sdk/python/tests/test_stream.py

   1 from __future__ import absolute_import
   2 from builtins import object
   3 import bz2
   4 import gzip
   5 import io
   6 import mock
   7 import os
   8 import unittest
   9 import hashlib
  10
  11 import arvados
  12 from arvados import StreamReader, StreamFileReader
  13 from arvados._ranges import Range
  14
  15 from . import arvados_testutil as tutil
  16 from . import run_test_server
  17
  18 class StreamFileReaderTestCase(unittest.TestCase):
  19     def make_count_reader(self):
  20         stream = tutil.MockStreamReader('.', '01234', '34567', '67890')
  21         return StreamFileReader(stream, [Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)],
  22                                 'count.txt')
  23
  24     def test_read_block_crossing_behavior(self):
  25         # read() calls will be aligned on block boundaries - see #3663.
  26         sfile = self.make_count_reader()
  27         self.assertEqual(b'123', sfile.read(10))
  28
  29     def test_small_read(self):
  30         sfile = self.make_count_reader()
  31         self.assertEqual(b'12', sfile.read(2))
  32
  33     def test_successive_reads(self):
  34         sfile = self.make_count_reader()
  35         for expect in [b'123', b'456', b'789', b'']:
  36             self.assertEqual(expect, sfile.read(10))
  37
  38     def test_readfrom_spans_blocks(self):
  39         sfile = self.make_count_reader()
  40         self.assertEqual(b'6789', sfile.readfrom(5, 12))
  41
  42     def test_small_readfrom_spanning_blocks(self):
  43         sfile = self.make_count_reader()
  44         self.assertEqual(b'2345', sfile.readfrom(1, 4))
  45
  46     def test_readall(self):
  47         sfile = self.make_count_reader()
  48         self.assertEqual(b'123456789', b''.join(sfile.readall()))
  49
  50     def test_one_arg_seek(self):
  51         self.test_absolute_seek([])
  52
  53     def test_absolute_seek(self, args=[os.SEEK_SET]):
  54         sfile = self.make_count_reader()
  55         sfile.seek(6, *args)
  56         self.assertEqual(b'78', sfile.read(2))
  57         sfile.seek(4, *args)
  58         self.assertEqual(b'56', sfile.read(2))
  59
  60     def test_relative_seek(self, args=[os.SEEK_CUR]):
  61         sfile = self.make_count_reader()
  62         self.assertEqual(b'12', sfile.read(2))
  63         sfile.seek(2, *args)
  64         self.assertEqual(b'56', sfile.read(2))
  65
  66     def test_end_seek(self):
  67         sfile = self.make_count_reader()
  68         sfile.seek(-6, os.SEEK_END)
  69         self.assertEqual(b'45', sfile.read(2))
  70
  71     def test_seek_min_zero(self):
  72         sfile = self.make_count_reader()
  73         self.assertEqual(0, sfile.tell())
  74         with self.assertRaises(IOError):
  75             sfile.seek(-2, os.SEEK_SET)
  76         self.assertEqual(0, sfile.tell())
  77
  78     def test_seek_max_size(self):
  79         sfile = self.make_count_reader()
  80         sfile.seek(2, os.SEEK_END)
  81         # POSIX permits seeking past end of file.
