MDL-51863 packer: ensure empty zip files behavior remains consistent
[moodle.git] / lib / filestorage / zip_archive.php
blob70b9d336f5dc68331cb59df4b75ce73f6b3b317d
1 <?php
2 // This file is part of Moodle - http://moodle.org/
3 //
4 // Moodle is free software: you can redistribute it and/or modify
5 // it under the terms of the GNU General Public License as published by
6 // the Free Software Foundation, either version 3 of the License, or
7 // (at your option) any later version.
8 //
9 // Moodle is distributed in the hope that it will be useful,
10 // but WITHOUT ANY WARRANTY; without even the implied warranty of
11 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 // GNU General Public License for more details.
14 // You should have received a copy of the GNU General Public License
15 // along with Moodle. If not, see <http://www.gnu.org/licenses/>.
17 /**
18 * Implementation of zip file archive.
20 * @package core_files
21 * @copyright 2008 Petr Skoda (http://skodak.org)
22 * @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
25 defined('MOODLE_INTERNAL') || die();
27 require_once("$CFG->libdir/filestorage/file_archive.php");
29 /**
30 * Zip file archive class.
32 * @package core_files
33 * @category files
34 * @copyright 2008 Petr Skoda (http://skodak.org)
35 * @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
37 class zip_archive extends file_archive {
39 /** @var string Pathname of archive */
40 protected $archivepathname = null;
42 /** @var int archive open mode */
43 protected $mode = null;
45 /** @var int Used memory tracking */
46 protected $usedmem = 0;
48 /** @var int Iteration position */
49 protected $pos = 0;
51 /** @var ZipArchive instance */
52 protected $za;
54 /** @var bool was this archive modified? */
55 protected $modified = false;
57 /** @var array unicode decoding array, created by decoding zip file */
58 protected $namelookup = null;
60 /** @var string base64 encoded contents of empty zip file */
61 protected static $emptyzipcontent = 'UEsFBgAAAAAAAAAAAAAAAAAAAAAAAA==';
63 /** @var bool ugly hack for broken empty zip handling in < PHP 5.3.10 */
64 protected $emptyziphack = false;
66 /**
67 * Create new zip_archive instance.
69 public function __construct() {
70 $this->encoding = null; // Autodetects encoding by default.
73 /**
74 * Open or create archive (depending on $mode).
76 * @todo MDL-31048 return error message
77 * @param string $archivepathname
78 * @param int $mode OPEN, CREATE or OVERWRITE constant
79 * @param string $encoding archive local paths encoding, empty means autodetect
80 * @return bool success
82 public function open($archivepathname, $mode=file_archive::CREATE, $encoding=null) {
83 $this->close();
85 $this->usedmem = 0;
86 $this->pos = 0;
87 $this->encoding = $encoding;
88 $this->mode = $mode;
90 $this->za = new ZipArchive();
92 switch($mode) {
93 case file_archive::OPEN: $flags = 0; break;
94 case file_archive::OVERWRITE: $flags = ZIPARCHIVE::CREATE | ZIPARCHIVE::OVERWRITE; break; //changed in PHP 5.2.8
95 case file_archive::CREATE:
96 default : $flags = ZIPARCHIVE::CREATE; break;
99 $result = $this->za->open($archivepathname, $flags);
101 if ($flags == 0 and $result === ZIPARCHIVE::ER_NOZIP and filesize($archivepathname) === 22) {
102 // Legacy PHP versions < 5.3.10 can not deal with empty zip archives.
