Source Code 1994 March

home *** CD-ROM | disk | FTP | other *** search

/ Source Code 1994 March / Source_Code_CD-ROM_Walnut_Creek_March_1994.iso / compsrcs / misc / volume38 / procmail / part06 < prev next >

Wrap

Text File | 1993-07-04 | 48.5 KB | 1,235 lines

Newsgroups: comp.sources.misc From: berg@pool.informatik.rwth-aachen.de (Stephen R. van den Berg) Subject: v38i025: procmail - mail processing package v2.90, Part06/11 Message-ID: <1993Jul1.151140.21429@sparky.imd.sterling.com> X-Md4-Signature: 67d7a60a2d5b53ceec60c1634904ee1f Sender: kent@sparky.imd.sterling.com (Kent Landfield) Organization: Sterling Software Date: Thu, 1 Jul 1993 15:11:40 GMT Approved: kent@sparky.imd.sterling.com Submitted-by: berg@pool.informatik.rwth-aachen.de (Stephen R. van den Berg) Posting-number: Volume 38, Issue 25 Archive-name: procmail/part06 Environment: sendmail, smail, MMDF, mailsurr, UNIX, POSIX Supersedes: procmail: Volume 35, Issue 21-32,124,125 #! /bin/sh # This is a shell archive. Remove anything before this line, then unpack # it by saving it into a file and typing "sh file". To overwrite existing # files, type "sh file -c". You can also feed this as standard input via # unshar, or by typing "sh <file", e.g.. If this archive is complete, you # will see the following message at the end: # "End of archive 6 (of 11)." # Contents: procmail/INSTALL procmail/mailinglist/INSTALL # procmail/mailinglist/INTRO procmail/mailinglist/bin/delink # procmail/src/goodies.c procmail/src/lockfile.c # Wrapped by berg@tubastos on Thu Jul 1 14:06:16 1993 PATH=/bin:/usr/bin:/usr/ucb ; export PATH if test -f 'procmail/INSTALL' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/INSTALL'\" else echo shar: Extracting \"'procmail/INSTALL'\" $8379 characters$ sed "s/^X//" >'procmail/INSTALL' <<'END_OF_FILE' XDiscusses: X 1. Getting the lot to compile X 2. IMPORTANT info for people with remote filesystems (like NFS) X 3. DEBUGGING AID X 4. Setting up the environment X 5. Extra options if you are a system administrator X X --- X X1. Getting the lot to compile X -------------------------- X XTo install procmail, lockfile and formail: edit Makefile & config.h accordingly Xand type 'make install'. XIntended configurable options in Makefile are: X the install-destinations and the LOCKINGTEST directories. XIntended configurable options in config.h are: X MMDF support, standard environment presettings, trusted userids. X X'make install' will: X - implicitly do a 'make init', which will check your basic utilities for X POSIX compliance, and generates correcting Makefiles accordingly X - execute autoconf (a shell script that repeatedly calls the C compiler X to determine if certain features/symbols are supported), which will X create a file named autoconf.h X - create three stripped binaries, a shell script and five man pages in X the new/ subdirectory (all that is needed to install): X procmail, lockfile, formail, mailstat, procmail.1, lockfile.1, X formail.1, procmailrc.5, procmailex.5 X - copy these binaries and mailstat to $(BINDIR) X - copy the man pages to $(MAN1DIR) and $(MAN5DIR) X X'make deinstall' will: X - remove the just installed files in $(BINDIR) X - remove the just installed files in $(MAN1DIR) and $(MAN5DIR) X X XMinimal requirements (for regular uses): X Xprocmail must be installed. X XOptional files (depending on your requirements): X Xlockfile only needs to be installed if you plan to read several mailboxes X with one of the standard mailers that don't support lockfiles. Xformail only needs to be installed if mail sometimes arrives in nonstandard X mailbox format (or if you want to generate auto replies, split up X mailboxes/digests etc., see the man page of formail for more info). Xmailstat is an "example" shell script that can be used as is to produce X summaries of the procmail generated logfiles; it is not needed by X procmail itself in any way. X Xprocmail, lockfile, formail and mailstat are all *standalone* programs, i.e. Xthey do *not* use any compiled-in paths or files, they all can be used and Xinstalled independently without the need to install the others. X XIf things don't compile automagically, I suggest you take a look at: Xsrc/autoconf, autoconf.h, config.h, src/includes.h X XFor autoconf to work as intended, your compiler should either be fully ANSI Xcompliant, or you should NOT turn off all warnings; enabling all warnings Xshouldn't hurt. In most cases the default options in the Makefile will do. X XThe sources are supposed to be fully ANSI, K&R and POSIX compliant. X XN.B. If you encounter any difficulty at all in installing procmail (e.g. if you X had to change Makefile or config.h in unpredicted ways, or a simple X "make install" doesn't work), I'd very much like to hear about it; who X knows, next time you might be able to simply "make install" as well. X X --- X X2. IMPORTANT info for people with remote filesystems (like NFS) X ------------------------------------------------------------ X XThe autoconf script tries to determine what kernel locking functions are Xavailable *and* reliable on your system. In order to do this it exercises Xall the available locking methods (fcntl(), lockf() and flock()) on some Xsample files extensively. X XThese tests produce results which depend on the filesystem which is being Xused to perform the tests. If you haven't defined LOCKINGTEST in the Makefile X(you can include as many directories as you want) autoconf will prompt you Xto enter any additional directories (preferably on distinct filesystems) Xyou want it to run the tests on. X XWhen specifying directories to test, it would probably be advisable to Xpick exactly those directories which belong to a unique fileserver-fileclient Xpair. I.e. one local file system will suffice. Only if you have remote Xfilesystems, you might have to specify several. XThis makes sure that the locking tests will