#!/usr/bin/perl -w
#
# Original Version
# ----------------
#	http://www.UCEAS.net/Scripts/AntiSpam.Collector.pl
#
# Release License
# ---------------
#	GPL
#
#
# Purpose
# -------
#	Separate the incoming spam into different files
#
# Process
# -------
#	Collect incoming spam at: SpamCollector@UCEAS.net
#	Send back acknowledgement status email back submitter of the unwanted spam
#		- majordomo config vs mailman
#	Process the new spam collected
#		- update /etc/mail/spam_domains
#		- update /etc/mail/spammers
#	Save original incoming spam submission for the day for "regeneration"
#
#
# Spam Prevention
# ---------------
#	use http://Spam.uceas.net	-- to view the daily statistics
#	download/install our "spam db files" into your mail server
#		( sendmail or exim or ?? )
#
#
# To Do
# -------
#	- watchout for From ( >From ) in the beginning of the sentence in the body of the emails
#		- not solved yet
#	- Fix "submitting user" as NOT the spammer
#	- Fix Respawning
#	- NO error checking yet ...
#	- folders for ~/Mail/1999, ~/Mail/2000 ~/Mail/customers, ~/Mail/friends
#	- watch out for additional incoming emails ( file locking )
#
#	- Collect data from other "spam collection" servers ( distributed database )
#
#
# Client Usage  ( for submitting their unwanted spam )
# -------------
#	submit.spam.sh
#	- aka -
#	AntiSpam.Collector.pl -update -submit -user /var/spool/mail/your-mailbox-with-spam
#
#
# SpamCollector Usage ( for processing )
# -------------------
#	mailbox2file.sh
#	- aka -
#	AntiSpam.Collector.pl -update -process -user /var/spool/mail/spamcollector -web ww -orig xx  -head yy
#		ww = top of the html directory		( /home/httpd/html )
#		xx = html directory of Orignal spam	( SpamCollector.Orig )
#		yy = html directory of spam headers	( SpamCollector.Head )
#
#		http://your-site.com/SpamCollector.Head
#		http://your-site.com/SpamCollector.Orig
#
#
# Setup & Permissions
# --------------------
#	chown root.mj	html/Scripts/AntiSpam.Collector.pl
#	chown 754	html/Scripts/AntiSpam.Collector.pl
#
#	crontab -e
#		...
#		#
#		# mailbox2file.sh invokes AntiSpam.Collector.pl to process /var/spool/mail/spamcoll
#		#
#		05,35 * * * *   /home/httpd/html/Scripts/mailbox2file.sh 
#		08,38 * * * *   /home/httpd/html/Scripts/mailbox2file.sh test
#
#
# Create the Working directories
# ------------------------------
#	cd /home/httpd/html
#
#	mkdir 		   SpamCollector.Head SpamCollector.Orig SpamCollector.Test SpamCollector.Test.Orig
#	chmod 770	   SpamCollector.Head SpamCollector.Orig SpamCollector.Test SpamCollector.Test.Orig
#	chown WebMaster.mj SpamCollector.Head SpamCollector.Orig SpamCollector.Test SpamCollector.Test.Orig
#
#
#
# 18-Oct-02 amo Date-of-Birth 
# 23-Dec-02 amo Added real code and comments - Ver 0.2 release
# 29-Dec-02 amo Ver-0.3 Fixed -orig and -head
# 02-Jun-03 amo Ver-0.4 Added SpamCollectorTest or SpamCollector
# 13-May-03 amo Added chmod  644 *.orig
#
#
#
# ---------------------  User Defined Variables/Options   -------------------------
#
my ( $DEBUG ) = 0;
#
#
# Allow users to submit the unwanted spam left in their /var/spool/mailbox
#
my ( $SubmitSpam ) = 0;		# user sending in their spam into the repository
#
my ( $ProcessSpam ) = 0;
