Example #1
0
    def test_copytree(self):
        src_dir_name = 'gs://gcsio-test/source/'
        dest_dir_name = 'gs://gcsio-test/dest/'
        file_size = 1024
        paths = ['a', 'b/c', 'b/d']
        for path in paths:
            src_file_name = src_dir_name + path
            dest_file_name = dest_dir_name + path
            self._insert_random_file(self.client, src_file_name, file_size)
            self.assertTrue(
                gcsio.parse_gcs_path(src_file_name) in
                self.client.objects.files)
            self.assertFalse(
                gcsio.parse_gcs_path(dest_file_name) in
                self.client.objects.files)

        self.gcs.copytree(src_dir_name, dest_dir_name)

        for path in paths:
            src_file_name = src_dir_name + path
            dest_file_name = dest_dir_name + path
            self.assertTrue(
                gcsio.parse_gcs_path(src_file_name) in
                self.client.objects.files)
            self.assertTrue(
                gcsio.parse_gcs_path(dest_file_name) in
                self.client.objects.files)
Example #2
0
  def test_delete(self):
    file_name = 'gs://gcsio-test/delete_me'
    file_size = 1024

    # Test deletion of non-existent file.
    self.gcs.delete(file_name)

    self._insert_random_file(self.client, file_name, file_size)
    self.assertTrue(
        gcsio.parse_gcs_path(file_name) in self.client.objects.files)

    self.gcs.delete(file_name)

    self.assertFalse(
        gcsio.parse_gcs_path(file_name) in self.client.objects.files)
Example #3
0
    def test_delete(self):
        file_name = 'gs://gcsio-test/delete_me'
        file_size = 1024

        # Test deletion of non-existent file.
        self.gcs.delete(file_name)

        self._insert_random_file(self.client, file_name, file_size)
        self.assertTrue(
            gcsio.parse_gcs_path(file_name) in self.client.objects.files)

        self.gcs.delete(file_name)

        self.assertFalse(
            gcsio.parse_gcs_path(file_name) in self.client.objects.files)
Example #4
0
    def test_rename(self):
        src_file_name = 'gs://gcsio-test/source'
        dest_file_name = 'gs://gcsio-test/dest'
        file_size = 1024
        self._insert_random_file(self.client, src_file_name, file_size)
        self.assertTrue(
            gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
        self.assertFalse(
            gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

        self.gcs.rename(src_file_name, dest_file_name)

        self.assertFalse(
            gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
        self.assertTrue(
            gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)
Example #5
0
  def test_rename(self):
    src_file_name = 'gs://gcsio-test/source'
    dest_file_name = 'gs://gcsio-test/dest'
    file_size = 1024
    self._insert_random_file(self.client, src_file_name, file_size)
    self.assertTrue(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertFalse(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.gcs.rename(src_file_name, dest_file_name)

    self.assertFalse(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertTrue(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)
Example #6
0
  def test_copy(self):
    src_file_name = 'gs://gcsio-test/source'
    dest_file_name = 'gs://gcsio-test/dest'
    file_size = 1024
    self._insert_random_file(self.client, src_file_name, file_size)
    self.assertTrue(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertFalse(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.gcs.copy(src_file_name, dest_file_name)

    self.assertTrue(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertTrue(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.assertRaises(IOError, self.gcs.copy, 'gs://gcsio-test/non-existent',
                      'gs://gcsio-test/non-existent-destination')
Example #7
0
  def test_copy(self):
    src_file_name = 'gs://gcsio-test/source'
    dest_file_name = 'gs://gcsio-test/dest'
    file_size = 1024
    self._insert_random_file(self.client, src_file_name, file_size)
    self.assertTrue(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertFalse(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.gcs.copy(src_file_name, dest_file_name)

