services/api/test/unit/salvage_collection_test.rb

   1 require 'test_helper'
   2 require 'salvage_collection'
   3
   4 # Valid manifest_text
   5 TEST_MANIFEST = ". 341dabea2bd78ad0d6fc3f5b926b450e+85626+Ad391622a17f61e4a254eda85d1ca751c4f368da9@55e076ce 0:85626:brca2-hg19.fa\n. d7321a918923627c972d8f8080c07d29+82570+A22e0a1d9b9bc85c848379d98bedc64238b0b1532@55e076ce 0:82570:brca1-hg19.fa\n"
   6 TEST_MANIFEST_STRIPPED = ". 341dabea2bd78ad0d6fc3f5b926b450e+85626 0:85626:brca2-hg19.fa\n. d7321a918923627c972d8f8080c07d29+82570 0:82570:brca1-hg19.fa\n"
   7
   8 # This invalid manifest_text has the following flaws:
   9 #   Missing stream name with locator in it's place
  10 #   Invalid locators:
  11 #     foofaafaafaabd78ad0d6fc3f5b926b450e+foo
  12 #     bar-baabaabaabd78ad0d6fc3f5b926b450e
  13 #     bad12345dae58ad0d6fc3f5b926b450e+
  14 #     341dabea2bd78ad0d6fc3f5b926b450e+abc
  15 #     341dabea2bd78ad0d6fc3f5b926abcdf
  16 # Expectation: All these locators are preserved in salvaged_data
  17 BAD_MANIFEST = "faaaafaafaafaabd78ad0d6fc3f5b926b450e+foo bar-baabaabaabd78ad0d6fc3f5b926b450e bad12345dae58ad0d6fc3f5b926b450e+ 341dabea2bd78ad0d6fc3f5b926b450e+abc 341dabea2bd78ad0d6fc3f5b926abcdf 0:85626:brca2-hg19.fa\n. abcdabea2bd78ad0d6fc3f5b926b450e+1000 0:1000:brca-hg19.fa\n. d7321a918923627c972d8f8080c07d29+2000+A22e0a1d9b9bc85c848379d98bedc64238b0b1532@55e076ce 0:2000:brca1-hg19.fa\n"
  18
  19 # Mock arv_put
  20 module SalvageCollection
  21   def self.salvage_collection_arv_put(cmd)
  22     file_contents = File.open(cmd.split[-1], "r").read
  23     # simulate arv-put error when it is 'user_agreement'
  24     if file_contents.include? 'GNU_General_Public_License'
  25       raise("Error during arv-put")
  26     else
  27       ". " +
  28       Digest::MD5.hexdigest(file_contents) + "+" + file_contents.length.to_s +
  29       " 0:" + file_contents.length.to_s + ":invalid_manifest_text.txt\n"
  30     end
  31   end
  32 end
  33
  34 class SalvageCollectionTest < ActiveSupport::TestCase
  35   include SalvageCollection
  36
  37   setup do
  38     set_user_from_auth :admin
  39     # arv-put needs ARV env variables
  40     ENV['ARVADOS_API_HOST'] = 'unused_by_test'
  41     ENV['ARVADOS_API_TOKEN'] = 'unused_by_test'
  42   end
  43
  44   teardown do
  45     ENV['ARVADOS_API_HOST'] = ''
  46     ENV['ARVADOS_API_TOKEN'] = ''
  47   end
  48
  49   test "salvage test collection with valid manifest text" do
  50     # create a collection to test salvaging
  51     src_collection = Collection.new name: "test collection", manifest_text: TEST_MANIFEST
  52     src_collection.save!
  53
  54     # salvage this collection
  55     SalvageCollection.salvage_collection src_collection.uuid, 'test salvage collection - see #6277, #6859'
  56
  57     # verify the updated src_collection data
  58     updated_src_collection = Collection.find_by_uuid src_collection.uuid
  59     updated_name = updated_src_collection.name
  60     assert_equal true, updated_name.include?(src_collection.name)
  61
  62     match = updated_name.match /^test collection.*salvaged data at (.*)\)$/
  63     assert_not_nil match
  64     assert_not_nil match[1]
  65     assert_empty updated_src_collection.manifest_text
  66
  67     # match[1] is the uuid of the new collection created from src_collection's salvaged data
  68     # use this to get the new collection and verify
  69     new_collection = Collection.find_by_uuid match[1]
  70     match = new_collection.name.match /^salvaged from (.*),.*/
  71     assert_not_nil match
  72     assert_equal src_collection.uuid, match[1]
  73
  74     # verify the new collection's manifest format
  75     expected_manifest = ". " + Digest::MD5.hexdigest(TEST_MANIFEST_STRIPPED) + "+" +
  76       TEST_MANIFEST_STRIPPED.length.to_s + " 0:" + TEST_MANIFEST_STRIPPED.length.to_s +
  77       ":invalid_manifest_text.txt\n. 341dabea2bd78ad0d6fc3f5b926b450e+85626 d7321a918923627c972d8f8080c07d29+82570 0:168196:salvaged_data\n"
  78     assert_equal expected_manifest, new_collection.manifest_text
  79   end
  80
  81   test "salvage collection with no uuid required argument" do
  82     e = assert_raises RuntimeError do
  83       SalvageCollection.salvage_collection nil
  84     end
  85   end
  86
  87   test "salvage collection with bogus uuid" do
  88     e = assert_raises RuntimeError do
  89       SalvageCollection.salvage_collection 'bogus-uuid'
  90     end
  91     assert_equal "No collection found for bogus-uuid.", e.message
