X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/5ff990155f06109c55bbb832a8dd670265d3f3d1..43cfd4220fea2ba6632bbb82e4d1779c68d827fc:/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 af4698eb32..7689d5016f 100644
--- a/sdk/ruby/test/test_keep_manifest.rb
+++ b/sdk/ruby/test/test_keep_manifest.rb
@@ -1,30 +1,22 @@
-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)
+    assert_equal(exp_blocks, stream[1].map(&:to_s))
+    assert_equal(exp_files, stream.last)
+  end
 
   def test_simple_each_line_array
     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
     stream_name, block_s, file = SIMPLEST_MANIFEST.strip.split
     stream_a = manifest.each_line.to_a
     assert_equal(1, stream_a.size, "wrong number of streams")
-    assert_equal(stream_name, stream_a[0][0])
-    assert_equal([block_s], stream_a[0][1].map(&:to_s))
-    assert_equal([file], stream_a[0][2])
+    check_stream(stream_a.first, stream_name, [block_s], [file])
   end
 
   def test_simple_each_line_block
@@ -46,21 +38,53 @@ 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
     assert_empty(Keep::Manifest.new("").each_line.to_a)
   end
 
+  def test_empty_each_file_spec
+    assert_empty(Keep::Manifest.new("").each_file_spec.to_a)
+  end
+
+  def test_empty_files
+    assert_empty(Keep::Manifest.new("").files)
+  end
+
+  def test_empty_files_count
+    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
+
+  def test_empty_line_within_manifest
+    block_s = random_block
+    manifest = Keep::Manifest.
+      new([". #{block_s} 0:1:file1 1:2:file2\n",
+           "\n",
+           ". #{block_s} 3:3:file3 6:4:file4\n"].join(""))
+    streams = manifest.each_line.to_a
+    assert_equal(2, streams.size)
+    check_stream(streams[0], ".", [block_s], ["0:1:file1", "1:2:file2"])
+    check_stream(streams[1], ".", [block_s], ["3:3:file3", "6:4:file4"])
+  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
 
@@ -85,10 +109,15 @@ 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(ESCAPED_FILENAME_MANIFEST)
+    assert_equal([[".", "a a.txt", 9]], manifest.files)
+  end
+
   def test_files_spanning_multiple_blocks
     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
     assert_equal([[".", "repfile", 5],
@@ -126,6 +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")
@@ -137,4 +175,108 @@ class ManifestTest < Minitest::Test
     refute(manifest.has_file?("./s2/repfile"), "one-arg missing stream found")
     refute(manifest.has_file?("./s2", "repfile"), "two-arg missing stream found")
   end
+
+  def test_has_file_with_spaces
+    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
+
+  def test_collection_with_dirs_in_filenames
+    manifest = Keep::Manifest.new(MANIFEST_WITH_DIRS_IN_FILENAMES)
+
+    seen = Hash.new { |this, key| this[key] = [] }
+
+    manifest.files.each do |stream, basename, size|
+      refute(seen[stream].include?(basename), "each_file repeated #{stream}/#{basename}")
+      assert_equal(3, size, "wrong size for #{stream}/#{basename}")
+      seen[stream] << basename
+    end
+
+    assert_equal(%w(. ./dir1 ./dir1/dir2), seen.keys)
+
+    seen.each_pair do |stream, basenames|
+      assert_equal(%w(file1), basenames.sort, "wrong file list for #{stream}")
+    end
+  end
+
+  def test_multilevel_collection_with_dirs_in_filenames
+    manifest = Keep::Manifest.new(MULTILEVEL_MANIFEST_WITH_DIRS_IN_FILENAMES)
+
+    seen = Hash.new { |this, key| this[key] = [] }
+    expected_sizes = {'.' => 3, './dir1' => 6, './dir1/dir2' => 11}
+
+    manifest.files.each do |stream, basename, size|
+      refute(seen[stream].include?(basename), "each_file repeated #{stream}/#{basename}")
+      assert_equal(expected_sizes[stream], size, "wrong size for #{stream}/#{basename}")
+      seen[stream] << basename
+    end
+
+    assert_equal(%w(. ./dir1 ./dir1/dir2), seen.keys)
+
+    seen.each_pair do |stream, basenames|
+      assert_equal(%w(file1), basenames.sort, "wrong file list for #{stream}")
+    end
+  end
+
+  [[false, nil],
+   [false, '+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e0+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0 '],
+   [false, "d41d8cd98f00b204e9800998ecf8427e+0\n"],
+   [false, ' d41d8cd98f00b204e9800998ecf8427e+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+K+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0+0'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e++'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K+'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0++K'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K++'],
+   [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K++Z'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e', nil,nil,nil],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+0', '+0','0',nil],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Fizz+Buzz','+0','0','+Fizz+Buzz'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+Fizz+Buzz', nil,nil,'+Fizz+Buzz'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Z', '+0','0','+Z'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+Z', nil,nil,'+Z'],
+  ].each do |ok, locator, match2, match3, match4|
+    define_method "test_LOCATOR_REGEXP_on_#{locator.inspect}" do
+      match = Keep::Locator::LOCATOR_REGEXP.match locator
+      assert_equal ok, !!match
+      if ok
+        assert_equal match2, match[2]
+        assert_equal match3, match[3]
+        assert_equal match4, match[4]
+      end
+    end
+  end
 end