X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=93ee80bbef27b51b88099889c416dc36798dff6a;hb=48dc8ff5e67714985897047c189bdf019b796a60;hp=16b860cbde641baee0cdf9bb7dbee9b7c4a948a4;hpb=6e517aa6dc310ec14c7f2e7cf5479d77f1be5069;p=public-inbox.git
diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm
index 16b860cb..93ee80bb 100644
--- a/lib/PublicInbox/Feed.pm
+++ b/lib/PublicInbox/Feed.pm
@@ -3,104 +3,195 @@
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 qw/find_encoding/;
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 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 $enc_mime = find_encoding('MIME-Header');
+
+# FIXME: workaround https://rt.cpan.org/Public/Bug/Display.html?id=22817
# main function
sub generate {
my ($class, $args) = @_;
- my $max = $args->{max} || 25;
+ 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 $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 => $addr || 'public-inbox@example.com',
+ updated => POSIX::strftime(DATEFMT, gmtime),
);
- each_recent_blob($max, sub {
+ my $git = try_git_pm($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, $top, $git);
});
$feed->as_string;
}
sub generate_html_index {
my ($class, $args) = @_;
- my $max = $args->{max} || 50;
+ 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 $title = escapeHTML($feed_opts->{description} || "");
+ my $title = xs_html($feed_opts->{description} || "");
my @messages;
- each_recent_blob($max, sub {
- my $str = `git cat-file blob $_[0]`;
- return 0 if $? != 0;
- my $simple = Email::Simple->new($str);
+ 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 $t = eval { str2time($simple->header('Date')) };
+ defined($t) or $t = 0;
+ $simple->header_set('X-PI-Date', $t);
push @messages, $simple;
1;
});
my $th = Mail::Thread->new(@messages);
$th->thread;
- my @args = (
+ my @out = (
"$title" .
- '' .
+ '' .
'');
- push @args, $feed_opts->{midurl};
- dump_html_line($_, 0, \@args) for $th->rootset;
- $args[0] . '';
+ push @out, $feed_opts->{midurl};
+
+ # 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;
+
+ my $footer = nav_footer($args->{cgi}, $first, $last);
+ $footer = "
$footer
" if $footer;
+ $out[0] . "$footer";
}
# 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);
+ $next = qq!next!;
+ }
+ if ($first && $old_r) {
+ $cgi->param('r', "$first..");
+ $prev = $cgi->url(%opts);
+ $prev = qq!prev!;
+ }
+ "$prev $next";
+}
+
sub each_recent_blob {
- my ($max, $cb) = @_;
+ my ($args, $cb) = @_;
+ my $max = $args->{max} || MAX_PER_PAGE;
+ my $refhex = qr/[a-f0-9]{4,40}(?:~\d+)?/;
+ my $cgi = $args->{cgi};
+
+ # 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;
+ }
+ }
# 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 @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 $last;
my $nr = 0;
- foreach my $line (<$log>) {
+ my @commits = ();
+ while (my $line = <$log>) {
if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
my $add = $1;
next if $deleted{$add};
$nr += $cb->($add);
- last if $nr >= $max;
+ if ($nr >= $max) {
+ $last = 1;
+ last;
+ }
} elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
$deleted{$1} = 1;
+ } elsif ($line =~ /^commit ([a-f0-9]{40})/) {
+ push @commits, $1;
}
}
- close $log;
+ if ($last) {
+ while (my $line = <$log>) {
+ if ($line =~ /^commit ([a-f0-9]{40})/) {
+ push @commits, $1;
+ last;
+ }
+ }
+ } elsif ($reverse) {
+ unshift @commits, undef
+ } else {
+ push @commits, undef;
+ }
+
+ close $log; # we may EPIPE here
+ # for pagination
+ $reverse ? ($commits[-1],$commits[0]) : ($commits[0],$commits[-1]);
}
# private functions below
@@ -110,25 +201,31 @@ 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;
+ 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!!) {
- $rv{atomurl} = "$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 . $cgi_url;
+ $url_base = $base . $url_base; # XXX is this needed?
}
} else {
$url_base = "http://example.com";
- $rv{atomurl} = "$url_base/index.atom.xml";
+ $rv{atomurl} = "$url_base/atom.xml";
}
$rv{url} ||= "$url_base/";
$rv{midurl} = "$url_base/m/";
@@ -141,27 +238,23 @@ 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));
+ $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($date) };
- scalar(@t) ? strftime(DATEFMT, @t) : 0;
+ scalar(@t) ? POSIX::strftime(DATEFMT, @t) : 0;
}
# returns 0 (skipped) or 1 (added)
sub add_to_feed {
- my ($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);
+ 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;
}
@@ -169,11 +262,13 @@ sub add_to_feed {
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\z//;
+ # FIXME: refactor
+ my (undef, $href) = PublicInbox::View::trim_message_id($mid);
+
+ my $content = PublicInbox::View->as_feed_entry($mime,
+ "$fullurl$href.html");
+ defined($content) or return 0;
my $subject = utf8_header($mime, "Subject") || "";
length($subject) or return 0;
@@ -186,7 +281,6 @@ sub add_to_feed {
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;
@@ -195,7 +289,7 @@ sub add_to_feed {
title => $subject,
updated => $date,
content => { type => "html", content => $content },
- link => $url,
+ link => $midurl . $href,
id => $add,
);
1;
@@ -203,26 +297,51 @@ sub add_to_feed {
sub dump_html_line {
my ($self, $level, $args) = @_; # args => [ $html, $midurl ]
- $args->[0] .= (' ' x $level);
if ($self->message) {
+ $args->[0] .= (' ' x $level);
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 $url = $args->[1] . xs_html(uri_escape($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;
- $from = escapeHTML($from);
- $subj = escapeHTML($subj);
- $args->[0] .= "`-> $subj $from\n";
- } else {
- $args->[0] .= "[ Message not available ]\n";
+ $from = xs_html($from);
+ $subj = xs_html($subj);
+ $args->[0] .= "$subj $from\n";
}
dump_html_line($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);
+ close $fh or die "failed to close string handle: $!\n";
+ return if $bytes <= 0;
+ } else {
+ $str = `git cat-file blob $sha1`;
+ return if $? != 0 || length($str) == 0;
+ }
+ $class->new($str);
+}
+
1;