14482: Allow unescaped " " on stream and file token regexes (WIP)
[arvados.git] / sdk / ruby / test / test_keep_manifest.rb
1 # Copyright (C) The Arvados Authors. All rights reserved.
2 #
3 # SPDX-License-Identifier: Apache-2.0
4
5 require "arvados/keep"
6 require "minitest/autorun"
7 require "sdk_fixtures"
8
9 class ManifestTest < Minitest::Test
10   include SDKFixtures
11
12   def check_stream(stream, exp_name, exp_blocks, exp_files)
13     assert_equal(exp_name, stream.first)
14     assert_equal(exp_blocks, stream[1].map(&:to_s))
15     assert_equal(exp_files, stream.last)
16   end
17
18   def test_simple_each_line_array
19     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
20     stream_name, block_s, file = SIMPLEST_MANIFEST.strip.split
21     stream_a = manifest.each_line.to_a
22     assert_equal(1, stream_a.size, "wrong number of streams")
23     check_stream(stream_a.first, stream_name, [block_s], [file])
24   end
25
26   def test_simple_each_line_block
27     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
28     result = []
29     manifest.each_line do |stream, blocks, files|
30       result << files
31     end
32     assert_equal([[SIMPLEST_MANIFEST.split.last]], result,
33                  "wrong result from each_line block")
34   end
35
36   def test_multilevel_each_line
37     manifest = Keep::Manifest.new(MULTILEVEL_MANIFEST)
38     seen = []
39     manifest.each_line do |stream, blocks, files|
40       refute(seen.include?(stream),
41              "each_line already yielded stream #{stream}")
42       seen << stream
43       assert_equal(3, files.size, "wrong file count for stream #{stream}")
44     end
45     assert_equal(MULTILEVEL_MANIFEST.count("\n"), seen.size,
46                  "wrong number of streams")
47   end
48
49   def test_empty_each_line
50     assert_empty(Keep::Manifest.new("").each_line.to_a)
51   end
52
53   def test_empty_each_file_spec
54     assert_empty(Keep::Manifest.new("").each_file_spec.to_a)
55   end
56
57   def test_empty_files
58     assert_empty(Keep::Manifest.new("").files)
59   end
60
61   def test_empty_files_count
62     assert_equal(0, Keep::Manifest.new("").files_count)
63   end
64
65   def test_empty_dir_files_count
66     assert_equal(0,
67       Keep::Manifest.new("./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n").files_count)
68   end
69
70   def test_empty_files_size
71     assert_equal(0, Keep::Manifest.new("").files_size)
72   end
73
74   def test_empty_has_file?
75     refute(Keep::Manifest.new("").has_file?(""))
76   end
77
78   def test_empty_line_within_manifest
79     block_s = random_block
80     manifest = Keep::Manifest.
81       new([". #{block_s} 0:1:file1 1:2:file2\n",
82            "\n",
83            ". #{block_s} 3:3:file3 6:4:file4\n"].join(""))
84     streams = manifest.each_line.to_a
85     assert_equal(2, streams.size)
86     check_stream(streams[0], ".", [block_s], ["0:1:file1", "1:2:file2"])
87     check_stream(streams[1], ".", [block_s], ["3:3:file3", "6:4:file4"])
88   end
89
90   def test_backslash_escape_parsing
91     manifest = Keep::Manifest.new(MANY_ESCAPES_MANIFEST)
92     streams = manifest.each_line.to_a
93     assert_equal(1, streams.size, "wrong number of streams with whitespace")
94     assert_equal("./dir name", streams.first.first,
95                  "wrong stream name with whitespace")
96     assert_equal(["0:9:file\\name\t\\here.txt"], streams.first.last,
97                  "wrong filename(s) with whitespace")
98   end
99
100   def test_simple_files
101     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
102     assert_equal([[".", "simple.txt", 9]], manifest.files)
103   end
104
105   def test_multilevel_files
106     manifest = Keep::Manifest.new(MULTILEVEL_MANIFEST)
107     seen = Hash.new { |this, key| this[key] = [] }
108     manifest.files.each do |stream, basename, size|
109       refute(seen[stream].include?(basename),
110              "each_file repeated #{stream}/#{basename}")
111       seen[stream] << basename
112       assert_equal(3, size, "wrong size for #{stream}/#{basename}")
113     end
114     seen.each_pair do |stream, basenames|
115       assert_equal(%w(file1 file2 file3), basenames.sort,
116                    "wrong file list for #{stream}")
117     end
118   end
119
120   def test_files_with_colons_in_names
121     manifest = Keep::Manifest.new(COLON_FILENAME_MANIFEST)
122     assert_equal([[".", "file:test.txt", 9]], manifest.files)
123   end
124
125   def test_files_with_escape_sequence_in_filename
126     manifest = Keep::Manifest.new(ESCAPED_FILENAME_MANIFEST)
127     assert_equal([[".", "a a.txt", 9]], manifest.files)
128   end
129
130   def test_files_spanning_multiple_blocks
