rlm@3
|
1 <?php
|
rlm@3
|
2 // +----------------------------------------------------------------------+
|
rlm@3
|
3 // | PHP version 5 |
|
rlm@3
|
4 // +----------------------------------------------------------------------+
|
rlm@3
|
5 // | Copyright (c) 2002-2006 James Heinrich, Allan Hansen |
|
rlm@3
|
6 // +----------------------------------------------------------------------+
|
rlm@3
|
7 // | This source file is subject to version 2 of the GPL license, |
|
rlm@3
|
8 // | that is bundled with this package in the file license.txt and is |
|
rlm@3
|
9 // | available through the world-wide-web at the following url: |
|
rlm@3
|
10 // | http://www.gnu.org/copyleft/gpl.html |
|
rlm@3
|
11 // +----------------------------------------------------------------------+
|
rlm@3
|
12 // | getID3() - http://getid3.sourceforge.net or http://www.getid3.org |
|
rlm@3
|
13 // +----------------------------------------------------------------------+
|
rlm@3
|
14 // | Authors: James Heinrich <infoØgetid3*org> |
|
rlm@3
|
15 // | Allan Hansen <ahØartemis*dk> |
|
rlm@3
|
16 // +----------------------------------------------------------------------+
|
rlm@3
|
17 // | module.archive.gzip.php |
|
rlm@3
|
18 // | module for analyzing GZIP files |
|
rlm@3
|
19 // | dependencies: PHP compiled with zlib support (optional) |
|
rlm@3
|
20 // +----------------------------------------------------------------------+
|
rlm@3
|
21 // | Module originally written by Mike Mozolin <teddybearØmail*ru> |
|
rlm@3
|
22 // +----------------------------------------------------------------------+
|
rlm@3
|
23 //
|
rlm@3
|
24 // $Id: module.archive.gzip.php,v 1.4 2006/12/04 16:00:35 ah Exp $
|
rlm@3
|
25
|
rlm@3
|
26
|
rlm@3
|
27
|
rlm@3
|
28 class getid3_gzip extends getid3_handler
|
rlm@3
|
29 {
|
rlm@3
|
30
|
rlm@3
|
31 // public: Optional file list - disable for speed.
|
rlm@3
|
32 public $option_gzip_parse_contents = true; // decode gzipped files, if possible, and parse recursively (.tar.gz for example)
|
rlm@3
|
33
|
rlm@3
|
34
|
rlm@3
|
35 // Reads the gzip-file
|
rlm@3
|
36 function Analyze() {
|
rlm@3
|
37
|
rlm@3
|
38 $info = &$this->getid3->info;
|
rlm@3
|
39
|
rlm@3
|
40 $info['fileformat'] = 'gzip';
|
rlm@3
|
41
|
rlm@3
|
42 $start_length = 10;
|
rlm@3
|
43 $unpack_header = 'a1id1/a1id2/a1cmethod/a1flags/a4mtime/a1xflags/a1os';
|
rlm@3
|
44
|
rlm@3
|
45 //+---+---+---+---+---+---+---+---+---+---+
|
rlm@3
|
46 //|ID1|ID2|CM |FLG| MTIME |XFL|OS |
|
rlm@3
|
47 //+---+---+---+---+---+---+---+---+---+---+
|
rlm@3
|
48
|
rlm@3
|
49 @fseek($this->getid3->fp, 0);
|
rlm@3
|
50 $buffer = @fread($this->getid3->fp, $info['filesize']);
|
rlm@3
|
51
|
rlm@3
|
52 $arr_members = explode("\x1F\x8B\x08", $buffer);
|
rlm@3
|
53
|
rlm@3
|
54 while (true) {
|
rlm@3
|
55 $is_wrong_members = false;
|
rlm@3
|
56 $num_members = intval(count($arr_members));
|
rlm@3
|
57 for ($i = 0; $i < $num_members; $i++) {
|
rlm@3
|
58 if (strlen($arr_members[$i]) == 0) {
|
rlm@3
|
59 continue;
|
rlm@3
|
60 }
|
rlm@3
|