properly reflect the (lowest common Xdenominator) locking capabilities on all filesystems available. X XIf you have a very heterogenous environment (like several OS versions Xon machines (perhaps even from different vendors) that have NFS mounted file Xsystems all over each other), then it could happen that some tests which Xare reliable with one remote filesystem, turn out to be unreliable across Xanother (it all depends on the OS versions of clients and servers). X XIf do not want to perform the locking tests on all those filesystems, but Xif you know what locking methods are unreliable anyway, then you can edit Xsome defines in the config.h file (NO_fcntl_LOCK, NO_lockf_LOCK and XNO_flock_LOCK). These can be uncommented by hand to prevent procmail Xfrom using them. X XThe most typical case would be if you happen to be using NFS. Autoconf Xnormally is very well capable of finding out if your lockd is reliable enough. XIn a very heterogenous environment (many different servers, many different Xlockd's (of perhaps different version and patchlevel)) however, it might Xbe hard to determine if all the lockd's are equally reliable. In such a Xcase (typically on SunOS :-), you might consider uncommenting both XNO_fcntl_LOCK and NO_lockf_LOCK (NO_flock_LOCK normally doesn't rely on the Xinfamous lockd, so you can leave it commented out). But, only do so if you Xthink you know it better than autoconf. X X --- X X3. DEBUGGING AID X ------------- X XSince procmail is intended to run completely independent of any terminals, it Xdoesn't use the stderr output to display error messages. It is recommended, Xespecially during debugging, to specify a LOGFILE (see man page) in the Xrcfile or on the command line. Procmail will log all serious problems it Xencounters. Of course, instead of a regular file, one could also specify a Xterminal as the default logfile. X XAlso, procmail can be persuaded to be a lot more verbose by inserting the Xfollowing assignment at the top of your rcfile: X X VERBOSE=on X XTherefore a suggested command line for your first trial run (no rcfiles Xneeded) would be: X X procmail VERBOSE=on X X(now type in a pseudo mail-message) X XIf all else fails, you can try uncommenting the "#define console" entry Xin the config.h file. This will provide you with the most verbose procmail Xyou can make. It is of course a good idea to comment out this line again Xafter your troubles have been solved. X XIf you run procmail by hand and pipe in some sample mail, then make Xsure that if you kill procmail, you use "kill pid" and NOT "kill -9 pid". XShould procmail seem to hang, check if any lockfiles are still present. XIf you kill procmail with "kill pid" it will clean up any lockfiles it created Xitself. X X --- X X4. Setting up the environment X -------------------------- X XEvery user that wants to use procmail should have a .forward and a X.procmailrc file in his HOME directory. For starters, you can look Xat the supplied example files in "examples". X(BTW, be sure to make .forward *world* readable). XMMDF users should note that they need a .maildelivery file *instead* of the X.forward file (see the procmail(1) man page for more information). X X --- X X5. Extra options if you are a system administrator X ----------------------------------------------- X XIf you are a system administrator you can decide to install procmail Xglobally (i.e. as a more robust drop-in replacement for the local- Xmaildelivery-capabilities of /bin/mail), this has the advantage that users do Xnot need to have a .forward file anymore that calls up procmail. Simply Xhaving a .procmailrc file in the HOME directory will suffice. Operation is Xtransparent in this case (i.e. if no .procmailrc file is present in the HOME Xdirectory, mail will be delivered as usual and procmail behaves like a more Xreliable /bin/mail substitute). X XFor direct examples on how to do this, look at the examples/advanced file. X X******************************************************************************* XHIGHLY RECOMMENDED: install "procmail" suid root (and/or sgid maildaemon) X install "lockfile" sgid maildaemon X XTo obtain specific instructions on the best installation, type "make recommend" X******************************************************************************* X X --- X XFor more info about the program, see the man pages or the FAQ list. END_OF_FILE if test 8379 -ne `wc -c <'procmail/INSTALL'`; then echo shar: \"'procmail/INSTALL'\" unpacked with wrong size! fi # end of 'procmail/INSTALL' fi if test -f 'procmail/mailinglist/INSTALL' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/mailinglist/INSTALL'\" else echo shar: Extracting \"'procmail/mailinglist/INSTALL'\" $6110 characters$ sed "s/^X//" >'procmail/mailinglist/INSTALL' <<'END_OF_FILE' XDiscusses: X 1. Prerequisites X 2. Installing the scripts X 3. Sharing the scripts on multiple architectures X 4. If you don't have a sendmail compatible mailer X 5. Upgrading from previous versions of these scripts X X --- X X1. Prerequisites X ------------- X XIn order to make sure that the mailinglist scripts work reliably for both Xremote and local mail, they have to be run under someone's account. XThe recommended procedure would be to create a new account and group named X"list" that is only going to be used to manage the mailinglist stuff. X XIf you are unable/unwilling to create a new user just for the mailinglist Xstuff, then the mailinglist scripts should be installed in a separate Xsubdirectory. This user will still be able to receive regular mail. X XIn the mailinglist scripts I made use of the -f option of sendmail. This Xoption makes sure that the sender address on the envelope (the From_ line) Xcontains the proper list address. In order for this option to work, the Xuser using it (either "list" or the regular user the lists run under) must Xbe on the trusted-user-list of