Merge branch '21666-provision-test-improvement'
[arvados.git] / sdk / python / tests / test_arvfile.py
index e0f6596741588cbcca783dbcca0774485ff9c5a9..6bcba9a81d6930480bbaca32d7a37cc73acbf031 100644 (file)
@@ -1,21 +1,18 @@
-from __future__ import absolute_import
-from builtins import hex
-from builtins import str
-from builtins import range
-from builtins import object
-import bz2
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import datetime
-import gzip
-import io
-import mock
 import os
-import unittest
 import time
+import unittest
+
+from unittest import mock
 
 import arvados
 from arvados._ranges import Range
 from arvados.keep import KeepLocator
-from arvados.collection import Collection, CollectionReader
+from arvados.collection import Collection
 from arvados.arvfile import ArvadosFile, ArvadosFileReader
 
 from . import arvados_testutil as tutil
@@ -26,16 +23,19 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         def __init__(self, blocks):
             self.blocks = blocks
             self.requests = []
-        def get(self, locator, num_retries=0):
+            self.num_prefetch_threads = 1
+        def get(self, locator, num_retries=0, prefetch=False):
             self.requests.append(locator)
             return self.blocks.get(locator)
         def get_from_cache(self, locator):
             self.requests.append(locator)
             return self.blocks.get(locator)
-        def put(self, data, num_retries=None, copies=None):
+        def put(self, data, num_retries=None, copies=None, classes=[]):
             pdh = tutil.str_keep_locator(data)
             self.blocks[pdh] = bytes(data)
             return pdh
+        def block_prefetch(self, loc):
+            self.requests.append(loc)
 
     class MockApi(object):
         def __init__(self, b, r):
@@ -79,7 +79,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
         })
         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
-                             api_client=api, keep_client=keep) as c:
+                        api_client=api, keep_client=keep) as c:
             writer = c.open("count.txt", "rb+")
             self.assertEqual(writer.size(), 10)
             self.assertEqual(b"0123456789", writer.read(12))
@@ -99,6 +99,76 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
             self.assertFalse(c.modified())
 
+
+    def test_truncate2(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({
+            "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
+        })
+        api = ArvadosFileWriterTestCase.MockApi({
+            "name": "test_truncate2",
+            "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
+            "replication_desired": None,
+        }, {
+            "uuid": "zzzzz-4zz18-mockcollection0",
+            "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
+            "portable_data_hash": "272da898abdf86ddc71994835e3155f8+95",
+        })
+        with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
+                        api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "rb+")
+            self.assertEqual(writer.size(), 10)
+            self.assertEqual(b"0123456789", writer.read(12))
+
+            # extend file size
+            writer.truncate(12)
+
+            self.assertEqual(writer.size(), 12)
+            writer.seek(0, os.SEEK_SET)
+            self.assertEqual(b"0123456789\x00\x00", writer.read(12))
+
+            self.assertIsNone(c.manifest_locator())
+            self.assertTrue(c.modified())
+            c.save_new("test_truncate2")
+            self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
+            self.assertFalse(c.modified())
+
+    def test_truncate3(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({
+            "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
+            "a925576942e94b2ef57a066101b48876+10": b"abcdefghij",
+        })
+        api = ArvadosFileWriterTestCase.MockApi({
+            "name": "test_truncate",
+            "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
+            "replication_desired": None,
+        }, {
+            "uuid": "zzzzz-4zz18-mockcollection0",
+            "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
+            "portable_data_hash": "7fcd0eaac3aad4c31a6a0e756475da92+52",
+        })
+        with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n',
+                        api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "rb+")
+            self.assertEqual(writer.size(), 20)
+            self.assertEqual(b"0123456789ab", writer.read(12))
+            self.assertEqual(12, writer.tell())
+
+            writer.truncate(8)
+
+            # Make sure reading off the end doesn't break
+            self.assertEqual(12, writer.tell())
+            self.assertEqual(b"", writer.read(12))
+
+            self.assertEqual(writer.size(), 8)
+            self.assertEqual(2, writer.seek(-10, os.SEEK_CUR))
+            self.assertEqual(b"234567", writer.read(12))
+
+            self.assertIsNone(c.manifest_locator())
+            self.assertTrue(c.modified())
+            c.save_new("test_truncate")
+            self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
+            self.assertFalse(c.modified())
+
     def test_write_to_end(self):
         keep = ArvadosFileWriterTestCase.MockKeep({
             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
@@ -117,7 +187,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             writer = c.open("count.txt", "rb+")
             self.assertEqual(writer.size(), 10)
 
