1<?php
2
3/////////////////////////////////////////////////////////////////
4/// getID3() by James Heinrich <info@getid3.org>               //
5//  available at https://github.com/JamesHeinrich/getID3       //
6//            or https://www.getid3.org                        //
7//            or http://getid3.sourceforge.net                 //
8//  see readme.txt for more details                            //
9/////////////////////////////////////////////////////////////////
10//                                                             //
11// module.archive.gzip.php                                     //
12// module for analyzing GZIP files                             //
13// dependencies: NONE                                          //
14//                                                            ///
15/////////////////////////////////////////////////////////////////
16//                                                             //
17// Module originally written by                                //
18//      Mike Mozolin <teddybearØmail*ru>                       //
19//                                                             //
20/////////////////////////////////////////////////////////////////
21
22if (!defined('GETID3_INCLUDEPATH')) { // prevent path-exposing attacks that access modules directly on public webservers
23	exit;
24}
25
26class getid3_gzip extends getid3_handler
27{
28	/**
29	 * Optional file list - disable for speed.
30	 *
31	 * Decode gzipped files, if possible, and parse recursively (.tar.gz for example).
32	 *
33	 * @var bool
34	 */
35	public $option_gzip_parse_contents = false;
36
37	/**
38	 * @return bool
39	 */
40	public function Analyze() {
41		$info = &$this->getid3->info;
42
43		$info['fileformat'] = 'gzip';
44
45		$start_length = 10;
46		$unpack_header = 'a1id1/a1id2/a1cmethod/a1flags/a4mtime/a1xflags/a1os';
47		//+---+---+---+---+---+---+---+---+---+---+
48		//|ID1|ID2|CM |FLG|     MTIME     |XFL|OS |
49		//+---+---+---+---+---+---+---+---+---+---+
50
51		if ($info['php_memory_limit'] && ($info['filesize'] > $info['php_memory_limit'])) {
52			$this->error('File is too large ('.number_format($info['filesize']).' bytes) to read into memory (limit: '.number_format($info['php_memory_limit'] / 1048576).'MB)');
53			return false;
54		}
55		$this->fseek(0);
56		$buffer = $this->fread($info['filesize']);
57
58		$arr_members = explode("\x1F\x8B\x08", $buffer);
59		$num_members = 0;
60		while (true) {
61			$is_wrong_members = false;
62			$num_members = intval(count($arr_members));
63			for ($i = 0; $i < $num_members; $i++) {
64				if (strlen($arr_members[$i]) == 0) {
65					continue;
66				}
67				$buf = "\x1F\x8B\x08".$arr_members[$i];
68
69				$attr = unpack($unpack_header, substr($buf, 0, $start_length));
70				if (!$this->get_os_type(ord($attr['os']))) {
71					// Merge member with previous if wrong OS type
72					$arr_members[($i - 1)] .= $buf;
73					$arr_members[$i] = '';
74					$is_wrong_members = true;
75					continue;
76				}
77			}
78			if (!$is_wrong_members) {
79				break;
80			}
81		}
82
83		$info['gzip']['files'] = array();
84
85		$fpointer = 0;
86		$idx = 0;
87		for ($i = 0; $i < $num_members; $i++) {
88			if (strlen($arr_members[$i]) == 0) {
89				continue;
90			}
91			$thisInfo = &$info['gzip']['member_header'][++$idx];
92
93			$buff = "\x1F\x8B\x08".$arr_members[$i];
94
95			$attr = unpack($unpack_header, substr($buff, 0, $start_length));
96			$thisInfo['filemtime']      = getid3_lib::LittleEndian2Int($attr['mtime']);
97			$thisInfo['raw']['id1']     = ord($attr['cmethod']);
