services/api/test/unit/salvage_collection_test.rb

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