X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/a7b16675337995c65b909a7519178efed98a3089..44c95f99098fa6c6acbfa82d4b6cbc6015eb6e39:/sdk/python/tests/test_collections.py diff --git a/sdk/python/tests/test_collections.py b/sdk/python/tests/test_collections.py index 892f146e90..e8c6dd163f 100644 --- a/sdk/python/tests/test_collections.py +++ b/sdk/python/tests/test_collections.py @@ -1,22 +1,26 @@ -# usage example: +# Copyright (C) The Arvados Authors. All rights reserved. # -# ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover +# SPDX-License-Identifier: Apache-2.0 +from __future__ import absolute_import + +from builtins import object import arvados import copy -import hashlib import mock import os import pprint +import random import re +import sys import tempfile +import time import unittest -import run_test_server -import arvados_testutil as tutil -from arvados.ranges import Range, LocatorAndRange -from arvados import import_manifest, export_manifest -from arvados.arvfile import SYNC_EXPLICIT +from . import run_test_server +from arvados._ranges import Range, LocatorAndRange +from arvados.collection import Collection, CollectionReader +from . import arvados_testutil as tutil class TestResumableWriter(arvados.ResumableCollectionWriter): KEEP_BLOCK_SIZE = 1024 # PUT to Keep every 1K. @@ -42,13 +46,13 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, self.assertEqual(cw.current_stream_name(), '.', 'current_stream_name() should be "." now') cw.set_current_file_name('foo.txt') - cw.write('foo') + cw.write(b'foo') self.assertEqual(cw.current_file_name(), 'foo.txt', 'current_file_name() should be foo.txt now') cw.start_new_file('bar.txt') - cw.write('bar') + cw.write(b'bar') cw.start_new_stream('baz') - cw.write('baz') + cw.write(b'baz') cw.set_current_file_name('baz.txt') self.assertEqual(cw.manifest_text(), ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" + @@ -57,9 +61,13 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, cw.finish() return cw.portable_data_hash() + def test_pdh_is_native_str(self): + pdh = self.write_foo_bar_baz() + self.assertEqual(type(''), type(pdh)) + def test_keep_local_store(self): - 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') + self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put') + self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get') def test_local_collection_writer(self): self.assertEqual(self.write_foo_bar_baz(), @@ -74,20 +82,20 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, for s in cr.all_streams(): for f in s.all_files(): got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]] - expected = [[3, '.', 'foo.txt', 'foo'], - [3, '.', 'bar.txt', 'bar'], - [3, './baz', 'baz.txt', 'baz']] + expected = [[3, '.', 'foo.txt', b'foo'], + [3, '.', 'bar.txt', b'bar'], + [3, './baz', 'baz.txt', b'baz']] self.assertEqual(got, expected) stream0 = cr.all_streams()[0] self.assertEqual(stream0.readfrom(0, 0), - '', + b'', 'reading zero bytes should have returned empty string') self.assertEqual(stream0.readfrom(0, 2**26), - 'foobar', + b'foobar', 'reading entire stream failed') self.assertEqual(stream0.readfrom(2**26, 0), - '', + b'', 'reading zero bytes should have returned empty string') def _test_subset(self, collection, expected): @@ -104,50 +112,50 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, def test_collection_manifest_subset(self): foobarbaz = self.write_foo_bar_baz() self._test_subset(foobarbaz, - [[3, '.', 'bar.txt', 'bar'], - [3, '.', 'foo.txt', 'foo'], - [3, './baz', 'baz.txt', 'baz']]) + [[3, '.', 'bar.txt', b'bar'], + [3, '.', 'foo.txt', b'foo'], + [3, './baz', 'baz.txt', b'baz']]) self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" % - (self.keep_client.put("foo"), - self.keep_client.put("bar"))), - [[3, '.', 'bar.txt', 'bar'], - [3, '.', 'foo.txt', 'foo']]) + (self.keep_client.put(b"foo"), + self.keep_client.put(b"bar"))), + [[3, '.', 'bar.txt', b'bar'], + [3, '.', 'foo.txt', b'foo']]) self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" % - (self.keep_client.put("foo"), - self.keep_client.put("bar"))), - [[2, '.', 'fo.txt', 'fo'], - [4, '.', 'obar.txt', 'obar']]) + (self.keep_client.put(b"foo"), + self.keep_client.put(b"bar"))), + [[2, '.', 'fo.txt', b'fo'], + [4, '.', 'obar.txt', b'obar']]) self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" % - (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', '']]) + (self.keep_client.put(b"foo"), + self.keep_client.put(b"bar"))), + [[2, '.', 'ar.txt', b'ar'], + [2, '.', 'fo.txt', b'fo'], + [2, '.', 'ob.txt', b'ob'], + [0, '.', 'zero.txt', b'']]) def test_collection_empty_file(self): cw = arvados.CollectionWriter(self.api_client) cw.start_new_file('zero.txt') - cw.write('') + cw.write(b'') self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n") self.check_manifest_file_sizes(cw.manifest_text(), [0]) cw = arvados.CollectionWriter(self.api_client) cw.start_new_file('zero.txt') - cw.write('') + cw.write(b'') cw.start_new_file('one.txt') - cw.write('1') + cw.write(b'1') cw.start_new_stream('foo') cw.start_new_file('zero.txt') - cw.write('') + cw.write(b'') self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0]) def test_no_implicit_normalize(self): cw = arvados.CollectionWriter(self.api_client) cw.start_new_file('b') - cw.write('b') + cw.write(b'b') cw.start_new_file('a') - cw.write('') + cw.write(b'') self.check_manifest_file_sizes(cw.manifest_text(), [1,0]) self.check_manifest_file_sizes( arvados.CollectionReader( @@ -313,14 +321,16 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, return self.content[locator] def test_stream_reader(self): - keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij', - 'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy', - 'cccccccccccccccccccccccccccccccc+5': 'z0123'} + keepblocks = { + 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij', + 'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy', + 'cccccccccccccccccccccccccccccccc+5': b'z0123', + } mk = self.MockKeep(keepblocks) sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk) - content = 'abcdefghijklmnopqrstuvwxyz0123456789' + content = b'abcdefghijklmnopqrstuvwxyz0123456789' self.assertEqual(sr.readfrom(0, 30), content[0:30]) self.assertEqual(sr.readfrom(2, 30), content[2:30]) @@ -334,7 +344,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, self.assertEqual(sr.readfrom(15, 5), content[15:20]) self.assertEqual(sr.readfrom(20, 5), content[20:25]) self.assertEqual(sr.readfrom(25, 5), content[25:30]) - self.assertEqual(sr.readfrom(30, 5), '') + self.assertEqual(sr.readfrom(30, 5), b'') def test_extract_file(self): m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt @@ -396,7 +406,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, def test_write_directory_tree_with_zero_recursion(self): cwriter = arvados.CollectionWriter(self.api_client) content = 'd1/d2/f3d1/f2f1' - blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content)) + blockhash = tutil.str_keep_locator(content) cwriter.write_directory_tree( self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']), max_manifest_depth=0) @@ -423,7 +433,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, def test_write_multiple_files(self): cwriter = arvados.CollectionWriter(self.api_client) for letter in 'ABC': - with self.make_test_file(letter) as testfile: + with self.make_test_file(letter.encode()) as testfile: cwriter.write_file(testfile.name, letter) self.assertEqual( cwriter.manifest_text(), @@ -434,7 +444,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, with self.make_test_file() as testfile: cwriter.write_file(testfile.name, 'test') resumed = TestResumableWriter.from_state(cwriter.current_state()) - self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(), + self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(), "resumed CollectionWriter had different manifest") def test_resume_fails_when_missing_dependency(self): @@ -466,7 +476,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, with self.make_test_file() as testfile: cwriter.write_file(testfile.name, 'test') orig_mtime = os.fstat(testfile.fileno()).st_mtime - testfile.write('extra') + testfile.write(b'extra') testfile.flush() os.utime(testfile.name, (orig_mtime, orig_mtime)) self.assertRaises(arvados.errors.StaleWriterStateError, @@ -503,17 +513,7 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers, ).manifest_text()) -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', - }]} +class CollectionTestMixin(tutil.ApiClientMock): API_COLLECTIONS = run_test_server.fixture('collections') DEFAULT_COLLECTION = API_COLLECTIONS['foo_file'] DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash'] @@ -523,32 +523,21 @@ class CollectionTestMixin(object): ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash'] ALT_MANIFEST = ALT_COLLECTION['manifest_text'] - 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) + def api_client_mock(self, status=200): + client = super(CollectionTestMixin, self).api_client_mock() + self.mock_keep_services(client, status=status, service_type='proxy', count=1) 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) + def mock_get_collection(self, api_mock, code, fixturename): + body = self.API_COLLECTIONS.get(fixturename) 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') + def api_client_mock(self, status=200): + client = super(CollectionReaderTestCase, self).api_client_mock() + self.mock_get_collection(client, status, 'foo_file') return client def test_init_no_default_retries(self): @@ -567,14 +556,13 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): 