#!/usr/bin/perl use Mail::Mbox::MessageParser; use Digest::SHA1 qw(sha1_hex); use strict; use warnings; # Boilerplate. See man page for Mail::Mbox::MessageParser. my $filename = $ARGV[0]; my $fh = FileHandle->new($filename); my $r = Mail::Mbox::MessageParser->new({ 'file_name' => $filename, 'file_handle' => $fh, 'enable_cache' => 0 }); die $r if !ref $r; my $prologue = $r->prologue; # Main loop. my $msgnum = 0; my %threadindex; my @threadlist; while (! $r->end_of_file) { my $emailref = $r->read_next_email(); my $email = $$emailref; ++$msgnum; my @line = split(/\r\n?|\n/, $email); # Extract message ID, subject, and from line. my ($message_id, $subject, $from, $date); for my $line (@line) { if ($line =~ /^Message-ID:\s+(.*)$/i) { $message_id = $1; $message_id =~ s/^$//; } elsif ($line =~ /^Subject:\s+(.*)$/i) { $subject = $1; $subject =~ s/\s+$//; } elsif ($line =~ /^From:\s+(.*)$/i) { $from = $1; $from =~ s/^\s+//; $from =~ s/\s+$//; $from =~ s/<.*// if $from =~ /^.+"]//g; $from =~ s/\s+$//; } elsif ($line =~ /^Date:\s+(.*)/i) { $date = $1; if ($date =~ /^\w\w\w,\s+(\d+ \w\w\w \d\d\d\d)/) { $date = $1; } else { warn "$msgnum: malformed date ($date)!\n"; next; } } elsif ($line =~ /^$/) { last; } } # Complain if no message-ID. if (!defined $message_id) { warn "$msgnum: no message-ID!\n"; next; } if (!defined $subject) { warn "$msgnum: no subject!\n"; next; } if (!defined $from) { warn "$msgnum: no from!\n"; next; } if (!defined $date) { warn "$msgnum: no date!\n"; next; } # Trim subject. my $trimsubject = $subject; $trimsubject =~s/\[(HACKERS|BUGS|PERFORM|COMMITTERS|ADMIN|DOCS|GENERAL|PATCHES|SQL|Pg-migrator-general)\]\s*//g; $trimsubject =~ s/^Re:\s*//; $trimsubject =~ s/\s*$//; $trimsubject =~ s/^\s*//; $trimsubject =~ s/\s+/ /g; $trimsubject = lc($trimsubject); # Build record for message. my $message = { 'from' => $from, 'subject' => $subject, 'message-id' => $message_id, 'date' => $date, 'msgnum' => $msgnum }; if (!exists $threadindex{$subject}) { push @threadlist, $message; $threadindex{$subject} = $message; } else { push @{$threadindex{$subject}->{'followup'}}, $message; } } print <<'EOM'; open mailbox

open mailbox

EOM my $threadnum = 0; my $previous_section = ''; for my $thread (@threadlist) { my $current_section = section_name($thread); if ($current_section ne $previous_section) { print "

Threads Starting In $current_section

"; $previous_section = $current_section; } print "
" . format_msg($thread) . "
\n"; for my $msg (@{$thread->{'followup'}}) { print "
" . format_msg($msg) . "
\n"; } print "\n"; } sub format_msg { my ($m) = @_; sprintf "%s %s %s %s", substr(sha1_hex($m->{'message-id'}), 0, 8), $m->{'message-id'}, $m->{'subject'}, $m->{'from'}, $m->{'date'}; } sub section_name { my ($m) = @_; my $x = $m->{'date'}; $x =~ s/^\d+\s*//; $x; } print < EOM