-#!/usr/bin/env python
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
from __future__ import absolute_import
from builtins import hex
"portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
- api_client=api, keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ 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))
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",
})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
api_client=api, keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ 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)
"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
})
c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
- writer = c.open("count.txt", "a+")
+ writer = c.open("count.txt", "ab+")
self.assertEqual(writer.read(20), b"0123456789")
- writer.seek(0, os.SEEK_SET)
+ writer.seek(0, os.SEEK_SET)
writer.write("hello")
- self.assertEqual(writer.read(20), b"0123456789hello")
+ self.assertEqual(writer.read(), b"")
+ writer.seek(-5, os.SEEK_CUR)
+ self.assertEqual(writer.read(3), b"hel")
+ self.assertEqual(writer.read(), b"lo")
writer.seek(0, os.SEEK_SET)
+ self.assertEqual(writer.read(), b"0123456789hello")
+ writer.seek(0)
writer.write("world")
- self.assertEqual(writer.read(20), b"0123456789helloworld")
+ self.assertEqual(writer.read(), b"")
+ writer.seek(0)
+ self.assertEqual(writer.read(), b"0123456789helloworld")
self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(b"0123456789", writer.readfrom(0, 13))
writer.seek(0, os.SEEK_SET)
writer.write("foo")
def test_write_empty(self):
keep = ArvadosFileWriterTestCase.MockKeep({})
with Collection(keep_client=keep) as c:
- writer = c.open("count.txt", "w")
+ writer = c.open("count.txt", "wb")
self.assertEqual(writer.size(), 0)
self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
def test_save_manifest_text(self):
keep = ArvadosFileWriterTestCase.MockKeep({})
with Collection(keep_client=keep) as c:
- writer = c.open("count.txt", "w")
+ writer = c.open("count.txt", "wb")
writer.write(b"0123456789")
self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
def test_get_manifest_text_commits(self):
keep = ArvadosFileWriterTestCase.MockKeep({})
with Collection(keep_client=keep) as c:
- writer = c.open("count.txt", "w")
+ writer = c.open("count.txt", "wb")
writer.write("0123456789")
self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(b"0123456789", writer.readfrom(0, 13))
writer.seek(3, os.SEEK_SET)
writer.write("foo")
keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(b"0123456789", writer.readfrom(0, 13))
writer.seek(7, os.SEEK_SET)
writer.write("foo")
keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(b"012345678901234", writer.readfrom(0, 15))
writer.seek(7, os.SEEK_SET)
writer.write("foobar")
keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(b"012301230123", writer.readfrom(0, 15))
writer.seek(2, os.SEEK_SET)
writer.write("abcdefg")
"portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
api_client=api, keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
text = "0123456789" * 100
for b in range(0, 100000):
writer.write(text)
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", "r+")
+ writer = c.open("count.txt", "rb+")
self.assertEqual(writer.size(), 0)
text = "0123456789"
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() 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() 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',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
for b in range(0, 10):
writer.seek(0, os.SEEK_SET)
writer.write("0123456789")
})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
for b in range(0, 10):
writer.seek(10, os.SEEK_SET)
writer.write("abcdefghij")
})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
keep_client=keep) as c:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
for b in range(0, 10):
writer.seek(5, os.SEEK_SET)
writer.write("abcdefghij")
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:
- writer = c.open("count.txt", "r+")
+ writer = c.open("count.txt", "rb+")
text = b''.join([b"0123456789" for a in range(0, 100)])
for b in range(0, 100000):
writer.write(text)
"portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51",
})
with Collection(api_client=api, keep_client=keep) as c:
- writer = c.open("count.txt", "w+")
+ writer = c.open("count.txt", "wb+")
self.assertEqual(writer.size(), 0)
writer.write("01234567")
self.assertEqual(writer.size(), 8)
"portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
with Collection(api_client=api, keep_client=keep) as c:
self.assertIsNone(c.api_response())
- writer = c.open("foo/bar/count.txt", "w+")
+ writer = c.open("foo/bar/count.txt", "wb+")
writer.write("01234567")
self.assertFalse(c.committed())
c.save_new("test_create")
"portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
api_client=api, keep_client=keep) as c:
- writer = c.open("count.txt", "w+")
+ writer = c.open("count.txt", "wb+")
self.assertEqual(writer.size(), 0)
writer.write("01234567")
self.assertEqual(writer.size(), 8)
def test_file_not_found(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
with self.assertRaises(IOError):
- writer = c.open("nocount.txt", "r")
+ writer = c.open("nocount.txt", "rb")
def test_cannot_open_directory(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
with self.assertRaises(IOError):
- writer = c.open(".", "r")
+ writer = c.open(".", "rb")
def test_create_multiple(self):
keep = ArvadosFileWriterTestCase.MockKeep({})
"manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
"portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
with Collection(api_client=api, keep_client=keep) as c:
- w1 = c.open("count1.txt", "w")
- w2 = c.open("count2.txt", "w")
+ w1 = c.open("count1.txt", "wb")
+ w2 = c.open("count2.txt", "wb")
w1.write("01234567")
w2.write("abcdefgh")
self.assertEqual(w1.size(), 8)
stream.append(Range(loc, n, len(d)))
n += len(d)
af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
- return ArvadosFileReader(af)
+ return ArvadosFileReader(af, mode="rb")
def test_read_block_crossing_behavior(self):
# read() needs to return all the data requested if possible, even if it
"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", "r")
+ 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)
def test__eq__from_writes(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
with Collection() as c2:
- f = c2.open("count1.txt", "w")
+ f = c2.open("count1.txt", "wb")
f.write("0123456789")
self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
def test__ne__(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
with Collection() as c2:
- f = c2.open("count1.txt", "w")
+ f = c2.open("count1.txt", "wb")
f.write("1234567890")
self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
af = ArvadosFile(col, "test",
stream=stream,
segments=segments)
+ kwargs.setdefault('mode', 'rb')
return ArvadosFileReader(af, **kwargs)
def read_for_test(self, reader, byte_count, **kwargs):
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
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