61 $buf = "\x1F\x8B\x08".$arr_members[$i];
|
rlm@3
|
62
|
rlm@3
|
63 $attr = unpack($unpack_header, substr($buf, 0, $start_length));
|
rlm@3
|
64 if (!$this->get_os_type(ord($attr['os']))) {
|
rlm@3
|
65
|
rlm@3
|
66 // Merge member with previous if wrong OS type
|
rlm@3
|
67 $arr_members[$i - 1] .= $buf;
|
rlm@3
|
68 $arr_members[$i] = '';
|
rlm@3
|
69 $is_wrong_members = true;
|
rlm@3
|
70 continue;
|
rlm@3
|
71 }
|
rlm@3
|
72 }
|
rlm@3
|
73 if (!$is_wrong_members) {
|
rlm@3
|
74 break;
|
rlm@3
|
75 }
|
rlm@3
|
76 }
|
rlm@3
|
77
|
rlm@3
|
78 $fpointer = 0;
|
rlm@3
|
79 $idx = 0;
|
rlm@3
|
80 for ($i = 0; $i < $num_members; $i++) {
|
rlm@3
|
81 if (strlen($arr_members[$i]) == 0) {
|
rlm@3
|
82 continue;
|
rlm@3
|
83 }
|
rlm@3
|
84 $info_gzip_member_header_idx = &$info['gzip']['member_header'][++$idx];
|
rlm@3
|
85
|
rlm@3
|
86 $buff = "\x1F\x8B\x08".$arr_members[$i];
|
rlm@3
|
87
|
rlm@3
|
88 $attr = unpack($unpack_header, substr($buff, 0, $start_length));
|
rlm@3
|
89 $info_gzip_member_header_idx['filemtime'] = getid3_lib::LittleEndian2Int($attr['mtime']);
|
rlm@3
|
90 $info_gzip_member_header_idx['raw']['id1'] = ord($attr['cmethod']);
|
rlm@3
|
91 $info_gzip_member_header_idx['raw']['id2'] = ord($attr['cmethod']);
|
rlm@3
|
92 $info_gzip_member_header_idx['raw']['cmethod'] = ord($attr['cmethod']);
|
rlm@3
|
93 $info_gzip_member_header_idx['raw']['os'] = ord($attr['os']);
|
rlm@3
|
94 $info_gzip_member_header_idx['raw']['xflags'] = ord($attr['xflags']);
|
rlm@3
|
95 $info_gzip_member_header_idx['raw']['flags'] = ord($attr['flags']);
|
rlm@3
|
96
|
rlm@3
|
97 $info_gzip_member_header_idx['flags']['crc16'] = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x02);
|
rlm@3
|
98 $info_gzip_member_header_idx['flags']['extra'] = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x04);
|
rlm@3
|
99 $info_gzip_member_header_idx['flags']['filename'] = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x08);
|
rlm@3
|
100 $info_gzip_member_header_idx['flags']['comment'] = (bool) ($info_gzip_member_header_idx['raw']['flags'] & 0x10);
|
rlm@3
|
101
|
rlm@3
|
102 $info_gzip_member_header_idx['compression'] = $this->get_xflag_type($info_gzip_member_header_idx['raw']['xflags']);
|
rlm@3
|
103
|
rlm@3
|
104 $info_gzip_member_header_idx['os'] = $this->get_os_type($info_gzip_member_header_idx['raw']['os']);
|
rlm@3
|
105 if (!$info_gzip_member_header_idx['os']) {
|
rlm@3
|
106 $info['error'][] = 'Read error on gzip file';
|
rlm@3
|
107 return false;
|
rlm@3
|
108 }
|
rlm@3
|
109
|
rlm@3
|
110 $fpointer = 10;
|
rlm@3
|
111 $arr_xsubfield = array ();
|
rlm@3
|
112
|
rlm@3
|
113 // bit 2 - FLG.FEXTRA
|
rlm@3
|
114 //+---+---+=================================+
|
rlm@3
|
115 //| XLEN |...XLEN bytes of "extra field"...|
|
rlm@3
|
116 //+---+---+=================================+
|
rlm@3
|
117
|
rlm@3
|
118 if ($info_gzip_member_header_idx['flags']['extra']) {
|
rlm@3
|
119 $w_xlen = substr($buff, $fpointer, 2);
|
rlm@3
|
120 $xlen = getid3_lib::LittleEndian2Int($w_xlen);
|
rlm@3
|
121 $fpointer += 2;
