-#! /usr/bin/env perl
-
-# $Id$
+#!/usr/bin/perl
=head1 NAME
B<minc> B<-d>
-B<minc> B<-h>
+B<minc> B<-r>
=head1 DESCRIPTION
use strict;
use warnings;
+$SIG{'PIPE'} = 'IGNORE';
+
use Data::Dumper;
use Errno;
use Fcntl qw(O_WRONLY O_EXCL O_CREAT);
use File::Temp qw(tempfile);
use File::stat;
use Getopt::Long qw(:config gnu_getopt no_ignore_case);
-use POSIX qw(strftime WEXITSTATUS WIFEXITED);
use Pod::Usage;
our $VERSION = 1;
my $MAGIC_TO_REGEX = '^((Original-)?(Resent-)?(To|Cc|Bcc)|(X-Envelope |Apparently(-Resent)?)-To)';
my $MAGIC_TO_TOKEN = ' TO';
-# List of SPAM message numbers, scanned at the end so the user can
-# check for false positives.
-my @SPAM;
+# Mapping of message numbers to array references. The first element is set by
+# filter_mail to a reference to a header hash for the message; the second is
+# set by maildier_spam to the name of the message file in the spam maildir.
+# scan_spam scans this at the end so the user can check for false positives.
+my %SPAM;
=head1 OPTIONS
Dry run; do not actually incorporate the mail, but log and report to
stdout/stderr as normal.
+=item B<-r>
+
+Rebuild `mhpath +`/.folders from scratch, processing no mail.
+
=back
=cut
my $help;
my $maxmsgs;
my $norun;
+my $rebuild_dot_folders;
+if (!caller()) {
GetOptions(
'd' => \$dumpfilters,
'h|help' => \$help,
'm=i' => \$maxmsgs,
'n' => \$norun,
+ 'r' => \$rebuild_dot_folders,
) or pod2usage();
$help and pod2usage(-exitstatus=>0, -verbose=>1);
@ARGV == 0 or pod2usage();
+}
our $run = !$norun;
B<minc> adds all folders it filters into to this file, which is used
by lukem's B<new(1)> (XXX need a link).
-=item `mhpath +`/logs/minc.log
-
-Where B<minc> logs what it does, unless in -n mode.
-
-=item `mhpath +`/logs/dryrun.log
-
-Where B<minc> logs what it would do; used in -n mode.
-
=item `mhpath +`/.minc.context
B<minc> uses this file for context (i.e. current folder) instead of
our $SCAN_P_FOLDER = 0.1;
our $SCAN_P_MESSAGE = 0.0625;
our $SCAN_P_FROM = 0.175;
+our @folder_sort_list = (qr/^inbox$/);
+our $folder_sorter = sub { sort_by_list(@_, @folder_sort_list) };
my $mh;
-my $logfile;
$mh = `mhpath +`;
chomp($mh);
-if ($run) {
- $logfile = $mh . '/logs/minc.log';
-} else {
- $logfile = $mh . '/logs/dryrun.log';
-}
-
$ENV{"MHCONTEXT"} = $mh . '/.minc.context';
\f
###############################################################################
-# Logging
-
-sub mylog {
- my $timestamp = strftime('%b %e %H:%M:%S', localtime);
- my $msg;
- foreach my $part (@_) {
- if (defined($part)) {
- $msg .= $part;
- }
- }
- # no newlines in the log message, thanks
- $msg =~ s/\n/ /gm;
-
- open(LOG, ">>$logfile") or die("open(>>$logfile): $!");
- print(LOG "$timestamp $msg\n") or die("print(>>$logfile): $!");
- close(LOG) or die("close($logfile): $!");
-}
+# Utility procedures
-sub logheader {
- my ($text, @contents) = @_;
- my $last;
+sub sort_by_list {
+ my $a = shift;
+ my $b = shift;
- if (@contents) {
- $last = $contents[-1];
- } else {
- $last = '';
+ for my $i (@_) {
+ my $am = $a =~ $i;
+ my $bm = $b =~ $i;
+ if ($am) {
+ if ($bm) {
+ last;
+ }
+ return -1;
+ } elsif ($bm) {
+ return 1;
+ }
}
- mylog('<< ', $text, $last);
+ return $a cmp $b;
}
-sub log_headers {
- my %headers = @_;
-
- # For an explanation of the %headers structure, see the
- # get_headers function below.