-            writer.seek(5, os.SEEK_SET)
+            self.assertEqual(5, writer.seek(5, os.SEEK_SET))
             self.assertEqual(b"56789", writer.read(8))
 
             writer.seek(10, os.SEEK_SET)
@@ -141,19 +211,41 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
         keep = ArvadosFileWriterTestCase.MockKeep({
             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
         })
-        c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
-        writer = c.open("count.txt", "ab+")
-        self.assertEqual(writer.read(20), b"0123456789")
-        writer.seek(0, os.SEEK_SET)
+        for (mode, convert) in (
+                ('a+', lambda data: data.decode(encoding='utf-8')),
+                ('at+', lambda data: data.decode(encoding='utf-8')),
+                ('ab+', lambda data: data)):
+            c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
+            writer = c.open("count.txt", mode)
+            self.assertEqual(writer.read(20), convert(b"0123456789"))
+
+            writer.seek(0, os.SEEK_SET)
+            writer.write(convert(b"hello"))
+            self.assertEqual(writer.read(), convert(b""))
+            if 'b' in mode:
+                writer.seek(-5, os.SEEK_CUR)
+                self.assertEqual(writer.read(3), convert(b"hel"))
+                self.assertEqual(writer.read(), convert(b"lo"))
+            else:
+                with self.assertRaises(IOError):
+                    writer.seek(-5, os.SEEK_CUR)
+                with self.assertRaises(IOError):
+                    writer.seek(-3, os.SEEK_END)
+            writer.seek(0, os.SEEK_SET)
+            writer.read(7)
+            self.assertEqual(7, writer.tell())
+            self.assertEqual(7, writer.seek(7, os.SEEK_SET))
 
-        writer.write("hello")
-        self.assertEqual(writer.read(20), b"0123456789hello")
-        writer.seek(0, os.SEEK_SET)
+            writer.seek(0, os.SEEK_SET)
+            self.assertEqual(writer.read(), convert(b"0123456789hello"))
 
-        writer.write("world")
-        self.assertEqual(writer.read(20), b"0123456789helloworld")
+            writer.seek(0)
+            writer.write(convert(b"world"))
+            self.assertEqual(writer.read(), convert(b""))
+            writer.seek(0)
+            self.assertEqual(writer.read(), convert(b"0123456789helloworld"))
 
-        self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
+            self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
 
     def test_write_at_beginning(self):
         keep = ArvadosFileWriterTestCase.MockKeep({
@@ -268,7 +360,6 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
             self.assertFalse(c.modified())
 
-
     def test_large_write(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({}, {})
@@ -285,6 +376,69 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
 
             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
 
+    def test_sparse_write(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
+                             api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "rb+")
+            self.assertEqual(writer.size(), 0)
+
+            text = b"0123456789"
+            writer.seek(2)
+            writer.write(text)
+            self.assertEqual(writer.size(), 12)
+            writer.seek(0, os.SEEK_SET)
+            self.assertEqual(writer.read(), b"\x00\x00"+text)
+
+            self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
+
+    def test_sparse_write2(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
+                             api_client=api, keep_client=keep) as c:
+            writer = c.open("count.txt", "rb+")
+            self.assertEqual(writer.size(), 0)
+
+            text = "0123456789"
+            writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
+            writer.write(text)
+            self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
+            writer.seek(0, os.SEEK_SET)
+
+            self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:67108864:count.txt 0:67108864:count.txt 0:2:count.txt 67108864:10:count.txt\n")
+
+    def test_sparse_write3(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
+            with Collection(api_client=api, keep_client=keep) as c:
+                writer = c.open("count.txt", "rb+")
+                self.assertEqual(writer.size(), 0)
+
+                for i in r:
+                    w = ("%s" % i) * 10
+                    writer.seek(i*10)
+                    writer.write(w.encode())
+                writer.seek(0)
+                self.assertEqual(writer.read(), b"00000000001111111111222222222233333333334444444444")
+
+    def test_sparse_write4(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({})
+        api = ArvadosFileWriterTestCase.MockApi({}, {})
+        for r in [[0, 1, 2, 4], [4, 2, 1, 0], [2, 0, 4, 1]]:
+            with Collection(api_client=api, keep_client=keep) as c:
+                writer = c.open("count.txt", "rb+")
+                self.assertEqual(writer.size(), 0)
+
+                for i in r:
+                    w = ("%s" % i) * 10
+                    writer.seek(i*10)
+                    writer.write(w.encode())
+                writer.seek(0)
+                self.assertEqual(writer.read(), b"000000000011111111112222222222\x00\x00\x00\x00\x00\x00\x00\x00\x00\x004444444444")
+
     def test_rewrite_on_empty_file(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
@@ -346,10 +500,10 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
     def test_write_large_rewrite(self):
         keep = ArvadosFileWriterTestCase.MockKeep({})
         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
-                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
+                                                 "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
                                                  "replication_desired":None},
                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
-                                                 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
+                                                 "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
                              api_client=api, keep_client=keep) as c:
@@ -472,6 +626,8 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
             def __init__(self, blocks, nocache):
                 self.blocks = blocks
                 self.nocache = nocache
+                self._keep = ArvadosFileWriterTestCase.MockKeep({})
+                self.prefetch_lookahead = 0
 
