import copy
import mock
import os
-import pprint
import random
import re
import sys
-import tempfile
import datetime
import ciso8601
import time
import unittest
+import parameterized
from . import run_test_server
from arvados._ranges import Range, LocatorAndRange
from arvados.collection import Collection, CollectionReader
from . import arvados_testutil as tutil
+from .arvados_testutil import make_block_cache
class TestResumableWriter(arvados.ResumableCollectionWriter):
KEEP_BLOCK_SIZE = 1024 # PUT to Keep every 1K.
def current_state(self):
return self.dump_state(copy.deepcopy)
-
+@parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}])
class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
tutil.ArvadosBaseTestCase):
+ disk_cache = False
MAIN_SERVER = {}
@classmethod
run_test_server.authorize_with('admin')
cls.api_client = arvados.api('v1')
cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
- local_store=cls.local_store)
+ local_store=cls.local_store,
+ block_cache=make_block_cache(cls.disk_cache))
def write_foo_bar_baz(self):
cw = arvados.CollectionWriter(self.api_client)
class MockKeep(object):
def __init__(self, content, num_retries=0):
self.content = content
+ self.num_prefetch_threads = 1
- def get(self, locator, num_retries=0):
+ def get(self, locator, num_retries=0, prefetch=False):
return self.content[locator]
def test_stream_reader(self):
self.mock_get_collection(client, status, 'foo_file')
return client
- def test_init_no_default_retries(self):
+ def test_init_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)
+ client.collections().get().execute.assert_called_with(num_retries=10)
def test_uuid_init_success(self):
client = self.api_client_mock(200)
# Ensure stripped_manifest() doesn't mangle our manifest in
# any way other than stripping hints.
self.assertEqual(
- re.sub('\+[^\d\s\+]+', '', nonnormal),
+ re.sub(r'\+[^\d\s\+]+', '', nonnormal),
reader.stripped_manifest())
# Ensure stripped_manifest() didn't mutate our reader.
self.assertEqual(nonnormal, reader.manifest_text())
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))
+ 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(0, os.SEEK_SET)
+ self.assertEqual('bar', f.read(3))
+
+
+class TextModes(run_test_server.TestCaseWithServers):
+
+ def setUp(self):
+ arvados.config.KEEP_BLOCK_SIZE = 4
+ if sys.version_info < (3, 0):
+ import unicodedata
+ self.sailboat = unicodedata.lookup('SAILBOAT')
+ self.snowman = unicodedata.lookup('SNOWMAN')
+ else:
+ self.sailboat = '\N{SAILBOAT}'
+ self.snowman = '\N{SNOWMAN}'
+
+ def tearDown(self):
+ arvados.config.KEEP_BLOCK_SIZE = 2 ** 26
+
+ def test_read_sailboat_across_block_boundary(self):
+ c = Collection()
+ f = c.open('sailboats', 'wb')
+ data = self.sailboat.encode('utf-8')
+ f.write(data)
+ f.write(data[:1])
+ f.write(data[1:])
+ f.write(b'\n')
+ f.close()
+ self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+3 ')
+
+ f = c.open('sailboats', 'r')
+ string = f.readline()
+ self.assertEqual(string, self.sailboat+self.sailboat+'\n')
+ f.close()
+
+ def test_write_snowman_across_block_boundary(self):
+ c = Collection()
+ f = c.open('snowmany', 'w')
+ data = self.snowman
+ f.write(data+data+'\n'+data+'\n')
+ f.close()
+ self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+4 .*\+3 ')
+
+ f = c.open('snowmany', 'r')
+ self.assertEqual(f.readline(), self.snowman+self.snowman+'\n')
+ self.assertEqual(f.readline(), self.snowman+'\n')
+ f.close()
class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
c1.save_new()
loc = c1.manifest_locator()
c2 = Collection(loc)
- self.assertEqual(c1.manifest_text, c2.manifest_text)
+ self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
self.assertEqual(c1.replication_desired, c2.replication_desired)
def test_replication_desired_not_loaded_if_provided(self):
c1.save_new()
loc = c1.manifest_locator()
c2 = Collection(loc, replication_desired=2)
- self.assertEqual(c1.manifest_text, c2.manifest_text)
+ self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
self.assertNotEqual(c1.replication_desired, c2.replication_desired)
+ def test_storage_classes_desired_kept_on_load(self):
+ m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
+ c1 = Collection(m, storage_classes_desired=['archival'])
+ c1.save_new()
+ loc = c1.manifest_locator()
+ c2 = Collection(loc)
+ self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
+ self.assertEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
+
+ def test_storage_classes_change_after_save(self):
+ m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
+ c1 = Collection(m, storage_classes_desired=['archival'])
+ c1.save_new()
+ loc = c1.manifest_locator()
+ c2 = Collection(loc)
+ self.assertEqual(['archival'], c2.storage_classes_desired())
+ c2.save(storage_classes=['highIO'])
+ self.assertEqual(['highIO'], c2.storage_classes_desired())