|
rlm@3
|
122
|
rlm@3
|
123 $info_gzip_member_header_idx['raw']['xfield'] = substr($buff, $fpointer, $xlen);
|
rlm@3
|
124
|
rlm@3
|
125 // Extra SubFields
|
rlm@3
|
126 //+---+---+---+---+==================================+
|
rlm@3
|
127 //|SI1|SI2| LEN |... LEN bytes of subfield data ...|
|
rlm@3
|
128 //+---+---+---+---+==================================+
|
rlm@3
|
129
|
rlm@3
|
130 $idx = 0;
|
rlm@3
|
131 while (true) {
|
rlm@3
|
132 if ($idx >= $xlen) {
|
rlm@3
|
133 break;
|
rlm@3
|
134 }
|
rlm@3
|
135 $si1 = ord(substr($buff, $fpointer + $idx++, 1));
|
rlm@3
|
136 $si2 = ord(substr($buff, $fpointer + $idx++, 1));
|
rlm@3
|
137 if (($si1 == 0x41) && ($si2 == 0x70)) {
|
rlm@3
|
138 $w_xsublen = substr($buff, $fpointer+$idx, 2);
|
rlm@3
|
139 $xsublen = getid3_lib::LittleEndian2Int($w_xsublen);
|
rlm@3
|
140 $idx += 2;
|
rlm@3
|
141 $arr_xsubfield[] = substr($buff, $fpointer+$idx, $xsublen);
|
rlm@3
|
142 $idx += $xsublen;
|
rlm@3
|
143 } else {
|
rlm@3
|
144 break;
|
rlm@3
|
145 }
|
rlm@3
|
146 }
|
rlm@3
|
147 $fpointer += $xlen;
|
rlm@3
|
148 }
|
rlm@3
|
149
|
rlm@3
|
150 // bit 3 - FLG.FNAME
|
rlm@3
|
151 //+=========================================+
|
rlm@3
|
152 //|...original file name, zero-terminated...|
|
rlm@3
|
153 //+=========================================+
|
rlm@3
|
154 // GZIP files may have only one file, with no filename, so assume original filename is current filename without .gz
|
rlm@3
|
155
|
rlm@3
|
156 $info_gzip_member_header_idx['filename'] = eregi_replace('.gz$', '', @$info['filename']);
|
rlm@3
|
157 if ($info_gzip_member_header_idx['flags']['filename']) {
|
rlm@3
|
158 while (true) {
|
rlm@3
|
159 if (ord($buff[$fpointer]) == 0) {
|
rlm@3
|
160 $fpointer++;
|
rlm@3
|
161 break;
|
rlm@3
|
162 }
|
rlm@3
|
163 $info_gzip_member_header_idx['filename'] .= $buff[$fpointer];
|
rlm@3
|
164 $fpointer++;
|
rlm@3
|
165 }
|
rlm@3
|
166 }
|
rlm@3
|
167
|
rlm@3
|
168 // bit 4 - FLG.FCOMMENT
|
rlm@3
|
169 //+===================================+
|
rlm@3
|
170 //|...file comment, zero-terminated...|
|
rlm@3
|
171 //+===================================+
|
rlm@3
|
172
|
rlm@3
|
173 if ($info_gzip_member_header_idx['flags']['comment']) {
|
rlm@3
|
174 while (true) {
|
rlm@3
|
175 if (ord($buff[$fpointer]) == 0) {
|
rlm@3
|
176 $fpointer++;
|
rlm@3
|
177 break;
|
rlm@3
|
178 }
|
rlm@3
|
179 $info_gzip_member_header_idx['comment'] .= $buff[$fpointer];
|
rlm@3
|
180 $fpointer++;
|
rlm@3
|
181 }
|
rlm@3
|
182 }
|
rlm@3
|
183
|
rlm@3
|
184 // bit 1 - FLG.FHCRC
|
rlm@3
|
185 //+---+---+
|
rlm@3
|
186 //| CRC16 |
|
rlm@3
|
187 //+---+---+
|
rlm@3
|
188
|
rlm@3
|
189 if ($info_gzip_member_header_idx['flags']['crc16']) {
|
rlm@3
|
190 $w_crc = substr($buff, $fpointer, 2);
|
rlm@3
|
191 $info_gzip_member_header_idx['crc16'] = getid3_lib::LittleEndian2Int($w_crc);
|
rlm@3
|
192 $fpointer += 2;
|
rlm@3
|
193 }
|
rlm@3
|
194
|
rlm@3
|
195 // bit 0 - FLG.FTEXT
|
rlm@3
|
196 //if ($info_gzip_member_header_idx['raw']['flags'] & 0x01) {
|
rlm@3
|
197 // Ignored...
