2 # Copyright (C) 2018-2021 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 $eml = PublicInbox::Eml->new(<<'EOF');
17 From oldbug-pre-a0c07cba0e5d8b6a Fri Oct 2 00:00:00 1993
20 Subject: this is a subject
22 Date: Fri, 02 Oct 1993 00:00:00 +0000
27 my $ibx = create_inbox 'v2', version => 2, indexlevel => 'medium',
28 tmpdir => "$tmpdir/v2", sub {
30 $im->add($eml) or BAIL_OUT;
31 $eml->body_set("hello world!\n");
33 local $SIG{__WARN__} = sub { push @warn, @_ };
34 $eml->header_set(Date => 'Fri, 02 Oct 1993 00:01:00 +0000');
35 $im->add($eml) or BAIL_OUT;
36 is(scalar(@warn), 1, 'got one warning');
37 my $mids = mids($eml->header_obj);
38 $new_mid = $mids->[1];
39 open my $fh, '>', "$ibx->{inboxdir}/new_mid" or BAIL_OUT;
40 print $fh $new_mid or BAIL_OUT;
41 close $fh or BAIL_OUT;
44 open my $fh, '<', "$ibx->{inboxdir}/new_mid" or BAIL_OUT;
48 my $cfgpath = "$ibx->{inboxdir}/pi_config";
50 open my $fh, '>', $cfgpath or BAIL_OUT $!;
51 print $fh <<EOF or BAIL_OUT $!;
52 [publicinbox "v2test"]
53 inboxdir = $ibx->{inboxdir}
54 address = $ibx->{-primary_address}
56 close $fh or BAIL_OUT;
59 my $msg = $ibx->msg_by_mid('a-mid@b');
60 like($$msg, qr/\AFrom oldbug/s,
61 '"From_" line stored to test old bug workaround');
62 my $cfg = PublicInbox::Config->new($cfgpath);
63 my $www = PublicInbox::WWW->new($cfg);
64 my ($res, $raw, @from_);
67 $res = $cb->(GET('/v2test/description'));
68 like($res->content, qr!\$INBOX_DIR/description missing!,
69 'got v2 description missing message');
70 $res = $cb->(GET('/v2test/a-mid@b/raw'));
72 unlike($raw, qr/^From oldbug/sm, 'buggy "From_" line omitted');
73 like($raw, qr/^hello world$/m, 'got first message');
74 like($raw, qr/^hello world!$/m, 'got second message');
75 @from_ = ($raw =~ m/^From /mg);
76 is(scalar(@from_), 2, 'two From_ lines');
78 $res = $cb->(GET("/v2test/$new_mid/raw"));
80 like($raw, qr/^hello world!$/m, 'second message with new Message-Id');
81 @from_ = ($raw =~ m/^From /mg);
82 is(scalar(@from_), 1, 'only one From_ line');
84 # Atom feed should sort by Date: (if Received is missing)
85 $res = $cb->(GET('/v2test/new.atom'));
86 my @bodies = ($res->content =~ />(hello [^<]+)</mg);
87 is_deeply(\@bodies, [ "hello world!\n", "hello world\n" ],
88 'Atom ordering is chronological');
90 # new.html should sort by Date:, too (if Received is missing)
91 $res = $cb->(GET('/v2test/new.html'));
92 @bodies = ($res->content =~ /^(hello [^<]+)$/mg);
93 is_deeply(\@bodies, [ "hello world!\n", "hello world\n" ],
94 'new.html ordering is chronological');
96 $res = $cb->(GET('/v2test/new.atom'));
97 my @dates = ($res->content =~ m!title><updated>([^<]+)</updated>!g);
98 is_deeply(\@dates, [ "1993-10-02T00:01:00Z", "1993-10-02T00:00:00Z" ],
99 'Date headers made it through');
101 test_psgi(sub { $www->call(@_) }, $client0);
102 my $env = { TMPDIR => $tmpdir, PI_CONFIG => $cfgpath };
103 test_httpd($env, $client0, 9);
105 $eml->header_set('Message-ID', 'a-mid@b');
106 $eml->body_set("hello ghosts\n");
107 my $im = $ibx->importer(0);
110 local $SIG{__WARN__} = sub { push @warn, @_ };
111 ok($im->add($eml), 'added 3rd duplicate-but-different message');
112 is(scalar(@warn), 1, 'got another warning');
113 like($warn[0], qr/mismatched/, 'warned about mismatched messages');
