#!/bin/perl
#-----------------------------------------------------------------------------#
# File:    	n2folder
# Function:	Convert VM/CMS notebook to Pine mail folder
# Usage:
#    n2folder  [-q] [-m maildir] [-r rscsnodes] [-h host] [notebook]...
#
# If no notebooks are specified, this program reads from standard input
# and writes to standard output.  -q means quiet.  If -m maildir
# is not identified, ~/mail/ is presumed as the mail folder directory.
#
# Folder format requires (1) a header line like...
#		From someplace Fri Sep  9 09:19:46 1994
# (2) the date here must be in Unix cdate format,
# and (3) each note ends with one null (not just blank) line.
# Besides insuring this, this program also attemps to reformat VM/CMS note
# header email addresses and dates into RFC822 format.
# 
# Update:	1995-07-24
# By:    	ccgreg@missouri.edu
#-----------------------------------------------------------------------------#

# The case-insensitive regular expression value of -r $rscsnodes is one or more 
# RSCS node names, which will be converted to the single internet-style 
# hostname -h $internethost.  -h $internethost will also be used to form
# the address where no hostname was given.

$rscsnodes ='^mizzou1$|^umcvmb$|^umvma$';
 
$internethost  ='Mizzou1.Missouri.edu';

### Let's hope you don't need to change more than the above. ###

require 'timelocal.pl';
require 'ctime.pl';
$verbose = 1;
$blankdelim = 1;	# -1/-2 Number of blank lines expected at end of note.
$error=0;
$script = $0;
if ($script =~ /.+\/(.*)/) { $script=$1;}
$MONTHS='jan feb mar apr may jun jul aug sep oct nov dec';

# Parse options:

foreach $i (0 .. $#ARGV) {
	   if (@ARGV[$i] eq '-0') {$blankdelim=0;}
	elsif (@ARGV[$i] eq '-1') {$blankdelim=1;}
	elsif (@ARGV[$i] eq '-2') {$blankdelim=2;}
	elsif (@ARGV[$i] eq '-q') {$verbose=0;}
	elsif (@ARGV[$i] eq '-m') {$i++; $maildir = @ARGV[$i]; }
	elsif (@ARGV[$i] eq '-r') {$i++; $rscsnodes = @ARGV[$i]; }
	elsif (@ARGV[$i] eq '-h') {$i++; $internethost = @ARGV[$i]; }
	elsif (@ARGV[$i] =~ /^-/) {
		print STDERR "usage: $script [-q] [-m maildir]" . 
			" [-r rscsnodes] [-h host] [notebook]...\n";
		$error++;
	}
	else {
		$infn = @ARGV[$i];
		push (@filelist,$infn);
		if (! (-r $infn && -f $infn )) {
			print "Can't read file $infn, $!\n";
			$error++;
		}
	}
}
if ($error != 0) {exit 1;}

# Check existence & permissions of  input and output files #
($login,$x,$x,$x,$x,$x,$x,$home,$x)=getpwuid($>);
if (!defined($maildir)) { $maildir = $home . "/mail";}

if (defined(@filelist)) {
	if (! -e $maildir) {
		mkdir($maildir,'700') || 
			die "Can't create output directory $maildir, $!\n";
	}
	if (! (-d $maildir && -w $maildir && -x $maildir) ) {
		die "Can't write to directory $maildir $!\n"; 
	}
	foreach $file (@filelist) {
		if (-e "$maildir/$file") {
			print STDERR "\n$script: $maildir/$file\n" .
			"Output folder exists! Quit/Append/Remove (q/a/r)? ";
			$ans = getc;
			if ($ans =~ /^q/i) {exit;}
			elsif ($ans =~ /^a/i) {}
			elsif ($ans =~ /^r/i) { unlink "$maildir/$file"; }
			else {redo};
		}
	}
}
else {@filelist = ('&STDIN');}

## Main loop:  convert each requested notebook ##

foreach $infn (@filelist) {
	open(INFILE,"<$infn") || die "Can't open input file $infn";
	"/$infn" =~ /.*\//;
	$newfn = $';
	if ($infn eq '&STDIN') {$outfn='&STDOUT';}
	else {$outfn =  "$maildir/$newfn";}
	open(OUTFILE,">>$outfn") || die "Can't open output file $outfn";
	if ($verbose) {print STDERR "$script: converting $infn ...\n";}
	$nnotes = 0;
	$blanksout=9;  #counts number of blank lines already output.
	# Read and dispatch each line of input:  
	while (<INFILE>) {
		if (/^\s*$/) { # we've read a blank line
			if ($inheader) {
				# End of header, output accumulated lines
				$nnotes++;
				print OUTFILE "From $from $date\n";
				if ($verbose) {print STDERR 
					"$script: $nnotes From $from $date\n";}
				foreach $line (@hdrlines) {
					print OUTFILE $line,"\n";
				}
				@hdrlines=();
				$inheader=0;
			}
			$blanksout++;
			print OUTFILE "\n";
		}

