2 # -*- coding: utf-8 -*-
16 from cStringIO import StringIO
19 import arvados.commands.put as arv_put
21 from arvados_testutil import ArvadosBaseTestCase, ArvadosKeepLocalStoreTestCase
22 import run_test_server
24 class ArvadosPutResumeCacheTest(ArvadosBaseTestCase):
28 ['/dev/null', '--filename', 'empty'],
30 ['/tmp', '--max-manifest-depth', '0'],
31 ['/tmp', '--max-manifest-depth', '1']
35 super(ArvadosPutResumeCacheTest, self).tearDown()
37 self.last_cache.destroy()
38 except AttributeError:
41 def cache_path_from_arglist(self, arglist):
42 return arv_put.ResumeCache.make_path(arv_put.parse_arguments(arglist))
44 def test_cache_names_stable(self):
45 for argset in self.CACHE_ARGSET:
46 self.assertEquals(self.cache_path_from_arglist(argset),
47 self.cache_path_from_arglist(argset),
48 "cache name changed for {}".format(argset))
50 def test_cache_names_unique(self):
52 for argset in self.CACHE_ARGSET:
53 path = self.cache_path_from_arglist(argset)
54 self.assertNotIn(path, results)
57 def test_cache_names_simple(self):
58 # The goal here is to make sure the filename doesn't use characters
59 # reserved by the filesystem. Feel free to adjust this regexp as
60 # long as it still does that.
61 bad_chars = re.compile(r'[^-\.\w]')
62 for argset in self.CACHE_ARGSET:
63 path = self.cache_path_from_arglist(argset)
64 self.assertFalse(bad_chars.search(os.path.basename(path)),
65 "path too exotic: {}".format(path))
67 def test_cache_names_ignore_argument_order(self):
69 self.cache_path_from_arglist(['a', 'b', 'c']),
70 self.cache_path_from_arglist(['c', 'a', 'b']))
72 self.cache_path_from_arglist(['-', '--filename', 'stdin']),
73 self.cache_path_from_arglist(['--filename', 'stdin', '-']))
75 def test_cache_names_differ_for_similar_paths(self):
76 # This test needs names at / that don't exist on the real filesystem.
78 self.cache_path_from_arglist(['/_arvputtest1', '/_arvputtest2']),
79 self.cache_path_from_arglist(['/_arvputtest1/_arvputtest2']))
81 def test_cache_names_ignore_irrelevant_arguments(self):
82 # Workaround: parse_arguments bails on --filename with a directory.
83 path1 = self.cache_path_from_arglist(['/tmp'])
84 args = arv_put.parse_arguments(['/tmp'])
86 path2 = arv_put.ResumeCache.make_path(args)
87 self.assertEquals(path1, path2,
88 "cache path considered --filename for directory")
90 self.cache_path_from_arglist(['-']),
91 self.cache_path_from_arglist(['-', '--max-manifest-depth', '1']),
92 "cache path considered --max-manifest-depth for file")
94 def test_cache_names_treat_negative_manifest_depths_identically(self):
95 base_args = ['/tmp', '--max-manifest-depth']
97 self.cache_path_from_arglist(base_args + ['-1']),
98 self.cache_path_from_arglist(base_args + ['-2']))
100 def test_cache_names_treat_stdin_consistently(self):
102 self.cache_path_from_arglist(['-', '--filename', 'test']),
103 self.cache_path_from_arglist(['/dev/stdin', '--filename', 'test']))
105 def test_cache_names_identical_for_synonymous_names(self):
107 self.cache_path_from_arglist(['.']),
108 self.cache_path_from_arglist([os.path.realpath('.')]))
109 testdir = self.make_tmpdir()
110 looplink = os.path.join(testdir, 'loop')
111 os.symlink(testdir, looplink)
113 self.cache_path_from_arglist([testdir]),
114 self.cache_path_from_arglist([looplink]))
116 def test_cache_names_different_by_api_host(self):
117 config = arvados.config.settings()
118 orig_host = config.get('ARVADOS_API_HOST')
120 name1 = self.cache_path_from_arglist(['.'])
