mailinspect.1

dbacl是一个通用目的的digramic贝叶斯文本分类器。它可以学习你提供的文本

\" t
.TH MAILINSPECT 1 "Bayesian Text Classification Tools" "Version 1.3" ""
.SH NAME
mailinspect \- sort an mbox by category and pipe emails to a command.
.SH SYNOPSIS
.HP
.B mailinspect [-zjiI]
-c 
.I category
FILE [-gG 
.IR regex ]...
[-s
.IR command ]
[-p 
.IR style ]
[-o
.IR scoring ]
.HP
.B mailinspect
-V
.SH DESCRIPTION
.PP
.B mailinspect
reads the single mbox folder named FILE and sorts it in order of similarity to
the 
.IR category , 
which must have been 
created by 
.BR dbacl (1). 
It can be used as a command line tool or interactively, when given the 
.B -I 
switch. 
.PP
When used as a command line tool, 
.B mailinspect
prints the sorted list of emails on STDOUT. Each line consists of a
seek position for the given email within FILE, followed by the score and a description
string in one of several styles chosen via the 
.B -p
option. 
.PP
When supplying a 
.I command
string in conjunction with the 
.B -s
option, 
.B mailinspect
spawns a shell and executes 
.I command 
for every email in FILE (possibly selected via the 
.BR -g " or" -G
options), in the sorted order. This is similar to the 
.BR formail (1)
functionality, except the latter doesn't order the emails.
.PP
In interactive mode, all the command line functionality is available via
keypresses. The sorted list of emails is displayed in a scrollable format,
and can be viewed, searched, tagged, resorted and sent to shell commands. Predefined
shell commands can be associated with function keys. See the usage section below.
.PP
The sorting heuristics are currently (and may always be) experimental, 
so there is no guarantee that the orderings are particularly well suited
for anything. Normally, a smaller score means the email is closer to the
selected category, while a bigger score means it is further away.
.SH EXIT STATUS
.B
mailinspect
returns 1 on success, 0 if some error occurred.
.SH OPTIONS
.IP -c
Use 
.I category
to compute the scores and sort the emails, which 
should be the file name of a 
.BR dbacl (1)
category.
.IP -g
Only emails matching the regular expression
.I regex
are sorted. All other emails are ignored. When several
.B -g
and 
.B -G
options are present on the command line,
earlier regular expressions are overridden
by later ones where applicable.
.IP -i
Force internationalized mode. 
.IP -j
Force regular expression searches to be case sensitive.
.IP -o
Determines the scoring formula to be used. The parameter 
.I scoring
must be an integer greater than or equal to zero. By default,
.I scoring 
equals zero.
.IP -p
Prints the email index in the given style. The parameter 
.I style
must be an integer greater than or equal to zero. By default,
.I style
equals zero.
.IP -s
For each email in the list, execute the shell
.IR command ,
with the email body on STDIN. Emails are processed in sorted order.
.IP -z
Reverse sort order. Normally, emails are sorted in order of closest 
to furthest relative to 
.IR category ,
but in this case, the opposite is true. 
.IP -I
Interactive mode. Instead of printing the sorted list of emails
on STDOUT, emails are displayed and can be scrolled, viewed, 
searched and piped interactively at the terminal.
.IP -G
Only emails 
.B not
matching the regular expression
.I regex
are sorted. Opposite of 
.B -g
switch.
.IP -V
Print the program version number and exit. 
.SH USAGE
.PP
.B mailinspect 
needs to read a prelearned
.I category 
before it can sort the emails in FILE. See 
.BR dbacl (1).
.PP
Suppose you have two mail folders named 
.I good.mbox
and
.IR bad.mbox 
respectively. You can create appropriate categories by typing the commands
.PP
.na
% dbacl -l good good.mbox -T email
.br
% dbacl -l bad bad.mbox -T email
.ad
.PP
Next, you can type the following command to view interactively the 
.I bad.mbox
file with the emails whose score is closest to the category
.I good
listed first:
.PP
.na
% mailinspect -I -c good bad.mbox 
.ad
.PP
Alternatively, you might be interested only in the five emails in the folder 
.I bad.mbox
whose score marks them as the furthest away from the category 
.BR bad ,
completely independently from any other category such as 
.I good 
(ie you want outliers in the scoring sense).
.PP
.na
% mailinspect -z -c bad bad.mbox | head -5
.ad
.PP
In interactive mode, the following keys are defined:
.IP o
toggles another scoring formula.
.IP p
toggles another display style.
.IP q
exits 
.BR mailinspect .
.IP s
sends the currently highlighted email to a shell command.
.IP S
sends all currently tagged emails to a shell command, in sorted order.
Every email executes the shell command independently.
.IP t
tags the currently highlighted email.
.IP T
tags all listed emails.
.IP v
sends the currently highlighted email to $PAGER for viewing. If the environment
variable PAGER is not defined, sends the email to 
.BR less (1).
.IP u
untags the highlighted email.
.IP U
untags all listed emails.
.IP z
reverses the sort order of displayed emails.
.IP /
searches for a regular expression (see 
.BR regex (7))
anywhere within the contents of all listed emails. Hides all emails
which don't match.
.IP ?
like /, but hides all emails which match, keeping all those which don't match.
.PP
As a convenience, the function keys F1-F10 can each be associated 
with a shell command string. In this
case, typing a function key has the same effect as the S key, but the command is 
already typed and ready to be edited/accepted.
The function key associations are read from the configuration file .mailinspectrc
if it exits. 
.SH FILES
.PP
.IP $HOME/.mailinspectrc
.B mailinspect 
reads the file .mailinspectrc in the $HOME directory, if it exists. 
This is a plain text file which contains entries of the form
.IP
.na
# this is a comment
.br
F2 cat >> interesting.mbox
.br
F5 mail [e-mail]
.ad
.SH ENVIRONMENT
.PP
.IP DBACL_PATH
When this variable is set, its value is prepended to every 
.I category
filename which doesn't start with a '/'.
.SH SOURCE
.PP
The source code for the latest version of this program is available at 
http://www.lbreyer.com/gpl.html
.SH AUTHOR
.PP
Laird A. Breyer <[e-mail]>
.SH SEE ALSO
.PP
.BR bayesol (1),
.BR dbacl (1),
.BR less (1),
.BR mailcross (1),
.BR regex (7)