3198: Implement rename() (efficient move within/between collections).
[arvados.git] / sdk / python / tests / test_collections.py
index 2af36227e9d7e6e043041e2701d095d52b8614a2..95b6dbe6d1943ad0b348d35bb5abf29eebf79985 100644 (file)
@@ -13,10 +13,9 @@ import tempfile
 import unittest
 
 import run_test_server
+from arvados._ranges import Range, LocatorAndRange
+from arvados.collection import Collection, CollectionReader
 import arvados_testutil as tutil
-from arvados.ranges import Range, LocatorAndRange
-from arvados import import_manifest, export_manifest
-from arvados.arvfile import SYNC_EXPLICIT
 
 class TestResumableWriter(arvados.ResumableCollectionWriter):
     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
@@ -434,7 +433,7 @@ 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):
@@ -546,9 +545,8 @@ 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)
@@ -571,11 +569,10 @@ 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 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
@@ -642,7 +639,7 @@ 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):
@@ -651,12 +648,6 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
         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')
-        self.check_open_file(cfile, '.', 'foo', 3)
-
     def test_open_deep_file(self):
         coll_name = 'collection_with_files_in_subdir'
         client = self.api_client_mock(200)
@@ -670,12 +661,12 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
     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
@@ -710,23 +701,26 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 
     def test_write_insufficient_replicas_via_disks(self):
         client = mock.MagicMock(name='api_client')
-        self.mock_keep_services(client, status=200, service_type='disk', count=2)
-        writer = self.foo_writer(api_client=client, replication=3)
         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')
-        self.mock_keep_services(client, status=200, service_type='disk', count=6)
-        writer = self.foo_writer(api_client=client, replication=3)
         with self.mock_keep(
-                None, 200, 500, 200, 500, 200, 200,
+                None, 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(5, keepmock.call_count)
+            # keepmock is the mock session constructor; keepmock.return_value
+            # is the mock session object, and keepmock.return_value.put is the
+            # actual mock method of interest.
+            self.assertEqual(6, keepmock.return_value.put.call_count)
 
     def test_write_whole_collection_through_retries(self):
         writer = self.foo_writer(num_retries=2)
@@ -813,75 +807,111 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 
 
 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
-    def test_import_manifest(self):
+
+    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.ReadOnlyCollection(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.ReadOnlyCollection(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):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
-        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 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.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", export_manifest(c))
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.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("")
 
     def test_remove_in_subdir(self):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+        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", export_manifest(c))
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 
     def test_remove_empty_subdir(self):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+        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", export_manifest(c))
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 
     def test_remove_nonempty_subdir(self):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+        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", rm_r=True)
-        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
+        c.remove("foo", recursive=True)
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 
-    def test_copy_to_dir1(self):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+    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", export_manifest(c))
-
-    def test_copy_to_dir2(self):
-        c = arvados.WritableCollection('. 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", export_manifest(c))
-
-    def test_copy_to_dir2(self):
-        c = arvados.WritableCollection('. 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", export_manifest(c))
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 
     def test_copy_file(self):
-        c = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+        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.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.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.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 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+        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", export_manifest(cl))
+        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.manifest_text())
 
-    def test_diff1(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+    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(d, [('del', './count2.txt', c2["count2.txt"]),
                              ('add', './count1.txt', c1["count1.txt"])])
@@ -892,9 +922,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff2(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+    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, [])
         d = c1.diff(c2)
@@ -904,9 +934,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff3(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
+    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)
@@ -916,9 +946,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff4(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
+    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(d, [('del', './count2.txt', c2["count2.txt"])])
         d = c1.diff(c2)
@@ -928,9 +958,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff5(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+    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(d, [('del', './foo', c2["foo"])])
         d = c1.diff(c2)
@@ -940,9 +970,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff6(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
-        c2 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
+    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(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
@@ -955,9 +985,9 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_diff7(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
-        c2 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
+    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(d, [('mod', './foo', c2["foo"], c1["foo"])])
         d = c1.diff(c2)
@@ -967,63 +997,159 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 
-    def test_conflict1(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+    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(d, [('del', './count1.txt', c1["count1.txt"]),
                              ('add', './count2.txt', c2["count2.txt"])])
-        with c1.open("count1.txt", "w") as f:
-            f.write("zzzzz")
+        f = c1.open("count1.txt", "w")
+        f.write("zzzzz")
 
         # c1 changed, so it should not be deleted.
         c1.apply(d)
         self.assertEqual(c1.manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
 
-    def test_conflict2(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
+    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"])])
-        with c1.open("count1.txt", "w") as f:
-            f.write("zzzzz")
+        f = c1.open("count1.txt", "w")
+        f.write("zzzzz")
 
         # c1 changed, so c2 mod will go to a conflict file
         c1.apply(d)
-        self.assertTrue(re.match(r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1.txt 5:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
-                                 c1.manifest_text()))
+        self.assertRegexpMatches(c1.manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$")
 
-    def test_conflict3(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
+    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(d, [('del', './count2.txt', c1["count2.txt"]),
                              ('add', './count1.txt', c2["count1.txt"])])
-        with c1.open("count1.txt", "w") as f:
-            f.write("zzzzz")
+        f = c1.open("count1.txt", "w")
+        f.write("zzzzz")
 
         # c1 added count1.txt, so c2 add will go to a conflict file
         c1.apply(d)
-        self.assertTrue(re.match(r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1.txt 5:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
-                                 c1.manifest_text()))
+        self.assertRegexpMatches(c1.manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$")
 
-    def test_conflict4(self):
-        c1 = arvados.WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
-        c2 = arvados.WritableCollection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
+    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.assertTrue(re.match(r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
-                                 c1.manifest_text()))
+        self.assertRegexpMatches(c1.manifest_text(), r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$")
 
-    def test_notify1(self):
-        c1 = arvados.WritableCollection(sync=SYNC_EXPLICIT)
+    def test_notify(self):
+        c1 = Collection()
         events = []
         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
-        c1.find("")
+        f = c1.open("foo.txt", "w")
+        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", "w").close()
+        self.assertEqual(c1["count1.txt"].size(), 0)
+
+
+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", "w") as f:
+            f.write("0123456789")
+
+        self.assertEqual(c.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.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+
+
+    def test_create_and_save_new(self):
+        c = self.create_count_txt()
+        c.save_new()
+        self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+
+    def test_create_diff_apply(self):
+        c1 = self.create_count_txt()
+        c1.save()
+
+        c2 = Collection(c1.manifest_locator())
+        with c2.open("count.txt", "w") as f:
+            f.write("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", "w") as f:
+            f.write("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", "w") as f:
+            f.write("XYZ")
+
+        c2 = arvados.collection.Collection(c1.manifest_locator())
+        with c2.open("count.txt", "w") as f:
+            f.write("abcdefg")
+
+        c2.save()
+
+        c1.update()
+        self.assertRegexpMatches(c1.manifest_text(), r"\. e65075d550f9b5bf9992fa1d71a131be\+3 7ac66c0f148de9519b8bd264312c4d64\+7\+A[a-f0-9]{40}@[a-f0-9]{8} 0:3:count\.txt 3:7:count\.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$")
+
 
 if __name__ == '__main__':
     unittest.main()