sendmail. To enter the user "list" on the Xtrusted-user-list simply edit the sendmail.cf file and add a line reading: X XTlist X XRestarting sendmail is not necessary. X X --- X X2. Installing the scripts X ---------------------- X XSuppose you have created this pseudo user "list" with a specified home Xdirectory of /home/list X XNow create /home/list by hand and make sure that you chown it to "list". X XNext, make sure that you are running as root (su root) and execute the X"install.sh" script present in this directory by typing something like: X X sh install.sh /home/list X XThis script will then create two subdirectories (.bin and .etc) in X/home/list. These directories will be filled with the files contained Xin the bin and etc subdirectories here. It will also make sure that Xthe "multigram" program is compiled from the procmail*/src directory and Xcopied into /home/list/.bin. X XFurthermore install.sh will link /home/list/.etc/rc.main to X/home/list/.procmailrc. This is of course superfluous (but nondestructive) if Xyou are still using this account to receive regular mail. X XDepending on your mail configuration, if procmail has not been integrated Xin the maildelivery system (see procmail*/examples/advanced for more Xinformation on that topic) you also have to create a .forward file with an Xappropriate content (see "man procmail", the NOTES section at the end). This, Xhowever, is only necessary if you created a seperate "list" account. X X Next, edit the /home/list/.etc/rc.init file. Make sure that X "domain" is set to the right value, and you can optionally specify a X "listmaster"; also check if the PATH is correct and make sure X that the procmail and formail are in the path. X XFor further instructions, you should read the "Manual" file, it has Xbeen copied to the /home/list/.etc directory as well, and can serve as a Xquick reference. X X --- X X3. Sharing the scripts on multiple architectures X --------------------------------------------- X XFor people that want to run the lists on a shared filesystem across different Xarchitectures there exists the possibility of using several .bin directories. XSimply use something like: X sh install.sh /home/list .bin.sun4 Xto install to a different .bin directory. Repeat the complete procmail and Xmailinglist installation for every architecture, specifying a different .bin Xdirectory every time. X X --- X X4. If you don't have a sendmail compatible mailer X ---------------------------------------------- X XAs is, these scripts are preconfigured to use sendmail (or a compatible Xmailer like smail) to actually transport the mails. If your system only Xsupplies a /bin/rmail or /bin/mail to transport mails, then several of Xthe options used when invoking sendmail will not work. X XFor these cases you have to edit the /home/list/.etc/rc.init file. XLook for a SENDMAIL assignment. Supplied with these scripts is a poor man's Xsendmail. It was installed in the .bin directory belonging to the Xlist. Make sure that the SENDMAIL assignment points at the sendmails script. XAnd, of course, don't forget to uncomment the SENDMAIL assignment and Xthe two sendmailOPT* assigments following it. X X --- X X5. Upgrading from previous versions of these scripts X ------------------------------------------------- X XBefore you start the upgrade, create the file /home/list/.etc/rc.lock XNow do a ps and see if any procmail-list instances are currently Xrunning. Wait till they are gone. Now touch the rc.lock file once more. XYou know have 17 minutes to complete the upgrade. If it should take Xlonger, touch the rc.lock file every so often. As soon as the difference Xbetween its mtime and the current time exceeds 17 minutes, the flist Xprograms start delivering mail to the lists again. X XYou can simply use the install.sh script again to install directly over Xthe old installation IF you customised ONLY the master rc.init and rc.custom Xfiles. Once you ran the install.sh script, you'll have to merge the changes Xyou made to the old rc.init file (still there) into the rc.init.new file. XThen cat (yes, use `cat', because you have to preserve the hardlinks) the Xrc.init.new file into the rc.init file. XThere are some new entries in the new rc.custom file template. Since they Xare commented out (i.e. rc.init already provides defaults) their omission Xin the old rc.custom files should not cause any problems. X XIf you customised more than just the rc.init or rc.custom files, you'll Xhave to make diffs between the old versions of the scripts, then install Xthe new scripts and then apply back the diffs (probably by hand). XThe reason why parts of the old scripts cannot operate with the new scripts Xis because they expect several new environment variables to be set (like Xdigest selection or not). The new rc.init sets them of course. X XYou can quickly verify which files are linked to other files by Xsimply typing something like: X X showlink rc.init X XAfter you are finished, remove the rc.lock file again. If it is older Xthan 17 minutes it is ignored anyway, but it's cleaner this way. END_OF_FILE if test 6110 -ne `wc -c <'procmail/mailinglist/INSTALL'`; then echo shar: \"'procmail/mailinglist/INSTALL'\" unpacked with wrong size! fi # end of 'procmail/mailinglist/INSTALL' fi if test -f 'procmail/mailinglist/INTRO' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/mailinglist/INTRO'\" else echo shar: Extracting \"'procmail/mailinglist/INTRO'\" $9350 characters$ sed "s/^X//" >'procmail/mailinglist/INTRO' <<'END_OF_FILE' X$Id: INTRO,v 1.5 1993/01/28 14:21:56 berg Exp $ X X How to set up mailing lists X --------------------------- X X Written by Stephen R. van den Berg. X berg@pool.informatik.rwth-aachen.de X berg@physik.tu-muenchen.de X XThis document mainly describes a sendmail environment, much of it applies Xto non-sendmail mail agents as well. X X XContents: X--------- 