2 // PukiWiki - Yet another WikiWikiWeb clone.
3 // $Id: file.php,v 1.67 2006/04/30 03:58:35 henoheno Exp $
5 // 2002-2006 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
12 define('PKWK_MAXSHOW_ALLOWANCE', 10);
13 define('PKWK_MAXSHOW_CACHE', 'recent.dat');
16 define('PKWK_AUTOLINK_REGEX_CACHE', 'autolink.dat');
18 // Get source(wiki text) data of the page
19 function get_source($page = NULL, $lock = TRUE, $join = FALSE)
21 $result = $join ? '' : array();
24 $path = get_filename($page);
27 $fp = @fopen($path, 'r');
28 if ($fp == FALSE) return $result;
34 $result = str_replace("\r", '', fread($fp, filesize($path)));
37 // Removing line-feeds: Because file() doesn't remove them.
38 $result = str_replace("\r", '', file($path));
50 // Get last-modified filetime of the page
51 function get_filetime($page)
53 return is_page($page) ? filemtime(get_filename($page)) - LOCALZONE : 0;
56 // Get physical file name of the page
57 function get_filename($page)
59 return DATA_DIR . encode($page) . '.txt';
62 // Put a data(wiki text) into a physical file(diff, backup, text)
63 function page_write($page, $postdata, $notimestamp = FALSE)
67 if (PKWK_READONLY) return; // Do nothing
69 $postdata = make_str_rules($postdata);
71 // Create and write diff
72 $oldpostdata = is_page($page) ? join('', get_source($page)) : '';
73 $diffdata = do_diff($oldpostdata, $postdata);
74 file_write(DIFF_DIR, $page, $diffdata);
77 make_backup($page, $postdata == ''); // Is $postdata null?
80 file_write(DATA_DIR, $page, $postdata, $notimestamp);
84 $_diff = explode("\n", $diffdata);
85 $plus = join("\n", preg_replace('/^\+/', '', preg_grep('/^\+/', $_diff)));
86 $minus = join("\n", preg_replace('/^-/', '', preg_grep('/^-/', $_diff)));
87 tb_send($page, $plus, $minus);
93 // Modify original text with user-defined / system-defined rules
94 function make_str_rules($source)
96 global $str_rules, $fixed_heading_anchor;
98 $lines = explode("\n", $source);
99 $count = count($lines);
104 for ($i = 0; $i < $count; $i++) {
105 $line = & $lines[$i]; // Modify directly
107 // Ignore null string and preformatted texts
108 if ($line == '' || $line{0} == ' ' || $line{0} == "\t") continue;
112 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
114 preg_match('/#[^{]*(\{\{+)\s*$/', $line, $matches)) {
115 // Multiline convert plugin start
117 $multiline = strlen($matches[1]); // Set specific number
120 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
122 preg_match('/^\}{' . $multiline . '}\s*$/', $line)) {
123 // Multiline convert plugin end
128 if ($modify === FALSE) continue;
130 // Replace with $str_rules
131 foreach ($str_rules as $pattern => $replacement)
132 $line = preg_replace('/' . $pattern . '/', $replacement, $line);
134 // Adding fixed anchor into headings
135 if ($fixed_heading_anchor &&
136 preg_match('/^(\*{1,3}.*?)(?:\[#([A-Za-z][\w-]*)\]\s*)?$/', $line, $matches) &&
137 (! isset($matches[2]) || $matches[2] == '')) {
138 // Generate unique id
139 $anchor = generate_fixed_heading_anchor_id($matches[1]);
140 $line = rtrim($matches[1]) . ' [#' . $anchor . ']';
144 // Multiline part has no stopper
145 if (! PKWKEXP_DISABLE_MULTILINE_PLUGIN_HACK &&
146 $modify === FALSE && $multiline != 0)
147 $lines[] = str_repeat('}', $multiline);
149 return implode("\n", $lines);
153 function generate_fixed_heading_anchor_id($seed)
155 // A random alphabetic letter + 7 letters of random strings from md()
156 return chr(mt_rand(ord('a'), ord('z'))) .
