MArch

MArch (march-20150126T142331.tbz), pronounced as the month, is a daemon program that saves email going through a mail server into a database. It plugs in the Sendmail mail delivery system as a milter. Messages can then be recovered, searched, read, forwarded via an HTML interface.

Here are some snapshots of the web interface (new versions of MArch may look slightly different):

	The search results page.
	The hot-spots show: Query entries with constraints on sender, recipient, subject, month, and category. The list order can be by date, subject, or sender. By default the search page shows today's messages, but you can choose another day, month or week. Menu bar that appears on every page. This includes some shortcuts to get to the list of all messages of this month/week/day and the one before. A navigation bar to move back and forth days/weeks/months. Below is the list of messages satisfying the query.
	The bottom of the search results page.
	The hot-spots show: Smile icon, when message is in the ham category (not spam). Devil icon, when message is in the spam category. Balloon icon, when message is neither in the spam nor in the ham category. That is, the Bayesian filter was unsure about the quality of the message contents. Outgoing icon, when message has been leaving the domain. Incoming icon, when message has been entering the domain. Number of messages satisfying the search query.
	The top of a message page displaying a non-spam message.
	The hot-spots show: Navigation bar to quickly jump to the most used pages. Some summary data, including information from the message headers (sender, subject, receiver, date), the internal MArch ID, the list of categories it belongs to and the spam score (normally not shown). The rest is the message body.
	The bottom of a message page displaying a non-spam message.
	The hot-spots show: Thumb down button to change the rank of the message to spam. Magnifying lens button to get to the raw message (including headers). Input box to forward the message to someone. Current position in the discussion thread. In the blue box you see the discussion thread.
	A non-spam message which the Bayesian filter hasn't been able to asses with certainty.
	A spam message which the Bayesian filter hasn't been able to asses with certainty.
	Statistics page about some general facts of the database and the type of traffic sen by MArch since it was started.
	Today's statistics of the database contents, divided per hour. Included is a line chart of the frequencies per hour and a pie chart of the overall contents divided per ham (green), spam (red), unsure (grey).
	Statistics for the current month, divided per day. Included is a line chart of the frequencies per day and a pie chart of the overall contents divided per ham (green), spam (red), unsure (gray).
	Two tables showing the top keywords (features) currently stored in the Bayesian table. The one on the left is the list of keywords with the highest spam rank. The one on the right is the list of most frequent keywords. On the top you see some figures regarding the overall spam table and the cache used by MArch to speed up their access.

To compile MArch you need SCLF, CLOT, demyltify, MIME4CL, HTML-sugar, and CLiconv, all from this page, plus a handful of other external packages/libraries. See the README for further details.

MArch has been developed and tested on FreeBSD running CMUCL. It should be rather portable, though.

All the photographs on these pages are copyright Walter C. Pelissero. All rights are reserved. No copy, manipulation or transfer, beside for the sake of personal view of the page where they appear on, is allowed. Also the direct linking of web pages to the single photograph image is forbidden.