X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/86f94e641462ddd78a454fc0a0cdc9dd4c4d9bef..63b03a39adfd78961c5bbb6a3a2d02ccd8c92e4d:/sdk/python/tests/test_arvfile.py diff --git a/sdk/python/tests/test_arvfile.py b/sdk/python/tests/test_arvfile.py index 91cfc79982..f50c446313 100644 --- a/sdk/python/tests/test_arvfile.py +++ b/sdk/python/tests/test_arvfile.py @@ -9,217 +9,276 @@ import unittest import hashlib import arvados -from arvados import ArvadosFile, ArvadosFileReader, Range, import_manifest, export_manifest +from arvados._ranges import Range +from arvados.keep import KeepLocator +from arvados.collection import Collection, CollectionReader +from arvados.arvfile import ArvadosFile, ArvadosFileReader import arvados_testutil as tutil -from test_stream import StreamFileReaderTestCase +from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin class ArvadosFileWriterTestCase(unittest.TestCase): class MockKeep(object): def __init__(self, blocks): self.blocks = blocks self.requests = [] - def get(self, locator, num_retries=0, cache_only=False): + def get(self, locator, num_retries=0): self.requests.append(locator) return self.blocks.get(locator) - def put(self, data): + 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)) self.blocks[pdh] = str(data) return pdh class MockApi(object): def __init__(self, b, r): - self.b = b - self.r = r + self.body = b + self.response = r class MockCollections(object): def __init__(self, b, r): - self.b = b - self.r = r + self.body = b + self.response = r class Execute(object): def __init__(self, r): - self.r = r + self.response = r def execute(self, num_retries=None): - return self.r + return self.response def create(self, ensure_unique_name=False, body=None): - if body != self.b: - raise Exception("Body %s does not match expectation %s" % (body, self.b)) - return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r) + if body != self.body: + raise Exception("Body %s does not match expectation %s" % (body, self.body)) + return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response) def update(self, uuid=None, body=None): - return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r) + return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response) def collections(self): - return ArvadosFileWriterTestCase.MockApi.MockCollections(self.b, self.r) + return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response) 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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - api_client=api, keep=keep) as c: + {"uuid":"zzzzz-4zz18-mockcollection0", + "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"}) + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + api_client=api, keep_client=keep) as c: writer = c.open("count.txt", "r+") self.assertEqual(writer.size(), 10) - writer.seek(5) - self.assertEqual("56789", writer.read(8)) + self.assertEqual("0123456789", writer.read(12)) + writer.truncate(8) - writer.seek(5, os.SEEK_SET) - self.assertEqual("567", writer.read(8)) - self.assertEqual(None, c._manifest_locator) + # Make sure reading off the end doesn't break + self.assertEqual("", writer.read(12)) + + self.assertEqual(writer.size(), 8) + writer.seek(0, os.SEEK_SET) + self.assertEqual("01234567", writer.read(12)) + + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) - c.save_as("test_truncate") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_truncate") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) - def test_append(self): + 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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - api_client=api, keep=keep) as c: + {"uuid":"zzzzz-4zz18-mockcollection0", + "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"}) + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + api_client=api, keep_client=keep) as c: writer = c.open("count.txt", "r+") + self.assertEqual(writer.size(), 10) + writer.seek(5, os.SEEK_SET) self.assertEqual("56789", writer.read(8)) + writer.seek(10, os.SEEK_SET) writer.write("foo") self.assertEqual(writer.size(), 13) + writer.seek(5, os.SEEK_SET) self.assertEqual("56789foo", writer.read(8)) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) - c.save_as("test_append") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + + c.save_new("test_append") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3")) - def test_write0(self): + + def test_append(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - keep=keep) as c: + c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep) + writer = c.open("count.txt", "a+") + self.assertEqual(writer.read(20), "0123456789") + writer.seek(0, os.SEEK_SET) + + writer.write("hello") + self.assertEqual(writer.read(20), "0123456789hello") + writer.seek(0, os.SEEK_SET) + + writer.write("world") + self.assertEqual(writer.read(20), "0123456789helloworld") + + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.manifest_text()) + + def test_write_at_beginning(self): + keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + keep_client=keep) as c: writer = c.open("count.txt", "r+") self.assertEqual("0123456789", writer.readfrom(0, 13)) writer.seek(0, os.SEEK_SET) 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", export_manifest(c)) + self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.manifest_text()) - def test_write1(self): + 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.manifest_text()) + + def test_write_in_middle(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - keep=keep) as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + keep_client=keep) as c: writer = c.open("count.txt", "r+") self.assertEqual("0123456789", writer.readfrom(0, 13)) writer.seek(3, os.SEEK_SET) 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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.manifest_text()) - def test_write2(self): + def test_write_at_end(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - keep=keep) as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + keep_client=keep) as c: writer = c.open("count.txt", "r+") self.assertEqual("0123456789", writer.readfrom(0, 13)) writer.seek(7, os.SEEK_SET) 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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.manifest_text()) - def test_write3(self): + def test_write_across_segment_boundary(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n', - keep=keep) as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n', + keep_client=keep) as c: writer = c.open("count.txt", "r+") self.assertEqual("012345678901234", writer.readfrom(0, 15)) writer.seek(7, os.SEEK_SET) 