    self.assertTrue(
        gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
    self.assertTrue(
        gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.assertRaises(IOError, self.gcs.copy, 'gs://gcsio-test/non-existent',
                      'gs://gcsio-test/non-existent-destination')
 def setUp(self):
     with open(testdata_util.get_full_file_path('Y.vcf.bgz'),
               mode='rb') as file_to_read:
         data = file_to_read.readlines()
     self._data = b''.join(data)
     self.client = gcsio_test.FakeGcsClient()
     self.gcs = gcsio.GcsIO(self.client)
     self._file_name = 'gs://bucket/test'
     bucket, name = gcsio.parse_gcs_path(self._file_name)
     self.client.objects.add_file(
         gcsio_test.FakeFile(bucket, name, self._data, 1))
Example #9
0
 def test_file_close(self):
   file_name = 'gs://gcsio-test/close_file'
   file_size = 5 * 1024 * 1024 + 2000
   contents = os.urandom(file_size)
   f = self.gcs.open(file_name, 'w')
   self.assertEqual(f.mode, 'w')
   f.write(contents)
   f.close()
   f.close()  # This should not crash.
   bucket, name = gcsio.parse_gcs_path(file_name)
   self.assertEqual(
       self.client.objects.get_file(bucket, name).contents, contents)
Example #10
0
 def test_file_close(self):
     file_name = 'gs://gcsio-test/close_file'
     file_size = 5 * 1024 * 1024 + 2000
     contents = os.urandom(file_size)
     f = self.gcs.open(file_name, 'w')
     self.assertEqual(f.mode, 'w')
     f.write(contents)
     f.close()
     f.close()  # This should not crash.
     bucket, name = gcsio.parse_gcs_path(file_name)
     self.assertEqual(
         self.client.objects.get_file(bucket, name).contents, contents)
Example #11
0
 def test_file_write(self):
     file_name = 'gs://gcsio-test/write_file'
     file_size = 5 * 1024 * 1024 + 2000
     contents = os.urandom(file_size)
     f = self.gcs.open(file_name, 'w')
     self.assertEqual(f.mode, 'w')
     f.write(contents[0:1000])
     f.write(contents[1000:1024 * 1024])
     f.write(contents[1024 * 1024:])
     f.close()
     bucket, name = gcsio.parse_gcs_path(file_name)
     self.assertEqual(
         self.client.objects.get_file(bucket, name).contents, contents)
Example #12
0
 def test_file_write(self):
   file_name = 'gs://gcsio-test/write_file'
   file_size = 5 * 1024 * 1024 + 2000
   contents = os.urandom(file_size)
   f = self.gcs.open(file_name, 'w')
   self.assertEqual(f.mode, 'w')
   f.write(contents[0:1000])
   f.write(contents[1000:1024 * 1024])
   f.write(contents[1024 * 1024:])
   f.close()
   bucket, name = gcsio.parse_gcs_path(file_name)
   self.assertEqual(
       self.client.objects.get_file(bucket, name).contents, contents)
Example #13
0
    def test_file_read_line(self):
        file_name = 'gs://gcsio-test/read_line_file'
        lines = []

        # Set a small buffer size to exercise refilling the buffer.
        # First line is carefully crafted so the newline falls as the last character
        # of the buffer to exercise this code path.
        read_buffer_size = 1024
        lines.append('x' * 1023 + '\n')

        for _ in range(1, 1000):
            line_length = random.randint(100, 500)
            line = os.urandom(line_length).replace('\n', ' ') + '\n'
            lines.append(line)
        contents = ''.join(lines)

        file_size = len(contents)
        bucket, name = gcsio.parse_gcs_path(file_name)
        self.client.objects.add_file(FakeFile(bucket, name, contents, 1))

        f = self.gcs.open(file_name, read_buffer_size=read_buffer_size)

        # Test read of first two lines.
        f.seek(0)
        self.assertEqual(f.readline(), lines[0])
        self.assertEqual(f.tell(), len(lines[0]))
        self.assertEqual(f.readline(), lines[1])

        # Test read at line boundary.
        f.seek(file_size - len(lines[-1]) - 1)
        self.assertEqual(f.readline(), '\n')

