comparison: delinker_fixer/defixer.pl
delinker_fixer/defixer.pl
- changeset 0
- 3b714bbb1347
equal
deleted
inserted
replaced
| -1:000000000000 | 0:3b714bbb1347 |
|---|---|
| 1 #!/usr/bin/perl | |
| 2 #$Id$ | |
| 3 # | |
| 4 # cleanup after delinker | |
| 5 # | |
| 6 # (c)Peter [[grin]] Gervai, 2022 | |
| 7 # cc-by-sa-4.0-int / gplv3+ | |
| 8 # | |
| 9 | |
| 10 use strict; | |
| 11 use warnings; | |
| 12 use utf8; | |
| 13 | |
| 14 use File::Basename; | |
| 15 use lib dirname(__FILE__); | |
| 16 use BotSecrets; | |
| 17 | |
| 18 use MediaWiki::Bot qw(:constants); | |
| 19 use DBI; | |
| 20 | |
| 21 binmode( STDOUT, ':utf8' ); | |
| 22 | |
| 23 $|=1; | |
| 24 | |
| 25 my ($db_name, $db_host, $db_port) = ("s52421__commonsdelinquent_p", "tools.db.svc.wikimedia.cloud", 3306); | |
| 26 ## my ($db_user, $db_pw) = ("xxxxxxxxx", "xxxxxxxxxx"); | |
| 27 | |
| 28 &d("Start"); | |
| 29 | |
| 30 ## connect db | |
| 31 my $dsn = "DBI:mysql:database=$db_name;host=$db_host;port=$db_port"; | |
| 32 my $dbh = DBI->connect( $dsn, $db_user, $db_pw, { mysql_enable_utf8=>1, RaiseError=>0, AutoCommit=>0 } ); | |
| 33 $dbh->{mysql_enable_utf8} = 1; | |
| 34 | |
| 35 ## prepare sql | |
| 36 #my $sth = $dbh->prepare( "SELECT id,action,file,wiki,page,revision,done,note,replace_with_file FROM event WHERE action=? AND file=? AND done=? AND timestamp BETWEEN ? AND ?" ); | |
| 37 my $sth = $dbh->prepare( "SELECT id,action,file,wiki,page,revision,done,note,replace_with_file FROM event WHERE action=? AND done=? AND timestamp BETWEEN ? AND ?" ); | |
| 38 | |
| 39 my $sth_update = $dbh->prepare( "UPDATE event SET done=? WHERE id=?" ); | |
| 40 | |
| 41 ## connect bot (enwp) | |
| 42 #my ($bu,$bp) = ('xxxxxxxxxxxxx', 'xxxxxxxxxxxxxx' ); | |
| 43 our $bot = MediaWiki::Bot->new({ | |
| 44 host => 'en.wikipedia.org', | |
| 45 login_data => { username => $bu, password=> $bp }, | |
| 46 do_sul => 1, | |
| 47 operator => 'grin', | |
| 48 protocol => 'https', | |
| 49 debug => 2, | |
| 50 assert => 'user', | |
| 51 }); | |
| 52 my $last_wiki='huwiki'; | |
| 53 | |
| 54 die "Cannot login" unless $bot; | |
| 55 | |
| 56 ## rev | |
| 57 my $revert_message = "Undoing CommonsDelinker bad replace, will be retried later."; | |
| 58 | |
| 59 ## prepare search | |
| 60 my $action = 'replace'; | |
| 61 #my $file = 'S-3A_MAD_DN-SC-87-05743.JPEG'; | |
| 62 my $done = 127; | |
| 63 my ($ts_from, $ts_to) = ('20220502000000', '20220506130000'); | |
| 64 #my $res = $sth->execute( $action, $file, $done, $ts_from, $ts_to ); | |
| 65 my $res = $sth->execute( $action, $done, $ts_from, $ts_to ); | |
| 66 if( $dbh->err ) { | |
| 67 die "Error doing SQL: " . $dbh->errstr; | |
| 68 } | |
| 69 print $sth->rows . " rows found.\n"; | |
| 70 | |
| 71 ## results | |
| 72 while( my $a = $sth->fetchrow_hashref ) { | |
| 73 for my $key (sort keys %$a) { | |
| 74 print "$key=" . $$a{$key} . " "; | |
| 75 } | |
| 76 print "\n"; | |
| 77 | |
| 78 # check data | |
| 79 if( $last_wiki ne $$a{wiki} ) { | |
| 80 my $wikidata = &get_wikidata( $$a{wiki} ); | |
| 81 next unless $wikidata; | |
| 82 $bot->set_wiki( $wikidata ); | |
| 83 } | |
| 84 | |
