import arvados
import bz2
import copy
+import mock
import os
import pprint
import subprocess
import tempfile
import unittest
-from arvados_testutil import ArvadosKeepLocalStoreTestCase
+import run_test_server
+import arvados_testutil as tutil
class TestResumableWriter(arvados.ResumableCollectionWriter):
KEEP_BLOCK_SIZE = 1024 # PUT to Keep every 1K.
return self.dump_state(copy.deepcopy)
-class ArvadosCollectionsTest(ArvadosKeepLocalStoreTestCase):
+class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
+ tutil.ArvadosBaseTestCase):
+ MAIN_SERVER = {}
+
+ @classmethod
+ def setUpClass(cls):
+ super(ArvadosCollectionsTest, cls).setUpClass()
+ run_test_server.authorize_with('active')
+ cls.api_client = arvados.api('v1')
+ cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
+ local_store=cls.local_store)
+
def write_foo_bar_baz(self):
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
self.assertEqual(cw.current_stream_name(), '.',
'current_stream_name() should be "." now')
cw.set_current_file_name('foo.txt')
return cw.finish()
def test_keep_local_store(self):
- self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
- self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
+ self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
+ self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
def test_local_collection_writer(self):
self.assertEqual(self.write_foo_bar_baz(),
def test_local_collection_reader(self):
self.write_foo_bar_baz()
- cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle')
+ cr = arvados.CollectionReader(
+ 'd6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle', self.api_client)
got = []
for s in cr.all_streams():
for f in s.all_files():
'reading zero bytes should have returned empty string')
def _test_subset(self, collection, expected):
- cr = arvados.CollectionReader(collection)
+ cr = arvados.CollectionReader(collection, self.api_client)
for s in cr.all_streams():
for ex in expected:
if ex[0] == s:
[3, '.', 'foo.txt', 'foo'],
[3, './baz', 'baz.txt', 'baz']])
self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
- (arvados.Keep.put("foo"),
- arvados.Keep.put("bar"))),
+ (self.keep_client.put("foo"),
+ self.keep_client.put("bar"))),
[[3, '.', 'bar.txt', 'bar'],
[3, '.', 'foo.txt', 'foo']])
self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
- (arvados.Keep.put("foo"),
- arvados.Keep.put("bar"))),
+ (self.keep_client.put("foo"),
+ self.keep_client.put("bar"))),
[[2, '.', 'fo.txt', 'fo'],
[4, '.', 'obar.txt', 'obar']])
self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
- (arvados.Keep.put("foo"),
- arvados.Keep.put("bar"))),
+ (self.keep_client.put("foo"),
+ self.keep_client.put("bar"))),
[[2, '.', 'ar.txt', 'ar'],
[2, '.', 'fo.txt', 'fo'],
[2, '.', 'ob.txt', 'ob'],
[0, '.', 'zero.txt', '']])
def _test_readline(self, what_in, what_out):
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
cw.start_new_file('test.txt')
cw.write(what_in)
test1 = cw.finish()
- cr = arvados.CollectionReader(test1)
+ cr = arvados.CollectionReader(test1, self.api_client)
got = []
for x in list(cr.all_files())[0].readlines():
got += [x]
["ab\n", "cd\n"])
def test_collection_empty_file(self):
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
cw.start_new_file('zero.txt')
cw.write('')
self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
self.check_manifest_file_sizes(cw.manifest_text(), [0])
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
cw.start_new_file('zero.txt')
cw.write('')
cw.start_new_file('one.txt')
self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
def check_manifest_file_sizes(self, manifest_text, expect_sizes):
- cr = arvados.CollectionReader(manifest_text)
+ cr = arvados.CollectionReader(manifest_text, self.api_client)
got_sizes = []
for f in cr.all_files():
got_sizes += [f.size()]
for x in xrange(0, 18):
data_in += data_in
compressed_data_in = bz2.compress(data_in)
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
cw.start_new_file('test.bz2')
cw.write(compressed_data_in)
bz2_manifest = cw.manifest_text()
- cr = arvados.CollectionReader(bz2_manifest)
+ cr = arvados.CollectionReader(bz2_manifest, self.api_client)
got = 0
for x in list(cr.all_files())[0].readlines():
shell=False, close_fds=True)
compressed_data_in, stderrdata = p.communicate(data_in)
- cw = arvados.CollectionWriter()
+ cw = arvados.CollectionWriter(self.api_client)
cw.start_new_file('test.gz')
cw.write(compressed_data_in)
gzip_manifest = cw.manifest_text()
- cr = arvados.CollectionReader(gzip_manifest)
+ cr = arvados.CollectionReader(gzip_manifest, self.api_client)
got = 0
for x in list(cr.all_files())[0].readlines():
self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
- self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(),
""". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
""")
m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
"""
- self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
