user/dev discussion of public-inbox itself
 help / Atom feed
* [RFC 0/2] support for /~/$MESSAGE_ID endpoint
@ 2019-01-09 11:43 Eric Wong
  2019-01-09 11:43 ` [RFC 1/2] config: inbox name checking matches git.git more closely Eric Wong
  2019-01-09 11:43 ` [RFC 2/2] www: add /~/$MESSAGE_ID global redirector endpoint Eric Wong
  0 siblings, 2 replies; 3+ messages in thread
From: Eric Wong @ 2019-01-09 11:43 UTC (permalink / raw)
  To: Konstantin Ryabitsev; +Cc: meta

Only lightly-tested at the moment, and unsure about the "/~/"
(I suppose "/_/" could conflict with a valid inbox name).

Eric Wong (2):
  config: inbox name checking matches git.git more closely
  www: add /~/$MESSAGE_ID global redirector endpoint

 MANIFEST                  |  1 +
 lib/PublicInbox/Config.pm | 20 ++++++++++--
 lib/PublicInbox/WWW.pm    | 48 ++++++++++++++++++++++++---
 t/config.t                | 36 ++++++++++++++++++++
 t/psgi_scan_all.t         | 69 +++++++++++++++++++++++++++++++++++++++
 5 files changed, 168 insertions(+), 6 deletions(-)
 create mode 100644 t/psgi_scan_all.t

-- 
EW

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [RFC 1/2] config: inbox name checking matches git.git more closely
  2019-01-09 11:43 [RFC 0/2] support for /~/$MESSAGE_ID endpoint Eric Wong
@ 2019-01-09 11:43 ` Eric Wong
  2019-01-09 11:43 ` [RFC 2/2] www: add /~/$MESSAGE_ID global redirector endpoint Eric Wong
  1 sibling, 0 replies; 3+ messages in thread
From: Eric Wong @ 2019-01-09 11:43 UTC (permalink / raw)
  To: Konstantin Ryabitsev; +Cc: meta

Actually, it turns out git.git/remote.c::valid_remote_nick
rules alone are insufficient.  More checking is performed as
part of the refname in the git.git/refs.c::check_refname_component

I also considered rejecting URL-unfriendly inbox names entirely,
but realized some users may intentionally configure names not
handled by our WWW endpoint for archives they don't want
accessible over HTTP.
---
 lib/PublicInbox/Config.pm | 20 ++++++++++++++++++--
 lib/PublicInbox/WWW.pm    |  4 +++-
 t/config.t                | 36 ++++++++++++++++++++++++++++++++++++
 3 files changed, 57 insertions(+), 3 deletions(-)

diff --git a/lib/PublicInbox/Config.pm b/lib/PublicInbox/Config.pm
index a2b721d..bea2617 100644
--- a/lib/PublicInbox/Config.pm
+++ b/lib/PublicInbox/Config.pm
@@ -152,6 +152,23 @@ sub git_config_dump {
 	\%rv;
 }
 
+sub valid_inbox_name ($) {
+	my ($name) = @_;
+
+	# Similar rules found in git.git/remote.c::valid_remote_nick
+	# and git.git/refs.c::check_refname_component
+	# We don't reject /\.lock\z/, however, since we don't lock refs
+	if ($name eq '' || $name =~ /\@\{/ ||
+	    $name =~ /\.\./ || $name =~ m![/:\?\[\]\^~\s\f[:cntrl:]\*]! ||
+	    $name =~ /\A\./ || $name =~ /\.\z/) {
+		return 0;
+	}
+
+	# Note: we allow URL-unfriendly characters; users may configure
+	# non-HTTP-accessible inboxes
+	1;
+}
+
 sub _fill {
 	my ($self, $pfx) = @_;
 	my $rv = {};
@@ -185,8 +202,7 @@ sub _fill {
 	my $name = $pfx;
 	$name =~ s/\Apublicinbox\.//;
 