98			$thisInfo['raw']['id2']     = ord($attr['cmethod']);
99			$thisInfo['raw']['cmethod'] = ord($attr['cmethod']);
100			$thisInfo['raw']['os']      = ord($attr['os']);
101			$thisInfo['raw']['xflags']  = ord($attr['xflags']);
102			$thisInfo['raw']['flags']   = ord($attr['flags']);
103
104			$thisInfo['flags']['crc16']    = (bool) ($thisInfo['raw']['flags'] & 0x02);
105			$thisInfo['flags']['extra']    = (bool) ($thisInfo['raw']['flags'] & 0x04);
106			$thisInfo['flags']['filename'] = (bool) ($thisInfo['raw']['flags'] & 0x08);
107			$thisInfo['flags']['comment']  = (bool) ($thisInfo['raw']['flags'] & 0x10);
108
109			$thisInfo['compression'] = $this->get_xflag_type($thisInfo['raw']['xflags']);
110
111			$thisInfo['os'] = $this->get_os_type($thisInfo['raw']['os']);
112			if (!$thisInfo['os']) {
113				$this->error('Read error on gzip file');
114				return false;
115			}
116
117			$fpointer = 10;
118			$arr_xsubfield = array();
119			// bit 2 - FLG.FEXTRA
120			//+---+---+=================================+
121			//| XLEN  |...XLEN bytes of "extra field"...|
122			//+---+---+=================================+
123			if ($thisInfo['flags']['extra']) {
124				$w_xlen = substr($buff, $fpointer, 2);
125				$xlen = getid3_lib::LittleEndian2Int($w_xlen);
126				$fpointer += 2;
127
128				$thisInfo['raw']['xfield'] = substr($buff, $fpointer, $xlen);
129				// Extra SubFields
130				//+---+---+---+---+==================================+
131				//|SI1|SI2|  LEN  |... LEN bytes of subfield data ...|
132				//+---+---+---+---+==================================+
133				$idx = 0;
134				while (true) {
135					if ($idx >= $xlen) {
136						break;
137					}
138					$si1 = ord(substr($buff, $fpointer + $idx++, 1));
139					$si2 = ord(substr($buff, $fpointer + $idx++, 1));
140					if (($si1 == 0x41) && ($si2 == 0x70)) {
141						$w_xsublen = substr($buff, $fpointer + $idx, 2);
142						$xsublen = getid3_lib::LittleEndian2Int($w_xsublen);
143						$idx += 2;
144						$arr_xsubfield[] = substr($buff, $fpointer + $idx, $xsublen);
145						$idx += $xsublen;
146					} else {
147						break;
148					}
149				}
150				$fpointer += $xlen;
151			}
152			// bit 3 - FLG.FNAME
153			//+=========================================+
154			//|...original file name, zero-terminated...|
155			//+=========================================+
156			// GZIP files may have only one file, with no filename, so assume original filename is current filename without .gz
157			$thisInfo['filename'] = preg_replace('#\\.gz$#i', '', $info['filename']);
158			if ($thisInfo['flags']['filename']) {
159				$thisInfo['filename'] = '';
160				while (true) {
161					if (ord($buff[$fpointer]) == 0) {
162						$fpointer++;
163						break;
164					}
165					$thisInfo['filename'] .= $buff[$fpointer];
166					$fpointer++;
167				}
168			}
169			// bit 4 - FLG.FCOMMENT
170			//+===================================+
171			//|...file comment, zero-terminated...|
172			//+===================================+
173			if ($thisInfo['flags']['comment']) {
174				while (true) {
175					if (ord($buff[$fpointer]) == 0) {
176						$fpointer++;
177						break;
178					}
179					$thisInfo['comment'] .= $buff[$fpointer];
180					$fpointer++;
181				}
182			}
183			// bit 1 - FLG.FHCRC
184			//+---+---+
185			//| CRC16 |
186			//+---+---+
187			if ($thisInfo['flags']['crc16']) {
188				$w_crc = substr($buff, $fpointer, 2);
189				$thisInfo['crc16'] = getid3_lib::LittleEndian2Int($w_crc);
190				$fpointer += 2;
191			}
192			// bit 0 - FLG.FTEXT
193			//if ($thisInfo['raw']['flags'] & 0x01) {
194			//	Ignored...