+ self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(), m2)
m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
- self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(),
""". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
""")
m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
- self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(),
"""./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
""")
m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
- self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(),
"""./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
""")
with self.data_file('1000G_ref_manifest') as f6:
m6 = f6.read()
- self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
+ self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(), m6)
with self.data_file('jlake_manifest') as f7:
m7 = f7.read()
- self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
+ self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(), m7)
m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
"""
- self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
+ self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(), m8)
def test_locators_and_ranges(self):
blocks2 = [['a', 10, 0],
class MockStreamReader(object):
def __init__(self, content):
self.content = content
+ self.num_retries = 0
- def readfrom(self, start, size):
+ def readfrom(self, start, size, num_retries=0):
return self.content[start:start+size]
def test_file_stream(self):
class MockKeep(object):
- def __init__(self, content):
+ def __init__(self, content, num_retries=0):
self.content = content
- def get(self, locator):
+ def get(self, locator, num_retries=0):
return self.content[locator]
def test_stream_reader(self):
f = sr.files()["foo"]
+ # f.read() calls will be aligned on block boundaries (as a
+ # result of ticket #3663).
+
f.seek(0)
- self.assertEqual(f.read(20), content[0:20])
+ self.assertEqual(f.read(20), content[0:10])
f.seek(0)
self.assertEqual(f.read(6), content[0:6])
- self.assertEqual(f.read(6), content[6:12])
- self.assertEqual(f.read(6), content[12:18])
+ self.assertEqual(f.read(6), content[6:10])
+ self.assertEqual(f.read(6), content[10:16])
+ self.assertEqual(f.read(6), content[16:20])
def test_extract_file(self):
m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
- m2 = arvados.CollectionReader(m1)
+ m2 = arvados.CollectionReader(m1, self.api_client)
self.assertEqual(m2.manifest_text(),
". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt 43:41:md6sum.txt 84:43:md7sum.txt 6:37:md8sum.txt 84:43:md8sum.txt 83:1:md9sum.txt 0:43:md9sum.txt 84:36:md9sum.txt\n")
- self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md5sum.txt'].as_manifest(),
". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
- self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md6sum.txt'].as_manifest(),
". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
- self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md7sum.txt'].as_manifest(),
". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
- self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md9sum.txt'].as_manifest(),
". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
def test_write_directory_tree(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
cwriter.write_directory_tree(self.build_directory_tree(
['basefile', 'subdir/subfile']))
self.assertEqual(cwriter.manifest_text(),
./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
def test_write_named_directory_tree(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
cwriter.write_directory_tree(self.build_directory_tree(
['basefile', 'subdir/subfile']), 'root')
self.assertEqual(
./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
def test_write_directory_tree_in_one_stream(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
cwriter.write_directory_tree(self.build_directory_tree(
['basefile', 'subdir/subfile']), max_manifest_depth=0)
self.assertEqual(cwriter.manifest_text(),
./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
def test_write_directory_tree_with_limited_recursion(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
cwriter.write_directory_tree(
self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
max_manifest_depth=1)
./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
def test_write_one_file(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
with self.make_test_file() as testfile:
cwriter.write_file(testfile.name)
self.assertEqual(
os.path.basename(testfile.name)))
def test_write_named_file(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
with self.make_test_file() as testfile:
cwriter.write_file(testfile.name, 'foo')
self.assertEqual(cwriter.manifest_text(),
". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
def test_write_multiple_files(self):
- cwriter = arvados.CollectionWriter()
+ cwriter = arvados.CollectionWriter(self.api_client)
