2 // PukiWiki - Yet another WikiWikiWeb clone.
3 // $Id: file.php,v 1.46 2006/01/12 01:00:51 teanan Exp $
5 // 2002-2005 PukiWiki Developers Team
6 // 2001-2002 Originally written by yu-ji
7 // License: GPL v2 or (at your option) any later version
9 // File related functions
11 // Get source(wiki text) data of the page
12 function get_source($page = NULL, $lock = TRUE)
17 $path = get_filename($page);
20 $fp = @fopen($path, 'r');
21 if ($fp == FALSE) return $array;
25 // Removing line-feeds: Because file() doesn't remove them.
26 $array = str_replace("\r", '', file($path));
37 // Get last-modified filetime of the page
38 function get_filetime($page)
40 return is_page($page) ? filemtime(get_filename($page)) - LOCALZONE : 0;
43 // Get physical file name of the page
44 function get_filename($page)
46 return DATA_DIR . encode($page) . '.txt';
49 // Put a data(wiki text) into a physical file(diff, backup, text)
50 function page_write($page, $postdata, $notimestamp = FALSE)
54 if (PKWK_READONLY) return; // Do nothing
56 $postdata = make_str_rules($postdata);
58 // Create and write diff
59 $oldpostdata = is_page($page) ? join('', get_source($page)) : '';
60 $diffdata = do_diff($oldpostdata, $postdata);
61 file_write(DIFF_DIR, $page, $diffdata);
64 make_backup($page, $postdata == ''); // Is $postdata null?
67 file_write(DATA_DIR, $page, $postdata, $notimestamp);
71 $_diff = explode("\n", $diffdata);
72 $plus = join("\n", preg_replace('/^\+/', '', preg_grep('/^\+/', $_diff)));
73 $minus = join("\n", preg_replace('/^-/', '', preg_grep('/^-/', $_diff)));
74 tb_send($page, $plus, $minus);
80 // Modify original text with user-defined / system-defined rules
81 function make_str_rules($source)
83 global $str_rules, $fixed_heading_anchor;
85 $lines = explode("\n", $source);
86 $count = count($lines);
91 for ($i = 0; $i < $count; $i++) {
92 $line = & $lines[$i]; // Modify directly
94 // Ignore null string and preformatted texts
95 if ($line == '' || $line{0} == ' ' || $line{0} == "\t") continue;
99 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
101 preg_match('/#[^{]*(\{\{+)\s*$/', $line, $matches)) {
102 // Multiline convert plugin start
104 $multiline = strlen($matches[1]); // Set specific number
107 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
109 preg_match('/^\}{' . $multiline . '}\s*$/', $line)) {
110 // Multiline convert plugin end
115 if ($modify === FALSE) continue;
117 // Replace with $str_rules
118 foreach ($str_rules as $pattern => $replacement)
119 $line = preg_replace('/' . $pattern . '/', $replacement, $line);
121 // Adding fixed anchor into headings
122 if ($fixed_heading_anchor &&
123 preg_match('/^(\*{1,3}.*?)(?:\[#([A-Za-z][\w-]*)\]\s*)?$/', $line, $matches) &&
124 (! isset($matches[2]) || $matches[2] == '')) {
125 // Generate unique id
126 $anchor = generate_fixed_heading_anchor_id($matches[1]);
127 $line = rtrim($matches[1]) . ' [#' . $anchor . ']';
131 // Multiline part has no stopper
132 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
133 $modify === FALSE && $multiline != 0)
134 $lines[] = str_repeat('}', $multiline);
136 return implode("\n", $lines);
140 function generate_fixed_heading_anchor_id($seed)
142 // A random alphabetic letter + 7 letters of random strings from md()
143 return chr(mt_rand(ord('a'), ord('z'))) .
