2 # Copyright (C) all contributors <meta@public-inbox.org>
3 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
6 use PublicInbox::TestCommon;
9 use PublicInbox::Config;
10 use PublicInbox::MID qw(mids);
11 require_mods(qw(DBD::SQLite Search::Xapian HTTP::Request::Common Plack::Test
12 URI::Escape Plack::Builder HTTP::Date));
13 use_ok($_) for (qw(HTTP::Request::Common Plack::Test));
14 use_ok 'PublicInbox::WWW';
15 my ($tmpdir, $for_destroy) = tmpdir();
16 my $enc_dup = 'ref-20150309094050.GO3427@x1.example';
18 my $dibx = create_inbox 'v2-dup', version => 2, indexlevel => 'medium',
19 tmpdir => "$tmpdir/dup", sub {
22 Date: Mon, 9 Mar 2015 09:40:50 +0000
26 Message-ID: <$enc_dup>
29 $im->add(PublicInbox::Eml->new($common.<<EOM)) or BAIL_OUT;
30 Content-Type: text/plain; charset=utf-8
31 Content-Disposition: inline
32 Content-Transfer-Encoding: 8bit
35 pipe \x{e2}\x{94}\x{82} or not
37 $im->add(PublicInbox::Eml->new($common.<<EOM)) or BAIL_OUT;
38 Content-Type: text/plain; charset="windows-1252"
39 Content-Transfer-Encoding: quoted-printable
46 my $eml = PublicInbox::Eml->new(<<'EOF');
47 From oldbug-pre-a0c07cba0e5d8b6a Fri Oct 2 00:00:00 1993
50 Subject: this is a subject
52 Date: Fri, 02 Oct 1993 00:00:00 +0000
53 Content-Type: text/plain; charset=iso-8859-1
58 my $ibx = create_inbox 'v2-1', version => 2, indexlevel => 'medium',
59 tmpdir => "$tmpdir/v2", sub {
61 $im->add($eml) or BAIL_OUT;
62 $eml->body_set("hello world!\n");
64 local $SIG{__WARN__} = sub { push @warn, @_ };
65 $eml->header_set(Date => 'Fri, 02 Oct 1993 00:01:00 +0000');
66 $im->add($eml) or BAIL_OUT;
67 is(scalar(@warn), 1, 'got one warning');
68 my $mids = mids($eml->header_obj);
69 $new_mid = $mids->[1];
70 open my $fh, '>', "$ibx->{inboxdir}/new_mid" or BAIL_OUT;
71 print $fh $new_mid or BAIL_OUT;
72 close $fh or BAIL_OUT;
75 open my $fh, '<', "$ibx->{inboxdir}/new_mid" or BAIL_OUT;
79 my $cfgpath = "$ibx->{inboxdir}/pi_config";
81 open my $fh, '>', $cfgpath or BAIL_OUT $!;
82 print $fh <<EOF or BAIL_OUT $!;
83 [publicinbox "v2test"]
84 inboxdir = $ibx->{inboxdir}
85 address = $ibx->{-primary_address}
87 inboxdir = $dibx->{inboxdir}
88 address = $dibx->{-primary_address}
90 close $fh or BAIL_OUT;
93 my $msg = $ibx->msg_by_mid('a-mid@b');
94 like($$msg, qr/\AFrom oldbug/s,
95 '"From_" line stored to test old bug workaround');
96 my $cfg = PublicInbox::Config->new($cfgpath);
97 my $www = PublicInbox::WWW->new($cfg);
98 my ($res, $raw, @from_);
101 $res = $cb->(GET('/v2test/description'));
102 like($res->content, qr!\$INBOX_DIR/description missing!,
103 'got v2 description missing message');
104 $res = $cb->(GET('/v2test/a-mid@b/raw'));
105 is($res->header('Content-Type'), 'text/plain; charset=iso-8859-1',
106 'charset from message used');
107 $raw = $res->content;
108 unlike($raw, qr/^From oldbug/sm, 'buggy "From_" line omitted');
109 like($raw, qr/^hello world$/m, 'got first message');
110 like($raw, qr/^hello world!$/m, 'got second message');
111 @from_ = ($raw =~ m/^From /mg);
112 is(scalar(@from_), 2, 'two From_ lines');
114 $res = $cb->(GET("/v2test/$new_mid/raw"));
115 $raw = $res->content;
116 like($raw, qr/^hello world!$/m, 'second message with new Message-Id');
117 @from_ = ($raw =~ m/^From /mg);
118 is(scalar(@from_), 1, 'only one From_ line');
120 # Atom feed should sort by Date: (if Received is missing)
121 $res = $cb->(GET('/v2test/new.atom'));
122 my @bodies = ($res->content =~ />(hello [^<]+)</mg);
123 is_deeply(\@bodies, [ "hello world!\n", "hello world\n" ],
124 'Atom ordering is chronological');
126 # new.html should sort by Date:, too (if Received is missing)
127 $res = $cb->(GET('/v2test/new.html'));
128 @bodies = ($res->content =~ /^(hello [^<]+)$/mg);
129 is_deeply(\@bodies, [ "hello world!\n", "hello world\n" ],
130 'new.html ordering is chronological');