121 config['ARVADOS_API_HOST'] = 'x' + (orig_host or 'localhost')
122 self.assertNotEqual(name1, self.cache_path_from_arglist(['.']))
124 if orig_host is None:
125 del config['ARVADOS_API_HOST']
127 config['ARVADOS_API_HOST'] = orig_host
129 def test_basic_cache_storage(self):
130 thing = ['test', 'list']
131 with tempfile.NamedTemporaryFile() as cachefile:
132 self.last_cache = arv_put.ResumeCache(cachefile.name)
133 self.last_cache.save(thing)
134 self.assertEquals(thing, self.last_cache.load())
136 def test_empty_cache(self):
137 with tempfile.NamedTemporaryFile() as cachefile:
138 cache = arv_put.ResumeCache(cachefile.name)
139 self.assertRaises(ValueError, cache.load)
141 def test_cache_persistent(self):
142 thing = ['test', 'list']
143 path = os.path.join(self.make_tmpdir(), 'cache')
144 cache = arv_put.ResumeCache(path)
147 self.last_cache = arv_put.ResumeCache(path)
148 self.assertEquals(thing, self.last_cache.load())
150 def test_multiple_cache_writes(self):
151 thing = ['short', 'list']
152 with tempfile.NamedTemporaryFile() as cachefile:
153 self.last_cache = arv_put.ResumeCache(cachefile.name)
154 # Start writing an object longer than the one we test, to make
155 # sure the cache file gets truncated.
156 self.last_cache.save(['long', 'long', 'list'])
157 self.last_cache.save(thing)
158 self.assertEquals(thing, self.last_cache.load())
160 def test_cache_is_locked(self):
161 with tempfile.NamedTemporaryFile() as cachefile:
162 cache = arv_put.ResumeCache(cachefile.name)
163 self.assertRaises(arv_put.ResumeCacheConflict,
164 arv_put.ResumeCache, cachefile.name)
166 def test_cache_stays_locked(self):
167 with tempfile.NamedTemporaryFile() as cachefile:
168 self.last_cache = arv_put.ResumeCache(cachefile.name)
169 path = cachefile.name
170 self.last_cache.save('test')
171 self.assertRaises(arv_put.ResumeCacheConflict,
172 arv_put.ResumeCache, path)
174 def test_destroy_cache(self):
175 cachefile = tempfile.NamedTemporaryFile(delete=False)
177 cache = arv_put.ResumeCache(cachefile.name)
181 arv_put.ResumeCache(cachefile.name)
182 except arv_put.ResumeCacheConflict:
183 self.fail("could not load cache after destroying it")
184 self.assertRaises(ValueError, cache.load)
186 if os.path.exists(cachefile.name):
187 os.unlink(cachefile.name)
189 def test_restart_cache(self):
190 path = os.path.join(self.make_tmpdir(), 'cache')
191 cache = arv_put.ResumeCache(path)
194 self.assertRaises(ValueError, cache.load)
195 self.assertRaises(arv_put.ResumeCacheConflict,
196 arv_put.ResumeCache, path)
199 class ArvadosPutCollectionWriterTest(ArvadosKeepLocalStoreTestCase):
201 super(ArvadosPutCollectionWriterTest, self).setUp()
202 with tempfile.NamedTemporaryFile(delete=False) as cachefile:
203 self.cache = arv_put.ResumeCache(cachefile.name)
204 self.cache_filename = cachefile.name
207 super(ArvadosPutCollectionWriterTest, self).tearDown()