144 substr(md5(uniqid(substr($seed, 0, 100), TRUE)),
149 function file_write($dir, $page, $str, $notimestamp = FALSE)
151 global $update_exec, $_msg_invalidiwn, $notify, $notify_diff_only, $notify_subject;
152 global $whatsdeleted, $maxshow_deleted;
154 if (PKWK_READONLY) return; // Do nothing
156 if (! is_pagename($page))
157 die_message(str_replace('$1', htmlspecialchars($page),
158 str_replace('$2', 'WikiName', $_msg_invalidiwn)));
160 $page = strip_bracket($page);
161 $file = $dir . encode($page) . '.txt';
165 if ($dir == DATA_DIR && file_exists($file)) {
168 add_recent($page, $whatsdeleted, '', $maxshow_deleted); // RecentDeleted
171 // File replacement (Edit)
172 $str = rtrim(preg_replace('/' . "\r" . '/', '', $str)) . "\n";
174 if ($notimestamp && file_exists($file))
175 $timestamp = filemtime($file) - LOCALZONE;
177 $fp = fopen($file, 'a') or die('fopen() failed: ' .
178 htmlspecialchars(basename($dir) . '/' . encode($page) . '.txt') .
180 'Maybe permission is not writable or filename is too long');
181 set_file_buffer($fp, 0);
194 if ($timestamp) pkwk_touch_file($file, $timestamp + LOCALZONE);
197 // Clear is_page() cache
198 is_page($page, TRUE);
200 if (! $timestamp && $dir == DATA_DIR)
203 // Execute $update_exec here
204 if ($update_exec && $dir == DATA_DIR)
205 system($update_exec . ' > /dev/null &');
207 if ($notify && $dir == DIFF_DIR) {
208 if ($notify_diff_only) $str = preg_replace('/^[^-+].*\n/m', '', $str);
210 $footer['ACTION'] = 'Page update';
211 $footer['PAGE'] = & $page;
212 $footer['URI'] = get_script_uri() . '?' . rawurlencode($page);
213 $footer['USER_AGENT'] = TRUE;
214 $footer['REMOTE_ADDR'] = TRUE;
216 pkwk_mail_notify($notify_subject, $str, $footer) or
217 die('pkwk_mail_notify(): Failed');
221 // Update RecentDeleted
222 function add_recent($page, $recentpage, $subject = '', $limit = 0)
224 if (PKWK_READONLY || $limit == 0 || $page == '' || $recentpage == '' ||
225 check_non_list($page)) return;
228 $lines = $matches = array();
229 foreach (get_source($recentpage) as $line)
230 if (preg_match('/^-(.+) - (\[\[.+\]\])$/', $line, $matches))
231 $lines[$matches[2]] = $line;
233 $_page = '[[' . $page . ']]';
235 // Remove a report about the same page
236 if (isset($lines[$_page])) unset($lines[$_page]);
239 array_unshift($lines, '-' . format_date(UTIME) . ' - ' . $_page .
240 htmlspecialchars($subject) . "\n");
242 // Get latest $limit reports
243 $lines = array_splice($lines, 0, $limit);
246 $fp = fopen(get_filename($recentpage), 'w') or
247 die_message('Cannot write page file ' .
248 htmlspecialchars($recentpage) .
249 '<br />Maybe permission is not writable or filename is too long');
250 set_file_buffer($fp, 0);
253 fputs($fp, '#freeze' . "\n");
254 fputs($fp, '#norelated' . "\n"); // :)
255 fputs($fp, join('', $lines));
260 // Update RecentChanges
261 function put_lastmodified()
263 global $maxshow, $whatsnew, $autolink;
265 if (PKWK_READONLY) return; // Do nothing
267 $pages = get_existpages();
268 $recent_pages = array();
269 foreach($pages as $page)
270 if ($page != $whatsnew && ! check_non_list($page))
271 $recent_pages[$page] = get_filetime($page);
273 // Sort decending order of last-modification date
274 arsort($recent_pages, SORT_NUMERIC);
276 // Create recent.dat (for recent.inc.php)
277 $fp = fopen(CACHE_DIR . 'recent.dat', 'w') or
278 die_message('Cannot write cache file ' .
