services/api/test/unit/collection_test.rb

   1 require 'test_helper'
   2 require 'sweep_trashed_collections'
   3
   4 class CollectionTest < ActiveSupport::TestCase
   5   include DbCurrentTime
   6
   7   def create_collection name, enc=nil
   8     txt = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:#{name}.txt\n"
   9     txt.force_encoding(enc) if enc
  10     return Collection.create(manifest_text: txt)
  11   end
  12
  13   test 'accept ASCII manifest_text' do
  14     act_as_system_user do
  15       c = create_collection 'foo', Encoding::US_ASCII
  16       assert c.valid?
  17     end
  18   end
  19
  20   test 'accept UTF-8 manifest_text' do
  21     act_as_system_user do
  22       c = create_collection "f\xc3\x98\xc3\x98", Encoding::UTF_8
  23       assert c.valid?
  24     end
  25   end
  26
  27   test 'refuse manifest_text with invalid UTF-8 byte sequence' do
  28     act_as_system_user do
  29       c = create_collection "f\xc8o", Encoding::UTF_8
  30       assert !c.valid?
  31       assert_equal [:manifest_text], c.errors.messages.keys
  32       assert_match(/UTF-8/, c.errors.messages[:manifest_text].first)
  33     end
  34   end
  35
  36   test 'refuse manifest_text with non-UTF-8 encoding' do
  37     act_as_system_user do
  38       c = create_collection "f\xc8o", Encoding::ASCII_8BIT
  39       assert !c.valid?
  40       assert_equal [:manifest_text], c.errors.messages.keys
  41       assert_match(/UTF-8/, c.errors.messages[:manifest_text].first)
  42     end
  43   end
  44
  45   [
  46     ". 0:0:foo.txt",
  47     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  48     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  49     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  50   ].each do |manifest_text|
  51     test "create collection with invalid manifest text #{manifest_text} and expect error" do
  52       act_as_system_user do
  53         c = Collection.create(manifest_text: manifest_text)
  54         assert !c.valid?
  55       end
  56     end
  57   end
  58
  59   [
  60     nil,
  61     "",
  62     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  63   ].each do |manifest_text|
  64     test "create collection with valid manifest text #{manifest_text.inspect} and expect success" do
  65       act_as_system_user do
  66         c = Collection.create(manifest_text: manifest_text)
  67         assert c.valid?
  68       end
  69     end
  70   end
  71
  72   [
  73     ". 0:0:foo.txt",
  74     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  75     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  76     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  77   ].each do |manifest_text|
  78     test "update collection with invalid manifest text #{manifest_text} and expect error" do
  79       act_as_system_user do
  80         c = create_collection 'foo', Encoding::US_ASCII
  81         assert c.valid?
  82
  83         c.update_attribute 'manifest_text', manifest_text
  84         assert !c.valid?
  85       end
  86     end
  87   end
  88
  89   [
  90     nil,
  91     "",
  92     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  93   ].each do |manifest_text|
  94     test "update collection with valid manifest text #{manifest_text.inspect} and expect success" do
  95       act_as_system_user do
  96         c = create_collection 'foo', Encoding::US_ASCII
  97         assert c.valid?
  98
  99         c.update_attribute 'manifest_text', manifest_text
 100         assert c.valid?
 101       end
 102     end
 103   end
 104
 105   test 'create and update collection and verify file_names' do
 106     act_as_system_user do
 107       c = create_collection 'foo', Encoding::US_ASCII
 108       assert c.valid?
 109       created_file_names = c.file_names
 110       assert created_file_names
 111       assert_match(/foo.txt/, c.file_names)
 112
 113       c.update_attribute 'manifest_text', ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo2.txt\n"
 114       assert_not_equal created_file_names, c.file_names
 115       assert_match(/foo2.txt/, c.file_names)
 116     end
 117   end
 118
 119   [
 120     [2**8, false],
 121     [2**18, true],
 122   ].each do |manifest_size, allow_truncate|
 123     test "create collection with manifest size #{manifest_size} with allow_truncate=#{allow_truncate},
 124           and not expect exceptions even on very large manifest texts" do
 125       # file_names has a max size, hence there will be no errors even on large manifests
 126       act_as_system_user do
 127         manifest_text = ''
 128         index = 0
 129         while manifest_text.length < manifest_size
 130           manifest_text += "./blurfl d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylongfilename000000000000#{index}.txt\n"
 131           index += 1
 132         end
 133         manifest_text += "./laststreamname d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylastfilename.txt\n"
 134         c = Collection.create(manifest_text: manifest_text)
 135
 136         assert c.valid?
 137         assert c.file_names
 138         assert_match(/veryverylongfilename0000000000001.txt/, c.file_names)
 139         assert_match(/veryverylongfilename0000000000002.txt/, c.file_names)
 140         if not allow_truncate
 141           assert_match(/veryverylastfilename/, c.file_names)
 142           assert_match(/laststreamname/, c.file_names)
 143         end
 144       end
 145     end
 146   end
 147
 148   test "full text search for collections" do
 149     # file_names column does not get populated when fixtures are loaded, hence setup test data
 150     act_as_system_user do
 151       Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n")
 152       Collection.create(manifest_text: ". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n")
 153       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")
 154     end
 155
 156     [
 157       ['foo', true],
 158       ['foo bar', false],                     # no collection matching both
 159       ['foo&bar', false],                     # no collection matching both
 160       ['foo|bar', true],                      # works only no spaces between the words
 161       ['Gnu public', true],                   # both prefixes found, though not consecutively
 162       ['Gnu&public', true],                   # both prefixes found, though not consecutively
 163       ['file4', true],                        # prefix match
 164       ['file4.txt', true],                    # whole string match
 165       ['filex', false],                       # no such prefix
 166       ['subdir', true],                       # prefix matches
 167       ['subdir2', true],
 168       ['subdir2/', true],
 169       ['subdir2/subdir3', true],
 170       ['subdir2/subdir3/subdir4', true],
 171       ['subdir2 file4', true],                # look for both prefixes
 172       ['subdir4', false],                     # not a prefix match
 173     ].each do |search_filter, expect_results|
 174       search_filters = search_filter.split.each {|s| s.concat(':*')}.join('&')
 175       results = Collection.where("#{Collection.full_text_tsvector} @@ to_tsquery(?)",
 176                                  "#{search_filters}")
 177       if expect_results
 178         refute_empty results
 179       else
 180         assert_empty results
 181       end
 182     end
 183   end
 184
 185   test 'portable data hash with missing size hints' do
 186     [[". d41d8cd98f00b204e9800998ecf8427e+0+Bar 0:0:x\n",
 187       ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"],
 188      [". d41d8cd98f00b204e9800998ecf8427e+Foo 0:0:x\n",
 189       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 190      [". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n",
 191       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 192     ].each do |unportable, portable|
 193       c = Collection.new(manifest_text: unportable)
 194       assert c.valid?
 195       assert_equal(Digest::MD5.hexdigest(portable)+"+#{portable.length}",
 196                    c.portable_data_hash)
 197     end
 198   end
 199
 200   pdhmanifest = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"
 201   pdhmd5 = Digest::MD5.hexdigest pdhmanifest
 202   [[true, nil],
 203    [true, pdhmd5],
 204    [true, pdhmd5+'+12345'],
 205    [true, pdhmd5+'+'+pdhmanifest.length.to_s],
 206    [true, pdhmd5+'+12345+Foo'],
 207    [true, pdhmd5+'+Foo'],
 208    [false, Digest::MD5.hexdigest(pdhmanifest.strip)],
 209    [false, Digest::MD5.hexdigest(pdhmanifest.strip)+'+'+pdhmanifest.length.to_s],
 210    [false, pdhmd5[0..30]],
 211    [false, pdhmd5[0..30]+'z'],
 212    [false, pdhmd5[0..24]+'000000000'],
 213    [false, pdhmd5[0..24]+'000000000+0']].each do |isvalid, pdh|
 214     test "portable_data_hash #{pdh.inspect} valid? == #{isvalid}" do
 215       c = Collection.new manifest_text: pdhmanifest, portable_data_hash: pdh
 216       assert_equal isvalid, c.valid?, c.errors.full_messages.to_s
 217     end
 218   end
 219
 220   [0, 2, 4, nil].each do |ask|
 221     test "set replication_desired to #{ask.inspect}" do
 222       Rails.configuration.default_collection_replication = 2
 223       act_as_user users(:active) do
 224         c = collections(:replication_undesired_unconfirmed)
 225         c.update_attributes replication_desired: ask
 226         assert_equal ask, c.replication_desired
 227       end
 228     end
 229   end
 230
 231   test "replication_confirmed* can be set by admin user" do
 232     c = collections(:replication_desired_2_unconfirmed)
 233     act_as_user users(:admin) do
 234       assert c.update_attributes(replication_confirmed: 2,
 235                                  replication_confirmed_at: Time.now)
 236     end
 237   end
 238
 239   test "replication_confirmed* cannot be set by non-admin user" do
 240     act_as_user users(:active) do
 241       c = collections(:replication_desired_2_unconfirmed)
 242       # Cannot set just one at a time.
