# 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::Address; use Email::MIME; use Date::Parse qw(strptime); use PublicInbox::Hval; use PublicInbox::Git; use PublicInbox::View; use PublicInbox::MID qw/mid_clean mid2path/; 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 Encode qw/find_encoding/; my $enc_utf8 = find_encoding('UTF-8'); # main function sub generate { my ($ctx) = @_; sub { emit_atom($_[0], $ctx) }; } sub generate_thread_atom { my ($ctx) = @_; sub { emit_atom_thread($_[0], $ctx) }; } sub generate_html_index { my ($ctx) = @_; sub { emit_html_index($_[0], $ctx) }; } # private subs sub title_tag { my ($title) = @_; # try to avoid the type attribute in title: $title = PublicInbox::Hval->new_oneline($title)->as_html; my $type = index($title, '&') >= 0 ? "\ntype=\"html\"" : ''; "
$footer"; } $fh->write("$footer"); $fh->close; } sub emit_index_nosrch { my ($ctx, $state, $fh) = @_; my $git = $ctx->{git} ||= PublicInbox::Git->new($ctx->{git_dir}); my (undef, $last) = each_recent_blob($ctx, sub { my ($path, $commit, $ts, $u, $subj) = @_; $state->{first} ||= $commit; my $mime = do_cat_mail($git, $path) or return 0; PublicInbox::View::index_entry($fh, $mime, 0, $state); 1; }); Email::Address->purge_cache; $last; } sub nav_footer { my ($cgi, $last, $feed_opts, $state, $param) = @_; $cgi or return ''; my $old_r = $cgi->param($param); my $head = ' '; my $next = ' '; my $first = $state->{first}; my $anchor = $state->{anchor_idx}; if ($last) { $next = qq!next!; } if ($old_r) { $head = $cgi->path_info; $head = qq!head!; } my $atom = "{atomurl}\">Atom"; "page: $next $head $atom"; } sub each_recent_blob { my ($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 $cgi = $ctx->{cgi}; # revision ranges may be specified my $range = 'HEAD'; my $r = $cgi->param('r') if $cgi; if ($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 $git = $ctx->{git} ||= PublicInbox::Git->new($ctx->{git_dir}); my $log = $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); 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) { while (my $line = <$log>) { if ($line =~ /^(${hex}{7,40})/o) { $last_commit = $1; last; } } } close $log; # we may EPIPE here # for pagination ($first_commit, $last_commit); } # private functions below sub get_feedopts { my ($ctx) = @_; my $pi_config = $ctx->{pi_config}; my $listname = $ctx->{listname}; my $cgi = $ctx->{cgi}; my %rv; if (open my $fh, '<', "$ctx->{git_dir}/description") { chomp($rv{description} = <$fh>); close $fh; } else { $rv{description} = '($GIT_DIR/description missing)'; } if ($pi_config && defined $listname && $listname ne '') { my $addr = $pi_config->get($listname, 'address') || ""; $rv{address} = $addr; $addr = $addr->[0] if ref($addr); $rv{id_addr} = $addr; } $rv{id_addr} ||= 'public-inbox@example.com'; my $url_base; if ($cgi) { my $base; if (ref($cgi) eq 'CGI') { $base = $cgi->url(-base); } else { $base = $cgi->base->as_string; $base =~ s!/\z!!; } $url_base = "$base/$listname"; if (my $mid = $ctx->{mid}) { # per-thread feed: $rv{atomurl} = "$url_base/$mid/t.atom"; } else { $rv{atomurl} = "$url_base/new.atom"; } } else { $url_base = "http://example.com"; $rv{atomurl} = "$url_base/new.atom"; } $rv{url} ||= "$url_base/"; $rv{midurl} = "$url_base/"; \%rv; } sub mime_header { my ($mime, $name) = @_; PublicInbox::Hval->new_oneline($mime->header($name))->raw; } sub feed_updated { my ($date, $ts) = @_; my @t = eval { strptime($date) } if defined $date; @t = gmtime($ts || time) unless scalar @t; '