103 if (file_get_contents($archivepathname) === base64_decode(self::$emptyzipcontent)) {
104 if ($temp = make_temp_directory('zip')) {
105 $this->emptyziphack = tempnam($temp, 'zip');
106 $this->za = new ZipArchive();
107 $result = $this->za->open($this->emptyziphack, ZIPARCHIVE::CREATE);
112 if ($result === true) {
113 if (file_exists($archivepathname)) {
114 $this->archivepathname = realpath($archivepathname);
115 } else {
116 $this->archivepathname = $archivepathname;
118 return true;
120 } else {
121 $message = 'Unknown error.';
122 switch ($result) {
123 case ZIPARCHIVE::ER_EXISTS: $message = 'File already exists.'; break;
124 case ZIPARCHIVE::ER_INCONS: $message = 'Zip archive inconsistent.'; break;
125 case ZIPARCHIVE::ER_INVAL: $message = 'Invalid argument.'; break;
126 case ZIPARCHIVE::ER_MEMORY: $message = 'Malloc failure.'; break;
127 case ZIPARCHIVE::ER_NOENT: $message = 'No such file.'; break;
128 case ZIPARCHIVE::ER_NOZIP: $message = 'Not a zip archive.'; break;
129 case ZIPARCHIVE::ER_OPEN: $message = 'Can\'t open file.'; break;
130 case ZIPARCHIVE::ER_READ: $message = 'Read error.'; break;
131 case ZIPARCHIVE::ER_SEEK: $message = 'Seek error.'; break;
133 debugging($message.': '.$archivepathname, DEBUG_DEVELOPER);
134 $this->za = null;
135 $this->archivepathname = null;
136 return false;
141 * Normalize $localname, always keep in utf-8 encoding.
143 * @param string $localname name of file in utf-8 encoding
144 * @return string normalised compressed file or directory name
146 protected function mangle_pathname($localname) {
147 $result = str_replace('\\', '/', $localname); // no MS \ separators
148 $result = preg_replace('/\.\.+/', '', $result); // prevent /.../
149 $result = ltrim($result, '/'); // no leading slash
151 if ($result === '.') {
152 $result = '';
155 return $result;
159 * Tries to convert $localname into utf-8
160 * please note that it may fail really badly.
161 * The resulting file name is cleaned.
163 * @param string $localname name (encoding is read from zip file or guessed)
164 * @return string in utf-8
166 protected function unmangle_pathname($localname) {
167 $this->init_namelookup();
169 if (!isset($this->namelookup[$localname])) {
170 $name = $localname;
171 // This should not happen.
172 if (!empty($this->encoding) and $this->encoding !== 'utf-8') {
173 $name = @core_text::convert($name, $this->encoding, 'utf-8');
175 $name = str_replace('\\', '/', $name); // no MS \ separators
176 $name = clean_param($name, PARAM_PATH); // only safe chars
177 return ltrim($name, '/'); // no leading slash
180 return $this->namelookup[$localname];
184 * Close archive, write changes to disk.
186 * @return bool success
188 public function close() {
189 if (!isset($this->za)) {
190 return false;
193 if ($this->emptyziphack) {
194 @$this->za->close();
195 $this->za = null;
196 $this->mode = null;
197 $this->namelookup = null;
198 $this->modified = false;
199 @unlink($this->emptyziphack);
200 $this->emptyziphack = false;
201 return true;
203 } else if ($this->za->numFiles == 0) {
204 // PHP can not create empty archives, so let's fake it.
205 @$this->za->close();
206 $this->za = null;
207 $this->mode = null;
208 $this->namelookup = null;
209 $this->modified = false;
210 @unlink($this->archivepathname);
211 $data = base64_decode(self::$emptyzipcontent);
212 if (!file_put_contents($this->archivepathname, $data)) {
213 return false;
215 return true;
218 $res = $this->za->close();
219 $this->za = null;
220 $this->mode = null;
221 $this->namelookup = null;
223 if ($this->modified) {
224 $this->fix_utf8_flags();
225 $this->modified = false;
228 return $res;
232 * Returns file stream for reading of content.
234 * @param int $index index of file
235 * @return resource|bool file handle or false if error
237 public function get_stream($index) {
238 if (!isset($this->za)) {
239 return false;
242 $name = $this->za->getNameIndex($index);
243 if ($name === false) {
244 return false;
247 return $this->za->getStream($name);
251 * Returns file information.
253 * @param int $index index of file
254 * @return stdClass|bool info object or false if error
256 public function get_info($index) {
257 if (!isset($this->za)) {
258 return false;
261 // Need to use the ZipArchive's numfiles, as $this->count() relies on this function to count actual files (skipping OSX junk).