-	# same rules as git.git/remote.c::valid_remote_nick
-	if ($name eq '' || $name =~ m!/! || $name eq '.' || $name eq '..') {
+	if (!valid_inbox_name($name)) {
 		warn "invalid inbox name: '$name'\n";
 		return;
 	}
diff --git a/lib/PublicInbox/WWW.pm b/lib/PublicInbox/WWW.pm
index c1c3926..3562e46 100644
--- a/lib/PublicInbox/WWW.pm
+++ b/lib/PublicInbox/WWW.pm
@@ -19,7 +19,9 @@ use URI::Escape qw(uri_unescape);
 use PublicInbox::MID qw(mid_escape);
 require PublicInbox::Git;
 use PublicInbox::GitHTTPBackend;
-our $INBOX_RE = qr!\A/([\w\.\-]+)!;
+
+# TODO: consider a routing tree now that we have more endpoints:
+our $INBOX_RE = qr!\A/([\w\-][\w\.\-]*)!;
 our $MID_RE = qr!([^/]+)!;
 our $END_RE = qr!(T/|t/|t\.mbox(?:\.gz)?|t\.atom|raw|)!;
 our $ATTACH_RE = qr!(\d[\.\d]*)-([[:alnum:]][\w\.-]+[[:alnum:]])!i;
diff --git a/t/config.t b/t/config.t
index 6a6b98c..5f0a95b 100644
--- a/t/config.t
+++ b/t/config.t
@@ -114,4 +114,40 @@ my $tmpdir = tempdir('pi-config-XXXXXX', TMPDIR => 1, CLEANUP => 1);
 		}, 'known addresses populated');
 }
 
+my @invalid = (
+	# git rejects this because it locks refnames, but we don't have
+	# this problem with inbox names:
+	# 'inbox.lock',
+
+	# git rejects these:
+	'', '..', '.', 'stash@{9}', 'inbox.', '^caret', '~tilde',
+	'*asterisk', 's p a c e s', ' leading-space', 'trailing-space ',
+	'question?', 'colon:', '[square-brace]', "\fformfeed",
+	"\0zero", "\bbackspace",
+
+);
+
+require Data::Dumper;
+for my $s (@invalid) {
+	my $d = Data::Dumper->new([$s])->Terse(1)->Indent(0)->Dump;
+	ok(!PublicInbox::Config::valid_inbox_name($s), "$d name rejected");
+}
+
+# obviously-valid examples
+my @valid = qw(a a@example a@example.com);
+
+# Rejecting more was considered, but then it dawned on me that
+# people may intentionally use inbox names which are not URL-friendly
+# to prevent the PSGI interface from displaying them...
+# URL-unfriendly
+# '<', '>', '%', '#', '?', '&', '(', ')',
+
+# maybe these aren't so bad, they're common in Message-IDs, even:
+# '!', '$', '=', '+'
+push @valid, qw[bang! ca$h less< more> 1% (parens) &more eql= +plus], '#hash';
+for my $s (@valid) {
+	my $d = Data::Dumper->new([$s])->Terse(1)->Indent(0)->Dump;
+	ok(PublicInbox::Config::valid_inbox_name($s), "$d name accepted");
+}
+
 done_testing();
-- 
EW


^ permalink raw reply	[flat|nested] 3+ messages in thread

* [RFC 2/2] www: add /~/$MESSAGE_ID global redirector endpoint
  2019-01-09 11:43 [RFC 0/2] support for /~/$MESSAGE_ID endpoint Eric Wong
  2019-01-09 11:43 ` [RFC 1/2] config: inbox name checking matches git.git more closely Eric Wong
@ 2019-01-09 11:43 ` Eric Wong
  1 sibling, 0 replies; 3+ messages in thread
From: Eric Wong @ 2019-01-09 11:43 UTC (permalink / raw)
  To: Konstantin Ryabitsev; +Cc: meta

The "/~/" is not finalized, yet.  Initially I chose "/_/", but
it could conflict with valid git remote names.

