def test_chunking_by_multiprocessing(self):
"""[fops] chunk by multiprocessing"""
input = 'test-data/test.read1.fa'
f_type, delim = fops.file_type(input)
chunks = fops.get_chunks(input, delim, split_type='pieces')
tf = fops.make_chunks(chunks, mp=True)
assert len(tf) == multiprocessing.cpu_count() - 1
self.clean(tf)
def test_chunking_by_size(self):
"""[fops] chunk by size"""
input = 'test-data/test.read1.fa'
f_type, delim = fops.file_type(input)
chunks = fops.get_chunks(input, delim, mb = 0.1, split_type='size')
tf = fops.make_chunks(chunks, mp=False)
# make sure chunks are ~ mb in size. problem is that the last one will almost alway be off
# so just round up and check if w/in 1 mb.
for f in tf:
sz = os.path.getsize(f)/1024.**2
self.failUnlessAlmostEqual(math.ceil(sz), 0.1, delta=1)
self.clean(tf)
def test_fastq_file_type(self):
"""[fops] fastq file type and delimiter check"""
input = 'test-data/test.read1.fq'
ft, delim = fops.file_type(input)
assert [ft, delim] == ['fastq','@']
def get_chunks(self, input):
f_type, delim = fops.file_type(input)
chunks = fops.get_chunks(input, delim, split_type='pieces')
values = fops.make_chunks(chunks, mp=False)
return values
