services/api/test/unit/collection_test.rb

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 require 'test_helper'
   6 require 'sweep_trashed_collections'
   7
   8 class CollectionTest < ActiveSupport::TestCase
   9   include DbCurrentTime
  10
  11   def create_collection name, enc=nil
  12     txt = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:#{name}.txt\n"
  13     txt.force_encoding(enc) if enc
  14     return Collection.create(manifest_text: txt)
  15   end
  16
  17   test 'accept ASCII manifest_text' do
  18     act_as_system_user do
  19       c = create_collection 'foo', Encoding::US_ASCII
  20       assert c.valid?
  21     end
  22   end
  23
  24   test 'accept UTF-8 manifest_text' do
  25     act_as_system_user do
  26       c = create_collection "f\xc3\x98\xc3\x98", Encoding::UTF_8
  27       assert c.valid?
  28     end
  29   end
  30
  31   test 'refuse manifest_text with invalid UTF-8 byte sequence' do
  32     act_as_system_user do
  33       c = create_collection "f\xc8o", Encoding::UTF_8
  34       assert !c.valid?
  35       assert_equal [:manifest_text], c.errors.messages.keys
  36       assert_match(/UTF-8/, c.errors.messages[:manifest_text].first)
  37     end
  38   end
  39
  40   test 'refuse manifest_text with non-UTF-8 encoding' do
  41     act_as_system_user do
  42       c = create_collection "f\xc8o", Encoding::ASCII_8BIT
  43       assert !c.valid?
  44       assert_equal [:manifest_text], c.errors.messages.keys
  45       assert_match(/UTF-8/, c.errors.messages[:manifest_text].first)
  46     end
  47   end
  48
  49   [
  50     ". 0:0:foo.txt",
  51     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  52     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  53     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  54   ].each do |manifest_text|
  55     test "create collection with invalid manifest text #{manifest_text} and expect error" do
  56       act_as_system_user do
  57         c = Collection.create(manifest_text: manifest_text)
  58         assert !c.valid?
  59       end
  60     end
  61   end
  62
  63   [
  64     nil,
  65     "",
  66     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  67   ].each do |manifest_text|
  68     test "create collection with valid manifest text #{manifest_text.inspect} and expect success" do
  69       act_as_system_user do
  70         c = Collection.create(manifest_text: manifest_text)
  71         assert c.valid?
  72       end
  73     end
  74   end
  75
  76   [
  77     ". 0:0:foo.txt",
  78     ". d41d8cd98f00b204e9800998ecf8427e foo.txt",
  79     "d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  80     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt",
  81   ].each do |manifest_text|
  82     test "update collection with invalid manifest text #{manifest_text} and expect error" do
  83       act_as_system_user do
  84         c = create_collection 'foo', Encoding::US_ASCII
  85         assert c.valid?
  86
  87         c.update_attribute 'manifest_text', manifest_text
  88         assert !c.valid?
  89       end
  90     end
  91   end
  92
  93   [
  94     nil,
  95     "",
  96     ". d41d8cd98f00b204e9800998ecf8427e 0:0:foo.txt\n",
  97   ].each do |manifest_text|
  98     test "update collection with valid manifest text #{manifest_text.inspect} and expect success" do
  99       act_as_system_user do
 100         c = create_collection 'foo', Encoding::US_ASCII
 101         assert c.valid?
 102
 103         c.update_attribute 'manifest_text', manifest_text
 104         assert c.valid?
 105       end
 106     end
 107   end
 108
 109   test 'create and update collection and verify file_names' do
 110     act_as_system_user do
 111       c = create_collection 'foo', Encoding::US_ASCII
 112       assert c.valid?
 113       created_file_names = c.file_names
 114       assert created_file_names
 115       assert_match(/foo.txt/, c.file_names)
 116
 117       c.update_attribute 'manifest_text', ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo2.txt\n"
 118       assert_not_equal created_file_names, c.file_names
 119       assert_match(/foo2.txt/, c.file_names)
 120     end
 121   end
 122
 123   [
 124     [2**8, false],
 125     [2**18, true],
 126   ].each do |manifest_size, allow_truncate|
 127     test "create collection with manifest size #{manifest_size} with allow_truncate=#{allow_truncate},
 128           and not expect exceptions even on very large manifest texts" do
 129       # file_names has a max size, hence there will be no errors even on large manifests
 130       act_as_system_user do
 131         manifest_text = ''
 132         index = 0
 133         while manifest_text.length < manifest_size
 134           manifest_text += "./blurfl d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylongfilename000000000000#{index}.txt\n"
 135           index += 1
 136         end
 137         manifest_text += "./laststreamname d41d8cd98f00b204e9800998ecf8427e+0 0:0:veryverylastfilename.txt\n"
 138         c = Collection.create(manifest_text: manifest_text)
 139
 140         assert c.valid?
