def test_pipeline_read_single_file(self): with TempDir() as tempdir: file_name = self._create_temp_vcf_file( _SAMPLE_HEADER_LINES + _SAMPLE_TEXT_LINES, tempdir) with TestPipeline() as pipeline: pcoll = pipeline | 'Read' >> ReadFromVcf(file_name) assert_that(pcoll, _count_equals_to(len(_SAMPLE_TEXT_LINES)))
def test_pipeline_read_file_pattern_large(self): pipeline = TestPipeline() pcoll = pipeline | 'Read' >> ReadFromVcf( os.path.join(get_full_dir(), 'valid-*.vcf')) assert_that(pcoll, _count_equals_to(9900)) pipeline.run()
def test_pipeline_read_single_file_large(self): pipeline = TestPipeline() pcoll = pipeline | 'Read' >> ReadFromVcf( get_full_file_path('valid-4.0.vcf')) assert_that(pcoll, _count_equals_to(5)) pipeline.run()