131     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
132     assert_equal([[".", "repfile", 5],
133                   [".", "uniqfile", 4],
134                   [".", "uniqfile2", 7],
135                   ["./s1", "repfile", 3],
136                   ["./s1", "uniqfile", 3]],
137                  manifest.files.sort)
138   end
139
140   def test_minimum_file_count_simple
141     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
142     assert(manifest.minimum_file_count?(1), "real minimum file count false")
143     refute(manifest.minimum_file_count?(2), "fake minimum file count true")
144   end
145
146   def test_minimum_file_count_multiblock
147     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
148     assert(manifest.minimum_file_count?(2), "low minimum file count false")
149     assert(manifest.minimum_file_count?(5), "real minimum file count false")
150     refute(manifest.minimum_file_count?(6), "fake minimum file count true")
151   end
152
153   def test_exact_file_count_simple
154     manifest = Keep::Manifest.new(SIMPLEST_MANIFEST)
155     assert(manifest.exact_file_count?(1), "exact file count false")
156     refute(manifest.exact_file_count?(0), "-1 file count true")
157     refute(manifest.exact_file_count?(2), "+1 file count true")
158   end
159
160   def test_exact_file_count_multiblock
161     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
162     assert(manifest.exact_file_count?(5), "exact file count false")
163     refute(manifest.exact_file_count?(4), "-1 file count true")
164     refute(manifest.exact_file_count?(6), "+1 file count true")
165   end
166
167   def test_files_size_multiblock
168     assert_equal(22, Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST).files_size)
169   end
170
171   def test_files_size_with_skipped_overlapping_data
172     manifest = Keep::Manifest.new(". #{random_block(9)} 3:3:f1 5:3:f2\n")
173     assert_equal(6, manifest.files_size)
174   end
175
176   def test_has_file
177     manifest = Keep::Manifest.new(MULTIBLOCK_FILE_MANIFEST)
178     assert(manifest.has_file?("./repfile"), "one-arg repfile not found")
179     assert(manifest.has_file?(".", "repfile"), "two-arg repfile not found")
180     assert(manifest.has_file?("./s1/repfile"), "one-arg s1/repfile not found")
181     assert(manifest.has_file?("./s1", "repfile"), "two-arg s1/repfile not found")
182     refute(manifest.has_file?("./s1/uniqfile2"), "one-arg missing file found")
183     refute(manifest.has_file?("./s1", "uniqfile2"), "two-arg missing file found")
184     refute(manifest.has_file?("./s2/repfile"), "one-arg missing stream found")
185     refute(manifest.has_file?("./s2", "repfile"), "two-arg missing stream found")
186   end
187
188   def test_has_file_with_spaces
189     manifest = Keep::Manifest.new(ESCAPED_FILENAME_MANIFEST)
190     assert(manifest.has_file?("./a a.txt"), "one-arg path not found")
191     assert(manifest.has_file?(".", "a a.txt"), "two-arg path not found")
192     refute(manifest.has_file?("a\\040\\141"), "one-arg unescaped found")
193     refute(manifest.has_file?(".", "a\\040\\141"), "two-arg unescaped found")
194   end
195
196   def test_parse_all_fixtures
197     fixtures('collections').each do |name, collection|
198       parse_collection_manifest name, collection
199     end
200   end
201
202   def test_raise_on_bogus_fixture
203     assert_raises ArgumentError do
204       parse_collection_manifest('bogus collection',
205                                 {'manifest_text' => ". zzz 0:\n"})
206     end
207   end
208
209   def parse_collection_manifest name, collection
210     manifest = Keep::Manifest.new(collection['manifest_text'])
211     manifest.each_file_spec do |stream_name, start_pos, file_size, file_name|
212       assert_kind_of String, stream_name
213       assert_kind_of Integer, start_pos
214       assert_kind_of Integer, file_size
215       assert_kind_of String, file_name
216       assert !stream_name.empty?, "empty stream_name in #{name} fixture"
217       assert !file_name.empty?, "empty file_name in #{name} fixture"
218     end
219   end
220
221   def test_collection_with_dirs_in_filenames
222     manifest = Keep::Manifest.new(MANIFEST_WITH_DIRS_IN_FILENAMES)
223
224     seen = Hash.new { |this, key| this[key] = [] }
225
226     manifest.files.each do |stream, basename, size|
227       refute(seen[stream].include?(basename), "each_file repeated #{stream}/#{basename}")
228       assert_equal(3, size, "wrong size for #{stream}/#{basename}")
229       seen[stream] << basename
230     end
231
232     assert_equal(%w(. ./dir1 ./dir1/dir2), seen.keys)
233
234     seen.each_pair do |stream, basenames|
235       assert_equal(%w(file1), basenames.sort, "wrong file list for #{stream}")
236     end
237   end
