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   [
  43     ". 0:0:foo.txt",
  44     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  45     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  46     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  47   ].each do |manifest_text|
  48     test "create collection with invalid manifest text #{manifest_text} and expect error" do
  49       act_as_system_user do
  50         c = Collection.create(manifest_text: manifest_text)
  51         assert !c.valid?
  52       end
  53     end
  54   end
  55
  56   [
  57     nil,
  58     "",
  59     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  60   ].each do |manifest_text|
  61     test "create collection with valid manifest text #{manifest_text.inspect} and expect success" do
  62       act_as_system_user do
  63         c = Collection.create(manifest_text: manifest_text)
  64         assert c.valid?
  65       end
  66     end
  67   end
  68
  69   [
  70     ". 0:0:foo.txt",
  71     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  72     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  73     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  74   ].each do |manifest_text|
  75     test "update collection with invalid manifest text #{manifest_text} and expect error" do
  76       act_as_system_user do
  77         c = create_collection 'foo', Encoding::US_ASCII
  78         assert c.valid?
  79
  80         c.update_attribute 'manifest_text', manifest_text
  81         assert !c.valid?
  82       end
  83     end
  84   end
  85
  86   [
  87     nil,
  88     "",
  89     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  90   ].each do |manifest_text|
  91     test "update collection with valid manifest text #{manifest_text.inspect} and expect success" do
  92       act_as_system_user do
  93         c = create_collection 'foo', Encoding::US_ASCII
  94         assert c.valid?
  95
  96         c.update_attribute 'manifest_text', manifest_text
  97         assert c.valid?
  98       end
  99     end
 100   end
 101
 102   test 'create and update collection and verify file_names' do
 103     act_as_system_user do
 104       c = create_collection 'foo', Encoding::US_ASCII
 105       assert c.valid?
 106       created_file_names = c.file_names
 107       assert created_file_names
 108       assert_match /foo.txt/, c.file_names
 109
 110       c.update_attribute 'manifest_text', ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo2.txt\n"
 111       assert_not_equal created_file_names, c.file_names
 112       assert_match /foo2.txt/, c.file_names
 113     end
 114   end
 115
 116   [
 117     [2**8, false],
 118     [2**18, true],
 119   ].each do |manifest_size, allow_truncate|
 120     test "create collection with manifest size #{manifest_size} with allow_truncate=#{allow_truncate},
 121           and not expect exceptions even on very large manifest texts" do
 122       # file_names has a max size, hence there will be no errors even on large manifests
 123       act_as_system_user do
 124         manifest_text = ''
 125         index = 0
 126         while manifest_text.length < manifest_size
 127           manifest_text += "./blurfl d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylongfilename000000000000#{index}.txt\n"
 128           index += 1
 129         end
 130         manifest_text += "./laststreamname d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylastfilename.txt\n"
 131         c = Collection.create(manifest_text: manifest_text)
 132
 133         assert c.valid?
