The uploaded corpora on Buildbot

There are two ways to perform MassChecks on HandClassifiedCorpora; one, by downloading the SpamAssassin SVN trunk to your machine and running them locally, and two, by uploading your corpora to "buildbot", our mass-check server, as described in this document.

The "bbmass" preflight mass-checks use this corpus; the log from the selection stage can be read at http://buildbot.spamassassin.org/bbmass/corpus_makeup.txt .

It is possible to mass-check these corpora nightly ; results also at http://buildbot.spamassassin.org/bbmass/ .

The "bbmass" preflight mass-checks use this corpus; the log from the selection stage can be read at http://buildbot.spamassassin.org/bbmass/corpus_makeup.txt .

It is possible to mass-check these corpora nightly ; results also at http://buildbot.spamassassin.org/bbmass/ .

Administrivia: how the corpus is laid out

The filesystem layout of the corpora rsynced up to the server, is like this:

/home/bbmass/rawcor/WHO/TYPE/FOLDER

"WHO" is the person who submitted it via rsync, e.g. "doc", "jm", "zmi".

Under that, we have "TYPE", which is either "ham" or "spam".

Under that, "FOLDER", which is whatever the person feels is appropriate. For example, some of us use date-stamped dirs here. It is also possible to use mboxes, as long as they are files and their filename ends in ".mbox".

Note that only files which (a) are directly in the "ham" or "spam" directory, not a subdirectory, and (b) are named ending in ".mbox", will be treated as mboxes. Anything else will be considered a single email message.

How to get your corpus up there

This is done via rsync.

Give somebody on the PMC a shout, since they have privileges to create an rsync area for you to upload stuff to. The easiest way is to mail the dev list. (If you're on the PMC, just SSH in and copy over a tarball yourself! or create yourself an rsync account using a random password.)

Once they've done this, they'll send you the username and password; you can then sync your files like so:

  export RSYNC_PASSWORD=$YOURPASS
  rsync -vr /path/to/your/files \
      rsync://$YOURUSER@rsync.spamassassin.org/mailcorpus_$YOURUSER

(where $YOURPASS, $YOURUSER, $YOU are whatever the PMC guy mailed to you.)

It's important that you have 2 dirs in the /path/to/your/files directory,
ham and spam. Any files ending in .mbox inside those dirs will be treated as UNIX mbox-format files; any other files will be treated as individual messages (one message per file).

Privacy

Uploaded corpora are not considered public knowledge. The people with accounts on that machine should treat the uploaded messages responsibly, and respect the uploader's privacy. If you are concerned about the privacy of these messages, you may be advised to remove the more private mails before uploading, or mass-check on your own machine instead.

(Administrivia: Creating a new rsync area for someone to upload corpora)

Some stuff for PMC people hacking on this...

sudo vi /etc/rsyncd.conf

add something like this to the end, changing "CORPUSUSER" to the username you want to give out:

[mailcorpus_CORPUSUSER]
        path = /home/bbmass/rawcor/CORPUSUSER
        read only = false
        auth users = CORPUSUSER
        secrets file = /home/corpus-rsync/secrets

CORPUSUSER="[username you want to give out]"
cd /home/bbmass/rawcor/
mkdir $CORPUSUSER
chmod 1777 $CORPUSUSER

Then create a random password string, and add a line to /home/corpus-rsync/secrets with $CORPUSUSER and that password.

Finally, let the submitter know their new username and password.