- logheader('From: ', @{$headers{'return-path'}});
- logheader('To: ', @{$headers{'to'}});
- logheader('Subject: ', @{$headers{'subject'}});
- logheader('Message-Id: ', @{$headers{'message-id'}});
-}
-
-\f
-###############################################################################
-# Utility procedures
-
sub mkfolder {
my $folder = shift;
my $target;
}
if (@result <= 2) {
- exit(0);
+ return ();
}
STDOUT->autoflush(1);
- print(@result - 2, " messages...");
+ if (@result == 3) {
+ print('1 message...');
+ } else {
+ print(@result - 2, ' messages...');
+ }
closedir(DIR);
return @result;
}
+my %msgnum_cache;
sub get_highest_msgnum {
my $mhfolder = shift;
my $dir;
my $highest;
my $msgnum;
+ if (defined($msgnum_cache{$mhfolder})) {
+ return $msgnum_cache{$mhfolder}++;
+ }
+
$dir = "$mh/$mhfolder";
if (not opendir(DIR, $dir)) {
die("opendir($dir): $!");
}
}
- return $highest;
+ $msgnum_cache{$mhfolder} = $highest;
+ return $msgnum_cache{$mhfolder}++;
+}
+
+sub mark {
+ my $folder = shift;
+ my $msgnum = shift;
+ my $seq = shift;
+ my $fn = "$mh/$folder/.mh_sequences";
+ my $fh;
+ my @sequences;
+
+ if (not open($fh, $fn)) {
+ $!{ENOENT} or die("open($fn): $!");
+ } else {
+ @sequences = <$fh>;
+ chomp(@sequences);
+ }
+
+ my $marked = 0;
+ open($fh, '>', $fn) or die("open(>$fn): $!");
+ for $_ (@sequences) {
+ if (/^$seq: (.*)/) {
+ my @parts;
+ my @result;
+ my $done = 0;
+ for my $part (split(' ', $1)) {
+ if (not $done) {
+ my ($st, $en) = split('-', $part);
+ if ((defined($en) and ($msgnum >= $st and $msgnum <= $en))
+ or $msgnum == $st) {
+ # It's already there.
+ $done = 1;
+ }
+ if (defined($en)) {
+ if ($st - 1 == $msgnum) {
+ $part = "$msgnum-$en";
+ $done = 1;
+ } elsif ($en + 1 == $msgnum) {
+ $part = "$st-$msgnum";
+ $done = 1;
+ }
+ } else {
+ if ($part - 1 == $msgnum) {
+ $part = "$msgnum-$part";
+ $done = 1;
+ } elsif ($part + 1 == $msgnum) {
+ $part = "$part-$msgnum";
+ $done = 1;
+ }
+ }
+ }
+ push(@result, $part);
+ }
+ if (not $done) {
+ push(@result, $msgnum);
+ }
+ print($fh "$seq: ", join(' ', @result), "\n");
+ $marked = 1;
+ } else {
+ print($fh "$_\n");
+ }
+ }
+ $marked or print($fh "$seq: $msgnum\n");
+ close($fh) or die("close(>$fn): $!");
}
sub store_message {
my $msgnum;
my $try;
my $mhmsg;
- my $status;
# We must do this even in -n mode because later steps fail without
# it. This should be harmless.
#sleep(2);
}
- mylog('+', $mhfolder);
-
if ($run) {
if (not rename($msg, $mhmsg)) {
die("rename($msg, $mhmsg): $!");
# fails. While it is slow, it is not safe to store multiple
# messages and then have a failure before marking some (or
# all).
- if ($mhfolder eq 'SPAM') {
- push(@SPAM, $msgnum);
- } else {
- $status = system('mark', "+$mhfolder", "$msgnum", '-sequence',
- 'unseen', '-add');
- # XXX need to handle signalled and stopped, and print
- # the exit code or signal number.
