#!/usr/bin/env python
import bz2
+import datetime
import gzip
import io
import mock
import os
import unittest
-import hashlib
+import time
import arvados
from arvados._ranges import Range
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
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
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',
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',
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"))
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',
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):
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',
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:
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',
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:
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:
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"))
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)
- af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
- return ArvadosFileReader(af, "count.txt")
+ 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)
def test_read_block_crossing_behavior(self):
# read() needs to return all the data requested if possible, even if it
def test__eq__from_writes(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
with Collection() as c2:
- with c2.open("count1.txt", "w") as f:
- f.write("0123456789")
+ f = c2.open("count1.txt", "w")
+ f.write("0123456789")
self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
def test__ne__(self):
with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
with Collection() as c2:
- with c2.open("count1.txt", "w") as f:
- f.write("1234567890")
+ f = c2.open("count1.txt", "w")
+ f.write("1234567890")
self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
blockmanager = arvados.arvfile._BlockManager(self.keep_client())
blockmanager.prefetch_enabled = False
col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
- af = ArvadosFile(col,
+ af = ArvadosFile(col, "test",
stream=stream,
segments=segments)
- return ArvadosFileReader(af, "test", **kwargs)
+ return ArvadosFileReader(af, **kwargs)
def read_for_test(self, reader, byte_count, **kwargs):
return reader.read(byte_count, **kwargs)
def read_for_test(self, reader, byte_count, **kwargs):
return ''.join(reader.readlines(**kwargs))
+
+class ArvadosFileTestCase(unittest.TestCase):
+ def datetime_to_hex(self, dt):
+ return hex(int(time.mktime(dt.timetuple())))[2:]
+
+ def test_permission_expired(self):
+ base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
+ now = datetime.datetime.now()
+ a_week_ago = now - datetime.timedelta(days=7)
+ a_month_ago = now - datetime.timedelta(days=30)
+ a_week_from_now = now + datetime.timedelta(days=7)
+ with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
+ self.assertFalse(c.find('count.txt').permission_expired())
+ with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
+ f = c.find('count.txt')
+ self.assertTrue(f.permission_expired())
+ self.assertTrue(f.permission_expired(a_week_from_now))
+ self.assertFalse(f.permission_expired(a_month_ago))
+
+
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: blockmanager.commit_bufferblock(bufferblock)
- 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: blockmanager.commit_bufferblock(bufferblock)
- 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)