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() + reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client) def test_locator_init(self): client = self.api_client_mock(200) # Ensure Keep will not return anything if asked. - with tutil.mock_get_responses(None, 404): + with tutil.mock_keep_responses(None, 404): reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH, api_client=client) self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text()) @@ -584,7 +572,7 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): # been written to Keep. client = self.api_client_mock(200) self.mock_get_collection(client, 404, None) - with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200): + with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200): reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH, api_client=client) self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text()) @@ -592,17 +580,16 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): def test_uuid_init_no_fallback_to_keep(self): # Do not look up a collection UUID in Keep. client = self.api_client_mock(404) - reader = arvados.CollectionReader(self.DEFAULT_UUID, - api_client=client) - with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200): + with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200): with self.assertRaises(arvados.errors.ApiError): - reader.manifest_text() + reader = arvados.CollectionReader(self.DEFAULT_UUID, + api_client=client) def test_try_keep_first_if_permission_hint(self): # To verify that CollectionReader tries Keep first here, we # mock API server to return the wrong data. client = self.api_client_mock(200) - with tutil.mock_get_responses(self.ALT_MANIFEST, 200): + with tutil.mock_keep_responses(self.ALT_MANIFEST, 200): self.assertEqual( self.ALT_MANIFEST, arvados.CollectionReader( @@ -614,9 +601,9 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): client = self.api_client_mock(200) reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client, num_retries=3) - with tutil.mock_get_responses('foo', 500, 500, 200): - self.assertEqual('foo', - ''.join(f.read(9) for f in reader.all_files())) + with tutil.mock_keep_responses('foo', 500, 500, 200): + self.assertEqual(b'foo', + b''.join(f.read(9) for f in reader.all_files())) def test_read_nonnormalized_manifest_with_collection_reader(self): # client should be able to use CollectionReader on a manifest without normalizing it @@ -654,7 +641,7 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): def test_api_response_with_collection_from_keep(self): client = self.api_client_mock() self.mock_get_collection(client, 404, 'foo') - with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200): + with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200): reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH, api_client=client) api_response = reader.api_response() @@ -663,19 +650,13 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): def check_open_file(self, coll_file, stream_name, file_name, file_size): self.assertFalse(coll_file.closed, "returned file is not open") self.assertEqual(stream_name, coll_file.stream_name()) - self.assertEqual(file_name, coll_file.name()) + self.assertEqual(file_name, coll_file.name) self.assertEqual(file_size, coll_file.size()) def test_open_collection_file_one_argument(self): client = self.api_client_mock(200) reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client) - cfile = reader.open('./foo') - self.check_open_file(cfile, '.', 'foo', 3) - - def test_open_collection_file_two_arguments(self): - client = self.api_client_mock(200) - reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client) - cfile = reader.open('.', 'foo') + cfile = reader.open('./foo', 'rb') self.check_open_file(cfile, '.', 'foo', 3) def test_open_deep_file(self): @@ -684,32 +665,32 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin): self.mock_get_collection(client, 200, coll_name) reader = arvados.CollectionReader( self.API_COLLECTIONS[coll_name]['uuid'], api_client=client) - cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt') + cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt', 'rb') self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt', 32) def test_open_nonexistent_stream(self): client = self.api_client_mock(200) reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client) - self.assertRaises(ValueError, reader.open, './nonexistent', 'foo') + self.assertRaises(IOError, reader.open, './nonexistent/foo') def test_open_nonexistent_file(self): client = self.api_client_mock(200) reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client) - self.assertRaises(ValueError, reader.open, '.', 'nonexistent') + self.assertRaises(IOError, reader.open, 'nonexistent') @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_put_responses(body, *codes, **headers) + return tutil.mock_keep_responses(body, *codes, **headers) def foo_writer(self, **kwargs): - api_client = self.api_client_mock() - writer = arvados.CollectionWriter(api_client, **kwargs) + kwargs.setdefault('api_client', self.api_client_mock()) + writer = arvados.CollectionWriter(**kwargs) writer.start_new_file('foo') - writer.write('foo') + writer.write(b'foo') return writer def test_write_whole_collection(self): @@ -723,6 +704,32 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin): with self.assertRaises(arvados.errors.KeepWriteError): writer.finish() + def test_write_insufficient_replicas_via_proxy(self): + writer = self.foo_writer(replication=3) + with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}): + with self.assertRaises(arvados.errors.KeepWriteError): + writer.manifest_text() + + def test_write_insufficient_replicas_via_disks(self): + client = mock.MagicMock(name='api_client') + with self.mock_keep( + None, 200, 200, + **{'x-keep-replicas-stored': 1}) as keepmock: + self.mock_keep_services(client, status=200, service_type='disk', count=2) + writer = self.foo_writer(api_client=client, replication=3) + with self.assertRaises(arvados.errors.KeepWriteError): + writer.manifest_text() + + def test_write_three_replicas(self): + client = mock.MagicMock(name='api_client') + with self.mock_keep( + "", 500, 500, 500, 200, 200, 200, + **{'x-keep-replicas-stored': 1}) as keepmock: + self.mock_keep_services(client, status=200, service_type='disk', count=6) + writer = self.foo_writer(api_client=client, replication=3) + writer.manifest_text() + self.assertEqual(6, keepmock.call_count) + def test_write_whole_collection_through_retries(self): writer = self.foo_writer(num_retries=2) with self.mock_keep(self.DEFAULT_DATA_HASH, @@ -742,8 +749,8 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin): with writer.open('out') as out_file: self.assertEqual('.', writer.current_stream_name()) self.assertEqual('out', writer.current_file_name()) - out_file.write('test data') - data_loc = hashlib.md5('test data').hexdigest() + '+9' + out_file.write(b'test data') + data_loc = tutil.str_keep_locator('test data') self.assertTrue(out_file.closed, "writer file not closed after context") self.assertRaises(ValueError, out_file.write, 'extra text') with self.mock_keep(data_loc, 200) as keep_mock: @@ -755,21 +762,21 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin): writer = arvados.CollectionWriter(client) with writer.open('six') as out_file: out_file.writelines(['12', '34', '56']) - data_loc = hashlib.md5('123456').hexdigest() + '+6' + data_loc = tutil.str_keep_locator('123456') with self.mock_keep(data_loc, 200) as keep_mock: self.assertEqual(". {} 0:6:six\n".format(data_loc), writer.manifest_text()) def test_open_flush(self): client = self.api_client_mock() - data_loc1 = hashlib.md5('flush1').hexdigest() + '+6' - data_loc2 = hashlib.md5('flush2').hexdigest() + '+6' + data_loc1 = tutil.str_keep_locator('flush1') + data_loc2 = tutil.str_keep_locator('flush2') with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock: writer = arvados.CollectionWriter(client) with writer.open('flush_test') as out_file: - out_file.write('flush1') + out_file.write(b'flush1') out_file.flush() - out_file.write('flush2') + out_file.write(b'flush2') self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1, data_loc2), writer.manifest_text()) @@ -778,24 +785,24 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin): client = self.api_client_mock() writer = arvados.CollectionWriter(client) with writer.open('.', '1') as out_file: - out_file.write('1st') + out_file.write(b'1st') with writer.open('.', '2') as out_file: - out_file.write('2nd') - data_loc = hashlib.md5('1st2nd').hexdigest() + '+6' + out_file.write(b'2nd') + data_loc = tutil.str_keep_locator('1st2nd') with self.mock_keep(data_loc, 200) as keep_mock: self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc), writer.manifest_text()) def test_two_opens_two_streams(self): client = self.api_client_mock() - data_loc1 = hashlib.md5('file').hexdigest() + '+4' - data_loc2 = hashlib.md5('indir').hexdigest() + '+5' + data_loc1 = tutil.str_keep_locator('file') + data_loc2 = tutil.str_keep_locator('indir') with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock: writer = arvados.CollectionWriter(client) with writer.open('file') as out_file: - out_file.write('file') + out_file.write(b'file') with writer.open('./dir', 'indir') as out_file: - out_file.write('indir') + out_file.write(b'indir') expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format( data_loc1, data_loc2) self.assertEqual(expected, writer.manifest_text()) @@ -807,67 +814,603 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin): self.assertRaises(arvados.errors.AssertionError, writer.open, 'two') +class CollectionMethods(run_test_server.TestCaseWithServers): + + def test_keys_values_items_support_indexing(self): + c = Collection() + with c.open('foo', 'wb') as f: + f.write(b'foo') + with c.open('bar', 'wb') as f: + f.write(b'bar') + self.assertEqual(2, len(c.keys())) + if sys.version_info < (3, 0): + # keys() supports indexing only for python2 callers. + fn0 = c.keys()[0] + fn1 = c.keys()[1] + else: + fn0, fn1 = c.keys() + self.assertEqual(2, len(c.values())) + f0 = c.values()[0] + f1 = c.values()[1] + self.assertEqual(2, len(c.items())) + self.assertEqual(fn0, c.items()[0][0]) + self.assertEqual(fn1, c.items()[1][0]) + + +class CollectionOpenModes(run_test_server.TestCaseWithServers): + + def test_open_binary_modes(self): + c = Collection() + for mode in ['wb', 'wb+', 'ab', 'ab+']: + with c.open('foo', mode) as f: + f.write(b'foo') + + def test_open_invalid_modes(self): + c = Collection() + for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]: + with self.assertRaises(Exception): + c.open('foo', mode) + + def test_open_text_modes(self): + c = Collection() + with c.open('foo', 'wb') as f: + f.write('foo') + for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']: + if sys.version_info >= (3, 0): + with self.assertRaises(NotImplementedError): + c.open('foo', mode) + else: + with c.open('foo', mode) as f: + if mode[0] == 'r' and '+' not in mode: + self.assertEqual('foo', f.read(3)) + else: + f.write('bar') + f.seek(-3, os.SEEK_CUR) + self.assertEqual('bar', f.read(3)) + + class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin): - def test_import_manifest(self): + + def test_replication_desired_kept_on_load(self): + m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n' + c1 = Collection(m, replication_desired=1) + c1.save_new() + loc = c1.manifest_locator() + c2 = Collection(loc) + self.assertEqual(c1.manifest_text, c2.manifest_text) + self.assertEqual(c1.replication_desired, c2.replication_desired) + + def test_replication_desired_not_loaded_if_provided(self): + m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n' + c1 = Collection(m, replication_desired=1) + c1.save_new() + loc = c1.manifest_locator() + c2 = Collection(loc, replication_desired=2) + self.assertEqual(c1.manifest_text, c2.manifest_text) + self.assertNotEqual(c1.replication_desired, c2.replication_desired) + + def test_init_manifest(self): m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt """ - self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(arvados.import_manifest(m1))) + self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False)) + self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True)) - def test_init_manifest(self): + def test_init_manifest_with_collision(self): m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt -. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt -. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt +./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt """ - self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(arvados.Collection(m1))) + with self.assertRaises(arvados.errors.ArgumentError): + self.assertEqual(m1, CollectionReader(m1)) + def test_init_manifest_with_error(self): + m1 = """. 0:43:md5sum.txt""" + with self.assertRaises(arvados.errors.ArgumentError): + self.assertEqual(m1, CollectionReader(m1)) def test_remove(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n', sync=SYNC_EXPLICIT) as c: - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c)) - self.assertTrue("count1.txt" in c) - c.remove("count1.txt") - self.assertFalse("count1.txt" in c) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c)) + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n') + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text()) + self.assertIn("count1.txt", c) + c.remove("count1.txt") + self.assertNotIn("count1.txt", c) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text()) + with self.assertRaises(arvados.errors.ArgumentError): + c.remove("") + + def test_find(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n') + self.assertIs(c.find("."), c) + self.assertIs(c.find("./count1.txt"), c["count1.txt"]) + self.assertIs(c.find("count1.txt"), c["count1.txt"]) + with self.assertRaises(IOError): + c.find("/.") + with self.assertRaises(arvados.errors.ArgumentError): + c.find("") + self.assertIs(c.find("./nonexistant.txt"), None) + self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None) def test_remove_in_subdir(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT) as c: - c.remove("foo/count2.txt") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + c.remove("foo/count2.txt") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text()) def test_remove_empty_subdir(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT) as