X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/682dd5b6cc23a455766a7651e3e841257660b31c..62d56bce0d714cc2df2ab5e7f1005dc3d76f783b:/sdk/python/tests/test_arvfile.py diff --git a/sdk/python/tests/test_arvfile.py b/sdk/python/tests/test_arvfile.py index c5dbb1692a..6b3562602a 100644 --- a/sdk/python/tests/test_arvfile.py +++ b/sdk/python/tests/test_arvfile.py @@ -6,7 +6,7 @@ import io import mock import os import unittest -import hashlib +import time import arvados from arvados._ranges import Range @@ -28,8 +28,8 @@ class ArvadosFileWriterTestCase(unittest.TestCase): def get_from_cache(self, locator): self.requests.append(locator) return self.blocks.get(locator) - def put(self, data, num_retries=None): - pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data)) + def put(self, data, num_retries=None, copies=None): + pdh = tutil.str_keep_locator(data) self.blocks[pdh] = str(data) return pdh @@ -37,6 +37,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase): def __init__(self, b, r): self.body = b self.response = r + self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema() + self._rootDesc = {} + class MockSchema(object): + def __init__(self): + self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}} class MockCollections(object): def __init__(self, b, r): self.body = b @@ -59,7 +64,8 @@ class ArvadosFileWriterTestCase(unittest.TestCase): def test_truncate(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate", - "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"}, + "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"}) with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', @@ -77,16 +83,17 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.seek(0, os.SEEK_SET) self.assertEqual("01234567", writer.read(12)) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) c.save_new("test_truncate") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) def test_write_to_end(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_append", - "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"}, + "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"}) with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', @@ -104,13 +111,13 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.seek(5, os.SEEK_SET) self.assertEqual("56789foo", writer.read(8)) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) - self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) + self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) c.save_new("test_append") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) @@ -222,22 +229,40 @@ class ArvadosFileWriterTestCase(unittest.TestCase): def test_write_large(self): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large", - "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"}, + "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"}) 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+") - text = ''.join(["0123456789" for a in xrange(0, 100)]) + text = "0123456789" * 100 for b in xrange(0, 100000): writer.write(text) self.assertEqual(writer.size(), 100000000) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) c.save_new("test_write_large") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) + + + def test_large_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", "r+") + self.assertEqual(writer.size(), 0) + + text = "0123456789" + writer.write(text) + text = "0123456789" * 9999999 + writer.write(text) + self.assertEqual(writer.size(), 100000000) + + self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n") def test_rewrite_on_empty_file(self): keep = ArvadosFileWriterTestCase.MockKeep({}) @@ -296,7 +321,8 @@ 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": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000: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"}) with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt', @@ -309,16 +335,17 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foo") self.assertEqual(writer.size(), 100000000) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) c.save_new("test_write_large") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) def test_create(self): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_create", - "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}, + "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}) with Collection(api_client=api, keep_client=keep) as c: @@ -327,32 +354,36 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("01234567") self.assertEqual(writer.size(), 8) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) - self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8")) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) + self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8")) c.save_new("test_create") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8")) def test_create_subdir(self): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_create", - "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}, + "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}) with Collection(api_client=api, keep_client=keep) as c: self.assertIsNone(c.api_response()) writer = c.open("foo/bar/count.txt", "w+") writer.write("01234567") + self.assertFalse(c.committed()) c.save_new("test_create") + self.assertTrue(c.committed()) self.assertEqual(c.api_response(), api.response) def test_overwrite(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite", - "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}, + "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}) with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', @@ -362,11 +393,11 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("01234567") self.assertEqual(writer.size(), 8) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) c.save_new("test_overwrite") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) def test_file_not_found(self): with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: @@ -381,7 +412,8 @@ class ArvadosFileWriterTestCase(unittest.TestCase): def test_create_multiple(self): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple", - "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"}, + "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n", + "replication_desired":None}, {"uuid":"zzzzz-4zz18-mockcollection0", "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"}) with Collection(api_client=api, keep_client=keep) as c: @@ -392,12 +424,12 @@ class ArvadosFileWriterTestCase(unittest.TestCase): self.assertEqual(w1.size(), 8) self.assertEqual(w2.size(), 8) - self.assertEqual(None, c.manifest_locator()) - self.assertEqual(True, c.modified()) - self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8")) + self.assertIsNone(c.manifest_locator()) + self.assertTrue(c.modified()) + self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8")) c.save_new("test_create_multiple") self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) - self.assertEqual(False, c.modified()) + self.assertFalse(c.modified()) self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8")) @@ -433,7 +465,7 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): n = 0 blocks = {} for d in ['01234', '34567', '67890']: - loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d)) + loc = tutil.str_keep_locator(d) blocks[loc] = d stream.append(Range(loc, n, len(d))) n += len(d) @@ -541,81 +573,99 @@ class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase): class BlockManagerTest(unittest.TestCase): def test_bufferblock_append(self): keep = ArvadosFileWriterTestCase.MockKeep({}) - blockmanager = arvados.arvfile._BlockManager(keep) - bufferblock = blockmanager.alloc_bufferblock() - bufferblock.append("foo") + with arvados.arvfile._BlockManager(keep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.append("foo") - self.assertEqual(bufferblock.size(), 3) - self.assertEqual(bufferblock.buffer_view[0:3], "foo") - self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") + self.assertEqual(bufferblock.size(), 3) + self.assertEqual(bufferblock.buffer_view[0:3], "foo") + self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") - bufferblock.append("bar") + bufferblock.append("bar") - self.assertEqual(bufferblock.size(), 6) - self.assertEqual(bufferblock.buffer_view[0:6], "foobar") - self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6") + self.assertEqual(bufferblock.size(), 6) + self.assertEqual(bufferblock.buffer_view[0:6], "foobar") + self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6") - bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING) - with self.assertRaises(arvados.errors.AssertionError): - bufferblock.append("bar") + bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING) + with self.assertRaises(arvados.errors.AssertionError): + bufferblock.append("bar") def test_bufferblock_dup(self): keep = ArvadosFileWriterTestCase.MockKeep({}) - blockmanager = arvados.arvfile._BlockManager(keep) - bufferblock = blockmanager.alloc_bufferblock() - bufferblock.append("foo") + with arvados.arvfile._BlockManager(keep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.append("foo") - self.assertEqual(bufferblock.size(), 3) - self.assertEqual(bufferblock.buffer_view[0:3], "foo") - self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") - bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING) + self.assertEqual(bufferblock.size(), 3) + self.assertEqual(bufferblock.buffer_view[0:3], "foo") + self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") + bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING) - bufferblock2 = blockmanager.dup_block(bufferblock, None) - self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid) + bufferblock2 = blockmanager.dup_block(bufferblock, None) + self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid) - bufferblock2.append("bar") + bufferblock2.append("bar") - self.assertEqual(bufferblock2.size(), 6) - self.assertEqual(bufferblock2.buffer_view[0:6], "foobar") - self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6") + self.assertEqual(bufferblock2.size(), 6) + self.assertEqual(bufferblock2.buffer_view[0:6], "foobar") + self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6") - self.assertEqual(bufferblock.size(), 3) - self.assertEqual(bufferblock.buffer_view[0:3], "foo") - self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") + self.assertEqual(bufferblock.size(), 3) + self.assertEqual(bufferblock.buffer_view[0:3], "foo") + self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3") def test_bufferblock_get(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - blockmanager = arvados.arvfile._BlockManager(keep) - bufferblock = blockmanager.alloc_bufferblock() - bufferblock.append("foo") + with arvados.arvfile._BlockManager(keep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.append("foo") - self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789") - self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo") + self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789") + self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo") def test_bufferblock_commit(self): mockkeep = mock.MagicMock() - blockmanager = arvados.arvfile._BlockManager(mockkeep) - bufferblock = blockmanager.alloc_bufferblock() - bufferblock.owner = mock.MagicMock() - bufferblock.owner.flush.side_effect = lambda x: blockmanager.commit_bufferblock(bufferblock, False) - bufferblock.append("foo") - blockmanager.commit_all() - self.assertTrue(bufferblock.owner.flush.called) - self.assertTrue(mockkeep.put.called) - self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED) - self.assertIsNone(bufferblock.buffer_view) + with arvados.arvfile._BlockManager(mockkeep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.owner = mock.MagicMock() + def flush(sync=None): + blockmanager.commit_bufferblock(bufferblock, sync) + bufferblock.owner.flush.side_effect = flush + bufferblock.append("foo") + blockmanager.commit_all() + self.assertTrue(bufferblock.owner.flush.called) + self.assertTrue(mockkeep.put.called) + self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED) + self.assertIsNone(bufferblock.buffer_view) + + def test_bufferblock_commit_pending(self): + # Test for bug #7225 + mockkeep = mock.MagicMock() + mockkeep.put.side_effect = lambda x: time.sleep(1) + with arvados.arvfile._BlockManager(mockkeep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.append("foo") + + blockmanager.commit_bufferblock(bufferblock, False) + self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING) + + 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") - blockmanager = arvados.arvfile._BlockManager(mockkeep) - bufferblock = blockmanager.alloc_bufferblock() - bufferblock.owner = mock.MagicMock() - bufferblock.owner.flush.side_effect = lambda x: blockmanager.commit_bufferblock(bufferblock, False) - bufferblock.append("foo") - with self.assertRaises(arvados.errors.KeepWriteError) as err: - blockmanager.commit_all() - self.assertTrue(bufferblock.owner.flush.called) - self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)") - self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING) + with arvados.arvfile._BlockManager(mockkeep) as blockmanager: + bufferblock = blockmanager.alloc_bufferblock() + bufferblock.owner = mock.MagicMock() + def flush(sync=None): + blockmanager.commit_bufferblock(bufferblock, sync) + bufferblock.owner.flush.side_effect = flush + bufferblock.append("foo") + with self.assertRaises(arvados.errors.KeepWriteError) as err: + blockmanager.commit_all() + self.assertTrue(bufferblock.owner.flush.called) + self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)") + self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)