# Copyright (C) 2013-2015 all contributors # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt) # # 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::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 }; # main function sub generate { my ($ctx) = @_; sub { emit_atom($_[0], $ctx) }; } sub generate_thread_atom { my ($ctx) = @_; sub { emit_atom_thread($_[0], $ctx) }; } sub generate_html_index { my ($ctx) = @_; # if the 'r' query parameter is given, it is a legacy permalink # which we must continue supporting: my $qp = $ctx->{qp}; if ($qp && !$qp->{r} && $ctx->{srch}) { return PublicInbox::View::index_topics($ctx); } my $env = $ctx->{env}; my $url = $ctx->{-inbox}->base_url($env) . 'new.html'; my $qs = $env->{QUERY_STRING}; $url .= "?$qs" if $qs ne ''; [302, [ 'Location', $url, 'Content-Type', 'text/plain'], [ "Redirecting to $url\n" ] ]; } sub new_html { my ($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) { return [404, ['Content-Type', 'text/plain'], ["No messages, yet\n"] ]; } $ctx->{-html_tip} = '
'; $ctx->{-upfx} = ''; PublicInbox::WwwStream->response($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 $s; } new_html_footer($ctx, $last); }); } # private subs sub title_tag { my ($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\"" : ''; "
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 $refhex = qr/(?:HEAD|${hex}{4,40})(?:~\d+)?/; my $qp = $ctx->{qp}; # revision ranges may be specified my $range = 'HEAD'; my $r = $qp->{r} if $qp; if ($r && ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o)) { $range = $r; } # 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 --no-notes --no-color --raw -r --abbrev=16 --abbrev-commit/, "--format=%h%x00%ct%x00%an%x00%s%x00", $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); local $/ = "\n"; 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) { $last = 1; last; } } elsif ($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) { $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; '
) . 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!