        # Test read at end of file.
        f.seek(file_size)
        self.assertEqual(f.readline(), '')

        # Test reads at random positions.
        random.seed(0)
        for _ in range(0, 10):
            start = random.randint(0, file_size - 1)
            line_index = 0
            # Find line corresponding to start index.
            chars_left = start
            while True:
                next_line_length = len(lines[line_index])
                if chars_left - next_line_length < 0:
                    break
                chars_left -= next_line_length
                line_index += 1
            f.seek(start)
            self.assertEqual(f.readline(), lines[line_index][chars_left:])
Example #14
0
  def test_file_read_line(self):
    file_name = 'gs://gcsio-test/read_line_file'
    lines = []

    # Set a small buffer size to exercise refilling the buffer.
    # First line is carefully crafted so the newline falls as the last character
    # of the buffer to exercise this code path.
    read_buffer_size = 1024
    lines.append('x' * 1023 + '\n')

    for _ in range(1, 1000):
      line_length = random.randint(100, 500)
      line = os.urandom(line_length).replace('\n', ' ') + '\n'
      lines.append(line)
    contents = ''.join(lines)

    file_size = len(contents)
    bucket, name = gcsio.parse_gcs_path(file_name)
    self.client.objects.add_file(FakeFile(bucket, name, contents, 1))

    f = self.gcs.open(file_name, read_buffer_size=read_buffer_size)

    # Test read of first two lines.
    f.seek(0)
    self.assertEqual(f.readline(), lines[0])
    self.assertEqual(f.tell(), len(lines[0]))
    self.assertEqual(f.readline(), lines[1])

    # Test read at line boundary.
    f.seek(file_size - len(lines[-1]) - 1)
    self.assertEqual(f.readline(), '\n')

    # Test read at end of file.
    f.seek(file_size)
    self.assertEqual(f.readline(), '')

    # Test reads at random positions.
    random.seed(0)
    for _ in range(0, 10):
      start = random.randint(0, file_size - 1)
      line_index = 0
      # Find line corresponding to start index.
      chars_left = start
      while True:
        next_line_length = len(lines[line_index])
        if chars_left - next_line_length < 0:
          break
        chars_left -= next_line_length
        line_index += 1
      f.seek(start)
      self.assertEqual(f.readline(), lines[line_index][chars_left:])
Example #15
0
  def test_copytree(self):
    src_dir_name = 'gs://gcsio-test/source/'
    dest_dir_name = 'gs://gcsio-test/dest/'
    file_size = 1024
    paths = ['a', 'b/c', 'b/d']
    for path in paths:
      src_file_name = src_dir_name + path
      dest_file_name = dest_dir_name + path
      self._insert_random_file(self.client, src_file_name, file_size)
      self.assertTrue(
          gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
      self.assertFalse(
          gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)

    self.gcs.copytree(src_dir_name, dest_dir_name)

    for path in paths:
      src_file_name = src_dir_name + path
      dest_file_name = dest_dir_name + path
      self.assertTrue(
          gcsio.parse_gcs_path(src_file_name) in self.client.objects.files)
      self.assertTrue(
          gcsio.parse_gcs_path(dest_file_name) in self.client.objects.files)
Example #16
0
 def test_file_flush(self):
   file_name = 'gs://gcsio-test/flush_file'
   file_size = 5 * 1024 * 1024 + 2000
   contents = os.urandom(file_size)
   bucket, name = gcsio.parse_gcs_path(file_name)
   f = self.gcs.open(file_name, 'w')
   self.assertEqual(f.mode, 'w')
   f.write(contents[0:1000])
   f.flush()
   f.write(contents[1000:1024 * 1024])
   f.flush()
   f.flush()  # Should be a NOOP.
   f.write(contents[1024 * 1024:])
   f.close()  # This should already call the equivalent of flush() in its body.
   self.assertEqual(
       self.client.objects.get_file(bucket, name).contents, contents)
Example #17
0
 def test_file_flush(self):
   file_name = 'gs://gcsio-test/flush_file'
   file_size = 5 * 1024 * 1024 + 2000
   contents = os.urandom(file_size)
   bucket, name = gcsio.parse_gcs_path(file_name)
   f = self.gcs.open(file_name, 'w')
   self.assertEqual(f.mode, 'w')
   f.write(contents[0:1000])
   f.flush()
   f.write(contents[1000:1024 * 1024])
   f.flush()
   f.flush()  # Should be a NOOP.
   f.write(contents[1024 * 1024:])
   f.close()  # This should already call the equivalent of flush() in its body.
   self.assertEqual(
       self.client.objects.get_file(bucket, name).contents, contents)
Example #18
0
    def test_context_manager(self):
        # Test writing with a context manager.
        file_name = 'gs://gcsio-test/context_manager_file'
        file_size = 1024
        contents = os.urandom(file_size)
        with self.gcs.open(file_name, 'w') as f:
            f.write(contents)
        bucket, name = gcsio.parse_gcs_path(file_name)
        self.assertEqual(
            self.client.objects.get_file(bucket, name).contents, contents)

