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   test 'create and update collection and verify file_names' do
  43     act_as_system_user do
  44       c = create_collection 'foo', Encoding::US_ASCII
  45       assert c.valid?
  46       created_file_names = c.file_names
  47       assert created_file_names
  48       assert_match /foo.txt/, c.file_names
  49
  50       c.update_attribute 'manifest_text', ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo2.txt\n"
  51       assert_not_equal created_file_names, c.file_names
  52       assert_match /foo2.txt/, c.file_names
  53     end
  54   end
  55
  56   [
  57     [2**8, false],
  58     [2**18, true],
  59   ].each do |manifest_size, gets_truncated|
  60     test "create collection with manifest size #{manifest_size} which gets truncated #{gets_truncated},
  61           and not expect exceptions even on very large manifest texts" do
  62       # file_names has a max size, hence there will be no errors even on large manifests
  63       act_as_system_user do
  64         manifest_text = './blurfl d41d8cd98f00b204e9800998ecf8427e+0'
  65         index = 0
  66         while manifest_text.length < manifest_size
  67           manifest_text += ' ' + "0:0:veryverylongfilename000000000000#{index}.txt\n./subdir1"
  68           index += 1
  69         end
  70         manifest_text += "\n"
  71         c = Collection.create(manifest_text: manifest_text)
  72
  73         assert c.valid?
  74         assert c.file_names
  75         assert_match /veryverylongfilename0000000000001.txt/, c.file_names
  76         assert_match /veryverylongfilename0000000000002.txt/, c.file_names
  77         if !gets_truncated
  78           assert_match /blurfl/, c.file_names
  79           assert_match /subdir1/, c.file_names
  80         end
  81       end
  82     end
  83   end
  84
  85   [
  86     ['foo', true],
  87     ['foo bar', true],
  88     ['foox barx', false],                               # no match for both
  89     ['foox bar', true],                                 # bar matches
  90     ['foo barx', true],
  91     ['file2_in_subdir4', true],                         # whole string match
  92     ['filex_in_subdir4', false],                        # looks for the whole string and fails
  93     ['filex in subdir4', true],                         # matches subdir4
  94     ['6a4ff0499484c6c79c95cd8c566bd25f+249025', true],
  95     ['6a4ff0499484c6c79c95cd8c566bd25f+249024', false], # matches the whole string and fails
  96     ['6a4ff0499484c6c79c95cd8', true],                  # prefix matches
  97     ['499484c6c79c95cd8c566bd', false],                 # not a prefix match
  98     ['no-such-file', false],                            # looks for whole string and fails
  99     ['no such file', true],                             # matches "file"
 100   ].each do |search_filter, expect_results|
 101     test "full text search collection for #{search_filter} and expect results #{expect_results}" do
 102       search_filters = search_filter.split.each {|s| s.concat(':*')}
 103       results = Collection.where("to_tsvector('english', translate(manifest_text, '/.', '  ')) @@ to_tsquery(?)",
 104                                  "#{search_filters.join('|')}")
 105       if expect_results
 106         assert_equal true, results.length>0
 107       else
 108         assert_equal 0, results.length
 109       end
 110     end
 111   end
 112 end