X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=6996f374e555a4f55d565fda80fe3b119369e3b0;hb=ec1d0105e6f758ced83902b92b25df0c4b75613d;hp=75e5d9532de3e7f0a21ec9344487b3c35b18e2aa;hpb=98352cd1c066d8cc16594f000509997cfd590985;p=public-inbox.git diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm index 75e5d953..6996f374 100644 --- a/lib/PublicInbox/Feed.pm +++ b/lib/PublicInbox/Feed.pm @@ -3,100 +3,174 @@ package PublicInbox::Feed; use strict; use warnings; -use XML::Atom::SimpleFeed; -use Email::MIME; use Email::Address; -use URI::Escape qw/uri_escape/; -use Encode qw/encode decode/; -use Encode::MIME::Header; -use CGI qw(escapeHTML); -use POSIX qw(strftime); -use Date::Parse qw(strptime); -use constant DATEFMT => '%Y-%m-%dT%H:%M:%SZ'; -use PublicInbox::View; -use Mail::Thread; +use Email::MIME; +use Date::Parse qw(strptime str2time); +use PublicInbox::Hval; +use PublicInbox::GitCatFile; +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 ($class, $args) = @_; - my $max = $args->{max} || 25; - my $top = $args->{top}; # bool + require XML::Atom::SimpleFeed; + require PublicInbox::View; + require POSIX; + my $max = $args->{max} || MAX_PER_PAGE; - local $ENV{GIT_DIR} = $args->{git_dir}; my $feed_opts = get_feedopts($args); - + my $addr = $feed_opts->{address}; + $addr = $addr->[0] if ref($addr); 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", + "http://example.com/atom.xml", }, - id => $feed_opts->{address} || 'public-inbox@example.com', - updated => strftime(DATEFMT, gmtime), + id => 'mailto:' . ($addr || 'public-inbox@example.com'), + updated => POSIX::strftime(DATEFMT, gmtime), ); - each_recent_blob($max, sub { + my $git = PublicInbox::GitCatFile->new($args->{git_dir}); + each_recent_blob($args, sub { my ($add) = @_; - add_to_feed($feed_opts, $feed, $add, $top); + add_to_feed($feed_opts, $feed, $add, $git); }); + $git = undef; # destroy pipes + Email::Address->purge_cache; $feed->as_string; } sub generate_html_index { my ($class, $args) = @_; - my $max = $args->{max} || 50; - my $top = $args->{top}; # bool - local $ENV{GIT_DIR} = $args->{git_dir}; + require PublicInbox::Thread; + + my $max = $args->{max} || MAX_PER_PAGE; my $feed_opts = get_feedopts($args); - my $title = escapeHTML($feed_opts->{description} || ""); + + my $title = $feed_opts->{description} || ''; + $title = PublicInbox::Hval->new_oneline($title)->as_html; + my @messages; - each_recent_blob($max, sub { - my $str = `git cat-file blob $_[0]`; - return 0 if $? != 0; - my $simple = Email::Simple->new($str); - $simple->body_set(""); # save some memory - push @messages, $simple; + my $git = PublicInbox::GitCatFile->new($args->{git_dir}); + my $last = each_recent_blob($args, sub { + my $mime = do_cat_mail($git, $_[0]) or return 0; + $mime->body_set(''); # save some memory + + my $t = eval { str2time($mime->header('Date')) }; + defined($t) or $t = 0; + $mime->header_set('X-PI-Date', $t); + push @messages, $mime; 1; }); + $git = undef; # destroy pipes. - my $th = Mail::Thread->new(@messages); + my $th = PublicInbox::Thread->new(@messages); $th->thread; - my @args = ( - "$title" . - '' . - '
');
-	push @args, $feed_opts->{midurl};
-	dump_html_line($_, 0, \@args) for $th->rootset;
-	$args[0] . '
'; + my $html = "$title" . + '' . + '
';
+
+	# 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, \$html) for $th->rootset;
+
+	Email::Address->purge_cache;
