X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=d7df07ee9c4577a1d15fe5410f68e6bdd68ee3e2;hb=be4d92e74dd0b1c8de85bc08c3c522af2d8d8656;hp=a697a4326fcaefcb64f2be0a17432f71cd1a2163;hpb=2ce1cf89680840c86af136b69372adb3881d7aec;p=public-inbox.git diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm index a697a432..d7df07ee 100644 --- a/lib/PublicInbox/Feed.pm +++ b/lib/PublicInbox/Feed.pm @@ -1,31 +1,44 @@ -# Copyright (C) 2013-2015 all contributors -# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt) +# Copyright (C) 2013-2018 all contributors +# License: AGPL-3.0+ # # Used for generating Atom feeds for web-accessible mailing list archives. package PublicInbox::Feed; use strict; use warnings; -use Email::MIME; -use Date::Parse qw(strptime); -use PublicInbox::Hval qw/ascii_html/; +use PublicInbox::MIME; use PublicInbox::View; -use PublicInbox::MID qw/mid_clean mid2path/; -use PublicInbox::Address; -use POSIX qw/strftime/; -use constant { - DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # Atom standard - MAX_PER_PAGE => 25, # this needs to be tunable -}; +use PublicInbox::WwwAtomStream; +use PublicInbox::SearchMsg; # this loads w/o Search::Xapian # main function sub generate { my ($ctx) = @_; - sub { emit_atom($_[0], $ctx) }; + my $msgs = recent_msgs($ctx); + return _no_thread() unless @$msgs; + + my $ibx = $ctx->{-inbox}; + PublicInbox::WwwAtomStream->response($ctx, 200, sub { + while (my $smsg = shift @$msgs) { + $ibx->smsg_mime($smsg) and return $smsg; + } + }); } sub generate_thread_atom { my ($ctx) = @_; - sub { emit_atom_thread($_[0], $ctx) }; + my $mid = $ctx->{mid}; + my $msgs = $ctx->{srch}->get_thread($mid); + return _no_thread() unless @$msgs; + + my $ibx = $ctx->{-inbox}; + my $html_url = $ibx->base_url($ctx->{env}); + $html_url .= PublicInbox::Hval->new_msgid($mid)->{href}; + $ctx->{-html_url} = $html_url; + PublicInbox::WwwAtomStream->response($ctx, 200, sub { + while (my $smsg = shift @$msgs) { + $ibx->smsg_mime($smsg) and return $smsg; + } + }); } sub generate_html_index { @@ -47,132 +60,48 @@ sub generate_html_index { sub new_html { my ($ctx) = @_; - my @paths; - my (undef, $last) = each_recent_blob($ctx, sub { - my ($path, $commit, $ts, $u, $subj) = @_; - $ctx->{first} ||= $commit; - push @paths, $path; - }); - if (!@paths) { + my $msgs = recent_msgs($ctx); + if (!@$msgs) { return [404, ['Content-Type', 'text/plain'], ["No messages, yet\n"] ]; } $ctx->{-html_tip} = '
';
 	$ctx->{-upfx} = '';
+	$ctx->{-hr} = 1;
+	my $ibx = $ctx->{-inbox};
 	PublicInbox::WwwStream->response($ctx, 200, sub {
-		while (my $path = shift @paths) {
-			my $m = do_cat_mail($ctx->{-inbox}, $path) or next;
-			my $more = scalar @paths;
-			my $s = PublicInbox::View::index_entry($m, $ctx, $more);
-			$s .= '