+ c3 = Collection(loc)
+ self.assertEqual(c1.manifest_text(strip=True), c3.manifest_text(strip=True))
+ self.assertEqual(['highIO'], c3.storage_classes_desired())
+
+ def test_storage_classes_desired_not_loaded_if_provided(self):
+ m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
+ c1 = Collection(m, storage_classes_desired=['archival'])
+ c1.save_new()
+ loc = c1.manifest_locator()
+ c2 = Collection(loc, storage_classes_desired=['default'])
+ self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
+ self.assertNotEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
+
def test_init_manifest(self):
m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
with self.assertRaises(arvados.errors.ArgumentError):
c.remove("")
+ def test_remove_recursive(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:a/b/c/d/efg.txt 0:10:xyz.txt\n')
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:xyz.txt\n./a/b/c/d 781e5e245d69b566979b86e28d23f2c7+10 0:10:efg.txt\n", c.portable_manifest_text())
+ self.assertIn("a", c)
+ self.assertEqual(1, len(c["a"].keys()))
+ # cannot remove non-empty directory with default recursive=False
+ with self.assertRaises(OSError):
+ c.remove("a/b")
+ with self.assertRaises(OSError):
+ c.remove("a/b/c/d")
+ c.remove("a/b", recursive=True)
+ self.assertEqual(0, len(c["a"].keys()))
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:xyz.txt\n./a d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
+
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("./nonexistant.txt"), None)
self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
+ def test_escaped_paths_dont_get_unescaped_on_manifest(self):
+ # Dir & file names are literally '\056' (escaped form: \134056)
+ manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
+ c = Collection(manifest)
+ self.assertEqual(c.portable_manifest_text(), manifest)
+
+ def test_other_special_chars_on_file_token(self):
+ cases = [
+ ('\\000', '\0'),
+ ('\\011', '\t'),
+ ('\\012', '\n'),
+ ('\\072', ':'),
+ ('\\134400', '\\400'),
+ ]
+ for encoded, decoded in cases:
+ manifest = '. d41d8cd98f00b204e9800998ecf8427e+0 0:0:some%sfile.txt\n' % encoded
+ c = Collection(manifest)
+ self.assertEqual(c.portable_manifest_text(), manifest)
+ self.assertIn('some%sfile.txt' % decoded, c.keys())
+
+ def test_escaped_paths_do_get_unescaped_on_listing(self):
+ # Dir & file names are literally '\056' (escaped form: \134056)
+ manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
+ c = Collection(manifest)
+ self.assertIn('\\056 Test', c.keys())
+ self.assertIn('\\056', c['\\056 Test'].keys())
+
+ def test_make_empty_dir_with_escaped_chars(self):
+ c = Collection()
+ c.mkdirs('./Empty\\056Dir')
+ self.assertEqual(c.portable_manifest_text(),
+ './Empty\\134056Dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
+
+ def test_make_empty_dir_with_spaces(self):
+ c = Collection()
+ c.mkdirs('./foo bar/baz waz')
+ self.assertEqual(c.portable_manifest_text(),
+ './foo\\040bar/baz\\040waz d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
+
def test_remove_in_subdir(self):
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())
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
def test_remove_empty_subdir(self):
c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
def setUp(self):
self.keep_put = getattr(arvados.keep.KeepClient, 'put')
+ @mock.patch('arvados.keep.KeepClient.put', autospec=True)
+ def test_storage_classes_desired(self, put_mock):
+ put_mock.side_effect = self.keep_put
+ c = Collection(storage_classes_desired=['default'])
+ with c.open("file.txt", 'wb') as f:
+ f.write('content')
+ c.save_new()
+ _, kwargs = put_mock.call_args
+ self.assertEqual(['default'], kwargs['classes'])
+
@mock.patch('arvados.keep.KeepClient.put', autospec=True)
def test_repacked_block_submission_get_permission_token(self, mocked_put):
'''
class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
+ def test_preserve_version_on_save(self):
+ c = Collection()
+ c.save_new(preserve_version=True)
+ coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
+ self.assertEqual(coll_record['version'], 1)
+ self.assertEqual(coll_record['preserve_version'], True)
+ with c.open("foo.txt", "wb") as foo:
+ foo.write(b"foo")
+ c.save(preserve_version=True)
+ coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
+ self.assertEqual(coll_record['version'], 2)
+ self.assertEqual(coll_record['preserve_version'], True)
+ with c.open("bar.txt", "wb") as foo:
+ foo.write(b"bar")
+ c.save(preserve_version=False)
+ coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
+ self.assertEqual(coll_record['version'], 3)
+ self.assertEqual(coll_record['preserve_version'], False)
+
def test_get_manifest_text_only_committed(self):
c = Collection()
with c.open("count.txt", "wb") as f: