.\" $Revision: 3.10.0 $
.TH SUCK 1
.SH NAME
suck - Pull a small newsfeed from an NNTP server, avoiding the NEWNEWS command.
.SH SYNOPSIS
.I suck
[
.BI
hostname
]
[
.BI @filename
]
[
.BI \-V
]
[
.BI \-K
]
[
.BI \-L[SL]
]
[
.Bi \-LF
filename
]
[
.BI \-H
]
[
.BI \-d[tmd]
dirname
]
[
.BI \-s\ |\ \-S
filename
]
[
.BI \-e\ |\ \-E
filename
]
[
.BI \-a
]
[
.BI \-m
]
[
.BI \-b[irlf]
batchfile
]
[
.BI \-r
filesize
]
[
.BI  \-p
extension
]
[
.BI \-U
userid
]
[
.BI \-P
password
]
[
.BI \-c
]
[
.BI \-M
]
[
.BI \-N
port_number
]
[
.BI \-W
pause_time pause_nr_msgs
]
[
.BI \-w
pause_time pause_nr_msgs
]
[
.BI \-l
phrase_file
]
[
.BI \-D
]
[
.BI \-R
]
[
.BI \-q
]
[
.BI \-C
count
]
[
.BI \-k
]
[
.BI \-A
]
[
.BI \-AL
activefile
]
[
.BI \-hl
localhost
]
[
.BI \-bp
]
[
.BI \-T
timeout
]
[
.BI \-n
]
[
.BI \-u
]
[
.BI \-z
]
[
.BI \-x
]
[
.BI \-B
]
[
.BI \-O
]
[
.BI \-G
]
[
.BI \-X
]
[
.BI \-f
]

Options valid in all modes

\@filename 

This option tells suck to read other options from a file in addition to the
commandline.

\-a

This option forces suck to always batch up any downloaded articles,
even if suck aborts for any reason.  Without this option, suck will
only batch up articles if it finishes successfully or is cancelled by
a signal (see below).

\-A

This option tells suck to scan the localhost (specified with the \-hl option) and use its active file
to build and update the sucknewsrc.   If you add a group to your local server, suck will add it to
sucknewsrc and download articles.  Or, if you delete a group from your local server, it will be deleted
from sucknewsrc.  If posting is not allowed to a particular group, then the line in sucknewsrc is
just commented out.  With this option, you should never have to edit your sucknewsrc.  In case you have
newsgroups (like control and junk) that you don't want downloaded, you can put these newsgroups in a
file "active-ignore", one per line, and suck will ignore these newsgroups when it scans the localhost.
If your system supports regex(), you may use regular expressions in the active-ignore file to skip multiple groups, eg: fred\.*.
If you use the -p (postfix) option, suck will check for the existence of an active-ignore file with the
postfix.  If that doesn't exist, then suck will check for the existence of the file without the postfix.

NOTE: If you use regular expressions, suck will silently add a "^" to the beginning of the group name, 
and a "$" to the end of the group name if they aren't already present, so that if you have
"comp.os.linux", it won't match "comp.os.linux.answers" or if you have alt.test it doesn't match comp.alt.test.

\-AL activefile

This option is identical to the -A option, except it reads the active file from the local file specified instead of 
reading it from the localhost.  All the caveats from the -A option apply to this option as well.  If both
options are used on the command line, suck first tries to use the -A option, then if that fails it uses
this option.

\-B

This option tells suck to attempt to batch up any articles in its directory
BEFORE starting to download messages.  This can be useful if you have a 
problem with the previous download.  This option will only work if you specify
a batch option (see below).   If there are no messages to batch up, some
of the batch options may produce warning messages.  They may be safely ignored.
Also, if the batch files exist at the end of the run, in inn-batch mode, it
will be overwritten, since the new batch file will contain all messages.  In
rnews mode, if the batch file exists, it will abort and not batch up any messages.