- if (not WIFEXITED($status)) {
- die("Failed to run mark");
- } elsif (WEXITSTATUS($status) != 0) {
- die("Failed to mark message unseen.");
- }
+ if ($mhfolder ne 'SPAM') {
+ mark($mhfolder, $msgnum, 'unseen');
}
}
}
close(MSG);
- return %headers;
+ return \%headers;
}
\f
sub find_mh_folder {
my $msg = shift;
- my %headers = @_;
- my $filterref;
- my @filter;
- my $header;
- my $contents;
- my $pair;
- my $match;
- my $expression;
- my $result;
-
- if (not %headers) {
+ my $header = shift;
+
+ if (not %$header) {
return 'malformed';
}
# Walk the list of filters. This structure is documented in
# pod at the end of the program.
- foreach $filterref (@FILTERS) {
- @filter = @$filterref;
- $header = shift(@filter);
-
- # Handle filters using the magic TO header specially.
- if ($header eq $MAGIC_TO_TOKEN) {
- foreach $header (keys(%headers)) {
- if ($header =~ /$MAGIC_TO_REGEX/i) {
- foreach $contents (@{$headers{lc($header)}}) {
- foreach $pair (@filter) {
- ($match, $expression) = @$pair;
- if ($contents =~ /$match/) {
- return $expression;
- }
- }
- }
- }
- }
-
- # Now that it's been processed specially, skip normal handling.
+ for my $filterref (@FILTERS) {
+ if (ref($filterref) eq 'CODE') {
+ my $m = $filterref->($header, $msg);
+ $m && return $m;
next;
}
- # Walk the list of message headers matching the filter's
- # specified header.
- foreach $contents (@{$headers{lc($header)}}) {
- # Walk the filter's list of match/expression pairs.
- foreach $pair (@filter) {
- ($match, $expression) = @$pair;
- if ($contents =~ /$match/i) {
- if (eval "\$result = \"$expression\"") {
- return $result;
- }
+ my $m = match($header, @$filterref);
+ $m && return $m;
+ }
+
+ return 'inbox';
+}
+
+# Test all the header fields against each [regexp, folder-expression] pair.
+sub match {
+ my $header = shift;
+ my $filter_field = shift;
+ my @filters = @_;
+
+ # Handle filters using the magic TO header specially.
+ if ($filter_field eq $MAGIC_TO_TOKEN) {
+ for my $field_name (keys(%$header)) {
+ if ($field_name =~ /$MAGIC_TO_REGEX/i) {
+ my $m = match_one_field($header->{$field_name}, \@filters);
+ $m && return $m;
+ }
+ }
+ # Now that it's been processed specially, skip normal handling.
+ return;
+ }
+
+ # Walk the list of header fields matching the filter's specified header.
+ my $m = match_one_field($header->{lc($filter_field)}, \@filters);
+ $m && return $m;
+}
+
+# Test all the values of one header field against each [regexp,
+# folder-expression] pair.
+sub match_one_field {
+ my $values = shift;
+ my $filters = shift;
+ for my $value (@$values) {
+ for my $pair (@$filters) {
+ my ($regexp, $expression) = @$pair;
+ if ($value =~ $regexp) {
+ my $result;
+ if (eval "\$result = \"$expression\"") {
+ return $result;
}
}
}
}
+}
- return 'inbox';
+sub scan_line {
+ my ($headers, $mhfolder, $msgnum, $nf, $nm, $nF, $ns) = @_;
+ my $from = '';
+ my $subject = '';
+ # Sometimes these headers are missing...
+ eval { $from = [@{$headers->{'from'}}]->[-1] };
+ eval { $subject = [@{$headers->{'subject'}}]->[-1] };
+ # Replace garbage characters.
