]> Sergey Matveev's repositories - public-inbox.git/blobdiff - lib/PublicInbox/MboxGz.pm
mboxgz: do asynchronous git blob retrievals
[public-inbox.git] / lib / PublicInbox / MboxGz.pm
index 2919ad6a9ce51022408d09dba212f7bd4af91a54..8c9010afb549c5ceebe7f4bba78a13a5d4e13e03 100644 (file)
-# Copyright (C) 2015-2019 all contributors <meta@public-inbox.org>
+# Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 package PublicInbox::MboxGz;
 use strict;
-use warnings;
-use Email::Simple;
+use parent 'PublicInbox::GzipFilter';
+use PublicInbox::Eml;
 use PublicInbox::Hval qw/to_filename/;
 use PublicInbox::Mbox;
-use IO::Compress::Gzip;
+use PublicInbox::GitAsyncCat;
+*msg_hdr = \&PublicInbox::Mbox::msg_hdr;
+*msg_body = \&PublicInbox::Mbox::msg_body;
 
 sub new {
        my ($class, $ctx, $cb) = @_;
-       my $buf = '';
        $ctx->{base_url} = $ctx->{-inbox}->base_url($ctx->{env});
        bless {
-               buf => \$buf,
-               gz => IO::Compress::Gzip->new(\$buf, Time => 0),
+               gz => PublicInbox::GzipFilter::gzip_or_die(),
                cb => $cb,
-               ctx => $ctx,
+               ctx => $ctx
        }, $class;
 }
 
+# this is public-inbox-httpd-specific
+sub mboxgz_blob_cb { # git->cat_async callback
+       my ($bref, $oid, $type, $size, $self) = @_;
+       my $http = $self->{ctx}->{env}->{'psgix.io'} or return; # client abort
+       my $smsg = delete $self->{smsg} or die 'BUG: no smsg';
+       if (!defined($oid)) {
+               # it's possible to have TOCTOU if an admin runs
+               # public-inbox-(edit|purge), just move onto the next message
+               return $http->next_step(\&async_next);
+       } else {
+               $smsg->{blob} eq $oid or die "BUG: $smsg->{blob} != $oid";
+       }
+       $self->zmore(msg_hdr($self->{ctx},
+                               PublicInbox::Eml->new($bref)->header_obj,
+                               $smsg->{mid}));
+
+       # PublicInbox::HTTP::{Chunked,Identity}::write
+       $self->{http_out}->write($self->translate(msg_body($$bref)));
+
+       $http->next_step(\&async_next);
+}
+
+# this is public-inbox-httpd-specific
+sub async_step ($) {
+       my ($self) = @_;
+       if (my $smsg = $self->{smsg} = $self->{cb}->($self->{ctx})) {
+               git_async_cat($self->{ctx}->{-inbox}->git, $smsg->{blob},
+                               \&mboxgz_blob_cb, $self);
+       } elsif (my $out = delete $self->{http_out}) {
+               $out->write($self->zflush);
+               $out->close;
+       }
+}
+
+# called by PublicInbox::DS::write
+sub async_next {
+       my ($http) = @_; # PublicInbox::HTTP
+       async_step($http->{forward});
+}
+
+# called by PublicInbox::HTTP::close, or any other PSGI server
+sub close { !!delete($_[0]->{http_out}) }
+
 sub response {
        my ($class, $ctx, $cb, $fn) = @_;
-       my $body = $class->new($ctx, $cb);
+       my $self = $class->new($ctx, $cb);
        # http://www.iana.org/assignments/media-types/application/gzip
-       my @h = qw(Content-Type application/gzip);
-       if ($fn) {
-               $fn = to_filename($fn);
-               push @h, 'Content-Disposition', "inline; filename=$fn.mbox.gz";
+       $fn = defined($fn) && $fn ne '' ? to_filename($fn) : 'no-subject';
+       my $h = [ qw(Content-Type application/gzip),
+               'Content-Disposition', "inline; filename=$fn.mbox.gz" ];
+       if ($ctx->{env}->{'pi-httpd.async'}) {
+               sub {
+                       my ($wcb) = @_; # -httpd provided write callback
+                       $self->{http_out} = $wcb->([200, $h]);
+                       $self->{ctx}->{env}->{'psgix.io'}->{forward} = $self;
+                       async_step($self); # start stepping
+               };
+       } else { # generic PSGI
+               [ 200, $h, $self ];
        }
-       [ 200, \@h, $body ];
 }
 
-# called by Plack::Util::foreach or similar
+# called by Plack::Util::foreach or similar (generic PSGI)
 sub getline {
        my ($self) = @_;
        my $ctx = $self->{ctx} or return;
-       my $gz = $self->{gz};
-       while (my $smsg = $self->{cb}->()) {
+       while (my $smsg = $self->{cb}->($ctx)) {
                my $mref = $ctx->{-inbox}->msg_by_smsg($smsg) or next;
-               my $h = Email::Simple->new($mref)->header_obj;
-               $gz->write(PublicInbox::Mbox::msg_hdr($ctx, $h, $smsg->{mid}));
-               $gz->write(PublicInbox::Mbox::msg_body($$mref));
-
-               my $bref = $self->{buf};
-               if (length($$bref) >= 8192) {
-                       my $ret = $$bref; # copy :<
-                       ${$self->{buf}} = '';
-                       return $ret;
-               }
-
-               # be fair to other clients on public-inbox-httpd:
-               return '';
+               my $h = PublicInbox::Eml->new($mref)->header_obj;
+               $self->zmore(msg_hdr($ctx, $h, $smsg->{mid}));
+               return $self->translate(msg_body($$mref));
        }
-       delete($self->{gz})->close;
        # signal that we're done and can return undef next call:
        delete $self->{ctx};
-       ${delete $self->{buf}};
+       $self->zflush;
 }
 
-sub close {} # noop
-
 1;