2 rcs_id('$Id: ziplib.php,v 1.11 2001-12-12 05:12:46 carstenklapp Exp $');
7 * Note that we use gzopen()/gzwrite() instead of gzcompress() even
8 * if gzcompress() is available. Gzcompress() puts out data with different
9 * headers --- in particular it includes an "adler-32" checksum rather than
10 * a "CRC32" checksum. Since we need the CRC-32 checksum, and since not
11 * all PHP's have gzcompress(), we'll just stick with gzopen().
13 function gzip_cleanup () {
17 @unlink($gzip_tmpfile);
20 function gzip_tempnam () {
25 //FIXME: does this work on non-unix machines?
26 $gzip_tmpfile = tempnam("/tmp", "wkzip");
27 register_shutdown_function("gzip_cleanup");
32 function gzip_compress ($data) {
33 $filename = gzip_tempnam();
34 if (!($fp = gzopen($filename, "wb")))
35 ExitWiki("gzopen failed");
36 gzwrite($fp, $data, strlen($data));
38 ExitWiki("gzclose failed");
40 $size = filesize($filename);
42 if (!($fp = fopen($filename, "rb")))
43 ExitWiki("fopen failed");
44 if (!($z = fread($fp, $size)) || strlen($z) != $size)
45 ExitWiki("fread failed");
47 ExitWiki("fclose failed");
53 function gzip_uncompress ($data) {
54 $filename = gzip_tempnam();
55 if (!($fp = fopen($filename, "wb")))
56 ExitWiki("fopen failed");
57 fwrite($fp, $data, strlen($data));
59 ExitWiki("fclose failed");
61 if (!($fp = gzopen($filename, "rb")))
62 ExitWiki("gzopen failed");
64 while ($buf = gzread($fp, 4096))
67 ExitWiki("gzclose failed");
74 * CRC32 computation. Hacked from Info-zip's zip-2.3 source code.
77 function zip_crc32 ($str, $crc = 0)
79 static $zip_crc_table;
81 if (empty($zip_crc_table)) {
82 /* NOTE: The range of PHP ints seems to be -0x80000000 to 0x7fffffff.
83 * So, had to munge these constants.
86 = array (0x00000000, 0x77073096, -0x11f19ed4, -0x66f6ae46, 0x076dc419,
87 0x706af48f, -0x169c5acb, -0x619b6a5d, 0x0edb8832, 0x79dcb8a4,
88 -0x1f2a16e2, -0x682d2678, 0x09b64c2b, 0x7eb17cbd, -0x1847d2f9,
89 -0x6f40e26f, 0x1db71064, 0x6ab020f2, -0x0c468eb8, -0x7b41be22,
90 0x1adad47d, 0x6ddde4eb, -0x0b2b4aaf, -0x7c2c7a39, 0x136c9856,
91 0x646ba8c0, -0x029d0686, -0x759a3614, 0x14015c4f, 0x63066cd9,
92 -0x05f0c29d, -0x72f7f20b, 0x3b6e20c8, 0x4c69105e, -0x2a9fbe1c,
93 -0x5d988e8e, 0x3c03e4d1, 0x4b04d447, -0x2df27a03, -0x5af54a95,
94 0x35b5a8fa, 0x42b2986c, -0x2444362a, -0x534306c0, 0x32d86ce3,
95 0x45df5c75, -0x2329f231, -0x542ec2a7, 0x26d930ac, 0x51de003a,
96 -0x3728ae80, -0x402f9eea, 0x21b4f4b5, 0x56b3c423, -0x30456a67,
97 -0x47425af1, 0x2802b89e, 0x5f058808, -0x39f3264e, -0x4ef416dc,
98 0x2f6f7c87, 0x58684c11, -0x3e9ee255, -0x4999d2c3, 0x76dc4190,
99 0x01db7106, -0x672ddf44, -0x102aefd6, 0x71b18589, 0x06b6b51f,
100 -0x60401b5b, -0x17472bcd, 0x7807c9a2, 0x0f00f934, -0x69f65772,
