X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/fa739e32888d42bb99cef4781dc0c3f1c6190ac7..0e3df3f0e82757edf23e548149324d3eb3d23726:/sdk/python/test_collections.py diff --git a/sdk/python/test_collections.py b/sdk/python/test_collections.py index 08589ddc50..7df620d977 100644 --- a/sdk/python/test_collections.py +++ b/sdk/python/test_collections.py @@ -42,7 +42,7 @@ class LocalCollectionReaderTest(unittest.TestCase): os.environ['KEEP_LOCAL_STORE'] = '/tmp' LocalCollectionWriterTest().runTest() def runTest(self): - cr = arvados.CollectionReader('23ca013983d6239e98931cc779e68426+114') + cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle') got = [] for s in cr.all_streams(): for f in s.all_files(): @@ -51,19 +51,15 @@ class LocalCollectionReaderTest(unittest.TestCase): [3, '.', 'foo.txt', 'foo'], [3, './baz', 'baz.txt', 'baz']] self.assertEqual(got, - expected, - 'resulting file list is not what I expected') + expected) stream0 = cr.all_streams()[0] - self.assertEqual(stream0.read(0), + self.assertEqual(stream0.readfrom(0, 0), '', 'reading zero bytes should have returned empty string') - self.assertEqual(stream0.read(2**26), + self.assertEqual(stream0.readfrom(0, 2**26), 'foobar', 'reading entire stream failed') - self.assertEqual(stream0.read(2**26), - None, - 'reading past end of stream should have returned None') - self.assertEqual(stream0.read(0), + self.assertEqual(stream0.readfrom(2**26, 0), '', 'reading zero bytes should have returned empty string') @@ -72,7 +68,7 @@ class LocalCollectionManifestSubsetTest(unittest.TestCase): os.environ['KEEP_LOCAL_STORE'] = '/tmp' LocalCollectionWriterTest().runTest() def runTest(self): - self._runTest('23ca013983d6239e98931cc779e68426+114', + self._runTest('d6c3b8e571f1b81ebb150a45ed06c884+114', [[3, '.', 'bar.txt', 'bar'], [3, '.', 'foo.txt', 'foo'], [3, './baz', 'baz.txt', 'baz']]) @@ -91,24 +87,19 @@ class LocalCollectionManifestSubsetTest(unittest.TestCase): arvados.Keep.put("bar"))), [[2, '.', 'ar.txt', 'ar'], [2, '.', 'fo.txt', 'fo'], - [1, '.', 'ob.txt', 'o'], - [1, '.', 'ob.txt', 'b'], - [0, '.', 'zero.txt', ''],]) + [2, '.', 'ob.txt', 'ob'], + [0, '.', 'zero.txt', '']]) + def _runTest(self, collection, expected): cr = arvados.CollectionReader(collection) - manifest_subsets = [] for s in cr.all_streams(): - for f in s.all_files(): - manifest_subsets += [f.as_manifest()] - expect_i = 0 - for m in manifest_subsets: - cr = arvados.CollectionReader(m) - for f in cr.all_files(): - got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))] - self.assertEqual(got, - expected[expect_i], - 'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, expected[expect_i])) - expect_i += 1 + for ex in expected: + if ex[0] == s: + f = s.files()[ex[2]] + got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))] + self.assertEqual(got, + ex, + 'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex)) class LocalCollectionReadlineTest(unittest.TestCase): def setUp(self): @@ -138,6 +129,8 @@ class LocalCollectionEmptyFileTest(unittest.TestCase): cw = arvados.CollectionWriter() cw.start_new_file('zero.txt') cw.write('') + + self.assertEqual(cw.manifest_text(), ". 0:0:zero.txt\n") self.check_manifest_file_sizes(cw.manifest_text(), [0]) cw = arvados.CollectionWriter() cw.start_new_file('zero.txt') @@ -148,6 +141,7 @@ class LocalCollectionEmptyFileTest(unittest.TestCase): cw.start_new_file('zero.txt') cw.write('') self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0]) + def check_manifest_file_sizes(self, manifest_text, expect_sizes): cr = arvados.CollectionReader(manifest_text) got_sizes = [] @@ -170,6 +164,7 @@ class LocalCollectionBZ2DecompressionTest(unittest.TestCase): bz2_manifest = cw.manifest_text() cr = arvados.CollectionReader(bz2_manifest) + got = 0 for x in list(cr.all_files())[0].readlines(): self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x) @@ -272,6 +267,23 @@ class LocatorsAndRangesTest(unittest.TestCase): self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]]) self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), []) self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), []) + + self.assertEqual(arvados.locators_and_ranges(blocks2, 0, 2), [['a', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), []) + self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), []) + + self.assertEqual(arvados.locators_and_ranges(blocks2, 9, 2), [['a', 10, 9, 1], ['b', 10, 0, 1]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]]) + self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]]) + blocks3 = [['a', 10, 0], ['b', 10, 10], @@ -332,3 +344,130 @@ class