Merge branch '11220-manifest-fetch-error'
[arvados.git] / sdk / python / tests / test_collections.py
index 86215f535a21634b18a0167636c79431657abd1f..49c00191bebe02cc8e267b397212a893a33f246a 100644 (file)
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 from __future__ import absolute_import
 
 from builtins import object
@@ -6,9 +10,11 @@ import copy
 import mock
 import os
 import pprint
+import random
 import re
 import sys
 import tempfile
+import time
 import unittest
 
 from . import run_test_server
@@ -30,7 +36,8 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
     @classmethod
     def setUpClass(cls):
         super(ArvadosCollectionsTest, cls).setUpClass()
-        run_test_server.authorize_with('active')
+        # need admin privileges to make collections with unsigned blocks
+        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)
@@ -52,9 +59,13 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
                          "wrong manifest: got {}".format(cw.manifest_text()))
-        cw.finish()
+        cw.save_new()
         return cw.portable_data_hash()
 
+    def test_pdh_is_native_str(self):
+        pdh = self.write_foo_bar_baz()
+        self.assertEqual(type(''), type(pdh))
+
     def test_keep_local_store(self):
         self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get')
@@ -87,6 +98,8 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
         self.assertEqual(stream0.readfrom(2**26, 0),
                          b'',
                          'reading zero bytes should have returned empty string')
+        self.assertEqual(3, len(cr))
+        self.assertTrue(cr)
 
     def _test_subset(self, collection, expected):
         cr = arvados.CollectionReader(collection, self.api_client)
@@ -493,15 +506,6 @@ class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
         self.assertRaises(arvados.errors.AssertionError,
                           cwriter.write, "badtext")
 
-    def test_read_arbitrary_data_with_collection_reader(self):
-        # arv-get relies on this to do "arv-get {keep-locator} -".
-        self.write_foo_bar_baz()
-        self.assertEqual(
-            'foobar',
-            arvados.CollectionReader(
-                '3858f62230ac3c915f300c664312c63f+6'
-                ).manifest_text())
-
 
 class CollectionTestMixin(tutil.ApiClientMock):
     API_COLLECTIONS = run_test_server.fixture('collections')
@@ -557,34 +561,13 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
                                               api_client=client)
             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 
-    def test_locator_init_fallback_to_keep(self):
-        # crunch-job needs this to read manifests that have only ever
-        # been written to Keep.
-        client = self.api_client_mock(200)
-        self.mock_get_collection(client, 404, None)
-        with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
-            reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
-                                              api_client=client)
-            self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
-
-    def test_uuid_init_no_fallback_to_keep(self):
-        # Do not look up a collection UUID in Keep.
-        client = self.api_client_mock(404)
-        with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
-            with self.assertRaises(arvados.errors.ApiError):
-                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
-        # mock API server to return the wrong data.
-        client = self.api_client_mock(200)
-        with tutil.mock_keep_responses(self.ALT_MANIFEST, 200):
-            self.assertEqual(
-                self.ALT_MANIFEST,
-                arvados.CollectionReader(
-                    self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
-                    api_client=client).manifest_text())
+    def test_init_no_fallback_to_keep(self):
+        # Do not look up a collection UUID or PDH in Keep.
+        for key in [self.DEFAULT_UUID, self.DEFAULT_DATA_HASH]:
+            client = self.api_client_mock(404)
+            with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
+                with self.assertRaises(arvados.errors.ApiError):
+                    reader = arvados.CollectionReader(key, api_client=client)
 
     def test_init_num_retries_propagated(self):
         # More of an integration test...
@@ -622,21 +605,14 @@ class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
                                           api_client=client)
         self.assertEqual('', reader.manifest_text())
+        self.assertEqual(0, len(reader))
+        self.assertFalse(reader)
 
     def test_api_response(self):
         client = self.api_client_mock()
         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 
-    def test_api_response_with_collection_from_keep(self):
-        client = self.api_client_mock()
-        self.mock_get_collection(client, 404, 'foo')
-        with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
-            reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
-                                              api_client=client)
-            api_response = reader.api_response()
-        self.assertIsNone(api_response)
-
     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())
@@ -804,12 +780,35 @@ class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 
 
+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])
+
+
 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', 'wb') as f:
+            with c.open('foo', mode) as f:
                 f.write(b'foo')
 
     def test_open_invalid_modes(self):
@@ -1146,6 +1145,38 @@ class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
         self.assertEqual(c1["count1.txt"].size(), 0)
 
 
+class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers):
+    MAIN_SERVER = {}
+    KEEP_SERVER = {}
+
+    def setUp(self):
+        self.keep_put = getattr(arvados.keep.KeepClient, 'put')
+
+    def test_repacked_block_submission_get_permission_token(self):
+        '''
+        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)
+
+        re_locator = "[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
+
+        with mock.patch('arvados.keep.KeepClient.put', autospec=True) as mocked_put:
+            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(re_locator, c.manifest_text())), 2)
+
+
 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
     def test_get_manifest_text_only_committed(self):
         c = Collection()
@@ -1173,7 +1204,7 @@ class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
 
     def test_only_small_blocks_are_packed_together(self):
         c = Collection()
-        # Write a couple of small files, 
+        # Write a couple of small files,
         f = c.open("count.txt", "wb")
         f.write(b"0123456789")
         f.close(flush=False)
@@ -1188,6 +1219,64 @@ class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
             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 = {}
@@ -1284,6 +1373,11 @@ class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
             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))
+
 
 if __name__ == '__main__':
     unittest.main()