X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FLeiToMail.pm;h=d3253d9b01139bf6a2ea7b29cd0a5619485d2f41;hb=47d4e53734820b4e;hp=64061788eae05c219f71de5d5bea974819b2baee;hpb=d98abf13946beaf685804e4ec2d79019129415b9;p=public-inbox.git diff --git a/lib/PublicInbox/LeiToMail.pm b/lib/PublicInbox/LeiToMail.pm index 64061788..d3253d9b 100644 --- a/lib/PublicInbox/LeiToMail.pm +++ b/lib/PublicInbox/LeiToMail.pm @@ -9,7 +9,6 @@ use parent qw(PublicInbox::IPC); use PublicInbox::Eml; use PublicInbox::ProcessPipe; use PublicInbox::Spawn qw(spawn); -use PublicInbox::PktOp qw(pkt_do); use Symbol qw(gensym); use IO::Handle; # ->autoflush use Fcntl qw(SEEK_SET SEEK_END O_CREAT O_EXCL O_WRONLY); @@ -41,11 +40,14 @@ sub _mbox_hdr_buf ($$$) { warn "# keyword `$k' not supported for mbox\n"; } } - # Messages are always 'O' (non-\Recent in IMAP), it saves - # MUAs the trouble of rewriting the mbox if no other - # changes are made. We put 'O' at the end (e.g. "Status: RO") - # to match mutt(1) output. - $eml->header_set('Status', join('', sort(@{$hdr{Status}})). 'O'); + # When writing to empty mboxes, messages are always 'O' + # (not-\Recent in IMAP), it saves MUAs the trouble of + # rewriting the mbox if no other changes are made. + # We put 'O' at the end (e.g. "Status: RO") to match mutt(1) output. + # We only set smsg->{-recent} if augmenting existing stores. + my $status = join('', sort(@{$hdr{Status}})); + $status .= 'O' unless $smsg->{-recent}; + $eml->header_set('Status', $status) if $status; if (my $chars = delete $hdr{'X-Status'}) { $eml->header_set('X-Status', join('', sort(@$chars))); } @@ -107,39 +109,38 @@ sub _mboxcl_common ($$$) { $$buf .= 'Content-Length: '.length($$bdy).$crlf. 'Lines: '.$lines.$crlf.$crlf; substr($$bdy, 0, 0, $$buf); # prepend header - $_[0] = $bdy; + $$bdy .= $crlf; + $bdy; } # mboxcl still escapes "From " lines sub eml2mboxcl { my ($eml, $smsg) = @_; my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg); - my $crlf = $eml->{crlf}; - if (my $bdy = delete $eml->{bdy}) { - $$bdy =~ s/^From />From /gm; - _mboxcl_common($buf, $bdy, $crlf); - } - $$buf .= $crlf; - $buf; + my $bdy = delete($eml->{bdy}) // \(my $empty = ''); + $$bdy =~ s/^From />From /gm; + _mboxcl_common($buf, $bdy, $eml->{crlf}); } # mboxcl2 has no "From " escaping sub eml2mboxcl2 { my ($eml, $smsg) = @_; my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg); - my $crlf = $eml->{crlf}; - if (my $bdy = delete $eml->{bdy}) { - _mboxcl_common($buf, $bdy, $crlf); - } - $$buf .= $crlf; - $buf; + my $bdy = delete($eml->{bdy}) // \(my $empty = ''); + _mboxcl_common($buf, $bdy, $eml->{crlf}); } sub git_to_mail { # git->cat_async callback my ($bref, $oid, $type, $size, $arg) = @_; + my ($write_cb, $smsg) = @$arg; + if ($type eq 'missing' && $smsg->{-lms_ro}) { + if ($bref = $smsg->{-lms_ro}->local_blob($oid, 1)) { + $type = 'blob'; + $size = length($$bref); + } + } return warn("W: $oid is $type (!