\-c

If this option is specified, suck will clean up after itself.  This includes:
.RS
1. Moving sucknewsrc to sucknewsrc.old
.RE
.RS
2. Moving suck.newrc to sucknewsrc
.RE
.RS
3. rm suck.sorted and suckothermsgs.
.RE

\-C count

This option tells suck to drop the connection and reopen it every count number of articles.
This is designed to battle INN's LIKE_PULLERS=DONT option, that some folks compile in. With
LIKE_PULLERS=DONT, after 100 messages INN will pause between every message, dramatically 
reducing your download speed. I don't recommend the use of this, but if you have no other choice....  

\-dd dirname

\-dm dirname

\-dt dirname

Specify the location of the various files used by suck.

\-dd dirname = directory of data files used by suck (sucknewsrc suckkillfile suckothermsgs active-ignore)

\-dm dirname = directory for storage of articles created in Multifile mode
or batch mode.  DO NOT make this the same as the directories used for the
\-dt or -\dd options, or you will lose all your configuration files.

\-dt dirname = directory of temp files created by suck (suck.newrc, suck.sort, suck.restart, suck.killlog, suck.post).

\-D

This option tells suck to log various debugging messages to "debug.suck", primarily
for use by the maintainer.

\-e | \-E filename

These options will send all error messages (normally displayed on stderr), to 
an alternate file.  The lower case version, -e, will send the error messages
to the compiled-in default defined in suck_config.h.  The default is suck.errlog.
The upper case version, -E, requires the filename parameter.  All error messages
will then be sent to this file.

\-f 

This option tells suck to reconnect after deduping, and before downloading the articles.  This is in case
long dedupe times cause timeouts on the remote end. 

\-G

This option causes suck to display the message count and BPS status lines in a slightly different format,
more suitable for use by a filter program (such as a GUI).

\-H

This option will cause suck to bypass the history check.

\-hl localhost

This option specifies the localhost name.  This option is required with both the \-A and the \-bp option.  

\-k

This option tells suck to NOT attach the postfix from the \-p option to the names of the killfiles,
both the master killfile and any group files.  This allows you to maintain one set of killfiles for
multiple servers.

\-K

This option will cause suck to bypass checking the killfile(s).

\-l phrase_file

This option tells suck to load in an alternate phrase file, instead of using
the built-in messages.  This allows you to have suck print phrases in another
language, or to allow you to customize the messages without re-building suck.
See below.

\-L

This option tells suck to NOT log killed articles to suck.killlog.

\-LF filename

This option allows you to override the built-in default of "suck.killlog" for the
file which contains the log entries for killed articles.

\-LL

This option tells suck to create long log entries for each killed article.  The long
entry contains the short log entry and the header for the killed message.

\-LS

This option tells suck to create short log entries for each killed article.  The short
entry contains which group and which pattern was matched, as well as the MsgID of the
killed article.

\-M

This option tells suck to send the "mode reader" command to the remote
server.  If you get an invalid command message immediately
after the welcome announcement, then try this option.

\-n

This option tells suck to use the article number vice the MsgId to retrieve the articles.  This
option is supposedly less harsh on the remote server.  It can also eliminate problems if your
ISP ages off articles quickly and you frequently get "article not found" errors.

\-N port_number

This option tells suck to use an alternate NNRP port number when connecting
to the host, instead of the default, 119.

\-O

This option tells suck to skip the first article upon restart.  This is used whenever
there is a problem with an article on the remote server.  For some reasons, some
NNTP servers, when they have a problem with a particular article, they time out.  
Yet, when you restart, you're back on the same article, and you time out again.
This option tells suck to skip the first article upon restart, so that you can
get the rest of the articles.

\-p extension

This extension is added to all files so that you can have multiple site feeds.
For example, if you specify -p .dummy, then suck looks for sucknewsrc.dummy, suckkillfile.dummy,
etc, and creates its temp files with the same extension.  This will allow you to keep
multiple sucknewsrc files, one for each site.

\-q

This option tells suck to not display the BPS and article count messages during download.
Handy when running suck unattended, such as from a crontab.