LocatorsAndRangesTest(unittest.TestCase): self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14], ['c', 5, 0, 1]]) +class FileStreamTest(unittest.TestCase): + class MockStreamReader(object): + def __init__(self, content): + self.content = content + + def readfrom(self, start, size): + return self.content[start:start+size] + + def runTest(self): + content = 'abcdefghijklmnopqrstuvwxyz0123456789' + msr = FileStreamTest.MockStreamReader(content) + segments = [[0, 10, 0], + [10, 15, 10], + [25, 5, 25]] + + sfr = arvados.StreamFileReader(msr, segments, "test") + + self.assertEqual(sfr.name(), "test") + self.assertEqual(sfr.size(), 30) + + self.assertEqual(sfr.readfrom(0, 30), content[0:30]) + self.assertEqual(sfr.readfrom(2, 30), content[2:30]) + + self.assertEqual(sfr.readfrom(2, 8), content[2:10]) + self.assertEqual(sfr.readfrom(0, 10), content[0:10]) + + self.assertEqual(sfr.tell(), 0) + self.assertEqual(sfr.read(5), content[0:5]) + self.assertEqual(sfr.tell(), 5) + self.assertEqual(sfr.read(5), content[5:10]) + self.assertEqual(sfr.tell(), 10) + self.assertEqual(sfr.read(5), content[10:15]) + self.assertEqual(sfr.tell(), 15) + self.assertEqual(sfr.read(5), content[15:20]) + self.assertEqual(sfr.tell(), 20) + self.assertEqual(sfr.read(5), content[20:25]) + self.assertEqual(sfr.tell(), 25) + self.assertEqual(sfr.read(5), content[25:30]) + self.assertEqual(sfr.tell(), 30) + self.assertEqual(sfr.read(5), '') + self.assertEqual(sfr.tell(), 30) + + segments = [[26, 10, 0], + [0, 15, 10], + [15, 5, 25]] + + sfr = arvados.StreamFileReader(msr, segments, "test") + + self.assertEqual(sfr.size(), 30) + + self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20]) + self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20]) + + self.assertEqual(sfr.readfrom(2, 8), content[28:36]) + self.assertEqual(sfr.readfrom(0, 10), content[26:36]) + + self.assertEqual(sfr.tell(), 0) + self.assertEqual(sfr.read(5), content[26:31]) + self.assertEqual(sfr.tell(), 5) + self.assertEqual(sfr.read(5), content[31:36]) + self.assertEqual(sfr.tell(), 10) + self.assertEqual(sfr.read(5), content[0:5]) + self.assertEqual(sfr.tell(), 15) + self.assertEqual(sfr.read(5), content[5:10]) + self.assertEqual(sfr.tell(), 20) + self.assertEqual(sfr.read(5), content[10:15]) + self.assertEqual(sfr.tell(), 25) + self.assertEqual(sfr.read(5), content[15:20]) + self.assertEqual(sfr.tell(), 30) + self.assertEqual(sfr.read(5), '') + self.assertEqual(sfr.tell(), 30) + + +class StreamReaderTest(unittest.TestCase): + + class MockKeep(object): + def __init__(self, content): + self.content = content + + def get(self, locator): + return self.content[locator] + + def runTest(self): + keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij', + 'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy', + 'cccccccccccccccccccccccccccccccc+5': 'z0123'} + mk = StreamReaderTest.MockKeep(keepblocks) + + sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk) + + content = 'abcdefghijklmnopqrstuvwxyz0123456789' + + self.assertEqual(sr.readfrom(0, 30), content[0:30]) + self.assertEqual(sr.readfrom(2, 30), content[2:30]) + + self.assertEqual(sr.readfrom(2, 8), content[2:10]) + self.assertEqual(sr.readfrom(0, 10), content[0:10]) + + self.assertEqual(sr.readfrom(0, 5), content[0:5]) + self.assertEqual(sr.readfrom(5, 5), content[5:10]) + self.assertEqual(sr.readfrom(10, 5), content[10:15]) + self.assertEqual(sr.readfrom(15, 5), content[15:20]) + self.assertEqual(sr.readfrom(20, 5), content[20:25]) + self.assertEqual(sr.readfrom(25, 5), content[25:30]) + self.assertEqual(sr.readfrom(30, 5), '') + +class ExtractFileTest(unittest.TestCase): + def runTest(self): + m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt +. 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt +. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt +. 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt +. 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt""" + + m2 = arvados.CollectionReader(m1) + + self.assertEqual(m2.manifest_text(), + ". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt 43:41:md6sum.txt 84:43:md7sum.txt 6:37:md8sum.txt 84:43:md8sum.txt 83:1:md9sum.txt 0:43:md9sum.txt 84:36:md9sum.txt\n") + + self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(), + ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n") + self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(), + ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n") + self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(), + ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n") + self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(), + ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")