Perhaps even "/.$MESSAGE_ID" or "/~$MESSAGE_ID" could work to
save a byte.

Requested-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
cf. https://public-inbox.org/meta/20190107190719.GE9442@pure.paranoia.local/
---
 MANIFEST               |  1 +
 lib/PublicInbox/WWW.pm | 44 +++++++++++++++++++++++++--
 t/psgi_scan_all.t      | 69 ++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 111 insertions(+), 3 deletions(-)
 create mode 100644 t/psgi_scan_all.t

diff --git a/MANIFEST b/MANIFEST
index e4f3df8..73d1047 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -193,6 +193,7 @@ t/psgi_attach.t
 t/psgi_bad_mids.t
 t/psgi_mount.t
 t/psgi_multipart_not.t
+t/psgi_scan_all.t
 t/psgi_search.t
 t/psgi_text.t
 t/psgi_v2.t
diff --git a/lib/PublicInbox/WWW.pm b/lib/PublicInbox/WWW.pm
index 3562e46..9e0973f 100644
--- a/lib/PublicInbox/WWW.pm
+++ b/lib/PublicInbox/WWW.pm
@@ -360,11 +360,23 @@ sub legacy_redirects {
 		r301($ctx, $1, $2, $3 eq 't' ? 't/#u' : $3);
 	} elsif ($path_info =~ m!$INBOX_RE/(\S+/\S+)/f\z!o) {
 		r301($ctx, $1, $2);
+
+	# scan across all inboxes
+	# XXX '/~/$MESSAGE_ID' not finalized
+	} elsif ($path_info =~ m!\A/~/(\S+)\z!) {
+		scan_all($ctx, $1);
 	} else {
 		$ctx->{www}->news_www->call($ctx->{env});
 	}
 }
 
+sub redirect ($$) {
+	my ($code, $url) = @_;
+	[ $code,
+	  [ Location => $url, 'Content-Type' => 'text/plain' ],
+	  [ "Redirecting to $url\n" ] ]
+}
+
 sub r301 {
 	my ($ctx, $inbox, $mid_ue, $suffix) = @_;
 	my $obj = $ctx->{-inbox};
@@ -383,9 +395,7 @@ sub r301 {
 	$url .= $suffix if (defined $suffix);
 	$url .= "?$qs" if $qs ne '';
 
-	[ 301,
-	  [ Location => $url, 'Content-Type' => 'text/plain' ],
-	  [ "Redirecting to $url\n" ] ]
+	redirect(301, $url);
 }
 
 sub msg_page {
@@ -446,4 +456,32 @@ sub get_attach {
 	PublicInbox::WwwAttach::get_attach($ctx, $idx, $fn);
 }
 
