SLURP - An advanced passive NNTP client ======================================= Version 1.08 Written by Stephen Hebditch (C) 1992/3 TQM Communications ------ Slurp is an advanced passive NNTP client for UNIX. It will connect to a remote NNTP server and retrieve articles in a specified set of Usenet newsgroups that have arrived after a particular date (typically the last time it was invoked) for processing by your local news system or forwarding on via UUCP to another news system. It replaces nntpxfer from the NNTP 1.5.11 reference implementation and nntpget from the INN distribution. The Feature List. - No restriction on the number of newsgroups that can be retrieved in a session. The client does a series of NEWNEWS requests if necessary to build up a list of articles which need to be retrieved from the server. - Compatible with C News and INN. - No lost articles due to unsynchronised clocks. By default, the time for the next NEWNEWS is set by a call to the tcp time service on the remote machine. This fixes a problem in nntpxfer and nntpget where articles can be missed if the clock on the server is behind that on the client. - Faster article throughput. If configured to do so, slurp stacks up the next article request before the current article is received, decreasing the total article transfer time by around 30%. Unlike nntpget, article batches can be written to the rnews spool directory for processing with 'rnews -U' later, rather than being offered to innd in turn. When used with C News, batches are written to the in.coming directory. This provides a big speedup over nntpxfer which created a new pipe to rnews for each article, lowering the transfer speed and increasing the load on the client machine. These speed improvements are of most use to the growing number of people retrieving news over a dialup SLIP/PPP connection. - Error recovery. In the event of an error occurring while articles are being fetched, the message IDs of unretrieved articles are dumped to a file. The next session can then retrieve the remaining articles without having to restart from the previous start time. - Easy configuration. slurp.sys defines which groups you wish to get from which server in a similar way to the normal C News sys file, e.g. news.demon.co.uk:demon.*,!demon.msdos.announce,uk.*,\ !uk.net.maps/world,uk slurp. contains the time to be used for the NEWNEWS request at each server. The newsgroups list, start time and distributions can be overridden on the command line. There are also settings to enable debugging output, prevent writing the time for the next NEWNEWS to slurp., issue a 'MODE READER' command to switch innd to nnrpd, not dump unfetched message IDs in the event of an error, and use the local time for the next NEWNEWS rather than the time at the server. - Support for the NNTP simple authorisation protocol. - Logging to syslog of the number of new, missing and duplicate articles at the end of a session plus, if required, the transfer speed of the article fetching stage. - Independent of the NNTP 1.5.11 code. Slurp was written in ANSI C under Dell's SVR4/386, but ports easily to SunOs 4.x, Linux, BSDI BSD/386, HPUX, SCO, ISC, NeXT, Ultrix and most other recent versions of UNIX. CONTENTS -------- In this archive you should have the following files:- HISTORY How we got from there to here. Makefile The Makefile (yes, really!) README What you're reading now articles.c Routines for reading articles and batching them up conf.h Compilation configuration file fakesyslog.c Fake version of syslog for those without fakesyslog.h Definitions needed for the above history.c Routines to open and interrogate the news history file misc.c Miscellaneous routines newnews.c Routines to retrieve new message ids from the server nntp.h Definitions of nntp response codes slurp.1 Slurp manual page slurp.c Main program slurp.h Definitions used by everything slurp.sys Configuration file for groups to take from a server sockets.c Server interface routines space.c Determines space available on news disk time.c Retrieves current time at remote server COMPILATION AND INSTALLATION ---------------------------- 1. Edit conf.h to reflect your reflect your local configuration. This file contains full details about each item that you may need to change. 2. Edit Makefile. If you have not installed dbz.h in /usr/include then you will need to add -Ipathname to CFLAGS so that the compiler does a search of the directory where it can be found. LIBS will need to contain entries to pick up the dbz library and possibly the sockets library depending on your system. If you don't have syslog then you will need to uncomment the FAKESRC and FAKEOBJ lines. 3. Type 'make' and hopefully you will get the slurp executable. 4. Copy slurp to wherever it is that you store your news binaries. 5. Copy slurp.sys to whichever location you selected for it in conf.h. 6. Copy the manual page to wherever you keep your manual pages. QUICK START ----------- The slurp manual page contains full details on how to operate slurp. As a quick start all you need to do is:- 1. Edit slurp.sys to set the hostname of the server you wish to connect to for retrieving articles and the groups you wish to receive. 2. Edit slurp. and enter a single line containing the time a couple of days previously, say, in 'YYMMDD HHMMSS' format. 3. Run 'slurp ' and sit back while the articles flood into your local news system. If you are running C News, then the next time that newsrun is invoked from cron, the articles in the in.coming directory will be added to the news system. If you are running INN and do not have RNEWS defined, then it will be necessary to run 'rnews -U' to have the files from the rnews spool directory processed by the news system. If RNEWS is defined then they will be piped to innd as they arrive - doing this will lower the speed at which articles are transferred from the remote server. Patches for the NNTP 1.5.11 server are available on request which implement caching of article locations during a NEWNEWS so that a further access of the history file is not required when an article is retrieved. I hope to have an equivalent patch available for INN soon. CONDITIONS OF USE ----------------- The Slurp package and individual code as detailed is Copyright 1992/3 TQM Communications, BCM Box 225, London, WC1N 3XX. All rights reserved. This package may be freely distributed provided the following conditions are met. The files detailed in the contents section of this file must all be included. No charge, other than basic online charges or media costs, may be levied. You are permitted to make alterations to the code for your own personal use, but please feed back any bug fixes or enhancements to us so they may be included in any future releases. DISCLAIMER ---------- TQM Communications makes no warranty of any kind in respect to this documentation and the software described. The user assumes any risk as to the quality and performance of this product. In no event will TQM Communications be liable for direct, indirect, incidental or consequential damages arising from any defect in the performance and use of this product. All trademarks acknowledged. ACKNOWLEDGEMENTS ---------------- The original inspiration for slurp came from nntpxfer, written by Brian Kantor. The article batching code is inspired by batch.c in the NNTP 1.5.11 package, written by Stan Barber and 'based on work by Henry Spencer and Geoff Collyer'. The fakesyslog code is taken from the NNTP 1.5.11 package where it is credited to John Robert LoVerso and Paul McKenny. The space code to determine whether there is enough space on the news disk was originally taken from the NNTP 1.5.11 package where it is credited to Stan Barber, Tad Guy, Chris Jepeway and Tom Lane. Regular expression expansion courtesy of Nick Lai. Network knowledge thanks to W. Richard Stevens, author of the most readable UNIX books around. For bug fixing, porting and enhancement ideas, thanks go to Tim Rylance, Pete Bentley, Stefan Schwarz, David Clunie, David Boyce, Barrie Spence and Philip Shearer. CONTACT INFORMATION ------------------- All comments, bug reports, bug fixes, etc. should be sent to steveh@orbital.demon.co.uk. I'm also available for consultancy work if you want help in getting news and mail working. ---- END OF DOCUMENTATION