1. Intro X 2. Bouncing mail X 3. The disadvantages X 4. How to circumvent these disadvantages X 5. Why use procmail to filter the mailinglist mail? X 6. How do I use procmail to filter the mailinglist mail? X X1. Intro X ----- X XThe simplest and most direct way to setup a mailinglist is by inserting a line Xin the /usr/lib/aliases file looking like: X Xmylist: fred,john, wilma, barney@bedrock, pebbles X XNow all the mail arriving at your machine for "mylist" (either local or Xmylist@your.domain) will be automatically forwarded to all the mentioned Xaddresses (fred, john, etc.). X XThe address mylist@your.domain is intended for submissions to the list that Xare supposed to be forwarded to all the subscribers. For the administrative Xtasks like removals from the list, new subscriptions to the list, or address Xchanges of subscribers, it is common practice to create a second entry in the X/usr/lib/aliases file: X Xmylist-request: your_login_name@your.domain X X X2. Bouncing mail X ------------- X XIn order to deal with bouncing mail gracefully, an extra precaution should Xbe taken. If for example mail to wilma bounces (user non-existent, mail Xfilesystem full, etc.), it will bounce back to the original sender. XNow, the only person that should be concerned with distribution failures Xshould be the mylist-request holder. Therefore you should be using a Xsendmail special alias like: X Xowner-mylist: mylist-request@your.domain X XThis way local mail will bounce back to mylist-request@your.domain. X X X3. The disadvantages X ----------------- X XIf you are using the above methods, some obvious disadvantages come to mind Xhowever: X Xa. The subscriber list cannot exceed 1000 bytes (on many sendmails). Xb. The subscriber list cannot be changed on-the-fly (/usr/lib/aliases needs X to be edited, and newaliases has to be run). Xc. People cannot be prevented from submitting messages like "Please remove X me from this mailinglist" to mylist (and thereby annoying all subscribers). Xd. People cannot be guarded from themselves in case they insert X "Return-Receipt-To:" fields in their headers (if they are particularly X unlucky, they will receive an acknowledge mail from *every* subscriber's X sendmail). Xe. People including "Errors-To:" or "Sender:" fields can cause the bounce X messages to bypass owner-mylist anyway. Xf. There is no way of limiting the number of submitters, i.e. every person X who knows the name of the mailing list and who can send mail to your.domain X is able to submit messages to the list. This means, for example, that you X cannot limit a mailing list to local users (i.e. only local users can X submit). Xg. You are unable to insert a "Reply-To: mylist@your.domain" in case you X would want to (this makes replying to the list easier, too easy as some X people say). X X X4. How to circumvent these disadvantages X ------------------------------------- X Xa. Can be circumvented by using nested aliases like: X mylist: mylist1, mylist2 X mylist1: fred,john X mylist2: wilma,barney@bedrock,pebbles X This can however, become extremely messy to maintain. X Xb. This can be avoided if you use aliases like: X mylist: :include:/path/to/the/memberfile X The memberfile should contain: X fred,john,wilma,barney@bedrock,pebbles X This will also take care of the upper limit on aliases expansions and X newaliases need not be run again every time you change the file. X Xc. Can only be taken care of by using a mailfilter like procmail. X Xd. Can only be taken care of by using a mailfilter like procmail. X Xe. Can only be taken care of by using a mailfilter like procmail. X Xf. Can only be taken care of by using a mailfilter like procmail. X Xg. Can only be taken care of by using a mailfilter like procmail. X X X5. Why use procmail to filter the mailinglist mail? X ------------------------------------------------ X XInstead of using a mailfilter you could also take care of most of the problems Xthree till seven by editing the sendmail.cf file. I would strongly recommend Xagainst this approach however, since this will be too much of a customising Xoperation and surely will not be a trivial task (in all cases). As a general Xrule: don't mess with a sendmail.cf file once it works :-). X XNow, you could, instead of procmail, simply use immediate VNIX commands Xlike grep, sed or awk to do the mail filtering. Again, there are some obvious Xdisadvantages with this approach: X XA. In case any system resources go out (no more file descriptors, no more X swap space, process table full, file system full (for temporary files)) X your awk or shell script will fail generously (i.e. several bad things X could happen: mail munged, truncated, lost, hanging awk or sh programs, X etc., you get the picture). X XB. All mail headers (including From: and Reply-To:) could very well be X multi-line headers; it will be very difficult to make it understandable X to awk that somehow the header line could continue on the next line X (in case you want to remove a header, or do some complicated substitution). X XC. Another hairy problem will be determining the end of the header, of course X that is solvable, but you have to make some extra precautions in your X awk script to ensure that any substitutions/changes will not occur in X the body of the message (further degrading performance and increasing the X load on your machine). X XD. Starting programs directly from within aliases or .forward files can get X extremely messy, since the environment the program starts in is X potentially hostile. X XProcmail does not *directly* allow you to change any headers, but that Xfeature is not really necessary since you can tell procmail to send ONLY the Xheader through some filter of your choice. X XTo comment on the previously mentioned three disadvantages: X XA. Procmail takes care of that. Should the filter have problems anyway, X procmail will