             def block_prefetch(self, loc):
                 pass
@@ -533,8 +689,60 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
             r = c.open("count.txt", "rb")
             self.assertEqual(b"0123", r.read(4))
-        self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
-        self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
+        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
+                          "e8dc4081b13434b45189a720b77b6818+8"], keep.requests)
+
+    def test_prefetch_disabled(self):
+        keep = ArvadosFileWriterTestCase.MockKeep({
+            "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
+            "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
+        })
+        keep.num_prefetch_threads = 0
+        with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
+            r = c.open("count.txt", "rb")
+            self.assertEqual(b"0123", r.read(4))
+
+        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
+
+    def test_prefetch_first_read_only(self):
+        # test behavior that prefetch only happens every 128 reads
+        # check that it doesn't make a prefetch request on the second read
+        keep = ArvadosFileWriterTestCase.MockKeep({
+            "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
+            "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
+        })
+        with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
+            r = c.open("count.txt", "rb")
+            self.assertEqual(b"0123", r.read(4))
+            self.assertEqual(b"45", r.read(2))
+        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
+                          "e8dc4081b13434b45189a720b77b6818+8",
+                          "2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
+        self.assertEqual(3, len(keep.requests))
+
+    def test_prefetch_again(self):
+        # test behavior that prefetch only happens every 128 reads
+        # check that it does make another prefetch request after 128 reads
+        keep = ArvadosFileWriterTestCase.MockKeep({
+            "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
+            "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
+        })
+        with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
+            r = c.open("count.txt", "rb")
+            for i in range(0, 129):
+                r.seek(0)
+                self.assertEqual(b"0123", r.read(4))
+        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
+                          "e8dc4081b13434b45189a720b77b6818+8",
+                          "2e9ec317e197819358fbc43afca7d837+8",
+                          "2e9ec317e197819358fbc43afca7d837+8"], keep.requests[0:4])
+        self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
+                          "2e9ec317e197819358fbc43afca7d837+8",
+                          "2e9ec317e197819358fbc43afca7d837+8",
+                          "e8dc4081b13434b45189a720b77b6818+8"], keep.requests[127:131])
+        # gets the 1st block 129 times from keep (cache),
+        # and the 2nd block twice to get 131 requests
+        self.assertEqual(131, len(keep.requests))
 
     def test__eq__from_manifest(self):
         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
@@ -687,7 +895,7 @@ class BlockManagerTest(unittest.TestCase):
         mockkeep = mock.MagicMock()
         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
             bufferblock = blockmanager.alloc_bufferblock()
-            bufferblock.owner = mock.MagicMock()
+            bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
             def flush(sync=None):
                 blockmanager.commit_bufferblock(bufferblock, sync)
             bufferblock.owner.flush.side_effect = flush
@@ -701,7 +909,7 @@ class BlockManagerTest(unittest.TestCase):
     def test_bufferblock_commit_pending(self):
         # Test for bug #7225
         mockkeep = mock.MagicMock()
-        mockkeep.put.side_effect = lambda x: time.sleep(1)
+        mockkeep.put.side_effect = lambda *args, **kwargs: time.sleep(1)
         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
             bufferblock = blockmanager.alloc_bufferblock()
             bufferblock.append("foo")
@@ -712,13 +920,12 @@ class BlockManagerTest(unittest.TestCase):
             blockmanager.commit_bufferblock(bufferblock, True)
             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 
-
     def test_bufferblock_commit_with_error(self):
         mockkeep = mock.MagicMock()
         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
             bufferblock = blockmanager.alloc_bufferblock()
-            bufferblock.owner = mock.MagicMock()
+            bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
             def flush(sync=None):
                 blockmanager.commit_bufferblock(bufferblock, sync)
             bufferblock.owner.flush.side_effect = flush