157 substr(md5(uniqid(substr($seed, 0, 100), TRUE)),
161 // Read top N lines as an array
162 // (Use PHP file() function if you want to get ALL lines)
163 function file_head($file, $count = 1, $lock = TRUE, $buffer = 8192)
167 $fp = @fopen($file, 'r');
168 if ($fp === FALSE) return FALSE;
169 set_file_buffer($fp, 0);
170 if ($lock) flock($fp, LOCK_SH);
173 while (! feof($fp)) {
174 $line = fgets($fp, $buffer);
175 if ($line != FALSE) $array[] = $line;
176 if (++$index >= $count) break;
178 if ($lock) flock($fp, LOCK_UN);
179 if (! fclose($fp)) return FALSE;
185 function file_write($dir, $page, $str, $notimestamp = FALSE)
187 global $update_exec, $_msg_invalidiwn, $notify, $notify_diff_only, $notify_subject;
188 global $whatsdeleted, $maxshow_deleted;
190 if (PKWK_READONLY) return; // Do nothing
191 if ($dir != DATA_DIR && $dir != DIFF_DIR) die('file_write(): Invalid directory');
193 $page = strip_bracket($page);
194 $file = $dir . encode($page) . '.txt';
195 $file_exists = file_exists($file);
200 if ($dir == DATA_DIR && $str === '') {
202 if (! $file_exists) return; // Ignore null posting for DATA_DIR
204 // Update RecentDeleted (Add the $page)
205 add_recent($page, $whatsdeleted, '', $maxshow_deleted);
210 // Update RecentDeleted, and remove the page from RecentChanges
211 lastmodified_add($whatsdeleted, $page);
213 // Clear is_page() cache
214 is_page($page, TRUE);
218 } else if ($dir == DIFF_DIR && $str === " \n") {
219 return; // Ignore null posting for DIFF_DIR
223 // File replacement (Edit)
225 if (! is_pagename($page))
226 die_message(str_replace('$1', htmlspecialchars($page),
227 str_replace('$2', 'WikiName', $_msg_invalidiwn)));
229 $str = rtrim(preg_replace('/' . "\r" . '/', '', $str)) . "\n";
230 $timestamp = ($file_exists && $notimestamp) ? filemtime($file) : FALSE;
232 $fp = fopen($file, 'a') or die('fopen() failed: ' .
233 htmlspecialchars(basename($dir) . '/' . encode($page) . '.txt') .
235 'Maybe permission is not writable or filename is too long');
236 set_file_buffer($fp, 0);
244 if ($timestamp) pkwk_touch_file($file, $timestamp);
247 if ($dir == DATA_DIR) {
248 // Update RecentChanges (Add or renew the $page)
249 if ($timestamp === FALSE) lastmodified_add($page);
251 // Execute $update_exec here
252 if ($update_exec) system($update_exec . ' > /dev/null &');
254 } else if ($dir == DIFF_DIR && $notify) {
255 if ($notify_diff_only) $str = preg_replace('/^[^-+].*\n/m', '', $str);
256 $footer['ACTION'] = 'Page update';
257 $footer['PAGE'] = & $page;
258 $footer['URI'] = get_script_uri() . '?' . rawurlencode($page);
259 $footer['USER_AGENT'] = TRUE;
260 $footer['REMOTE_ADDR'] = TRUE;
261 pkwk_mail_notify($notify_subject, $str, $footer) or
262 die('pkwk_mail_notify(): Failed');
265 is_page($page, TRUE); // Clear is_page() cache
268 // Update RecentDeleted
269 function add_recent($page, $recentpage, $subject = '', $limit = 0)
271 if (PKWK_READONLY || $limit == 0 || $page == '' || $recentpage == '' ||
272 check_non_list($page)) return;
275 $lines = $matches = array();
276 foreach (get_source($recentpage) as $line)
277 if (preg_match('/^-(.+) - (\[\[.+\]\])$/', $line, $matches))
278 $lines[$matches[2]] = $line;
280 $_page = '[[' . $page . ']]';
282 // Remove a report about the same page
283 if (isset($lines[$_page])) unset($lines[$_page]);
286 array_unshift($lines, '-' . format_date(UTIME) . ' - ' . $_page .
