def estimate_current_remote_usage(remote, username, r_dir, l_dir, fastq2rsem_ratio):
"""
estimate the space that has already been or will be consumed by rsem_output
by walking through each GSM and computing the sum of their estimated usage,
if rsem.COMPLETE exists for a GSM, then ignore that GSM
mechanism: fetch the list of files in r_dir, and find the
fastq.gz for each GSM, then find the corresponding fastq.gz in
l_dir, and estimate sizes based on them
:param find_cmd: should be in the form of find {remote_dir}
:param r_dir: remote rsem output directory
:param l_dir: local rsem output directory
"""
files = fetch_remote_file_list(remote, username, r_dir)
usage = 0
for dir_ in sorted(files):
match = re.search(r'(GSM\d+$)', os.path.basename(dir_))
if match:
rsem_comp = os.path.join(dir_, 'rsem.COMPLETE')
if (not rsem_comp in files) and (not misc.is_empty_dir(dir_, files)):
# only count the disk spaces used by those GSMs that are
# being processed
gsm_dir = dir_.replace(r_dir, l_dir)
usage += estimate_rsem_usage(gsm_dir, fastq2rsem_ratio)
return usage
def test_is_empty_dir(self):
self.assertFalse(misc.is_empty_dir('/p', ['/p', '/p/a.txt']))
self.assertTrue(misc.is_empty_dir('/p', ['/p', '/s', '/s/a.txt']))
def test_is_empty_dir(self):
self.assertFalse(misc.is_empty_dir('/p', ['/p', '/p/a.txt']))
self.assertTrue(misc.is_empty_dir('/p', ['/p', '/s', '/s/a.txt']))
