-# Copyright (C) 2013, Eric Wong <normalperson@yhbt.net> and all contributors
-# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
+# Copyright (C) 2013-2021 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <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 URI::Escape qw/uri_escape/;
-use Encode qw/find_encoding/;
-use Encode::MIME::Header;
-use CGI qw(escapeHTML);
-use Date::Parse qw(strptime str2time);
-eval { require Git }; # this is GPLv2+, so we are OK to use it
-use constant {
- DATEFMT => '%Y-%m-%dT%H:%M:%SZ',
- MAX_PER_PAGE => 25,
-};
-my $enc_utf8 = find_encoding('utf8');
-my $enc_ascii = find_encoding('us-ascii');
-my $enc_mime = find_encoding('MIME-Header');
+use PublicInbox::View;
+use PublicInbox::WwwAtomStream;
+use PublicInbox::Smsg; # this loads w/o Search::Xapian
-# FIXME: workaround https://rt.cpan.org/Public/Bug/Display.html?id=22817
+sub generate_i {
+ my ($ctx) = @_;
+ shift @{$ctx->{msgs}};
+}
# main function
sub generate {
- my ($class, $args) = @_;
- require XML::Atom::SimpleFeed;
- require PublicInbox::View;
- require Email::MIME;
- require POSIX;
- my $max = $args->{max} || MAX_PER_PAGE;
- my $top = $args->{top}; # bool
-
- local $ENV{GIT_DIR} = $args->{git_dir};
- my $feed_opts = get_feedopts($args);
-
- my $feed = XML::Atom::SimpleFeed->new(
- title => $feed_opts->{description} || "unnamed feed",
- link => $feed_opts->{url} || "http://example.com/",
- link => {
- rel => 'self',
- href => $feed_opts->{atomurl} ||
- "http://example.com/atom",
- },
- id => $feed_opts->{address} || 'public-inbox@example.com',
- updated => POSIX::strftime(DATEFMT, gmtime),
- );
+ my ($ctx) = @_;
+ my $msgs = $ctx->{msgs} = recent_msgs($ctx);
+ return _no_thread() unless @$msgs;
+ PublicInbox::WwwAtomStream->response($ctx, 200, \&generate_i);
+}
- my $git = try_git_pm($args->{git_dir});
- each_recent_blob($args, sub {
- my ($add) = @_;
- add_to_feed($feed_opts, $feed, $add, $top, $git);
- });
- $feed->as_string;
+sub generate_thread_atom {
+ my ($ctx) = @_;
+ my $msgs = $ctx->{msgs} = $ctx->{ibx}->over->get_thread($ctx->{mid});
+ return _no_thread() unless @$msgs;
+ PublicInbox::WwwAtomStream->response($ctx, 200, \&generate_i);
}
sub generate_html_index {
- my ($class, $args) = @_;
- require Mail::Thread;
-
- my $max = $args->{max} || MAX_PER_PAGE;
- my $top = $args->{top}; # bool
- local $ENV{GIT_DIR} = $args->{git_dir};
- my $feed_opts = get_feedopts($args);
- my $title = xs_html($feed_opts->{description} || "");
- my @messages;
- my $git = try_git_pm($args->{git_dir});
- my ($first, $last) = each_recent_blob($args, sub {
- my $simple = do_cat_mail($git, 'Email::Simple', $_[0])
- or return 0;
- if ($top && ($simple->header("In-Reply-To") ||
- $simple->header("References"))) {
- return 0;
- }
- $simple->body_set(""); # save some memory
+ my ($ctx) = @_;
+ # if the 'r' query parameter is given, it is a legacy permalink
+ # which we must continue supporting:
+ my $qp = $ctx->{qp};
+ my $ibx = $ctx->{ibx};
+ if ($qp && !$qp->{r} && $ibx->over) {
+ return PublicInbox::View::index_topics($ctx);
+ }
- my $t = eval { str2time($simple->header('Date')) };
- defined($t) or $t = 0;
- $simple->header_set('X-PI-Date', $t);
- push @messages, $simple;
- 1;
- });
+ my $env = $ctx->{env};
+ my $url = $ibx->base_url($env) . 'new.html';
+ my $qs = $env->{QUERY_STRING};
+ $url .= "?$qs" if $qs ne '';
+ [302, [ 'Location', $url, 'Content-Type', 'text/plain'],
+ [ "Redirecting to $url\n" ] ];
+}
- my $th = Mail::Thread->new(@messages);
- $th->thread;
- my @out = (
- "<html><head><title>$title</title>" .
- '<link rel=alternate title=Atom.feed href="' .
- $feed_opts->{atomurl} . '" type="application/atom+xml"/>' .
