closes #5826
[arvados.git] / services / api / test / unit / collection_test.rb
1 require 'test_helper'
2
3 class CollectionTest < ActiveSupport::TestCase
4   def create_collection name, enc=nil
5     txt = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:#{name}.txt\n"
6     txt.force_encoding(enc) if enc
7     return Collection.create(manifest_text: txt)
8   end
9
10   test 'accept ASCII manifest_text' do
11     act_as_system_user do
12       c = create_collection 'foo', Encoding::US_ASCII
13       assert c.valid?
14     end
15   end
16
17   test 'accept UTF-8 manifest_text' do
18     act_as_system_user do
19       c = create_collection "f\xc3\x98\xc3\x98", Encoding::UTF_8
20       assert c.valid?
21     end
22   end
23
24   test 'refuse manifest_text with invalid UTF-8 byte sequence' do
25     act_as_system_user do
26       c = create_collection "f\xc8o", Encoding::UTF_8
27       assert !c.valid?
28       assert_equal [:manifest_text], c.errors.messages.keys
29       assert_match /UTF-8/, c.errors.messages[:manifest_text].first
30     end
31   end
32
33   test 'refuse manifest_text with non-UTF-8 encoding' do
34     act_as_system_user do
35       c = create_collection "f\xc8o", Encoding::ASCII_8BIT
36       assert !c.valid?
37       assert_equal [:manifest_text], c.errors.messages.keys
38       assert_match /UTF-8/, c.errors.messages[:manifest_text].first
39     end
40   end
41
42   test 'create and update collection and verify file_names' do
43     act_as_system_user do
44       c = create_collection 'foo', Encoding::US_ASCII
45       assert c.valid?
46       created_file_names = c.file_names
47       assert created_file_names
48       assert_match /foo.txt/, c.file_names
49
50       c.update_attribute 'manifest_text', ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo2.txt\n"
51       assert_not_equal created_file_names, c.file_names
52       assert_match /foo2.txt/, c.file_names
53     end
54   end
55
56   [
57     [2**8, false],
58     [2**18, true],
59   ].each do |manifest_size, gets_truncated|
60     test "create collection with manifest size #{manifest_size} which gets truncated #{gets_truncated},
61           and not expect exceptions even on very large manifest texts" do
62       # file_names has a max size, hence there will be no errors even on large manifests
63       act_as_system_user do
64         manifest_text = './blurfl d41d8cd98f00b204e9800998ecf8427e+0'
65         index = 0
66         while manifest_text.length < manifest_size
67           manifest_text += ' ' + "0:0:veryverylongfilename000000000000#{index}.txt\n./subdir1"
68           index += 1
69         end
70         manifest_text += "\n"
71         c = Collection.create(manifest_text: manifest_text)
72
73         assert c.valid?
74         assert c.file_names
75         assert_match /veryverylongfilename0000000000001.txt/, c.file_names
76         assert_match /veryverylongfilename0000000000002.txt/, c.file_names
77         if !gets_truncated
78           assert_match /blurfl/, c.file_names
79           assert_match /subdir1/, c.file_names
80         end
81       end
82     end
83   end
84
85   test "full text search for collections" do
86     # file_names column does not get populated when fixtures are loaded, hence setup test data
87     act_as_system_user do
88       Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n")
89       Collection.create(manifest_text: ". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n")
90       Collection.create(manifest_text: ". 85877ca2d7e05498dd3d109baf2df106+95+A3a4e26a366ee7e4ed3e476ccf05354761be2e4ae@545a9920 0:95:file_in_subdir1\n./subdir2/subdir3 2bbc341c702df4d8f42ec31f16c10120+64+A315d7e7bad2ce937e711fc454fae2d1194d14d64@545a9920 0:32:file1.txt 32:32:file2.txt\n./subdir2/subdir3/subdir4 2bbc341c702df4d8f42ec31f16c10120+64+A315d7e7bad2ce937e711fc454fae2d1194d14d64@545a9920 0:32:file3.txt 32:32:file4.txt")
91     end
92
93     [
94       ['foo', true],
95       ['foo bar', false],                     # no collection matching both
96       ['foo&bar', false],                     # no collection matching both
97       ['foo|bar', true],                      # works only no spaces between the words
98       ['Gnu public', true],                   # both prefixes found, though not consecutively
99       ['Gnu&public', true],                   # both prefixes found, though not consecutively
100       ['file4', true],                        # prefix match
101       ['file4.txt', true],                    # whole string match
102       ['filex', false],                       # no such prefix
103       ['subdir', true],                       # prefix matches
104       ['subdir2', true],
105       ['subdir2/', true],
106       ['subdir2/subdir3', true],
107       ['subdir2/subdir3/subdir4', true],
108       ['subdir2 file4', true],                # look for both prefixes
109       ['subdir4', false],                     # not a prefix match
110     ].each do |search_filter, expect_results|
111       search_filters = search_filter.split.each {|s| s.concat(':*')}.join('&')
112       results = Collection.where("#{Collection.full_text_tsvector} @@ to_tsquery(?)",
113                                  "#{search_filters}")
114       if expect_results
115         refute_empty results
116       else
117         assert_empty results
118       end
119     end
120   end
121
122   test 'portable data hash with missing size hints' do
123     [[". d41d8cd98f00b204e9800998ecf8427e+0+Bar 0:0:x",
124       ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x"],
125      [". d41d8cd98f00b204e9800998ecf8427e+Foo 0:0:x",
126       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x"],
127      [". d41d8cd98f00b204e9800998ecf8427e 0:0:x",
128       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x"],
129     ].each do |unportable, portable|
130       c = Collection.new(manifest_text: unportable)
131       assert c.valid?