graciously notice that the filter was in some kind of X trouble, and will try something else with the original unmunged mail X (you can specify what it should do of course, obvious choices: try X the same filter again, drop the mail in a file and send you a notice, X forward the mail to you instead (unfiltered), etc.) X XB. In order to make consistent scanning of the header possible using the X egrep regular expressions built in to procmail, procmail will internally X concatenate any headers that were continued according to the RCF 822 X recommendations, in order for external filters to benefit from this, you X would use the formail program to pre-filter the mail. X XC. Procmail can be told to send the header, the body or both through the X filter, hence your filter need not watch out to avoid doing any X substitutions in the body, and the filter can therefore be a lot simpler. X XD. Procmail makes no assumptions about the environment it is started in, it X assumes that everything is hostile and fights its way back to the X civilised world by initialising *everything* to sane and expected default X values. Thereby providing a warm bed for any program started from within X procmail. X XBut procmail has some additional advantages as well: X X -- It will probably all go a bit faster, since only the header of the mail X is being piped through the filter. Also, procmail reads in the mail in X 16KB chunks, not line by line as sed does. X X -- You could use procmail to filter out any messages to the normal mailing X list that should have gone to the mylist-request and remail them to X mylist-request. X XWell, anyway, as you see, procmail does not give you everything you would want, Xbut this was intentional in accordance to the true VNIX spirit (modularity). XWhat procmail does provide is a *very* reliable hook (you might say it Xprovides an anchor :-) for any mail processing you might do. For the more Xcomplex things you still have to use shell scripts or call other programs Xfrom within procmail, but then again, that saves you from learning any Xparticular syntax procmail would have had to do the same. X XAs it happens, the accompanying formail program is able to cater to most X(if not all) of your needs regarding mail munging. X X X6. How do I use procmail to filter the mailinglist mail? X ----------------------------------------------------- X XIn order to cater for most wishes regarding mailinglist setup, I took the Xliberty to write some rcfiles for procmail that can readily be used to create Xany number of mailinglists in a comfortable and simple way. X XSee the FEATURES file in this directory for more information on what Xthese mailinglist scripts will do for you. X XIf the scripts do not exactly do what you would have liked, you are invited to Xedit them to taste. Perhaps you could send your changes to the procmail Xmailinglist if you feel that they could be useful to others. X XTo get started I suggest you read the INSTALL file in this directory. X XFor operating instructions you should read the Manual file in this directory. X X XP.S. Any suggestions/corrections/improvements on this document are welcome. END_OF_FILE if test 9350 -ne `wc -c <'procmail/mailinglist/INTRO'`; then echo shar: \"'procmail/mailinglist/INTRO'\" unpacked with wrong size! fi # end of 'procmail/mailinglist/INTRO' fi if test -f 'procmail/mailinglist/bin/delink' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/mailinglist/bin/delink'\" else echo shar: Extracting \"'procmail/mailinglist/bin/delink'\" $756 characters$ sed "s/^X//" >'procmail/mailinglist/bin/delink' <<'END_OF_FILE' X#! /bin/sh X: &&O= || exec /bin/sh $0 $argv:q # we're in a csh, feed myself to sh X######################################################################### X# delink Gracefully disconnects a hardlinked file # X# # X# Created by S.R. van den Berg, The Netherlands # X######################################################################### X#$Id: delink,v 1.4 1993/06/02 15:30:42 berg Exp $ X Xecho=echo # /bin/echo Xtest=test # /bin/test Xrm=rm # /bin/rm Xmv=mv # /bin/mv Xcp=cp # /bin/cp Xdirname=dirname # /bin/dirname X X$test 0 = $# && $echo "Usage: delink filename ..." 1>&2 && exit 64 X XTMPF=delink.$$ X Xtrap "$rm -f \$TMPF; exit 1" 1 2 3 15 X Xfor a in "$@" Xdo X TMPF="`dirname \"$a\"`"/delink.$$ X $cp -p "$a" "$TMPF" && $mv -f "$TMPF" "$a" Xdone END_OF_FILE if test 756 -ne `wc -c <'procmail/mailinglist/bin/delink'`; then echo shar: \"'procmail/mailinglist/bin/delink'\" unpacked with wrong size! fi chmod +x 'procmail/mailinglist/bin/delink' # end of 'procmail/mailinglist/bin/delink' fi if test -f 'procmail/src/goodies.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/src/goodies.c'\" else echo shar: Extracting \"'procmail/src/goodies.c'\" $9686 characters$ sed "s/^X//" >'procmail/src/goodies.c' <<'END_OF_FILE' X/************************************************************************ X * Collection of library-worthy routines * X * * X * Copyright (c) 1990-1992, S.R. van den Berg, The Netherlands * X * #include "README" * X ************************************************************************/ X#ifdef RCS Xstatic /*const*/char rcsid[]= X "$Id: goodies.c,v 1.15 1993/06/21 14:24:17 berg Exp $"; X#endif X#include "procmail.h" X#include "sublib.h" X#include "robust.h" X#include "shell.h" X#include "misc.h" X#include "pipes.h" X#include "common.h" X#include "cstdio.h" X#include "goodies.h" X Xlong Stdfilled; X#ifndef GOT_bin_test Xconst char test[]="test"; X#endif Xconst char*Tmnate,*All_args; X X#define NOTHING_YET (-1) /* readparse understands a very complete */ X#define SKIPPING_SPACE 0 /* subset of the standard /bin/sh syntax */ X#define NORMAL_TEXT 1 /* that includes single-, double- and back- */ X#define DOUBLE_QUOTED 2 /* quotes, backslashes and $subtitutions */ X#define SINGLE_QUOTED 3 X X#define fgetc() (*fpgetc)() /* some compilers previously choked on it */ X X/* sarg==0 : normal parsing, split up arguments like in /bin/sh X * sarg==1 : environment assignment parsing, parse up till first whitespace X * sarg==2 : normal parsing, split up arguments by single spaces X */ Xvoid readparse(p,fpgetc,sarg)register char*p;int(*const fpgetc)(); X const int sarg; X{ static i;int got;char*startb; X All_args=0; X for(got=NOTHING_YET;;) /* buf2 is used as scratch space */ Xloop: X { i=fgetc(); X if(buf+linebuf-3<p) /* doesn't catch everything, just a hint */ X { nlog("Exceeded LINEBUF\n");p=buf+linebuf-3;goto ready; X } Xnewchar: X switch(i) X { case EOF: /* check sarg too to prevent warnings in the recipe- */ X if(sarg!