X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/0c082b748011ee94679204056cb52c145af35b08..c6c2f3518bc745eed95b5f5b81db5d17db4366ff:/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 c49bd941d5..eee8b39699 100644 --- a/sdk/ruby/test/test_keep_manifest.rb +++ b/sdk/ruby/test/test_keep_manifest.rb @@ -1,3 +1,7 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + require "arvados/keep" require "minitest/autorun" require "sdk_fixtures" @@ -58,6 +62,11 @@ class ManifestTest < Minitest::Test assert_equal(0, Keep::Manifest.new("").files_count) end + def test_empty_dir_files_count + assert_equal(0, + Keep::Manifest.new("./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n").files_count) + end + def test_empty_files_size assert_equal(0, Keep::Manifest.new("").files_size) end @@ -266,6 +275,8 @@ class ManifestTest < Minitest::Test [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+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', '+0','0','+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'], + [true, 'd41d8cd98f00b204e9800998ecf8427e+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', nil,nil,'+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'], [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Z', '+0','0','+Z'], [true, 'd41d8cd98f00b204e9800998ecf8427e+Z', nil,nil,'+Z'], ].each do |ok, locator, match2, match3, match4| @@ -278,37 +289,222 @@ class ManifestTest < Minitest::Test assert_equal match4, match[4] end end + define_method "test_parse_method_on_#{locator.inspect}" do + loc = Keep::Locator.parse locator + if !ok + assert_nil loc + else + refute_nil loc + assert loc.is_a?(Keep::Locator) + #assert loc.hash + #assert loc.size + #assert loc.hints.is_a?(Array) + end + end end [ - [false, 'abc d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt', 'invalid stream name abc'], - [false, 'd41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt', 'invalid stream name d41d8cd98f00b204e9800998ecf8427e'], - [false, '. d41d8cd98f00b204e9800998ecf8427 0:0:abc.txt', 'invalid locator d41d8cd98f00b204e9800998ecf8427'], - [true, '. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt'], - [true, '. d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt'], - [true, '. d41d8cd98f00b204e9800998ecf8427e a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt'], # 2 locators - [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n/dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt", - 'Manifest invalid for stream 2. Missing or invalid stream name /dir1'], - [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt", - 'Manifest invalid for stream 2. Missing or invalid locator d41d8cd98f00b204e9800998ecf842'], - [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 abc.txt", - 'Manifest invalid for stream 2. Missing or invalid file name abc.txt'], - [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:abc.txt", - 'Manifest invalid for stream 2. Missing or invalid file name 0:abc.txt'], - [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt xyz.txt", - 'Manifest invalid for stream 2. Missing or invalid file name xyz.txt'], + [false, nil, "No manifest found"], + [true, ""], + [false, " ", "Invalid manifest: does not end with newline"], + [true, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"], # 2 locators + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:...\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.../.foo./.../bar\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/...\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.../bar\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.bar/baz.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar./baz.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 000000000000000000000000000000:0777:foo.txt\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:0:0\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\040\n"], + [true, ". 00000000000000000000000000000000+0 0:0:0\n"], + [true, ". 00000000000000000000000000000000+0 0:0:d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff 0:0:empty.txt\n"], + [true, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:.\n"], + [false, '. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt', + "Invalid manifest: does not end with newline"], + [false, "abc d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"abc\""], + [false, "abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"abc/./foo\""], + [false, "./abc/../foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"./abc/../foo\""], + [false, "./abc/. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"./abc/.\""], + [false, "./abc/.. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"./abc/..\""], + [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"./abc/./foo\""], + # non-empty '.'-named file tokens aren't acceptable. Empty ones are used as empty dir placeholders. + [false, ". 8cf8463b34caa8ac871a52d5dd7ad1ef+1 0:1:.\n", + "invalid file token \"0:1:.\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:..\n", + "invalid file token \"0:0:..\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:./abc.txt\n", + "invalid file token \"0:0:./abc.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:../abc.txt\n", + "invalid file token \"0:0:../abc.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/.\n", + "invalid file token \"0:0:abc.txt/.\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/..\n", + "invalid file token \"0:0:abc.txt/..\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/./bc.txt\n", + "invalid file token \"0:0:a/./bc.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/../bc.txt\n", + "invalid file token \"0:0:a/../bc.txt\""], + [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n", + "invalid stream name \"./abc/./foo\""], + [false, "d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n", + "invalid stream name \"d41d8cd98f00b204e9800998ecf8427e+0\""], + [false, ". d41d8cd98f00b204e9800998ecf8427 0:0:abc.txt\n", + "invalid locator \"d41d8cd98f00b204e9800998ecf8427\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e\n", + "Manifest invalid for stream 1: no file tokens"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n/dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n", + "Manifest invalid for stream 2: missing or invalid stream name \"/dir1\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n", + "Manifest invalid for stream 2: missing or invalid locator \"d41d8cd98f00b204e9800998ecf842\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 abc.txt\n", + "Manifest invalid for stream 2: invalid file token \"abc.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:abc.txt\n", + "Manifest invalid for stream 2: invalid file token \"0:abc.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt xyz.txt\n", + "Manifest invalid for stream 2: invalid file token \"xyz.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n", + "Manifest invalid for stream 1: invalid file token \"d41d8cd98f00b204e9800998ecf8427e+0\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\n", + "Manifest invalid for stream 1: invalid file token \"0:0:\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0\n", + "Manifest invalid for stream 1: no file tokens"], + [false, ". 