= blob)\n") if $type ne 'blob'; return warn("E: $oid is empty\n") unless $size; - my ($write_cb, $smsg) = @$arg; die "BUG: expected=$smsg->{blob} got=$oid" if $smsg->{blob} ne $oid; $write_cb->($bref, $smsg); } @@ -190,11 +191,13 @@ sub _mbox_write_cb ($$) { my $dedupe = $lei->{dedupe}; $dedupe->prepare_dedupe; my $lse = $lei->{lse}; # may be undef + my $set_recent = $dedupe->has_entries; sub { # for git_to_mail my ($buf, $smsg, $eml) = @_; $eml //= PublicInbox::Eml->new($buf); return if $dedupe->is_dup($eml, $smsg); $lse->xsmsg_vmd($smsg) if $lse; + $smsg->{-recent} = 1 if $set_recent; $buf = $eml2mbox->($eml, $smsg); if ($atomic_append) { atomic_append($lei, $buf); @@ -237,10 +240,14 @@ sub _rand () { sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq); } -sub _buf2maildir { - my ($dst, $buf, $smsg) = @_; +sub kw2suffix ($;@) { + my $kw = shift; + join('', sort(map { $kw2char{$_} // () } @$kw, @_)); +} + +sub _buf2maildir ($$$$) { + my ($dst, $buf, $smsg, $dir) = @_; my $kw = $smsg->{kw} // []; - my $sfx = join('', sort(map { $kw2char{$_} // () } @$kw)); my $rand = ''; # chosen by die roll :P my ($tmp, $fh, $base, $ok); my $common = $smsg->{blob} // _rand; @@ -250,14 +257,10 @@ sub _buf2maildir { } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) && $!{EEXIST} && ($rand = _rand.',')); if ($ok && print $fh $$buf and close($fh)) { - # ignore new/ and write only to cur/, otherwise MUAs - # with R/W access to the Maildir will end up doing - # a mass rename which can take a while with thousands - # of messages. - $dst .= 'cur/'; + $dst .= $dir; # 'new/' or 'cur/' $rand = ''; do { - $base = $rand.$common.':2,'.$sfx + $base = $rand.$common.':2,'.kw2suffix($kw); } while (!($ok = link($tmp, $dst.$base)) && $!{EEXIST} && ($rand = _rand.',')); die "link($tmp, $dst$base): $!" unless $ok; @@ -278,7 +281,12 @@ sub _maildir_write_cb ($$) { my $dst = $lei->{ovv}->{dst}; my $lse = $lei->{lse}; # may be undef my $sto = $lei->{opt}->{'mail-sync'} ? $lei->{sto} : undef; - my $out = $sto ? 'maildir:'.$lei->rel2abs($dst) : undef; + my $out = $sto ? 'maildir:'.$lei->abs_path($dst) : undef; + + # Favor cur/ and only write to new/ when augmenting. This + # saves MUAs from having to do a mass rename when the initial + # search result set is huge. + my $dir = $dedupe && $dedupe->has_entries ? 'new/' : 'cur/'; sub { # for git_to_mail my ($bref, $smsg, $eml) = @_; $dst // return $lei->fail; # dst may be undef-ed in last run @@ -286,7 +294,8 @@ sub _maildir_write_cb ($$) { PublicInbox::Eml->new($$bref), $smsg); $lse->xsmsg_vmd($smsg) if $lse; - my $n = _buf2maildir($dst, $bref // \($eml->as_string), $smsg); + my $n = _buf2maildir($dst, $bref // \($eml->as_string), + $smsg, $dir); $sto->ipc_do('set_sync_info', $smsg->{blob}, $out, $n) if $sto; ++$lei->{-nr_write}; } @@ -297,11 +306,12 @@ sub _imap_write_cb ($$) { my $dedupe = $lei->{dedupe}; $dedupe->prepare_dedupe if $dedupe; my $append = $lei->{net}->can('imap_append'); - my $mic = $lei->{net}->mic_get($self->{uri}); - my $folder = $self->{uri}->mailbox; + my $uri = $self->{uri}; + my $mic = $lei->{net}->mic_get($uri); + my $folder = $uri->mailbox; + $uri->uidvalidity($mic->uidvalidity($folder)); my $lse = $lei->{lse}; # may be undef my $sto = $lei->{opt}->{'mail-sync'} ? $lei->{sto} : undef; - my $out = $lei->{ovv}->{dst}; sub { # for git_to_mail my ($bref, $smsg, $eml) = @_; $mic // return $lei->fail; # mic may be undef-ed in last run @@ -317,7 +327,7 @@ sub _imap_write_cb ($$) { # imap_append returns UID if IMAP server has UIDPLUS extension ($sto && $uid =~ /\A[0-9]+\z/) and $sto->ipc_do('set_sync_info', - $smsg->{blob}, $out, $uid + 0); + $smsg->{blob}, $$uri, $uid + 0); ++$lei->{-nr_write}; } } @@ -335,16 +345,29 @@ sub _text_write_cb ($$) { sub { # for git_to_mail my ($bref, $smsg, $eml) = @_; $lse->xsmsg_vmd($smsg) if $lse; - $eml //= PublicInbox::Eml->new($bref); # copy bref + $eml //= PublicInbox::Eml->new($bref); return if $dedupe && $dedupe->is_dup($eml, $smsg); my $lk = $ovv->lock_for_scope; $lei->out(${$lvt->eml_to_text($smsg, $eml)}, "\n"); } } +sub _v2_write_cb ($$) { + my ($self, $lei) = @_; + my $dedupe = $lei->{dedupe}; + $dedupe->prepare_dedupe if $dedupe; + sub { # for git_to_mail + my ($bref, $smsg, $eml) = @_; + $eml //= PublicInbox::Eml->new($bref); + return if $dedupe && $dedupe->is_dup($eml, $smsg); + $lei->{v2w}->ipc_do('add', $eml); # V2Writable->add + ++$lei->{-nr_write}; + } +} + sub write_cb { # returns a callback for git_to_mail my ($self, $lei) = @_; - # _mbox_write_cb, _maildir_write_cb or _imap_write_cb + # _mbox_write_cb, _maildir_write_cb, _imap_write_cb, _v2_write_cb my $m = "_$self->{base_type}_write_cb"; $self->$m($lei); } @@ -354,23 +377,31 @@ sub new { my $fmt = $lei->{ovv}->{fmt}; my $dst = $lei->{ovv}->{dst}; my $self = bless {}, $cls; + my @conflict; if ($fmt eq 'maildir') { require PublicInbox::MdirReader; $self->{base_type} = 'maildir'; -e $dst && !-d _ and die "$dst exists and is not a directory\n"; $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/'; + $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q'; } elsif (substr($fmt, 0, 4) eq 'mbox') { require PublicInbox::MboxReader; $self->can("eml2$fmt") or die "bad mbox format: $fmt\n"; $self->{base_type} = 'mbox'; + if ($lei->{cmd} eq 'q' && + (($lei->path_to_fd($dst) // -1) < 0) && + (-f $dst || !-e _)) { + $lei->{opt}->{save} //= \1; + } } elsif ($fmt =~ /\Aimaps?\z/) { require PublicInbox::NetWriter; require PublicInbox::URIimap; - my $net = PublicInbox::NetWriter->new; + # {net} may exist from "lei up" for auth + my $net = $lei->{net} // PublicInbox::NetWriter->new; $net->{quiet} = $lei->{opt}->{quiet}; my $uri = PublicInbox::URIimap->new($dst)->canonical; - $net->add_url($uri); + $net->add_url($$uri); my $err = $net->errors($lei); return $lei->fail($err) if $err; $uri->mailbox or return $lei->fail("No mailbox: $dst"); @@ -378,10 +409,19 @@ sub new { $dst = $lei->{ovv}->{dst} = $$uri; # canonicalized $lei->{net} = $net; $self->{base_type} = 'imap'; - } elsif ($fmt eq 'text') { + $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q'; + } elsif ($fmt eq 'text' || $fmt eq 'reply') { require PublicInbox::LeiViewText; - $lei->{lvt} = PublicInbox::LeiViewText->new($lei); + $lei->{lvt} = PublicInbox::LeiViewText->new($lei, $fmt); $self->{base_type} = 'text'; + @conflict = qw(mua save); + } elsif ($fmt eq 'v2') { + die "--dedupe=oid and v2 are incompatible\n" if + ($lei->{opt}->{dedupe}//'') eq 'oid'; + $self->{base_type} = 'v2'; + $lei->{opt}->{save} = \1; + $dst = $lei->{ovv}->{dst} = $lei->abs_path($dst); + @conflict = qw(mua sort); } else { die "bad mail --format=$fmt\n"; } @@ -389,12 +429,8 @@ sub new { (-d $dst || (-e _ && !-w _)) and die "$dst exists and is not a writable file\n"; } - if ($self->{base_type} eq 'text') { - my @err = map { - defined($lei->{opt}->{$_}) ? "--$_" : (); - } (qw(mua save)); - die "@err incompatible with $fmt\n" if @err; - } + my @err = map { defined($lei->{opt}->{$_}) ? "--$_" : () } @conflict; + die "@err incompatible with $fmt\n" if @err; $self->{dst} = $dst; $lei->{dedupe} = $lei->{lss} // do { my $dd_cls = 'PublicInbox::'. @@ -420,9 +456,19 @@ sub _pre_augment_maildir { open $self->{poke_dh}, '<', "${dst}cur" or die "open ${dst}cur: $!"; } +sub clobber_dst_prepare ($;$) { + my ($lei, $f) = @_; + if (my $lms = defined($f) ? $lei->lms : undef) { + $lms->lms_write_prepare; + $lms->forget_folders($f); + } + my $dedupe = $lei->{dedupe} or return; + $dedupe->reset_dedupe if $dedupe->can('reset_dedupe'); +} + sub _do_augment_maildir { my ($self, $lei) = @_; - return if ($lei->{opt}->{save} // 0) < 0; + return if $lei->{cmd} eq 'up'; my $dst = $lei->{ovv}->{dst}; my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef; my $mdr = PublicInbox::MdirReader->new; @@ -434,18 +480,20 @@ sub _do_augment_maildir { $dedupe->pause_dedupe; } } elsif ($lse) { + clobber_dst_prepare($lei, "maildir:$dst"); $mdr->{shard_info} = $self->{shard_info}; $mdr->maildir_each_eml($dst, \&_md_update, $lei, $lse, 1); } else {# clobber existing Maildir + clobber_dst_prepare($lei, "maildir:$dst"); $mdr->maildir_each_file($dst, \&_unlink); } } sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb - my ($url, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_; + my ($uri, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_; update_kw_maybe($lei, $lse, $eml, $kw); if ($delete_mic) { - $lei->{net}->imap_delete_1($url, $uid, $delete_mic); + $lei->{net}->imap_delete_1($uri, $uid, $delete_mic); } else { _augment($eml, $lei); } @@ -453,7 +501,7 @@ sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb sub _do_augment_imap { my ($self, $lei) = @_; - return if ($lei->{opt}->{save} // 0) < 0; + return if $lei->{cmd} eq 'up'; my $net = $lei->{net}; my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef; if ($lei->{opt}->{augment}) { @@ -465,11 +513,13 @@ sub _do_augment_imap { } } elsif ($lse) { my $delete_mic; + clobber_dst_prepare($lei, "$self->{uri}"); $net->imap_each($self->{uri}, \&_imap_augment_or_delete, $lei, $lse, \$delete_mic); $delete_mic->expunge if $delete_mic; } elsif (!$self->{-wq_worker_nr}) { # undef or 0 # clobber existing IMAP folder + clobber_dst_prepare($lei, "$self->{uri}"); $net->imap_delete_all($self->{uri}); } } @@ -529,6 +579,8 @@ sub _pre_augment_mbox { if $imp_before && !ref($imp_before); die "--augment specified but $dst is not seekable\n" if $lei->{opt}->{augment}; + die "cannot --save with unseekable $dst\n" if + $lei->{dedupe} && $lei->{dedupe}->can('reset_dedupe'); } if ($self->{zsfx} = PublicInbox::MboxReader::zsfx($dst)) { pipe(my ($r, $w)) or die "pipe: $!"; @@ -547,10 +599,10 @@ sub _do_augment_mbox { my ($self, $lei) = @_; return unless $self->{seekable}; my $opt = $lei->{opt}; - return if ($opt->{save} // 0) < 0; + return if $lei->{cmd} eq 'up'; my $out = $lei->{1}; my ($fmt, $dst) = @{$lei->{ovv}}{qw(fmt dst)}; - return unless -s $out; + return clobber_dst_prepare($lei) unless -s $out; unless ($opt->{augment} || $opt->{'import-before'}) { truncate($out, 0) or die "truncate($dst): $!"; return; @@ -565,6 +617,8 @@ sub _do_augment_mbox { if ($opt->{augment}) { $dedupe = $lei->{dedupe}; $dedupe->prepare_dedupe if $dedupe; + } else { + clobber_dst_prepare($lei); } if ($opt->{'import-before'}) { # the default my $lse = $lei->{lse}; @@ -581,9 +635,43 @@ sub _do_augment_mbox { $dedupe->pause_dedupe if $dedupe; } +sub _pre_augment_v2 { + my ($self, $lei) = @_; + my $dir = $self->{dst}; + require PublicInbox::InboxWritable; + my ($ibx, @creat); + if (-d $dir) { + my $opt = { -min_inbox_version => 2 }; + require PublicInbox::Admin; + my @ibx = PublicInbox::Admin::resolve_inboxes([ $dir ], $opt); + $ibx = $ibx[0] or die "$dir is not a v2 inbox\n"; + } else { + $creat[0] = {}; + $ibx = PublicInbox::Inbox->new({ + name => 'lei-result', # XXX configurable + inboxdir => $dir, + version => 2, + address => [ 'lei@example.com' ], + }); + } + PublicInbox::InboxWritable->new($ibx, @creat); + $ibx->init_inbox if @creat; + my $v2w = $lei->{v2w} = $ibx->importer; + $v2w->ipc_lock_init("$dir/ipc.lock"); + $v2w->ipc_worker_spawn("lei/v2w $dir", $lei->oldset, { lei => $lei }); + return if !