101 -0x1ef167e8, 0x7f6a0dbb, 0x086d3d2d, -0x6e9b9369, -0x199ca3ff,
102 0x6b6b51f4, 0x1c6c6162, -0x7a9acf28, -0x0d9dffb2, 0x6c0695ed,
103 0x1b01a57b, -0x7df70b3f, -0x0af03ba9, 0x65b0d9c6, 0x12b7e950,
104 -0x74414716, -0x03467784, 0x62dd1ddf, 0x15da2d49, -0x732c830d,
105 -0x042bb39b, 0x4db26158, 0x3ab551ce, -0x5c43ff8c, -0x2b44cf1e,
106 0x4adfa541, 0x3dd895d7, -0x5b2e3b93, -0x2c290b05, 0x4369e96a,
107 0x346ed9fc, -0x529877ba, -0x259f4730, 0x44042d73, 0x33031de5,
108 -0x55f5b3a1, -0x22f28337, 0x5005713c, 0x270241aa, -0x41f4eff0,
109 -0x36f3df7a, 0x5768b525, 0x206f85b3, -0x46992bf7, -0x319e1b61,
110 0x5edef90e, 0x29d9c998, -0x4f2f67de, -0x3828574c, 0x59b33d17,
111 0x2eb40d81, -0x4842a3c5, -0x3f459353, -0x12477ce0, -0x65404c4a,
112 0x03b6e20c, 0x74b1d29a, -0x152ab8c7, -0x622d8851, 0x04db2615,
113 0x73dc1683, -0x1c9cf4ee, -0x6b9bc47c, 0x0d6d6a3e, 0x7a6a5aa8,
114 -0x1bf130f5, -0x6cf60063, 0x0a00ae27, 0x7d079eb1, -0x0ff06cbc,
115 -0x78f75c2e, 0x1e01f268, 0x6906c2fe, -0x089da8a3, -0x7f9a9835,
116 0x196c3671, 0x6e6b06e7, -0x012be48a, -0x762cd420, 0x10da7a5a,
117 0x67dd4acc, -0x06462091, -0x71411007, 0x17b7be43, 0x60b08ed5,
118 -0x29295c18, -0x5e2e6c82, 0x38d8c2c4, 0x4fdff252, -0x2e44980f,
119 -0x5943a899, 0x3fb506dd, 0x48b2364b, -0x27f2d426, -0x50f5e4b4,
120 0x36034af6, 0x41047a60, -0x209f103d, -0x579820ab, 0x316e8eef,
121 0x4669be79, -0x349e4c74, -0x43997ce6, 0x256fd2a0, 0x5268e236,
122 -0x33f3886b, -0x44f4b8fd, 0x220216b9, 0x5505262f, -0x3a45c442,
123 -0x4d42f4d8, 0x2bb45a92, 0x5cb36a04, -0x3d280059, -0x4a2f30cf,
124 0x2cd99e8b, 0x5bdeae1d, -0x649b3d50, -0x139c0dda, 0x756aa39c,
125 0x026d930a, -0x63f6f957, -0x14f1c9c1, 0x72076785, 0x05005713,
126 -0x6a40b57e, -0x1d4785ec, 0x7bb12bae, 0x0cb61b38, -0x6d2d7165,
127 -0x1a2a41f3, 0x7cdcefb7, 0x0bdbdf21, -0x792c2d2c, -0x0e2b1dbe,
128 0x68ddb3f8, 0x1fda836e, -0x7e41e933, -0x0946d9a5, 0x6fb077e1,
129 0x18b74777, -0x77f7a51a, -0x00f09590, 0x66063bca, 0x11010b5c,
130 -0x709a6101, -0x079d5197, 0x616bffd3, 0x166ccf45, -0x5ff51d88,
131 -0x28f22d12, 0x4e048354, 0x3903b3c2, -0x5898d99f, -0x2f9fe909,
132 0x4969474d, 0x3e6e77db, -0x512e95b6, -0x2629a524, 0x40df0b66,
133 0x37d83bf0, -0x564351ad, -0x2144613b, 0x47b2cf7f, 0x30b5ffe9,
134 -0x42420de4, -0x35453d76, 0x53b39330, 0x24b4a3a6, -0x452fc9fb,
135 -0x3228f96d, 0x54de5729, 0x23d967bf, -0x4c9985d2, -0x3b9eb548,
136 0x5d681b02, 0x2a6f2b94, -0x4bf441c9, -0x3cf3715f, 0x5a05df1b,
141 for ($i = 0; $i < strlen($str); $i++) {
142 $crc = ( $zip_crc_table[($crc ^ ord($str[$i])) & 0xff]
143 ^ (($crc >> 8) & 0xffffff) );
148 define('GZIP_MAGIC', "\037\213");
149 define('GZIP_DEFLATE', 010);
151 function zip_deflate ($content)
153 // Compress content, and suck information from gzip header.
