sdk/python/tests/test_stream.py

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