scripts/slrnspool2maildir: use MHreader and LeiToMail

This contains gmane-specific header munging to unmunge the
things gmane dones to headers.  While we're at it, document the
generic `lei convert' invocation for users who don't need the
gmane-specific header munging.

1 files changed, 47 insertions, 43 deletions

diff --git a/scripts/slrnspool2maildir b/scripts/slrnspool2maildir
index 8e2ba08a..ba0729ec 100755
--- a/scripts/slrnspool2maildir
+++ b/scripts/slrnspool2maildir
@@ -1,51 +1,55 @@
 #!/usr/bin/perl -w
-# Copyright (C) 2013-2021 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
-#
-# One-off script to convert an slrnpull news spool to Maildir
 =begin usage
+One-off script to convert an slrnpull spool from gmane to Maildir
+Note: this contains Gmane-specific header munging to workaround
+the munging done by Gmane.
+
         ./slrnspool2maildir SLRNPULL_ROOT/news/foo/bar /path/to/maildir/
-=cut
-use strict;
-use warnings;
-use Email::Filter;
-use Email::LocalDelivery;
-use File::Glob qw(bsd_glob GLOB_NOSORT);
-sub usage { "Usage:\n".join('',grep(/\t/, `head -n 12 $0`)) }
-my $spool = shift @ARGV or die usage();
-my $dir = shift @ARGV or die usage();
--d $dir or die "$dir is not a directory\n";
-$dir .= '/' unless $dir =~ m!/\z!;
-foreach my $sub (qw(cur new tmp)) {
-        my $nd = "$dir/$sub";
-        -d $nd and next;
-        mkdir $nd or die "mkdir $nd failed: $!\n";
-}
 
-foreach my $n (grep(/\d+\z/, bsd_glob("$spool/*", GLOB_NOSORT))) {
-        if (open my $fh, '<', $n) {
-                my $f = Email::Filter->new(data => do { local $/; <$fh> });
-                my $s = $f->simple;
+A generic replacement w/o Gmane-specific munging could treat
+the slrnpull spool as an MH folder with lei:
 
-                # gmane rewrites Received headers, which increases spamminess
-                # Some older archives set Original-To
-                foreach my $x (qw(Received To)) {
-                        my @h = $s->header("Original-$x");
-                        if (@h) {
-                                $s->header_set($x, @h);
-                                $s->header_set("Original-$x");
-                        }
+        lei convert mh:SLRNPULL_ROOT/news/foo/bar -o /path/to/maildir
+        # (and `lei daemon-kill' if you don't want the daemon to linger)
+=cut
+use v5.12;
+use autodie;
+# warning: unstable internal APIs:
+use PublicInbox::Eml;
+use PublicInbox::LeiToMail;
+use PublicInbox::MHreader;
+use PublicInbox::IO qw(read_all);
+use File::Path qw(make_path);
+use File::Spec ();
+sub usage {
+        open my $fh, '<', __FILE__;
+        ("Usage:\n", grep { /^=begin usage/../^=cut/ and !/^=/m } <$fh>);
+}
+my $spool = shift @ARGV or die usage();
+my $dst = shift @ARGV or die usage();
+$dst .= '/' unless $dst =~ m!/\z!;
+File::Path::make_path(map { $dst.$_ } qw(tmp new cur));
+$dst = File::Spec->rel2abs($dst).'/';
+opendir my $cwdfh, '.';
+my $mhr = PublicInbox::MHreader->new($spool, $cwdfh);
+my $smsg;
+$mhr->mh_each_eml(sub {
+        my ($d, $n, $kw, $eml) = @_;
+        # gmane rewrites Received headers, which increases spamminess
+        # Some older archives set Original-To
+        for my $x (qw(Received To)) {
+                my @h = $eml->header_raw("Original-$x");
+                if (@h) {
+                        $eml->header_set($x, @h);
+                        $eml->header_set("Original-$x");
                 }
-
-                # triggers for the SA HEADER_SPAM rule
-                foreach my $drop (qw(Approved)) { $s->header_set($drop) }
-
-                # appears to be an old gmane bug:
-                $s->header_set('connect()');
-
-                $f->exit(0);
-                $f->accept($dir);
-        } else {
-                warn "Failed to open $n: $!\n";
         }
-}
+        # `Approved' triggers the SA HEADER_SPAM rule
+        # `connect()' appears to be an old gmane bug:
+        $eml->header_set($_) for ('Approved', 'connect()');
+        my $buf = $eml->as_string;
+        $smsg->{blob} = $n;
+        PublicInbox::LeiToMail::_buf2maildir($dst, \$buf, $smsg, 'new/');
+});