#
#
my ( $Update ) = 0;	# update the script before processing
#
#
#
# if  ~/Mail/Family/xxx, ~/Mail/Friends/yyy  and ~/Mail/Customers/zzz...
#	- use those names ( family, friends, .. ) as folders
#
#
# The default spam-filled mailbox to convert and save as files
#
#y ( $SRC ) = "/var/spool/mail/root" ;
#y ( $SRC ) = "/var/spool/mail/your-left-over-spam" ;	# your mailbox with the spam
my ( $SRC ) = "/var/spool/mail/spamcollector" ;		# incoming spam mailbox
#
#
#
my ( $WEB ) = "/home/httpd/html";	# top of the web tree
my ( $DST ) = "SpamCollector.Head" ;	# Spam database for web browsing
my ( $ORG ) = "SpamCollector.Orig" ;	# original spam report
#
#
# ---------------------  End of user cconfig  -------------------------------------
#
my ( $NM ) = "AntiSpam.Collector.pl";
#
my ( $Ver) = "0.5test";	# Release Version
#
#
# Where to Find the binaries on your System
# -----------------------------------------
#
my ( $WGET ) = "/usr/bin/wget";
my ( $DATE ) = "/bin/date";
my ( $MV ) = "/bin/mv";
#
#
# Email address to submit your Unwanted spam
#
my ( $SpamCollectorEmail ) = "SpamCollector\@UCEAS.net";
#
#
my ( $SpamDir ) = "Undefined";		# DST/yr/mon/date
my ( $SpamFile ) = "Undefined";		# time.unique_name
#
my ( $SpamSubmitorFile ) = "Undefined";
my ( $SpamSubmitorHeader) = "Undefined";
#
#
# Process Command line options
# ----------------------------
&cmd_opt;
#
printf "..$NM-$Ver..\n";	# which version
#
#
# Now Check and update the Versions
# ---------------------------------
&check_updates;
#
#
&submit_spam if ( $SubmitSpam );
#
#
&process_spam if ( $ProcessSpam );
#
printf "Done\n\n";
#
exit 0;	# done
#
#
#
# =================================================================
#
# Just Submit the Spam ( just send it in )
# --------------------
#
sub submit_spam 
{
  #
  #
  # Spam to submit to the repository for analysis
  #
  if ( ! -f $SRC ) {
    printf "\nERROR: Aborting, No such mailbox=$SRC\n\n";
    exit 1;
  }
  #
  #
  my ( $cmd ) = "mail -v -s 'incoming spam' $SpamCollectorEmail < $SRC ";
  #
  #
  printf "\n";
  printf "Finished Submitting the unwanted spam in $SRC\n";
  printf "You should receive an automated confirmation reply in a few minutes\n";
  printf "\t$cmd\n\n";
  ` $cmd `;
  #
  # Sun Dec 29 09:00:06 PST 2002
  my ( @dt ) = split ( /\s+/, ` $DATE ` );
  my ( $yr, $mid, $date, $uniqtime ) = &year_mon_date_time ( $dt[5], $dt[1], $dt[2], $dt[3] );
  #
  ` mv $SRC $SRC.$mid.$date.$uniqtime `;
  #
  #
} # submit_spam
#
#
# Where to put the Processed Spam
# -------------------------------
#
sub process_spam
{
  #
  if ( ! -d "$WEB/$DST" ) {
    printf "WARNING: Spam directory=$DST does NOT exist\n\n";
    # exit 1;
  }
  if ( ! -d "$WEB/$ORG" ) {
    printf "WARNING: Spam directory=$ORG does NOT exist\n\n";
    # exit 1;
  }
  #
  #
  # wait a little for mj to finish writing and syncing the mailbox
  #
  if ( ! -f $SRC ) {
    printf "\nERROR: Aborting, No such mailbox=$SRC\n\n";
    exit 1;
  }
  #
  # For some reason, /var/spool/mail/spamcoll mailbox does NOT yet have the incoming spam 
  #
  # if ( $DEBUG ) {
    @ls = `ls -la $SRC`;
    $l = join ( " ", @ls );
    #
    printf "Processing incoming Spam=$SRC..\n\t$l..\n";
    #
  # }
  #
  #
  if ( -z $SRC ) {
    printf "\nWARNING: No un-processed incoming spam=$SRC\n\n";
    return 0;
  }
  #
  #
  # Need to clock it to prevent other incoming emails, 
  # since we are gonna delete the mailbox after processing
  #
  my ( $LOCK ) = "${SRC}.lock";
  #
  ` touch $LOCK ` ;
  #
  printf "Processing incoming Spam=$SRC..lock=$LOCK..\n";
  #
  my ( $status ) = open ( FH, "< $SRC" );
  #
  if ( $status <= 0 ) {
    printf "ERROR: Cannot read $SRC..\n";