262 if ($index < 0 or $index >=$this->za->numFiles) {
263 return false;
266 // PHP 5.6 introduced encoding guessing logic, we need to fall back
267 // to raw ZIP_FL_ENC_RAW (== 64) to get consistent results as in PHP 5.5.
268 $result = $this->za->statIndex($index, 64);
270 if ($result === false) {
271 return false;
274 $info = new stdClass();
275 $info->index = $index;
276 $info->original_pathname = $result['name'];
277 $info->pathname = $this->unmangle_pathname($result['name']);
278 $info->mtime = (int)$result['mtime'];
280 if ($info->pathname[strlen($info->pathname)-1] === '/') {
281 $info->is_directory = true;
282 $info->size = 0;
283 } else {
284 $info->is_directory = false;
285 $info->size = (int)$result['size'];
288 if ($this->is_system_file($info)) {
289 // Don't return system files.
290 return false;
293 return $info;
297 * Returns array of info about all files in archive.
299 * @return array of file infos
301 public function list_files() {
302 if (!isset($this->za)) {
303 return false;
306 $infos = array();
308 foreach ($this as $info) {
309 // Simply iterating over $this will give us info only for files we're interested in.
310 array_push($infos, $info);
313 return $infos;
316 public function is_system_file($fileinfo) {
317 if (substr($fileinfo->pathname, 0, 8) === '__MACOSX' or substr($fileinfo->pathname, -9) === '.DS_Store') {
318 // Mac OSX system files.
319 return true;
321 if (substr($fileinfo->pathname, -9) === 'Thumbs.db') {
322 $stream = $this->za->getStream($fileinfo->pathname);
323 $info = base64_encode(fread($stream, 8));
324 fclose($stream);
325 if ($info === '0M8R4KGxGuE=') {
326 // It's an OLE Compound File - so it's almost certainly a Windows thumbnail cache.
327 return true;
330 return false;
334 * Returns number of files in archive.
336 * @return int number of files
338 public function count() {
339 if (!isset($this->za)) {
340 return false;
343 return count($this->list_files());
347 * Returns approximate number of files in archive. This may be a slight
348 * overestimate.
350 * @return int|bool Estimated number of files, or false if not opened
352 public function estimated_count() {
353 if (!isset($this->za)) {
354 return false;
357 return $this->za->numFiles;
361 * Add file into archive.
363 * @param string $localname name of file in archive
364 * @param string $pathname location of file
365 * @return bool success
367 public function add_file_from_pathname($localname, $pathname) {
368 if ($this->emptyziphack) {
369 $this->close();
370 $this->open($this->archivepathname, file_archive::OVERWRITE, $this->encoding);
373 if (!isset($this->za)) {
374 return false;
377 if ($this->archivepathname === realpath($pathname)) {
378 // Do not add self into archive.
379 return false;
382 if (!is_readable($pathname) or is_dir($pathname)) {
383 return false;
386 if (is_null($localname)) {
387 $localname = clean_param($pathname, PARAM_PATH);
389 $localname = trim($localname, '/'); // No leading slashes in archives!
390 $localname = $this->mangle_pathname($localname);
392 if ($localname === '') {
393 // Sorry - conversion failed badly.
394 return false;
397 if (!$this->za->addFile($pathname, $localname)) {
398 return false;
400 $this->modified = true;
401 return true;
405 * Add content of string into archive.
407 * @param string $localname name of file in archive
408 * @param string $contents contents
409 * @return bool success
411 public function add_file_from_string($localname, $contents) {
412 if ($this->emptyziphack) {
413 $this->close();
414 $this->open($this->archivepathname, file_archive::OVERWRITE, $this->encoding);
417 if (!isset($this->za)) {
418 return false;
421 $localname = trim($localname, '/'); // No leading slashes in archives!
422 $localname = $this->mangle_pathname($localname);
424 if ($localname === '') {
425 // Sorry - conversion failed badly.
426 return false;
429 if ($this->usedmem > 2097151) {
430 // This prevents running out of memory when adding many large files using strings.
431 $this->close();
432 $res = $this->open($this->archivepathname, file_archive::OPEN, $this->encoding);
433 if ($res !== true) {
434 print_error('cannotopenzip');
437 $this->usedmem += strlen($contents);
439 if (!$this->za->addFromString($localname, $contents)) {
440 return false;
442 $this->modified = true;
443 return true;
447 * Add empty directory into archive.
