#!/usr/bin/perl

$USAGE = "
  SYNOPSIS: Scan mail files for new email addresses for mailrc
  AUTHOR: Mohsin Ahmed, http://www.cs.albany.edu/~mosh
  USAGE:   $0 < mbox >> ~/.mailrc
          mail-alias mailfiles | more
   options: -v verbose
            -a all addresses
            -ns input in Netscape mail address format
";

while( $_ = $ARGV[0],  m/^-/ ){
    shift; last if /^--$/;
    $verbose++, next if m/^-v/;
    $all++, next if m/^-a/;
    $netscape++, next if m/^-ns/;
    die $USAGE if m/^-?/;
}

open( MAILRC, "< $ENV{HOME}/.mailrc")
    or die "Cannot read ~/.mailrc\n";

my( %alias, %mailrc );

# Scan the mailrc for old aliases.
while( <MAILRC> ){
    my $addr;
    #  alias name, name is $1.
    if( s/alias\s+(\S+)// ){
        $alias{$1}++;
    }
    # '@' mean address.
    while( s/([\w\d-_.]+\@[\w\d-._]+)// && ($addr = $1) ){
        $mailrc{$addr}++;
    }
}
close MAILRC;

# Now scan the mail files, format: ^From: Full Name <addr>
# Lower case the addr, remove spaces from Full Name and
# add it to %name

my %name;
LINE:
while(<>){
    my( $name, $addr );
    if( ($name,$addr) = m/^From:\s+\"?(.*?)\"?\s+<(.*)>/ ){
        $addr = "\L$addr";
        $name =~ s/\W+/_/g;
        # print "$name is $addr\n";
        next if( $name{$addr} && (index($name{$addr},$name)>=0));
        $name{$addr} .= $name;
        next LINE;
    }
    if( $netscape ){
        # "'Hindustan Organic Chemicals,Cochin'" <hoclchin@md2.vsnl.net.in>,
        while( m/</ ){
            my $name, $addr;
            s,^\s+,,;
            if( s,['"]+(.+)['"]+,, ){
                $name = $1;
            }elsif( s,^(\S+.*?)\s*<,<, ){
                $name = $1;
            }
            if( s/<([^>]+)>,?\s*// ){
                $addr = $1;
            }
            $name =~ s/\s+$//; $name =~ s/^\s+//;
            $name =~ s/\s+/_/g;
            # print "name=$name, addr=$addr, _='$_'\n";
            if( $name and ! $addr ){
                die "No addr for name=$name.\n";
            }
            $name{$addr} .= $name;
        }
    }
    # # This picks up too many addresses from headers.
    if( $all ){
        while( s/([\w\d-_]+\@[\w.\d-_]+)// && ($addr = "\L$1") ){
            $name{$addr} ||= "?";  # Missing Name is '?'.
        }
    }
}

# Now print them out.

foreach $addr (sort keys %name){

    next if $mailrc{$addr};
    next if $addr =~ m/sasi\.com$/i; # don't keep local addresses.
    next if $addr =~ m/microsoft\.com$/i; # don't keep local addresses.    
    my $name = $name{$addr};
    my $alias;

    if( $name =~ m/([A-Za-z]+)/ ){
        $alias = $1;
    }elsif( $name =~ m/(\w+)/ ){
        $alias = $1;
    }else{
        $alias = $addr;
        $alias =~ s/@.*//;
    }
    $alias = lc($alias);

    # Create a unique alias name, ie. name_N
    if( $alias{$alias} ){
        $alias .= "_2";
        $alias++ while( exists $alias{$alias} );
    }

    $alias{$alias}++;
    printf "alias %-5s\t(%s)\t%s\n", $alias, $name{$addr}, $addr;
}

# EOF