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 DateTime::Format::Mail;
-use CGI qw(escapeHTML);
-use POSIX qw(strftime);
-use constant DATEFMT => '%Y-%m-%dT%H:%M:%SZ';
-our $dt_parser = DateTime::Format::Mail->new(loose => 1);
+use Email::MIME;
+use Date::Parse qw(strptime str2time);
+use PublicInbox::Hval;
+use PublicInbox::GitCatFile;
+use constant {
+ DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # atom standard
+ MAX_PER_PAGE => 25, # this needs to be tunable
+};
# main function
-# FIXME: takes too many args, cleanup
sub generate {
- my ($class, $git_dir, $max, $pi_config, $listname, $cgi, $top) = @_;
- $max ||= 25;
-
- local $ENV{GIT_DIR} = $git_dir;
- my $feed_opts = get_feedopts($pi_config, $listname, $cgi);
+ my ($class, $args) = @_;
+ require XML::Atom::SimpleFeed;
+ require PublicInbox::View;
+ require POSIX;
+ my $max = $args->{max} || MAX_PER_PAGE;
+ 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),
);
- my @entries;
+ my $git = PublicInbox::GitCatFile->new($args->{git_dir});
+ each_recent_blob($args, sub {
+ my ($add) = @_;
+ 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) = @_;
+ require PublicInbox::Thread;
+
+ my $max = $args->{max} || MAX_PER_PAGE;
+ my $feed_opts = get_feedopts($args);
+
+ my $title = $feed_opts->{description} || '';
+ $title = PublicInbox::Hval->new_oneline($title)->as_html;
+
+ my @messages;
+ 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 = PublicInbox::Thread->new(@messages);
+ $th->thread;
+ my $html = "<html><head><title>$title</title>" .
+ '<link rel="alternate" title="Atom feed" href="' .
+ $feed_opts->{atomurl} . '" type="application/atom+xml"/>' .
+ '</head><body><pre>';
+
+ # 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);
+ $footer = "<hr /><pre>$footer</pre>" if $footer;
+ $html . "</pre>$footer</html>";
+}
+
+# private subs
+
+sub nav_footer {
+ my ($cgi, $last) = @_;
+ $cgi or return '';
+ my $old_r = $cgi->param('r');
+ my $head = ' ';
+ my $next = ' ';
+
+ if ($last) {
+ $next = qq!<a href="?r=$last">next</a>!;
+ }
+ if ($old_r) {
+ $head = $cgi->path_info;
+ $head = qq!<a href="$head">head</a>!;
+ }
+ "$next $head";
+}
+
+sub each_recent_blob {
+ my ($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-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};
- $nr += add_to_feed($feed_opts, $feed, $add, $top);
- last if $nr >= $max;
- } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
+ $nr += $cb->($add);
+ 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;
+ }
- $feed->as_string;
+ close $log; # we may EPIPE here
+ # for pagination
+ $commits[-1];
}
# private functions below
sub get_feedopts {
- my ($pi_config, $listname, $cgi) = @_;
+ my ($args) = @_;
+ my $pi_config = $args->{pi_config};
+ 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 url atomurl midurl)) {
- $rv{$key} = $pi_config->get($listname, $key);
+ foreach my $key (qw(address)) {
+ $rv{$key} = $pi_config->get($listname, $key) || "";
}
}
+ my $url_base;
if ($cgi) {
- my $cgi_url = $cgi->self_url;
- my $url_base = $cgi_url;
- $url_base =~ s!/?(?:index|all)\.atom\.xml\z!!;
- $rv{url} ||= "$url_base/";
- $rv{midurl} ||= "$url_base/mid/%s.html";
- $rv{atomurl} = $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!!) {
+ $rv{atomurl} = "$base$url_base/atom.xml";
+ } else {
+ $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/atom.xml";
}
+ $rv{url} ||= "$url_base/";
+ $rv{midurl} = "$url_base/m/";
+ $rv{fullurl} = "$url_base/f/";
\%rv;
}
