X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/ad69b48e324c3ce29a4d2c84732dfd3d0288ebb3..62d56bce0d714cc2df2ab5e7f1005dc3d76f783b:/sdk/python/tests/test_arvfile.py diff --git a/sdk/python/tests/test_arvfile.py b/sdk/python/tests/test_arvfile.py index 9c67fbd53f..6b3562602a 100644 --- a/sdk/python/tests/test_arvfile.py +++ b/sdk/python/tests/test_arvfile.py @@ -6,13 +6,13 @@ import io import mock import os import unittest -import hashlib +import time import arvados from arvados._ranges import Range from arvados.keep import KeepLocator from arvados.collection import Collection, CollectionReader -from arvados.arvfile import ArvadosFile, ArvadosFileReader, SYNC_READONLY, SYNC_EXPLICIT +from arvados.arvfile import ArvadosFile, ArvadosFileReader import arvados_testutil as tutil from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin @@ -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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + 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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + self.assertFalse(c.modified()) self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) @@ -128,7 +135,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("world") self.assertEqual(writer.read(20), "0123456789helloworld") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.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({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -140,7 +147,36 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foo") self.assertEqual(writer.size(), 10) self.assertEqual("foo3456789", writer.readfrom(0, 13)) - self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.manifest_text()) + self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text()) + + def test_write_empty(self): + keep = ArvadosFileWriterTestCase.MockKeep({}) + with Collection(keep_client=keep) as c: + writer = c.open("count.txt", "w") + 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.write("0123456789") + self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text()) + self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks) + + self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False)) + self.assertIn('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.write("0123456789") + self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text()) + self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks) + self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text()) + self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks) + def test_write_in_middle(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -152,7 +188,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foo") self.assertEqual(writer.size(), 10) self.assertEqual("012foo6789", writer.readfrom(0, 13)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text()) def test_write_at_end(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -164,7 +200,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foo") self.assertEqual(writer.size(), 10) self.assertEqual("0123456foo", writer.readfrom(0, 13)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text()) def test_write_across_segment_boundary(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -176,7 +212,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foobar") self.assertEqual(writer.size(), 20) self.assertEqual("0123456foobar34", writer.readfrom(0, 15)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text()) def test_write_across_several_segments(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -188,27 +224,45 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("abcdefg") self.assertEqual(writer.size(), 12) self.assertEqual("01abcdefg123", writer.readfrom(0, 15)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text()) 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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + 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({}) @@ -218,10 +272,14 @@ class ArvadosFileWriterTestCase(unittest.TestCase): for b in xrange(0, 10): writer.seek(0, os.SEEK_SET) writer.write("0123456789") - writer.arvadosfile._repack_writes() + + self.assertEqual(writer.size(), 10) + self.assertEqual("0123456789", writer.readfrom(0, 20)) + self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text()) + writer.flush() self.assertEqual(writer.size(), 10) self.assertEqual("0123456789", writer.readfrom(0, 20)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text()) def test_rewrite_append_existing_file(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -231,10 +289,15 @@ class ArvadosFileWriterTestCase(unittest.TestCase): for b in xrange(0, 10): writer.seek(10, os.SEEK_SET) writer.write("abcdefghij") - writer.arvadosfile._repack_writes() + self.assertEqual(writer.size(), 20) self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text()) + + writer.arvadosfile.flush() + self.assertEqual(writer.size(), 20) + self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text()) def test_rewrite_over_existing_file(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) @@ -244,15 +307,22 @@ class ArvadosFileWriterTestCase(unittest.TestCase): for b in xrange(0, 10): writer.seek(5, os.SEEK_SET) writer.write("abcdefghij") - writer.arvadosfile._repack_writes() + self.assertEqual(writer.size(), 15) self.assertEqual("01234abcdefghij", writer.readfrom(0, 20)) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.manifest_text()) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text()) + + writer.arvadosfile.flush() + + self.assertEqual(writer.size(), 15) + self.assertEqual("01234abcdefghij", writer.readfrom(0, 20)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text()) 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', @@ -265,16 +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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + 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: @@ -283,32 +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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + 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', @@ -318,11 +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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + self.assertFalse(c.modified()) def test_file_not_found(self): with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: @@ -337,7 +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: @@ -348,12 +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.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) + self.assertFalse(c.modified()) self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8")) @@ -383,36 +459,38 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): def _my_block_manager(self): return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache) - def sync_mode(self): - return SYNC_READONLY - def make_count_reader(self, nocache=False): stream = [] 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_returns_first_block(self): - # read() calls will be aligned on block boundaries - see #3663. + def test_read_block_crossing_behavior(self): + # read() needs to return all the data requested if possible, even if it + # crosses uncached blocks: https://arvados.org/issues/5856 sfile = self.make_count_reader(nocache=True) - self.assertEqual('123', sfile.read(10)) + self.assertEqual('12345678', sfile.read(8)) def test_successive_reads(self): + # Override StreamFileReaderTestCase.test_successive_reads sfile = self.make_count_reader(nocache=True) - for expect in ['123', '456', '789', '']: - self.assertEqual(expect, sfile.read(10)) + self.assertEqual('1234', sfile.read(4)) + self.assertEqual('5678', sfile.read(4)) + self.assertEqual('9', sfile.read(4)) + self.assertEqual('', sfile.read(4)) def test_tell_after_block_read(self): + # Override StreamFileReaderTestCase.test_tell_after_block_read sfile = self.make_count_reader(nocache=True) - sfile.read(5) - self.assertEqual(3, sfile.tell()) + self.assertEqual('12345678', sfile.read(8)) + self.assertEqual(8, sfile.tell()) def test_prefetch(self): keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"}) @@ -431,8 +509,8 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): 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"]) @@ -440,8 +518,8 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): 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"]) @@ -460,12 +538,14 @@ class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin): n += k.size except ValueError: pass - col = Collection(keep_client=self.keep_client()) - col._my_block_manager().prefetch_enabled = False - af = ArvadosFile(col, + + blockmanager = arvados.arvfile._BlockManager(self.keep_client()) + blockmanager.prefetch_enabled = False + col = Collection(keep_client=self.keep_client(), block_manager=blockmanager) + 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) @@ -493,75 +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.append("foo") - blockmanager.commit_all() - 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.append("foo") - with self.assertRaises(arvados.errors.KeepWriteError) as err: - blockmanager.commit_all() - self.assertEquals(str(err.exception), "Error writing some blocks: 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)