208 if os.path.exists(self.cache_filename):
212 def test_writer_caches(self):
213 cwriter = arv_put.ArvPutCollectionWriter(self.cache)
214 cwriter.write_file('/dev/null')
215 cwriter.cache_state()
216 self.assertTrue(self.cache.load())
217 self.assertEquals(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:null\n", cwriter.manifest_text())
219 def test_writer_works_without_cache(self):
220 cwriter = arv_put.ArvPutCollectionWriter()
221 cwriter.write_file('/dev/null')
222 self.assertEquals(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:null\n", cwriter.manifest_text())
224 def test_writer_resumes_from_cache(self):
225 cwriter = arv_put.ArvPutCollectionWriter(self.cache)
226 with self.make_test_file() as testfile:
227 cwriter.write_file(testfile.name, 'test')
228 cwriter.cache_state()
229 new_writer = arv_put.ArvPutCollectionWriter.from_cache(
232 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:test\n",
233 new_writer.manifest_text())
235 def test_new_writer_from_stale_cache(self):
236 cwriter = arv_put.ArvPutCollectionWriter(self.cache)
237 with self.make_test_file() as testfile:
238 cwriter.write_file(testfile.name, 'test')
239 new_writer = arv_put.ArvPutCollectionWriter.from_cache(self.cache)
240 new_writer.write_file('/dev/null')
241 self.assertEquals(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:null\n", new_writer.manifest_text())
243 def test_new_writer_from_empty_cache(self):
244 cwriter = arv_put.ArvPutCollectionWriter.from_cache(self.cache)
245 cwriter.write_file('/dev/null')
246 self.assertEquals(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:null\n", cwriter.manifest_text())
248 def test_writer_resumable_after_arbitrary_bytes(self):
249 cwriter = arv_put.ArvPutCollectionWriter(self.cache)
250 # These bytes are intentionally not valid UTF-8.
251 with self.make_test_file('\x00\x07\xe2') as testfile:
252 cwriter.write_file(testfile.name, 'test')
253 cwriter.cache_state()
254 new_writer = arv_put.ArvPutCollectionWriter.from_cache(
256 self.assertEquals(cwriter.manifest_text(), new_writer.manifest_text())
258 def make_progress_tester(self):
260 def record_func(written, expected):
261 progression.append((written, expected))
262 return progression, record_func
264 def test_progress_reporting(self):
265 for expect_count in (None, 8):
266 progression, reporter = self.make_progress_tester()
267 cwriter = arv_put.ArvPutCollectionWriter(
268 reporter=reporter, bytes_expected=expect_count)
269 with self.make_test_file() as testfile:
270 cwriter.write_file(testfile.name, 'test')
271 cwriter.finish_current_stream()
272 self.assertIn((4, expect_count), progression)
274 def test_resume_progress(self):
275 cwriter = arv_put.ArvPutCollectionWriter(self.cache, bytes_expected=4)
276 with self.make_test_file() as testfile:
277 # Set up a writer with some flushed bytes.