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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.manifest_text()) - def test_write4(self): + def test_write_across_several_segments(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt', - keep=keep) as c: + 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+") self.assertEqual("012301230123", writer.readfrom(0, 15)) writer.seek(2, os.SEEK_SET) 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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt', - api_client=api, keep=keep) as c: + {"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)]) for b in xrange(0, 100000): writer.write(text) self.assertEqual(writer.size(), 100000000) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) - c.save_as("test_write_large") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_write_large") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) - def test_write_rewrite0(self): + def test_rewrite_on_empty_file(self): keep = ArvadosFileWriterTestCase.MockKeep({}) - with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt', - keep=keep) as c: + with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt', + keep_client=keep) as c: writer = c.open("count.txt", "r+") 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.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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.manifest_text()) - def test_write_rewrite1(self): + def test_rewrite_append_existing_file(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt', - keep=keep) as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt', + keep_client=keep) as c: writer = c.open("count.txt", "r+") 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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.manifest_text()) - def test_write_rewrite2(self): + 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.manifest_text()) + + def test_rewrite_over_existing_file(self): keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt', - keep=keep) as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt', + keep_client=keep) as c: writer = c.open("count.txt", "r+") 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 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.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", export_manifest(c)) + self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.manifest_text()) - def test_write_large_rewrite0(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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt', - api_client=api, keep=keep) as c: + {"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', + api_client=api, keep_client=keep) as c: writer = c.open("count.txt", "r+") text = ''.join(["0123456789" for a in xrange(0, 100)]) for b in xrange(0, 100000): @@ -228,28 +287,29 @@ class ArvadosFileWriterTestCase(unittest.TestCase): writer.write("foo") self.assertEqual(writer.size(), 100000000) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) - c.save_as("test_write_large") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_write_large") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) def test_create(self): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_create", "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with arvados.Collection(api_client=api, keep_client=keep) as c: + {"uuid":"zzzzz-4zz18-mockcollection0", + "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}) + with Collection(api_client=api, keep_client=keep) as c: writer = c.open("count.txt", "w+") self.assertEqual(writer.size(), 0) writer.write("01234567") self.assertEqual(writer.size(), 8) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8")) - c.save_as("test_create") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_create") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8")) @@ -258,37 +318,41 @@ class ArvadosFileWriterTestCase(unittest.TestCase): keep = ArvadosFileWriterTestCase.MockKeep({}) api = ArvadosFileWriterTestCase.MockApi({"name":"test_create", "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with arvados.Collection(api_client=api, keep_client=keep) as c: + {"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") - c.save_as("test_create") + c.save_new("test_create") + 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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', - api_client=api, keep=keep) as c: + {"uuid":"zzzzz-4zz18-mockcollection0", + "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"}) + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', + api_client=api, keep_client=keep) as c: writer = c.open("count.txt", "w+") self.assertEqual(writer.size(), 0) writer.write("01234567") self.assertEqual(writer.size(), 8) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) - c.save_as("test_overwrite") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_overwrite") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) def test_file_not_found(self): - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: with self.assertRaises(IOError): writer = c.open("nocount.txt", "r") def test_cannot_open_directory(self): - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c: with self.assertRaises(IOError): writer = c.open(".", "r") @@ -296,8 +360,9 @@ class ArvadosFileWriterTestCase(unittest.TestCase): 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"}, - {"uuid":"zzzzz-4zz18-mockcollection0"}) - with arvados.Collection(api_client=api, keep_client=keep) as c: + {"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: w1 = c.open("count1.txt", "w") w2 = c.open("count2.txt", "w") w1.write("01234567") @@ -305,39 +370,14 @@ class ArvadosFileWriterTestCase(unittest.TestCase): self.assertEqual(w1.size(), 8) self.assertEqual(w2.size(), 8) - self.assertEqual(None, c._manifest_locator) + self.assertEqual(None, c.manifest_locator()) self.assertEqual(True, c.modified()) self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8")) - c.save_as("test_create_multiple") - self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator) + c.save_new("test_create_multiple") + self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator()) self.assertEqual(False, c.modified()) self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8")) - def test_remove(self): - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n') as c: - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c)) - self.assertTrue("count1.txt" in c) - c.remove("count1.txt") - self.assertFalse("count1.txt" in c) - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c)) - - def test_remove_in_subdir(self): - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c: - c.remove("foo/count2.txt") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) - - def test_remove_subdir(self): - with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c: - c.remove("foo") - self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c)) - - def test_prefetch(self): - keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"}) - with import_manifest(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep=keep) as c: - r = c.open("count.txt", "r") - self.assertEqual("0123", r.read(4)) - self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8", "2e9ec317e197819358fbc43afca7d837+8", "e8dc4081b13434b45189a720b77b6818+8"], keep.requests) - class ArvadosFileReaderTestCase(StreamFileReaderTestCase): class MockParent(object): @@ -349,7 +389,7 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): def block_prefetch(self, loc): pass - def get_block(self, loc, num_retries=0, cache_only=False): + def get_block_contents(self, loc, num_retries=0, cache_only=False): if self.nocache and cache_only: return None return self.blocks[loc] @@ -357,10 +397,18 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): def __init__(self, blocks, nocache): self.blocks = blocks self.nocache = nocache + self.lock = arvados.arvfile.NoopLock() + + def root_collection(self): + return self 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 @@ -388,45 +436,156 @@ class ArvadosFileReaderTestCase(StreamFileReaderTestCase): sfile.read(5) self.assertEqual(3, sfile.tell()) -# class StreamReaderTestCase(unittest.TestCase, StreamRetryTestMixin): -# def reader_for(self, coll_name, **kwargs): -# return StreamReader(self.manifest_for(coll_name).split(), -# self.keep_client(), **kwargs) + def test_prefetch(self): + keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"}) + with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c: + r = c.open("count.txt", "r") + self.assertEqual("0123", r.read(4)) + self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests) + self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests) -# def read_for_test(self, reader, byte_count, **kwargs): -# return reader.readfrom(0, byte_count, **kwargs) + def test__eq__from_manifest(self): + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1: + with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2: + self.assertTrue(c1["count1.txt"] == c2["count1.txt"]) + self.assertFalse(c1["count1.txt"] != c2["count1.txt"]) -# def test_manifest_text_without_keep_client(self): -# mtext = self.manifest_for('multilevel_collection_1') -# for line in mtext.rstrip('\n').split('\n'): -# reader = StreamReader(line.split()) -# self.assertEqual(line + '\n', reader.manifest_text()) + 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") + self.assertTrue(c1["count1.txt"] == c2["count1.txt"]) + self.assertFalse(c1["count1.txt"] != c2["count1.txt"]) -# class StreamFileReadTestCase(unittest.TestCase, StreamRetryTestMixin): -# def reader_for(self, coll_name, **kwargs): -# return StreamReader(self.manifest_for(coll_name).split(), -# self.keep_client(), **kwargs).all_files()[0] + 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") -# def read_for_test(self, reader, byte_count, **kwargs): -# return reader.read(byte_count, **kwargs) + self.assertTrue(c1["count1.txt"] != c2["count1.txt"]) + self.assertFalse(c1["count1.txt"] == c2["count1.txt"]) -# class StreamFileReadFromTestCase(StreamFileReadTestCase): -# def read_for_test(self, reader, byte_count, **kwargs): -# return reader.readfrom(0, byte_count, **kwargs) +class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin): + def reader_for(self, coll_name, **kwargs): + stream = [] + segments = [] + n = 0 + for d in self.manifest_for(coll_name).split(): + try: + k = KeepLocator(d) + segments.append(Range(n, n, k.size)) + stream.append(Range(d, n, k.size)) + n += k.size + except ValueError: + pass + + blockmanager = arvados.arvfile._BlockManager(self.keep_client()) + blockmanager.prefetch_enabled = False + col = Collection(keep_client=self.keep_client(), block_manager=blockmanager) + af = ArvadosFile(col, + stream=stream, + segments=segments) + return ArvadosFileReader(af, "test", **kwargs) + def read_for_test(self, reader, byte_count, **kwargs): + return reader.read(byte_count, **kwargs) -# class StreamFileReadAllTestCase(StreamFileReadTestCase): -# def read_for_test(self, reader, byte_count, **kwargs): -# return ''.join(reader.readall(**kwargs)) +class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase): + def read_for_test(self, reader, byte_count, **kwargs): + return reader.readfrom(0, byte_count, **kwargs) -# class StreamFileReadAllDecompressedTestCase(StreamFileReadTestCase): -# def read_for_test(self, reader, byte_count, **kwargs): -# return ''.join(reader.readall_decompressed(**kwargs)) +class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase): + def read_for_test(self, reader, byte_count, **kwargs): + return ''.join(reader.readall(**kwargs)) -# class StreamFileReadlinesTestCase(StreamFileReadTestCase): -# def read_for_test(self, reader, byte_count, **kwargs): -# return ''.join(reader.readlines(**kwargs)) + +class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase): + def read_for_test(self, reader, byte_count, **kwargs): + return ''.join(reader.readall_decompressed(**kwargs)) + + +class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase): + def read_for_test(self, reader, byte_count, **kwargs): + return ''.join(reader.readlines(**kwargs)) + +class BlockManagerTest(unittest.TestCase): + def test_bufferblock_append(self): + keep = ArvadosFileWriterTestCase.MockKeep({}) + blockmanager = arvados.arvfile._BlockManager(keep) + 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.append("bar") + + 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") + + def test_bufferblock_dup(self): + keep = ArvadosFileWriterTestCase.MockKeep({}) + blockmanager = arvados.arvfile._BlockManager(keep) + 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) + + bufferblock2 = blockmanager.dup_block(bufferblock, None) + self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid) + + bufferblock2.append("bar") + + 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") + + def test_bufferblock_get(self): + keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"}) + blockmanager = arvados.arvfile._BlockManager(keep) + 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") + + 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) + + + 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.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)") + self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)