    exit 1;
  }
  #
  my ( $line ) = "";
  my ( $lcnt ) = 0;
  #
  my ( $header ) = 1;
  my ( $spam ) = "";
  my ( $msg ) = "";
  #
  # Read the Incoming Emails
  # ------------------------
  #
  while ( <FH> ) {
    #
    chomp ( $line = $_ );
    #
    # printf "..header=$header..line[$lcnt]=$line..\n";
    #
    # Local files to process, or incoming Spam being submitted
    #
    #	From	-- will be the user submitting the unwanted spam
    #	>From	-- will be the start of the Spam
    #
    #
    ( $header, $spamtype, $spam ) = &check_from ( $line ) if ( ( s/^From /From / ) || ( s/^>From />From / ));
    #
    #
    # so that an email reader can read the spam 
    #
    $line =~ s/^>From /From /;
    #
    #
    # Process/Save only the header, original content is intact in *.done
    #
    printf SP "%s\n", $line if ( $header );
    #
    #
    # Blank line between email header and email body
    #	- watch out for blank lines in the email body
    #
    #
    if ( $header && ( "$line" eq "" ) ) {
      #
      $header = 0;
      #
      $msg = "This is the spammer header info, the spam itself was truncated ";
      $msg = "This user submited their unwanted Spam" if ( $spamtype eq "Submitor" );
      #
      printf SP "-- $msg \n-- See $ORG/$SpamSubmitorFile.orig for the spam in its entirty\n";
      printf SP "\n";
      #
    } # done headers
    #
    $lcnt += 1;
    #
  } # end of incoming emails
  #
  # Close the last spam
  #
  &spam_done ( 1, $spam );	# relative to WEB/DST
  #
  close ( FH );
  #
  #
  # use the First From time as its unique Date/Time stamp
  #
  $cmd = "$MV $SRC $WEB/$ORG/$SpamSubmitorFile.orig ";	# SDIR/$Time.orig `;
  #
  printf "$cmd\n";
  #
  # 13-May-03 amo Added chmod  644 *.orig
  #
  #
  # ` $cmd `;
  ` $cmd ; chmod 644 $WEB/$ORG/$SpamSubmitorFile.orig `;
  #
  unlink $LOCK ;
  #
  #
} # process_spam
#
#
# Use the Date stamp of the spam as its unique filename, ( From-Date.Process-Date )
#
#	From	-- will be the user submitting the unwanted spam
#	>From	-- will be the start of the Spam
#
sub check_from 
{
  my ( $line ) = $_[0];
  #
  printf "..From=$line..\n";	# if ( $DEBUG );
  #
  #
  #	From user@Their-Domain.com Mon Dec 23 11:36:36 2002
  #
  my ( $From, $ud, $Dow, $Mon, $Date, $Time, $Year ) = split ( /\s+/, $line );
  $From = "x";  $ud = "x";	# get rid of dumb warnings
  #
  #
  #
  # Check the Spam Directory tree
  # -----------------------------
  my ( $ymdt ) = &check_spam_dir ( $Year, $Mon, $Date, $Time );		# yr/mon/date/hh.mm.ss
  #
  #
  # Close the previous spam
  # -----------------------
  my ( $spamtype, $spam ) = &spam_done ( 0, $ymdt );	# year/mon/date/hh.mm.ss[.submitor] elative path to WEB/DST
  #
  return ( 1, $spamtype, $spam );	# start new spam
  #
} # check_from
#
#
# Done extracting the headers of the submittor or the unwanted Spam 
# -----------------------------------------------------------------
#
sub spam_done
{
  #
  my ( $done ) = $_[0];		# done
  my ( $spam ) = $_[1];		# year/mon/date/hh.mm.ss
  #
  my ( $SpamType ) = "spam";
  my ( $status ) = 0;
  #
  #
  # Check for existing spam filenames
  #
  if ( -f "$WEB/$DST/$spam" || -f "$WEB/$ORG/$spam.Submitor" ) {
    #
    printf "WARNING: spam=$DST/$spam already exists\n" if ( $DEBUG );
    #
    # Add local processing time for unique spamFileName
    #
    # Sun Dec 29 09:00:06 PST 2002
    my ( @dt ) = split ( /\s+/, ` $DATE ` );
    my ( $yr, $mid, $date, $uniqtime ) = &year_mon_date_time ( $dt[5], $dt[1], $dt[2], $dt[3] );
    #
    $spam .= ",$uniqtime";
    #
  } # uniq spam file name
  #
  #
  #  First (From) time is the submittor of the unwanted spam
  #
  if ( $SpamSubmitorFile eq "Undefined" ) {