195			//}
196			// bits 5, 6, 7 - reserved
197
198			$thisInfo['crc32']    = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 8, 4));
199			$thisInfo['filesize'] = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 4));
200
201			$info['gzip']['files'] = getid3_lib::array_merge_clobber($info['gzip']['files'], getid3_lib::CreateDeepArray($thisInfo['filename'], '/', $thisInfo['filesize']));
202
203			if ($this->option_gzip_parse_contents) {
204				// Try to inflate GZip
205				$csize = 0;
206				$inflated = '';
207				$chkcrc32 = '';
208				if (function_exists('gzinflate')) {
209					$cdata = substr($buff, $fpointer);
210					$cdata = substr($cdata, 0, strlen($cdata) - 8);
211					$csize = strlen($cdata);
212					$inflated = gzinflate($cdata);
213
214					// Calculate CRC32 for inflated content
215					$thisInfo['crc32_valid'] = sprintf('%u', crc32($inflated)) == $thisInfo['crc32'];
216
217					// determine format
218					$formattest = substr($inflated, 0, 32774);
219					$getid3_temp = new getID3();
220					$determined_format = $getid3_temp->GetFileFormat($formattest);
221					unset($getid3_temp);
222
223					// file format is determined
224					$determined_format['module'] = (isset($determined_format['module']) ? $determined_format['module'] : '');
225					switch ($determined_format['module']) {
226						case 'tar':
227							// view TAR-file info
228							if (file_exists(GETID3_INCLUDEPATH.$determined_format['include']) && include_once(GETID3_INCLUDEPATH.$determined_format['include'])) {
229								if (($temp_tar_filename = tempnam(GETID3_TEMP_DIR, 'getID3')) === false) {
230									// can't find anywhere to create a temp file, abort
231									$this->error('Unable to create temp file to parse TAR inside GZIP file');
232									break;
233								}
234								if ($fp_temp_tar = fopen($temp_tar_filename, 'w+b')) {
235									fwrite($fp_temp_tar, $inflated);
236									fclose($fp_temp_tar);
237									$getid3_temp = new getID3();
238									$getid3_temp->openfile($temp_tar_filename);
239									$getid3_tar = new getid3_tar($getid3_temp);
240									$getid3_tar->Analyze();
241									$info['gzip']['member_header'][$idx]['tar'] = $getid3_temp->info['tar'];
242									unset($getid3_temp, $getid3_tar);
243									unlink($temp_tar_filename);
244								} else {
245									$this->error('Unable to fopen() temp file to parse TAR inside GZIP file');
246									break;
247								}
248							}
249							break;
250
251						case '':
252						default:
253							// unknown or unhandled format
254							break;
255					}
256				} else {
257					$this->warning('PHP is not compiled with gzinflate() support. Please enable PHP Zlib extension or recompile with the --with-zlib switch');
258				}
259			}
260		}
261		return true;
262	}
263
264	/**
265	 * Converts the OS type.
266	 *
267	 * @param string $key
268	 *
269	 * @return string
270	 */
271	public function get_os_type($key) {
272		static $os_type = array(
273			'0'   => 'FAT filesystem (MS-DOS, OS/2, NT/Win32)',
274			'1'   => 'Amiga',
275			'2'   => 'VMS (or OpenVMS)',
276			'3'   => 'Unix',
277			'4'   => 'VM/CMS',
278			'5'   => 'Atari TOS',
279			'6'   => 'HPFS filesystem (OS/2, NT)',
280			'7'   => 'Macintosh',
281			'8'   => 'Z-System',
282			'9'   => 'CP/M',
283			'10'  => 'TOPS-20',
284			'11'  => 'NTFS filesystem (NT)',
285			'12'  => 'QDOS',
286			'13'  => 'Acorn RISCOS',
287			'255' => 'unknown'
288		);
289		return (isset($os_type[$key]) ? $os_type[$key] : '');
290	}
291
292	/**
293	 * Converts the eXtra FLags.
294	 *
295	 * @param string $key
296	 *
297	 * @return string
298	 */
299	public function get_xflag_type($key) {
300		static $xflag_type = array(
301			'0' => 'unknown',
302			'2' => 'maximum compression',
303			'4' => 'fastest algorithm'
304		);
305		return (isset($xflag_type[$key]) ? $xflag_type[$key] : '');
306	}
307}
308
309