278 cwriter.write_file(testfile.name, 'test')
279 cwriter.finish_current_stream()
280 cwriter.cache_state()
281 new_writer = arv_put.ArvPutCollectionWriter.from_cache(self.cache)
282 self.assertEqual(new_writer.bytes_written, 4)
285 class ArvadosExpectedBytesTest(ArvadosBaseTestCase):
286 TEST_SIZE = os.path.getsize(__file__)
288 def test_expected_bytes_for_file(self):
289 self.assertEquals(self.TEST_SIZE,
290 arv_put.expected_bytes_for([__file__]))
292 def test_expected_bytes_for_tree(self):
293 tree = self.make_tmpdir()
294 shutil.copyfile(__file__, os.path.join(tree, 'one'))
295 shutil.copyfile(__file__, os.path.join(tree, 'two'))
296 self.assertEquals(self.TEST_SIZE * 2,
297 arv_put.expected_bytes_for([tree]))
298 self.assertEquals(self.TEST_SIZE * 3,
299 arv_put.expected_bytes_for([tree, __file__]))
301 def test_expected_bytes_for_device(self):
302 self.assertIsNone(arv_put.expected_bytes_for(['/dev/null']))
303 self.assertIsNone(arv_put.expected_bytes_for([__file__, '/dev/null']))
306 class ArvadosPutReportTest(ArvadosBaseTestCase):
307 def test_machine_progress(self):
308 for count, total in [(0, 1), (0, None), (1, None), (235, 9283)]:
309 expect = ": {} written {} total\n".format(
310 count, -1 if (total is None) else total)
312 arv_put.machine_progress(count, total).endswith(expect))
314 def test_known_human_progress(self):
315 for count, total in [(0, 1), (2, 4), (45, 60)]:
316 expect = '{:.1%}'.format(float(count) / total)
317 actual = arv_put.human_progress(count, total)
318 self.assertTrue(actual.startswith('\r'))
319 self.assertIn(expect, actual)
321 def test_unknown_human_progress(self):
322 for count in [1, 20, 300, 4000, 50000]:
323 self.assertTrue(re.search(r'\b{}\b'.format(count),
324 arv_put.human_progress(count, None)))
327 class ArvadosPutProjectLinkTest(ArvadosBaseTestCase):
328 Z_UUID = 'zzzzz-zzzzz-zzzzzzzzzzzzzzz'
331 self.stderr = StringIO()
332 super(ArvadosPutProjectLinkTest, self).setUp()
336 super(ArvadosPutProjectLinkTest, self).tearDown()
338 def prep_link_from_arguments(self, args, uuid_found=True):
340 link = arv_put.prep_project_link(arv_put.parse_arguments(args),
342 lambda uuid: uuid_found)
347 def check_link(self, link, project_uuid, link_name=None):
348 self.assertEqual(project_uuid, link.get('tail_uuid'))
349 self.assertEqual('name', link.get('link_class'))
350 if link_name is None:
351 self.assertNotIn('name', link)
353 self.assertEqual(link_name, link.get('name'))
354 self.assertNotIn('head_uuid', link)
356 def check_stderr_empty(self):
357 self.assertEqual('', self.stderr.getvalue())
359 def test_project_link_with_name(self):
360 link = self.prep_link_from_arguments(['--project-uuid', self.Z_UUID,
361 '--name', 'test link AAA'])
362 self.check_link(link, self.Z_UUID, 'test link AAA')
363 self.check_stderr_empty()
365 def test_project_link_without_name(self):
366 link = self.prep_link_from_arguments(['--project-uuid', self.Z_UUID])
367 self.check_link(link, self.Z_UUID)
368 self.check_stderr_empty()
370 def test_collection_without_project_warned(self):
371 self.assertIsNone(self.prep_link_from_arguments([]))
372 for line in self.stderr:
373 if "--project-uuid or --name" in line:
376 self.fail("no warning emitted about the lack of project name")
378 def test_no_link_or_warning_with_no_collection(self):
379 self.assertIsNone(self.prep_link_from_arguments(['--raw']))
380 self.check_stderr_empty()
382 def test_error_when_project_not_found(self):
383 self.assertRaises(ValueError,
384 self.prep_link_from_arguments,
385 ['--project-uuid', self.Z_UUID], False)
387 def test_name_without_project_is_error(self):
388 self.assertRaises(ValueError,
389 self.prep_link_from_arguments,