238
239   def test_multilevel_collection_with_dirs_in_filenames
240     manifest = Keep::Manifest.new(MULTILEVEL_MANIFEST_WITH_DIRS_IN_FILENAMES)
241
242     seen = Hash.new { |this, key| this[key] = [] }
243     expected_sizes = {'.' => 3, './dir1' => 6, './dir1/dir2' => 11}
244
245     manifest.files.each do |stream, basename, size|
246       refute(seen[stream].include?(basename), "each_file repeated #{stream}/#{basename}")
247       assert_equal(expected_sizes[stream], size, "wrong size for #{stream}/#{basename}")
248       seen[stream] << basename
249     end
250
251     assert_equal(%w(. ./dir1 ./dir1/dir2), seen.keys)
252
253     seen.each_pair do |stream, basenames|
254       assert_equal(%w(file1), basenames.sort, "wrong file list for #{stream}")
255     end
256   end
257
258   [[false, nil],
259    [false, '+0'],
260    [false, 'd41d8cd98f00b204e9800998ecf8427'],
261    [false, 'd41d8cd98f00b204e9800998ecf8427+0'],
262    [false, 'd41d8cd98f00b204e9800998ecf8427e0'],
263    [false, 'd41d8cd98f00b204e9800998ecf8427e0+0'],
264    [false, 'd41d8cd98f00b204e9800998ecf8427e+0 '],
265    [false, "d41d8cd98f00b204e9800998ecf8427e+0\n"],
266    [false, ' d41d8cd98f00b204e9800998ecf8427e+0'],
267    [false, 'd41d8cd98f00b204e9800998ecf8427e+K+0'],
268    [false, 'd41d8cd98f00b204e9800998ecf8427e+0+0'],
269    [false, 'd41d8cd98f00b204e9800998ecf8427e++'],
270    [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K+'],
271    [false, 'd41d8cd98f00b204e9800998ecf8427e+0++K'],
272    [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K++'],
273    [false, 'd41d8cd98f00b204e9800998ecf8427e+0+K++Z'],
274    [true, 'd41d8cd98f00b204e9800998ecf8427e', nil,nil,nil],
275    [true, 'd41d8cd98f00b204e9800998ecf8427e+0', '+0','0',nil],
276    [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Fizz+Buzz','+0','0','+Fizz+Buzz'],
277    [true, 'd41d8cd98f00b204e9800998ecf8427e+Fizz+Buzz', nil,nil,'+Fizz+Buzz'],
278    [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', '+0','0','+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'],
279    [true, 'd41d8cd98f00b204e9800998ecf8427e+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', nil,nil,'+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'],
280    [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Z', '+0','0','+Z'],
281    [true, 'd41d8cd98f00b204e9800998ecf8427e+Z', nil,nil,'+Z'],
282   ].each do |ok, locator, match2, match3, match4|
283     define_method "test_LOCATOR_REGEXP_on_#{locator.inspect}" do
284       match = Keep::Locator::LOCATOR_REGEXP.match locator
285       assert_equal ok, !!match
286       if ok
287         assert_equal match2, match[2]
288         assert_equal match3, match[3]
289         assert_equal match4, match[4]
290       end
291     end
292     define_method "test_parse_method_on_#{locator.inspect}" do
293       loc = Keep::Locator.parse locator
294       if !ok
295         assert_nil loc
296       else
297         refute_nil loc
298         assert loc.is_a?(Keep::Locator)
299         #assert loc.hash
300         #assert loc.size
301         #assert loc.hints.is_a?(Array)
302       end
303     end
304   end
305
306   [
307     [false, nil, "No manifest found"],
308     [true, ""],
309     [false, " ", "Invalid manifest: does not end with newline"],
310     [true, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n"],
311     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"],
312     [true, ". d41d8cd98f00b204e9800998ecf8427e a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"], # 2 locators
313     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar.txt\n"],
314     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo.txt\n"],
315     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo\n"],
316     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:...\n"],
317     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.../.foo./.../bar\n"],
318     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/...\n"],
319     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.../bar\n"],
320     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.bar/baz.txt\n"],
321     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar./baz.txt\n"],
322     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 000000000000000000000000000000:0777:foo.txt\n"],
323     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:0:0\n"],
324     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\040\n"],
325     [true, ". 00000000000000000000000000000000+0 0:0:0\n"],
326     [true, ". 00000000000000000000000000000000+0 0:0:d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff\n"],
327     [true, ". d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff 0:0:empty.txt\n"],