287 htmlspecialchars($subject) . "\n");
289 // Get latest $limit reports
290 $lines = array_splice($lines, 0, $limit);
293 $fp = fopen(get_filename($recentpage), 'w') or
294 die_message('Cannot write page file ' .
295 htmlspecialchars($recentpage) .
296 '<br />Maybe permission is not writable or filename is too long');
297 set_file_buffer($fp, 0);
300 fputs($fp, '#freeze' . "\n");
301 fputs($fp, '#norelated' . "\n"); // :)
302 fputs($fp, join('', $lines));
307 // Update PKWK_MAXSHOW_CACHE itself (Add or renew about the $page) (Light)
308 // Use without $autolink
309 function lastmodified_add($update = '', $remove = '')
311 global $maxshow, $whatsnew, $autolink;
313 // AutoLink implimentation needs everything, for now
315 put_lastmodified(); // Try to (re)create ALL
319 if (($update == '' || check_non_list($update)) && $remove == '')
322 $file = CACHE_DIR . PKWK_MAXSHOW_CACHE;
323 if (! file_exists($file)) {
324 put_lastmodified(); // Try to (re)create ALL
329 pkwk_touch_file($file);
330 $fp = fopen($file, 'r+') or
331 die_message('Cannot open ' . 'CACHE_DIR/' . PKWK_MAXSHOW_CACHE);
332 set_file_buffer($fp, 0);
335 // Read (keep the order of the lines)
336 $recent_pages = $matches = array();
337 foreach(file_head($file, $maxshow + PKWK_MAXSHOW_ALLOWANCE, FALSE) as $line)
338 if (preg_match('/^([0-9]+)\t(.+)/', $line, $matches))
339 $recent_pages[$matches[2]] = $matches[1];
341 // Remove if it exists inside
342 if (isset($recent_pages[$update])) unset($recent_pages[$update]);
343 if (isset($recent_pages[$remove])) unset($recent_pages[$remove]);
345 // Add to the top: like array_unshift()
347 $recent_pages = array($update => get_filetime($update)) + $recent_pages;
350 $abort = count($recent_pages) < $maxshow;
356 foreach ($recent_pages as $_page=>$time)
357 fputs($fp, $time . "\t" . $_page . "\n");
364 put_lastmodified(); // Try to (re)create ALL
371 // Update the page 'RecentChanges'
373 $recent_pages = array_splice($recent_pages, 0, $maxshow);
374 $file = get_filename($whatsnew);
377 pkwk_touch_file($file);
378 $fp = fopen($file, 'r+') or
379 die_message('Cannot open ' . htmlspecialchars($whatsnew));
380 set_file_buffer($fp, 0);
386 foreach ($recent_pages as $_page=>$time)
387 fputs($fp, '-' . htmlspecialchars(format_date($time)) .