 243       assert_raise ArvadosModel::PermissionDeniedError do
 244         c.update_attributes replication_confirmed: 1
 245       end
 246       assert_raise ArvadosModel::PermissionDeniedError do
 247         c.update_attributes replication_confirmed_at: Time.now
 248       end
 249       # Cannot set both at once, either.
 250       assert_raise ArvadosModel::PermissionDeniedError do
 251         c.update_attributes(replication_confirmed: 1,
 252                             replication_confirmed_at: Time.now)
 253       end
 254     end
 255   end
 256
 257   test "replication_confirmed* can be cleared (but only together) by non-admin user" do
 258     act_as_user users(:active) do
 259       c = collections(:replication_desired_2_confirmed_2)
 260       # Cannot clear just one at a time.
 261       assert_raise ArvadosModel::PermissionDeniedError do
 262         c.update_attributes replication_confirmed: nil
 263       end
 264       c.reload
 265       assert_raise ArvadosModel::PermissionDeniedError do
 266         c.update_attributes replication_confirmed_at: nil
 267       end
 268       # Can clear both at once.
 269       c.reload
 270       assert c.update_attributes(replication_confirmed: nil,
 271                                  replication_confirmed_at: nil)
 272     end
 273   end
 274
 275   test "clear replication_confirmed* when introducing a new block in manifest" do
 276     c = collections(:replication_desired_2_confirmed_2)
 277     act_as_user users(:active) do
 278       assert c.update_attributes(manifest_text: collections(:user_agreement).signed_manifest_text)
 279       assert_nil c.replication_confirmed
 280       assert_nil c.replication_confirmed_at
 281     end
 282   end
 283
 284   test "don't clear replication_confirmed* when just renaming a file" do
 285     c = collections(:replication_desired_2_confirmed_2)
 286     act_as_user users(:active) do
 287       new_manifest = c.signed_manifest_text.sub(':bar', ':foo')
 288       assert c.update_attributes(manifest_text: new_manifest)
 289       assert_equal 2, c.replication_confirmed
 290       assert_not_nil c.replication_confirmed_at
 291     end
 292   end
 293
 294   test "don't clear replication_confirmed* when just deleting a data block" do
 295     c = collections(:replication_desired_2_confirmed_2)
 296     act_as_user users(:active) do
 297       new_manifest = c.signed_manifest_text
 298       new_manifest.sub!(/ \S+:bar/, '')
 299       new_manifest.sub!(/ acbd\S+/, '')
 300
 301       # Confirm that we did just remove a block from the manifest (if
 302       # not, this test would pass without testing the relevant case):
 303       assert_operator new_manifest.length+40, :<, c.signed_manifest_text.length
 304
 305       assert c.update_attributes(manifest_text: new_manifest)
 306       assert_equal 2, c.replication_confirmed
 307       assert_not_nil c.replication_confirmed_at
 308     end
 309   end
 310
 311   test 'signature expiry does not exceed trash_at' do
 312     act_as_user users(:active) do
 313       t0 = db_current_time
 314       c = Collection.create!(manifest_text: ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n", name: 'foo')
 315       c.update_attributes! trash_at: (t0 + 1.hours)
 316       c.reload
 317       sig_exp = /\+A[0-9a-f]{40}\@([0-9]+)/.match(c.signed_manifest_text)[1].to_i
 318       assert_operator sig_exp.to_i, :<=, (t0 + 1.hours).to_i
 319     end
 320   end
 321
 322   test 'far-future expiry date cannot be used to circumvent configured permission ttl' do
 323     act_as_user users(:active) do
 324       c = Collection.create!(manifest_text: ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n",
 325                              name: 'foo',
 326                              trash_at: db_current_time + 1.years)
 327       sig_exp = /\+A[0-9a-f]{40}\@([0-9]+)/.match(c.signed_manifest_text)[1].to_i
 328       expect_max_sig_exp = db_current_time.to_i + Rails.configuration.blob_signature_ttl
 329       assert_operator c.trash_at.to_i, :>, expect_max_sig_exp
 330       assert_operator sig_exp.to_i, :<=, expect_max_sig_exp
 331     end
 332   end
 333
 334   test "create collection with properties" do
 335     act_as_system_user do
 336       c = Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n",
 337                             properties: {'property_1' => 'value_1'})
 338       assert c.valid?
