X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FFeed.pm;h=6996f374e555a4f55d565fda80fe3b119369e3b0;hb=ec1d0105e6f758ced83902b92b25df0c4b75613d;hp=4671323cc1bd74b48a07c905b9065d889d28bcc9;hpb=bbe5d24accf4f10af1a9a5ae46df10604c1410f5;p=public-inbox.git
diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm
index 4671323c..6996f374 100644
--- a/lib/PublicInbox/Feed.pm
+++ b/lib/PublicInbox/Feed.pm
@@ -4,35 +4,26 @@ 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 Email::MIME;
use Date::Parse qw(strptime str2time);
-eval { require Git }; # this is GPLv2+, so we are OK to use it
+use PublicInbox::Hval;
+use PublicInbox::GitCatFile;
use constant {
- DATEFMT => '%Y-%m-%dT%H:%M:%SZ',
- MAX_PER_PAGE => 25,
+ DATEFMT => '%Y-%m-%dT%H:%M:%SZ', # atom standard
+ MAX_PER_PAGE => 25, # this needs to be tunable
};
-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) = @_;
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/",
@@ -41,53 +32,50 @@ sub generate {
href => $feed_opts->{atomurl} ||
"http://example.com/atom.xml",
},
- id => $feed_opts->{address} || 'public-inbox@example.com',
+ id => 'mailto:' . ($addr || 'public-inbox@example.com'),
updated => POSIX::strftime(DATEFMT, gmtime),
);
- my $git = try_git_pm($args->{git_dir});
+ my $git = PublicInbox::GitCatFile->new($args->{git_dir});
each_recent_blob($args, sub {
my ($add) = @_;
- add_to_feed($feed_opts, $feed, $add, $top, $git);
+ add_to_feed($feed_opts, $feed, $add, $git);
});
+ $git = undef; # destroy pipes
+ Email::Address->purge_cache;
$feed->as_string;
}
sub generate_html_index {
my ($class, $args) = @_;
- require Mail::Thread;
+ require PublicInbox::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 $title = $feed_opts->{description} || '';
+ $title = PublicInbox::Hval->new_oneline($title)->as_html;
+
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 $git = PublicInbox::GitCatFile->new($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 $t = eval { str2time($simple->header('Date')) };
+ my $t = eval { str2time($mime->header('Date')) };
defined($t) or $t = 0;
- $simple->header_set('X-PI-Date', $t);
- push @messages, $simple;
+ $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);
$th->thread;
- my @out = (
- "$title" .
- '$title" .
+ '' .
- '');
- push @out, $feed_opts->{midurl};
+ '';
# sort by date, most recent at top
$th->order(sub {
@@ -96,71 +84,65 @@ sub generate_html_index {
$a->topmost->message->header('X-PI-Date')
} @_;
});
- dump_html_line($_, 0, \@out) for $th->rootset;
+ dump_html_line($_, 0, \$html) for $th->rootset;
+
+ Email::Address->purge_cache;
- my $footer = nav_footer($args->{cgi}, $first, $last);
+ my $footer = nav_footer($args->{cgi}, $last);
$footer = "
$footer
" if $footer;
- $out[0] . "$footer";
+ $html . "$footer";
}
# private subs
sub nav_footer {
- my ($cgi, $first, $last) = @_;
+ my ($cgi, $last) = @_;
$cgi or return '';
my $old_r = $cgi->param('r');
- my $prev = ' ';
+ my $head = ' ';
my $next = ' ';
- my %opts = (-path => 1, -query => 1, -relative => 1);
if ($last) {
- $cgi->param('r', $last);
- $next = $cgi->url(%opts);
- $next = qq!next!;
+ $next = qq!next!;
}
- if ($first && $old_r) {
- $cgi->param('r', "$first..");
- $prev = $cgi->url(%opts);
- $prev = qq!prev!;
+ if ($old_r) {
+ $head = $cgi->path_info;
+ $head = qq!head!;
}
- "$prev $next";
+ "$next $head";
}
sub each_recent_blob {
my ($args, $cb) = @_;
my $max = $args->{max} || MAX_PER_PAGE;
- my $refhex = qr/[a-f0-9]{4,40}(?:~\d+)?/;