115 my $mids = mids($eml->header_obj);
116 my $third = $mids->[-1];
121 $res = $cb->(GET('/v2test/_/text/config/raw'));
122 my $lm = $res->header('Last-Modified');
123 ok($lm, 'Last-Modified set w/ ->mm');
124 $lm = HTTP::Date::str2time($lm);
125 is($lm, $ibx->mm->created_at,
126 'Last-Modified for text/config/raw matches ->created_at');
129 $res = $cb->(GET("/v2test/$third/raw"));
130 $raw = $res->content;
131 like($raw, qr/^hello ghosts$/m, 'got third message');
132 @from_ = ($raw =~ m/^From /mg);
133 is(scalar(@from_), 1, 'one From_ line');
135 $res = $cb->(GET('/v2test/a-mid@b/raw'));
136 $raw = $res->content;
137 like($raw, qr/^hello world$/m, 'got first message');
138 like($raw, qr/^hello world!$/m, 'got second message');
139 like($raw, qr/^hello ghosts$/m, 'got third message');
140 @from_ = ($raw =~ m/^From /mg);
141 is(scalar(@from_), 3, 'three From_ lines');
142 $cfg->each_inbox(sub { $_[0]->search->reopen });
145 eval { require IO::Uncompress::Gunzip };
146 skip 'IO::Uncompress::Gunzip missing', 6 if $@;
147 my ($in, $out, $status);
148 my $req = GET('/v2test/a-mid@b/raw');
149 $req->header('Accept-Encoding' => 'gzip');
151 is($res->header('Content-Encoding'), 'gzip', 'gzip encoding');
153 IO::Uncompress::Gunzip::gunzip(\$in => \$out);
154 is($out, $raw, 'gzip response matches');
156 $res = $cb->(GET('/v2test/a-mid@b/t.mbox.gz'));
158 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
159 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
160 like($out, qr/^hello world$/m, 'got first in t.mbox.gz');
161 like($out, qr/^hello world!$/m, 'got second in t.mbox.gz');
162 like($out, qr/^hello ghosts$/m, 'got third in t.mbox.gz');
163 @from_ = ($out =~ m/^From /mg);
164 is(scalar(@from_), 3, 'three From_ lines in t.mbox.gz');
167 $res = $cb->(POST('/v2test/?q=m:a-mid@b&x=m'));
169 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
170 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
171 like($out, qr/^hello world$/m, 'got first in mbox POST');
172 like($out, qr/^hello world!$/m, 'got second in mbox POST');
173 like($out, qr/^hello ghosts$/m, 'got third in mbox POST');
174 @from_ = ($out =~ m/^From /mg);
175 is(scalar(@from_), 3, 'three From_ lines in mbox POST');
177 # all.mbox.gz interface
178 $res = $cb->(GET('/v2test/all.mbox.gz'));
180 $status = IO::Uncompress::Gunzip::gunzip(\$in => \$out);
181 unlike($out, qr/^From oldbug/sm, 'buggy "From_" line omitted');
182 like($out, qr/^hello world$/m, 'got first in all.mbox');
183 like($out, qr/^hello world!$/m, 'got second in all.mbox');
184 like($out, qr/^hello ghosts$/m, 'got third in all.mbox');
185 @from_ = ($out =~ m/^From /mg);
186 is(scalar(@from_), 3, 'three From_ lines in all.mbox');
189 $res = $cb->(GET('/v2test/?q=m:a-mid@b&x=t'));
190 is($res->code, 200, 'success with threaded search');
191 my $raw = $res->content;
192 ok($raw =~ s/\A.*>Results 1-3 of 3\b//s, 'got all results');
193 my @over = ($raw =~ m/\d{4}-\d+-\d+\s+\d+:\d+ +(?:\d+\% )?(.+)$/gm);
194 is_deeply(\@over, [ '<a', '` <a', '` <a' ], 'threaded messages show up');
196 $res = $cb->(GET('/v2test/?q=m:a-mid@b&x=A'));
197 is($res->code, 200, 'success with Atom search');
199 require_mods(qw(XML::TreePP), 2);
200 my $t = XML::TreePP->new->parse($res->content);