388 ' - ' . '[[' . htmlspecialchars($_page) . ']]' . "\n");
389 fputs($fp, '#norelated' . "\n"); // :)
395 // Re-create PKWK_MAXSHOW_CACHE (Heavy)
396 function put_lastmodified()
398 global $maxshow, $whatsnew, $autolink;
400 if (PKWK_READONLY) return; // Do nothing
402 // Get WHOLE page list
403 $pages = get_existpages();
405 // Check ALL filetime
406 $recent_pages = array();
407 foreach($pages as $page)
408 if ($page != $whatsnew && ! check_non_list($page))
409 $recent_pages[$page] = get_filetime($page);
411 // Sort decending order of last-modification date
412 arsort($recent_pages, SORT_NUMERIC);
415 $recent_pages = array_splice($recent_pages, 0, $maxshow + PKWK_MAXSHOW_ALLOWANCE);
417 // Re-create PKWK_MAXSHOW_CACHE
418 $file = CACHE_DIR . PKWK_MAXSHOW_CACHE;
419 pkwk_touch_file($file);
420 $fp = fopen($file, 'r+') or
421 die_message('Cannot open' . 'CACHE_DIR/' . PKWK_MAXSHOW_CACHE);
422 set_file_buffer($fp, 0);
426 foreach ($recent_pages as $page=>$time)
427 fputs($fp, $time . "\t" . $page . "\n");
431 // Create RecentChanges
432 $file = get_filename($whatsnew);
433 pkwk_touch_file($file);
434 $fp = fopen($file, 'r+') or
435 die_message('Cannot open ' . htmlspecialchars($whatsnew));
436 set_file_buffer($fp, 0);
440 foreach (array_keys($recent_pages) as $page) {
441 $time = $recent_pages[$page];
442 $s_lastmod = htmlspecialchars(format_date($time));
443 $s_page = htmlspecialchars($page);
444 fputs($fp, '-' . $s_lastmod . ' - [[' . $s_page . ']]' . "\n");
446 fputs($fp, '#norelated' . "\n"); // :)
452 list($pattern, $pattern_a, $forceignorelist) =
453 get_autolink_pattern($pages);
455 $file = CACHE_DIR . PKWK_AUTOLINK_REGEX_CACHE;
456 pkwk_touch_file($file);
457 $fp = fopen($file, 'r+') or
458 die_message('Cannot open ' . 'CACHE_DIR/' . PKWK_AUTOLINK_REGEX_CACHE);
459 set_file_buffer($fp, 0);
463 fputs($fp, $pattern . "\n");
464 fputs($fp, $pattern_a . "\n");
465 fputs($fp, join("\t", $forceignorelist) . "\n");
471 // Get elapsed date of the page
472 function get_pg_passage($page, $sw = TRUE)
474 global $show_passage;
475 if (! $show_passage) return '';
477 $time = get_filetime($page);
478 $pg_passage = ($time != 0) ? get_passage($time) : '';
480 return $sw ? '<small>' . $pg_passage . '</small>' : ' ' . $pg_passage;
483 // Last-Modified header
484 function header_lastmod($page = NULL)
488 if ($lastmod && is_page($page)) {
490 header('Last-Modified: ' .
491 date('D, d M Y H:i:s', get_filetime($page)) . ' GMT');
495 // Get a page list of this wiki
496 function get_existpages($dir = DATA_DIR, $ext = '.txt')
500 $pattern = '((?:[0-9A-F]{2})+)';
501 if ($ext != '') $ext = preg_quote($ext, '/');
502 $pattern = '/^' . $pattern . $ext . '$/';
504 $dp = @opendir($dir) or
505 die_message($dir . ' is not found or not readable.');
507 while ($file = readdir($dp))
508 if (preg_match($pattern, $file, $matches))
509 $aryret[$file] = decode($matches[1]);
515 // Get PageReading(pronounce-annotated) data in an array()
516 function get_readings()
518 global $pagereading_enable, $pagereading_kanji2kana_converter;
519 global $pagereading_kanji2kana_encoding, $pagereading_chasen_path;
520 global $pagereading_kakasi_path, $pagereading_config_page;
521 global $pagereading_config_dict;
523 $pages = get_existpages();
526 foreach ($pages as $page)
527 $readings[$page] = '';
529 $deletedPage = FALSE;
531 foreach (get_source($pagereading_config_page) as $line) {