132 $res = $cb->(GET('/v2test/new.atom'));
133 my @dates = ($res->content =~ m!title><updated>([^<]+)</updated>!g);
134 is_deeply(\@dates, [ "1993-10-02T00:01:00Z", "1993-10-02T00:00:00Z" ],
135 'Date headers made it through');
137 test_psgi(sub { $www->call(@_) }, $client0);
138 my $env = { TMPDIR => $tmpdir, PI_CONFIG => $cfgpath };
139 test_httpd($env, $client0, 9);
141 $eml->header_set('Message-ID', 'a-mid@b');
142 $eml->body_set("hello ghosts\n");
143 my $im = $ibx->importer(0);
146 local $SIG{__WARN__} = sub { push @warn, @_ };
147 ok($im->add($eml), 'added 3rd duplicate-but-different message');
148 is(scalar(@warn), 1, 'got another warning');
149 like($warn[0], qr/mismatched/, 'warned about mismatched messages');
151 my $mids = mids($eml->header_obj);
152 my $third = $mids->[-1];
157 $res = $cb->(GET('/v2test/_/text/config/raw'));
158 my $lm = $res->header('Last-Modified');
159 ok($lm, 'Last-Modified set w/ ->mm');
160 $lm = HTTP::Date::str2time($lm);
161 is($lm, $ibx->mm->created_at,
162 'Last-Modified for text/config/raw matches ->created_at');
165 $res = $cb->(GET("/v2test/$third/raw"));
166 $raw = $res->content;
167 like($raw, qr/^hello ghosts$/m, 'got third message');
168 @from_ = ($raw =~ m/^From /mg);
169 is(scalar(@from_), 1, 'one From_ line');
171 $res = $cb->(GET('/v2test/a-mid@b/raw'));
172 $raw = $res->content;
173 like($raw, qr/^hello world$/m, 'got first message');
174 like($raw, qr/^hello world!$/m, 'got second message');
175 like($raw, qr/^hello ghosts$/m, 'got third message');
176 @from_ = ($raw =~ m/^From /mg);
177 is(scalar(@from_), 3, 'three From_ lines');
178 $cfg->each_inbox(sub { $_[0]->search->reopen });
181 eval { require IO::Uncompress::Gunzip };
182 skip 'IO::Uncompress::Gunzip missing', 6 if $@;
183 my ($in, $out, $status);
184 my $req = GET('/v2test/a-mid@b/raw');
185 $req->header('Accept-Encoding' => 'gzip');
187 is($res->header('Content-Encoding'), 'gzip', 'gzip encoding');
189 IO::Uncompress::Gunzip::gunzip(\$in => \$out);
190 is($out, $raw, 'gzip response matches');
192 $res = $cb->(GET('/v2test/a-mid@b/t.mbox.gz'));
194 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
195 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
196 like($out, qr/^hello world$/m, 'got first in t.mbox.gz');
197 like($out, qr/^hello world!$/m, 'got second in t.mbox.gz');
198 like($out, qr/^hello ghosts$/m, 'got third in t.mbox.gz');
199 @from_ = ($out =~ m/^From /mg);
200 is(scalar(@from_), 3, 'three From_ lines in t.mbox.gz');
203 $res = $cb->(POST('/v2test/?q=m:a-mid@b&x=m'));
205 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
206 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
207 like($out, qr/^hello world$/m, 'got first in mbox POST');
208 like($out, qr/^hello world!$/m, 'got second in mbox POST');
209 like($out, qr/^hello ghosts$/m, 'got third in mbox POST');
210 @from_ = ($out =~ m/^From /mg);
211 is(scalar(@from_), 3, 'three From_ lines in mbox POST');
213 # all.mbox.gz interface
214 $res = $cb->(GET('/v2test/all.mbox.gz'));
216 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
217 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
218 like($out, qr/^hello world$/m, 'got first in all.mbox');
219 like($out, qr/^hello world!$/m, 'got second in all.mbox');
220 like($out, qr/^hello ghosts$/m, 'got third in all.mbox');
221 @from_ = ($out =~ m/^From /mg);
222 is(scalar(@from_), 3, 'three From_ lines in all.mbox');
225 $res = $cb->(GET('/v2test/?q=m:a-mid@b&x=t'));
226 is($res->code, 200, 'success with threaded search');
227 my $raw = $res->content;
228 ok($raw =~ s/\A.*>Results 1-3 of 3\b//s, 'got all results');
229 my @over = ($raw =~ m/\d{4}-\d+-\d+\s+\d+:\d+ +(?:\d+\% )?(.+)$/gm);
230 is_deeply(\@over, [ '<a', '` <a', '` <a' ], 'threaded messages show up');
232 $res = $cb->(GET('/v2test/?q=m:a-mid@b&x=A'));