392 def test_link_without_collection_is_error(self):
393 self.assertRaises(ValueError,
394 self.prep_link_from_arguments,
395 ['--project-uuid', self.Z_UUID, '--stream'])
398 class ArvadosPutTest(ArvadosKeepLocalStoreTestCase):
399 def call_main_with_args(self, args):
400 self.main_stdout = StringIO()
401 self.main_stderr = StringIO()
402 return arv_put.main(args, self.main_stdout, self.main_stderr)
404 def call_main_on_test_file(self):
405 with self.make_test_file() as testfile:
407 self.call_main_with_args(['--stream', '--no-progress', path])
409 os.path.exists(os.path.join(os.environ['KEEP_LOCAL_STORE'],
410 '098f6bcd4621d373cade4e832627b4f6')),
411 "did not find file stream in Keep store")
414 for outbuf in ['main_stdout', 'main_stderr']:
415 if hasattr(self, outbuf):
416 getattr(self, outbuf).close()
417 delattr(self, outbuf)
418 super(ArvadosPutTest, self).tearDown()
420 def test_simple_file_put(self):
421 self.call_main_on_test_file()
423 def test_put_with_unwriteable_cache_dir(self):
424 orig_cachedir = arv_put.ResumeCache.CACHE_DIR
425 cachedir = self.make_tmpdir()
426 os.chmod(cachedir, 0o0)
427 arv_put.ResumeCache.CACHE_DIR = cachedir
429 self.call_main_on_test_file()
431 arv_put.ResumeCache.CACHE_DIR = orig_cachedir
432 os.chmod(cachedir, 0o700)
434 def test_put_with_unwritable_cache_subdir(self):
435 orig_cachedir = arv_put.ResumeCache.CACHE_DIR
436 cachedir = self.make_tmpdir()
437 os.chmod(cachedir, 0o0)
438 arv_put.ResumeCache.CACHE_DIR = os.path.join(cachedir, 'cachedir')
440 self.call_main_on_test_file()
442 arv_put.ResumeCache.CACHE_DIR = orig_cachedir
443 os.chmod(cachedir, 0o700)
445 def test_link_without_project_uuid_aborts(self):
446 self.assertRaises(SystemExit, self.call_main_with_args,
447 ['--name', 'test without project UUID', '/dev/null'])
449 def test_link_without_collection_aborts(self):
450 self.assertRaises(SystemExit, self.call_main_with_args,
451 ['--name', 'test without Collection',
452 '--stream', '/dev/null'])
454 class ArvPutIntegrationTest(unittest.TestCase):
455 PROJECT_UUID = run_test_server.fixture('groups')['aproject']['uuid']
457 ENVIRON['PYTHONPATH'] = ':'.join(sys.path)
462 del os.environ['KEEP_LOCAL_STORE']
466 # Use the blob_signing_key from the Rails "test" configuration
467 # to provision the Keep server.
468 with open(os.path.join(os.path.dirname(__file__),
469 run_test_server.ARV_API_SERVER_DIR,
471 "application.yml")) as f:
472 rails_config = yaml.load(f.read())
474 config_blob_signing_key = rails_config["test"]["blob_signing_key"]
476 config_blob_signing_key = rails_config["common"]["blob_signing_key"]
477 run_test_server.run()
478 run_test_server.run_keep(blob_signing_key=config_blob_signing_key,
479 enforce_permissions=True)
482 def tearDownClass(cls):
483 run_test_server.stop()
484 run_test_server.stop_keep()
486 def authorize_with(self, token_name):
487 run_test_server.authorize_with(token_name)
488 for v in ["ARVADOS_API_HOST",
489 "ARVADOS_API_HOST_INSECURE",
490 "ARVADOS_API_TOKEN"]:
491 os.environ[v] = arvados.config.settings()[v]
493 def test_check_real_project_found(self):
494 self.assertTrue(arv_put.check_project_exists(self.PROJECT_UUID),
495 "did not correctly find test fixture project")
497 def test_check_error_finding_nonexistent_project(self):
498 BAD_UUID = 'zzzzz-zzzzz-zzzzzzzzzzzzzzz'
500 result = arv_put.check_project_exists(BAD_UUID)
501 except ValueError as error:
502 self.assertIn(BAD_UUID, error.message)
504 self.assertFalse(result, "incorrectly found nonexistent project")
506 def test_short_put_from_stdin(self):
507 # Have to run this as an integration test since arv-put can't
508 # read from the tests' stdin.