154 $z = gzip_compress($content);
156 // Suck OS type byte from gzip header. FIXME: this smells bad.
158 extract(unpack("a2magic/Ccomp_type/Cflags/@9/Cos_type", $z));
160 if ($magic != GZIP_MAGIC)
161 ExitWiki("Bad gzip magic");
162 if ($comp_type != GZIP_DEFLATE)
163 ExitWiki("Bad gzip comp type");
164 if (($flags & 0x3e) != 0)
165 ExitWiki(sprintf("Bad flags (0x%02x)", $flags));
168 $gz_data_len = strlen($z) - $gz_header_len - 8;
169 if ($gz_data_len < 0)
170 ExitWiki("not enough gzip output?");
172 extract(unpack("Vcrc32", substr($z, $gz_header_len + $gz_data_len)));
174 return array(substr($z, $gz_header_len, $gz_data_len), // gzipped data
180 function zip_inflate ($data, $crc32, $uncomp_size)
182 if (!function_exists('gzopen'))
183 ExitWiki("Can't inflate data: zlib support not enabled in this PHP");
185 // Reconstruct gzip header and ungzip the data.
186 $mtime = time(); //(Bogus mtime)
188 return gzip_uncompress( pack("a2CxV@10", GZIP_MAGIC, GZIP_DEFLATE, $mtime)
190 . pack("VV", $crc32, $uncomp_size) );
193 function unixtime2dostime ($unix_time) {
195 $unix_time++; // Round up to even seconds.
197 list ($year,$month,$mday,$hour,$min,$sec)
198 = explode(" ", date("Y n j G i s", $unix_time));
201 list ($year,$month,$mday,$hour,$min,$sec) = array(1980, 1, 1, 0, 0, 0);
203 $dosdate = (($year - 1980) << 9) | ($month << 5) | $mday;
204 $dostime = ($hour << 11) | ($min << 5) | ($sec >> 1);
206 return array($dosdate, $dostime);
209 function dostime2unixtime ($dosdate, $dostime) {
210 $mday = $dosdate & 0x1f;
211 $month = ($dosdate >> 5) & 0x0f;
212 $year = 1980 + (($dosdate >> 9) & 0x7f);
214 $sec = ($dostime & 0x1f) * 2;
215 $min = ($dostime >> 5) & 0x3f;
216 $hour = ($dostime >> 11) & 0x1f;
218 return mktime($hour, $min, $sec, $month, $mday, $year);
223 * Class for zipfile creation.
225 define('ZIP_DEFLATE', GZIP_DEFLATE);
226 define('ZIP_STORE', 0);
227 define('ZIP_CENTHEAD_MAGIC', "PK\001\002");
228 define('ZIP_LOCHEAD_MAGIC', "PK\003\004");
229 define('ZIP_ENDDIR_MAGIC', "PK\005\006");
233 function ZipWriter ($comment = "", $zipname = "archive.zip") {
234 $this->comment = $comment;
236 $this->dir = ""; // "Central directory block"
237 $this->offset = 0; // Current file position.
239 $zipname = addslashes($zipname);
240 header("Content-Type: application/zip; name=\"$zipname\"");
241 header("Content-Disposition: attachment; filename=\"$zipname\"");
244 function addRegularFile ($filename, $content, $attrib = false) {
248 $size = strlen($content);
249 if (function_exists('gzopen'))
251 list ($data, $crc32, $os_type) = zip_deflate($content);
252 if (strlen($data) < $size)
254 $content = $data; // Use compressed data.
255 $comp_type = ZIP_DEFLATE;
258 unset($crc32); // force plain store.
262 $comp_type = ZIP_STORE;
263 $crc32 = zip_crc32($content);
266 if (!empty($attrib['write_protected']))
267 $atx = (0100444 << 16) | 1; // S_IFREG + read permissions to everybody.
