import arvados
import copy
-import hashlib
import mock
import os
import pprint
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)
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'
+ 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:
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('1st')
with writer.open('.', '2') as out_file:
out_file.write('2nd')
- data_loc = hashlib.md5('1st2nd').hexdigest() + '+6'
+ 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:
class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
+ 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
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):
c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
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, [])
+ self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
d = c1.diff(c2)
- self.assertEqual(d, [])
+ 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)
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(d, [('del', './count2.txt', c2["count2.txt"])])
+ self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
d = c1.diff(c2)
- self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"])])
+ self.assertEqual(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)
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(d, [('del', './foo', c2["foo"])])
+ self.assertEqual(d, [('del', './foo', c2["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
d = c1.diff(c2)
- self.assertEqual(d, [('add', './foo', c2["foo"])])
+ self.assertEqual(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)
d = c2.diff(c1)
self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
- ('add', './foo/count2.txt', c1.find("foo/count2.txt"))])
+ ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
d = c1.diff(c2)
self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
- ('add', './foo/count3.txt', c2.find("foo/count3.txt"))])
+ ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
c1.apply(d)
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(d, [('mod', './foo', c2["foo"], c1["foo"])])
+ self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
d = c1.diff(c2)
- self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"])])
+ self.assertEqual(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["count1.txt"].size(), 0)
+class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
+ def test_get_manifest_text_only_committed(self):
+ c = Collection()
+ with c.open("count.txt", "w") as f:
+ # One file committed
+ with c.open("foo.txt", "w") as foo:
+ foo.write("foo")
+ foo.flush() # Force block commit
+ f.write("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", "w")
+ f.write("0123456789")
+ f.close(flush=False)
+ foo = c.open("foo.txt", "w")
+ foo.write("foo")
+ foo.close(flush=False)
+ # Then, write a big file, it shouldn't be packed with the ones above
+ big = c.open("bigfile.txt", "w")
+ big.write("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')
+
+
class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
MAIN_SERVER = {}
KEEP_SERVER = {}
projects / arvados.git / blobdiff