 339       assert_equal 'value_1', c.properties['property_1']
 340     end
 341   end
 342
 343   test 'create, delete, recreate collection with same name and owner' do
 344     act_as_user users(:active) do
 345       # create collection with name
 346       c = Collection.create(manifest_text: '',
 347                             name: "test collection name")
 348       assert c.valid?
 349       uuid = c.uuid
 350
 351       # mark collection as expired
 352       c.update_attributes!(trash_at: Time.new.strftime("%Y-%m-%d"))
 353       c = Collection.where(uuid: uuid)
 354       assert_empty c, 'Should not be able to find expired collection'
 355
 356       # recreate collection with the same name
 357       c = Collection.create(manifest_text: '',
 358                             name: "test collection name")
 359       assert c.valid?
 360     end
 361   end
 362
 363   test 'trash_at cannot be set too far in the past' do
 364     act_as_user users(:active) do
 365       t0 = db_current_time
 366       c = Collection.create!(manifest_text: '', name: 'foo')
 367       c.update_attributes! trash_at: (t0 - 2.weeks)
 368       c.reload
 369       assert_operator c.trash_at, :>, t0
 370     end
 371   end
 372
 373   now = Time.now
 374   [['trash-to-delete interval negative',
 375     :collection_owned_by_active,
 376     {trash_at: now+2.weeks, delete_at: now},
 377     {state: :invalid}],
 378    ['now-to-delete interval short',
 379     :collection_owned_by_active,
 380     {trash_at: now+3.days, delete_at: now+7.days},
 381     {state: :trash_future}],
 382    ['now-to-delete interval short, trash=delete',
 383     :collection_owned_by_active,
 384     {trash_at: now+3.days, delete_at: now+3.days},
 385     {state: :trash_future}],
 386    ['trash-to-delete interval ok',
 387     :collection_owned_by_active,
 388     {trash_at: now, delete_at: now+15.days},
 389     {state: :trash_now}],
 390    ['trash-to-delete interval short, but far enough in future',
 391     :collection_owned_by_active,
 392     {trash_at: now+13.days, delete_at: now+15.days},
 393     {state: :trash_future}],
 394    ['trash by setting is_trashed bool',
 395     :collection_owned_by_active,
 396     {is_trashed: true},
 397     {state: :trash_now}],
 398    ['trash in future by setting just trash_at',
 399     :collection_owned_by_active,
 400     {trash_at: now+1.week},
 401     {state: :trash_future}],
 402    ['trash in future by setting trash_at and delete_at',
 403     :collection_owned_by_active,
 404     {trash_at: now+1.week, delete_at: now+4.weeks},
 405     {state: :trash_future}],
 406    ['untrash by clearing is_trashed bool',
 407     :expired_collection,
 408     {is_trashed: false},
 409     {state: :not_trash}],
 410   ].each do |test_name, fixture_name, updates, expect|
 411     test test_name do
 412       act_as_user users(:active) do
 413         min_exp = (db_current_time +
 414                    Rails.configuration.blob_signature_ttl.seconds)
 415         if fixture_name == :expired_collection
 416           # Fixture-finder shorthand doesn't find trashed collections
 417           # because they're not in the default scope.