233 is($res->code, 200, 'success with Atom search');
235 require_mods(qw(XML::TreePP), 2);
236 my $t = XML::TreePP->new->parse($res->content);
237 like($t->{feed}->{-xmlns}, qr/\bAtom\b/,
238 'looks like an an Atom feed');
239 is(scalar @{$t->{feed}->{entry}}, 3, 'parsed three entries');
242 local $SIG{__WARN__} = 'DEFAULT';
243 $res = $cb->(GET('/v2test/a-mid@b/'));
244 $raw = $res->content;
245 like($raw, qr/WARNING: multiple messages have this Message-ID/,
246 'warned about duplicate Message-IDs');
247 like($raw, qr/^hello world$/m, 'got first message');
248 like($raw, qr/^hello world!$/m, 'got second message');
249 like($raw, qr/^hello ghosts$/m, 'got third message');
250 @from_ = ($raw =~ m/>From: /mg);
251 is(scalar(@from_), 3, 'three From: lines');
252 foreach my $mid ('a-mid@b', $new_mid, $third) {
253 like($raw, qr!>\Q$mid\E</a>!s, "Message-ID $mid shown");
255 like($raw, qr/\b3\+ messages\b/, 'thread overview shown');
257 $res = $cb->(GET("/dup/$enc_dup/d/"));
258 is($res->code, 200, '/d/ (diff) endpoint works');
259 $raw = $res->content;
260 like($raw, qr!</span> cr_mismatch\n!s,
261 'cr_mismatch is only diff context');
262 like($raw, qr!>\-pipe !s, 'pipe diff del line');
263 like($raw, qr!>\+pipe !s, 'pipe diff ins line');
266 test_psgi(sub { $www->call(@_) }, $client1);
267 test_httpd($env, $client1, 38);
270 my $exp = [ qw(<a-mid@b> <reuse@mid>) ];
271 $eml->header_set('Message-Id', @$exp);
272 $eml->header_set('Subject', '4th dupe');
273 local $SIG{__WARN__} = sub {};
274 ok($im->add($eml), 'added one message');
276 my @h = $eml->header('Message-ID');
277 is_deeply($exp, \@h, 'reused existing Message-ID');
278 $cfg->each_inbox(sub { $_[0]->search->reopen });
283 my $res = $cb->(GET('/v2test/new.atom'));
284 my @ids = ($res->content =~ m!<id>urn:uuid:([^<]+)</id>!sg);
287 is_deeply([qw(1 1 1 1)], [values %ids], 'feed ids unique');
289 $res = $cb->(GET('/v2test/reuse@mid/T/'));
290 $raw = $res->content;
291 like($raw, qr/\b4\+ messages\b/, 'thread overview shown with /T/');
292 my @over = ($raw =~ m/^\d{4}-\d+-\d+\s+\d+:\d+ (.+)$/gm);
293 is_deeply(\@over, [ '<a', '` <a', '` <a', '` <a' ],
294 'duplicate messages share the same root');
296 $res = $cb->(GET('/v2test/reuse@mid/t/'));
297 $raw = $res->content;
298 like($raw, qr/\b4\+ messages\b/, 'thread overview shown with /t/');
300 $res = $cb->(GET('/v2test/0/info/refs'));
301 is($res->code, 200, 'got info refs for dumb clones');
302 $res = $cb->(GET('/v2test/0.git/info/refs'));
303 is($res->code, 200, 'got info refs for dumb clones w/ .git suffix');
304 $res = $cb->(GET('/v2test/info/refs'));
305 is($res->code, 404, 'v2 git URL w/o shard fails');
308 test_psgi(sub { $www->call(@_) }, $client2);
309 test_httpd($env, $client2, 8);
311 # ensure conflicted attachments can be resolved
312 local $SIG{__WARN__} = sub {};
313 foreach my $body (qw(old new)) {
314 $im->add(eml_load "t/psgi_v2-$body.eml") or BAIL_OUT;
318 $cfg->each_inbox(sub { $_[0]->search->reopen });
322 my $res = $cb->(GET('/v2test/a@dup/'));
323 my @links = ($res->content =~ m!"\.\./([^/]+/2-attach\.txt)\"!g);
324 is(scalar(@links), 2, 'both attachment links exist');
325 isnt($links[0], $links[1], 'attachment links are different');
327 my $old = $cb->(GET('/v2test/' . $links[0]));
328 my $new = $cb->(GET('/v2test/' . $links[1]));
329 is($old->content, 'old', 'got expected old content');
330 is($new->content, 'new', 'got expected new content');
332 $res = $cb->(GET('/v2test/?t=1970'.'01'.'01'.'000000'));
333 is($res->code, 404, '404 for out-of-range t= param');
335 local $SIG{__WARN__} = sub { push @warn, @_ };
336 $res = $cb->(GET('/v2test/?t=1970'.'01'.'01'));
337 is_deeply(\@warn, [], 'no warnings on YYYYMMDD only');
339 test_psgi(sub { $www->call(@_) }, $client3);
340 test_httpd($env, $client3, 4);