  82         self.assertEqual(11, sfile.tell())
  83
  84     def test_size(self):
  85         self.assertEqual(9, self.make_count_reader().size())
  86
  87     def test_tell_after_block_read(self):
  88         sfile = self.make_count_reader()
  89         sfile.read(5)
  90         self.assertEqual(3, sfile.tell())
  91
  92     def test_tell_after_small_read(self):
  93         sfile = self.make_count_reader()
  94         sfile.read(1)
  95         self.assertEqual(1, sfile.tell())
  96
  97     def test_no_read_after_close(self):
  98         sfile = self.make_count_reader()
  99         sfile.close()
 100         self.assertRaises(ValueError, sfile.read, 2)
 101
 102     def test_context(self):
 103         with self.make_count_reader() as sfile:
 104             self.assertFalse(sfile.closed, "reader is closed inside context")
 105             self.assertEqual(b'12', sfile.read(2))
 106         self.assertTrue(sfile.closed, "reader is open after context")
 107
 108     def make_newlines_reader(self):
 109         stream = tutil.MockStreamReader('.', 'one\ntwo\n\nth', 'ree\nfour\n\n')
 110         return StreamFileReader(stream, [Range(0, 0, 11), Range(11, 11, 10)], 'count.txt')
 111
 112     def check_lines(self, actual):
 113         self.assertEqual(['one\n', 'two\n', '\n', 'three\n', 'four\n', '\n'],
 114                          actual)
 115
 116     def test_readline(self):
 117         reader = self.make_newlines_reader()
 118         actual = []
 119         while True:
 120             data = reader.readline()
 121             if not data:
 122                 break
 123             actual.append(data)
 124         self.check_lines(actual)
 125
 126     def test_readlines(self):
 127         self.check_lines(self.make_newlines_reader().readlines())
 128
 129     def test_iteration(self):
 130         self.check_lines(list(iter(self.make_newlines_reader())))
 131
 132     def test_readline_size(self):
 133         reader = self.make_newlines_reader()
 134         self.assertEqual('on', reader.readline(2))
 135         self.assertEqual('e\n', reader.readline(4))
 136         self.assertEqual('two\n', reader.readline(6))
 137         self.assertEqual('\n', reader.readline(8))
 138         self.assertEqual('thre', reader.readline(4))
 139
 140     def test_readlines_sizehint(self):
 141         result = self.make_newlines_reader().readlines(8)
 142         self.assertEqual(['one\n', 'two\n'], result[:2])
 143         self.assertNotIn('three\n', result)
 144
 145     def test_name_attribute(self):
 146         # Test both .name and .name() (for backward compatibility)
 147         stream = tutil.MockStreamReader()
 148         sfile = StreamFileReader(stream, [Range(0, 0, 0)], 'nametest')
 149         self.assertEqual('nametest', sfile.name)
 150         self.assertEqual('nametest', sfile.name())
 151
 152     def check_decompressed_name(self, filename, expect):
 153         stream = tutil.MockStreamReader('.', '')
 154         reader = StreamFileReader(stream, [Range(0, 0, 0)], filename)
 155         self.assertEqual(expect, reader.decompressed_name())
 156
 157     def test_decompressed_name_uncompressed_file(self):
 158         self.check_decompressed_name('test.log', 'test.log')
 159
 160     def test_decompressed_name_gzip_file(self):
 161         self.check_decompressed_name('test.log.gz', 'test.log')
 162
 163     def test_decompressed_name_bz2_file(self):
 164         self.check_decompressed_name('test.log.bz2', 'test.log')
 165
 166     def check_decompression(self, compress_ext, compress_func):
 167         test_text = b'decompression\ntest\n'
 168         test_data = compress_func(test_text)
 169         stream = tutil.MockStreamReader('.', test_data)
 170         reader = StreamFileReader(stream, [Range(0, 0, len(test_data))],
 171                                   'test.' + compress_ext)
 172         self.assertEqual(test_text, b''.join(reader.readall_decompressed()))
 173
 174     @staticmethod
 175     def gzip_compress(data):
 176         compressed_data = io.BytesIO()
 177         with gzip.GzipFile(fileobj=compressed_data, mode='wb') as gzip_file:
 178             gzip_file.write(data)
 179         return compressed_data.getvalue()
 180
 181     def test_no_decompression(self):
 182         self.check_decompression('log', lambda s: s)
 183
 184     def test_gzip_decompression(self):
 185         self.check_decompression('gz', self.gzip_compress)
 186
 187     def test_bz2_decompression(self):
 188         self.check_decompression('bz2', bz2.compress)
 189
 190     def test_readline_then_readlines(self):
 191         reader = self.make_newlines_reader()
 192         data = reader.readline()
 193         self.assertEqual('one\n', data)
 194         data = reader.readlines()
 195         self.assertEqual(['two\n', '\n', 'three\n', 'four\n', '\n'], data)
 196
 197     def test_readline_then_readall(self):
 198         reader = self.make_newlines_reader()
 199         data = reader.readline()
 200         self.assertEqual('one\n', data)
 201         self.assertEqual(b''.join([b'two\n', b'\n', b'three\n', b'four\n', b'\n']), b''.join(reader.readall()))
 202
 203
 204 class StreamRetryTestMixin(object):
 205     # Define reader_for(coll_name, **kwargs)
 206     # and read_for_test(reader, size, **kwargs).