449 * @param string $localname name of file in archive
450 * @return bool success
452 public function add_directory($localname) {
453 if ($this->emptyziphack) {
454 $this->close();
455 $this->open($this->archivepathname, file_archive::OVERWRITE, $this->encoding);
458 if (!isset($this->za)) {
459 return false;
461 $localname = trim($localname, '/'). '/';
462 $localname = $this->mangle_pathname($localname);
464 if ($localname === '/') {
465 // Sorry - conversion failed badly.
466 return false;
469 if ($localname !== '') {
470 if (!$this->za->addEmptyDir($localname)) {
471 return false;
473 $this->modified = true;
475 return true;
479 * Returns current file info.
481 * @return stdClass
483 public function current() {
484 if (!isset($this->za)) {
485 return false;
488 return $this->get_info($this->pos);
492 * Returns the index of current file.
494 * @return int current file index
496 public function key() {
497 return $this->pos;
501 * Moves forward to next file.
503 public function next() {
504 $this->pos++;
508 * Rewinds back to the first file.
510 public function rewind() {
511 $this->pos = 0;
515 * Did we reach the end?
517 * @return bool
519 public function valid() {
520 if (!isset($this->za)) {
521 return false;
524 // Skip over unwanted system files (get_info will return false).
525 while (!$this->get_info($this->pos) && $this->pos < $this->za->numFiles) {
526 $this->next();
529 // No files left - we're at the end.
530 if ($this->pos >= $this->za->numFiles) {
531 return false;
534 return true;
538 * Create a map of file names used in zip archive.
539 * @return void
541 protected function init_namelookup() {
542 if ($this->emptyziphack) {
543 $this->namelookup = array();
544 return;
547 if (!isset($this->za)) {
548 return;
550 if (isset($this->namelookup)) {
551 return;
554 $this->namelookup = array();
556 if ($this->mode != file_archive::OPEN) {
557 // No need to tweak existing names when creating zip file because there are none yet!
558 return;
561 if (!file_exists($this->archivepathname)) {
562 return;
565 if (!$fp = fopen($this->archivepathname, 'rb')) {
566 return;
568 if (!$filesize = filesize($this->archivepathname)) {
569 return;
572 $centralend = self::zip_get_central_end($fp, $filesize);
574 if ($centralend === false or $centralend['disk'] !== 0 or $centralend['disk_start'] !== 0 or $centralend['offset'] === 0xFFFFFFFF) {
575 // Single disk archives only and o support for ZIP64, sorry.
576 fclose($fp);
577 return;
580 fseek($fp, $centralend['offset']);
581 $data = fread($fp, $centralend['size']);
582 $pos = 0;
583 $files = array();
584 for($i=0; $i<$centralend['entries']; $i++) {
585 $file = self::zip_parse_file_header($data, $centralend, $pos);
586 if ($file === false) {
587 // Wrong header, sorry.
588 fclose($fp);
589 return;
591 $files[] = $file;
593 fclose($fp);
595 foreach ($files as $file) {
596 $name = $file['name'];
597 if (preg_match('/^[a-zA-Z0-9_\-\.]*$/', $file['name'])) {
598 // No need to fix ASCII.
599 $name = fix_utf8($name);
601 } else if (!($file['general'] & pow(2, 11))) {
602 // First look for unicode name alternatives.
603 $found = false;
604 foreach($file['extra'] as $extra) {
605 if ($extra['id'] === 0x7075) {
606 $data = unpack('cversion/Vcrc', substr($extra['data'], 0, 5));
607 if ($data['crc'] === crc32($name)) {
608 $found = true;
609 $name = substr($extra['data'], 5);
613 if (!$found and !empty($this->encoding) and $this->encoding !== 'utf-8') {
614 // Try the encoding from open().
615 $newname = @core_text::convert($name, $this->encoding, 'utf-8');
616 $original = core_text::convert($newname, 'utf-8', $this->encoding);
617 if ($original === $name) {
618 $found = true;
619 $name = $newname;
622 if (!$found and $file['version'] === 0x315) {
623 // This looks like OS X build in zipper.
624 $newname = fix_utf8($name);
625 if ($newname === $name) {
626 $found = true;
627 $name = $newname;
630 if (!$found and $file['version'] === 0) {
631 // This looks like our old borked Moodle 2.2 file.
632 $newname = fix_utf8($name);
633 if ($newname === $name) {
634 $found = true;
635 $name = $newname;
638 if (!$found and $encoding = get_string('oldcharset', 'langconfig')) {
639 // Last attempt - try the dos/unix encoding from current language.
640 $windows = true;
641 foreach($file['extra'] as $extra) {
642 // In Windows archivers do not usually set any extras with the exception of NTFS flag in WinZip/WinRar.
643 $windows = false;
644 if ($extra['id'] === 0x000a) {
645 $windows = true;
646 break;
650 if ($windows === true) {
651 switch(strtoupper($encoding)) {
652 case 'ISO-8859-1': $encoding = 'CP850'; break;
653 case 'ISO-8859-2': $encoding = 'CP852'; break;
654 case 'ISO-8859-4': $encoding = 'CP775'; break;
655 case 'ISO-8859-5': $encoding = 'CP866'; break;
656 case 'ISO-8859-6': $encoding = 'CP720'; break;
657 case 'ISO-8859-7': $encoding = 'CP737'; break;
658 case 'ISO-8859-8': $encoding = 'CP862'; break;
659 case 'EUC-JP':
660 case 'UTF-8':
661 if ($winchar = get_string('localewincharset', 'langconfig')) {
662 // Most probably works only for zh_cn,
663 // if there are more problems we could add zipcharset to langconfig files.
664 $encoding = $winchar;
666 break;
669 $newname = @core_text::convert($name, $encoding, 'utf-8');
670 $original = core_text::convert($newname, 'utf-8', $encoding);
672 if ($original === $name) {
673 $name = $newname;
677 $name = str_replace('\\', '/', $name); // no MS \ separators
678 $name = clean_param($name, PARAM_PATH); // only safe chars
679 $name = ltrim($name, '/'); // no leading slash
681 if (function_exists('normalizer_normalize')) {
682 $name = normalizer_normalize($name, Normalizer::FORM_C);
685 $this->namelookup[$file['name']] = $name;
690 * Add unicode flag to all files in archive.
692 * NOTE: single disk archives only, no ZIP64 support.
694 * @return bool success, modifies the file contents
696 protected function fix_utf8_flags() {
697 if ($this->emptyziphack) {
698 return true;
701 if (!file_exists($this->archivepathname)) {
702 return true;
705 // Note: the ZIP structure is described at http://www.pkware.com/documents/casestudies/APPNOTE.TXT
706 if (!$fp = fopen($this->archivepathname, 'rb+')) {
707 return false;
709 if (!$filesize = filesize($this->archivepathname)) {
710 return false;
713 $centralend = self::zip_get_central_end($fp, $filesize);
715 if ($centralend === false or $centralend['disk'] !== 0 or $centralend['disk_start'] !== 0 or $centralend['offset'] === 0xFFFFFFFF) {
716 // Single disk archives only and o support for ZIP64, sorry.
717 fclose($fp);
718 return false;
721 fseek($fp, $centralend['offset']);
722 $data = fread($fp, $centralend['size']);
723 $pos = 0;
724 $files = array();
725 for($i=0; $i<$centralend['entries']; $i++) {
726 $file = self::zip_parse_file_header($data, $centralend, $pos);
727 if ($file === false) {
728 // Wrong header, sorry.
729 fclose($fp);
730 return false;
733 $newgeneral = $file['general'] | pow(2, 11);
734 if ($newgeneral === $file['general']) {
735 // Nothing to do with this file.
736 continue;
739 if (preg_match('/^[a-zA-Z0-9_\-\.]*$/', $file['name'])) {
740 // ASCII file names are always ok.
741 continue;
743 if ($file['extra']) {
744 // Most probably not created by php zip ext, better to skip it.
