Follow @Openwall on Twitter for new release announcements and other news

blists is a web-based interface to mailing list archives that works off indexed mbox files. There are two programs: bindex and bit. bindex generates or updates the index file (yes, incremental updates are supported). bit is a CGI program, which generates web pages on the fly.

  blists homepage:
  Live example with a high volume mailing list:

To compile, simply run "make". There's currently no "install" target; you're supposed to copy the bindex and bit programs in place on your own, as appropriate for your setup. Also you may need to setup for your httpd: cgi, SSI (shtml), and mod_rewrite.

You will likely want to have bindex run after new messages arrive. For example, you may invoke it from a .procmailrc file like this:

    * ^TOlistname
      :0 c
      | /usr/bin/bindex Mail/listname

This delivers new messages to an mbox file called "listname" and it immediately triggers update of the index file for it. You can also accomplish this from .forward and .qmail files. Alternatively, you may choose to run bindex on cron.

The index file name is produced by adding the .idx suffix to the mbox filename, so in this example it will be "listname.idx" in the same directory. With the default params.h settings, the index file size is typically 100 KB plus around 3.5% of the mbox file's size.

bit is meant to be invoked via SSI (it will refuse to work otherwise), and it has only been tested with Apache so far. Here's an example SSI-enabled HTML file (usually with extension .shtml):

    <!DOCTYPE html>
    <!--#include virtual="/cgi-bin/bit?header"-->
    <style type="text/css">
    .cal_brief { text-align: center; }
    .cal_brief td:first-child { background: inherit; }
    .cal_brief td { background: #ccc; width: 5ex; padding: 2px; }
    .cal_big { text-align: center; padding: 0; margin: 0; }
    .cal_big td { padding: 0 2px; }
    .cal_mon { text-align: center; }
    .cal_mon th { font-size: small; padding: 0; margin: 0; }
    .cal_mon td { background: #ccc; width: 5ex; height: 1.5em;
                padding: 2px; text-align: right; }
    .cal_mon td[colspan] { background: inherit; }
    .cal_mon sup { color: #F0F0F0; text-align: left; float: left;
                margin-top: -2pt; font-weight: bold; }
    .cal_mon a { text-align: right; margin-left: -4em; float: right; }
    <!--#include virtual="/cgi-bin/bit?body"-->

bit output it in UTF-8, so you will need to configure charset, for Apache add this (for example to to .htaccess in the directory where bit.shtml is):

    AddCharset UTF-8 .shtml

Obviously, you'll also need to adjust the /cgi-bin/bit paths, and you might need to add a filename suffix to match your web server configuration (for example it could be bit.cgi).

You may need to configure MAIL_SPOOL_PATH definition in params.h to tell bit where mboxes are located, otherwise bit will assume they are in ../../blists/ relative to cgi-bin directory (where bit is).

In order for the links generated by bit to point to valid URLs, as well as for the URLs to look pretty, you may use mod_rewrite rules like this:

    RewriteEngine On
    RewriteRule ^((listname1|listname2)/([0-9]{4}/([0-9]{2}/([0-9]{2}/([1-9][0-9]*)?)?)?)?)$ list.shtml?$1 [L]
    RewriteRule ^((listname1|listname2)/[0-9]{4}/[0-9]{2}/[0-9]{2}/[1-9][0-9]*/[1-9][0-9]*)$ /cgi-bin/bit?attachment+$1 [L]

Direct call to bit is required to set HTTP headers for attachments.

To workaround a bug in Lynx where it would omit the trailing slash when following links to "..", add:

    RewriteRule ^[a-z-]+([/0-9]*[0-9])?$ https://%{SERVER_NAME}%{REQUEST_URI}/ [R,L]

(where "[a-z-]+" is supposed to match your list names; adjust it if not).

To have separate HTML wrapper pages for different lists (such as to include different additional info on those pages), use:

    RewriteRule ^(listname1|listname2)/(([0-9]{4}/([0-9]{2}/([0-9]{2}/([1-9][0-9]*)?)?)?)?)$ list-$1.shtml?$1/$2 [L]

To make use of the censorship feature (to hide spam messages), create a separate HTML wrapper page with:

    <!--#include virtual="/cgi-bin/bit?header-censored"-->
    <!--#include virtual="/cgi-bin/bit?body-censored"-->

then refer to it in more specific RewriteRule directives, which you need to place above the catch-all ones:

    RewriteRule ^(listname1/2011/01/02/3)$ list-censor.shtml?$1 [L]

You may match multiple messages at once with trickier regexps:

    RewriteRule ^(listname1/2011/01/(09/1|12/1|12/2))$ list-censor.shtml?$1 [L]

Good luck!