+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])
+
+ def test_get_properties(self):
+ c = Collection()
+ self.assertEqual(c.get_properties(), {})
+ c.save_new(properties={"foo":"bar"})
+ self.assertEqual(c.get_properties(), {"foo":"bar"})
+
+ def test_get_trash_at(self):
+ c = Collection()
+ self.assertEqual(c.get_trash_at(), None)
+ c.save_new(trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
+ self.assertEqual(c.get_trash_at(), ciso8601.parse_datetime('2111-01-01T11:11:11.111111000Z'))
+
+
+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']:
+ 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):
+
+ 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(strip=True), c2.manifest_text(strip=True))
+ 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(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
+. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
+"""
+ 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_with_collision(self):
+ m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
+./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
+"""
+ 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):
+ 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_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("./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_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./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')
+ 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):
+ 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):
+ 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 = {}
+ local_locator_re = r"[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
+ remote_locator_re = r"[0-9a-f]{32}\+\d+\+R[a-z]{5}-[a-f0-9]{40}@[a-f0-9]{8}"
+
+ 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):
+ '''
+ 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)
+
+ 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(self.local_locator_re, c.manifest_text())), 2)
+
+ @mock.patch('arvados.keep.KeepClient.refresh_signature')
+ def test_copy_remote_blocks_on_save_new(self, rs_mock):
+ remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
+ local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
+ rs_mock.return_value = local_block_loc
+ c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, c.manifest_text())), 1)
+ self.assertEqual(
+ len(re.findall(self.local_locator_re, c.manifest_text())), 0)
+ c.save_new()
+ rs_mock.assert_called()
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, c.manifest_text())), 0)
+ self.assertEqual(
+ len(re.findall(self.local_locator_re, c.manifest_text())), 1)
+
+ @mock.patch('arvados.keep.KeepClient.refresh_signature')
+ def test_copy_remote_blocks_on_save(self, rs_mock):
+ remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
+ local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
+ rs_mock.return_value = local_block_loc
+ # Remote collection
+ remote_c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, remote_c.manifest_text())), 1)
+ # Local collection
+ local_c = Collection()
+ with local_c.open('barfile.txt', 'wb') as f:
+ f.write('bar')
+ local_c.save_new()
+ self.assertEqual(
+ len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
+ # Copy remote file to local collection
+ local_c.copy('./foofile.txt', './copied/foofile.txt', remote_c)
+ self.assertEqual(
+ len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, local_c.manifest_text())), 1)
+ # Save local collection: remote block should be copied
+ local_c.save()
+ rs_mock.assert_called()
+ self.assertEqual(
+ len(re.findall(self.local_locator_re, local_c.manifest_text())), 2)
+ self.assertEqual(
+ len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
+
+
+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:
+ # 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.assertEqual(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n")
+ self.assertEqual(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n")
+ self.assertEqual(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(properties={'type' : 'Intermediate'},
+ storage_classes=['archive'],
+ trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
+
+ self.assertRegex(
+ c.manifest_text(),
+ r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+ self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
+ self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
+ self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
+
+
+ def test_create_and_save_new(self):
+ c = self.create_count_txt()
+ c.save_new(properties={'type' : 'Intermediate'},
+ storage_classes=['archive'],
+ trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
+
+ self.assertRegex(
+ c.manifest_text(),
+ r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+ self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
+ self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
+ self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
+
+ def test_create_and_save_after_commiting(self):
+ c = self.create_count_txt()
+ c.save(properties={'type' : 'Intermediate'},
+ storage_classes=['hot'],
+ trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
+ c.save(properties={'type' : 'Output'},
+ storage_classes=['cold'],
+ trash_at=datetime.datetime(2222, 2, 2, 22, 22, 22, 222222))
+
+ self.assertEqual(c.api_response()["storage_classes_desired"], ['cold'])
+ self.assertEqual(c.api_response()["properties"], {'type' : 'Output'})
+ self.assertEqual(c.api_response()["trash_at"], '2222-02-02T22:22:22.222222000Z')
+
+ 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))
+
+