10383: Merge branch 'master' into 10383-arv-put-incremental-upload
[arvados.git] / sdk / python / tests / test_arv_put.py
index f35e4c725c1ebfb74062b44b4c4c4d5477684f9a..bc933e27f8de55d7611865bf10d92993a54937f1 100644 (file)
@@ -32,9 +32,7 @@ class ArvadosPutResumeCacheTest(ArvadosBaseTestCase):
         [],
         ['/dev/null'],
         ['/dev/null', '--filename', 'empty'],
-        ['/tmp'],
-        ['/tmp', '--max-manifest-depth', '0'],
-        ['/tmp', '--max-manifest-depth', '1']
+        ['/tmp']
         ]
 
     def tearDown(self):
@@ -241,6 +239,7 @@ class ArvadosPutResumeCacheTest(ArvadosBaseTestCase):
 
 class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
                           ArvadosBaseTestCase):
+
     def setUp(self):
         super(ArvPutUploadJobTest, self).setUp()
         run_test_server.authorize_with('active')
@@ -271,7 +270,7 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
 
     def test_writer_works_without_cache(self):
         cwriter = arv_put.ArvPutUploadJob(['/dev/null'], resume=False)
-        cwriter.start()
+        cwriter.start(save_collection=False)
         self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:null\n", cwriter.manifest_text())
 
     def test_writer_works_with_cache(self):
@@ -279,13 +278,13 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
             f.write('foo')
             f.flush()
             cwriter = arv_put.ArvPutUploadJob([f.name])
-            cwriter.start()
-            self.assertEqual(3, cwriter.bytes_written)
+            cwriter.start(save_collection=False)
+            self.assertEqual(3, cwriter.bytes_written - cwriter.bytes_skipped)
             # Don't destroy the cache, and start another upload
             cwriter_new = arv_put.ArvPutUploadJob([f.name])
-            cwriter_new.start()
+            cwriter_new.start(save_collection=False)
             cwriter_new.destroy_cache()
-            self.assertEqual(0, cwriter_new.bytes_written)
+            self.assertEqual(0, cwriter_new.bytes_written - cwriter_new.bytes_skipped)
 
     def make_progress_tester(self):
         progression = []
@@ -301,13 +300,13 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
                 progression, reporter = self.make_progress_tester()
                 cwriter = arv_put.ArvPutUploadJob([f.name],
                     reporter=reporter, bytes_expected=expect_count)
-                cwriter.start()
+                cwriter.start(save_collection=False)
                 cwriter.destroy_cache()
                 self.assertIn((3, expect_count), progression)
 
     def test_writer_upload_directory(self):
         cwriter = arv_put.ArvPutUploadJob([self.tempdir])
-        cwriter.start()
+        cwriter.start(save_collection=False)
         cwriter.destroy_cache()
         self.assertEqual(1024*(1+2+3+4+5), cwriter.bytes_written)
 