- '</head><body><pre>');
- push @out, $feed_opts->{midurl};
+sub new_html_i {
+ my ($ctx, $eml) = @_;
+ $ctx->zmore($ctx->html_top) if exists $ctx->{-html_tip};
- # sort by date, most recent at top
- $th->order(sub {
- sort {
- $b->topmost->message->header('X-PI-Date') <=>
- $a->topmost->message->header('X-PI-Date')
- } @_;
- });
- dump_html_line($_, 0, \@out) for $th->rootset;
+ $eml and return PublicInbox::View::eml_entry($ctx, $eml);
+ my $smsg = shift @{$ctx->{msgs}} or
+ $ctx->zmore(PublicInbox::View::pagination_footer(
+ $ctx, './new.html'));
+ $smsg;
+}
- my $footer = nav_footer($args->{cgi}, $first, $last);
- $footer = "<hr /><pre>$footer</pre>" if $footer;
- $out[0] . "</pre>$footer</html>";
+sub new_html {
+ my ($ctx) = @_;
+ my $msgs = $ctx->{msgs} = recent_msgs($ctx);
+ if (!@$msgs) {
+ return [404, ['Content-Type', 'text/plain'],
+ ["No messages, yet\n"] ];
+ }
+ $ctx->{-html_tip} = '<pre>';
+ $ctx->{-upfx} = '';
+ $ctx->{-hr} = 1;
+ PublicInbox::WwwStream::aresponse($ctx, 200, \&new_html_i);
}
# private subs
-sub nav_footer {
- my ($cgi, $first, $last) = @_;
- $cgi or return '';
- my $old_r = $cgi->param('r');
- my $prev = ' ';
- my $next = ' ';
- my %opts = (-path => 1, -query => 1, -relative => 1);
-
- if ($last) {
- $cgi->param('r', $last);
- $next = $cgi->url(%opts);
- $next = qq!<a href="$next">next</a>!;
- }
- if ($first && $old_r) {
- $cgi->param('r', "$first..");
- $prev = $cgi->url(%opts);
- $prev = qq!<a href="$prev">prev</a>!;
- }
- "$prev $next";
+sub _no_thread () {
+ [404, ['Content-Type', 'text/plain'], ["No feed found for thread\n"]];
}
-sub each_recent_blob {
- my ($args, $cb) = @_;
- my $max = $args->{max} || MAX_PER_PAGE;
- my $refhex = qr/[a-f0-9]{4,40}(?:~\d+)?/;
- my $cgi = $args->{cgi};
+sub recent_msgs {
+ my ($ctx) = @_;
+ my $ibx = $ctx->{ibx};
+ my $max = $ibx->{feedmax} // 25;
+ return PublicInbox::View::paginate_recent($ctx, $max) if $ibx->over;
+
+ # only for rare v1 inboxes which aren't indexed at all
+ my $qp = $ctx->{qp};
+ 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,})(?:~[0-9]+)?/;
# revision ranges may be specified
- my $reverse;
my $range = 'HEAD';
- my $r = $cgi->param('r') if $cgi;
- if ($r) {
- if ($r =~ /\A(?:$refhex\.\.)?$refhex\z/o) {
- $range = $r;
- } elsif ($r =~ /\A(?:$refhex\.\.)\z/o) {
- $reverse = 1;
- $range = $r;
- }
+ 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 @cmd = qw/git log --no-notes --no-color --raw -r --no-abbrev/;
- push @cmd, '--reverse' if $reverse;
- push @cmd, $range;
- my $first;
-
- my $pid = open(my $log, '-|', @cmd) or
- die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
- my %deleted;
+ my $log = $ibx->git->popen(qw/log
+ --no-notes --no-color --raw -r
+ --no-abbrev --abbrev-commit/,
+ "--format=%H", $range);
+ my %deleted; # only an optimization at this point
my $last;
- my $nr = 0;
- my @commits = ();
- while (my $line = <$log>) {
- if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
+ my $last_commit;
+ local $/ = "\n";
+ my @oids;
+ while (defined(my $line = <$log>)) {
+ if ($line =~ /$addmsg/o) {
my $add = $1;
- next if $deleted{$add};
- $nr += $cb->($add);
- if ($nr >= $max) {
+ next if $deleted{$add}; # optimization-only
+ push @oids, $add;
+ if (scalar(@oids) >= $max) {
$last = 1;
last;
}
- } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
+ } elsif ($line =~ /$delmsg/o) {
$deleted{$1} = 1;
- } elsif ($line =~ /^commit ([a-f0-9]{40})/) {
- push @commits, $1;
}
}
if ($last) {
+ local $/ = "\n";
while (my $line = <$log>) {
- if ($line =~ /^commit ([a-f0-9]{40})/) {
- push @commits, $1;
+ if ($line =~ /^(${hex}{7,})/) {
+ $last_commit = $1;
last;
}
}
}
- close $log; # we may EPIPE here
- # for pagination
- $reverse ? ($commits[-1],$commits[0]) : ($commits[0],$commits[-1]);
-}
-
-# private functions below
-sub get_feedopts {
- my ($args) = @_;
- my $pi_config = $args->{pi_config};
- my $listname = $args->{listname};
- my $cgi = $args->{cgi};
- my %rv;
+ $last_commit and
+ $ctx->{next_page} = qq[<a\nhref="?r=$last_commit"\nrel=next>] .