328     [true, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n"],
329     [false, '. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt',
330       "Invalid manifest: does not end with newline"],
331     [false, "abc d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
332       "invalid stream name \"abc\""],
333     [false, "abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
334       "invalid stream name \"abc/./foo\""],
335     [false, "./abc/../foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
336       "invalid stream name \"./abc/../foo\""],
337     [false, "./abc/. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
338       "invalid stream name \"./abc/.\""],
339     [false, "./abc/.. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
340       "invalid stream name \"./abc/..\""],
341     [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
342       "invalid stream name \"./abc/./foo\""],
343     # non-empty '.'-named file tokens aren't acceptable. Empty ones are used as empty dir placeholders.
344     [false, ". 8cf8463b34caa8ac871a52d5dd7ad1ef+1 0:1:.\n",
345       "invalid file token \"0:1:.\""],
346     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:..\n",
347       "invalid file token \"0:0:..\""],
348     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:./abc.txt\n",
349       "invalid file token \"0:0:./abc.txt\""],
350     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:../abc.txt\n",
351       "invalid file token \"0:0:../abc.txt\""],
352     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/.\n",
353       "invalid file token \"0:0:abc.txt/.\""],
354     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/..\n",
355       "invalid file token \"0:0:abc.txt/..\""],
356     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/./bc.txt\n",
357       "invalid file token \"0:0:a/./bc.txt\""],
358     [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/../bc.txt\n",
359       "invalid file token \"0:0:a/../bc.txt\""],
360     [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
361       "invalid stream name \"./abc/./foo\""],
362     [false, "d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n",
363       "invalid stream name \"d41d8cd98f00b204e9800998ecf8427e+0\""],
364     [false, ". d41d8cd98f00b204e9800998ecf8427 0:0:abc.txt\n",
365       "invalid locator \"d41d8cd98f00b204e9800998ecf8427\""],
366     [false, ". d41d8cd98f00b204e9800998ecf8427e\n",
367       "Manifest invalid for stream 1: no file tokens"],
368     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n/dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n",
369       "Manifest invalid for stream 2: missing or invalid stream name \"/dir1\""],
370     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n",
371       "Manifest invalid for stream 2: missing or invalid locator \"d41d8cd98f00b204e9800998ecf842\""],
372     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 abc.txt\n",
373       "Manifest invalid for stream 2: invalid file token \"abc.txt\""],
374     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:abc.txt\n",
375       "Manifest invalid for stream 2: invalid file token \"0:abc.txt\""],
376     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt xyz.txt\n",
377       "Manifest invalid for stream 2: invalid file token \"xyz.txt\""],
378     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n",
379       "Manifest invalid for stream 1: invalid file token \"d41d8cd98f00b204e9800998ecf8427e+0\""],
380     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\n",
381       "Manifest invalid for stream 1: invalid file token \"0:0:\""],
382     [false, ". d41d8cd98f00b204e9800998ecf8427e+0\n",
383       "Manifest invalid for stream 1: no file tokens"],
384     [false, ". 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n",
385       "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""],
386     [false, ". 0:0:foo.txt\n",
387       "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""],
388     [false, ".\n", "Manifest invalid for stream 1: missing or invalid locator"],
389     [false, ".", "Invalid manifest: does not end with newline"],
390     [false, ". \n", "Manifest invalid for stream 1: missing or invalid locator"],
391     [false, ".  \n", "Manifest invalid for stream 1: missing or invalid locator"],
392     [false, " . d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
393       "Manifest invalid for stream 1: missing or invalid stream name"],
394     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt \n",
395       "stream 1: trailing space"],
396    # TAB and other tricky whitespace characters:
397     [false, "\v. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