| 85 my $revid = $bot->get_last( $$a{page}, 'CommonsDelinker' ); | |
| 86 if( !defined( $revid ) ) { | |
| 87 &d("Revid is missing!! skipping $$a{wiki}:$$a{page}!!"); | |
| 88 #&error("missing revid"); | |
| 89 next; | |
| 90 } | |
| 91 | |
| 92 if( $revid == $$a{revision} ) { | |
| 93 &d(" Page unchanged, undo possible! REVERTING $$a{wiki}:$$a{page}"); | |
| 94 | |
| 95 # revert | |
| 96 if( $bot->revert( $$a{page}, $$a{revision}, $revert_message ) ) { | |
| 97 &d( "Success. Updating."); | |
| 98 $sth_update->execute( 0, $$a{id} ); | |
| 99 if( $dbh->err ) { | |
| 100 &d( "*** Error updating db for $$a{wiki}:$$a{page} id=$$a{id}!!"); | |
| 101 $dbh->commit; | |
| 102 exit; | |
| 103 } | |
| 104 | |
| 105 } else { | |
| 106 &d( " Revert #1 failed, try to login."); | |
| 107 if( $bot->login( { username=>$bu, password=>$bp } ) ) { | |
| 108 if( $bot->revert( $$a{page}, $$a{revision}, $revert_message ) ) { | |
| 109 &d( "Success^2. Updating."); | |
| 110 $sth_update->execute( 0, $$a{id} ); | |
| 111 if( $dbh->err ) { | |
| 112 &d( "*** Error updating db for $$a{wiki}:$$a{page} id=$$a{id}!!"); | |
| 113 $dbh->commit; | |
| 114 exit; | |
| 115 } | |
| 116 } | |
| 117 | |
| 118 } else { | |
| 119 &d( "Login failed into $$a{wiki}. Skipping"); | |
| 120 &error("login failed"); | |
| 121 $sth_update->execute( 43, $$a{id} ); | |
| 122 } | |
| 123 } | |
| 124 | |
| 125 } else { | |
| 126 &d( " Page changed, oldid $$a{revision} - newid $revid; skipping (update db)."); | |
| 127 $sth_update->execute( 42, $$a{id} ); | |
| 128 if( $dbh->err ) { | |
| 129 &d( "*** Error updating db for non-changed $$a{wiki}:$$a{page} id=$$a{id}!!"); | |
| 130 } | |
| 131 } | |
| 132 | |
| 133 $dbh->commit; | |
| 134 } | |
| 135 &d( "Commit."); | |
| 136 $dbh->commit; | |
| 137 | |
| 138 exit; | |
| 139 | |
| 140 sub d { | |
| 141 my ($s) = @_; | |
| 142 print scalar(localtime) . " [$$] $s\n"; | |
| 143 } | |
| 144 | |
| 145 | |
| 146 sub error { | |
| 147 return; # doesn't work | |
| 148 my ($s) = @_; | |
| 149 print "error: $s; " . $bot->{error}->{code} . "; " . $bot->{error}->{details} . "\n"; | |
| 150 | |
| 151 #use Data::Dumper; | |
| 152 #die Dumper($bot); | |
| 153 #exit; | |
| 154 } | |
| 155 | |
| 156 ## decypher short wikinames | |
| 157 sub get_wikidata { | |
| 158 my ($name) = @_; | |
| 159 | |
| 160 &d("Decode $name"); | |
| 161 | |
| 162 my $host = $bot->db_to_domain( $name ); | |
| 163 return { host => $host }; | |
| 164 | |
| 165 if( $name =~ /^(.{2,3}|simple)wiki$/ ) { | |
| 166 return { host => "$1.wikipedia.org" }; | |
| 167 } | |
| 168 | |
| 169 if( $name =~ /^(.{2,3})wikivoyage$/ ) { | |
| 170 return { host => "$1.wikivoyage.org" }; | |
| 171 } | |
| 172 | |
| 173 if( $name =~ /^(.{2,3})wikisource$/ ) { | |
| 174 return { host => "$1.wikisource.org" }; | |
| 175 } | |
| 176 | |
| 177 if( $name =~ /^(.{2,3})wikiquote$/ ) { | |
| 178 return { host => "$1.wikiquote.org" }; | |
| 179 } | |
| 180 | |
| 181 if( $name =~ /^(.{2,3})wikibooks$/ ) { | |
| 182 return { host => "$1.wikibooks.org" }; | |
| 183 } | |
| 184 | |
| 185 if( $name =~ /^(.{2,3})wiktionary$/ ) { | |
| 186 return { host => "$1.wiktionary.org" }; | |
| 187 } | |
| 188 | |