 141         assert c.file_names
 142         assert_match(/veryverylongfilename0000000000001.txt/, c.file_names)
 143         assert_match(/veryverylongfilename0000000000002.txt/, c.file_names)
 144         if not allow_truncate
 145           assert_match(/veryverylastfilename/, c.file_names)
 146           assert_match(/laststreamname/, c.file_names)
 147         end
 148       end
 149     end
 150   end
 151
 152   test "full text search for collections" do
 153     # file_names column does not get populated when fixtures are loaded, hence setup test data
 154     act_as_system_user do
 155       Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n")
 156       Collection.create(manifest_text: ". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n")
 157       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")
 158     end
 159
 160     [
 161       ['foo', true],
 162       ['foo bar', false],                     # no collection matching both
 163       ['foo&bar', false],                     # no collection matching both
 164       ['foo|bar', true],                      # works only no spaces between the words
 165       ['Gnu public', true],                   # both prefixes found, though not consecutively
 166       ['Gnu&public', true],                   # both prefixes found, though not consecutively
 167       ['file4', true],                        # prefix match
 168       ['file4.txt', true],                    # whole string match
 169       ['filex', false],                       # no such prefix
 170       ['subdir', true],                       # prefix matches
 171       ['subdir2', true],
 172       ['subdir2/', true],
 173       ['subdir2/subdir3', true],
 174       ['subdir2/subdir3/subdir4', true],
 175       ['subdir2 file4', true],                # look for both prefixes
 176       ['subdir4', false],                     # not a prefix match
 177     ].each do |search_filter, expect_results|
 178       search_filters = search_filter.split.each {|s| s.concat(':*')}.join('&')
 179       results = Collection.where("#{Collection.full_text_tsvector} @@ to_tsquery(?)",
 180                                  "#{search_filters}")
 181       if expect_results
 182         refute_empty results
 183       else
 184         assert_empty results
 185       end
 186     end
 187   end
 188
 189   test 'portable data hash with missing size hints' do
 190     [[". d41d8cd98f00b204e9800998ecf8427e+0+Bar 0:0:x\n",
 191       ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"],
 192      [". d41d8cd98f00b204e9800998ecf8427e+Foo 0:0:x\n",
 193       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 194      [". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n",
 195       ". d41d8cd98f00b204e9800998ecf8427e 0:0:x\n"],
 196     ].each do |unportable, portable|
 197       c = Collection.new(manifest_text: unportable)
 198       assert c.valid?
 199       assert_equal(Digest::MD5.hexdigest(portable)+"+#{portable.length}",
 200                    c.portable_data_hash)
 201     end
 202   end
 203
 204   pdhmanifest = ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n"
 205   pdhmd5 = Digest::MD5.hexdigest pdhmanifest
 206   [[true, nil],
 207    [true, pdhmd5],
 208    [true, pdhmd5+'+12345'],
 209    [true, pdhmd5+'+'+pdhmanifest.length.to_s],
 210    [true, pdhmd5+'+12345+Foo'],
 211    [true, pdhmd5+'+Foo'],
 212    [false, Digest::MD5.hexdigest(pdhmanifest.strip)],
 213    [false, Digest::MD5.hexdigest(pdhmanifest.strip)+'+'+pdhmanifest.length.to_s],
 214    [false, pdhmd5[0..30]],
 215    [false, pdhmd5[0..30]+'z'],
 216    [false, pdhmd5[0..24]+'000000000'],
 217    [false, pdhmd5[0..24]+'000000000+0']].each do |isvalid, pdh|
 218     test "portable_data_hash #{pdh.inspect} valid? == #{isvalid}" do
 219       c = Collection.new manifest_text: pdhmanifest, portable_data_hash: pdh
 220       assert_equal isvalid, c.valid?, c.errors.full_messages.to_s
 221     end
 222   end
 223
 224   [0, 2, 4, nil].each do |ask|
 225     test "set replication_desired to #{ask.inspect}" do
 226       Rails.configuration.default_collection_replication = 2
 227       act_as_user users(:active) do
 228         c = collections(:replication_undesired_unconfirmed)
 229         c.update_attributes replication_desired: ask
 230         assert_equal ask, c.replication_desired
 231       end
 232     end
 233   end
 234
 235   test "replication_confirmed* can be set by admin user" do
 236     c = collections(:replication_desired_2_unconfirmed)
 237     act_as_user users(:admin) do
 238       assert c.update_attributes(replication_confirmed: 2,
 239                                  replication_confirmed_at: Time.now)
 240     end
 241   end
 242
 243   test "replication_confirmed* cannot be set by non-admin user" do
 244     act_as_user users(:active) do
 245       c = collections(:replication_desired_2_unconfirmed)
 246       # Cannot set just one at a time.
 247       assert_raise ArvadosModel::PermissionDeniedError do
 248         c.update_attributes replication_confirmed: 1
 249       end
 250       assert_raise ArvadosModel::PermissionDeniedError do
 251         c.update_attributes replication_confirmed_at: Time.now
 252       end
 253       # Cannot set both at once, either.
 254       assert_raise ArvadosModel::PermissionDeniedError do
 255         c.update_attributes(replication_confirmed: 1,
 256                             replication_confirmed_at: Time.now)
 257       end
 258     end
 259   end
 260
 261   test "replication_confirmed* can be cleared (but only together) by non-admin user" do
 262     act_as_user users(:active) do
 263       c = collections(:replication_desired_2_confirmed_2)
 264       # Cannot clear just one at a time.
 265       assert_raise ArvadosModel::PermissionDeniedError do
 266         c.update_attributes replication_confirmed: nil
 267       end
 268       c.reload
 269       assert_raise ArvadosModel::PermissionDeniedError do
 270         c.update_attributes replication_confirmed_at: nil
 271       end
 272       # Can clear both at once.
 273       c.reload
 274       assert c.update_attributes(replication_confirmed: nil,
 275                                  replication_confirmed_at: nil)
 276     end
 277   end
 278
 279   test "clear replication_confirmed* when introducing a new block in manifest" do
 280     c = collections(:replication_desired_2_confirmed_2)
 281     act_as_user users(:active) do
 282       assert c.update_attributes(manifest_text: collections(:user_agreement).signed_manifest_text)
 283       assert_nil c.replication_confirmed
 284       assert_nil c.replication_confirmed_at
 285     end
 286   end
 287
 288   test "don't clear replication_confirmed* when just renaming a file" do
 289     c = collections(:replication_desired_2_confirmed_2)
 290     act_as_user users(:active) do
 291       new_manifest = c.signed_manifest_text.sub(':bar', ':foo')
 292       assert c.update_attributes(manifest_text: new_manifest)
 293       assert_equal 2, c.replication_confirmed
 294       assert_not_nil c.replication_confirmed_at
 295     end
 296   end
 297
 298   test "don't clear replication_confirmed* when just deleting a data block" do
 299     c = collections(:replication_desired_2_confirmed_2)
 300     act_as_user users(:active) do
 301       new_manifest = c.signed_manifest_text
 302       new_manifest.sub!(/ \S+:bar/, '')
 303       new_manifest.sub!(/ acbd\S+/, '')
 304
 305       # Confirm that we did just remove a block from the manifest (if
 306       # not, this test would pass without testing the relevant case):
 307       assert_operator new_manifest.length+40, :<, c.signed_manifest_text.length
 308
 309       assert c.update_attributes(manifest_text: new_manifest)
 310       assert_equal 2, c.replication_confirmed
 311       assert_not_nil c.replication_confirmed_at
 312     end
 313   end
 314
 315   test 'signature expiry does not exceed trash_at' do
 316     act_as_user users(:active) do
 317       t0 = db_current_time
 318       c = Collection.create!(manifest_text: ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n", name: 'foo')
 319       c.update_attributes! trash_at: (t0 + 1.hours)
 320       c.reload
 321       sig_exp = /\+A[0-9a-f]{40}\@([0-9]+)/.match(c.signed_manifest_text)[1].to_i
 322       assert_operator sig_exp.to_i, :<=, (t0 + 1.hours).to_i
 323     end
 324   end
 325
 326   test 'far-future expiry date cannot be used to circumvent configured permission ttl' do
 327     act_as_user users(:active) do
 328       c = Collection.create!(manifest_text: ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:x\n",
 329                              name: 'foo',
 330                              trash_at: db_current_time + 1.years)
 331       sig_exp = /\+A[0-9a-f]{40}\@([0-9]+)/.match(c.signed_manifest_text)[1].to_i
 332       expect_max_sig_exp = db_current_time.to_i + Rails.configuration.blob_signature_ttl
 333       assert_operator c.trash_at.to_i, :>, expect_max_sig_exp
 334       assert_operator sig_exp.to_i, :<=, expect_max_sig_exp
 335     end
 336   end
 337
 338   test "create collection with properties" do
 339     act_as_system_user do
 340       c = Collection.create(manifest_text: ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo\n",
 341                             properties: {'property_1' => 'value_1'})
 342       assert c.valid?