+
+	my $footer = nav_footer($args->{cgi}, $last);
+	$footer = "
$footer
" if $footer; + $html . "
$footer"; } # private subs +sub nav_footer { + my ($cgi, $last) = @_; + $cgi or return ''; + my $old_r = $cgi->param('r'); + my $head = ' '; + my $next = ' '; + + if ($last) { + $next = qq!next!; + } + if ($old_r) { + $head = $cgi->path_info; + $head = qq!head!; + } + "$next $head"; +} + sub each_recent_blob { - my ($max, $cb) = @_; + my ($args, $cb) = @_; + my $max = $args->{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 $refhex = qr/${hex}{4,40}(?:~\d+)?/; + my $cgi = $args->{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 $cmd = "git log --no-notes --no-color --raw -r --no-abbrev HEAD |"; - my $pid = open my $log, $cmd or die "open `$cmd' pipe failed: $!\n"; - my %deleted; + my @cmd = ('git', "--git-dir=$args->{git_dir}", + qw/log --no-notes --no-color --raw -r/); + push @cmd, $range; + + my $pid = open(my $log, '-|', @cmd) or + die('open `'.join(' ', @cmd) . " pipe failed: $!\n"); + my %deleted; # only an optimization at this point + my $last; my $nr = 0; - foreach my $line (<$log>) { - if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) { + my @commits = (); + while (my $line = <$log>) { + if ($line =~ /$addmsg/o) { my $add = $1; next if $deleted{$add}; $nr += $cb->($add); - last if $nr >= $max; - } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) { + if ($nr >= $max) { + $last = 1; + last; + } + } elsif ($line =~ /$delmsg/o) { $deleted{$1} = 1; + } elsif ($line =~ /^commit (${hex}{40})/) { + push @commits, $1; } } - close $log; + if ($last) { + while (my $line = <$log>) { + if ($line =~ /^commit (${hex}{40})/) { + push @commits, $1; + last; + } + } + } else { + push @commits, undef; + } + + close $log; # we may EPIPE here + # for pagination + $commits[-1]; } # private functions below @@ -106,25 +180,37 @@ sub get_feedopts { my $listname = $args->{listname}; my $cgi = $args->{cgi}; my %rv; + if (open my $fh, '<', "$args->{git_dir}/description") { + chomp($rv{description} = <$fh>); + close $fh; + } if ($pi_config && defined $listname && length $listname) { - foreach my $key (qw(description address)) { + foreach my $key (qw(address)) { $rv{$key} = $pi_config->get($listname, $key) || ""; } } my $url_base; if ($cgi) { - my $cgi_url = $cgi->self_url; - $url_base = $cgi_url; + my $path_info = $cgi->path_info; + my $base; + if (ref($cgi) eq 'CGI') { + $base = $cgi->url(-base); + } else { + $base = $cgi->base->as_string; + $base =~ s!/\z!!; + } + $url_base = $path_info; if ($url_base =~ s!/(?:|index\.html)?\z!!) { - $rv{atomurl} = "$url_base/index.atom.xml"; + $rv{atomurl} = "$base$url_base/atom.xml"; } else { - $url_base =~ s!/?(?:index|all)\.atom\.xml\z!!; - $rv{atomurl} = $cgi_url; + $url_base =~ s!/atom\.xml\z!!; + $rv{atomurl} = $base . $path_info; + $url_base = $base . $url_base; # XXX is this needed? } } else { $url_base = "http://example.com"; - $rv{atomurl} = "$url_base/index.atom.xml"; + $rv{atomurl} = "$url_base/atom.xml"; } $rv{url} ||= "$url_base/"; $rv{midurl} = "$url_base/m/"; @@ -133,92 +219,87 @@ sub get_feedopts { \%rv; } -sub utf8_header { - my ($simple, $name) = @_; - my $val = $simple->header($name); - return "" unless defined $val; - $val =~ tr/\t\r\n / /s; - encode('utf8', decode('MIME-Header', $val)); +sub mime_header { + my ($mime, $name) = @_; + PublicInbox::Hval->new_oneline($mime->header($name))->raw; } sub