533 if(preg_match('/^-\[\[([^]]+)\]\]\s+(.+)$/', $line, $matches)) {
534 if(isset($readings[$matches[1]])) {
535 // This page is not clear how to be pronounced
536 $readings[$matches[1]] = $matches[2];
538 // This page seems deleted
544 // If enabled ChaSen/KAKASI execution
545 if($pagereading_enable) {
547 // Check there's non-clear-pronouncing page
548 $unknownPage = FALSE;
549 foreach ($readings as $page => $reading) {
556 // Execute ChaSen/KAKASI, and get annotation
558 switch(strtolower($pagereading_kanji2kana_converter)) {
560 if(! file_exists($pagereading_chasen_path))
561 die_message('ChaSen not found: ' . $pagereading_chasen_path);
563 $tmpfname = tempnam(realpath(CACHE_DIR), 'PageReading');
564 $fp = fopen($tmpfname, 'w') or
565 die_message('Cannot write temporary file "' . $tmpfname . '".' . "\n");
566 foreach ($readings as $page => $reading) {
567 if($reading != '') continue;
568 fputs($fp, mb_convert_encoding($page . "\n",
569 $pagereading_kanji2kana_encoding, SOURCE_ENCODING));
573 $chasen = "$pagereading_chasen_path -F %y $tmpfname";
574 $fp = popen($chasen, 'r');
577 die_message('ChaSen execution failed: ' . $chasen);
579 foreach ($readings as $page => $reading) {
580 if($reading != '') continue;
583 $line = mb_convert_encoding($line, SOURCE_ENCODING,
584 $pagereading_kanji2kana_encoding);
586 $readings[$page] = $line;
591 die_message('Temporary file can not be removed: ' . $tmpfname);
594 case 'kakasi': /*FALLTHROUGH*/
596 if(! file_exists($pagereading_kakasi_path))
597 die_message('KAKASI not found: ' . $pagereading_kakasi_path);
599 $tmpfname = tempnam(realpath(CACHE_DIR), 'PageReading');
600 $fp = fopen($tmpfname, 'w') or
601 die_message('Cannot write temporary file "' . $tmpfname . '".' . "\n");
602 foreach ($readings as $page => $reading) {
603 if($reading != '') continue;
604 fputs($fp, mb_convert_encoding($page . "\n",
605 $pagereading_kanji2kana_encoding, SOURCE_ENCODING));
609 $kakasi = "$pagereading_kakasi_path -kK -HK -JK < $tmpfname";
610 $fp = popen($kakasi, 'r');
613 die_message('KAKASI execution failed: ' . $kakasi);
616 foreach ($readings as $page => $reading) {
617 if($reading != '') continue;
620 $line = mb_convert_encoding($line, SOURCE_ENCODING,
621 $pagereading_kanji2kana_encoding);
623 $readings[$page] = $line;
628 die_message('Temporary file can not be removed: ' . $tmpfname);
632 $patterns = $replacements = $matches = array();
633 foreach (get_source($pagereading_config_dict) as $line) {
635 if(preg_match('|^ /([^/]+)/,\s*(.+)$|', $line, $matches)) {
636 $patterns[] = $matches[1];
637 $replacements[] = $matches[2];
640 foreach ($readings as $page => $reading) {
641 if($reading != '') continue;
643 $readings[$page] = $page;
644 foreach ($patterns as $no => $pattern)
645 $readings[$page] = mb_convert_kana(mb_ereg_replace($pattern,
646 $replacements[$no], $readings[$page]), 'aKCV');
651 die_message('Unknown kanji-kana converter: ' . $pagereading_kanji2kana_converter . '.');
656 if($unknownPage || $deletedPage) {
658 asort($readings); // Sort by pronouncing(alphabetical/reading) order
660 foreach ($readings as $page => $reading)
661 $body .= '-[[' . $page . ']] ' . $reading . "\n";
663 page_write($pagereading_config_page, $body);
667 // Pages that are not prounouncing-clear, return pagenames of themselves
668 foreach ($pages as $page) {
669 if($readings[$page] == '')
670 $readings[$page] = $page;