+ 'next (older)</a>';
- if ($pi_config && defined $listname && length $listname) {
- foreach my $key (qw(description address)) {
- $rv{$key} = $pi_config->get($listname, $key) || "";
- }
- }
- my $url_base;
- if ($cgi) {
- my $cgi_url = $cgi->url(-path=>1, -relative=>1);
- my $base = $cgi->url(-base);
- $url_base = $cgi_url;
- if ($url_base =~ s!/(?:|index\.html)?\z!!) {
- $rv{atomurl} = "$base$url_base/index.atom.xml";
- } else {
- $url_base =~ s!/?(?:index|all)\.atom\.xml\z!!;
- $rv{atomurl} = $base . $cgi_url;
- $url_base = $base . $url_base; # XXX is this needed?
- }
- } else {
- $url_base = "http://example.com";
- $rv{atomurl} = "$url_base/index.atom.xml";
- }
- $rv{url} ||= "$url_base/";
- $rv{midurl} = "$url_base/m/";
- $rv{fullurl} = "$url_base/f/";
-
- \%rv;
-}
-
-sub utf8_header {
- my ($simple, $name) = @_;
- my $val = $simple->header($name);
- return "" unless defined $val;
- $val =~ tr/\t\n / /s;
- $val =~ tr/\r//d;
- $enc_utf8->encode($enc_mime->decode($val));
-}
-
-sub feed_date {
- my ($date) = @_;
- my @t = eval { strptime($date) };
-
- scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0;
-}
-
-# returns 0 (skipped) or 1 (added)
-sub add_to_feed {
- my ($feed_opts, $feed, $add, $top, $git) = @_;
-
- my $mime = do_cat_mail($git, 'Email::MIME', $add) or return 0;
- if ($top && $mime->header("In-Reply-To")) {
- return 0;
- }
-
- my $midurl = $feed_opts->{midurl} || 'http://example.com/m/';
- my $fullurl = $feed_opts->{fullurl} || 'http://example.com/f/';
-
- my $content = PublicInbox::View->as_feed_entry($mime, $fullurl);
- defined($content) or return 0;
-
- my $mid = utf8_header($mime, "Message-ID") or return 0;
- $mid =~ s/\A<//; $mid =~ s/>\z//;
-
- my $subject = utf8_header($mime, "Subject") || "";
- length($subject) or return 0;
-
- my $from = utf8_header($mime, "From") or return 0;
-
- my @from = Email::Address->parse($from);
- my $name = $from[0]->name;
- defined $name or $name = "";
- my $email = $from[0]->address;
- defined $email or $email = "";
-
- my $url = $midurl . uri_escape($mid);
- my $date = utf8_header($mime, "Date");
- $date or return 0;
- $date = feed_date($date) or return 0;
- $feed->add_entry(
- author => { name => $name, email => $email },
- title => $subject,
- updated => $date,
- content => { type => "html", content => $content },
- link => $url,
- id => $add,
- );
- 1;
-}
-
-sub dump_html_line {
- my ($self, $level, $args) = @_; # args => [ $html, $midurl ]
- $args->[0] .= (' ' x $level);
- if ($self->message) {
- my $simple = $self->message;
- my $subj = utf8_header($simple, "Subject");
- my $mid = utf8_header($simple, "Message-ID");
- $mid =~ s/\A<//;
- $mid =~ s/>\z//;
- my $url = $args->[1] . uri_escape($mid);
- my $from = utf8_header($simple, "From");
- my @from = Email::Address->parse($from);
- $from = $from[0]->name;
- (defined($from) && length($from)) or $from = $from[0]->address;
- $from = xs_html($from);
- $subj = xs_html($subj);
- $args->[0] .= "<a href=\"$url.html\">$subj</a> $from\n";
- } else {
- $args->[0] .= "[ Message not available ]\n";
- }
- dump_html_line($self->child, $level+1, $args) if $self->child;
- dump_html_line($self->next, $level, $args) if $self->next;
-}
-
-sub xs_html {
- $enc_ascii->encode(escapeHTML($enc_utf8->decode($_[0])),
- Encode::HTMLCREF);
-}
-
-sub try_git_pm {
- my ($dir) = @_;
- eval { Git->repository(Directory => $dir) };
-};
-
-sub do_cat_mail {
- my ($git, $class, $sha1) = @_;
- my $str;
- if ($git) {
- open my $fh, '>', \$str or
- die "failed to setup string handle: $!\n";
- binmode $fh;
- my $bytes = $git->cat_blob($sha1, $fh);
- return if $bytes <= 0;
- close $fh or die "failed to close string handle: $!\n";
- } else {
- $str = `git cat-file blob $sha1`;
- return if $? != 0 || length($str) == 0;
- }
- $class->new($str);
+ [ map { bless {blob => $_ }, 'PublicInbox::Smsg' } @oids ];
}
1;