feed_date { my ($date) = @_; my @t = eval { strptime($date) }; - scalar(@t) ? strftime(DATEFMT, @t) : 0; + scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0; } # returns 0 (skipped) or 1 (added) sub add_to_feed { - my ($feed_opts, $feed, $add, $top) = @_; - - # we can use git cat-file --batch if performance becomes a - # problem, but I doubt it... - my $str = `git cat-file blob $add`; - return 0 if $? != 0; - my $mime = Email::MIME->new($str); - - if ($top && $mime->header("In-Reply-To")) { - return 0; - } + my ($feed_opts, $feed, $add, $git) = @_; + my $mime = do_cat_mail($git, $add) or 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); + my $mid = $mime->header_obj->header_raw('Message-ID'); + defined $mid or return 0; + $mid = PublicInbox::Hval->new_msgid($mid); + my $href = $mid->as_href . '.html'; + my $content = PublicInbox::View->feed_entry($mime, $fullurl . $href); defined($content) or return 0; - my $mid = utf8_header($mime, "Message-ID") or return 0; - $mid =~ s/\A\z//; - - my $subject = utf8_header($mime, "Subject") || ""; - length($subject) or return 0; - - my $from = utf8_header($mime, "From") or return 0; + my $subject = mime_header($mime, 'Subject') or return 0; + my $from = mime_header($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; + my $date = $mime->header('Date'); + $date = PublicInbox::Hval->new_oneline($date); + $date = feed_date($date->raw) or return 0; + $add =~ tr!/!!d; + my $h = '[a-f0-9]'; + my (@uuid5) = ($add =~ m!\A($h{8})($h{4})($h{4})($h{4})($h{12})!o); + $feed->add_entry( author => { name => $name, email => $email }, title => $subject, updated => $date, - content => { type => "html", content => $content }, - link => $url, - id => $add, + content => { type => 'xhtml', content => $content }, + link => $midurl . $href, + id => 'urn:uuid:' . join('-', @uuid5), ); 1; } sub dump_html_line { - my ($self, $level, $args) = @_; # args => [ $html, $midurl ] - $args->[0] .= (' ' x $level); + my ($self, $level, $html) = @_; if ($self->message) { - my $simple = $self->message; - my $subj = utf8_header($simple, "Subject"); - my $mid = utf8_header($simple, "Message-ID"); - $mid =~ s/\A\z//; - my $url = $args->[1] . uri_escape($mid); - my $from = utf8_header($simple, "From"); + $$html .= (' ' x $level); + my $mime = $self->message; + my $subj = $mime->header('Subject'); + my $mid = $mime->header_obj->header_raw('Message-ID'); + $mid = PublicInbox::Hval->new_msgid($mid); + my $href = 'm/' . $mid->as_href . '.html'; + my $from = mime_header($mime, 'From'); + my @from = Email::Address->parse($from); $from = $from[0]->name; (defined($from) && length($from)) or $from = $from[0]->address; - $from = escapeHTML($from); - $subj = escapeHTML($subj); - $args->[0] .= "`-> $subj $from\n"; - } else { - $args->[0] .= "[ Message not available ]\n"; + + $from = PublicInbox::Hval->new_oneline($from)->as_html; + $subj = PublicInbox::Hval->new_oneline($subj)->as_html; + $$html .= "$subj $from\n"; } - dump_html_line($self->child, $level+1, $args) if $self->child; - dump_html_line($self->next, $level, $args) if $self->next; + dump_html_line($self->child, $level+1, $html) if $self->child; + dump_html_line($self->next, $level, $html) if $self->next; +} + +sub do_cat_mail { + my ($git, $path) = @_; + my $str = $git->cat_file("HEAD:$path"); + Email::MIME->new($str); } 1;