    #
    # about to process the submitors header info
    $SpamSubmitorHeader = 1;
    #
    #
    $spam .= ".Submitor";
    #
    $SpamSubmitorFile = "$spam";	# assign it once
    $SpamType = "Submitor";
    #
  } elsif ( $SpamSubmitorHeader ) {
    #
    $SpamSubmitorHeader = 0;
    #
    # 2nd time thru, we just finished processing the submitors header info
    #
    $SpamType = "Submitor";
    #
    # printf "..done $SpamType..submitor=$SpamSubmitorFile..\n";
    #
    # About to process the spam
    #
    $SpamType = "spam";
    #
  } else {
    #
    $SpamType = "spam";		# just to do something
    #
    # rest of the file is all spammers
    #
    # printf "..done $SpamType..submitor=$SpamSubmitorFile..\n";
    #
  }
  #
  close ( SP );	# close previous spam
  #
  #
  # Nothing left to do
  #
  if ( ! $done ) {
    #
    # Full path to spam file only where needed
    # -----------------------------------------
    if ( $SpamType eq "Submitor" ) {
      $status = open ( SP, "> $WEB/$ORG/$spam" );
    } else {
      $status = open ( SP, "> $WEB/$DST/$spam" );
    }
    #
    printf "ERROR: Could not open file=$spam..\n" if ( $status != 1 );
    #
    # printf "....Saving a new $SpamType email[$status]: $spam\n";
    printf ".... new $SpamType: $spam\n";
    #
  } # done
  #
  return ( $SpamType, $spam );	# submitor or spam
  #
} # spam_done
#
#
# Check where to Save all the incoming Spams
# ------------------------------------------
#
sub check_spam_dir
{
  my ( $y ) = "$_[0]";	# 2002
  my ( $m ) = "$_[1]";	# Dec
  my ( $d ) = "$_[2]";	# 29
  my ( $t ) = "$_[3]";	# time  hh.mm.ss
  #
  #
  # Create unique filename for each incoming spam
  #
  my ( $yr, $mid, $date, $time ) = &year_mon_date_time ( $y, $m, $d, $t );
  #
  # printf "..$yr..$mid..$date..$time..\n";
  #
  #
  # Main Spam directory
  # --------------------
  #
  my ( $sdir ) = "$WEB/$DST/$yr/$mid/$date";
  #
  if ( ! -d "$sdir" ) {
    printf "WARNING: Creating new SpamDirectory.Head=$sdir..\n";
    ` mkdir -p $sdir `;
  } # spam directory
  #
  my ( $odir ) = "$WEB/$ORG/$yr/$mid/$date";
  if ( ! -d "$odir" ) {
    printf "WARNING: Creating new SpamDirectory.Orig=$odir..\n";
    ` mkdir -p $odir `;
  } # spam directory
  #
  #
  return ( "$yr/$mid/$date/$time" );	# ymdt
  #
} # check_spam_dir
#
#
# Multiple people could have gotten a spam at the same mon/day/mm.hh.ss
# use a counter for distinction
#
sub year_mon_date_time
{
  my ( $y ) = "$_[0]";	# 2002
  my ( $m ) = "$_[1]";	# Dec
  my ( $d ) = "$_[2]";	# 29
  my ( $t ) = "$_[3]";	# time  hh.mm.ss
  #
  # Convert Months into a number
  #
  my ( @MID ) = ();
	$MID{ "Jan" } = "01";
	$MID{ "Feb" } = "02";
	$MID{ "Mar" } = "03";
	$MID{ "Apr" } = "04";
	$MID{ "May" } = "05";
	$MID{ "Jun" } = "06";
	$MID{ "Jul" } = "07";
	$MID{ "Aug" } = "08";
	$MID{ "Sep" } = "09";
	$MID{ "Oct" } = "10";
	$MID{ "Nov" } = "11";
	$MID{ "Dec" } = "12";
  #
  #
  my ( $date ) = sprintf "%02d", $d;		# 01  for the 1st of the month
  #
  my ( $mid ) = $MID{ $m };
  #
  $t =~ s/:/./g;	# use time as part of the filename
  #
  return ( $y, $mid, $date, $t );
  #
} # year_mon_date_time
#
#
# Check for any updates
# ---------------------
#
sub check_updates
{
  #
  # Get the Latest copy and compare it
  # ----------------------------------
  #
  if ( ! -x $WGET ) {
    #
    printf "\n";
    printf "WARNING: Could NOT find $WGET for updating $0..\n";
    #
    return ( 0 );	# just return, pretend its up to date
  }
  #
  my ( $new ) = "cd /tmp ; rm -f $NM ; $WGET -q http://www.UCEAS.net/Scripts/$NM ";
  #
  printf "\n";
  printf "Checking against the master copy: http://www.UCEAS.net/Scripts \n";
