v', strlen($name)); // file name length 746 $header .= pack('v', 0); // extra field length 747 $header .= $name; 748 return $header; 749 } 750} 751