maintenance/importLogs.inc

   1 <?php
   2 # Copyright (C) 2004 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 # Attempt to import existing log pages into the log tables.
  21
  22 # Not yet complete.
  23
  24 require_once( 'GlobalFunctions.php' );
  25 require_once( 'Database.php' );
  26 require_once( 'Article.php' );
  27 require_once( 'LogPage.php' );
  28
  29 # Log importer
  30 class LogImporter {
  31         var $dummy = false;
  32
  33         function LogImporter( $type ) {
  34                 $this->type = $type;
  35                 $this->db =& wfGetDB( DB_MASTER );
  36                 $this->actions = $this->setupActions();
  37         }
  38
  39         function setupActions() {
  40                 $actions = array();
  41                 foreach( LogPage::validActions( $this->type ) as $action ) {
  42                         $key = "{$this->type}/$action";
  43                         $actions[$key] = $this->makeLineRegexp( $this->type, $action );
  44                 }
  45                 return $actions;
  46         }
  47
  48         function makeLineRegexp( $type, $action ) {
  49                 $linkRegexp = '(?:\[\[)?([^|\]]+?)(?:\|[^\]]+?)?(?:\]\])?';
  50                 $linkRegexp2 = '\[\[([^|\]]+?)(?:\|[^\]]+?)?\]\]';
  51
  52                 $text = LogPage::actionText( $type, $action );
  53                 $text = preg_quote( $text, '/' );
  54                 $text = str_replace( '\$1', $linkRegexp, $text );
  55                 $text = '^(.*?) ' . $linkRegexp2 . ' ' . $text;
  56                 $text .= '(?: <em>\((.*)\)<\/em>)?';
  57                 $text = "/$text/";
  58                 return $text;
  59         }
  60
  61         function importText( $text ) {
  62                 if( $this->dummy ) {
  63                         print $text;
  64                         var_dump( $this->actions );
  65                 }
  66                 $lines = explode( '<li>', $text );
  67                 foreach( $lines as $line ) {
  68                         if( preg_match( '!^(.*)</li>!', $line, $matches ) ) {
  69                                 $this->importLine( $matches[1] );
  70                         }
  71                 }
  72         }
  73
  74         function fixDate( $date ) {
  75                 # Yuck! Parsing multilingual date formats??!!!!???!!??!
  76                 # 01:55, 23 Aug 2004 - won't take in strtotimr
  77                 # "Aug 23 2004 01:55" - seems ok
  78                 # TODO: multilingual attempt to extract from the data in Language
  79                 if( preg_match( '/^(\d+:\d+(?::\d+)?), (.*)$/', $date, $matches ) ) {
  80                         $date = $matches[2] . ' ' . $matches[1];
  81                 }
  82                 $n = strtotime( $date ) + date("Z");
  83                 # print gmdate( 'D, d M Y H:i:s T', $n ) . "\n";
  84                 $timestamp = wfTimestamp( TS_MW, $n );
  85                 return $timestamp;
  86         }
  87
  88         function importLine( $line ) {
  89                 foreach( $this->actions as $action => $regexp ) {
  90                         if( preg_match( $regexp, $line, $matches ) ) {
  91                                 if( $this->dummy ) {
  92                                         #var_dump( $matches );
  93                                 }
  94                                 $date = $this->fixDate( $matches[1] );
  95                                 $user = Title::newFromText( $matches[2] );
  96                                 $target = Title::newFromText( $matches[3] );
  97                                 if( isset( $matches[4] ) ) {
  98                                         $comment = $matches[4];
  99                                 } else {
 100                                         $comment = '';
 101                                 }
 102
 103                                 $insert = array(
 104                                         'log_type' => $this->type,
 105                                         'log_action' => preg_replace( '!^.*/!', '', $action ),
 106                                         'log_timestamp' => $date,
 107                                         'log_user' => IntVal( User::idFromName( $user->getText() ) ),
 108                                         'log_namespace' => $target->getNamespace(),
 109                                         'log_title' => $target->getDBkey(),
 110                                         'log_comment' => wfUnescapeWikiText( $comment ),
 111                                 );
 112                                 if( $this->dummy ) {
 113                                         var_dump( $insert );
 114                                 } else {
 115                                         # FIXME: avoid duplicates!
 116                                         $this->db->insertArray( 'logging', $insert );
 117                                 }
 118                                 break;
 119                         }
 120                 }
 121         }
 122 }
 123
 124 function wfUnescapeWikiText( $text ) {
 125         $text = str_replace(
 126                 array( '&#91;', '&#124;', '&#39;', 'ISBN&#32;', '&#58;//' , "\n&#61;", '&#123;&#123;' ),
 127                 array( '[',             '|',      "'",     'ISBN '        , '://'         , "\n=", '{{' ),
 128                 $text );
 129         return $text;
 130 }
 131
 132 ?>