+ for ($from, $subject) {
+ tr/\x00-\x1f\x80-\xff/?/;
+ }
+ return sprintf("\%-${nf}s \%${nm}d \%-${nF}s \%s",
+ substr($mhfolder, 0, $nf), substr($msgnum, 0, $nm),
+ substr($from, 0, $nF),
+ substr($subject, 0, $ns));
}
sub filter_mail {
+ @_ or return ();
my $msgcount = @_ - 2; # don't count . and ..
my $len = length($msgcount);
my @baton;
}
}
+ my $tty = -t STDOUT;
+ if (not $tty) {
+ # Print a newline after the incomplete "N messages..." line.
+ print("\n");
+ }
+
for $msg (@_) {
($msg eq '.' or $msg eq '..') and next;
- my %headers = get_headers($msg);
- log_headers(%headers);
+ my $headers = get_headers($msg);
undef($mhfolder);
for my $hook (@filter_hooks) {
- my $result = $hook->(\%batons, \%headers, $msg);
+ my $result = $hook->(\%batons, $headers, $msg);
defined($result) and ($mhfolder = $result);
}
- defined($mhfolder) or ($mhfolder = find_mh_folder($msg, %headers));
+ defined($mhfolder) or ($mhfolder = find_mh_folder($msg, $headers));
$msgnum = store_message($msg, $mhfolder);
$folders{$mhfolder}++;
+ if ($tty) {
+ print("\r");
+ }
+
if ($mhfolder eq 'SPAM') {
$spam++;
+ $SPAM{$msgnum} = [$headers, undef];
} else {
$saved++;
- my $from = [@{$headers{'from'}}]->[-1];
- my $subject = [@{$headers{'subject'}}]->[-1];
- for ($from, $subject) {
- tr/\x00-\x1f\x80-\xff/ /;
- }
- print("\r");
- print(' ' x $COLUMNS);
- printf("\r\%-${nf}s \%${nm}d \%-${nF}s \%s\n",
- substr($mhfolder, 0, $nf), substr($msgnum, 0, $nm),
- # XXX shouldn't pop, as these are about to be
- # passed to post_store_hooks
- substr($from, 0, $nF),
- substr($subject, 0, $ns));
+ print(scan_line($headers, $mhfolder, $msgnum, $nf, $nm, $nF, $ns),
+ "\n");
}
for my $hook (@post_store_hooks) {
- $hook->(\%batons, \%headers, $mhfolder, $msgnum);
+ $hook->(\%batons, $headers, $mhfolder, $msgnum);
}
- print("\r");
- printf(" \%${len}d SPAM \%${len}d saved \%${len}d/%1d",
- $spam, $saved, $spam + $saved, $msgcount);
+ if ($tty) {
+ printf(" \%${len}d SPAM \%${len}d saved \%${len}d/%1d",
+ $spam, $saved, $spam + $saved, $msgcount);
+ }
defined($maxmsgs) and ($spam + $saved < $maxmsgs or last);
}
return %folders;
}
-sub update_folders {
+sub build_dot_folders {
my $folders = shift;
- my ($fh, $fn);
+ my $fh = shift;
+ my $fn;
- if (not open($fh, "$mh/.folders")) {
- warn("open($mh/.folders): $!");
- return;
- }
-
- while (<$fh>) {
- chomp;
- $folders->{$_}++;
+ if (defined($fh)) {
+ while (<$fh>) {
+ chomp;
+ $folders->{$_}++;
+ }
}
eval { ($fh, $fn) = tempfile("$mh/.folders.XXXXX") };
return;
}
- print($fh "inbox\n");
- for my $folder (sort(keys(%$folders))) {
- $folder eq 'inbox' and next;
+ for my $folder (sort { $folder_sorter->($a,$b) } keys(%$folders)) {
print($fh "$folder\n");
}
rename($fn, "$mh/.folders") or warn("rename($fn, $mh/.folders): $!");
}
+sub create_dot_folders {
+ if (-f "$HOME/.minc") {
+ require "$HOME/.minc";
+ }
+
+ my %folders;
+ open(my $fh, '-|', 'folders', '-fast', '-recur')
+ or die("open(folders|): $!");
+ build_dot_folders(\%folders, $fh);
+ return 0;
+}
+
+sub update_dot_folders {
+ my $folders = shift;
+ my $fh;
+
+ if (not open($fh, "$mh/.folders") and not $!{ENOENT}) {
+ # For ENOENT, we go ahead and create it, else we error and
+ # don't clobber it.