201 like($t->{feed}->{-xmlns}, qr/\bAtom\b/,
202 'looks like an an Atom feed');
203 is(scalar @{$t->{feed}->{entry}}, 3, 'parsed three entries');
206 local $SIG{__WARN__} = 'DEFAULT';
207 $res = $cb->(GET('/v2test/a-mid@b/'));
208 $raw = $res->content;
209 like($raw, qr/^hello world$/m, 'got first message');
210 like($raw, qr/^hello world!$/m, 'got second message');
211 like($raw, qr/^hello ghosts$/m, 'got third message');
212 @from_ = ($raw =~ m/>From: /mg);
213 is(scalar(@from_), 3, 'three From: lines');
214 foreach my $mid ('a-mid@b', $new_mid, $third) {
215 like($raw, qr!>\Q$mid\E</a>!s, "Message-ID $mid shown");
217 like($raw, qr/\b3\+ messages\b/, 'thread overview shown');
220 test_psgi(sub { $www->call(@_) }, $client1);
221 test_httpd($env, $client1, 38);
224 my $exp = [ qw(<a-mid@b> <reuse@mid>) ];
225 $eml->header_set('Message-Id', @$exp);
226 $eml->header_set('Subject', '4th dupe');
227 local $SIG{__WARN__} = sub {};
228 ok($im->add($eml), 'added one message');
230 my @h = $eml->header('Message-ID');
231 is_deeply($exp, \@h, 'reused existing Message-ID');
232 $cfg->each_inbox(sub { $_[0]->search->reopen });
237 my $res = $cb->(GET('/v2test/new.atom'));
238 my @ids = ($res->content =~ m!<id>urn:uuid:([^<]+)</id>!sg);
241 is_deeply([qw(1 1 1 1)], [values %ids], 'feed ids unique');
243 $res = $cb->(GET('/v2test/reuse@mid/T/'));
244 $raw = $res->content;
245 like($raw, qr/\b4\+ messages\b/, 'thread overview shown with /T/');
246 my @over = ($raw =~ m/^\d{4}-\d+-\d+\s+\d+:\d+ (.+)$/gm);
247 is_deeply(\@over, [ '<a', '` <a', '` <a', '` <a' ],
248 'duplicate messages share the same root');
250 $res = $cb->(GET('/v2test/reuse@mid/t/'));
251 $raw = $res->content;
252 like($raw, qr/\b4\+ messages\b/, 'thread overview shown with /t/');
254 $res = $cb->(GET('/v2test/0/info/refs'));
255 is($res->code, 200, 'got info refs for dumb clones');
256 $res = $cb->(GET('/v2test/0.git/info/refs'));
257 is($res->code, 200, 'got info refs for dumb clones w/ .git suffix');
258 $res = $cb->(GET('/v2test/info/refs'));
259 is($res->code, 404, 'v2 git URL w/o shard fails');
262 test_psgi(sub { $www->call(@_) }, $client2);
263 test_httpd($env, $client2, 8);
265 # ensure conflicted attachments can be resolved
266 local $SIG{__WARN__} = sub {};
267 foreach my $body (qw(old new)) {
268 $im->add(eml_load "t/psgi_v2-$body.eml") or BAIL_OUT;
272 $cfg->each_inbox(sub { $_[0]->search->reopen });
276 my $res = $cb->(GET('/v2test/a@dup/'));
277 my @links = ($res->content =~ m!"\.\./([^/]+/2-attach\.txt)\"!g);
278 is(scalar(@links), 2, 'both attachment links exist');
279 isnt($links[0], $links[1], 'attachment links are different');
281 my $old = $cb->(GET('/v2test/' . $links[0]));
282 my $new = $cb->(GET('/v2test/' . $links[1]));
283 is($old->content, 'old', 'got expected old content');
284 is($new->content, 'new', 'got expected new content');
286 $res = $cb->(GET('/v2test/?t=1970'.'01'.'01'.'000000'));
287 is($res->code, 404, '404 for out-of-range t= param');
289 local $SIG{__WARN__} = sub { push @warn, @_ };
290 $res = $cb->(GET('/v2test/?t=1970'.'01'.'01'));
291 is_deeply(\@warn, [], 'no warnings on YYYYMMDD only');
293 test_psgi(sub { $www->call(@_) }, $client3);
294 test_httpd($env, $client3, 4);