+ 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 $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 = $r;
- }
+ if ($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 $reverse;
+ my @cmd = ('git', "--git-dir=$args->{git_dir}",
+ qw/log --no-notes --no-color --raw -r/);
push @cmd, $range;
- my $first;
my $pid = open(my $log, '-|', @cmd) or
die('open `'.join(' ', @cmd) . " pipe failed: $!\n");
- my %deleted;
+ my %deleted; # only an optimization at this point
my $last;
my $nr = 0;
my @commits = ();
while (my $line = <$log>) {
- if ($line =~ /^:000000 100644 0{40} ([a-f0-9]{40})/) {
+ if ($line =~ /$addmsg/o) {
my $add = $1;
next if $deleted{$add};
$nr += $cb->($add);
@@ -168,29 +150,27 @@ sub each_recent_blob {
$last = 1;
last;
}
- } elsif ($line =~ /^:100644 000000 ([a-f0-9]{40}) 0{40}/) {
+ } elsif ($line =~ /$delmsg/o) {
$deleted{$1} = 1;
- } elsif ($line =~ /^commit ([a-f0-9]{40})/) {
+ } elsif ($line =~ /^commit (${hex}{40})/) {
push @commits, $1;
}
}
if ($last) {
while (my $line = <$log>) {
- if ($line =~ /^commit ([a-f0-9]{40})/) {
+ if ($line =~ /^commit (${hex}{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]);
+ $commits[-1];
}
# private functions below
@@ -200,22 +180,32 @@ 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->url(-path=>1, -relative=>1);
- my $base = $cgi->url(-base);
- $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!!) {
$rv{atomurl} = "$base$url_base/atom.xml";
} else {
$url_base =~ s!/atom\.xml\z!!;
- $rv{atomurl} = $base . $cgi_url;
+ $rv{atomurl} = $base . $path_info;
$url_base = $base . $url_base; # XXX is this needed?
}
} else {
@@ -229,13 +219,9 @@ sub get_feedopts {
\%rv;
}
-sub utf8_header {
- my ($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($val));
+sub mime_header {
+ my ($mime, $name) = @_;
+ PublicInbox::Hval->new_oneline($mime->header($name))->raw;
}
sub feed_date {
@@ -247,95 +233,73 @@ sub feed_date {
# returns 0 (skipped) or 1 (added)
sub add_to_feed {
- my ($feed_opts, $feed, $add, $top, $git) = @_;
-
- my $mime = do_cat_mail($git, 'Email::MIME', $add) or return 0;
- if ($top && $mime->header("In-Reply-To")) {
- return 0;
- }
+ my ($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, $midurl ]
+ my ($self, $level, $html) = @_;
if ($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 $from = utf8_header($simple, "From");
+ $$html .= (' ' x $level);
+ my $mime = $self->message;
+ my $subj = $mime->header('Subject');
+ my $mid = $mime->header_obj->header_raw('Message-ID');
+ $mid = PublicInbox::Hval->new_msgid($mid);
+ my $href = 'm/' . $mid->as_href . '.html';
+ my $from = mime_header($mime, 'From');
+
my @from = Email::Address->parse($from);
$from = $from[0]->name;
(defined($from) && length($from)) or $from = $from[0]->address;
- $from = xs_html($from);
- $subj = xs_html($subj);
- $args->[0] .= "$subj $from\n";
- }
- dump_html_line($self->child, $level+1, $args) if $self->child;
- dump_html_line($self->next, $level, $args) if $self->next;
-}
-sub xs_html {
- $enc_ascii->encode(escapeHTML($enc_utf8->decode($_[0])),
- Encode::HTMLCREF);
+ $from = PublicInbox::Hval->new_oneline($from)->as_html;
+ $subj = PublicInbox::Hval->new_oneline($subj)->as_html;
+ $$html .= "$subj $from\n";
+ }
+ dump_html_line($self->child, $level+1, $html) if $self->child;
+ dump_html_line($self->next, $level, $html) if $self->next;
}
-sub try_git_pm {
- my ($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);
+ my ($git, $path) = @_;
+ my $str = $git->cat_file("HEAD:$path");
+ Email::MIME->new($str);
}
1;