+ warn("open($mh/.folders): $!");
+ return;
+ }
+
+ build_dot_folders($folders, $fh);
+}
+
+# XXX Could use some unification with getfiles...
+sub maildir_spam {
+ my $dir = "$MAILDIR/spam/new";
+
+ if (not chdir($dir)) {
+ $!{ENOENT} or print(STDERR "skipping maildir spam: chdir($dir): $!\n");
+ return;
+ }
+
+ if (not opendir(DIR, '.')) {
+ print(STDERR "skipping maildir spam: opendir($dir): $!\n");
+ return;
+ }
+
+ $! = 0;
+ my @spams = readdir(DIR);
+ if ($! != 0) {
+ print(STDERR "skipping maildir spam: readdir($dir): $!\n");
+ return;
+ }
+
+ closedir(DIR);
+
+ for my $msg (@spams) {
+ ($msg eq '.' or $msg eq '..') and next;
+ my $msgnum = store_message($msg, 'SPAM');
+ # Store the original file name for scan_spam in -n mode.
+ $SPAM{$msgnum} = [undef, $msg];
+ }
+}
+
+sub scan_spam {
+ my ($msgnum, $header, $tuple, $msg);
+
+ # Unlike filter_mail, we don't need to print the folder name.
+ # Calculate how many columns would be allocated to it...
+ my $nf = int($COLUMNS * $SCAN_P_FOLDER);
+ # ...and add that amount to COLUMNS to calculate the number of columns to
+ # allocate to msgnum and from snippet, thus filling the line without
+ # printing the folder name.
+ my $nm = int(($COLUMNS + $nf) * $SCAN_P_MESSAGE);
+ my $nF = int(($COLUMNS + $nf) * $SCAN_P_FROM);
+ my $ns = $COLUMNS - $nm - $nF - 3;
+
+ for $msgnum (sort(keys(%SPAM))) {
+ $tuple = $SPAM{$msgnum};
+ if (defined($tuple->[0])) {
+ # Filed by filter_mail, so we have the header.
+ $header = $tuple->[0];
+ } elsif (defined($tuple->[1])) {
+ # Filed by maildir_spam, so we don't have the header.
+ if ($run) {
+ # The message has been filed, load it from $mh.
+ $msg = "$mh/SPAM/$msgnum";
+ } else {
+ # The message has not been filed, load it from the maildir.
+ # $tuple->[1] is just a basename, not a path; this works
+ # because maildir_spam did chdir(Maildir/spam/new).
+ $msg = $tuple->[1];
+ }
+ $header = get_headers($msg);
+ } else {
+ print(STDERR
+ "BUG: corrupt SPAM tuple, neither element defined",
+ " for message $msgnum\n");
+ next;
+ }
+ print(scan_line($header, '', $msgnum, 0, $nm, $nF, $ns),
+ "\n");
+ }
+}
+
\f
-MAIN: {
+if (!caller()) {
my $st;
if ($dumpfilters) {
exit;
}
+ $rebuild_dot_folders and exit(create_dot_folders);
+
chdir("$MAILDIR/new") or die("chdir($MAILDIR/new): $!");
my %folders = filter_mail(map { $_->[1] }
sort { $a->[0] <=> $b->[0] }
}
getfiles('.'));
- update_folders(\%folders);
+ $run and %folders and update_dot_folders(\%folders);
- @SPAM and (exec('scan', '+SPAM', @SPAM) or die);
+ maildir_spam();
+ scan_spam();
}
+1;
+
\f
__END__
projects / minc / blobdiff