269 $atx = (0100644 << 16); // Add owner write perms.
271 $ati = $attrib['is_ascii'] ? 1 : 0;
273 if (empty($attrib['mtime']))
274 $attrib['mtime'] = time();
275 list ($mod_date, $mod_time) = unixtime2dostime($attrib['mtime']);
277 // Construct parts common to "Local file header" and "Central
278 // directory file header."
279 if (!isset($attrib['extra_field']))
280 $attrib['extra_field'] = '';
281 if (!isset($attrib['file_comment']))
282 $attrib['file_comment'] = '';
284 $head = pack("vvvvvVVVvv",
285 20, // Version needed to extract (FIXME: is this right?)
286 0, // Gen purp bit flag
294 strlen($attrib['extra_field']));
296 // Construct the "Local file header"
297 $lheader = ZIP_LOCHEAD_MAGIC . $head . $filename . $attrib['extra_field'];
299 // Construct the "central directory file header"
300 $this->dir .= pack("a4CC",
302 23, // Version made by (FIXME: is this right?)
305 $this->dir .= pack("vvvVV",
306 strlen($attrib['file_comment']),
307 0, // Disk number start
308 $ati, // Internal file attributes
309 $atx, // External file attributes
310 $this->offset); // Relative offset of local header
311 $this->dir .= $filename . $attrib['extra_field'] . $attrib['file_comment'];
313 // Output the "Local file header" and file contents.
317 $this->offset += strlen($lheader) + strlen($content);
322 // Output the central directory
325 // Construct the "End of central directory record"
326 echo ZIP_ENDDIR_MAGIC;
328 0, // Number of this disk.
329 0, // Number of disk with start of c dir
330 $this->nfiles, // Number entries on this disk
331 $this->nfiles, // Number entries
332 strlen($this->dir), // Size of central directory
333 $this->offset, // Offset of central directory
334 strlen($this->comment));
341 * Class for reading zip files.
345 * Many of the ExitWiki()'s should probably be warn()'s (eg. CRC mismatch).
347 * Only a subset of zip formats is recognized. (I think that unsupported
348 * formats will be recognized as such rather than silently munged.)
350 * We don't read the central directory. This means we don't see the
351 * file attributes (text? read-only?), or file comments.
353 * Right now we ignore the file mod date and time, since we don't need it.
357 function ZipReader ($zipfile) {
358 if (!is_string($zipfile))
359 $this->fp = $zipfile; // File already open
360 else if (!($this->fp = fopen($zipfile, "rb")))
361 ExitWiki("Can't open zip file '$zipfile' for reading");
364 function _read ($nbytes) {
365 $chunk = fread($this->fp, $nbytes);
366 if (strlen($chunk) != $nbytes)
367 ExitWiki("Unexpected EOF in zip file");
376 function readFile () {
377 $head = $this->_read(30);
379 extract(unpack("a4magic/vreq_version/vflags/vcomp_type"
380 . "/vmod_time/vmod_date"
381 . "/Vcrc32/Vcomp_size/Vuncomp_size"
382 . "/vfilename_len/vextrafld_len",
385 //FIXME: we should probably check $req_version.
386 $attrib['mtime'] = dostime2unixtime($mod_date, $mod_time);
388 if ($magic != ZIP_LOCHEAD_MAGIC)
390 if ($magic != ZIP_CENTHEAD_MAGIC)
391 ExitWiki("Bad header type: " . htmlspecialchars($magic)); // FIXME: better message?
392 return $this->done();
394 if (($flags & 0x21) != 0)
395 ExitWiki("Encryption and/or zip patches not supported.");
396 if (($flags & 0x08) != 0)
397 ExitWiki("Postponed CRC not yet supported."); // FIXME: ???
399 $filename = $this->_read($filename_len);
400 if ($extrafld_len != 0)
401 $attrib['extra_field'] = $this->_read($extrafld_len);
403 $data = $this->_read($comp_size);
405 if ($comp_type == ZIP_DEFLATE)
407 $data = zip_inflate($data, $crc32, $uncomp_size);
409 else if ($comp_type == ZIP_STORE)
411 $crc = zip_crc32($data);
413 ExitWiki(sprintf("CRC mismatch %x != %x", $crc, $crc32));
416 ExitWiki("Compression method $comp_method unsupported");
418 if (strlen($data) != $uncomp_size)
419 ExitWiki(sprintf("Uncompressed size mismatch %d != %d",
420 strlen($data), $uncomp_size));
422 return array($filename, $data, $attrib);
427 * Routines for Mime mailification of pages.