@@ -325,14 +324,75 @@ class ArvPutUploadJobTest(run_test_server.TestCaseWithServers,
             writer = arv_put.ArvPutUploadJob([self.large_file_name],
                                              replication_desired=1)
             with self.assertRaises(SystemExit):
-                writer.start()
-                self.assertLess(writer.bytes_written,
-                                os.path.getsize(self.large_file_name))
+                writer.start(save_collection=False)
+            # Confirm that the file was partially uploaded
+            self.assertGreater(writer.bytes_written, 0)
+            self.assertLess(writer.bytes_written,
+                            os.path.getsize(self.large_file_name))
         # Retry the upload
         writer2 = arv_put.ArvPutUploadJob([self.large_file_name],
                                           replication_desired=1)
-        writer2.start()
-        self.assertEqual(writer.bytes_written + writer2.bytes_written,
+        writer2.start(save_collection=False)
+        self.assertEqual(writer.bytes_written + writer2.bytes_written - writer2.bytes_skipped,
+                         os.path.getsize(self.large_file_name))
+        writer2.destroy_cache()
+
+    def test_no_resume_when_asked(self):
+        def wrapped_write(*args, **kwargs):
+            data = args[1]
+            # Exit only on last block
+            if len(data) < arvados.config.KEEP_BLOCK_SIZE:
+                raise SystemExit("Simulated error")
+            return self.arvfile_write(*args, **kwargs)
+
+        with mock.patch('arvados.arvfile.ArvadosFileWriter.write',
+                        autospec=True) as mocked_write:
+            mocked_write.side_effect = wrapped_write
+            writer = arv_put.ArvPutUploadJob([self.large_file_name],
+                                             replication_desired=1)
+            with self.assertRaises(SystemExit):
+                writer.start(save_collection=False)
+            # Confirm that the file was partially uploaded
+            self.assertGreater(writer.bytes_written, 0)
+            self.assertLess(writer.bytes_written,
+                            os.path.getsize(self.large_file_name))
+        # Retry the upload, this time without resume
+        writer2 = arv_put.ArvPutUploadJob([self.large_file_name],
+                                          replication_desired=1,
+                                          resume=False)
+        writer2.start(save_collection=False)
+        self.assertEqual(writer2.bytes_skipped, 0)
+        self.assertEqual(writer2.bytes_written,
+                         os.path.getsize(self.large_file_name))
+        writer2.destroy_cache()
+
+    def test_no_resume_when_no_cache(self):
+        def wrapped_write(*args, **kwargs):
+            data = args[1]
+            # Exit only on last block
+            if len(data) < arvados.config.KEEP_BLOCK_SIZE:
+                raise SystemExit("Simulated error")
+            return self.arvfile_write(*args, **kwargs)
+
+        with mock.patch('arvados.arvfile.ArvadosFileWriter.write',
+                        autospec=True) as mocked_write:
+            mocked_write.side_effect = wrapped_write
+            writer = arv_put.ArvPutUploadJob([self.large_file_name],
+                                             replication_desired=1)
+            with self.assertRaises(SystemExit):
+                writer.start(save_collection=False)
+            # Confirm that the file was partially uploaded
+            self.assertGreater(writer.bytes_written, 0)
+            self.assertLess(writer.bytes_written,
+                            os.path.getsize(self.large_file_name))
+        # Retry the upload, this time without cache usage
+        writer2 = arv_put.ArvPutUploadJob([self.large_file_name],
+                                          replication_desired=1,
+                                          resume=False,
+                                          use_cache=False)
+        writer2.start(save_collection=False)
+        self.assertEqual(writer2.bytes_skipped, 0)
+        self.assertEqual(writer2.bytes_written,
                          os.path.getsize(self.large_file_name))
         writer2.destroy_cache()
 
@@ -634,6 +694,21 @@ class ArvPutIntegrationTest(run_test_server.TestCaseWithServers,
         self.assertEqual(1, len(collection_list))
         return collection_list[0]
 
+    def test_put_collection_with_later_update(self):
+        tmpdir = self.make_tmpdir()
+        with open(os.path.join(tmpdir, 'file1'), 'w') as f:
+            f.write('Relaxing in basins at the end of inlets terminates the endless tests from the box')
+        col = self.run_and_find_collection("", ['--no-progress', tmpdir])
+        self.assertNotEqual(None, col['uuid'])
+        # Add a new file to the directory
+        with open(os.path.join(tmpdir, 'file2'), 'w') as f:
+            f.write('The quick brown fox jumped over the lazy dog')
+        updated_col = self.run_and_find_collection("", ['--no-progress', '--update-collection', col['uuid'], tmpdir])
+        self.assertEqual(col['uuid'], updated_col['uuid'])
+        # Get the manifest and check that the new file is being included
+        c = arv_put.api_client.collections().get(uuid=updated_col['uuid']).execute()
+        self.assertRegexpMatches(c['manifest_text'], r'^\. .*:44:file2\n')
+
     def test_put_collection_with_high_redundancy(self):
         # Write empty data: we're not testing CollectionWriter, just
         # making sure collections.create tells the API server what our