$lei->{opt}->{shared}; + my $d = "$lei->{ale}->{git}->{git_dir}/objects"; + my $al = "$dir/git/0.git/objects/info/alternates"; + open my $fh, '+>>', $al or die "open($al): $!"; + seek($fh, 0, SEEK_SET) or die "seek($al): $!"; + grep(/\A\Q$d\E\n/, <$fh>) and return; + print $fh "$d\n" or die "print($al): $!"; + close $fh or die "close($al): $!"; +} + sub pre_augment { # fast (1 disk seek), runs in same process as post_augment my ($self, $lei) = @_; - # _pre_augment_maildir, _pre_augment_mbox + # _pre_augment_maildir, _pre_augment_mbox, _pre_augment_v2 my $m = $self->can("_pre_augment_$self->{base_type}") or return; $m->($self, $lei); } @@ -610,7 +698,7 @@ sub do_post_auth { my ($self) = @_; my $lei = $self->{lei}; # lei_xsearch can start as soon as all l2m workers get here - pkt_do($lei->{pkt_op_p}, 'incr_start_query') or + $lei->{pkt_op_p}->pkt_do('incr_start_query') or die "incr_start_query: $!"; my $aug; if (lock_free($self)) { # all workers do_augment @@ -629,7 +717,7 @@ sub do_post_auth { local $0 = 'do_augment'; eval { do_augment($self, $lei) }; $lei->fail($@) if $@; - pkt_do($lei->{pkt_op_p}, $aug) == 1 or + $lei->{pkt_op_p}->pkt_do($aug) == 1 or die "do_post_augment trigger: $!"; } # done augmenting, connect the compressor pipe for each worker @@ -637,13 +725,25 @@ sub do_post_auth { $lei->{1} = $zpipe->[1]; close $zpipe->[0]; } + my $au_peers = delete $self->{au_peers}; + if ($au_peers) { # wait for peer l2m to finish augmenting: + $au_peers->[1] = undef; + sysread($au_peers->[0], my $barrier1, 1); + } $self->{wcb} = $self->write_cb($lei); + if ($au_peers) { # wait for peer l2m to set write_cb + $au_peers->[3] = undef; + sysread($au_peers->[2], my $barrier2, 1); + } } sub ipc_atfork_child { my ($self) = @_; my $lei = $self->{lei}; $lei->_lei_atfork_child; + if (my $lse = $lei->{lse}) { + $self->{-lms_ro} = $lse->{-lms_ro} //= $lse->lms; + } $lei->{auth}->do_auth_atfork($self) if $lei->{auth}; $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb(); $self->SUPER::ipc_atfork_child; @@ -665,19 +765,19 @@ sub poke_dst { sub write_mail { # via ->wq_io_do my ($self, $smsg, $eml) = @_; return $self->{wcb}->(undef, $smsg, $eml) if $eml; + $smsg->{-lms_ro} = $self->{-lms_ro}; $self->{lei}->{ale}->git->cat_async($smsg->{blob}, \&git_to_mail, [$self->{wcb}, $smsg]); } sub wq_atexit_child { my ($self) = @_; - delete $self->{wcb}; my $lei = $self->{lei}; + delete $self->{wcb}; $lei->{ale}->git->async_wait_all; my $nr = delete($lei->{-nr_write}) or return; return if $lei->{early_mua} || !$lei->{-progress} || !$lei->{pkt_op_p}; - require PublicInbox::PktOp; - PublicInbox::PktOp::pkt_do($lei->{pkt_op_p}, 'l2m_progress', $nr); + $lei->{pkt_op_p}->pkt_do('l2m_progress', $nr); } # called in top-level lei-daemon when LeiAuth is done @@ -687,6 +787,4 @@ sub net_merge_all_done { $self->wq_close(1); } -no warnings 'once'; # the following works even when LeiAuth is lazy-loaded -*net_merge_all = \&PublicInbox::LeiAuth::net_merge_all; 1;