c: - c.remove("foo/count2.txt") - c.remove("foo") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + c.remove("foo/count2.txt") + c.remove("foo") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text()) def test_remove_nonempty_subdir(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT) as c: - with self.assertRaises(IOError): - c.remove("foo") - c.remove("foo", rm_r=True) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) - - def test_copy_to_dir1(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT) as c: - c.copy("count1.txt", "foo/count2.txt") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c)) - - def test_copy_to_dir2(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT) as c: - c.copy("count1.txt", "foo") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) - - def test_copy_to_dir2(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT) as c: - c.copy("count1.txt", "foo/") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + with self.assertRaises(IOError): + c.remove("foo") + c.remove("foo", recursive=True) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text()) + + def test_copy_to_file_in_dir(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c.copy("count1.txt", "foo/count2.txt") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text()) def test_copy_file(self): - with arvados.import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT) as c: - c.copy("count1.txt", "count2.txt") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c)) + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c.copy("count1.txt", "count2.txt") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text()) + + def test_copy_to_existing_dir(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + c.copy("count1.txt", "foo") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text()) + + def test_copy_to_new_dir(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c.copy("count1.txt", "foo/") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text()) + + def test_rename_file(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c.rename("count1.txt", "count2.txt") + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text()) + + def test_move_file_to_dir(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c.mkdirs("foo") + c.rename("count1.txt", "foo/count2.txt") + self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text()) + + def test_move_file_to_other(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection() + c2.rename("count1.txt", "count2.txt", source_collection=c1) + self.assertEqual("", c1.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text()) + + def test_clone(self): + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + cl = c.clone() + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text()) + + def test_diff_del_add(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n') + d = c2.diff(c1) + self.assertEqual(sorted(d), [ + ('add', './count1.txt', c1["count1.txt"]), + ('del', './count2.txt', c2["count2.txt"]), + ]) + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './count2.txt', c2["count2.txt"]), + ('del', './count1.txt', c1["count1.txt"]), + ]) + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_same(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + d = c2.diff(c1) + self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])]) + d = c1.diff(c2) + self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])]) + + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_mod(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n') + d = c2.diff(c1) + self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])]) + d = c1.diff(c2) + self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])]) + + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_add(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n') + d = c2.diff(c1) + self.assertEqual(sorted(d), [ + ('del', './count2.txt', c2["count2.txt"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './count2.txt', c2["count2.txt"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_add_in_subcollection(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n') + d = c2.diff(c1) + self.assertEqual(sorted(d), [ + ('del', './foo', c2["foo"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './foo', c2["foo"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_del_add_in_subcollection(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n') + c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n') + d = c2.diff(c1) + self.assertEqual(sorted(d), [ + ('add', './foo/count2.txt', c1.find("foo/count2.txt")), + ('del', './foo/count3.txt', c2.find("foo/count3.txt")), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './foo/count3.txt', c2.find("foo/count3.txt")), + ('del', './foo/count2.txt', c1.find("foo/count2.txt")), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_diff_mod_in_subcollection(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n') + c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n') + d = c2.diff(c1) + self.assertEqual(sorted(d), [ + ('mod', './foo', c2["foo"], c1["foo"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('mod', './foo', c1["foo"], c2["foo"]), + ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]), + ]) + + self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text()) + + def test_conflict_keep_local_change(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n') + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './count2.txt', c2["count2.txt"]), + ('del', './count1.txt', c1["count1.txt"]), + ]) + f = c1.open("count1.txt", "wb") + f.write(b"zzzzz") + + # c1 changed, so it should not be deleted. + c1.apply(d) + self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n") + + def test_conflict_mod(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt') + d = c1.diff(c2) + self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])]) + f = c1.open("count1.txt", "wb") + f.write(b"zzzzz") + + # c1 changed, so c2 mod will go to a conflict file + c1.apply(d) + self.assertRegex( + c1.portable_manifest_text(), + r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$") + + def test_conflict_add(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n') + d = c1.diff(c2) + self.assertEqual(sorted(d), [ + ('add', './count1.txt', c2["count1.txt"]), + ('del', './count2.txt', c1["count2.txt"]), + ]) + f = c1.open("count1.txt", "wb") + f.write(b"zzzzz") + + # c1 added count1.txt, so c2 add will go to a conflict file + c1.apply(d) + self.assertRegex( + c1.portable_manifest_text(), + r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$") + + def test_conflict_del(self): + c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') + c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt') + d = c1.diff(c2) + self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])]) + c1.remove("count1.txt") + + # c1 deleted, so c2 mod will go to a conflict file + c1.apply(d) + self.assertRegex( + c1.portable_manifest_text(), + r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$") + + def test_notify(self): + c1 = Collection() + events = [] + c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item))) + f = c1.open("foo.txt", "wb") + self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile)) + + def test_open_w(self): + c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n") + self.assertEqual(c1["count1.txt"].size(), 10) + c1.open("count1.txt", "wb").close() + self.assertEqual(c1["count1.txt"].size(), 0) + + +class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers): + MAIN_SERVER = {} + KEEP_SERVER = {} + + def setUp(self): + self.keep_put = getattr(arvados.keep.KeepClient, 'put') + + def test_repacked_block_submission_get_permission_token(self): + ''' + Make sure that those blocks that are committed after repacking small ones, + get their permission tokens assigned on the collection manifest. + ''' + def wrapped_keep_put(*args, **kwargs): + # Simulate slow put operations + time.sleep(1) + return self.keep_put(*args, **kwargs) + + re_locator = "[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}" + + with mock.patch('arvados.keep.KeepClient.put', autospec=True) as mocked_put: + mocked_put.side_effect = wrapped_keep_put + c = Collection() + # Write 70 files ~1MiB each so we force to produce 1 big block by repacking + # small ones before finishing the upload. + for i in range(70): + f = c.open("file_{}.txt".format(i), 'wb') + f.write(random.choice('abcdefghijklmnopqrstuvwxyz') * (2**20+i)) + f.close(flush=False) + # We should get 2 blocks with their tokens + self.assertEqual(len(re.findall(re_locator, c.manifest_text())), 2) + + +class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers): + def test_get_manifest_text_only_committed(self): + c = Collection() + with c.open("count.txt", "wb") as f: + # One file committed + with c.open("foo.txt", "wb") as foo: + foo.write(b"foo") + foo.flush() # Force block commit + f.write(b"0123456789") + # Other file not committed. Block not written to keep yet. + self.assertEqual( + c._get_manifest_text(".", + strip=False, + normalize=False, + only_committed=True), + '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n') + # And now with the file closed... + f.flush() # Force block commit + self.assertEqual( + c._get_manifest_text(".", + strip=False, + normalize=False, + only_committed=True), + ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n") + + def test_only_small_blocks_are_packed_together(self): + c = Collection() + # Write a couple of small files, + f = c.open("count.txt", "wb") + f.write(b"0123456789") + f.close(flush=False) + foo = c.open("foo.txt", "wb") + foo.write(b"foo") + foo.close(flush=False) + # Then, write a big file, it shouldn't be packed with the ones above + big = c.open("bigfile.txt", "wb") + big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2 + big.close(flush=False) + self.assertEqual( + c.manifest_text("."), + '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n') + + def test_flush_after_small_block_packing(self): + c = Collection() + # Write a couple of small files, + f = c.open("count.txt", "wb") + f.write(b"0123456789") + f.close(flush=False) + foo = c.open("foo.txt", "wb") + foo.write(b"foo") + foo.close(flush=False) + + self.assertEqual( + c.manifest_text(), + '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n') + + f = c.open("count.txt", "rb+") + f.close(flush=True) + + self.assertEqual( + c.manifest_text(), + '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n') + + def test_write_after_small_block_packing2(self): + c = Collection() + # Write a couple of small files, + f = c.open("count.txt", "wb") + f.write(b"0123456789") + f.close(flush=False) + foo = c.open("foo.txt", "wb") + foo.write(b"foo") + foo.close(flush=False) + + self.assertEqual( + c.manifest_text(), + '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n') + + f = c.open("count.txt", "rb+") + f.write(b"abc") + f.close(flush=False) + + self.assertEqual( + c.manifest_text(), + '. 900150983cd24fb0d6963f7d28e17f72+3 a8430a058b8fbf408e1931b794dbd6fb+13 0:3:count.txt 6:7:count.txt 13:3:foo.txt\n') + + + def test_small_block_packing_with_overwrite(self): + c = Collection() + c.open("b1", "wb").close() + c["b1"].writeto(0, b"b1", 0) + + c.open("b2", "wb").close() + c["b2"].writeto(0, b"b2", 0) + + c["b1"].writeto(0, b"1b", 0) + + self.assertEquals(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n") + self.assertEquals(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n") + self.assertEquals(c["b2"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 2:2:b2\n") + + +class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers): + MAIN_SERVER = {} + KEEP_SERVER = {} + + def create_count_txt(self): + # Create an empty collection, save it to the API server, then write a + # file, but don't save it. + + c = Collection() + c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True) + self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0") + self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" ) + + with c.open("count.txt", "wb") as f: + f.write(b"0123456789") + + self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n") + + return c + + def test_create_and_save(self): + c = self.create_count_txt() + c.save() + self.assertRegex( + c.manifest_text(), + r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",) + + def test_create_and_save_new(self): + c = self.create_count_txt() + c.save_new() + self.assertRegex( + c.manifest_text(), + r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",) + + def test_create_diff_apply(self): + c1 = self.create_count_txt() + c1.save() + + c2 = Collection(c1.manifest_locator()) + with c2.open("count.txt", "wb") as f: + f.write(b"abcdefg") + + diff = c1.diff(c2) + + self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"])) + + c1.apply(diff) + self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash()) + + def test_diff_apply_with_token(self): + baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n") + c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n") + other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n") + + diff = baseline.diff(other) + self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])]) + + c.apply(diff) + + self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n") + + + def test_create_and_update(self): + c1 = self.create_count_txt() + c1.save() + + c2 = arvados.collection.Collection(c1.manifest_locator()) + with c2.open("count.txt", "wb") as f: + f.write(b"abcdefg") + + c2.save() + + self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash()) + c1.update() + self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash()) + + + def test_create_and_update_with_conflict(self): + c1 = self.create_count_txt() + c1.save() + + with c1.open("count.txt", "wb") as f: + f.write(b"XYZ") + + c2 = arvados.collection.Collection(c1.manifest_locator()) + with c2.open("count.txt", "wb") as f: + f.write(b"abcdefg") + + c2.save() + + c1.update() + self.assertRegex( + c1.manifest_text(), + r"\. e65075d550f9b5bf9992fa1d71a131be\+3\S* 7ac66c0f148de9519b8bd264312c4d64\+7\S* 0:3:count\.txt 3:7:count\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$") + + def test_pdh_is_native_str(self): + c1 = self.create_count_txt() + pdh = c1.portable_data_hash() + self.assertEqual(type(''), type(pdh)) if __name__ == '__main__':