X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/e2da84c1fd9052791ed2b684741469570e09ea35..a239e2db534cc36aa8c3e08077383d84bf6ba8e8:/sdk/ruby/test/test_keep_manifest.rb

diff --git a/sdk/ruby/test/test_keep_manifest.rb b/sdk/ruby/test/test_keep_manifest.rb
index 50a3d36210..f1f1a530ce 100644
--- a/sdk/ruby/test/test_keep_manifest.rb
+++ b/sdk/ruby/test/test_keep_manifest.rb
@@ -1,21 +1,9 @@
-require "minitest/autorun"
 require "arvados/keep"
-
-def random_block(size=nil)
-  sprintf("%032x+%d", rand(16 ** 32), size || rand(64 * 1024 * 1024))
-end
+require "minitest/autorun"
+require "sdk_fixtures"
 
 class ManifestTest < Minitest::Test
-  SIMPLEST_MANIFEST = ". #{random_block(9)} 0:9:simple.txt\n"
-  MULTIBLOCK_FILE_MANIFEST =
-    [". #{random_block(8)} 0:4:repfile 4:4:uniqfile",
-     "./s1 #{random_block(6)} 0:3:repfile 3:3:uniqfile",
-     ". #{random_block(8)} 0:7:uniqfile2 7:1:repfile\n"].join("\n")
-  MULTILEVEL_MANIFEST =
-    [". #{random_block(9)} 0:3:file1 3:3:file2 6:3:file3\n",
-     "./dir1 #{random_block(9)} 0:3:file1 3:3:file2 6:3:file3\n",
-     "./dir1/subdir #{random_block(9)} 0:3:file1 3:3:file2 6:3:file3\n",
-     "./dir2 #{random_block(9)} 0:3:file1 3:3:file2 6:3:file3\n"].join("")
+  include SDKFixtures
 
   def check_stream(stream, exp_name, exp_blocks, exp_files)
     assert_equal(exp_name, stream.first)
@@ -50,7 +38,8 @@ class ManifestTest < Minitest::Test
       seen << stream
       assert_equal(3, files.size, "wrong file count for stream #{stream}")
     end
-    assert_equal(4, seen.size, "wrong number of streams")
+    assert_equal(MULTILEVEL_MANIFEST.count("\n"), seen.size,
+                 "wrong number of streams")
   end
 
   def test_empty_each_line
@@ -69,6 +58,10 @@ class ManifestTest < Minitest::Test
     assert_equal(0, Keep::Manifest.new("").files_count)
   end
 
+  def test_empty_files_size
+    assert_equal(0, Keep::Manifest.new("").files_size)
+  end
+
   def test_empty_has_file?
     refute(Keep::Manifest.new("").has_file?(""))
   end
@@ -86,13 +79,12 @@ class ManifestTest < Minitest::Test
   end
 
   def test_backslash_escape_parsing
-    m_text = "./dir\\040name #{random_block} 0:0:file\\\\name\\011\\here.txt\n"
-    manifest = Keep::Manifest.new(m_text)
+    manifest = Keep::Manifest.new(MANY_ESCAPES_MANIFEST)
     streams = manifest.each_line.to_a
     assert_equal(1, streams.size, "wrong number of streams with whitespace")
     assert_equal("./dir name", streams.first.first,
                  "wrong stream name with whitespace")
-    assert_equal(["0:0:file\\name\t\\here.txt"], streams.first.last,
+    assert_equal(["0:9:file\\name\t\\here.txt"], streams.first.last,
                  "wrong filename(s) with whitespace")
   end
 
@@ -117,12 +109,12 @@ class ManifestTest < Minitest::Test
   end
 
   def test_files_with_colons_in_names
-    manifest = Keep::Manifest.new(". #{random_block(9)} 0:9:file:test.txt\n")
+    manifest = Keep::Manifest.new(COLON_FILENAME_MANIFEST)
     assert_equal([[".", "file:test.txt", 9]], manifest.files)
   end
 
   def test_files_with_escape_sequence_in_filename
-    manifest = Keep::Manifest.new(". #{random_block(9)} 0:9:a\\040\\141.txt\n")
+    manifest = Keep::Manifest.new(ESCAPED_FILENAME_MANIFEST)
     assert_equal([[".", "a a.txt", 9]], manifest.files)
   end
 
@@ -163,6 +155,15 @@ class ManifestTest < Minitest::Test
     refute(manifest.exact_file_count?(6), "+1 file count true")
   end
 
+  def test_files_size_multiblock
+    assert_equal(22, Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST).files_size)
+  end
+
+  def test_files_size_with_skipped_overlapping_data
+    manifest = Keep::Manifest.new(". #{random_block(9)} 3:3:f1 5:3:f2\n")
+    assert_equal(6, manifest.files_size)
+  end
+
   def test_has_file
     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
     assert(manifest.has_file?("./repfile"), "one-arg repfile not found")
@@ -176,10 +177,35 @@ class ManifestTest < Minitest::Test
   end
 
   def test_has_file_with_spaces
-    manifest = Keep::Manifest.new(". #{random_block(3)} 0:3:a\\040b\\040c\n")
-    assert(manifest.has_file?("./a b c"), "one-arg 'a b c' not found")
-    assert(manifest.has_file?(".", "a b c"), "two-arg 'a b c' not found")
-    refute(manifest.has_file?("a\\040b\\040c"), "one-arg unescaped found")
-    refute(manifest.has_file?(".", "a\\040b\\040c"), "two-arg unescaped found")
+    manifest = Keep::Manifest.new(ESCAPED_FILENAME_MANIFEST)
+    assert(manifest.has_file?("./a a.txt"), "one-arg path not found")
+    assert(manifest.has_file?(".", "a a.txt"), "two-arg path not found")
+    refute(manifest.has_file?("a\\040\\141"), "one-arg unescaped found")
+    refute(manifest.has_file?(".", "a\\040\\141"), "two-arg unescaped found")
+  end
+
+  def test_parse_all_fixtures
+    fixtures('collections').each do |name, collection|
+      parse_collection_manifest name, collection
+    end
+  end
+
+  def test_raise_on_bogus_fixture
+    assert_raises ArgumentError do
+      parse_collection_manifest('bogus collection',
+                                {'manifest_text' => ". zzz 0:\n"})
+    end
+  end
+
+  def parse_collection_manifest name, collection
+    manifest = Keep::Manifest.new(collection['manifest_text'])
+    manifest.each_file_spec do |stream_name, start_pos, file_size, file_name|
+      assert_kind_of String, stream_name
+      assert_kind_of Integer, start_pos
+      assert_kind_of Integer, file_size
+      assert_kind_of String, file_name
+      assert !stream_name.empty?, "empty stream_name in #{name} fixture"
+      assert !file_name.empty?, "empty file_name in #{name} fixture"
+    end
   end
 end