 207     API_COLLECTIONS = run_test_server.fixture('collections')
 208
 209     def keep_client(self):
 210         return arvados.KeepClient(proxy='http://[%s]:1' % (tutil.TEST_HOST,),
 211                                   local_store='')
 212
 213     def manifest_for(self, coll_name):
 214         return self.API_COLLECTIONS[coll_name]['manifest_text']
 215
 216     @tutil.skip_sleep
 217     def test_success_without_retries(self):
 218         with tutil.mock_keep_responses('bar', 200):
 219             reader = self.reader_for('bar_file')
 220             self.assertEqual(b'bar', self.read_for_test(reader, 3))
 221
 222     @tutil.skip_sleep
 223     def test_read_no_default_retry(self):
 224         with tutil.mock_keep_responses('', 500):
 225             reader = self.reader_for('user_agreement')
 226             with self.assertRaises(arvados.errors.KeepReadError):
 227                 self.read_for_test(reader, 10)
 228
 229     @tutil.skip_sleep
 230     def test_read_with_instance_retries(self):
 231         with tutil.mock_keep_responses('foo', 500, 200):
 232             reader = self.reader_for('foo_file', num_retries=3)
 233             self.assertEqual(b'foo', self.read_for_test(reader, 3))
 234
 235     @tutil.skip_sleep
 236     def test_read_with_method_retries(self):
 237         with tutil.mock_keep_responses('foo', 500, 200):
 238             reader = self.reader_for('foo_file')
 239             self.assertEqual(b'foo',
 240                              self.read_for_test(reader, 3, num_retries=3))
 241
 242     @tutil.skip_sleep
 243     def test_read_instance_retries_exhausted(self):
 244         with tutil.mock_keep_responses('bar', 500, 500, 500, 500, 200):
 245             reader = self.reader_for('bar_file', num_retries=3)
 246             with self.assertRaises(arvados.errors.KeepReadError):
 247                 self.read_for_test(reader, 3)
 248
 249     @tutil.skip_sleep
 250     def test_read_method_retries_exhausted(self):
 251         with tutil.mock_keep_responses('bar', 500, 500, 500, 500, 200):
 252             reader = self.reader_for('bar_file')
 253             with self.assertRaises(arvados.errors.KeepReadError):
 254                 self.read_for_test(reader, 3, num_retries=3)
 255
 256     @tutil.skip_sleep
 257     def test_method_retries_take_precedence(self):
 258         with tutil.mock_keep_responses('', 500, 500, 500, 200):
 259             reader = self.reader_for('user_agreement', num_retries=10)
 260             with self.assertRaises(arvados.errors.KeepReadError):
 261                 self.read_for_test(reader, 10, num_retries=1)
 262
 263
 264 class StreamReaderTestCase(unittest.TestCase, StreamRetryTestMixin):
 265     def reader_for(self, coll_name, **kwargs):
 266         return StreamReader(self.manifest_for(coll_name).split(),
 267                             self.keep_client(), **kwargs)
 268
 269     def read_for_test(self, reader, byte_count, **kwargs):
 270         return reader.readfrom(0, byte_count, **kwargs)
 271
 272     def test_manifest_text_without_keep_client(self):
 273         mtext = self.manifest_for('multilevel_collection_1')
 274         for line in mtext.rstrip('\n').split('\n'):
 275             reader = StreamReader(line.split())
 276             self.assertEqual(line + '\n', reader.manifest_text())
 277
 278
 279 class StreamFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 280     def reader_for(self, coll_name, **kwargs):
 281         return StreamReader(self.manifest_for(coll_name).split(),
 282                             self.keep_client(), **kwargs).all_files()[0]
 283
 284     def read_for_test(self, reader, byte_count, **kwargs):
 285         return reader.read(byte_count, **kwargs)
 286
 287
 288 class StreamFileReadFromTestCase(StreamFileReadTestCase):
 289     def read_for_test(self, reader, byte_count, **kwargs):
 290         return reader.readfrom(0, byte_count, **kwargs)
 291
 292
 293 class StreamFileReadAllTestCase(StreamFileReadTestCase):
 294     def read_for_test(self, reader, byte_count, **kwargs):
 295         return b''.join(reader.readall(**kwargs))
 296
 297
 298 class StreamFileReadAllDecompressedTestCase(StreamFileReadTestCase):
 299     def read_for_test(self, reader, byte_count, **kwargs):
 300         return b''.join(reader.readall_decompressed(**kwargs))
 301
 302
 303 class StreamFileReadlinesTestCase(StreamFileReadTestCase):
 304     def read_for_test(self, reader, byte_count, **kwargs):
 305         return ''.join(reader.readlines(**kwargs)).encode()
 306
 307 if __name__ == '__main__':
 308     unittest.main()