676 // Get a list of encoded files (must specify a directory and a suffix)
677 function get_existfiles($dir, $ext)
679 $pattern = '/^(?:[0-9A-F]{2})+' . preg_quote($ext, '/') . '$/';
681 $dp = @opendir($dir) or die_message($dir . ' is not found or not readable.');
682 while ($file = readdir($dp))
683 if (preg_match($pattern, $file))
684 $aryret[] = $dir . $file;
689 // Get a list of related pages of the page
690 function links_get_related($page)
692 global $vars, $related;
693 static $links = array();
695 if (isset($links[$page])) return $links[$page];
697 // If possible, merge related pages generated by make_link()
698 $links[$page] = ($page == $vars['page']) ? $related : array();
700 // Get repated pages from DB
701 $links[$page] += links_get_related_db($vars['page']);
703 return $links[$page];
706 // _If needed_, re-create the file to change/correct ownership into PHP's
707 // NOTE: Not works for Windows
708 function pkwk_chown($filename, $preserve_time = TRUE)
710 static $php_uid; // PHP's UID
712 if (! isset($php_uid)) {
713 if (extension_loaded('posix')) {
714 $php_uid = posix_getuid(); // Unix
716 $php_uid = 0; // Windows
720 // Lock for pkwk_chown()
721 $lockfile = CACHE_DIR . 'pkwk_chown.lock';
722 $flock = fopen($lockfile, 'a') or
723 die('pkwk_chown(): fopen() failed for: CACHEDIR/' .
724 basename(htmlspecialchars($lockfile)));
725 flock($flock, LOCK_EX) or die('pkwk_chown(): flock() failed for lock');
728 $stat = stat($filename) or
729 die('pkwk_chown(): stat() failed for: ' . basename(htmlspecialchars($filename)));
730 if ($stat[4] === $php_uid) {
731 // NOTE: Windows always here
732 $result = TRUE; // Seems the same UID. Nothing to do
734 $tmp = $filename . '.' . getmypid() . '.tmp';
736 // Lock source $filename to avoid file corruption
737 // NOTE: Not 'r+'. Don't check write permission here
738 $ffile = fopen($filename, 'r') or
739 die('pkwk_chown(): fopen() failed for: ' .
740 basename(htmlspecialchars($filename)));
742 // Try to chown by re-creating files
744 // * touch() before copy() is for 'rw-r--r--' instead of 'rwxr-xr-x' (with umask 022).
745 // * (PHP 4 < PHP 4.2.0) touch() with the third argument is not implemented and retuns NULL and Warn.
746 // * @unlink() before rename() is for Windows but here's for Unix only
747 flock($ffile, LOCK_EX) or die('pkwk_chown(): flock() failed');
748 $result = touch($tmp) && copy($filename, $tmp) &&
749 ($preserve_time ? (touch($tmp, $stat[9], $stat[8]) || touch($tmp, $stat[9])) : TRUE) &&
750 rename($tmp, $filename);
751 flock($ffile, LOCK_UN) or die('pkwk_chown(): flock() failed');
753 fclose($ffile) or die('pkwk_chown(): fclose() failed');
755 if ($result === FALSE) @unlink($tmp);
758 // Unlock for pkwk_chown()
759 flock($flock, LOCK_UN) or die('pkwk_chown(): flock() failed for lock');
760 fclose($flock) or die('pkwk_chown(): fclose() failed for lock');
765 // touch() with trying pkwk_chown()
766 function pkwk_touch_file($filename, $time = FALSE, $atime = FALSE)
768 // Is the owner incorrected and unable to correct?
769 if (! file_exists($filename) || pkwk_chown($filename)) {
770 if ($time === FALSE) {
771 $result = touch($filename);
772 } else if ($atime === FALSE) {
773 $result = touch($filename, $time);
775 $result = touch($filename, $time, $atime);
779 die('pkwk_touch_file(): Invalid UID and (not writable for the directory or not a flie): ' .
780 htmlspecialchars(basename($filename)));