def test_keep_filtered_read_input():
import sys
sys.argv = ['--keep_filtered', 'test.fastq']
(min_len, keep_filtered, min_GC, max_GC, output_base_name,
file_name) = read_input()
assert keep_filtered
sys.argv = ['test.fastq']
(min_len, keep_filtered, min_GC, max_GC, output_base_name,
file_name) = read_input()
assert not keep_filtered
def test_GC_bounds_read_input():
import sys
sys.argv = ['--gc_bounds', '10', '75', 'test.fastq']
(min_len, keep_filtered, min_GC, max_GC, output_base_name,
file_name) = read_input()
assert max_GC == 75
assert min_GC == 10
print(OVERWRITE_WARNING)
if min_len == 0 and min_GC == 0 and max_GC == 0:
copy_fastq(output_base_name, file_name)
elif min_len == 0:
sort_only_GC_trash_filtered(min_GC, max_GC,
output_base_name, file_name)
elif min_len > 0:
if min_GC == 0 and max_GC == 100:
sort_only_len_trash_filtered(min_len,
output_base_name, file_name)
else:
sort_by_len_GC_trash_filtered(min_len, min_GC, max_GC,
output_base_name, file_name)
def copy_fastq(output_base_name: str, file_name: str):
print('Filtration arguments not found. Output file is equal to\
non filtrated file')
original = file_name
target = '{}_output.fastq'.format(output_base_name)
copyfile(original, target)
if __name__ == '__main__':
(min_len, keep_filtered, min_GC,
max_GC, output_base_name, file_name) = read_input()
main_sorter(min_len, keep_filtered, min_GC,
max_GC, output_base_name, file_name)
def test_file_name_read_input():
import sys
sys.argv = ['--min_length', '50', 'test.fastq']
(min_len, keep_filtered, min_GC, max_GC, output_base_name,
file_name) = read_input()
assert file_name == 'test.fastq'