745 continue;
747 if (fix_utf8($file['name']) !== $file['name']) {
748 // Does not look like a valid utf-8 encoded file name, skip it.
749 continue;
752 // Read local file header.
753 fseek($fp, $file['local_offset']);
754 $localfile = unpack('Vsig/vversion_req/vgeneral/vmethod/vmtime/vmdate/Vcrc/Vsize_compressed/Vsize/vname_length/vextra_length', fread($fp, 30));
755 if ($localfile['sig'] !== 0x04034b50) {
756 // Borked file!
757 fclose($fp);
758 return false;
761 $file['local'] = $localfile;
762 $files[] = $file;
765 foreach ($files as $file) {
766 $localfile = $file['local'];
767 // Add the unicode flag in central file header.
768 fseek($fp, $file['central_offset'] + 8);
769 if (ftell($fp) === $file['central_offset'] + 8) {
770 $newgeneral = $file['general'] | pow(2, 11);
771 fwrite($fp, pack('v', $newgeneral));
773 // Modify local file header too.
774 fseek($fp, $file['local_offset'] + 6);
775 if (ftell($fp) === $file['local_offset'] + 6) {
776 $newgeneral = $localfile['general'] | pow(2, 11);
777 fwrite($fp, pack('v', $newgeneral));
781 fclose($fp);
782 return true;
786 * Read end of central signature of ZIP file.
787 * @internal
788 * @static
789 * @param resource $fp
790 * @param int $filesize
791 * @return array|bool
793 public static function zip_get_central_end($fp, $filesize) {
794 // Find end of central directory record.
795 fseek($fp, $filesize - 22);
796 $info = unpack('Vsig', fread($fp, 4));
797 if ($info['sig'] === 0x06054b50) {
798 // There is no comment.
799 fseek($fp, $filesize - 22);
800 $data = fread($fp, 22);
801 } else {
802 // There is some comment with 0xFF max size - that is 65557.
803 fseek($fp, $filesize - 65557);
804 $data = fread($fp, 65557);
807 $pos = strpos($data, pack('V', 0x06054b50));
808 if ($pos === false) {
809 // Borked ZIP structure!
810 return false;
812 $centralend = unpack('Vsig/vdisk/vdisk_start/vdisk_entries/ventries/Vsize/Voffset/vcomment_length', substr($data, $pos, 22));
813 if ($centralend['comment_length']) {
814 $centralend['comment'] = substr($data, 22, $centralend['comment_length']);
815 } else {
816 $centralend['comment'] = '';
819 return $centralend;
823 * Parse file header.
824 * @internal
825 * @param string $data
826 * @param array $centralend
827 * @param int $pos (modified)
828 * @return array|bool file info
830 public static function zip_parse_file_header($data, $centralend, &$pos) {
831 $file = unpack('Vsig/vversion/vversion_req/vgeneral/vmethod/Vmodified/Vcrc/Vsize_compressed/Vsize/vname_length/vextra_length/vcomment_length/vdisk/vattr/Vattrext/Vlocal_offset', substr($data, $pos, 46));
832 $file['central_offset'] = $centralend['offset'] + $pos;
833 $pos = $pos + 46;
834 if ($file['sig'] !== 0x02014b50) {
835 // Borked ZIP structure!
836 return false;
838 $file['name'] = substr($data, $pos, $file['name_length']);
839 $pos = $pos + $file['name_length'];
840 $file['extra'] = array();
841 $file['extra_data'] = '';
842 if ($file['extra_length']) {
843 $extradata = substr($data, $pos, $file['extra_length']);
844 $file['extra_data'] = $extradata;
845 while (strlen($extradata) > 4) {
846 $extra = unpack('vid/vsize', substr($extradata, 0, 4));
847 $extra['data'] = substr($extradata, 4, $extra['size']);
848 $extradata = substr($extradata, 4+$extra['size']);
849 $file['extra'][] = $extra;
851 $pos = $pos + $file['extra_length'];
853 if ($file['comment_length']) {
854 $pos = $pos + $file['comment_length'];
855 $file['comment'] = substr($data, $pos, $file['comment_length']);
856 } else {
857 $file['comment'] = '';
859 return $file;