132       assert_equal(Digest::MD5.hexdigest(portable)+"+#{portable.length}",
133                    c.portable_data_hash)
134     end
135   end
136
137   [0, 2, 4, nil].each do |ask|
138     test "set replication_desired to #{ask.inspect}" do
139       Rails.configuration.default_collection_replication = 2
140       act_as_user users(:active) do
141         c = collections(:replication_undesired_unconfirmed)
142         c.update_attributes replication_desired: ask
143         assert_equal ask, c.replication_desired
144       end
145     end
146   end
147
148   test "replication_confirmed* can be set by admin user" do
149     c = collections(:replication_desired_2_unconfirmed)
150     act_as_user users(:admin) do
151       assert c.update_attributes(replication_confirmed: 2,
152                                  replication_confirmed_at: Time.now)
153     end
154   end
155
156   test "replication_confirmed* cannot be set by non-admin user" do
157     act_as_user users(:active) do
158       c = collections(:replication_desired_2_unconfirmed)
159       # Cannot set just one at a time.
160       assert_raise ArvadosModel::PermissionDeniedError do
161         c.update_attributes replication_confirmed: 1
162       end
163       assert_raise ArvadosModel::PermissionDeniedError do
164         c.update_attributes replication_confirmed_at: Time.now
165       end
166       # Cannot set both at once, either.
167       assert_raise ArvadosModel::PermissionDeniedError do
168         c.update_attributes(replication_confirmed: 1,
169                             replication_confirmed_at: Time.now)
170       end
171     end
172   end
173
174   test "replication_confirmed* can be cleared (but only together) by non-admin user" do
175     act_as_user users(:active) do
176       c = collections(:replication_desired_2_confirmed_2)
177       # Cannot clear just one at a time.
178       assert_raise ArvadosModel::PermissionDeniedError do
179         c.update_attributes replication_confirmed: nil
180       end
181       c.reload
182       assert_raise ArvadosModel::PermissionDeniedError do
183         c.update_attributes replication_confirmed_at: nil
184       end
185       # Can clear both at once.
186       c.reload
187       assert c.update_attributes(replication_confirmed: nil,
188                                  replication_confirmed_at: nil)
189     end
190   end
191
192   test "clear replication_confirmed* when introducing a new block in manifest" do
193     c = collections(:replication_desired_2_confirmed_2)
194     act_as_user users(:active) do
195       assert c.update_attributes(manifest_text: collections(:user_agreement).signed_manifest_text)
196       assert_nil c.replication_confirmed
197       assert_nil c.replication_confirmed_at
198     end
199   end
200
201   test "don't clear replication_confirmed* when just renaming a file" do
202     c = collections(:replication_desired_2_confirmed_2)
203     act_as_user users(:active) do
204       new_manifest = c.signed_manifest_text.sub(':bar', ':foo')
205       assert c.update_attributes(manifest_text: new_manifest)
206       assert_equal 2, c.replication_confirmed
207       assert_not_nil c.replication_confirmed_at
208     end
209   end
210
211   test "don't clear replication_confirmed* when just deleting a data block" do
212     c = collections(:replication_desired_2_confirmed_2)
213     act_as_user users(:active) do
214       new_manifest = c.signed_manifest_text
215       new_manifest.sub!(/ \S+:bar/, '')
216       new_manifest.sub!(/ acbd\S+/, '')
217
218       # Confirm that we did just remove a block from the manifest (if
219       # not, this test would pass without testing the relevant case):
220       assert_operator new_manifest.length+40, :<, c.signed_manifest_text.length
221
222       assert c.update_attributes(manifest_text: new_manifest)
223       assert_equal 2, c.replication_confirmed
224       assert_not_nil c.replication_confirmed_at
225     end
226   end
227
228   test "create collection with properties" do
229     act_as_system_user do
230       c = Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n",
231                             properties: {'property_1' => 'value_1'})
232       assert c.valid?
233       assert_equal 'value_1', c.properties['property_1']
234     end
235   end
236
237   test 'create, delete, recreate collection with same name and owner' do
238     act_as_user users(:active) do
239       # create collection with name
240       c = Collection.create(manifest_text: '',
241                             name: "test collection name")
242       assert c.valid?
243       uuid = c.uuid
244
245       # mark collection as expired
246       c.update_attribute 'expires_at', Time.new.strftime("%Y-%m-%d")
247       c = Collection.where(uuid: uuid)
248       assert_empty c, 'Should not be able to find expired collection'
249
250       # recreate collection with the same name
251       c = Collection.create(manifest_text: '',
252                             name: "test collection name")
253       assert c.valid?
254     end
255   end
256
257   test "find_all_for_docker_image resolves names that look like hashes" do
258     coll_list = Collection.
259       find_all_for_docker_image('a' * 64, nil, [users(:active)])
260     coll_uuids = coll_list.map(&:uuid)
261     assert_includes(coll_uuids, collections(:docker_image).uuid)
262   end
263 end