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n", + "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""], + [false, ". 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""], + [false, ".\n", "Manifest invalid for stream 1: missing or invalid locator"], + [false, ".", "Invalid manifest: does not end with newline"], + [false, ". \n", "Manifest invalid for stream 1: missing or invalid locator"], + [false, ". \n", "Manifest invalid for stream 1: missing or invalid locator"], + [false, " . d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt \n", + "stream 1: trailing space"], + # TAB and other tricky whitespace characters: + [false, "\v. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"\\v."], + [false, "./foo\vbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\vbar"], + [false, "\t. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"\\t"], + [false, ".\td41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \".\\t"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\t\n", + "stream 1: invalid file token \"0:0:foo.txt\\t\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0\t 0:0:foo.txt\n", + "stream 1: missing or invalid locator \"d41d8cd98f00b204e9800998ecf8427e+0\\t\""], + [false, "./foo\tbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "stream 1: missing or invalid stream name \"./foo\\tbar\""], + # other whitespace errors: + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: invalid file token \"\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n \n", + "Manifest invalid for stream 2: missing stream name"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n\n", + "Manifest invalid for stream 2: missing stream name"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n ", + "Invalid manifest: does not end with newline"], + [false, "\n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing stream name"], + [false, " \n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing stream name"], + # empty file and stream name components: + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:/foo.txt\n", + "Manifest invalid for stream 1: invalid file token \"0:0:/foo.txt\""], + [false, "./ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./\""], + [false, ".//foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \".//foo\""], + [false, "./foo/ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo/\""], + [false, "./foo//bar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo//bar\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo//bar.txt\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo//bar.txt\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo/\""], + # escaped chars + [true, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n"], + [false, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\\056\n", + "Manifest invalid for stream 1: invalid file token \"0:0:\\\\056\\\\056\""], + [false, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\\056\\057foo\n", + "Manifest invalid for stream 1: invalid file token \"0:0:\\\\056\\\\056\\\\057foo\""], + [false, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 0\\0720\\072foo\n", + "Manifest invalid for stream 1: invalid file token \"0\\\\0720\\\\072foo\""], + [false, "./empty_dir d41d8cd98f00b204e9800998ecf8427e+0 \\060:\\060:foo\n", + "Manifest invalid for stream 1: invalid file token \"\\\\060:\\\\060:foo\""], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\057bar\n"], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\072\n"], + [true, ".\\057Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"], + [true, "\\056\\057Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"], + [true, "./\\134444 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"], + [false, "./\\\\444 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./\\\\\\\\444\""], + [true, "./\\011foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"], + [false, "./\\011/.. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./\\\\011/..\""], + [false, ".\\056\\057 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \".\\\\056\\\\057\""], + [false, ".\\057\\056 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \".\\\\057\\\\056\""], + [false, ".\\057Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\444\n", + "Manifest invalid for stream 1: >8-bit encoded chars not allowed on file token \"0:0:foo\\\\444\""], + [false, "./\\444 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: >8-bit encoded chars not allowed on stream token \"./\\\\444\""], + [false, "./\tfoo d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./\\tfoo\""], + [false, "./foo\\ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\\\\""], + [false, "./foo\\r d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\\\r\""], + [false, "./foo\\444 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: >8-bit encoded chars not allowed on stream token \"./foo\\\\444\""], + [false, "./foo\\888 d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\\\888\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo\\\\\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\r\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo\\\\r\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\444\n", + "Manifest invalid for stream 1: >8-bit encoded chars not allowed on file token \"0:0:foo\\\\444\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\888\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo\\\\888\""], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\\057/bar\n", + "Manifest invalid for stream 1: invalid file token \"0:0:foo\\\\057/bar\""], + [false, ".\\057/Data d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n", + "Manifest invalid for stream 1: missing or invalid stream name \".\\\\057/Data\""], + [true, "./Data\\040Folder d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo\n"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\057foo/bar\n", + "Manifest invalid for stream 1: invalid file token \"0:0:\\\\057foo/bar\""], + [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134057foo/bar\n"], + [false, ". d41d8cd98f00b204e9800998ecf8427e+0 \\040:\\040:foo.txt\n", + "Manifest invalid for stream 1: invalid file token \"\\\\040:\\\\040:foo.txt\""], ].each do |ok, manifest, expected_error=nil| - define_method "test_manifest_valid_#{ok}_#{manifest}_and_expect_error_#{expected_error}" do + define_method "test_validate manifest #{manifest.inspect}" do + assert_equal ok, Keep::Manifest.valid?(manifest) if ok - assert Keep::Manifest.valid? manifest + assert Keep::Manifest.validate! manifest else begin - Keep::Manifest.valid? manifest + Keep::Manifest.validate! manifest rescue ArgumentError => e msg = e.message end refute_nil msg, "Expected ArgumentError" - assert msg.include?(expected_error), "Did not find error message #{expected_error} in #{msg}" + assert msg.include?(expected_error), "Did not find expected error message. Expected: #{expected_error}; Actual: #{msg}" end end end