\-R

This option tells suck to skip a rescan of the remote newserver upon a restart.  The
default is to rescan the newserver for any new articles whenever suck runs, including
restarts.

\-s | \-S filename

These options will send all status messages (normally displayed on stdout), to
an alternate file.  The lower case version, -s, will send the status messages
to the compiled-in default defined in suck_config.h.  The default is /dev/null,
so no status messages will be displayed.  The upper case version, -S, requires
the filename parameter.  All status messages will then be sent to this file.

\-T timeout

This option overrides the compiled-in TIMEOUT value. This is how long suck waits for data from the
remote host before timing out and aborting.  The timeout value is in seconds.

\-u 

This option tells suck to send the AUTHINFO USER command immediately upon connect to the
remote server, rather than wait for a request for authorization.  You must supply the
\-U and \-P options when you use this option.

\-U userid

\-P password

These two options let you specify a userid and password, if your NNTP server
requires them.

\-V

This option will cause suck to print out the version number and then exit.

\-w pause_timer pause_nr_msgs

This option allows you to slow down suck while pulling articles.  If you
send suck a predefined signal (default SIGUSR1, see suck_config.h),
suck will swap the default pause options (if specified by the -W option),
with the values from this option.  For example, you run suck with -w 2 2,
and you send suck a SIGUSR1 (using kill), suck will then pause 2 seconds
between every other message, allowing the server to "catch its breath."
If you send suck another SIGUSR1, then suck will put back the default
pause options.  If no pause options were specified on the command line
(you omitted -W), then suck will return to the default full speed pull.

\-W pause_time pause_nr_msgs

This option tells suck to pause between the download of articles.  You need
to specify how long to pause (in seconds), and how often to pause (every X nr
of articles). Ex: \-W 10 100 would cause suck to pause for 10 seconds every
100 articles.  Why would you want to do this?  Suck can cause heavy loads on 
a remote server, and this pause allows the server to "catch its breath."

\-x

This option tells suck to not check the Message-IDs for the ending > character.  This option
is for brain dead NNTP servers that truncate the XHDR information at 72 characters.

\-X

This option tells suck to bypass the XOVER killfiles.

\-z

This option tells suck to bypass the normal deduping process.  This is primarily for
slow machines where the deduping takes longer than the download of messages would.  Not
recommended.

.SH LONG OPTION EQUIVALENTS
.RS
\-a  \-\-always_batch
.RE
.RS
\-bi \-\-batch-inn
.RE
.RS
\-br \-\-batch_rnews
.RE
.RS
\-bl \-\-batch_lmove
.RE
.RS
\-bf \-\-batch_innfeed
.RE
.RS
\-bp \-\-batch_post
.RE
.RS
\-c  \-\-cleanup
.RE
.RS
\-dt \-\-dir_temp
.RE
.RS
\-dd \-\-dir_data
.RE
.RS
\-dm \-\-dir_msgs
.RE
.RS
\-e  \-\-def_error_log
.RE
.RS
\-f  \-\-reconnect_dedupe
.RS
\-h  \-\-host
.RE
.RS
\-hl \-\-localhost
.RE
.RS
\-k  \-\-kill_no_postfix
.RE
.RS
\-l  \-\-language_file
.RE
.RS
\-m  \-\-multifile
.RE
.RS
\-n  \-\-number_mode
.RE
.RS
\-p  \-\-postfix
.RE
.RS
\-q  \-\-quiet
.RE
.RS
\-r  \-\-rnews_size
.RE
.RS
\-s  \-\-def_status_log
.RE
.RS
\-w  \-\-wait_signal
.RE
.RS
\-x  \-\-no_chk_msgid
.RE
.RS
\-z  \-\-no_dedupe
.RE
.RS
\-A  \-\-active
.RE
.RS
\-AL \-\-read_active
.RS
.RE
\-B   \-\-pre-batch
.RE
.RS
\-C  \-\-reconnect
.RE
.RS
\-D  \-\-debug
.RE
.RS
\-E  \-\-error_log
.RE
.RS
\-G  \-\-use_gui
.RE
.RS
\-H  \-\-no_history
.RE
.RS
\-K  \-\-killfile
.RE
.RS
\-L  \-\-kill_log_none
.RE
.RS
\-LS \-\-kill_log_short
.RE
.RS
\-LL \-\-kill_log_long
.RE
.RS
\-M  \-\-mode_reader
.RE
.RS
\-N  \-\-portnr
.RE
.RS
\-O \-\-skip_on_restart
.RE
.RS
\-P  \-\-password
.RE
.RS
\-R  \-\-no_rescan
.RE
.RS
\-S  \-\-status_log
.RE
.RS
\-T  \-\-timeout
.RE
.RS
\-U  \-\-userid
.RE
.RS
\-V  \-\-version
.RE
.RS
\-W  \-\-wait
.RE
.RS
\-X  \-\-no_xover