		elsif (/^={72,}/) { # VM/CMS Note separator
			$inheader=1;
			while ($blanksout < 2) {
				print OUTFILE "\n";
				$blanksout++;
			}
			$from=$login;
			$date='Mon Jan  1 00:00:00 1970';
			$blanksout=0;
		}
		
		elsif ($inheader) {
			chop;
			if    (/^Date: /i) { $date=&rfcdate; $inaddr=0;}
			elsif (/^From: /i) { $from=&address; $inaddr=1;}
			elsif (/^To: |^Cc: /i)   { &address; $inaddr=1;}
			# Reply-to, Bcc, etc. don't need CMS Note conversion.
			elsif ($inaddr  && (! /^\S+: /)) {  #addr continuation
				&address;
				#append any missing comma to preceding header:
				if (@hdrlines[$#hdrlines] !~ /,\s*$/) {
					@hdrlines[$#hdrlines] .= ',';
				}
			}
			else {$inaddr=0;}
			push (@hdrlines, $_);
		}
		else {
			print OUTFILE;
			$blanksout=0;
		}
		
	} #while INFILE#
	while ($blanksout < 2) {
		print OUTFILE "\n";
		$blanksout++;
	}

	# Done with this file. 
	close INFILE;
	close OUTFILE;
	if ($verbose) {
		print STDERR "$script: $nnotes notes converted to $outfn\n\n";
	}
} #main loop# 
exit; # program n2folder #


# Try to convert IBM VM/CMS NOTE addresses to RFC822
# Input $_ is line containing addresses to convert.
# $_ may return modified, with proper addresses, 
# Return value is first email address found, converted to RFC822 form.
sub address {
	@line = split;
	if (@line[$#line-1] =~ /at/i) { # VM/CMS style address:  userid at node
		$userid = @line[$#line-2];
		$node =  @line[$#line];
		if ($node =~ /$rscsnodes/i) { 
			# convert known local node(s)
			$node=$internethost;
		}
		else {$node = "$node.bitnet";} # otherwise use .bitnet gateway
		$email = "$userid\@$node";
		$#line = $#line - 3;
		$_ = join(' ',@line) . " <$email>";
	}
	elsif (($email) = grep(/\S+\@\S+/,@line)) { 
		# Blindly assume "@" implies RFC822
		if ($email eq '') {$email = 'unknown';}
		else {$email =~ tr/<>,//d;}
	}
	else {  # VM/CMS style local address:  userid[, userid]...
		($email = "@line[1]\@$internethost") =~ tr/,//d;
		$_ = shift(@line);
		foreach $addr (@line) {
			$comma = substr(' ,',($addr =~ tr/,//d),1);
			$_ .= " $addr\@$internethost$comma";
		}
	}
	return ( $email );
} # end of sub address #


# Folder headers require cdate format... 
# 		From someuser Sat Feb  1 14:02:47 1995
# So try to extract date from the Date: line
# Return correct cdate format for even if content is bogus.
# Also return typical RFC 822 format of original Date: line.
# Typical (!) input formats:  
# RFC 822...	Date: Wed, 1 Feb 1995 14:02:46 CST
# VM/CMS....	Date: 28 November 94, 15:25:00 CST

sub rfcdate {
	@line = split;
	shift(@line);			# Discard Date: tag
	if ($#line > 4) {shift(@line);}	# Discard day of week
	$mday=@line[0];
	($mo=substr(@line[1],0,3)) =~ tr/A-Z/a-z/;
	$mon= index($MONTHS,$mo)/4;
	($yr = @line[2]) =~ tr/,//d;
	if ($yr > 1969 & $yr < 2000) {$yr=$yr-1900;} # I'm taking that cruise!
	($hh,$mm,$ss) = split(/:/,@line[3]);
	$date = &ctime( &timelocal(($ss,$mm,$hh,$mday,$mon,$yr)) );
	chop $date;
	$_ = "Date: " . substr($date,0,3) . "," . 
		substr($date,7,3) . substr($date,3,4) . 
		substr($date,19,5) . substr($date,10,10) ;
	return ($date);
} # end of sub rfcdate #

# end of program n2folder #