398       "Manifest invalid for stream 1: missing or invalid stream name \"\\v."],
399     [false, "./foo\vbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
400       "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\vbar"],
401     [false, "\t. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
402       "Manifest invalid for stream 1: missing or invalid stream name \"\\t"],
403     [false, ".\td41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
404       "Manifest invalid for stream 1: missing or invalid stream name \".\\t"],
405     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\t\n",
406       "stream 1: invalid file token \"0:0:foo.txt\\t\""],
407     [false, ". d41d8cd98f00b204e9800998ecf8427e+0\t 0:0:foo.txt\n",
408       "stream 1: missing or invalid locator \"d41d8cd98f00b204e9800998ecf8427e+0\\t\""],
409     [false, "./foo\tbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
410       "stream 1: missing or invalid stream name \"./foo\\tbar\""],
411     # other whitespace errors:
412     [false, ". d41d8cd98f00b204e9800998ecf8427e+0  0:0:foo.txt\n",
413       "Manifest invalid for stream 1: invalid file token \"\""],
414     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n \n",
415       "Manifest invalid for stream 2: missing stream name"],
416     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n\n",
417       "Manifest invalid for stream 2: missing stream name"],
418     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n ",
419       "Invalid manifest: does not end with newline"],
420     [false, "\n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
421       "Manifest invalid for stream 1: missing stream name"],
422     [false, " \n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
423       "Manifest invalid for stream 1: missing stream name"],
424     # empty file and stream name components:
425     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:/foo.txt\n",
426       "Manifest invalid for stream 1: invalid file token \"0:0:/foo.txt\""],
427     [false, "./ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
428       "Manifest invalid for stream 1: missing or invalid stream name \"./\""],
429     [false, ".//foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
430       "Manifest invalid for stream 1: missing or invalid stream name \".//foo\""],
431     [false, "./foo/ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
432       "Manifest invalid for stream 1: missing or invalid stream name \"./foo/\""],
433     [false, "./foo//bar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
434       "Manifest invalid for stream 1: missing or invalid stream name \"./foo//bar\""],
435     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo//bar.txt\n",
436       "Manifest invalid for stream 1: invalid file token \"0:0:foo//bar.txt\""],
437     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/\n",
438       "Manifest invalid for stream 1: invalid file token \"0:0:foo/\""],
439     # escaped chars
440     [true, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n"],
441     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\057bar\n"],
442     [true, ".\\057Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"],
443     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\057/bar\n",
444       "Manifest invalid for stream 1: invalid file token \"0:0:foo\\\\057/bar\""],
445     [false, ".\\057/Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n",
446       "Manifest invalid for stream 1: missing or invalid stream name \".\\\\057/Data\""],
447     [true, "./Data\\040Folder d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"],
448     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\057foo/bar\n",
449       "Manifest invalid for stream 1: invalid file token \"0:0:\\\\057foo/bar\""],
450     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134057foo/bar\n"],
451     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 \\040:\\040:foo.txt\n",
452       "Manifest invalid for stream 1: invalid file token \"\\\\040:\\\\040:foo.txt\""],
453   ].each do |ok, manifest, expected_error=nil|
454     define_method "test_validate manifest #{manifest.inspect}" do
455       assert_equal ok, Keep::Manifest.valid?(manifest)
456       if ok
457         assert Keep::Manifest.validate! manifest
458       else
459         begin
460           Keep::Manifest.validate! manifest
461         rescue ArgumentError => e
462           msg = e.message
463         end
464         refute_nil msg, "Expected ArgumentError"
465         assert msg.include?(expected_error), "Did not find expected error message. Expected: #{expected_error}; Actual: #{msg}"
466       end
467     end
468   end
469 end