 134         assert c.file_names
 135         assert_match /veryverylongfilename0000000000001.txt/, c.file_names
 136         assert_match /veryverylongfilename0000000000002.txt/, c.file_names
 137         if not allow_truncate
 138           assert_match /veryverylastfilename/, c.file_names
 139           assert_match /laststreamname/, c.file_names
 140         end
 141       end
 142     end
 143   end
 144
 145   test "full text search for collections" do
 146     # file_names column does not get populated when fixtures are loaded, hence setup test data
 147     act_as_system_user do
 148       Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n")
 149       Collection.create(manifest_text: ". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n")
 150       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\n")
 151     end
 152
 153     [
 154       ['foo', true],
 155       ['foo bar', false],                     # no collection matching both
 156       ['foo&bar', false],                     # no collection matching both
 157       ['foo|bar', true],                      # works only no spaces between the words
 158       ['Gnu public', true],                   # both prefixes found, though not consecutively
 159       ['Gnu&public', true],                   # both prefixes found, though not consecutively
 160       ['file4', true],                        # prefix match
 161       ['file4.txt', true],                    # whole string match
 162       ['filex', false],                       # no such prefix
 163       ['subdir', true],                       # prefix matches
 164       ['subdir2', true],
 165       ['subdir2/', true],
 166       ['subdir2/subdir3', true],
 167       ['subdir2/subdir3/subdir4', true],
 168       ['subdir2 file4', true],                # look for both prefixes
 169       ['subdir4', false],                     # not a prefix match
 170     ].each do |search_filter, expect_results|
 171       search_filters = search_filter.split.each {|s| s.concat(':*')}.join('&')
 172       results = Collection.where("#{Collection.full_text_tsvector} @@ to_tsquery(?)",
 173                                  "#{search_filters}")
 174       if expect_results
 175         refute_empty results
 176       else
 177         assert_empty results
 178       end
 179     end
 180   end
 181
 182   test 'portable data hash with missing size hints' do
 183     [[". d41d8cd98f00b204e9800998ecf8427e+0+Bar 0:0:x\n",
 184       ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"],
 185      [". d41d8cd98f00b204e9800998ecf8427e+Foo 0:0:x\n",
 186       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 187      [". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n",
 188       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 189     ].each do |unportable, portable|
 190       c = Collection.new(manifest_text: unportable)
 191       assert c.valid?
 192       assert_equal(Digest::MD5.hexdigest(portable)+"+#{portable.length}",
 193                    c.portable_data_hash)
 194     end
 195   end
 196
 197   pdhmanifest = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"
 198   pdhmd5 = Digest::MD5.hexdigest pdhmanifest
 199   [[true, nil],
 200    [true, pdhmd5],
 201    [true, pdhmd5+'+12345'],
 202    [true, pdhmd5+'+'+pdhmanifest.length.to_s],
 203    [true, pdhmd5+'+12345+Foo'],
 204    [true, pdhmd5+'+Foo'],
 205    [false, Digest::MD5.hexdigest(pdhmanifest.strip)],
 206    [false, Digest::MD5.hexdigest(pdhmanifest.strip)+'+'+pdhmanifest.length.to_s],
 207    [false, pdhmd5[0..30]],
 208    [false, pdhmd5[0..30]+'z'],
 209    [false, pdhmd5[0..24]+'000000000'],
 210    [false, pdhmd5[0..24]+'000000000+0']].each do |isvalid, pdh|
 211     test "portable_data_hash #{pdh.inspect} valid? == #{isvalid}" do
 212       c = Collection.new manifest_text: pdhmanifest, portable_data_hash: pdh
 213       assert_equal isvalid, c.valid?, c.errors.full_messages.to_s
 214     end
 215   end
 216
 217   [0, 2, 4, nil].each do |ask|
 218     test "set replication_desired to #{ask.inspect}" do
 219       Rails.configuration.default_collection_replication = 2
 220       act_as_user users(:active) do
 221         c = collections(:replication_undesired_unconfirmed)
 222         c.update_attributes replication_desired: ask
 223         assert_equal ask, c.replication_desired
 224       end
 225     end
 226   end
 227
 228   test "replication_confirmed* can be set by admin user" do
 229     c = collections(:replication_desired_2_unconfirmed)
 230     act_as_user users(:admin) do
 231       assert c.update_attributes(replication_confirmed: 2,
 232                                  replication_confirmed_at: Time.now)
 233     end
 234   end
 235
 236   test "replication_confirmed* cannot be set by non-admin user" do
 237     act_as_user users(:active) do
 238       c = collections(:replication_desired_2_unconfirmed)
 239       # Cannot set just one at a time.
 240       assert_raise ArvadosModel::PermissionDeniedError do
 241         c.update_attributes replication_confirmed: 1
 242       end
 243       assert_raise ArvadosModel::PermissionDeniedError do
 244         c.update_attributes replication_confirmed_at: Time.now
 245       end
 246       # Cannot set both at once, either.