 343       assert_equal 'value_1', c.properties['property_1']
 344     end
 345   end
 346
 347   test 'create, delete, recreate collection with same name and owner' do
 348     act_as_user users(:active) do
 349       # create collection with name
 350       c = Collection.create(manifest_text: '',
 351                             name: "test collection name")
 352       assert c.valid?
 353       uuid = c.uuid
 354
 355       c = Collection.readable_by(current_user).where(uuid: uuid)
 356       assert_not_empty c, 'Should be able to find live collection'
 357
 358       # mark collection as expired
 359       c.first.update_attributes!(trash_at: Time.new.strftime("%Y-%m-%d"))
 360       c = Collection.readable_by(current_user).where(uuid: uuid)
 361       assert_empty c, 'Should not be able to find expired collection'
 362
 363       # recreate collection with the same name
 364       c = Collection.create(manifest_text: '',
 365                             name: "test collection name")
 366       assert c.valid?
 367     end
 368   end
 369
 370   test 'trash_at cannot be set too far in the past' do
 371     act_as_user users(:active) do
 372       t0 = db_current_time
 373       c = Collection.create!(manifest_text: '', name: 'foo')
 374       c.update_attributes! trash_at: (t0 - 2.weeks)
 375       c.reload
 376       assert_operator c.trash_at, :>, t0
 377     end
 378   end
 379
 380   now = Time.now
 381   [['trash-to-delete interval negative',
 382     :collection_owned_by_active,
 383     {trash_at: now+2.weeks, delete_at: now},
 384     {state: :invalid}],
 385    ['now-to-delete interval short',
 386     :collection_owned_by_active,
 387     {trash_at: now+3.days, delete_at: now+7.days},
 388     {state: :trash_future}],
 389    ['now-to-delete interval short, trash=delete',
 390     :collection_owned_by_active,
 391     {trash_at: now+3.days, delete_at: now+3.days},
 392     {state: :trash_future}],
 393    ['trash-to-delete interval ok',
 394     :collection_owned_by_active,
 395     {trash_at: now, delete_at: now+15.days},
 396     {state: :trash_now}],
 397    ['trash-to-delete interval short, but far enough in future',
 398     :collection_owned_by_active,
 399     {trash_at: now+13.days, delete_at: now+15.days},
 400     {state: :trash_future}],
 401    ['trash by setting is_trashed bool',
 402     :collection_owned_by_active,
 403     {is_trashed: true},
 404     {state: :trash_now}],
 405    ['trash in future by setting just trash_at',
 406     :collection_owned_by_active,
 407     {trash_at: now+1.week},
 408     {state: :trash_future}],
 409    ['trash in future by setting trash_at and delete_at',
 410     :collection_owned_by_active,
 411     {trash_at: now+1.week, delete_at: now+4.weeks},
 412     {state: :trash_future}],
 413    ['untrash by clearing is_trashed bool',
 414     :expired_collection,
 415     {is_trashed: false},
 416     {state: :not_trash}],
 417   ].each do |test_name, fixture_name, updates, expect|
 418     test test_name do
 419       act_as_user users(:active) do
 420         min_exp = (db_current_time +
 421                    Rails.configuration.blob_signature_ttl.seconds)
 422         if fixture_name == :expired_collection
 423           # Fixture-finder shorthand doesn't find trashed collections
 424           # because they're not in the default scope.
