# 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::Git; 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) = @_; sub { emit_html_index($_[0], $ctx) }; } # private subs sub title_tag { my ($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\"" : ''; "
} } else { $top = '' . $top . "\n"; } $fh->write("$title " . "" . PublicInbox::Hval::STYLE . "$top"); # if the 'r' query parameter is given, it is a legacy permalink # which we must continue supporting: my $cgi = $ctx->{cgi}; if ($cgi && !$cgi->param('r') && $srch) { $state->{srch} = $srch; $last = PublicInbox::View::emit_index_topics($state); $param = 'o'; } else { $last = emit_index_nosrch($ctx, $state); $param = 'r'; } $footer = nav_footer($cgi, $last, $feed_opts, $state, $param); if ($footer) { my $list_footer = $ctx->{footer}; $footer .= "\n\n" . $list_footer if $list_footer; $footer = "$footer"; } $fh->write("$footer"); $fh->close; } sub emit_index_nosrch { my ($ctx, $state) = @_; 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, $path) or return 0; PublicInbox::View::index_entry($mime, 0, $state); 1; }); $last; } sub nav_footer { my ($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 feed"; "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); 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 $pi_config = $ctx->{pi_config}; my $inbox = $ctx->{inbox}; my $obj = $ctx->{-inbox}; my $cgi = $ctx->{cgi}; my %rv = ( description => $obj ? $obj->description : 'FIXME' ); if ($obj) { $rv{address} = $obj->{address}; $rv{id_addr} = $obj->{-primary_address}; } elsif ($pi_config && defined $inbox && $inbox ne '') { # TODO: remove my $addr = $pi_config->get($inbox, 'address') || ""; $rv{address} = $addr; $addr = $addr->[0] if ref($addr); $rv{id_addr} = $addr; } $rv{id_addr} ||= 'public-inbox@example.com'; my $url_base; if ($obj) { $url_base = $obj->base_url($cgi); # CGI may be undef 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} = $url_base; \%rv; } sub feed_updated { my ($date, $ts) = @_; my @t = eval { strptime($date) } if defined $date; @t = gmtime($ts || time) unless scalar @t; '' . strftime(DATEFMT, @t) . ' '; } # returns 0 (skipped) or 1 (added) sub add_to_feed { my ($feed_opts, $fh, $add, $git) = @_; my $mime = do_cat_mail($git, $add) or return 0; 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 0; $mid = PublicInbox::Hval->new_msgid($mid); my $href = $midurl.$mid->as_href; my $content = qq() . PublicInbox::View::multipart_text_as_html($mime, $href) . ''; my $date = $header_obj->header('Date'); my $updated = feed_updated($date); my $title = $header_obj->header('Subject'); defined $title or return 0; $title = title_tag($title); my $from = $header_obj->header('From') or return 0; my ($email) = PublicInbox::Address::emails($from); my $name = PublicInbox::Address::from_name($from); $name = ascii_html($name); $email = ascii_html($email); if (delete $feed_opts->{emit_header}) { $fh->write(atom_header($feed_opts, $title) . $updated); } $fh->write(""); 1; } sub do_cat_mail { my ($git, $path) = @_; my $mime = eval { my $str = $git->cat_file("HEAD:$path"); Email::MIME->new($str); }; $@ ? undef : $mime; } 1; $title$updated" . qq{ $name } . qq{ !. "}); $fh->write($content); $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); $fh->write(qq!$id