.SH DESCRIPTION

.SH MODE 1 \- stdout mode
.RS
%suck
.RE
.RS
%suck myhost.com
.RE
.PP
Suck grabs news from an NNTP server and sends the articles to
stdout. Suck accepts as argument the name of an NNTP server or
if you don't give an argument it will take the environment variable
NNTPSERVER. You can redirect the articles to a file or compress them
on the fly like "suck server.domain | gzip \-9 > output.gz".
Now it's up to you what you do with the articles.  Maybe
you have the output already on your local machine because you
used a slip line or you still have to transfer the output to your
local machine.
.SH MODE 2 \- Multifile mode
.RS
%suck \-m
.RE
.RS
%suck myhost.com \-m
.RE
.PP
Suck grabs news from an NNTP server and stores each article in a
separate file.  They are stored in the directory specified in suck_config.h or
by the \-dm command line option.
.SH MODE 3 \- Batch mode
.RS
%suck myhost.com \-b[irlf] batchfile
.RE
.RS
or %suck myhost.com \-bp -hl localhost
.RE
.RS
%suck myhost.com \-b[irlf] batchfile
.RE
.PP
Suck will grab news articles from an NNTP server and store them
into files, one for each article (Multifile mode).  The location of the files
is based on the defines in suck_config.h and the command line \-dm.  
Once suck is done downloading the articles, it will build a batch file
which can be processed by either innxmit or rnews, or it will call lmove
to put the files directly into the news/group/number format.

\-bi \- build batch file for innxmit.  The articles are left intact,
and a batchfile is built with a one\-up listing of the full path of each article.
Then innxmit can be called:

.RS
%innxmit localhost batchfile
.RE

\-bl \- suck will call lmove to put the articles into
news/group/number format.  You must provide the name of the
configuration file on the command line.  The following arguments from suck 
are passed to lmove:

.RS
The configuration file name (the batchfile name provided with this option)
.RE
.RS
The directory specified for articles (-dm or built-in default).
.RE
.RS
The errorlog to log errors to (-e or -E), if provided on the command line.
.RE
.RS
The phrases file (-l), if provided on the command line.
.RE
.RS
The Debug option, if provided on the command line.
.RE

\-br \- build batch file for rnews.  The articles are
concatenated together, with the #!rnews size
article separator.  This can the be fed to rnews:

.RS
%rnews \-S localhost batchfile
.RE

\-r filesize  specify maximum batch file size for rnews.  This option
allows you to specify the maximum size of a batch file to be fed to rnews.
When this limit is reached, a new batch file is created AFTER I finish
writing the current article to the old batch file.  The second and
successive batch files get a 1 up sequence number attached to the
file name specified with the -br.  Note that since I have to finish
writing out the current article after reaching the limit, the
max file size is only approximate.

\-bf \- build a batch file for innfeed.  This batchfile contains the
MsgID and full path of each article.  The main difference between this
and the innxmit option is that the innfeed file is built as the articles
are downloaded, so that innfeed can be posting the articles, even while
more articles are downloaded.