  printf "\t $new \n" if ( $DEBUG );
  ` $new `;
  #
  #
  #
  my ( $diff ) = "diff /tmp/$NM $0" ;
  my ( @diffs ) = ` $diff `;
  #
  my ( $update ) = "chmod 754 /tmp/$NM ; cp /tmp/$NM $0 ";
  #
  #
  # Update before executing
  #
  if ( $#diffs > 0 ) {
    #
    # Found Differences
    #
    printf "\n";
    printf "WARNING: Differences encountered, Please update first\n";
    printf "\t $0 -update \n";
    printf "\t -- or --\n";
    printf "\t $update \n\n";
    #
    printf " $diff \n";
    printf "%s\n", join ( " ", @diffs );
    printf "\n";
    #
    if ( $Update == 1 ) {
      #
      printf "WARNING: Updating: $update\n";
      ` $update `;
      #
      # now need to respawn with the new updated script
      # ------------------------------------------------
      #
      ` exec $0 `;
      #
    } else {
      exit 1;
    }
    #
  } else {
    #
    printf "\t $0 is up to date\n";
    #
  }
  #
} # check_updates
#
#
# Help File
# ------------
#
sub usage
{
  printf "\n";
  printf "$NM\n";
  printf "\t -h	\t these help files\n";
  printf "\t -v	\t which version is this\n";
  printf "\t -update \t check for latest version and updates before executing\n";
  printf "\n";
  printf "    For Users:\n";
  printf "\t -submit \t\t\t Submit your unwanted spam to the SpamCollector Servers \n";
  printf "\t -test \t\t\t Submit your unwanted spam to the 'test' SpamCollector Servers \n";
  printf "\t -user /var/spool/mail/you \t your mailbox with spam in it\n";
  printf "\n";
  printf "\n";
  printf "    For SpamCollector Servers:\n";
  printf "\t -process \t\t\t Processing the incoming unwanted spam from submitters \n";
  printf "\t -user /var/spool/mail/spamcoll \t incoming spam to process \n";
  printf "\t -web /home/httpd/html/SpamCollector \t Deposit Processed Spam here\n";
  printf "\n";
  printf "\n";
  exit 1;
  #
} # usage
#
#
# Process command line options
# -----------------------------
#
sub cmd_opt
{
  my ( $id ) = 0;
  #
  my ( $argv ) = "";
  #
  #
  while ( $id <= $#ARGV ) {
    #
    $argv = $ARGV[$id];
    #
    # printf "..argv[$id]=$argv..\n";
    #
    if ( "$argv" eq "-h" ) {
      #
      &usage;	# help file
      #
    } elsif ( "$argv" eq "-v" ) {
      #
      printf "$NM-$Ver\n\n";	# which version
      exit 0;
      #
    } elsif ( "$argv" eq "-update" ) {
      #
      $Update = 1;	# check for updates before executing
      #
    } elsif ( "$argv" eq "-process" ) {
      #
      $ProcessSpam = 1;	# Process the incoming unwanted Spam
      #
      $SRC = "/var/spool/mail/spamcoll";
      #
    } elsif ( "$argv" eq "-submit" || "$argv" eq "-test" ) {
      #
      $SubmitSpam = 1;	# Submit your unwanted Spam
      #
      $SRC = "/var/spool/mail/your-mailbox-with-unwanted-spam";
      #
      $SpamCollectorEmail = "SpamCollectorTest\@UCEAS.net" if ( "$argv" eq "-test" );
      #
    } elsif ( "$argv" eq "-user" ) {
      #
      $id += 1;
      $SRC = $ARGV[$id];	# use /var/spool/mail/USER
      #
    } elsif ( "$argv" eq "-web" ) {
      #
      $SubmitSpam = 0;	# we are NOT submitting it, but are processing the incoming spam
      #
      $id += 1;
      $WEB = $ARGV[$id];	# use /home/httpd/html
      #
    } elsif ( "$argv" eq "-head" ) {
      #
      $SubmitSpam = 0;	# we are NOT submitting it, but are processing the incoming spam
      #
      $id += 1;
      $DST = $ARGV[$id];	# use WEB/SpamCollector.Head
      #
    } elsif ( "$argv" eq "-orig" ) {
      #
      $SubmitSpam = 0;	# we are NOT submitting it, but are processing the incoming spam
      #
      $id += 1;
      $ORG = $ARGV[$id];	# use WEB/SpamCollector.Orig
      #
    }
    #
    $id += 1;
  }
  #
} # cmd_opt
#
#
# End of file