' unless $more; - return $s; + while (my $smsg = shift @$msgs) { + my $m = $ibx->smsg_mime($smsg) or next; + my $more = scalar @$msgs; + return PublicInbox::View::index_entry($m, $ctx, $more); } - new_html_footer($ctx, $last); + PublicInbox::View::pagination_footer($ctx, './new.html'); }); } # private subs -sub title_tag { - my ($title) = @_; - $title =~ tr/\t\n / /s; # squeeze spaces - # try to avoid the type attribute in title: - $title = ascii_html($title); - my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : ''; - "$title"; -} - -sub atom_header { - my ($feed_opts, $title) = @_; - - $title = title_tag($feed_opts->{description}) unless (defined $title); - - qq(\n) . - qq{} . - qq{$title} . - qq() . - qq({atomurl}"/>) . - qq(mailto:$feed_opts->{id_addr}); -} - -sub emit_atom { - my ($cb, $ctx) = @_; - my $feed_opts = get_feedopts($ctx); - my $fh = $cb->([ 200, ['Content-Type' => 'application/atom+xml']]); - my $max = $ctx->{max} || MAX_PER_PAGE; - my $x = atom_header($feed_opts); - my $ibx = $ctx->{-inbox}; - each_recent_blob($ctx, sub { - my ($path, undef, $ts) = @_; - if (defined $x) { - $fh->write($x . feed_updated(undef, $ts)); - $x = undef; - } - my $s = feed_entry($feed_opts, $path, $ibx) or return 0; - $fh->write($s); - 1; - }); - end_feed($fh); -} - -sub _no_thread { - my ($cb) = @_; - $cb->([404, ['Content-Type', 'text/plain'], - ["No feed found for thread\n"]]); +sub _no_thread () { + [404, ['Content-Type', 'text/plain'], ["No feed found for thread\n"]]; } -sub end_feed { - my ($fh) = @_; - $fh->write(''); - $fh->close; -} - -sub emit_atom_thread { - my ($cb, $ctx) = @_; - my $mid = $ctx->{mid}; - my $res = $ctx->{srch}->get_thread($mid); - return _no_thread($cb) unless $res->{total}; - my $feed_opts = get_feedopts($ctx); - my $fh = $cb->([200, ['Content-Type' => 'application/atom+xml']]); +sub recent_msgs { + my ($ctx) = @_; my $ibx = $ctx->{-inbox}; - my $html_url = $ibx->base_url($ctx->{env}); - $html_url .= PublicInbox::Hval->new_msgid($mid)->as_href; - - $feed_opts->{url} = $html_url; - $feed_opts->{emit_header} = 1; - - foreach my $msg (@{$res->{msgs}}) { - my $s = feed_entry($feed_opts, mid2path($msg->mid), $ibx); - $fh->write($s) if defined $s; - } - end_feed($fh); -} - -sub new_html_footer { - my ($ctx, $last) = @_; - my $qp = delete $ctx->{qp} or return; - my $old_r = $qp->{r}; - my $latest = ''; - my $next = ' '; - - if ($last) { - $next = qq!next!; + my $max = $ibx->{feedmax}; + my $qp = $ctx->{qp}; + my $v = $ibx->{version} || 1; + if ($v > 2) { + die "BUG: unsupported inbox version: $v\n"; } - if ($old_r) { - $latest = qq! latest!; + if (my $srch = $ibx->search) { + return PublicInbox::View::paginate_recent($ctx, $max); } - "
page: $next$latest
"; -} -sub each_recent_blob { - my ($ctx, $cb) = @_; - my $max = $ctx->{max} || MAX_PER_PAGE; my $hex = '[a-f0-9]'; - my $addmsg = qr!^:000000 100644 \S+ \S+ A\t(${hex}{2}/${hex}{38})$!; - my $delmsg = qr!^:100644 000000 \S+ \S+ D\t(${hex}{2}/${hex}{38})$!; + my $addmsg = qr!^:000000 100644 \S+ (\S+) A\t${hex}{2}/${hex}{38}$!; + my $delmsg = qr!^:100644 000000 (\S+) \S+ D\t(${hex}{2}/${hex}{38})$!; my $refhex = qr/(?:HEAD|${hex}{4,40})(?:~\d+)?