\-bp \- This option tells suck to build a batch file, and post the articles
in that batchfile to the localhost (specified with the \-hl option).  This option
uses the IHAVE command to post all downloaded articles to the local host.
The batch file is called suck.post, and is put in the temporary directory (-dt).
It is deleted upon completion, as are the successfully posted articles.
If the article is not wanted by the server (usually because it already exists on
the server, or it is too old), the article is also deleted.  If other errors
occur, the article is NOT deleted.
With the following command line, you can download and post articles without
worrying if you are using INND or CNEWS.  

.RS
%suck news.server.com -bp -hl localhost -A -c
.RE

.SH SUCK ARGUMENT FILE
.PP
If you specify @filename on the command line, suck will read from filename and
parse it for any arguments that you wish to pass to suck.  You specify the
same arguments in this file as you do on the command line.  The arguments
can be on one line, or spread out among more than one line.  You may also
use comments.  Comments begin with '#' and go to the end of a line.  All
command line arguments override arguments in the file.

.RS
# Sample Argument file
.RE
.RS
-bi batch # batch file option
.RE
.RS
-M	# use mode reader option
.RE

.SH SUCKNEWSRC
.PP
Suck looks for a file
.I sucknewsrc
to see what articles you want and
which you already received. The format of sucknewsrc is very simple. It
consists of one line for each newsgroup.  The line contains two or
three fields.
The first field is the name of the group.  The next field is the highest
article number that was in the group when that group was last downloaded.
The third field, which is optional, limits the number of articles which
can be downloaded at any given time.  If there are more articles than this
number, only the newest are downloaded.  If the third field is 0, then
no new messages are downloaded.
The fields are separated by a space.

.RS
comp.os.linux.announce 1 [ 100 ]
.RE
.PP
When suck is finished, it creates the file suck.newrc which contains the
new sucknewsrc with the updated article numbers.
.PP
To add a new newsgroup, just stick it in sucknewsrc, with a
highest article number of \-1 (or any number less than 0).
Suck will then get the newest X number of messages for that newsgroup.
For example, a -100 would cause suck to download the newest 100
articles for that newsgroup.
.PP
To tell suck to skip a newsgroup, put a # as the first
character of a line.

.SH SUCKKILLFILE and SUCKXOVER
There are two types of killfiles supported in suck.  The first, via
the file suckkillfile, kills articles based on information in the
actual article header or body.   The second, via the file suckxover,
kills articles based on the information retreived via the NNTP command
XOVER.  They are implemented in two fundamentally different ways.  The
suckkillfile killing is done as the articles are downloaded, one at a
time.  The XOVER killing is done while suck is getting the list of articles
to download, and before a single article is downloaded.  You may use
either, none or both type of killfiles.

.SH SUCKKILLFILE and GROUP KEEP/KILLFILES
If
.I suckkillfile
exists,  the headers of  all articles will be scanned and the article downloaded or not, 
based on the parameters in the files.  If no logging option is specified (see the -L options
above), then the long logging option is used.
.PP
Comments lines are allowed in the killfiles.  A comment line has a "#" in the first position.
Everything on a comment line is ignored.
.PP
Here's how the whole keep/delete package works.  All articles are checked against the
master kill file (suckkillfile).  If an article is not killed by the master kill file,
then its group line is parsed.  If a group file exists for one of the groups then the
article is checked against that group file.  If it matches a keep file, then it is
kept, otherwise it is flagged for deletion.  If it matches a delete file, then it is
flagged for deletion, otherwise it is kept.  This is done for every group on the group line.
.PP
NOTES: With the exception of the USE_EXTENDED_REGEX parameter, none of these parameters are
passed from the master killfile to the individual group file.  Each killfile is separate
and independant.  Also, each search is case-insensitive unless specifically specified by starting the
search string with the QUOTE character (see below).
.SH
PARAMETERS
.RS
LOWLINES=#######
.RE
.RS
HILINES=#######
.RE
.RS
NRGRPS=####
.RE
.RS
QUOTE=c
.RE
.RS
NON_REGEX=c
.RE
.RS
GROUP=keep groupname filename  OR
GROUP=delete groupname filename
.RE
.RS
PROGRAM=pathname
.RE
.RS
PERL=pathname
.RE
.RS
TIEBREAKER_DELETE
.RE
.RS
GROUP_OVERRIDE_MASTER
.RE
.RS
USE_EXTENDED_REGEX
.RE
.RS
HEADER:
.RE
.RS
Any Valid Header Line:
.RE
.RS
BODY:
.RE
.RS
BODYSIZE>
.RE
.RS
BODYSIZE<
.RE