  92   end
  93
  94   test "salvage collection with no env ARVADOS_API_HOST" do
  95     e = assert_raises RuntimeError do
  96       ENV['ARVADOS_API_HOST'] = ''
  97       ENV['ARVADOS_API_TOKEN'] = ''
  98       SalvageCollection.salvage_collection collections('user_agreement').uuid
  99     end
 100     assert_equal "ARVADOS environment variables missing. Please set your admin user credentials as ARVADOS environment variables.", e.message
 101   end
 102
 103   test "salvage collection with error during arv-put" do
 104     # try to salvage collection while mimicking error during arv-put
 105     e = assert_raises RuntimeError do
 106       SalvageCollection.salvage_collection collections('user_agreement').uuid
 107     end
 108     assert_equal "Error during arv-put", e.message
 109   end
 110
 111   # This test uses BAD_MANIFEST, which has the following flaws:
 112   #   Missing stream name with locator in it's place
 113   #   Invalid locators:
 114   #     foo-faafaafaabd78ad0d6fc3f5b926b450e+foo
 115   #     bar-baabaabaabd78ad0d6fc3f5b926b450e
 116   #     bad12345dae58ad0d6fc3f5b926b450e+
 117   #     341dabea2bd78ad0d6fc3f5b926b450e+abc
 118   #     341dabea2bd78ad0d6fc3f5b926abcdf
 119   # Expectation: All these locators are preserved in salvaged_data
 120   test "invalid locators preserved during salvaging" do
 121     locator_data = SalvageCollection.salvage_collection_locator_data BAD_MANIFEST
 122     assert_equal 7, locator_data[0].size
 123     assert_equal false, locator_data[0].include?("foo-faafaafaabd78ad0d6fc3f5b926b450e+foo")
 124     assert_equal true,  locator_data[0].include?("faafaafaabd78ad0d6fc3f5b926b450e")
 125     assert_equal false, locator_data[0].include?("bar-baabaabaabd78ad0d6fc3f5b926b450e")
 126     assert_equal true,  locator_data[0].include?("baabaabaabd78ad0d6fc3f5b926b450e")
 127     assert_equal false, locator_data[0].include?("bad12345dae58ad0d6fc3f5b926b450e+")
 128     assert_equal true,  locator_data[0].include?("bad12345dae58ad0d6fc3f5b926b450e")
 129     assert_equal false, locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926b450e+abc")
 130     assert_equal true,  locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926b450e")
 131     assert_equal true,  locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926abcdf")
 132     assert_equal true,  locator_data[0].include?("abcdabea2bd78ad0d6fc3f5b926b450e+1000")
 133     assert_equal true,  locator_data[0].include?("d7321a918923627c972d8f8080c07d29+2000")
 134     assert_equal true,  locator_data[1].eql?(1000 + 2000)   # size
 135   end
 136
 137   test "salvage a collection with invalid manifest text" do
 138     # create a collection to test salvaging
 139     src_collection = Collection.new name: "test collection", manifest_text: BAD_MANIFEST, owner_uuid: 'zzzzz-tpzed-000000000000000'
 140     src_collection.save!(validate: false)
 141
 142     # salvage this collection
 143     SalvageCollection.salvage_collection src_collection.uuid, 'test salvage collection - see #6277, #6859'
 144
 145     # verify the updated src_collection data
 146     updated_src_collection = Collection.find_by_uuid src_collection.uuid
 147     updated_name = updated_src_collection.name
 148     assert_equal true, updated_name.include?(src_collection.name)
 149
 150     match = updated_name.match /^test collection.*salvaged data at (.*)\)$/
 151     assert_not_nil match
 152     assert_not_nil match[1]
 153     assert_empty updated_src_collection.manifest_text
 154
 155     # match[1] is the uuid of the new collection created from src_collection's salvaged data
 156     # use this to get the new collection and verify
 157     new_collection = Collection.find_by_uuid match[1]
 158     match = new_collection.name.match /^salvaged from (.*),.*/
 159     assert_not_nil match
 160     assert_equal src_collection.uuid, match[1]
 161     # verify the new collection's manifest includes the bad locators
 162     expected_manifest = ". " + Digest::MD5.hexdigest(BAD_MANIFEST) + "+" + BAD_MANIFEST.length.to_s +
 163       " 0:" + BAD_MANIFEST.length.to_s + ":invalid_manifest_text.txt\n. faafaafaabd78ad0d6fc3f5b926b450e baabaabaabd78ad0d6fc3f5b926b450e bad12345dae58ad0d6fc3f5b926b450e 341dabea2bd78ad0d6fc3f5b926b450e 341dabea2bd78ad0d6fc3f5b926abcdf abcdabea2bd78ad0d6fc3f5b926b450e+1000 d7321a918923627c972d8f8080c07d29+2000 0:3000:salvaged_data\n"
 164     assert_equal expected_manifest, new_collection.manifest_text
 165   end
 166 end