1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from __future__ import absolute_import
6 from future.utils import listitems
14 from unittest import mock
17 import arvados.collection as collection
18 import arvados.commands.get as arv_get
19 from . import run_test_server
21 from . import arvados_testutil as tutil
22 from .arvados_testutil import ArvadosBaseTestCase
24 class ArvadosGetTestCase(run_test_server.TestCaseWithServers,
31 super(ArvadosGetTestCase, self).setUp()
32 self.tempdir = tempfile.mkdtemp()
33 self.col_loc, self.col_pdh, self.col_manifest = self.write_test_collection()
35 self.stdout = tutil.BytesIO()
36 self.stderr = tutil.StringIO()
37 self.loggingHandler = logging.StreamHandler(self.stderr)
38 self.loggingHandler.setFormatter(logging.Formatter('%(levelname)s: %(message)s'))
39 logging.getLogger().addHandler(self.loggingHandler)
42 logging.getLogger().removeHandler(self.loggingHandler)
43 super(ArvadosGetTestCase, self).tearDown()
44 shutil.rmtree(self.tempdir)
46 def write_test_collection(self,
51 'subdir/baz.txt' : 'baz',
54 c = collection.Collection(api_client=api)
55 for path, data in listitems(contents):
56 with c.open(path, 'wb') as f:
60 api.close_connections()
62 return (c.manifest_locator(),
63 c.portable_data_hash(),
64 c.manifest_text(strip=strip_manifest))
66 def run_get(self, args):
67 self.stdout.seek(0, 0)
68 self.stdout.truncate(0)
69 self.stderr.seek(0, 0)
70 self.stderr.truncate(0)
71 return arv_get.main(args, self.stdout, self.stderr)
73 def test_version_argument(self):
74 with tutil.redirected_streams(
75 stdout=tutil.StringIO, stderr=tutil.StringIO) as (out, err):
76 with self.assertRaises(SystemExit):
77 self.run_get(['--version'])
78 self.assertVersionOutput(out, err)
80 def test_get_single_file(self):
81 # Get the file using the collection's locator
82 r = self.run_get(["{}/subdir/baz.txt".format(self.col_loc), '-'])
83 self.assertEqual(0, r)
84 self.assertEqual(b'baz', self.stdout.getvalue())
86 r = self.run_get(["{}/subdir/baz.txt".format(self.col_pdh), '-'])
87 self.assertEqual(0, r)
88 self.assertEqual(b'baz', self.stdout.getvalue())
90 def test_get_block(self):
91 # Get raw data using a block locator
92 blk = re.search(r' (acbd18\S+\+A\S+) ', self.col_manifest).group(1)
93 r = self.run_get([blk, '-'])
94 self.assertEqual(0, r)
95 self.assertEqual(b'foo', self.stdout.getvalue())
97 def test_get_multiple_files(self):
98 # Download the entire collection to the temp directory
99 r = self.run_get(["{}/".format(self.col_loc), self.tempdir])
100 self.assertEqual(0, r)
101 with open(os.path.join(self.tempdir, "foo.txt"), "r") as f:
102 self.assertEqual("foo", f.read())
103 with open(os.path.join(self.tempdir, "bar.txt"), "r") as f:
104 self.assertEqual("bar", f.read())
105 with open(os.path.join(self.tempdir, "subdir", "baz.txt"), "r") as f:
106 self.assertEqual("baz", f.read())
108 def test_get_collection_unstripped_manifest(self):
109 dummy_token = "+Axxxxxxx"
110 # Get the collection manifest by UUID
111 r = self.run_get([self.col_loc, self.tempdir])
112 self.assertEqual(0, r)
113 m_from_collection = re.sub(r"\+A[0-9a-f@]+", dummy_token, self.col_manifest)
114 with open(os.path.join(self.tempdir, self.col_loc), "r") as f:
115 # Replace manifest tokens before comparison to avoid races
116 m_from_file = re.sub(r"\+A[0-9a-f@]+", dummy_token, f.read())
117 self.assertEqual(m_from_collection, m_from_file)
118 # Get the collection manifest by PDH
119 r = self.run_get([self.col_pdh, self.tempdir])
120 self.assertEqual(0, r)
121 with open(os.path.join(self.tempdir, self.col_pdh), "r") as f:
122 # Replace manifest tokens before comparison to avoid races
123 m_from_file = re.sub(r"\+A[0-9a-f@]+", dummy_token, f.read())
124 self.assertEqual(m_from_collection, m_from_file)
126 def test_get_collection_stripped_manifest(self):
127 col_loc, col_pdh, col_manifest = self.write_test_collection(
129 # Get the collection manifest by UUID
130 r = self.run_get(['--strip-manifest', col_loc, self.tempdir])
131 self.assertEqual(0, r)
132 with open(os.path.join(self.tempdir, col_loc), "r") as f:
133 self.assertEqual(col_manifest, f.read())
134 # Get the collection manifest by PDH
135 r = self.run_get(['--strip-manifest', col_pdh, self.tempdir])
136 self.assertEqual(0, r)
137 with open(os.path.join(self.tempdir, col_pdh), "r") as f:
138 self.assertEqual(col_manifest, f.read())
140 def test_invalid_collection(self):
141 # Asking for an invalid collection should generate an error.