 418           c = Collection.unscoped.find_by_uuid('zzzzz-4zz18-mto52zx1s7sn3ih')
 419         else
 420           c = collections(fixture_name)
 421         end
 422         updates_ok = c.update_attributes(updates)
 423         expect_valid = expect[:state] != :invalid
 424         assert_equal expect_valid, updates_ok, c.errors.full_messages.to_s
 425         case expect[:state]
 426         when :invalid
 427           refute c.valid?
 428         when :trash_now
 429           assert c.is_trashed
 430           assert_not_nil c.trash_at
 431           assert_operator c.trash_at, :<=, db_current_time
 432           assert_not_nil c.delete_at
 433           assert_operator c.delete_at, :>=, min_exp
 434         when :trash_future
 435           refute c.is_trashed
 436           assert_not_nil c.trash_at
 437           assert_operator c.trash_at, :>, db_current_time
 438           assert_not_nil c.delete_at
 439           assert_operator c.delete_at, :>=, c.trash_at
 440           # Currently this minimum interval is needed to prevent early
 441           # garbage collection:
 442           assert_operator c.delete_at, :>=, min_exp
 443         when :not_trash
 444           refute c.is_trashed
 445           assert_nil c.trash_at
 446           assert_nil c.delete_at
 447         else
 448           raise "bad expect[:state]==#{expect[:state].inspect} in test case"
 449         end
 450       end
 451     end
 452   end
 453
 454   test 'default trash interval > blob signature ttl' do
 455     Rails.configuration.default_trash_lifetime = 86400 * 21 # 3 weeks
 456     start = db_current_time
 457     act_as_user users(:active) do
 458       c = Collection.create!(manifest_text: '', name: 'foo')
 459       c.update_attributes!(trash_at: start + 86400.seconds)
 460       assert_operator c.delete_at, :>=, start + (86400*22).seconds
 461       assert_operator c.delete_at, :<, start + (86400*22 + 30).seconds
 462       c.destroy
 463
 464       c = Collection.create!(manifest_text: '', name: 'foo')
 465       c.update_attributes!(is_trashed: true)
 466       assert_operator c.delete_at, :>=, start + (86400*21).seconds
 467     end
 468   end
 469
 470   test "find_all_for_docker_image resolves names that look like hashes" do
 471     coll_list = Collection.
 472       find_all_for_docker_image('a' * 64, nil, [users(:active)])
 473     coll_uuids = coll_list.map(&:uuid)
 474     assert_includes(coll_uuids, collections(:docker_image).uuid)
 475   end
 476
 477   test "move to trash in SweepTrashedCollections" do
 478     c = collections(:trashed_on_next_sweep)
 479     refute_empty Collection.where('uuid=? and is_trashed=false', c.uuid)
 480     assert_raises(ActiveRecord::RecordNotUnique) do
 481       act_as_user users(:active) do
 482         Collection.create!(owner_uuid: c.owner_uuid,
 483                            name: c.name)
 484       end
 485     end
 486     SweepTrashedCollections.sweep_now
 487     c = Collection.unscoped.where('uuid=? and is_trashed=true', c.uuid).first
 488     assert c
 489     act_as_user users(:active) do
 490       assert Collection.create!(owner_uuid: c.owner_uuid,
 491                                 name: c.name)
 492     end
 493   end
 494
 495   test "delete in SweepTrashedCollections" do
 496     uuid = 'zzzzz-4zz18-3u1p5umicfpqszp' # deleted_on_next_sweep
 497     assert_not_empty Collection.unscoped.where(uuid: uuid)
 498     SweepTrashedCollections.sweep_now
 499     assert_empty Collection.unscoped.where(uuid: uuid)
 500   end
 501
 502   test "delete referring links in SweepTrashedCollections" do
 503     uuid = collections(:trashed_on_next_sweep).uuid
 504     act_as_system_user do
 505       Link.create!(head_uuid: uuid,
 506                    tail_uuid: system_user_uuid,
 507                    link_class: 'whatever',
 508                    name: 'something')
 509     end
 510     past = db_current_time
 511     Collection.unscoped.where(uuid: uuid).
 512       update_all(is_trashed: true, trash_at: past, delete_at: past)
 513     assert_not_empty Collection.unscoped.where(uuid: uuid)
 514     SweepTrashedCollections.sweep_now
 515     assert_empty Collection.unscoped.where(uuid: uuid)
 516   end
 517 end