| 189 if( $name eq 'wikidatawiki' ) { | |
| 190 # wikidata probably not fucked up | |
| 191 return undef; | |
| 192 } | |
| 193 | |
| 194 &d("*** $name not implemented yet!!! ***"); | |
| 195 return undef; | |
| 196 #die "decode '$name' isn't implemented yet."; | |
| 197 } | |
| 198 | |
| 199 exit; | |
| 200 | |
| 201 my $article_name = 'a'; | |
| 202 | |
| 203 my $options = { revid => 13849803 }; | |
| 204 my $txt = $bot->get_text($article_name, $options); | |
| 205 die "error something" unless defined $options->{pageid}; | |
| 206 warn "page doesn't exist" if $options->{pageid} == MediaWiki::Bot::PAGE_NONEXISTENT; | |
| 207 print "Page length is ". length($txt) . "!\n"; | |
| 208 | |
| 209 my $pageid = $bot->get_id($article_name); | |
| 210 die "error something else" unless defined $pageid; | |
| 211 printf "Page id is %s\n", $pageid; | |
| 212 | |
| 213 # last _not_ by user | |
| 214 my $revid = $bot->get_last($article_name,'no such user'); | |
| 215 printf "Last revid is %s\n", $revid; | |
| 216 | |
| 217 $revid = $bot->get_last($article_name,'FoBe'); | |
| 218 printf "Last revid-2 is %s\n", $revid; | |
| 219 | |
| 220 $options = { oldid=> 20300641, revid=>20300648 }; | |
| 221 my $diff = $bot->diff($options); | |
| 222 print "Diff: $diff\n"; | |
| 223 | |
| 224 ## commons | |
| 225 $bot->set_wiki({ | |
| 226 host => 'commons.wikimedia.org' | |
| 227 }); | |
| 228 die "Cannot login to commons" unless $bot; | |
| 229 print "Logged over commons.\n"; | |
| 230 | |
| 231 $options = { revid=> 568734018, oldid=>628016329 }; | |
| 232 $diff = $bot->diff($options); | |
| 233 print "Diff: $diff\n"; | |
| 234 | |
| 235 ## a hiba: az elozo sor utolso szava + \n bekerult a replacementbe | |
| 236 ## javitas: | |
| 237 ## - ha ez az utolso edit | |
| 238 ## - revert | |
| 239 | |
| 240 __END__ | |
| 241 <tr> | |
| 242 <td class="diff-marker" data-marker="−"></td> | |
| 243 <td class="diff-deletedline diff-side-deleted"><div>File:Nestor Lakoba, Nikita Khrushchev, Lavrenti Beria and Aghasi Khanjian.jpg|Nestor Lakoba, Nikita Khrushchev, Lavrenti Beria and Aghasi Khanjian, <del class="diffchange diffchange-inline">1935</del></div></td> | |
| 244 <td class="diff-marker" data-marker="+"></td> | |
| 245 <td class="diff-addedline diff-side-added"><div>File:Nestor Lakoba, Nikita Khrushchev, Lavrenti Beria and Aghasi Khanjian.jpg|Nestor Lakoba, Nikita Khrushchev, Lavrenti Beria and Aghasi Khanjian,<ins class="diffchange diffchange-inline">|{{c|Georgy</ins> <ins class="diffchange diffchange-inline">Malenkov}} and Beria, 1940</ins></div></td> | |
| 246 </tr> | |
| 247 <tr> | |
| 248 <td class="diff-marker" data-marker="−"></td> | |
| 249 <td class="diff-deletedline diff-side-deleted"><div>File:Берия в суде.jpg|{{c|Georgy Malenkov}} and Beria, 1940</div></td> | |
| 250 <td colspan="2" class="diff-empty diff-side-added"></td> | |
| 251 </tr> | |
| 252 | |
| 253 | |
| 254 ## generic revert | |
| 255 ## - adatbazisbol ami done=42 | |
| 256 ## - revision | |
| 257 ## - ha az az utolso akkor: | |
| 258 ## - revert | |
| 259 ## - done=0 (pending) | |
| 260 ## - ha nem akkor | |
| 261 ## - done=666 (fixx it felix) | |
| 262 ## |