/; - my $qp = $ctx->{qp}; # revision ranges may be specified my $range = 'HEAD'; @@ -184,130 +113,41 @@ sub each_recent_blob { # get recent messages # we could use git log -z, but, we already know ssoma will not # leave us with filenames with spaces in them.. - my $log = $ctx->{-inbox}->git->popen(qw/log + my $log = $ibx->git->popen(qw/log --no-notes --no-color --raw -r - --abbrev=16 --abbrev-commit/, - "--format=%h%x00%ct%x00%an%x00%s%x00", - $range); + --no-abbrev --abbrev-commit/, + "--format=%h", $range); my %deleted; # only an optimization at this point my $last; - my $nr = 0; - my ($cur_commit, $first_commit, $last_commit); - my ($ts, $subj, $u); + my $last_commit; local $/ = "\n"; + my @oids; while (defined(my $line = <$log>)) { if ($line =~ /$addmsg/o) { my $add = $1; next if $deleted{$add}; # optimization-only - $nr += $cb->($add, $cur_commit, $ts, $u, $subj); - if ($nr >= $max) { + push @oids, $add; + if (scalar(@oids) >= $max) { $last = 1; last; } } elsif ($line =~ /$delmsg/o) { $deleted{$1} = 1; - } elsif ($line =~ /^${hex}{7,40}/o) { - ($cur_commit, $ts, $u, $subj) = split("\0", $line); - unless (defined $first_commit) { - $first_commit = $cur_commit; - } } } if ($last) { local $/ = "\n"; while (my $line = <$log>) { - if ($line =~ /^(${hex}{7,40})/o) { + if ($line =~ /^(${hex}{7,40})/) { $last_commit = $1; last; } } } - # for pagination - ($first_commit, $last_commit); -} - -# private functions below -sub get_feedopts { - my ($ctx) = @_; - my $inbox = $ctx->{inbox}; - my $obj = $ctx->{-inbox}; - my %rv = ( description => $obj->description ); - - $rv{address} = $obj->{address}; - $rv{id_addr} = $obj->{-primary_address}; - my $url_base = $obj->base_url($ctx->{env}); - if (my $mid = $ctx->{mid}) { # per-thread feed: - $rv{atomurl} = "$url_base$mid/t.atom"; - } else { - $rv{atomurl} = $url_base."new.atom"; - } - $rv{url} ||= $url_base; - $rv{midurl} = $url_base; - - \%rv; -} - -sub feed_updated { - my ($date, $ts) = @_; - my @t = eval { strptime($date) } if defined $date; - @t = gmtime($ts || time) unless scalar @t; - - '' . strftime(DATEFMT, @t) . ''; -} - -# returns undef or string -sub feed_entry { - my ($feed_opts, $add, $ibx) = @_; - - my $mime = do_cat_mail($ibx, $add) or return; - my $url = $feed_opts->{url}; - my $midurl = $feed_opts->{midurl}; - - my $header_obj = $mime->header_obj; - my $mid = $header_obj->header_raw('Message-ID'); - defined $mid or return; - $mid = PublicInbox::Hval->new_msgid($mid); - my $href = $midurl . $mid->as_href . '/'; - - my $date = $header_obj->header('Date'); - my $updated = feed_updated($date); - - my $title = $header_obj->header('Subject'); - defined $title or return; - $title = title_tag($title); - - my $from = $header_obj->header('From') or return; - my ($email) = PublicInbox::Address::emails($from); - my $name = join(', ',PublicInbox::Address::names($from)); - $name = ascii_html($name); - $email = ascii_html($email); - - my $s = ''; - if (delete $feed_opts->{emit_header}) { - $s .= atom_header($feed_opts, $title) . $updated; - } - $s .= "$name$email" . - "$title$updated" . - qq{} . - qq{} . - qq() . - PublicInbox::View::multipart_text_as_html($mime, $href) . - ''; - - $add =~ tr!/!!d; - my $h = '[a-f0-9]'; - my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o); - my $id = 'urn:uuid:' . join('-', @uuid5); - $s .= qq!!. - "$id"; -} - -sub do_cat_mail { - my ($ibx, $path) = @_; - my $mime = eval { $ibx->msg_by_path($path) } or return; - Email::MIME->new($mime); + $ctx->{next_page} = "r=$last_commit" if $last_commit; + [ map { bless {blob => $_ }, 'PublicInbox::SearchMsg' } @oids ]; } 1;