Poster of Linux kernelThe best gift for a Linux geek
mailliststat

mailliststat

Section: User Manuals (1) Updated: June 2003
Local index Up
 

NAME

mailliststat - Display useful statistics on email messages  

SYNOPSIS

mailliststat [-hvq] [-i file] [-o file] [-r|w|u file] [-t|T text] [-m mode] [-n XX] [-g xxxx]  

DESCRIPTION

MailListStat is program that prints some "useful" statistical info on email messages. It's main usage is in email conferences - mailing lists. Currently it displays both tables and graphs. You can select either TEXT or HTML output.  

OPTIONS

-h
print help text and exit
-q
be quiet (print only errors to stderr)
-v
turn on verbose mode - in this mode it will print more info to stderr - indication of progress (will print every 10th, 20th, ..., 90th, 100th, 200th, ..., 900th, 1000th, 2000th ... message being processed) and warnings about malformed headers found
-i file
name of input file (if not specified, use stdin). This file should be in MBOX format. It should exist and be readable.
-o file
name of output file (if not specified, use stdout). If exists, it will be overwritten.
-r file
read input from cache file instead of mailbox. You can read input either from mailbox or cache file, not both!
-w file
write cache file (no stats produced). You can either produce text output or write cache file, not both! When writing cache file, output-related options are ignored.
-u file
update cache file = read cache, read input, write cache. For use with .procmailrc/.forward
-t text
name of mailing list this statistics is computed for. If specified, it is just appended to the title of statistics, so it will be like "Statistics from 16.8.2001 to 7.9.2001 for text", where text is whatever you put as this parameter (it could be name of the mailing list or just its email, e.g. mobil@mobil.sk).
-T text
title text (only this will be printed as title); this can be used to supress normal title text (date of oldest/newest msg) and completely replace it with your text.
-m mode
select mode of output (text, html, html2).
-n XX
show TOP XX tables (default TOP 10). By default, mailliststat displays tables of TOP 10 people, subjects, quoting or whatever. Using this parameter, you can define how many lines shall these tables have.
-g xxxx
graphs to show (Day, Week, Month, Year, Xnone) - specify first letter (e.g. -g dmy).
 

EXIT STATUS

0
Everything went OK and no error occurred.
1
Error in sscanf() while reading & parsing cache file. It means that the format of cache file is invalid. Try to create the cache file again.
2
Invalid command-line option. You have specified an invalid command-line parameter.
3
Cannot open input/output file. Please check that you have typed correct filename and that you have read permissions for input file and write permissions to destination directory (because output file must be created). If output file exists, it's overwritten.
4
Not enough memory is available for dynamically allocated variables. This could be caused by user-limits, because mailliststat requires only few MBs of memory (it depends on number of messages processed and number of different subjects and authors).
5
Error compiling regex. This error should not occur in world-available versions.
 

USAGE

 

Input

On input, there should be mailbox file in standard MBOX format. If the file is in different format, the results are unpredictable. There should be at least one email message, otherwise no stats can be computed.

Warning: Be sure that no special messages are in input files (such as that with "DON'T DELETE THIS MESSAGE -- FOLDER INTERNAL DATA" subject), because they will be also analysed. Many programs (POP3/IMAP daemons, email readers) put their special messages to the mailbox. This message is only ignored when reporting oldest message found.  

Output

Statistics is put into output file (or stdout if unspecified). All diagnostic messages are written to stderr. Output consists of several statistical data - tables, graphs and summaries. The title has two formats depending on -t parameter. If it's not specified, it looks like "Statistics from 16.8.2001 to 7.9.2001", where first date is date of the oldest message found in input and second is date of the newest one. If there is for example -t mobil@mobil.sk parameter, it will look like "Statistics from 16.8.2001 to 7.9.2001 for mobil@mobil.sk". The problem is that date of oldest & newest msg is often wrong (thanks to bad date/time settings on PC of msg author), so you can specify entire title using command-line option -T. When used, only your text will be printed as title, nothing more. There you can put for example something like "Statistics for mobil@mobil.sk".