.PP
All parameters are valid in both the master kill file and the group files, with the
exception of GROUP, PROGRAM, PERL, TIEBREAKER_DELETE, and GROUP_OVERRIDE_MASTER.
These are only valid in the master kill file.

.SH KILL/KEEP Files Parameters
.PP
.I HILINES=
Match any article longer than the number of lines specified.
.PP
.I LOWLINES=
Match any article shorter than the number of lines specified.
.PP
.I NRGRPS=
This line will match any article which has more groups than the number specified.
Typically this is used in a killfile to prevent spammed articles.
(A spammed article is one that is posted to many many groups, such
as those get-rich quick schemes, etc.)
.PP
.I HEADER:
.I Any Valid Header Line:
Suck allows you to scan any single header line for a particular pattern/string, or
you may scan the entire article header.  To scan an individual line, just specify
it, for example to scan the From line for boby@pixi.com, you would put

.RS
From:boby@pixi.com
.RE

Note that the header line EXACTLY matches what is contained in the article.  To scan
the Followup-To: line, simply put \"Followup-To:\" as the parameter. 
To search the same header line for multiple search items, then each search
item must be on a separate line, eg:
.RS
From:boby@xxx
.RE
.RS
From:nerd@yyy
.RE
.RS
Subject:suck
.RE
.RS
Subject:help
.RE
The parameter HEADER: is a special case of the above.  If you use the HEADER: parameter,
then the entire header is searched for the item.  You are allowed multiple HEADER: lines
in each killfile. 
.PP
When suck searches for the pattern, it only searches for what follows
the :, and spaces following the : are significant.  With the above example "Subject:suck",
we will search the Subject header line for the string "suck".  If the example had read
"Subject: suck", suck would have searched for the string " suck".  Note the extra space.
.PP
If your system has regex() routines on it, then the items searched for can be POSIX
regular expressions, instead of just strings.  Note that the QUOTE= option is still
applied, even to regular expressions.
.PP
.I BODY:
This parameter allows you to search the body of an article for text.  Again,
if your system has regex(), you can use regular expressions, and the QUOTE= option is
also applied.  You are allowed multiple BODY: lines in each killfile.
WARNING:  Certain regex combinations, especially with .* at the beginning,
(eg BODY:.*jpg), in combination with large articles, can cause the regex code
to eat massive amounts of CPU, and suck will seem like it is doing nothing.
.PP
.I BODYSIZE>
This parameter will match an article if the size of its body (not including the
header) is greater than this parameter.  The size is specified in bytes.
.PP
.I BODYSIZE<
This parameter will match an article if the size of its body, is less than this parameter.
The size is specified in bytes.
.PP
.I QUOTE=
This item specifies the character that defines a quoted string.  The default
for this is a ".  If an item starts with the QUOTE character, then the item is
checked as-is (case significant).  If an item does not start with the QUOTE character,
then the item is checked with out regard to case.
.PP
.I NON_REGEX=
This items specifies the character that defines a non-regex string.  The default 
for this is a %.  If an item starts with the NON_REGEX character, then the item
is never checked for regular expressions.  If the item doesn't start with the QUOTE
character, then suck tries to determine if it is a regular expression, and if it
is, use regex() on it.  This item is so that you can tell suck to treat strings
like "$$$$ MONEY $$$$" as non-regex items.    IF YOU USE BOTH QUOTE and NON_REGEX
characters on a string, the NON_REGEX character MUST appear first.
.PP
.I GROUP=
This line allows you to specify either keep or delete parameters on a group
by group basis.  There are three parts to this line.  Each part of this line
must be separated by exactly one space.  The first part is either
"keep" or "delete".  If it is keep, then only articles in that group which match
the parameters in the group file are downloaded.  If it is delete, articles in that
group which match the parameters are not downloaded.  The second part, the group name
is the full group name for articles to check against the group file.   The group name
may contain an * as the last character, to match multiple groups, eg:  "comp.os.linux.*"
would match comp.os.linux.announce, comp.os.linux.answers, etc..  The third part
specifies the group file which contains the parameters to check the articles against.
Note, that if you specified a postfix with the \-p option, then this postfix is attached
to the name of the file when suck looks for it, UNLESS you use the \-k option above.
.PP
.I GROUP_OVERRIDE_MASTER
This allows you to override the default behavior of the master kill file.  If this
option is in the master kill file, then even if an article is flagged for deletion
by the master kill file, it is checked against the group files.  If the group files
says to not delete it, then the article is kept.
.PP
.I TIEBREAKER_DELETE
This option allows you to override the built-in tie-breaker default.  The potential
exists for a message to be flagged by one group file as kept, and another group
file as killed.  The built-in default is to then keep the message.  The TIEBREAKER_DELETE
option will override that, and caused the article to be deleted.
.PP
.I USE_EXTENDED_REGEX
This option tells suck to use extended regular expressions vice standard regular expressions.
It may used in the master killfile, in which case it applies to all killfiles, or in an
individual killfile, where it only applies to the parameters that follow it in the
killfile.
.PP
.I PROGRAM=
This line allows suck to call an external program to check each article.
You may specify any arguments in addition to the program name on this line.
If this line is in your suckkillfile, all other lines are ignored.  Instead, the
headers are passed to the external program, and the external program determines
whether or not to download the article.  Here's how it works.  Suck will fork
your program, with stdin and stdout redirected.  Suck will feed the headers
to your program thru stdin, and expect a reply back thru stdout.  Here's the 
data flow for each article:

.RS
1. suck will write a 8 byte long string, which represents the length of the
header record on stdin of the external program.  Then length is in ascii,
is left-aligned, and ends in a newline (example: "1234   \\n").
.RE
.RS
2. suck will then write the header on stdin of the external program.
.RE
.RS
3. suck will wait for a 2 character response code on stdout.  This response code is
either "0\\n" or "1\\n" (NOT BINARY ZERO OR ONE, ASCII ZERO OR ONE).  If the return
code is zero, suck will download the article, if it is one, suck won't.
.RE
.RS
4. When there are no more articles, the length written down (for step 1) will be zero
(again in ascii "0       \\n").  Suck will then wait for the external program to
exit before continuing on.  The external program can do any clean up it needs,
then exit.  Note: suck will not continue processing until the external program exits.
.RE

.PP
.I PERL=
This line allows suck to call a perl subroutine to check each article.  In order
to use this option, you must edit the Makefile, specifically the PERL* options. 
If the PERL=
line is in your suckkillfile, all other lines are ignored.  Instead, the header
is sent to your perl subroutine, and your subroutine determines if the article
is downloaded or not.  The parameter on the PERL= line specifies the file name
of the perl routine eg:

.RS
PERL=perl_kill.pl
.RE

.PP
See the sample/perl_kill.pl for a sample perl subroutine.  There are a couple of
key points in this sample.  The "package Embed::Persistant;" must be in the perl
file.  This is so that any variable names you create will not conflict with variable
names in suck.  In addition, the subroutine you define must be "perl_kill", unless
you change the PERL_PACKAGE_SUB define in suck_config.h.  Also, your subroutine must
return exactly one value, an integer, either 0 or 1.  If the subroutine returns
0, then the article is downloaded, otherwise, the article is not downloaded.

.PP
NOTES: The perl file is only compiled once, before any articles are downloaded. 
This is to prevent lengthy delays between articles while the perl routine
is re-compiled.  Also, you must use Perl 5.003 or newer.  In addition, you
are advised to run 'perl -wc filter' BEFORE using your filter, in order
to check for syntax errors and avoid problems.