|
rlm@3
|
198 //}
|
rlm@3
|
199 // bits 5, 6, 7 - reserved
|
rlm@3
|
200
|
rlm@3
|
201 $info_gzip_member_header_idx['crc32'] = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 8, 4));
|
rlm@3
|
202 $info_gzip_member_header_idx['filesize'] = getid3_lib::LittleEndian2Int(substr($buff, strlen($buff) - 4));
|
rlm@3
|
203
|
rlm@3
|
204 if ($this->option_gzip_parse_contents) {
|
rlm@3
|
205
|
rlm@3
|
206 // Try to inflate GZip
|
rlm@3
|
207
|
rlm@3
|
208 if (!function_exists('gzinflate')) {
|
rlm@3
|
209 $this->getid3->warning('PHP does not have zlib support - contents not parsed.');
|
rlm@3
|
210 return true;
|
rlm@3
|
211 }
|
rlm@3
|
212
|
rlm@3
|
213 $csize = 0;
|
rlm@3
|
214 $inflated = '';
|
rlm@3
|
215 $chkcrc32 = '';
|
rlm@3
|
216
|
rlm@3
|
217 $cdata = substr($buff, $fpointer);
|
rlm@3
|
218 $cdata = substr($cdata, 0, strlen($cdata) - 8);
|
rlm@3
|
219 $csize = strlen($cdata);
|
rlm@3
|
220 $inflated = gzinflate($cdata);
|
rlm@3
|
221
|
rlm@3
|
222 // Calculate CRC32 for inflated content
|
rlm@3
|
223 $info_gzip_member_header_idx['crc32_valid'] = (bool) (sprintf('%u', crc32($inflated)) == $info_gzip_member_header_idx['crc32']);
|
rlm@3
|
224
|
rlm@3
|
225
|
rlm@3
|
226 //// Analyse contents
|
rlm@3
|
227
|
rlm@3
|
228 // write content to temp file
|
rlm@3
|
229 if (($temp_file_name = tempnam('*', 'getID3')) === false) {
|
rlm@3
|
230 throw new getid3_exception('Unable to create temporary file.');
|
rlm@3
|
231 }
|
rlm@3
|
232
|
rlm@3
|
233 if ($tmp = fopen($temp_file_name, 'wb')) {
|
rlm@3
|
234 fwrite($tmp, $inflated);
|
rlm@3
|
235 fclose($tmp);
|
rlm@3
|
236
|
rlm@3
|
237 // clone getid3 - we want same settings
|
rlm@3
|
238 $clone = clone $this->getid3;
|
rlm@3
|
239 unset($clone->info);
|
rlm@3
|
240 try {
|
rlm@3
|
241 $clone->Analyze($temp_file_name);
|
rlm@3
|
242 $info_gzip_member_header_idx['parsed_content'] = $clone->info;
|
rlm@3
|
243 }
|
rlm@3
|
244 catch (getid3_exception $e) {
|
rlm@3
|
245 // unable to parse contents
|
rlm@3
|
246 }
|
rlm@3
|
247
|
rlm@3
|
248 unlink($temp_file_name);
|
rlm@3
|
249 }
|
rlm@3
|
250
|
rlm@3
|
251 // Unknown/unhandled format
|
rlm@3
|
252 else {
|
rlm@3
|
253
|
rlm@3
|
254 }
|
rlm@3
|
255 }
|
rlm@3
|
256 }
|
rlm@3
|
257 return true;
|
rlm@3
|
258 }
|
rlm@3
|
259
|
rlm@3
|
260
|
rlm@3
|
261 // Converts the OS type
|
rlm@3
|
262 public static function get_os_type($key) {
|
rlm@3
|
263 static $os_type = array (
|
rlm@3
|
264 '0' => 'FAT filesystem (MS-DOS, OS/2, NT/Win32)',
|
rlm@3
|
265 '1' => 'Amiga',
|
rlm@3
|
266 '2' => 'VMS (or OpenVMS)',
|
rlm@3
|
267 '3' => 'Unix',
|
rlm@3
|
268 '4' => 'VM/CMS',
|
rlm@3
|
269 '5' => 'Atari TOS',
|
rlm@3
|
270 '6' => 'HPFS filesystem (OS/2, NT)',
|
rlm@3
|
271 '7' => 'Macintosh',
|
rlm@3
|
272 '8' => 'Z-System',
|
rlm@3
|
273 '9' => 'CP/M',
|
rlm@3
|
274 '10' => 'TOPS-20',
|
rlm@3
|
275 '11' => 'NTFS filesystem (NT)',
|
rlm@3
|
276 '12' => 'QDOS',
|
rlm@3
|
277 '13' => 'Acorn RISCOS',
|
rlm@3
|
278 '255' => 'unknown'
|
rlm@3
|
279 );
|
rlm@3
|
280 return @$os_type[$key];
|
rlm@3
|
281 }
|
rlm@3
|
282
|
rlm@3
|
283
|
rlm@3
|
284 // Converts the eXtra FLags
|
rlm@3
|
285 public static function get_xflag_type($key) {
|
rlm@3
|
286 static $xflag_type = array (
|
rlm@3
|
287 '0' => 'unknown',
|
rlm@3
|
288 '2' => 'maximum compression',
|
rlm@3
|
289 '4' => 'fastest algorithm'
|
rlm@3
|
290 );
|
rlm@3
|
291 return @$xflag_type[$key];
|
rlm@3
|
292 }
|
rlm@3
|
293
|
rlm@3
|
294 }
|
rlm@3
|
295
|
rlm@3
|
296 ?> |