Now you have option ( -g) to specify which graphs you want to show - hours of Day, days of Week, days of Month, months of Year. Use 1st letters as argument to -g option (so -g dw will print just hours of Day and days of Week). Use -g x to disable printing of any graph. For example you don't want to show graph for months of Year if you are presenting stats for one month, but for full-year stats you probably want it.  

HTML output

You can choose between 2 modes of output - TEXT and HTML. When in HTML mode, mailliststat will produce the output as HTML page. When you specify HTML2 mode, only the body of HTML document is produced (no header/footer) - it can be used to have different HTML header/footer when calling mailliststat as CGI or when using PHP wrapper. The output consists of HTML tables and bar graphs. Almost every aspect of how it looks can be configured by modifying CSS style-sheet. Please note that files style_mls.css and bar.gif must be present in the same directory as produced HTML file. You can, however, modify both to best suit your needs. Everything should be clear after reading comments in CSS file and looking at the produced HTML source.

I was unsure what type of graphs to produce. I have tried also horizontal bar graphs and if you want to try them, just uncomment part of code in PrintGraphHtml() in mls_text.c.  

Cache file support

Instead of producing statistics in text format, you can save all the generated values/results into "cache" file. Retrieving information from this file is very fast, so it is useful for integration with web pages. Now you can update the cache file just after new mail was received. Users can view actual stats using mailliststat
 as CGI script. It has an advantage over static stats that user can choose options and it will be generated in a moment!

To update cache file, use the -u option. It works like this: first, the stats are loaded from cache file (doesn't have to exist) and then new message(s) to be added are read from stdin (or from -i file) and added to the stats. Finally the updated stats are written back to the cache file. The process is really quick, because usually only one message is added at a time. This is useful mainly for updating cache files upon receiving new message. In the "examples/" subdir, you can find examples of integration with your .forward and .procmailrc files. By running MLS more than once, you can generate cache files for individual months and also for whole years (see examples). Then use some PHP script to present list of these cache files to user.

Format of cache files was changed in version 1.3, because of new stats added. Now it contains version info, so mailliststat can inform you that you have to re-create that cache file with new version. Unfortunately, you have to re-create them also when you want new email clients to be recognized also in old (already processed) messages. Note that email clients detection was buggy in 1.2.2 (a lot of clients not recognized).  

PHP wrapper

I have written also PHP wrapper for mailliststat to make it more "interactive". It has one major advantage over plain HTML output from mailliststat: User can choose output number of TOP items to show. It works by running mailliststat with appopriate command-line options. It's safe, because only one item from user is topXX which is checked using regexp, so running arbitrary code is not possible. You can also alter mailliststat output - for example change @ in email addresses to (at) to prevent spamming.
   You can have normal MBOX file as input, but I recommend using cache file. When using cache file, the stats are produced in a moment. You can see how long it took to generate the page, see the last line of HTML source. However, there is minor speed problem. It takes longer when you specify to show many topXX (like 999). The problem is regexp that searches for @. It has to search for it in whole mailliststat output together and when it is large, it takes a while (1.1 seconds on my 2.1GHz pentium4). I have added an option which should use Perl-compatible regex function (preg_replace) instead of POSIX (ereg_replace), if available. This will result in MUCH faster execution (50ms instead of 1.1sec).  

NOTES

 

How it is all computed?

OK, so let's start from beginning - the format of MBOX file. It's plain text file containing some email messages delimited with one empty line. Each message starts with line like this From abc@a.sk Thu Aug 16 15:48:58 2001. After this line, there are few headers, one empty line and message text. Storing emails in this format is quite common - your incoming mail is usually saved in MBOX format and also your folders in mail-readers like elm(1), pine(1), mutt(1)...

Who is author of an email message? It's taken from From: header field and everything except the actual email address (like your full name) is stripped off using quite simple regular expression (regexp).

Subject is taken from Subject: header field. If it contains some Re:, those will be stripped off. There can be up to 5 of them. Also counted format ( Re[3]:) is supported. For example The Bat! email client uses it. MIME-decoding is applied to subject lines (see below).
   Date is just everything in the Date: header. This header is generated by the email client, so it's date of message creation and it doesn't have to be present in each message. If it isn't, you are warned by message like "Warning: 1 message(s) not counted." in output. Some clients don't put full date there and usually the day of week is missing and you are warned. No timezones are considered, the date is taken as-is.