.SH SUCKXOVER
If the file
.I suckxover
exists, then suck uses the XOVER command to get information
on the articles and decide whether or not to download the article.
Xover files use the same syntax as suckkillfiles, but supports a subset
of the commands.
.PP
The following killfile commands are not supported in suckxover files:
.RS
NRGROUPS:
.RE
.RS
HEADER:
.RE
.RS
BODY:
.RE
.RS
TIEBREAKER_DELETE:
.RE
.RS
PROGRAM:
.RE
.PP
Only the following header lines will be checked:
.RS
Subject:
.RE
.RS
From:
.RE
.RS
Message-ID:
.RE
.RS
References:
.RE
.PP
The behaviour of the size commands (
.I BODYSIZE>, BODYSIZE<, HILINES, and LOWLINES
) specify the total size of the article (not just the body) in
bytes or lines, respectively.
.PP
All other parameters are allowed.  However, if you use an invalid parameter,
it is silently ignored.

.SH SUCKOTHERMSGS
If
.I suckothermsgs
exists, It must contain a one-per-line listing
of Message-ID's, with the <> included.  These articles will then 
be downloaded, in addition to any articles generated by processing
the
.I sucknewsrc
file.  This can be used to get articles in other groups,
or to download an article that was killed.  These articles
.B ARE NOT
processed through the kill articles routines.
.SH FOREIGN LANGUAGE PHRASES
If the
.BI "-l phrases"
option is specified or the file /usr/local/lib/suck.phrases (defined in suck_config.h)
exists, then suck will load an alternate language phrase file, and use
it for all status & error messages, instead of the built-in defaults.  The command line
overrides the build in default, if both are present.
The phrase file contains all messages used by suck, rpost, testhost,
and lmove, each on a separate line and enclosed in quotes.  To generate
a sample phrase file, run
.BI "make phrases"
from the command line.  This will create "phrases.engl", which is a list
of the default phrases.  Simply edit this file, changing the english
phrases to the language of your choosing, being sure to keep the phrases
within the quotes.  These phrases may contain variables to print items
provided by the program, such as hostname.  Variables are designated
by %vN% where N is a one-up sequence per phrase.  These variables may
exist in any order on the phrase line, for example,
.RS
"Hello, %v1%, welcome to %v2%" 	or
.RE
.RS
"Welcome to %v2%, %v1%"	
.RE
are both valid phrases.  Phrases may contain,  \\n, \\r, or \\t to print a newline, carriage return,
or tab, respectively.  Note that the first line of the phrase file is the current version
number.  This is checked against the version of suck running, to be sure that the phrases
file is the correct version.

If you modify any of the source code, and add in new phrases, you will need to regenerate
phrases.h, so that everything works correctly.  To recreate, just run
.BI "make phrases.h"
from the command line.
.SH SIGNAL HANDLING
Suck accepts two signals, defined in
.I suck_config.h.
The first signal (default SIGTERM) will cause Suck to finish downloading the
current article, batch up whatever articles were downloaded, and
exit, without an error.

The second signal (default SIGUSR1) will cause suck to use the pause values defined with
the -w option (see above).

.SH EXIT CODES
Suck will exit with the following return codes:
.RS
0 = success
.RE
.RS
1 = no articles available for download.
.RE
.RS
2 = suck got an unexpected answer to a command it issued to the remote server.
.RE
.RS
3 = the -V option was used.
.RE
.RS
4 = suck was unable to perform NNTP authorization with the remote server.
.RE
.RS
-1 = general error.
.RE
.SH HISTORY
.RS
Original Author - Tim Smith (unknown address)
.RE
.RS
Maintainers - 
.RE
.RS
March 1995 - Sven Goldt (goldt@math.tu-berlin.de)
.RE
.RS
July 1995  - Robert A. Yetman (boby@pixi.com)
.RE
.de R$
Revision \\$$3, \\$$4
..
.SH "SEE ALSO"
testhost(1), rpost(1).