+sub scan_all {
+	my ($ctx, $mid) = @_; # mid may have trailing slash
+
+	# TODO: user-sortable
+
+	my @found;
+	do {
+		$ctx->{www}->{pi_config}->each_inbox(sub {
+			my ($ibx) = @_;
+			# do not pass $env, since HTTP_HOST can be different
+			my $url = $ibx->base_url or next;
+
+			my $n = eval { $ibx->mm->num_for($mid) } or return;
+
+			# ambiguous, so 302 instead of 301:
+			push @found, redirect(302, $url .= "$mid/");
+		});
+
+	# account for trailing slash, since the rest of our API uses it
+	} while (!@found && $mid =~ s!/+\z!!);
+
+	# FIXME: It's possible for a message to have the same Message-ID but
+	# different content across multiple groups...
+	@found ? $found[0] : r404();
+
+	# n.b. we use trailing slash in most URLs to allow "wget -r" mirrors :)
+}
+
 1;
diff --git a/t/psgi_scan_all.t b/t/psgi_scan_all.t
new file mode 100644
index 0000000..bf03f22
--- /dev/null
+++ b/t/psgi_scan_all.t
@@ -0,0 +1,69 @@
+# Copyright (C) 2019 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+use strict;
+use warnings;
+use Test::More;
+use Email::MIME;
+use File::Temp qw/tempdir/;
+use PublicInbox::Config;
+my @mods = qw(HTTP::Request::Common Plack::Test URI::Escape Search::Xapian
+	DBD::SQLite);
+foreach my $mod (@mods) {
+	eval "require $mod";
+	plan skip_all => "$mod missing for psgi_scan_all.t" if $@;
+}
+use_ok 'PublicInbox::V2Writable';
+foreach my $mod (@mods) { use_ok $mod; }
+my $tmp = tempdir('pi-scan_all-XXXXXX', TMPDIR => 1, CLEANUP => 1);
+my $cfg = {};
+
+foreach my $i (1..2) {
+	my $cfgpfx = "publicinbox.test-$i";
+	my $addr = $cfg->{"$cfgpfx.address"} = "test-$i\@example.com";
+	my $mainrepo = $cfg->{"$cfgpfx.mainrepo"} = "$tmp/$i";
+	$cfg->{"$cfgpfx.url"} = "http://example.com/$i";
+	my $opt = {
+		mainrepo => $mainrepo,
+		name => "test-$i",
+		version => 2,
+		-primary_address => $addr,
+	};
+	my $ibx = PublicInbox::Inbox->new($opt);
+	my $im = PublicInbox::V2Writable->new($ibx, 1);
+	$im->{parallel} = 0;
+	$im->init_inbox(0);
+	my $mime = PublicInbox::MIME->new(<<EOF);
+From: a\@example.com
+To: $addr
+Subject: s$i
+Message-ID: <a-mid-$i\@b>
+Date: Fri, 02 Oct 1993 00:00:00 +0000
+
+hello world
+EOF
+
+	ok($im->add($mime), "added message to $i");
+	$im->done;
+}
+my $config = PublicInbox::Config->new($cfg);
+use_ok 'PublicInbox::WWW';
+my $www = PublicInbox::WWW->new($config);
+
+test_psgi(sub { $www->call(@_) }, sub {
+	my ($cb) = @_;
+	foreach my $i (1..2) {
+		foreach my $end ('', '/') {
+			my $res = $cb->(GET("/~/a-mid-$i\@b$end"));
+			is($res->code, 302, 'got 302');
+			is($res->header('Location'),
+				"http://example.com/$i/a-mid-$i\@b/",
+				"redirected OK to $i");
+		}
+	}
+	foreach my $x (qw(inv@lid inv@lid/ i/v/a l/i/d/)) {
+		my $res = $cb->(GET("/~/$x"));
+		is($res->code, 404, "404 on $x");
+	}
+});
+
+done_testing();
-- 
EW


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, back to index

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-01-09 11:43 [RFC 0/2] support for /~/$MESSAGE_ID endpoint Eric Wong
2019-01-09 11:43 ` [RFC 1/2] config: inbox name checking matches git.git more closely Eric Wong
2019-01-09 11:43 ` [RFC 2/2] www: add /~/$MESSAGE_ID global redirector endpoint Eric Wong

user/dev discussion of public-inbox itself

Archives are clonable:
	git clone --mirror https://public-inbox.org/meta
	git clone --mirror http://czquwvybam4bgbro.onion/meta
	git clone --mirror http://hjrcffqmbrq6wope.onion/meta
	git clone --mirror http://ou63pmih66umazou.onion/meta

Newsgroups are available over NNTP:
	nntp://news.public-inbox.org/inbox.comp.mail.public-inbox.meta
	nntp://ou63pmih66umazou.onion/inbox.comp.mail.public-inbox.meta
	nntp://czquwvybam4bgbro.onion/inbox.comp.mail.public-inbox.meta
	nntp://hjrcffqmbrq6wope.onion/inbox.comp.mail.public-inbox.meta
	nntp://news.gmane.org/gmane.mail.public-inbox.general

 note: .onion URLs require Tor: https://www.torproject.org/
       or Tor2web: https://www.tor2web.org/

AGPL code for this site: git clone https://public-inbox.org/ public-inbox