=2&&got>NORMAL_TEXT) /* condition expansion code */ Xearly_eof: nlog(unexpeof); Xready: if(got!=SKIPPING_SPACE||sarg) /* not terminated yet or sarg==2 ? */ X *p++='\0'; X Tmnate=p;return; X case '\\': X if(got==SINGLE_QUOTED) X break; X switch(i=fgetc()) X { case EOF:goto early_eof; /* can't quote EOF */ X case '\n':continue; /* concatenate lines */ X case '#': X if(got>SKIPPING_SPACE) /* escaped comment at start of word? */ X goto noesc; /* apparently not, literally */ X case ' ':case '\t':case '\'': X if(got==DOUBLE_QUOTED) X goto noesc; X case '"':case '\\':case '$':case '`':goto nodelim; X } X if(got>NORMAL_TEXT) Xnoesc: *p++='\\'; /* nothing to escape, just echo both */ X break; X case '`': X if(got==SINGLE_QUOTED) X goto nodelim; X for(startb=p;;) /* mark your position */ X { switch(i=fgetc()) /* copy till next backquote */ X { case '\\': X switch(i=fgetc()) X { case EOF:nlog(unexpeof);goto forcebquote; X case '\n':continue; X case '"': X if(got!=DOUBLE_QUOTED) X break; X case '\\':case '$':case '`':goto escaped; X } X *p++='\\';break; X case '"': X if(got!=DOUBLE_QUOTED) /* missing closing backquote? */ X break; Xforcebquote: case EOF:case '`': X { int osh=sh; X *p='\0'; X if(!(sh=!!strpbrk(startb,tgetenv(shellmetas)))) X { const char*save=sgetcp,*sall_args=All_args; X sgetcp=p=tstrdup(startb);readparse(startb,sgetc,0); X if(!All_args) /* only one can be remembered */ X All_args=sall_args; /* this is a bug */ X#ifndef GOT_bin_test X if(!strcmp(test,startb)) X strcpy(startb,p),sh=1; /* oops, `test' found */ X#endif X free(p);sgetcp=save; /* chopped up */ X } /* drop source buffer, read from program */ X startb= X fromprog(p=startb,startb,(size_t)(buf-startb+linebuf-3)); X sh=osh; /* restore sh */ X if(got!=DOUBLE_QUOTED) X { i=0;startb=p;goto simplsplit; /* split it up */ X } X if(i=='"'||got<=SKIPPING_SPACE) /* missing closing ` ? */ X got=NORMAL_TEXT; /* or sarg!=0 ? */ X p=startb;goto loop; X } X case '\n':i=';'; /* newlines separate commands */ X } Xescaped: *p++=i; X } X case '"': X switch(got) X { case DOUBLE_QUOTED:got=NORMAL_TEXT;continue; /* closing " */ X case SINGLE_QUOTED:goto nodelim; X } X got=DOUBLE_QUOTED;continue; /* opening " */ X case '\'': X switch(got) X { case DOUBLE_QUOTED:goto nodelim; X case SINGLE_QUOTED:got=NORMAL_TEXT;continue;} /* closing ' */ X got=SINGLE_QUOTED;continue; /* opening ' */ X case '#': X if(got>SKIPPING_SPACE) /* comment at start of word? */ X break; X while((i=fgetc())!=EOF&&i!='\n'); /* skip till EOL */ X goto ready; X case '$': X if(got==SINGLE_QUOTED) X break; X startb=buf2; X switch(i=fgetc()) X { case EOF:*p++='$';goto ready; X case '@': X if(got!=DOUBLE_QUOTED) X goto normchar; X All_args=p;continue; X case '{': /* ${name} */ X while(EOF!=(i=fgetc())&&alphanum(i)) X *startb++=i; X *startb='\0'; X if(i!='}'||numeric(*buf2)&&buf2[1]) X { nlog("Bad substitution of");logqnl(buf2);continue; X } X i='\0';break; /* $$ =pid */ X case '$':ultstr(0,(unsigned long)thepid,p);goto ieofstr; X case '?':strcpy(p,"-1"); X if(lexitcode>=0) /* $? =exitcode from last started program */ X ultstr(0,(unsigned long)lexitcode,p); X goto ieofstr; /* $# =number of extra command-line arguments */ X case '#':ultstr(0,(unsigned long)crestarg,p);goto ieofstr; X case '-':strcpy(p,tgetenv(lastfolder)); Xieofstr: i='\0';goto eofstr; /* $- =lastfolder */ X default: X if(numeric(i)) /* $n positional argument */ X { *startb++=i;i='\0';goto finsb; X } X if(alphanum(i)) /* $name */ X { do *startb++=i; X while(EOF!=(i=fgetc())&&alphanum(i)); X if(i==EOF) X i='\0'; Xfinsb: *startb='\0';break; X } Xnormchar: *p++='$';goto newchar; /* not a substitution */ X } X ;{ int j; X startb=(unsigned)(j=(*buf2)-'0')>9?(char*)tgetenv(buf2): X !j?(char*)argv0:j<=crestarg?(char*)restargv[j-1]:""; X } X if(got!=DOUBLE_QUOTED) Xsimplsplit: for(;;startb++) /* simply split it up in arguments */ X { switch(*startb) X { case ' ':case '\t':case '\n': X if(got<=SKIPPING_SPACE) X continue; X *p++=sarg?' ':'\0';got=SKIPPING_SPACE;continue; X case '\0':goto eeofstr; X } X *p++= *startb;got=NORMAL_TEXT; X } X else X { strcpy(p,startb); /* simply copy it */ Xeofstr: if(got<=SKIPPING_SPACE) /* can only occur if sarg!=0 */ X got=NORMAL_TEXT; X p=strchr(p,'\0'); X } Xeeofstr: if(i) /* already read next character? */ X goto newchar; X continue; X case ' ':case '\t': X switch(got) X { case NORMAL_TEXT: X if(sarg==1) X goto ready; /* already fetched a single argument */ X got=SKIPPING_SPACE;*p++=sarg?' ':'\0'; /* space or \0 sep. */ X case NOTHING_YET:case SKIPPING_SPACE:continue; /* skip space */ X } X case '\n': X if(got<=NORMAL_TEXT) X goto ready; /* EOL means we're ready */ X } Xnodelim: X *p++=i; /* ah, a normal character */ X if(got<=SKIPPING_SPACE) /* should we bother to change mode? */ X got=NORMAL_TEXT; X } X} X Xwaitfor(pid)const pid_t pid; /* wait for a specific process */ X{ int i;pid_t j; X while(pid!=(j=wait(&i))||WIFSTOPPED(i)) X if(-1==j) X return -1; X return lexitcode=WIFEXITED(i)?WEXITSTATUS(i):-1; X} X Xstatic struct lienv{struct lienv*enext;char ename[255];}*myenv; Xstatic char**lastenv; X /* smart putenv, the way it was supposed to be */ Xvoid sputenv(a)const char*const a; X{ static alloced;int i,remove;char*split,**preenv;struct lienv*curr,**last; X yell("Assigning",a);remove=0; X if(!