def test_one_gzip_block(self):
with self._open_bgzf_block(self._file_name,
bgzf_io.Block(9287, 18988)) as file_to_read:
self._validate_first_line_is_complete(file_to_read.readline())
lines = self._read_all_lines(file_to_read)
self.assertEqual(len(lines), 43)
self._validate_last_line_is_complete(lines[-1])
def test_block_size_larger_than_gcs_buffer(self):
with self._open_bgzf_block(self._file_name,
bgzf_io.Block(9287, 949900),
read_buffer_size=64 * 1024) as file_to_read:
file_to_read._read_size = 64 * 1024
lines = self._read_all_lines(file_to_read)
self.assertEqual(len(lines), 425)
self._validate_last_line_is_complete(lines[-1])
def test_merge_chunks(self):
blocks = [bgzf_io.Block(start=0, end=100),
bgzf_io.Block(start=90, end=99),
bgzf_io.Block(start=100, end=150),
bgzf_io.Block(start=120, end=201),
bgzf_io.Block(start=201, end=300)]
merged_blocks = bgzf_io._merge_blocks(blocks, size_limit=200)
self.assertEqual(merged_blocks, [bgzf_io.Block(start=0, end=150),
bgzf_io.Block(start=150, end=300)])
def test_remove_invalid_blocks(self):
blocks = [bgzf_io.Block(start=5, end=5),
bgzf_io.Block(start=3, end=9)]
valid_blocks = bgzf_io._remove_invalid_blocks(blocks)
self.assertEqual(valid_blocks,
[bgzf_io.Block(start=3, end=9)])
def test_multiple_gzip_block(self):
with self._open_bgzf_block(self._file_name,
bgzf_io.Block(9287, 37595)) as file_to_read:
lines = self._read_all_lines(file_to_read)
self.assertEqual(len(lines), 52)
self._validate_last_line_is_complete(lines[-1])