        # Test reading with a context manager.
        with self.gcs.open(file_name) as f:
            self.assertEqual(f.read(), contents)

        # Test that exceptions are not swallowed by the context manager.
        with self.assertRaises(ZeroDivisionError):
            with self.gcs.open(file_name) as f:
                f.read(0 / 0)
Example #19
0
  def test_context_manager(self):
    # Test writing with a context manager.
    file_name = 'gs://gcsio-test/context_manager_file'
    file_size = 1024
    contents = os.urandom(file_size)
    with self.gcs.open(file_name, 'w') as f:
      f.write(contents)
    bucket, name = gcsio.parse_gcs_path(file_name)
    self.assertEqual(
        self.client.objects.get_file(bucket, name).contents, contents)

    # Test reading with a context manager.
    with self.gcs.open(file_name) as f:
      self.assertEqual(f.read(), contents)

    # Test that exceptions are not swallowed by the context manager.
    with self.assertRaises(ZeroDivisionError):
      with self.gcs.open(file_name) as f:
        f.read(0 / 0)
Example #20
0
    def test_file_iterator(self):
        file_name = 'gs://gcsio-test/iterating_file'
        lines = []
        line_count = 10
        for _ in range(line_count):
            line_length = random.randint(100, 500)
            line = os.urandom(line_length).replace('\n', ' ') + '\n'
            lines.append(line)

        contents = ''.join(lines)
        bucket, name = gcsio.parse_gcs_path(file_name)
        self.client.objects.add_file(FakeFile(bucket, name, contents, 1))

        f = self.gcs.open(file_name)

        read_lines = 0
        for line in f:
            read_lines += 1

        self.assertEqual(read_lines, line_count)
Example #21
0
  def test_file_iterator(self):
    file_name = 'gs://gcsio-test/iterating_file'
    lines = []
    line_count = 10
    for _ in range(line_count):
      line_length = random.randint(100, 500)
      line = os.urandom(line_length).replace('\n', ' ') + '\n'
      lines.append(line)

    contents = ''.join(lines)
    bucket, name = gcsio.parse_gcs_path(file_name)
    self.client.objects.add_file(FakeFile(bucket, name, contents, 1))

    f = self.gcs.open(file_name)

    read_lines = 0
    for line in f:
      read_lines += 1

    self.assertEqual(read_lines, line_count)
Example #22
0
 def test_gcs_path(self):
   self.assertEqual(
       gcsio.parse_gcs_path('gs://bucket/name'), ('bucket', 'name'))
   self.assertEqual(
       gcsio.parse_gcs_path('gs://bucket/name/sub'), ('bucket', 'name/sub'))
Example #23
0
 def _insert_random_file(self, client, path, size, generation=1):
   bucket, name = gcsio.parse_gcs_path(path)
   f = FakeFile(bucket, name, os.urandom(size), generation)
   client.objects.add_file(f)
   return f
Example #24
0
 def _insert_random_file(self, client, path, size, generation=1):
     bucket, name = gcsio.parse_gcs_path(path)
     f = FakeFile(bucket, name, os.urandom(size), generation)
     client.objects.add_file(f)
     return f
Example #25
0
 def test_gcs_path(self):
     self.assertEqual(gcsio.parse_gcs_path('gs://bucket/name'),
                      ('bucket', 'name'))
     self.assertEqual(gcsio.parse_gcs_path('gs://bucket/name/sub'),
                      ('bucket', 'name/sub'))