279 CACHE_DIR . 'recent.dat' .
280 '<br />Maybe permission is not writable or filename is too long');
282 set_file_buffer($fp, 0);
285 foreach ($recent_pages as $page=>$time)
286 fputs($fp, $time . "\t" . $page . "\n");
290 // Create RecentChanges
291 $fp = fopen(get_filename($whatsnew), 'w') or
292 die_message('Cannot write page file ' .
293 htmlspecialchars($whatsnew) .
294 '<br />Maybe permission is not writable or filename is too long');
296 set_file_buffer($fp, 0);
300 // BugTrack2/106: Only variables can be passed by reference from PHP 5.0.5
301 $tmp_array = array_keys($recent_pages); // with array_splice()
303 foreach (array_splice($tmp_array, 0, $maxshow) as $page) {
304 $time = $recent_pages[$page];
305 $s_lastmod = htmlspecialchars(format_date($time));
306 $s_page = htmlspecialchars($page);
307 fputs($fp, '-' . $s_lastmod . ' - [[' . $s_page . ']]' . "\n");
309 fputs($fp, '#norelated' . "\n"); // :)
315 list($pattern, $pattern_a, $forceignorelist) =
316 get_autolink_pattern($pages);
318 $fp = fopen(CACHE_DIR . 'autolink.dat', 'w') or
319 die_message('Cannot write autolink file ' .
320 CACHE_DIR . '/autolink.dat' .
321 '<br />Maybe permission is not writable');
322 set_file_buffer($fp, 0);
325 fputs($fp, $pattern . "\n");
326 fputs($fp, $pattern_a . "\n");
327 fputs($fp, join("\t", $forceignorelist) . "\n");
333 // Get elapsed date of the pate
334 function get_pg_passage($page, $sw = TRUE)
336 global $show_passage;
337 if (! $show_passage) return '';
339 $time = get_filetime($page);
340 $pg_passage = ($time != 0) ? get_passage($time) : '';
342 return $sw ? '<small>' . $pg_passage . '</small>' : ' ' . $pg_passage;
345 // Last-Modified header
346 function header_lastmod($page = NULL)
350 if ($lastmod && is_page($page)) {
352 header('Last-Modified: ' .
353 date('D, d M Y H:i:s', get_filetime($page)) . ' GMT');
357 // Get a page list of this wiki
358 function get_existpages($dir = DATA_DIR, $ext = '.txt')
362 $pattern = '((?:[0-9A-F]{2})+)';
363 if ($ext != '') $ext = preg_quote($ext, '/');
364 $pattern = '/^' . $pattern . $ext . '$/';
366 $dp = @opendir($dir) or
367 die_message($dir . ' is not found or not readable.');
369 while ($file = readdir($dp))
370 if (preg_match($pattern, $file, $matches))
371 $aryret[$file] = decode($matches[1]);
377 // Get PageReading(pronounce-annotated) data in an array()
378 function get_readings()
380 global $pagereading_enable, $pagereading_kanji2kana_converter;
381 global $pagereading_kanji2kana_encoding, $pagereading_chasen_path;
382 global $pagereading_kakasi_path, $pagereading_config_page;
383 global $pagereading_config_dict;
385 $pages = get_existpages();
388 foreach ($pages as $page)
389 $readings[$page] = '';
391 $deletedPage = FALSE;
393 foreach (get_source($pagereading_config_page) as $line) {
395 if(preg_match('/^-\[\[([^]]+)\]\]\s+(.+)$/', $line, $matches)) {
396 if(isset($readings[$matches[1]])) {
397 // This page is not clear how to be pronounced
398 $readings[$matches[1]] = $matches[2];
400 // This page seems deleted
406 // If enabled ChaSen/KAKASI execution
407 if($pagereading_enable) {
409 // Check there's non-clear-pronouncing page
410 $unknownPage = FALSE;
411 foreach ($readings as $page => $reading) {