(split=strchr(a,'='))) /* assignment or removal? */ X remove=1,split=strchr(a,'\0'); X i=split-a; X for(curr= *(last= &myenv);curr;curr= *(last= &curr->enext)) /* is it one */ X if(!strncmp(a,curr->ename,i)&&curr->ename[i]=='=') /* I made earlier? */ X { split=curr->ename;*last=curr->enext;free(curr); X for(preenv=environ;*preenv!=split;preenv++); X goto wipenv; X } X for(preenv=environ;*preenv;preenv++) /* is it in the standard */ X if(!strncmp(a,*preenv,i)&&(*preenv)[i]=='=') /* environment? */ Xwipenv: X { while(*preenv=preenv[1]) /* wipe this entry out of the environment */ X preenv++; X break; X } X i=(preenv-environ+2)*sizeof*environ; X if(alloced) /* have we ever alloced the environ array before? */ X environ=realloc(environ,i); X else X alloced=1,environ=tmemmove(malloc(i),environ,i-sizeof*environ); X if(!remove) /* if not remove, then add it to both environments */ X { for(preenv=environ;*preenv;preenv++); X curr=malloc(ioffsetof(struct lienv,ename[0])+(i=strlen(a)+1)); X tmemmove(*(lastenv=preenv)=curr->ename,a,i);preenv[1]=0;curr->enext=myenv; X myenv=curr; X } X} X /* between calling primeStdout() and retStdout() */ Xvoid primeStdout P((void)) /* *no* environment changes are allowed! */ X{ char*p; X if((p=strchr(buf,'\0'))[-1]!='=') /* does it end in an '='? */ X *p='=',p[1]='\0'; /* make sure it does */ X sputenv(buf);Stdout=(char*)myenv; X Stdfilled=ioffsetof(struct lienv,ename[0])+strlen(myenv->ename); X} X Xvoid retStdout(newmyenv)char*const newmyenv; /* see note on primeStdout() */ X{ if(newmyenv[Stdfilled-1]=='\n') /* strip one trailing newline */ X Stdfilled--; X newmyenv[Stdfilled]='\0';*lastenv=(myenv=(struct lienv*)newmyenv)->ename; X Stdout=0; X} X Xvoid postStdout P((void)) /* throw it into the keyword parser */ X{ const char*p;size_t i; X p= *lastenv;tmemmove(buf,p,i=strchr(p,'=')-p);buf[i]='\0';asenv(p+i+1); X} END_OF_FILE if test 9686 -ne `wc -c <'procmail/src/goodies.c'`; then echo shar: \"'procmail/src/goodies.c'\" unpacked with wrong size! fi # end of 'procmail/src/goodies.c' fi if test -f 'procmail/src/lockfile.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'procmail/src/lockfile.c'\" else echo shar: Extracting \"'procmail/src/lockfile.c'\" $9418 characters$ sed "s/^X//" >'procmail/src/lockfile.c' <<'END_OF_FILE' X/************************************************************************ X * lockfile - The conditional semaphore-file creator * X * * X * It has been designed to be able to be run sgid mail or * X * any gid you see fit (in case your mail spool area is *not* * X * world writeable, but group writeable), without creating * X * security holes. * X * * X * Seems to be relatively bug free. * X * * X * Copyright (c) 1990-1992, S.R. van den Berg, The Netherlands * X * #include "README" * X ************************************************************************/ X#ifdef RCS Xstatic /*const*/char rcsid[]= X "$Id: lockfile.c,v 1.15 1993/06/21 14:24:28 berg Exp $"; X#endif Xstatic /*const*/char rcsdate[]="$Date: 1993/06/21 14:24:28 $"; X#include "includes.h" X#include "sublib.h" X#include "exopen.h" X X#ifndef SYSTEM_MBOX X#define SYSTEM_MBOX SYSTEM_MAILBOX X#endif X Xstatic volatile exitflag; Xpid_t thepid; Xstatic char systm_mbox[]=SYSTEM_MBOX; Xstatic const char dirsep[]=DIRSEP,lockext[]=DEFlockext, X nameprefix[]="lockfile: ",lgname[]="LOGNAME",home[]="HOME"; X Xstatic void failure P((void)) /* signal trap */ X{ exitflag=2; /* merely sets a flag */ X} X /* see locking.c for comment on xcreat() */ Xstatic xcreat(name,tim)const char*const name;time_t*const tim; X{ char*p,*q;int j= -1,i;struct stat stbuf; X for(q=(char*)name;p=strpbrk(q,dirsep);q=p+1); X i=q-name; X if(!(p=malloc(i+UNIQnamelen))) X return exitflag=1; X strncpy(p,name,i); X if(unique(p,p+i,0,0)) X stat(p,&stbuf),*tim=stbuf.st_mtime,j=myrename(p,name); X free(p);return j; X} X Xvoid elog(a)const char*const a; X{ write(STDERR,a,strlen(a)); X} X Xvoid nlog(a)const char*const a; X{ elog(nameprefix);elog(a); /* decent error messages should start with this */ X} X Xstatic size_t parsecopy(dest,org,pass)char*const dest;const char*org; X const struct passwd*const pass; /* try and digest the mailbox-lockfile name */ X{ size_t len;const char*chp;char*p;unsigned i; X for(p=dest,len=STRLEN(lockext)+1;;) X { switch(*org) X { case '$': /* we substitute */ X if(!strncmp(++org,lgname,STRLEN(lgname))) X org+=STRLEN(lgname),chp=pass->pw_name; /* $LOGNAME and */ X else if(!strncmp(org,home,STRLEN(home))) X org+=STRLEN(home),chp=pass->pw_dir; /* $HOME */ X else X goto capac; /* no other fancy stuff */ X if((i= *org)-'a'<='z'-'a'||i-'A'<='Z'-'A'||numeric(i)||i=='_') X goto capac; X if(p) X p+=strlen(strcpy(p,chp)); /* paste it in */ X len+=strlen(chp);continue; X default: X if(p) X *p++= *org; /* simply copy everything else */ X len++;org++;continue; /* except suspicous looking characters */ X case '\'':case '`':case '"':case '\\':case '{':goto capac; X case '\0':; X } X if(p) X { if(p==dest||!strchr(dirsep,*dest)) /* absolute path wanted */ Xcapac: { nlog("Sorry, but turning this mess into a useable mailbox \ Xexceeds my humble\ncapacities");return 0; X } X strcpy(p,lockext); X } X return len; X } X} X Xstatic PROGID; X Xmain(argc,argv)const char*const argv[]; X{ const char*const*p,*const*lastf;char*cp;uid_t uid; X int sleepsec,retries,invert,force,suspend,retval=EX_OK,virgin=1; X static const char usage[]="Usage: lockfile -nnn | -r nnn | -l nnn | -s nnn \ X| -! | -ml | -mu | file ...\n"; X if(argc<=1) /* sanity check, any argument at all? */ X goto usg; X sleepsec=DEFlocksleep;force=invert=(char*)progid-(char*)progid;retries= -1; X suspend=DEFsuspend;thepid=getpid();uid=getuid();signal(SIGPIPE,SIG_IGN); Xagain: X signal(SIGHUP,(void(*)())failure);signal(SIGINT,(void(*)())failure); X signal(SIGQUIT,(void(*)())failure);signal(SIGTERM,(void(*)())failure); X for(lastf=p=argv;--argc;) X if(*(cp=(char*)*++p)=='-') X for(cp++;;) X { char*cp2=cp;int i; X switch(*cp++) X { case '!':invert^=1;continue; /* invert the exitcode */ X case 'r':case 'l':case 's': X if(!