-sub utf8_header {
+sub mime_header {
my ($mime, $name) = @_;
- encode('utf8', decode('MIME-Header', $mime->header($name)));
+ PublicInbox::Hval->new_oneline($mime->header($name))->raw;
}
sub feed_date {
my ($date) = @_;
- my $dt = $dt_parser->parse_datetime($date);
- $dt ? $dt->strftime(DATEFMT) : 0;
+ my @t = eval { strptime($date) };
+
+ scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0;
}
# returns 0 (skipped) or 1 (added)
sub add_to_feed {
- my ($feed_opts, $feed, $add, $top) = @_;
-
- # 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);
+ my ($feed_opts, $feed, $add, $git) = @_;
- if ($top && $mime->header("In-Reply-To")) {
- return 0;
- }
+ 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 $content = msg_content($mime);
+ 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 $midurl = $feed_opts->{midurl} || "http://example.com/mid/%s.html";
- my $mid = utf8_header($mime, "Message-ID") or return 0;
- $mid =~ s/\A<//;
- $mid =~ s/>\z//;
-
- my $subject = utf8_header($mime, "Subject") || "";
- defined($subject) && length($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 $url = sprintf($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 . $href,
+ id => 'urn:uuid:' . join('-', @uuid5),
);
1;
}
-# returns a plain-text message body without quoted text
-# returns undef if there was nothing
-sub msg_content {
- my ($mime) = @_;
- my $rv;
-
- # scan through all parts, looking for displayable text
- $mime->walk_parts(sub {
- return if $rv;
- my ($part) = @_;
- return if $part->subparts; # walk_parts already recurses
- my $ct = $part->content_type || 'text/plain';
- return if $ct !~ m!\btext/[a-z0-9\+\._-]+\b!i;
- my @body;
- my $killed_wrote; # omit "So-and-so wrote:" line
-
- # no quoted text in Atom feed summary
- # $part->body should already be decoded for us (no QP)
-
- my $state = 0; # 0: beginning, 1: keep, 2: quoted
- foreach my $l (split(/\r?\n/, $part->body)) {
- if ($state == 0) {
- # drop leading blank lines
- next if $l =~ /\A\s*\z/;
-
- $state = ($l =~ /\A>/) ? 2 : 1; # fall-through
- }
- if ($state == 2) { # quoted text, drop it
- if ($l !~ /\A>/) {
- push @body, "<quoted text snipped>";
- if ($l =~ /\S/) {
- push @body, $l;
- }
- $state = 1;
- }
- }
- if ($state == 1) { # stuff we may keep
- if ($l =~ /\A>/) {
- # drop "So-and-so wrote:" line
- if (@body && !$killed_wrote &&
- $body[-1] =~ /:\z/) {
- $killed_wrote = 1;
- pop @body;
- }
- $state = 2;
- } else {
- push @body, $l;
- }
- }
- }
- $rv = "<pre>" .
- join("\n", map { escapeHTML($_) } @body) .
- "</pre>";
- });
+sub dump_html_line {
+ my ($self, $level, $html) = @_;
+ if ($self->message) {
+ $$html .= (' ' x $level);
+ my $mime = $self->message;
+ my $subj = $mime->header('Subject');
+ my $mid = $mime->header_obj->header_raw('Message-ID');
+ $mid = PublicInbox::Hval->new_msgid($mid);
+ my $href = 'm/' . $mid->as_href . '.html';
+ my $from = mime_header($mime, 'From');
+
+ my @from = Email::Address->parse($from);
+ $from = $from[0]->name;
+ (defined($from) && length($from)) or $from = $from[0]->address;
+
+ $from = PublicInbox::Hval->new_oneline($from)->as_html;
+ $subj = PublicInbox::Hval->new_oneline($subj)->as_html;
+ $$html .= "<a href=\"$href\">$subj</a> $from\n";
+ }
+ dump_html_line($self->child, $level+1, $html) if $self->child;
+ dump_html_line($self->next, $level, $html) if $self->next;
+}
- $rv;
+sub do_cat_mail {
+ my ($git, $path) = @_;
+ my $str = $git->cat_file("HEAD:$path");
+ Email::MIME->new($str);
}
1;