def fixed_tempfolder_processor(self, tmpdir, num_cores):
"""
Create processor with known temporary folder.
Provide some basic required values to the constructor, and critically,
fix the temporary folder to the location into which the individual
reads chunk files were ensured to exist. The processor constructor
provides tempfolder parent as a parameter, but then it creates a
temporary folder within that, used to search for the individual
chunk output files. Those files have already been created in a test
temp folder, though, so the processor needs to know about that.
Cleanup of both locations is handled. The processor registers its
own temporary folder for removal while pytest cleans up its folder.
Parameters
----------
tmpdir : py._path.local.LocalPath
Path to where chunks' dummy output files have been placed.
num_cores : int
Number of cores to use for the test case, parameterized.
Returns
-------
pararead.ParaReadProcessor
New processor instance, with updated temp folder knowledge.
"""
path_output_file = tmpdir.join("test-output.txt").strpath
processor = IdentityProcessor(PATH_ALIGNED_FILE,
cores=num_cores,
outfile=path_output_file)
processor.temp_folder = tmpdir.strpath
return processor
def test_different_format(self, tmpdir, filetype, combined_output_type,
which_names, extant_files, num_cores):
""" File content is actually combined, and formats can differ. """
# Manual creation of the processor here to control output type.
path_output_file = tmpdir.join(
"testfile.{}".format(combined_output_type)).strpath
processor = IdentityProcessor(PATH_ALIGNED_FILE,
cores=num_cores,
outfile=path_output_file,
intermediate_output_type=filetype)
processor.temp_folder = tmpdir.strpath
# Write to the dummy output file for each chunk.
expected_lines = {
fp: "file{}: {}\n".format(i, fp)
for i, fp in enumerate(extant_files)
}
for fp, line in expected_lines.items():
with open(fp, 'w') as f:
f.write(line)
# For control, enforce that combined output doesn't already exist.
assert not os.path.exists(path_output_file)
processor.combine(self.CHUNK_NAMES[which_names], strict=True)
assert os.path.isfile(path_output_file)
# Check that output was combined accurately.
with open(path_output_file, 'r') as combined:
observed_lines = combined.readlines()
assert set(expected_lines.values()) == set(observed_lines)