for letter in 'ABC':
with self.make_test_file(letter) as testfile:
cwriter.write_file(testfile.name, letter)
cwriter.write, "badtext")
+class CollectionTestMixin(object):
+ PROXY_RESPONSE = {
+ 'items_available': 1,
+ 'items': [{
+ 'uuid': 'zzzzz-bi6l4-mockproxy012345',
+ 'owner_uuid': 'zzzzz-tpzed-mockowner012345',
+ 'service_host': tutil.TEST_HOST,
+ 'service_port': 65535,
+ 'service_ssl_flag': True,
+ 'service_type': 'proxy',
+ }]}
+ API_COLLECTIONS = run_test_server.fixture('collections')
+ DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
+ DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
+ DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
+ DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
+
+ def _mock_api_call(self, mock_method, code, body):
+ mock_method = mock_method().execute
+ if code == 200:
+ mock_method.return_value = body
+ else:
+ mock_method.side_effect = arvados.errors.ApiError(
+ tutil.fake_httplib2_response(code), "{}")
+
+ def mock_keep_services(self, api_mock, code, body):
+ self._mock_api_call(api_mock.keep_services().accessible, code, body)
+
+ def api_client_mock(self, code=200):
+ client = mock.MagicMock(name='api_client')
+ self.mock_keep_services(client, code, self.PROXY_RESPONSE)
+ return client
+
+
+@tutil.skip_sleep
+class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
+ def mock_get_collection(self, api_mock, code, body):
+ body = self.API_COLLECTIONS.get(body)
+ self._mock_api_call(api_mock.collections().get, code, body)
+
+ def api_client_mock(self, code=200):
+ client = super(CollectionReaderTestCase, self).api_client_mock(code)
+ self.mock_get_collection(client, code, 'foo_file')
+ return client
+
+ def test_init_no_default_retries(self):
+ client = self.api_client_mock(200)
+ reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
+ reader.manifest_text()
+ client.collections().get().execute.assert_called_with(num_retries=0)
+
+ def test_uuid_init_success(self):
+ client = self.api_client_mock(200)
+ reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
+ num_retries=3)
+ self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
+ reader.manifest_text())
+ client.collections().get().execute.assert_called_with(num_retries=3)
+
+ def test_uuid_init_failure_raises_api_error(self):
+ client = self.api_client_mock(500)
+ reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
+ with self.assertRaises(arvados.errors.ApiError):
+ reader.manifest_text()
+
+ def test_locator_init(self):
+ client = self.api_client_mock(200)
+ # Ensure Keep will not return anything if asked.
+ with tutil.mock_responses(None, 404):
+ reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
+ api_client=client)
+ self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
+
+ def test_locator_init_falls_back_to_keep(self):
+ # Reading manifests from Keep is deprecated. Feel free to
+ # remove this test when we remove the fallback.
+ client = self.api_client_mock(200)
+ self.mock_get_collection(client, 404, None)
+ with tutil.mock_responses(self.DEFAULT_MANIFEST, 200):
+ reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
+ api_client=client, num_retries=3)
+ self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
+
+ def test_init_num_retries_propagated(self):
+ # More of an integration test...
+ client = self.api_client_mock(200)
+ reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
+ num_retries=3)
+ with tutil.mock_responses('foo', 500, 500, 200):
+ self.assertEqual('foo',
+ ''.join(f.read(9) for f in reader.all_files()))
+
+
+@tutil.skip_sleep
+class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
+ def mock_keep(self, body, *codes, **headers):
+ headers.setdefault('x-keep-replicas-stored', 2)
+ return tutil.mock_responses(body, *codes, **headers)
+
+ def foo_writer(self, **kwargs):
+ api_client = self.api_client_mock()
+ writer = arvados.CollectionWriter(api_client, **kwargs)
+ writer.start_new_file('foo')
+ writer.write('foo')
+ return writer
+
+ def test_write_whole_collection(self):
+ writer = self.foo_writer()
+ with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
+ self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
+
+ def test_write_no_default(self):
+ writer = self.foo_writer()
+ with self.mock_keep(None, 500):
+ with self.assertRaises(arvados.errors.KeepWriteError):
+ writer.finish()
+
+ def test_write_whole_collection_through_retries(self):
+ writer = self.foo_writer(num_retries=2)
+ with self.mock_keep(self.DEFAULT_DATA_HASH,
+ 500, 500, 200, 500, 500, 200):
+ self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
+
+ def test_flush_data_retries(self):
+ writer = self.foo_writer(num_retries=2)
+ foo_hash = self.DEFAULT_MANIFEST.split()[1]
+ with self.mock_keep(foo_hash, 500, 200):
+ writer.flush_data()
+ self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
+
+
if __name__ == '__main__':
unittest.main()
projects / arvados.git / blobdiff