509 # arv-put usually can't stat(os.path.realpath('/dev/stdin')) in this
510 # case, because the /proc entry is already gone by the time it tries.
511 pipe = subprocess.Popen(
512 [sys.executable, arv_put.__file__, '--stream'],
513 stdin=subprocess.PIPE, stdout=subprocess.PIPE,
514 stderr=subprocess.STDOUT, env=self.ENVIRON)
515 pipe.stdin.write('stdin test\n')
517 deadline = time.time() + 5
518 while (pipe.poll() is None) and (time.time() < deadline):
520 returncode = pipe.poll()
521 if returncode is None:
523 self.fail("arv-put did not PUT from stdin within 5 seconds")
524 elif returncode != 0:
525 sys.stdout.write(pipe.stdout.read())
526 self.fail("arv-put returned exit code {}".format(returncode))
527 self.assertIn('4a9c8b735dce4b5fa3acf221a0b13628+11', pipe.stdout.read())
529 def test_ArvPutSignedManifest(self):
530 # ArvPutSignedManifest runs "arv-put foo" and then attempts to get
531 # the newly created manifest from the API server, testing to confirm
532 # that the block locators in the returned manifest are signed.
533 self.authorize_with('active')
535 # Before doing anything, demonstrate that the collection
536 # we're about to create is not present in our test fixture.
537 api = arvados.api('v1', cache=False)
538 manifest_uuid = "00b4e9f40ac4dd432ef89749f1c01e74+47"
539 with self.assertRaises(apiclient.errors.HttpError):
540 notfound = api.collections().get(uuid=manifest_uuid).execute()
542 datadir = tempfile.mkdtemp()
543 with open(os.path.join(datadir, "foo"), "w") as f:
544 f.write("The quick brown fox jumped over the lazy dog")
545 p = subprocess.Popen([sys.executable, arv_put.__file__, datadir],
546 stdout=subprocess.PIPE, env=self.ENVIRON)
547 (arvout, arverr) = p.communicate()
548 self.assertEqual(p.returncode, 0)
549 self.assertEqual(arverr, None)
550 self.assertEqual(arvout.strip(), manifest_uuid)
552 # The manifest text stored in the API server under the same
553 # manifest UUID must use signed locators.
554 c = api.collections().get(uuid=manifest_uuid).execute()
555 self.assertRegexpMatches(
557 r'^\. 08a008a01d498c404b0c30852b39d3b8\+44\+A[0-9a-f]+@[0-9a-f]+ 0:44:foo\n')
559 os.remove(os.path.join(datadir, "foo"))
562 def run_and_find_link(self, text, extra_args=[]):
563 self.authorize_with('active')
564 pipe = subprocess.Popen(
565 [sys.executable, arv_put.__file__,
566 '--project-uuid', self.PROJECT_UUID] + extra_args,
567 stdin=subprocess.PIPE, stdout=subprocess.PIPE,
568 stderr=subprocess.PIPE, env=self.ENVIRON)
569 stdout, stderr = pipe.communicate(text)
570 link_list = arvados.api('v1', cache=False).links().list(
571 filters=[['head_uuid', '=', stdout.strip()],
572 ['tail_uuid', '=', self.PROJECT_UUID],
573 ['link_class', '=', 'name']]).execute().get('items', [])
574 self.assertEqual(1, len(link_list))
577 def test_put_collection_with_unnamed_project_link(self):
578 link = self.run_and_find_link("Test unnamed collection")
579 username = pwd.getpwuid(os.getuid()).pw_name
580 self.assertRegexpMatches(
582 r'^Collection saved by {}@'.format(re.escape(username)))
584 def test_put_collection_with_named_project_link(self):
585 link_name = 'Test auto Collection Link'
586 link = self.run_and_find_link("Test named collection",
587 ['--name', link_name])
588 self.assertEqual(link_name, link['name'])
591 if __name__ == '__main__':