 247       assert_raise ArvadosModel::PermissionDeniedError do
 248         c.update_attributes(replication_confirmed: 1,
 249                             replication_confirmed_at: Time.now)
 250       end
 251     end
 252   end
 253
 254   test "replication_confirmed* can be cleared (but only together) by non-admin user" do
 255     act_as_user users(:active) do
 256       c = collections(:replication_desired_2_confirmed_2)
 257       # Cannot clear just one at a time.
 258       assert_raise ArvadosModel::PermissionDeniedError do
 259         c.update_attributes replication_confirmed: nil
 260       end
 261       c.reload
 262       assert_raise ArvadosModel::PermissionDeniedError do
 263         c.update_attributes replication_confirmed_at: nil
 264       end
 265       # Can clear both at once.
 266       c.reload
 267       assert c.update_attributes(replication_confirmed: nil,
 268                                  replication_confirmed_at: nil)
 269     end
 270   end
 271
 272   test "clear replication_confirmed* when introducing a new block in manifest" do
 273     c = collections(:replication_desired_2_confirmed_2)
 274     act_as_user users(:active) do
 275       assert c.update_attributes(manifest_text: collections(:user_agreement).signed_manifest_text)
 276       assert_nil c.replication_confirmed
 277       assert_nil c.replication_confirmed_at
 278     end
 279   end
 280
 281   test "don't clear replication_confirmed* when just renaming a file" do
 282     c = collections(:replication_desired_2_confirmed_2)
 283     act_as_user users(:active) do
 284       new_manifest = c.signed_manifest_text.sub(':bar', ':foo')
 285       assert c.update_attributes(manifest_text: new_manifest)
 286       assert_equal 2, c.replication_confirmed
 287       assert_not_nil c.replication_confirmed_at
 288     end
 289   end
 290
 291   test "don't clear replication_confirmed* when just deleting a data block" do
 292     c = collections(:replication_desired_2_confirmed_2)
 293     act_as_user users(:active) do
 294       new_manifest = c.signed_manifest_text
 295       new_manifest.sub!(/ \S+:bar/, '')
 296       new_manifest.sub!(/ acbd\S+/, '')
 297
 298       # Confirm that we did just remove a block from the manifest (if
 299       # not, this test would pass without testing the relevant case):
 300       assert_operator new_manifest.length+40, :<, c.signed_manifest_text.length
 301
 302       assert c.update_attributes(manifest_text: new_manifest)
 303       assert_equal 2, c.replication_confirmed
 304       assert_not_nil c.replication_confirmed_at
 305     end
 306   end
 307
 308   test "create collection with properties" do
 309     act_as_system_user do
 310       c = Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n",
 311                             properties: {'property_1' => 'value_1'})
 312       assert c.valid?
 313       assert_equal 'value_1', c.properties['property_1']
 314     end
 315   end
 316
 317   test 'create, delete, recreate collection with same name and owner' do
 318     act_as_user users(:active) do
 319       # create collection with name
 320       c = Collection.create(manifest_text: '',
 321                             name: "test collection name")
 322       assert c.valid?
 323       uuid = c.uuid
 324
 325       # mark collection as expired
 326       c.update_attribute 'expires_at', Time.new.strftime("%Y-%m-%d")
 327       c = Collection.where(uuid: uuid)
 328       assert_empty c, 'Should not be able to find expired collection'
 329
 330       # recreate collection with the same name
 331       c = Collection.create(manifest_text: '',
 332                             name: "test collection name")
 333       assert c.valid?
 334     end
 335   end
 336
 337   test "find_all_for_docker_image resolves names that look like hashes" do
 338     coll_list = Collection.
 339       find_all_for_docker_image('a' * 64, nil, [users(:active)])
 340     coll_uuids = coll_list.map(&:uuid)
 341     assert_includes(coll_uuids, collections(:docker_image).uuid)
 342   end
 343 end