418 // Execute ChaSen/KAKASI, and get annotation
420 switch(strtolower($pagereading_kanji2kana_converter)) {
422 if(! file_exists($pagereading_chasen_path))
423 die_message('ChaSen not found: ' . $pagereading_chasen_path);
425 $tmpfname = tempnam(realpath(CACHE_DIR), 'PageReading');
426 $fp = fopen($tmpfname, 'w') or
427 die_message('Cannot write temporary file "' . $tmpfname . '".' . "\n");
428 foreach ($readings as $page => $reading) {
429 if($reading != '') continue;
430 fputs($fp, mb_convert_encoding($page . "\n",
431 $pagereading_kanji2kana_encoding, SOURCE_ENCODING));
435 $chasen = "$pagereading_chasen_path -F %y $tmpfname";
436 $fp = popen($chasen, 'r');
439 die_message('ChaSen execution failed: ' . $chasen);
441 foreach ($readings as $page => $reading) {
442 if($reading != '') continue;
445 $line = mb_convert_encoding($line, SOURCE_ENCODING,
446 $pagereading_kanji2kana_encoding);
448 $readings[$page] = $line;
453 die_message('Temporary file can not be removed: ' . $tmpfname);
456 case 'kakasi': /*FALLTHROUGH*/
458 if(! file_exists($pagereading_kakasi_path))
459 die_message('KAKASI not found: ' . $pagereading_kakasi_path);
461 $tmpfname = tempnam(realpath(CACHE_DIR), 'PageReading');
462 $fp = fopen($tmpfname, 'w') or
463 die_message('Cannot write temporary file "' . $tmpfname . '".' . "\n");
464 foreach ($readings as $page => $reading) {
465 if($reading != '') continue;
466 fputs($fp, mb_convert_encoding($page . "\n",
467 $pagereading_kanji2kana_encoding, SOURCE_ENCODING));
471 $kakasi = "$pagereading_kakasi_path -kK -HK -JK < $tmpfname";
472 $fp = popen($kakasi, 'r');
475 die_message('KAKASI execution failed: ' . $kakasi);
478 foreach ($readings as $page => $reading) {
479 if($reading != '') continue;
482 $line = mb_convert_encoding($line, SOURCE_ENCODING,
483 $pagereading_kanji2kana_encoding);
485 $readings[$page] = $line;
490 die_message('Temporary file can not be removed: ' . $tmpfname);
494 $patterns = $replacements = $matches = array();
495 foreach (get_source($pagereading_config_dict) as $line) {
497 if(preg_match('|^ /([^/]+)/,\s*(.+)$|', $line, $matches)) {
498 $patterns[] = $matches[1];
499 $replacements[] = $matches[2];
502 foreach ($readings as $page => $reading) {
503 if($reading != '') continue;
505 $readings[$page] = $page;
506 foreach ($patterns as $no => $pattern)
507 $readings[$page] = mb_convert_kana(mb_ereg_replace($pattern,
508 $replacements[$no], $readings[$page]), 'aKCV');
513 die_message('Unknown kanji-kana converter: ' . $pagereading_kanji2kana_converter . '.');
518 if($unknownPage || $deletedPage) {
520 asort($readings); // Sort by pronouncing(alphabetical/reading) order
522 foreach ($readings as $page => $reading)
523 $body .= '-[[' . $page . ']] ' . $reading . "\n";
525 page_write($pagereading_config_page, $body);
529 // Pages that are not prounouncing-clear, return pagenames of themselves
530 foreach ($pages as $page) {
531 if($readings[$page] == '')
532 $readings[$page] = $page;
538 // Get a list of encoded files (must specify a directory and a suffix)
539 function get_existfiles($dir, $ext)
541 $pattern = '/^(?:[0-9A-F]{2})+' . preg_quote($ext, '/') . '$/';