 425           c = Collection.find_by_uuid('zzzzz-4zz18-mto52zx1s7sn3ih')
 426         else
 427           c = collections(fixture_name)
 428         end
 429         updates_ok = c.update_attributes(updates)
 430         expect_valid = expect[:state] != :invalid
 431         assert_equal expect_valid, updates_ok, c.errors.full_messages.to_s
 432         case expect[:state]
 433         when :invalid
 434           refute c.valid?
 435         when :trash_now
 436           assert c.is_trashed
 437           assert_not_nil c.trash_at
 438           assert_operator c.trash_at, :<=, db_current_time
 439           assert_not_nil c.delete_at
 440           assert_operator c.delete_at, :>=, min_exp
 441         when :trash_future
 442           refute c.is_trashed
 443           assert_not_nil c.trash_at
 444           assert_operator c.trash_at, :>, db_current_time
 445           assert_not_nil c.delete_at
 446           assert_operator c.delete_at, :>=, c.trash_at
 447           # Currently this minimum interval is needed to prevent early
 448           # garbage collection:
 449           assert_operator c.delete_at, :>=, min_exp
 450         when :not_trash
 451           refute c.is_trashed
 452           assert_nil c.trash_at
 453           assert_nil c.delete_at
 454         else
 455           raise "bad expect[:state]==#{expect[:state].inspect} in test case"
 456         end
 457       end
 458     end
 459   end
 460
 461   test 'default trash interval > blob signature ttl' do
 462     Rails.configuration.default_trash_lifetime = 86400 * 21 # 3 weeks
 463     start = db_current_time
 464     act_as_user users(:active) do
 465       c = Collection.create!(manifest_text: '', name: 'foo')
 466       c.update_attributes!(trash_at: start + 86400.seconds)
 467       assert_operator c.delete_at, :>=, start + (86400*22).seconds
 468       assert_operator c.delete_at, :<, start + (86400*22 + 30).seconds
 469       c.destroy
 470
 471       c = Collection.create!(manifest_text: '', name: 'foo')
 472       c.update_attributes!(is_trashed: true)
 473       assert_operator c.delete_at, :>=, start + (86400*21).seconds
 474     end
 475   end
 476
 477   test "find_all_for_docker_image resolves names that look like hashes" do
 478     coll_list = Collection.
 479       find_all_for_docker_image('a' * 64, nil, [users(:active)])
 480     coll_uuids = coll_list.map(&:uuid)
 481     assert_includes(coll_uuids, collections(:docker_image).uuid)
 482   end
 483
 484   test "move to trash in SweepTrashedCollections" do
 485     c = collections(:trashed_on_next_sweep)
 486     refute_empty Collection.where('uuid=? and is_trashed=false', c.uuid)
 487     assert_raises(ActiveRecord::RecordNotUnique) do
 488       act_as_user users(:active) do
 489         Collection.create!(owner_uuid: c.owner_uuid,
 490                            name: c.name)
 491       end
 492     end
 493     SweepTrashedCollections.sweep_now
 494     c = Collection.where('uuid=? and is_trashed=true', c.uuid).first
 495     assert c
 496     act_as_user users(:active) do
 497       assert Collection.create!(owner_uuid: c.owner_uuid,
 498                                 name: c.name)
 499     end
 500   end
 501
 502   test "delete in SweepTrashedCollections" do
 503     uuid = 'zzzzz-4zz18-3u1p5umicfpqszp' # deleted_on_next_sweep
 504     assert_not_empty Collection.where(uuid: uuid)
 505     SweepTrashedCollections.sweep_now
 506     assert_empty Collection.where(uuid: uuid)
 507   end
 508
 509   test "delete referring links in SweepTrashedCollections" do
 510     uuid = collections(:trashed_on_next_sweep).uuid
 511     act_as_system_user do
 512       Link.create!(head_uuid: uuid,
 513                    tail_uuid: system_user_uuid,
 514                    link_class: 'whatever',
 515                    name: 'something')
 516     end
 517     past = db_current_time
 518     Collection.where(uuid: uuid).
 519       update_all(is_trashed: true, trash_at: past, delete_at: past)
 520     assert_not_empty Collection.where(uuid: uuid)
 521     SweepTrashedCollections.sweep_now
 522     assert_empty Collection.where(uuid: uuid)
 523   end
 524 end