search

edsu / wikichanges

a NodeJS library for monitoring changes on Wikipedia sites
https://npmjs.org/package/wikichanges
70 stars 17 forks source link

What other flags might we see? #11

Closed WardCunningham closed 10 years ago

WardCunningham commented 10 years ago

I've been watching en.wikipedia for a day now and have seen the following flags.

screen shot 2014-09-01 at 10 06 59 am

Where can I find additional information about these flags? I'm interested to know how I can use them to make better sense of other things I see in the feed. Thanks.

reedy commented 10 years ago

The code that generates this (on the MediaWiki side) is actually at https://github.com/wikimedia/mediawiki-core/blob/master/includes/rcfeed/IRCColourfulRCFeedFormatter.php

$fullString = "$titleString\0034 $flag\00310 \00302$url\003 \0035\003 \00303$user\003 \0035\003 $szdiff \00310$comment\003\n"

rc_log_action is a db column

-- Store log action or null rc_log_action varbinary(255) NULL default NULL,

    if ( $attribs['rc_type'] == RC_LOG ) {
        $targetText = $rc->getTitle()->getPrefixedText();
        $comment = self::cleanupForIRC( str_replace(
            "[[$targetText]]",
            "[[\00302$targetText\00310]]",
            $actionComment
        ) );
        $flag = $attribs['rc_log_action'];
    } else {
        $comment = self::cleanupForIRC( $attribs['rc_comment'] );
        $flag = '';
        if ( !$attribs['rc_patrolled']
            && ( $wgUseRCPatrol || $attribs['rc_type'] == RC_NEW && $wgUseNPPatrol )
        ) {
            $flag .= '!';
        }
        $flag .= ( $attribs['rc_type'] == RC_NEW ? "N" : "" )
            . ( $attribs['rc_minor'] ? "M" : "" ) . ( $attribs['rc_bot'] ? "B" : "" );
    }
nemobis commented 10 years ago

Hello!

On m, b, N, ! see:

On the others, see

There's more ;)

I have no idea what "hit" is.

Glaisher commented 10 years ago

hit is for triggers of public abusefilters

edsu commented 10 years ago

Does this help @WardCunningham? I wonder if it makes sense for wikichanges to try to decode/normalize these to make them more accessible?

FYI: All I had to do was mention Ward Cunningham had a question on Github in #wikimedia-tech and ... BOOM ... answers!

WardCunningham commented 10 years ago

I'm traveling and having trouble with connectivity. Still, here is what I found looking at two days of data. I'm feeding this into NewRelic's Insights and finding lots to explore. More soon.

89281   MB
65107   M
27569   upload
12759   create
10327   B
8720    N
6825    !N
6742    overwrite
6398    hit
5753    delete
3978    !
3195    approve
2361    !M
2067    move
1520    thank
1303    patrol
1010    NB
977 block
594 reviewed
187 protect
164 approve-i
152 tag
150 move_redir
99  selfadd
67  restore
61  byemail
59  revision
44  create2
40  interwiki
34  group
33  unapprove
26  reblock
24  modify
17  update
17  autopromote
16  mark
15  unreviewed
15  message
11  rights
11  add
10  renameuser
7   move_prot
4   unblock
4   config
3   selfremove
2   merge
1   unprotect
1   reset
1   remove
WardCunningham commented 10 years ago

I've learned a lot by browsing this stream for a few days. Here is a sample report that I've enjoyed watching to see hard work being done many hours at a time:

http://c2.com/awb.html

I haven't felt any strong need to have the flag field further decoded. The bulk of these are related to updates in corresponding logs. I will reopen this issue if I discover otherwise.

Thanks all.