Message size is everything between end of message header and beginning of new email (or end of file). So only actual size of message text (body) is counted, not headers.

Email clients are taken from X-Mailer: or User-Agent: or X-Newsreader: headers and some grouping is done to avoid different versions of the same mailer to take the whole TOP 10. There is also work-around for Pine mailer (MLS will search also Message-ID: header).  

What is quoting? Why I have it 95%?

What is quoting? When you reply to some message, you can insert part of the original message there, you quote the author of original message. Every line of original text is usually prepended with > or MP>, where MP are initials of the original sender's name (for example The Bat! uses this second format).

And what is "quote ratio"? It's size of quoted text divided by total size of message, specified in percent. It's included in stats, because many people reply to message, add one line of text and leaving there for example 10 pages of original text, which makes the quote ratio even higher than 90%! In times of FIDONET, there were conferences, where quote ratio higher than 50% was forbidden. Try to think about it when replying to message in mailing list where more than 300 people will download and read it.  

And now all the stats

At first, there are TOP 10 tables (or TOP XX when using -n XX parameter). First table shows people who have written most messages, how much and how many percent of total message count it is. Last row shows the "other" - number of messages written by everyone not listed above and how many percent it is. Second and third tables are similar to this one - they also show best authors, but not by the number of messages written. Authors are sorted by total (or average) size of all their messages, but without quoting (size of message minus how much was quoted in that msg). Next table shows most successful subjects and how many messages with this subject have been posted. The other table shows most used email clients. The last table show people with maximal quote ratio. It's computed as sum of quoted text in all his/her messages divided by total size of those messages. Last row shows an average - sum of quoted text in all messages divided by total size of all messages.

Next part of stats are some graphs. They show how much messages have been written during different hours of day, days of month and days of week. From these you can see for example when (and how much) people sleep :) or if they work during the working-hours or just write tons of messages...

Next part contains info about messages which are BEST in something - message with max. quote ratio, longest message and some details about most successful subject.

At the end, there is final summary - total number of messages, their total and average size and number of different authors and subjects.  

MIME (Multipurpose Internet Mail Extensions)

What is it? Original implementation email permitted only 7bit ASCII messages. But during the time, there was need to send international or even binary files. MIME defines how can these be encoded into 7bit form suitable for emailing and how to decode it back to human readable form.

In email message, you can have MIME-encoded text (body of message), but also some headers - for example subject and From field. MLS tries to find out if subject lines are MIME-encoded and if so, it tries to decode it, to present it to you in human-readable form. You can read more about MIME in RFC 1521 and 1522.  

Inspiration

I was inspired by similar DOS program used before few years in FIDONET and Slovak ULTRANET. It was created by Ivan Friedlander.  

BUGS/TODO

doesn't support header fields splitted to more lines (you can use formail(1) to put them to one line before using MLS)
charset conversion in MIME-decoding
more stats
 

VERSION

This man page is written for mailliststat version 1.3.  

AUTHOR

mailliststat (MailListStat) is written by Marek -Marki- Podmaka <marki@nexin.sk>.  

SEE ALSO

Visit http://freshmeat.net/projects/mls for more information and latest version of mailliststat.  

COPYING

MailListStat - print useful statistics on email messages Copyright (C) 2001-2003 Marek Podmaka <marki@nexin.sk>

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA


 

Index

NAME
SYNOPSIS
DESCRIPTION
OPTIONS
EXIT STATUS
USAGE
Input
Output
HTML output
Cache file support
PHP wrapper
NOTES
How it is all computed?
What is quoting? Why I have it 95%?
And now all the stats
MIME (Multipurpose Internet Mail Extensions)
Inspiration
BUGS/TODO
VERSION
AUTHOR
SEE ALSO
COPYING

This document was created by man2html, using the manual pages.
Time: 21:14:47 GMT, April 16, 2011