429 //FIXME: these should go elsewhere (libmime?).
432 * Routines for quoted-printable en/decoding.
434 function QuotedPrintableEncode ($string)
436 // Quote special characters in line.
440 // The complicated regexp is to force quoting of trailing spaces.
441 preg_match('/^([ !-<>-~]*)(?:([!-<>-~]$)|(.))/s', $string, $match);
442 $quoted .= $match[1] . $match[2];
443 if (!empty($match[3]))
444 $quoted .= sprintf("=%02X", ord($match[3]));
445 $string = substr($string, strlen($match[0]));
448 // This splits the line (preferably after white-space) into lines
449 // which are no longer than 76 chars (after adding trailing '=' for
450 // soft line break, but before adding \r\n.)
451 return preg_replace('/(?=.{77})(.{10,74}[ \t]|.{71,73}[^=][^=])/s',
452 "\\1=\r\n", $quoted);
455 function QuotedPrintableDecode ($string)
457 // Eliminate soft line-breaks.
458 $string = preg_replace('/=[ \t\r]*\n/', '', $string);
459 return quoted_printable_decode($string);
462 define('MIME_TOKEN_REGEXP', "[-!#-'*+.0-9A-Z^-~]+");
464 function MimeContentTypeHeader ($type, $subtype, $params)
466 $header = "Content-Type: $type/$subtype";
468 while (list($key, $val) = each($params))
470 //FIXME: what about non-ascii printables in $val?
471 if (!preg_match('/^' . MIME_TOKEN_REGEXP . '$/', $val))
472 $val = '"' . addslashes($val) . '"';
473 $header .= ";\r\n $key=$val";
475 return "$header\r\n";
478 function MimeMultipart ($parts)
480 global $mime_multipart_count;
482 // The string "=_" can not occur in quoted-printable encoded data.
483 $boundary = "=_multipart_boundary_" . ++$mime_multipart_count;
485 $head = MimeContentTypeHeader('multipart', 'mixed',
486 array('boundary' => $boundary));
488 $sep = "\r\n--$boundary\r\n";
490 return $head . $sep . implode($sep, $parts) . "\r\n--${boundary}--\r\n";
493 function MimeifyPageRevision ($revision) {
494 $page = $revision->getPage();
495 // FIXME: add 'hits' to $params
496 $params = array('pagename' => rawurlencode($page->getName()),
497 'author' => rawurlencode($revision->get('author')),
498 'version' => $revision->getVersion(),
500 'lastmodified' => $revision->get('mtime'));
501 if ($page->get('mtime'))
502 $params['created'] = $page->get('mtime');
503 if ($page->get('locked'))
504 $params['flags'] = 'PAGE_LOCKED';
505 if ($revision->get('author_id'))
506 $params['author_id'] = $revision->get('author_id');
509 $out = MimeContentTypeHeader('application', 'x-phpwiki', $params);
510 $out .= "Content-Transfer-Encoding: quoted-printable\r\n";
513 foreach ($revision->getContent() as $line) {
514 $out .= QuotedPrintableEncode(chop($line)) . "\r\n";
520 * Routines for parsing Mime-ified phpwiki pages.
522 function ParseRFC822Headers (&$string)
524 if (preg_match("/^From (.*)\r?\n/", $string, $match))
526 $headers['from '] = preg_replace('/^\s+|\s+$/', '', $match[1]);
527 $string = substr($string, strlen($match[0]));
530 while (preg_match('/^([!-9;-~]+) [ \t]* : [ \t]* '
531 . '( .* \r?\n (?: [ \t] .* \r?\n)* )/x',
534 $headers[strtolower($match[1])]
535 = preg_replace('/^\s+|\s+$/', '', $match[2]);
536 $string = substr($string, strlen($match[0]));
542 if (! preg_match("/^\r?\n/", $string, $match))
544 // No blank line after headers.