142 r = self.run_get(['this-uuid-seems-to-be-fake', self.tempdir])
143 self.assertNotEqual(0, r)
145 def test_invalid_file_request(self):
146 # Asking for an inexistant file within a collection should generate an error.
147 r = self.run_get(["{}/im-not-here.txt".format(self.col_loc), self.tempdir])
148 self.assertNotEqual(0, r)
150 def test_invalid_destination(self):
151 # Asking to place the collection's files on a non existant directory
152 # should generate an error.
153 r = self.run_get([self.col_loc, "/fake/subdir/"])
154 self.assertNotEqual(0, r)
156 def test_preexistent_destination(self):
157 # Asking to place a file with the same path as a local one should
158 # generate an error and avoid overwrites.
159 with open(os.path.join(self.tempdir, "foo.txt"), "w") as f:
160 f.write("another foo")
161 r = self.run_get(["{}/foo.txt".format(self.col_loc), self.tempdir])
162 self.assertNotEqual(0, r)
163 with open(os.path.join(self.tempdir, "foo.txt"), "r") as f:
164 self.assertEqual("another foo", f.read())
166 def test_no_progress_when_stderr_not_a_tty(self):
167 # Create a collection with a big file (>64MB) to force the progress
169 c = collection.Collection()
170 with c.open('bigfile.txt', 'wb') as f:
172 f.write("x" * 1024 * 1024)
174 tmpdir = self.make_tmpdir()
175 # Simulate a TTY stderr
176 stderr = mock.MagicMock()
177 stdout = tutil.BytesIO()
179 # Confirm that progress is written to stderr when is a tty
180 stderr.isatty.return_value = True
181 r = arv_get.main(['{}/bigfile.txt'.format(c.manifest_locator()),
182 '{}/bigfile.txt'.format(tmpdir)],
184 self.assertEqual(0, r)
185 self.assertEqual(b'', stdout.getvalue())
186 self.assertTrue(stderr.write.called)
188 # Clean up and reset stderr mock
189 os.remove('{}/bigfile.txt'.format(tmpdir))
190 stderr = mock.MagicMock()
191 stdout = tutil.BytesIO()
193 # Confirm that progress is not written to stderr when isn't a tty
194 stderr.isatty.return_value = False
195 r = arv_get.main(['{}/bigfile.txt'.format(c.manifest_locator()),
196 '{}/bigfile.txt'.format(tmpdir)],
198 self.assertEqual(0, r)
199 self.assertEqual(b'', stdout.getvalue())
200 self.assertFalse(stderr.write.called)
202 request_id_regex = r'INFO: X-Request-Id: req-[a-z0-9]{20}\n'
204 def test_request_id_logging_on(self):
205 r = self.run_get(["-v", "{}/".format(self.col_loc), self.tempdir])
206 self.assertEqual(0, r)
207 self.assertRegex(self.stderr.getvalue(), self.request_id_regex)
209 def test_request_id_logging_off(self):
210 r = self.run_get(["{}/".format(self.col_loc), self.tempdir])
211 self.assertEqual(0, r)
212 self.assertNotRegex(self.stderr.getvalue(), self.request_id_regex)