*cp&&(cp=(char*)*++p,!--argc)) /* concatenated/seperate */ X goto eusg; X i=strtol(cp,&cp,10); X switch(*cp2) X { case 'r':retries=i;goto checkrdec; X case 'l':force=i;goto checkrdec; X default: X if(i<0) /* suspend should be >=0 */ X goto eusg; X suspend=i;goto checkrdec; X } X case HELPOPT1:case HELPOPT2:elog(usage); X elog( X "\t-nnn\twait nnn seconds between locking attempts\ X\n\t-r nnn\tmake at most nnn retries before giving up on a lock\ X\n\t-l nnn\tset locktimeout to nnn seconds\ X\n\t-s nnn\tsuspend nnn seconds after a locktimeout occurred\ X\n\t-!\tinvert the exit code of lockfile\ X\n\t-ml\tlock your system mail-spool file\ X\n\t-mu\tunlock your system mail-spool file\n");goto xusg; X default: X if(sleepsec>=0) /* is this still the first pass? */ X { if((sleepsec=strtol(cp2,&cp,10))<0) X goto eusg; Xcheckrdec: if(cp2==cp) Xeusg: { elog(usage); /* regular usage message */ Xxusg: retval=EX_USAGE;goto nfailure; X } X } X else /* no second pass, so leave sleepsec<0 */ X strtol(cp2,&cp,10); /* and discard the number */ X continue; X case 'm': /* take $LOGNAME as a hint, check if valid */ X { struct passwd*pass;static char*ma;size_t alen; X if(*cp&&cp[1]||ma&&sleepsec>=0) /* second pass? */ X goto eusg; X if(!ma) /* ma initialised last time? */ X { if(!((ma=(char*)getenv(lgname))&&(pass=getpwnam(ma))&& X pass->pw_uid==uid||(pass=getpwuid(uid)))) X { nlog("Can't determine your mailbox, who are you?\n"); X goto nfailure; /* panic, you're not in /etc/passwd */ X } X if(!(alen=parsecopy((char*)0,systm_mbox,pass))) X { cp=systm_mbox;goto lfailure; /* couldn't digest */ X } /* mailbox */ X if(!(ma=malloc(alen))) /* ok, make some room */ X goto outofmem; X parsecopy(ma,systm_mbox,pass); /* and fill her up */ X } X switch(*cp) X { default:goto eusg; /* somebody goofed again */ X case 'l': /* lock the mailbox */ X if(sleepsec>=0) /* first pass? */ X { cp=ma;goto stilv; /* yes, lock it! */ X } X case 'u': /* unlock the mailbox */ X if(unlink(ma)) X { nlog("Can't unlock \"");elog(ma);elog("\""); X if(*cp=='l') /* they messed up, give them a hint */ X elog(" again,\n already dropped my privileges"); X elog("\n"); X } X else X virgin=0; X } X } X case '\0':; X } X break; X } X else if(sleepsec<0) /* second pass, release everything we acquired */ X unlink(cp); X else X { time_t t;int permanent; X setgid(getgid()); /* just to be on the safe side */ Xstilv: virgin=0;permanent=nfsTRY; X while(0>xcreat(cp,&t)) /* try and lock */ X { struct stat stbuf; X if(exitflag) /* time to stop? */ X { if(exitflag==1) /* was it failure() or malloc() */ Xoutofmem: retval=EX_OSERR,nlog("Out of memory"); X else X retval=EX_TEMPFAIL,nlog("Signal received"); X goto lfailure; X } X switch(errno) /* why did the lock not succeed? */ X { case EEXIST: /* hmmm..., by force then? */ X if(force&&!lstat(cp,&stbuf)&&force<t-stbuf.st_mtime) X { nlog(unlink(cp)?"Forced unlock denied on \"": X "Forcing lock on \""); X elog(cp);elog("\"\n");sleep(suspend); /* important */ X } X else /* no forcing now */ X case ENOSPC: X#ifdef EDQUOT X case EDQUOT: X#endif X switch(retries) /* await your turn like everyone else */ X { case 0:nlog("Sorry");retval=EX_CANTCREAT; X goto lfailure; /* patience exhausted, give up */ X default:retries--; /* count sheep */ X case -1:sleep(sleepsec); /* wait and see */ X } X break; X case ENOENT:case ENOTDIR:case EIO:case EACCES: X if(!--permanent) /* NFS sporadically generates these */ X { sleep(sleepsec);continue; /* unwarranted */ X } /* so ignore them first */ X default: /* but, it seems to persist, so give up */ X nlog("Try praying");retval=EX_UNAVAILABLE; X#ifdef ENAMETOOLONG X goto lfailure; X case ENAMETOOLONG: X if(0<(permanent=strlen(cp)-1)&& /* can we truncate it? */ X !strchr(dirsep,cp[permanent-1])) X { nlog("Truncating \"");elog(cp); /* then try it */ X elog("\" and retrying lock\n");cp[permanent]='\0';break; X } /* otherwise, forget it */ X nlog("Filename too long");retval=EX_UNAVAILABLE; X#endif Xlfailure: elog(", giving up on \"");elog(cp);elog("\"\n"); Xnfailure: sleepsec= -1;argc=lastf-argv+1;goto again; /* mark sleepsec */ X } /* for second pass, and adjust argc to the no. of args parsed */ X permanent=nfsTRY; /* refresh the NFS-error-ignore count */ X } X lastf=p; /* last valid file */ X } X if(retval==EX_OK&&virgin) /* any errors? did we do anything? */ Xusg: X { elog(usage);return EX_USAGE; X } X if(invert) X switch(retval) /* we only invert the regular cases */ X { case EX_OK:return EX_CANTCREAT; X case EX_CANTCREAT:return EX_OK; X } X return retval; /* all other exitcodes remain */ X} X Xvoid*tmalloc(len)const size_t len; /* stub */ X{ return malloc(len); X} X Xropen(name,mode,mask)const char*const name;const int mode;const mode_t mask; X{ return open(name,mode,mask); /* stub */ X} X Xrclose(fd)const int fd; /* stub */ X{ return close(fd); X} X Xvoid writeerr(a)const char*const a; /* stub */ X{ X} END_OF_FILE if test 9418 -ne `wc -c <'procmail/src/lockfile.c'`; then echo shar: \"'procmail/src/lockfile.c'\" unpacked with wrong size! fi # end of 'procmail/src/lockfile.c' fi echo shar: End of archive 6 $of 11$. cp /dev/null ark6isdone MISSING="" for I in 1 2 3 4 5 6 7 8 9 10 11 ; do if test ! -f ark${I}isdone ; then MISSING="${MISSING} ${I}" fi done if test "${MISSING}" = "" ; then echo You have unpacked all 11 archives. rm -f ark[1-9]isdone ark[1-9][0-9]isdone else echo You still need to unpack the following archives: echo " " ${MISSING} fi ## End of shell archive. exit 0 -- Sincerely, berg@pool.informatik.rwth-aachen.de Stephen R. van den Berg (AKA BuGless). berg@physik.tu-muenchen.de "Always look on the bright side of life!" exit 0 # Just in case...