548 $string = substr($string, strlen($match[0]));
554 function ParseMimeContentType ($string)
556 // FIXME: Remove (RFC822 style comments).
559 if (!preg_match(':^\s*(' . MIME_TOKEN_REGEXP . ')\s*'
561 . '\s*(' . MIME_TOKEN_REGEXP . ')\s*:x',
563 ExitWiki ("Bad MIME content-type");
565 $type = strtolower($match[1]);
566 $subtype = strtolower($match[2]);
567 $string = substr($string, strlen($match[0]));
571 while (preg_match('/^;\s*(' . MIME_TOKEN_REGEXP . ')\s*=\s*'
572 . '(?:(' . MIME_TOKEN_REGEXP . ')|"((?:[^"\\\\]|\\.)*)") \s*/sx',
575 if (strlen($match[2]))
578 $val = preg_replace('/[\\\\](.)/s', '\\1', $match[3]);
580 $param[strtolower($match[1])] = $val;
582 $string = substr($string, strlen($match[0]));
585 return array($type, $subtype, $param);
588 function ParseMimeMultipart($data, $boundary)
591 ExitWiki("No boundary?");
593 $boundary = preg_quote($boundary);
595 while (preg_match("/^(|.*?\n)--$boundary((?:--)?)[^\n]*\n/s",
598 $data = substr($data, strlen($match[0]));
599 if ( ! isset($parts) )
600 $parts = array(); // First time through: discard leading chaff
603 if ($content = ParseMimeifiedPages($match[1]))
604 for (reset($content); $p = current($content); next($content))
609 return $parts; // End boundary found.
611 ExitWiki("No end boundary?");
614 function GenerateFootnotesFromRefs($params)
616 $footnotes = array();
618 while (list($p, $reference) = each($params))
620 if (preg_match('/^ref([1-9][0-9]*)$/', $p, $m))
621 $footnotes[$m[1]] = sprintf(gettext ("[%d] See [%s]"),
622 $m[1], rawurldecode($reference));
625 if (sizeof($footnotes) > 0)
629 . gettext ("!References") . "\n"
630 . join("\n%%%\n", $footnotes) . "\n";
637 // Convert references in meta-data to footnotes.
638 // Only zip archives generated by phpwiki 1.2.x or earlier should have
640 function ParseMimeifiedPages ($data)
642 if (!($headers = ParseRFC822Headers($data))
643 || empty($headers['content-type'])) {
644 //trigger_error("Can't find content-type header", E_USER_WARNING);
647 $typeheader = $headers['content-type'];
649 if (!(list ($type, $subtype, $params) = ParseMimeContentType($typeheader))) {
650 trigger_error("Can't parse content-type: ("
651 . htmlspecialchars($typeheader) . ")", E_USER_WARNING);
654 if ("$type/$subtype" == 'multipart/mixed') {
655 return ParseMimeMultipart($data, $params['boundary']);
657 else if ("$type/$subtype" != 'application/x-phpwiki') {
658 trigger_error("Bad content-type: $type/$subtype", E_USER_WARNING);
662 // FIXME: more sanity checking?
665 $versiondata = array();
667 foreach ($params as $key => $value) {
670 $value = rawurldecode($value);
674 $page[$key] = $value;
677 if (preg_match('/PAGE_LOCKED/', $value))
678 $pagedata['locked'] = 'yes';
681 $pagedata[$key] = $value;
684 $versiondata['mtime'] = $value;
688 $versiondata[$key] = $value;
693 // FIXME: do we need to try harder to find a pagename if we
694 // haven't got one yet?
695 if (!isset($versiondata['author'])) {
696 $versiondata['author'] = $GLOBALS['user']->id();
699 $encoding = strtolower($headers['content-transfer-encoding']);
700 if ($encoding == 'quoted-printable')
701 $data = QuotedPrintableDecode($data);
702 else if ($encoding && $encoding != 'binary')
703 ExitWiki("Unknown encoding type: $encoding");
705 $data .= GenerateFootnotesFromRefs($params);
707 $page['content'] = preg_replace('/[ \t\r]*\n/', "\n", chop($data));
708 $page['pagedata'] = $pagedata;
709 $page['versiondata'] = $versiondata;
718 // c-hanging-comment-ender-p: nil
719 // indent-tabs-mode: nil