543 $dp = @opendir($dir) or die_message($dir . ' is not found or not readable.');
544 while ($file = readdir($dp))
545 if (preg_match($pattern, $file))
546 $aryret[] = $dir . $file;
551 // Get a list of related pages of the page
552 function links_get_related($page)
554 global $vars, $related;
555 static $links = array();
557 if (isset($links[$page])) return $links[$page];
559 // If possible, merge related pages generated by make_link()
560 $links[$page] = ($page == $vars['page']) ? $related : array();
562 // Get repated pages from DB
563 $links[$page] += links_get_related_db($vars['page']);
565 return $links[$page];
568 // _If needed_, re-create the file to change/correct ownership into PHP's
569 // NOTE: Not works for Windows
570 function pkwk_chown($filename, $preserve_time = TRUE)
572 static $php_uid; // PHP's UID
574 if (! isset($php_uid)) {
575 if (extension_loaded('posix')) {
576 $php_uid = posix_getuid(); // Unix
578 $php_uid = 0; // Windows
582 // Lock for pkwk_chown()
583 $lockfile = CACHE_DIR . 'pkwk_chown.lock';
584 $flock = fopen($lockfile, 'a') or
585 die('pkwk_chown(): fopen() failed for: CACHEDIR/' .
586 basename(htmlspecialchars($lockfile)));
587 flock($flock, LOCK_EX) or die('pkwk_chown(): flock() failed for lock');
590 $stat = stat($filename) or
591 die('pkwk_chown(): stat() failed for: ' . basename(htmlspecialchars($filename)));
592 if ($stat[4] === $php_uid) {
593 // NOTE: Windows always here
594 $result = TRUE; // Seems the same UID. Nothing to do
596 $tmp = $filename . '.' . getmypid() . '.tmp';
598 // Lock source $filename to avoid file corruption
599 // NOTE: Not 'r+'. Don't check write permission here
600 $ffile = fopen($filename, 'r') or
601 die('pkwk_chown(): fopen() failed for: ' .
602 basename(htmlspecialchars($filename)));
604 // Try to chown by re-creating files
606 // * touch() before copy() is for 'rw-r--r--' instead of 'rwxr-xr-x' (with umask 022).
607 // * (PHP 4 < PHP 4.2.0) touch() with the third argument is not implemented and retuns NULL and Warn.
608 // * @unlink() before rename() is for Windows but here's for Unix only
609 flock($ffile, LOCK_EX) or die('pkwk_chown(): flock() failed');
610 $result = touch($tmp) && copy($filename, $tmp) &&
611 ($preserve_time ? (touch($tmp, $stat[9], $stat[8]) || touch($tmp, $stat[9])) : TRUE) &&
612 rename($tmp, $filename);
613 flock($ffile, LOCK_UN) or die('pkwk_chown(): flock() failed');
615 fclose($ffile) or die('pkwk_chown(): fclose() failed');
617 if ($result === FALSE) @unlink($tmp);
620 // Unlock for pkwk_chown()
621 flock($flock, LOCK_UN) or die('pkwk_chown(): flock() failed for lock');
622 fclose($flock) or die('pkwk_chown(): fclose() failed for lock');
627 // touch() with trying pkwk_chown()
628 function pkwk_touch_file($filename, $time = FALSE, $atime = FALSE)
630 // Is the owner incorrected and unable to correct?
631 if (! file_exists($filename) || pkwk_chown($filename)) {
632 if ($time === FALSE) {
633 $result = touch($filename);
634 } else if ($atime === FALSE) {
635 $result = touch($filename, $time);
637 $result = touch($filename, $time, $atime);
641 die('pkwk_touch_file(): Invalid UID and (not writable for the directory or not a flie): ' .
642 htmlspecialchars(basename($filename)));